diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,392689 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 29.999564130210363,
+  "eval_steps": 500,
+  "global_step": 653850,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4999999999999998e-06,
+      "loss": 10.6407,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999999999999997e-06,
+      "loss": 10.6418,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.499999999999999e-06,
+      "loss": 10.6347,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999999999999999e-06,
+      "loss": 10.6297,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.499999999999999e-06,
+      "loss": 10.6259,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.999999999999999e-06,
+      "loss": 10.6165,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.05e-05,
+      "loss": 10.6199,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1999999999999999e-05,
+      "loss": 10.6069,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3499999999999998e-05,
+      "loss": 10.5906,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4999999999999999e-05,
+      "loss": 10.5707,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6499999999999998e-05,
+      "loss": 10.5153,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7999999999999997e-05,
+      "loss": 10.4083,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.95e-05,
+      "loss": 10.2319,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1e-05,
+      "loss": 10.0177,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2499999999999998e-05,
+      "loss": 9.7888,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3999999999999997e-05,
+      "loss": 9.5083,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.55e-05,
+      "loss": 9.2598,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6999999999999996e-05,
+      "loss": 9.0536,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 8.6742,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 8.3333,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.149999999999999e-05,
+      "loss": 8.0746,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2999999999999996e-05,
+      "loss": 7.7664,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.45e-05,
+      "loss": 7.4577,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5999999999999994e-05,
+      "loss": 7.2412,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.75e-05,
+      "loss": 6.9368,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9e-05,
+      "loss": 6.8098,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.05e-05,
+      "loss": 6.7295,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2e-05,
+      "loss": 6.4451,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3499999999999993e-05,
+      "loss": 6.3628,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 6.3733,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.65e-05,
+      "loss": 6.1627,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7999999999999994e-05,
+      "loss": 6.3107,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.95e-05,
+      "loss": 6.1892,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1e-05,
+      "loss": 6.0842,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2499999999999995e-05,
+      "loss": 6.2253,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 6.064,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5499999999999994e-05,
+      "loss": 6.0779,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 6.0491,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.85e-05,
+      "loss": 6.0538,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 6.0028,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.149999999999999e-05,
+      "loss": 6.0339,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.299999999999999e-05,
+      "loss": 6.0346,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.45e-05,
+      "loss": 5.9224,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 6.0043,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.75e-05,
+      "loss": 5.9748,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.9e-05,
+      "loss": 6.0752,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.049999999999999e-05,
+      "loss": 6.0394,
+      "step": 470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.199999999999999e-05,
+      "loss": 6.1646,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.35e-05,
+      "loss": 6.0321,
+      "step": 490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.5e-05,
+      "loss": 6.032,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.499885207010024e-05,
+      "loss": 5.9603,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.499770414020049e-05,
+      "loss": 6.0339,
+      "step": 520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.499655621030076e-05,
+      "loss": 6.0068,
+      "step": 530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.4995408280401e-05,
+      "loss": 6.0533,
+      "step": 540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.499426035050126e-05,
+      "loss": 5.9862,
+      "step": 550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.49931124206015e-05,
+      "loss": 5.9503,
+      "step": 560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.499196449070177e-05,
+      "loss": 5.9781,
+      "step": 570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.4990816560802e-05,
+      "loss": 5.9835,
+      "step": 580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.498966863090227e-05,
+      "loss": 5.8957,
+      "step": 590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.498852070100252e-05,
+      "loss": 5.9596,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.498737277110277e-05,
+      "loss": 5.9407,
+      "step": 610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.498622484120302e-05,
+      "loss": 5.926,
+      "step": 620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.498507691130328e-05,
+      "loss": 5.9378,
+      "step": 630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.498392898140353e-05,
+      "loss": 5.9211,
+      "step": 640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.498278105150378e-05,
+      "loss": 5.8627,
+      "step": 650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.498163312160403e-05,
+      "loss": 5.9214,
+      "step": 660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.498048519170428e-05,
+      "loss": 5.851,
+      "step": 670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.497933726180455e-05,
+      "loss": 5.9169,
+      "step": 680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.49781893319048e-05,
+      "loss": 5.8954,
+      "step": 690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.497704140200504e-05,
+      "loss": 5.826,
+      "step": 700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.49758934721053e-05,
+      "loss": 5.7874,
+      "step": 710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.497474554220556e-05,
+      "loss": 5.7865,
+      "step": 720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.49735976123058e-05,
+      "loss": 5.799,
+      "step": 730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.497244968240606e-05,
+      "loss": 5.7739,
+      "step": 740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.497130175250631e-05,
+      "loss": 5.7765,
+      "step": 750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.497015382260656e-05,
+      "loss": 5.8061,
+      "step": 760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.496900589270681e-05,
+      "loss": 5.79,
+      "step": 770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.496785796280707e-05,
+      "loss": 5.734,
+      "step": 780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.496671003290732e-05,
+      "loss": 5.6869,
+      "step": 790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.496556210300757e-05,
+      "loss": 5.6226,
+      "step": 800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.496441417310782e-05,
+      "loss": 5.7269,
+      "step": 810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.496326624320807e-05,
+      "loss": 5.5789,
+      "step": 820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.496211831330833e-05,
+      "loss": 5.6637,
+      "step": 830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.496097038340858e-05,
+      "loss": 5.5877,
+      "step": 840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.495982245350883e-05,
+      "loss": 5.5361,
+      "step": 850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.495867452360908e-05,
+      "loss": 5.5174,
+      "step": 860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.495752659370935e-05,
+      "loss": 5.5792,
+      "step": 870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.495637866380958e-05,
+      "loss": 5.5549,
+      "step": 880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.495523073390985e-05,
+      "loss": 5.5175,
+      "step": 890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.49540828040101e-05,
+      "loss": 5.5307,
+      "step": 900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.495293487411035e-05,
+      "loss": 5.5238,
+      "step": 910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.49517869442106e-05,
+      "loss": 5.4601,
+      "step": 920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.495063901431086e-05,
+      "loss": 5.4516,
+      "step": 930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.494949108441111e-05,
+      "loss": 5.4186,
+      "step": 940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.494834315451136e-05,
+      "loss": 5.3155,
+      "step": 950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.494719522461161e-05,
+      "loss": 5.3367,
+      "step": 960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.494604729471186e-05,
+      "loss": 5.2882,
+      "step": 970
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.494489936481212e-05,
+      "loss": 5.269,
+      "step": 980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.494375143491237e-05,
+      "loss": 5.2941,
+      "step": 990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.494260350501262e-05,
+      "loss": 5.2201,
+      "step": 1000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.494145557511287e-05,
+      "loss": 5.1349,
+      "step": 1010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.494030764521313e-05,
+      "loss": 5.229,
+      "step": 1020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.493915971531337e-05,
+      "loss": 4.9812,
+      "step": 1030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.493801178541363e-05,
+      "loss": 5.1464,
+      "step": 1040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.493686385551388e-05,
+      "loss": 5.1863,
+      "step": 1050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.493571592561413e-05,
+      "loss": 4.9998,
+      "step": 1060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.493456799571438e-05,
+      "loss": 4.9957,
+      "step": 1070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.493342006581465e-05,
+      "loss": 5.0263,
+      "step": 1080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.49322721359149e-05,
+      "loss": 4.9606,
+      "step": 1090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.493112420601515e-05,
+      "loss": 5.1051,
+      "step": 1100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.49299762761154e-05,
+      "loss": 4.9715,
+      "step": 1110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.492882834621565e-05,
+      "loss": 4.9413,
+      "step": 1120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.492768041631591e-05,
+      "loss": 4.9109,
+      "step": 1130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.492653248641616e-05,
+      "loss": 4.7654,
+      "step": 1140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.492538455651641e-05,
+      "loss": 4.806,
+      "step": 1150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.492423662661666e-05,
+      "loss": 4.8094,
+      "step": 1160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.492308869671692e-05,
+      "loss": 4.7146,
+      "step": 1170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.492194076681716e-05,
+      "loss": 4.8585,
+      "step": 1180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.492079283691742e-05,
+      "loss": 4.8578,
+      "step": 1190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.491964490701767e-05,
+      "loss": 4.5756,
+      "step": 1200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.491849697711792e-05,
+      "loss": 4.8017,
+      "step": 1210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.491734904721817e-05,
+      "loss": 4.6589,
+      "step": 1220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.491620111731842e-05,
+      "loss": 4.691,
+      "step": 1230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.491505318741869e-05,
+      "loss": 4.5795,
+      "step": 1240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.491390525751894e-05,
+      "loss": 4.5849,
+      "step": 1250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.491275732761919e-05,
+      "loss": 4.6512,
+      "step": 1260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.491160939771944e-05,
+      "loss": 4.4761,
+      "step": 1270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.49104614678197e-05,
+      "loss": 4.3574,
+      "step": 1280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.490931353791994e-05,
+      "loss": 4.6313,
+      "step": 1290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.49081656080202e-05,
+      "loss": 4.3773,
+      "step": 1300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.490701767812045e-05,
+      "loss": 4.5851,
+      "step": 1310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.49058697482207e-05,
+      "loss": 4.3018,
+      "step": 1320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.490472181832095e-05,
+      "loss": 4.41,
+      "step": 1330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.490357388842121e-05,
+      "loss": 4.2652,
+      "step": 1340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.490242595852146e-05,
+      "loss": 4.2841,
+      "step": 1350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.490127802862171e-05,
+      "loss": 4.4485,
+      "step": 1360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.490013009872196e-05,
+      "loss": 4.1498,
+      "step": 1370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.489898216882221e-05,
+      "loss": 4.279,
+      "step": 1380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.489783423892247e-05,
+      "loss": 4.2893,
+      "step": 1390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.489668630902272e-05,
+      "loss": 4.1882,
+      "step": 1400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.489553837912297e-05,
+      "loss": 4.2559,
+      "step": 1410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.489439044922322e-05,
+      "loss": 4.3,
+      "step": 1420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.489324251932349e-05,
+      "loss": 4.0547,
+      "step": 1430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.489209458942372e-05,
+      "loss": 4.1308,
+      "step": 1440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.489094665952399e-05,
+      "loss": 4.1885,
+      "step": 1450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.488979872962424e-05,
+      "loss": 4.1336,
+      "step": 1460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.488865079972449e-05,
+      "loss": 4.3071,
+      "step": 1470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.488750286982474e-05,
+      "loss": 4.1681,
+      "step": 1480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.4886354939925e-05,
+      "loss": 4.0325,
+      "step": 1490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.488520701002525e-05,
+      "loss": 4.0338,
+      "step": 1500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.48840590801255e-05,
+      "loss": 4.1194,
+      "step": 1510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.488291115022575e-05,
+      "loss": 3.9066,
+      "step": 1520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.4881763220326e-05,
+      "loss": 3.9887,
+      "step": 1530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.488061529042626e-05,
+      "loss": 4.2705,
+      "step": 1540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.487946736052651e-05,
+      "loss": 3.8695,
+      "step": 1550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.487831943062676e-05,
+      "loss": 3.9867,
+      "step": 1560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.487717150072701e-05,
+      "loss": 4.0012,
+      "step": 1570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.487602357082728e-05,
+      "loss": 4.0483,
+      "step": 1580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.487487564092751e-05,
+      "loss": 3.7693,
+      "step": 1590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.487372771102778e-05,
+      "loss": 3.9504,
+      "step": 1600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.487257978112803e-05,
+      "loss": 4.0404,
+      "step": 1610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.487143185122828e-05,
+      "loss": 3.7446,
+      "step": 1620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.487028392132853e-05,
+      "loss": 3.7566,
+      "step": 1630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.486913599142879e-05,
+      "loss": 3.888,
+      "step": 1640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.486798806152904e-05,
+      "loss": 3.8926,
+      "step": 1650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.486684013162929e-05,
+      "loss": 3.7637,
+      "step": 1660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.486569220172954e-05,
+      "loss": 3.7601,
+      "step": 1670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.486454427182979e-05,
+      "loss": 3.661,
+      "step": 1680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.486339634193005e-05,
+      "loss": 3.7864,
+      "step": 1690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.48622484120303e-05,
+      "loss": 3.5535,
+      "step": 1700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.486110048213055e-05,
+      "loss": 3.5624,
+      "step": 1710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.48599525522308e-05,
+      "loss": 3.8567,
+      "step": 1720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.485880462233106e-05,
+      "loss": 3.8366,
+      "step": 1730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.48576566924313e-05,
+      "loss": 3.7124,
+      "step": 1740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.485650876253156e-05,
+      "loss": 3.6912,
+      "step": 1750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.485536083263181e-05,
+      "loss": 3.8285,
+      "step": 1760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.485421290273206e-05,
+      "loss": 3.6731,
+      "step": 1770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.485306497283231e-05,
+      "loss": 3.9717,
+      "step": 1780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.485191704293258e-05,
+      "loss": 3.597,
+      "step": 1790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.485076911303283e-05,
+      "loss": 3.6297,
+      "step": 1800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.484962118313308e-05,
+      "loss": 3.7053,
+      "step": 1810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.484847325323333e-05,
+      "loss": 3.4267,
+      "step": 1820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.484732532333358e-05,
+      "loss": 3.6189,
+      "step": 1830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.484617739343384e-05,
+      "loss": 3.6977,
+      "step": 1840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.484502946353409e-05,
+      "loss": 3.5919,
+      "step": 1850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.484388153363434e-05,
+      "loss": 3.5321,
+      "step": 1860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.484273360373459e-05,
+      "loss": 3.6587,
+      "step": 1870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.484158567383485e-05,
+      "loss": 3.5794,
+      "step": 1880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.484043774393509e-05,
+      "loss": 3.3874,
+      "step": 1890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.483928981403535e-05,
+      "loss": 3.4155,
+      "step": 1900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.48381418841356e-05,
+      "loss": 3.4902,
+      "step": 1910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.483699395423585e-05,
+      "loss": 3.5463,
+      "step": 1920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.48358460243361e-05,
+      "loss": 3.5145,
+      "step": 1930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.483469809443637e-05,
+      "loss": 3.5093,
+      "step": 1940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.483355016453662e-05,
+      "loss": 3.5678,
+      "step": 1950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.483240223463687e-05,
+      "loss": 3.6013,
+      "step": 1960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.483125430473712e-05,
+      "loss": 3.5172,
+      "step": 1970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.483010637483736e-05,
+      "loss": 3.3076,
+      "step": 1980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.482895844493763e-05,
+      "loss": 3.3164,
+      "step": 1990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.482781051503788e-05,
+      "loss": 3.3715,
+      "step": 2000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.482666258513813e-05,
+      "loss": 3.1562,
+      "step": 2010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.482551465523838e-05,
+      "loss": 3.4478,
+      "step": 2020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.482436672533864e-05,
+      "loss": 3.4829,
+      "step": 2030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.482321879543888e-05,
+      "loss": 3.2744,
+      "step": 2040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.482207086553914e-05,
+      "loss": 3.3584,
+      "step": 2050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.482092293563939e-05,
+      "loss": 3.3837,
+      "step": 2060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.481977500573964e-05,
+      "loss": 3.3173,
+      "step": 2070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.481862707583989e-05,
+      "loss": 3.2421,
+      "step": 2080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.481747914594015e-05,
+      "loss": 3.2653,
+      "step": 2090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.48163312160404e-05,
+      "loss": 3.4677,
+      "step": 2100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.481518328614065e-05,
+      "loss": 3.2253,
+      "step": 2110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.48140353562409e-05,
+      "loss": 3.2602,
+      "step": 2120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.481288742634115e-05,
+      "loss": 3.161,
+      "step": 2130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.481173949644142e-05,
+      "loss": 3.2883,
+      "step": 2140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.481059156654167e-05,
+      "loss": 3.3627,
+      "step": 2150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.480944363664192e-05,
+      "loss": 3.4091,
+      "step": 2160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.480829570674217e-05,
+      "loss": 3.2481,
+      "step": 2170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.480714777684243e-05,
+      "loss": 3.2871,
+      "step": 2180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.480599984694267e-05,
+      "loss": 3.5556,
+      "step": 2190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.480485191704293e-05,
+      "loss": 3.264,
+      "step": 2200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.480370398714318e-05,
+      "loss": 3.255,
+      "step": 2210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.480255605724343e-05,
+      "loss": 3.1735,
+      "step": 2220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.480140812734368e-05,
+      "loss": 3.1607,
+      "step": 2230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.480026019744394e-05,
+      "loss": 3.2667,
+      "step": 2240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.479911226754419e-05,
+      "loss": 3.1309,
+      "step": 2250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.479796433764444e-05,
+      "loss": 3.3623,
+      "step": 2260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.479681640774469e-05,
+      "loss": 3.0975,
+      "step": 2270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.479566847784494e-05,
+      "loss": 3.1824,
+      "step": 2280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.47945205479452e-05,
+      "loss": 3.2809,
+      "step": 2290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.479337261804546e-05,
+      "loss": 3.3306,
+      "step": 2300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.47922246881457e-05,
+      "loss": 3.2994,
+      "step": 2310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.479107675824595e-05,
+      "loss": 3.092,
+      "step": 2320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.478992882834622e-05,
+      "loss": 3.3619,
+      "step": 2330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.478878089844645e-05,
+      "loss": 3.1708,
+      "step": 2340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.478763296854672e-05,
+      "loss": 3.0974,
+      "step": 2350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.478648503864697e-05,
+      "loss": 3.2471,
+      "step": 2360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.478533710874722e-05,
+      "loss": 3.1651,
+      "step": 2370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.478418917884747e-05,
+      "loss": 3.1778,
+      "step": 2380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.478304124894773e-05,
+      "loss": 3.2344,
+      "step": 2390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.478189331904798e-05,
+      "loss": 3.2394,
+      "step": 2400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.478074538914823e-05,
+      "loss": 3.1922,
+      "step": 2410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.477959745924848e-05,
+      "loss": 3.0635,
+      "step": 2420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.477844952934873e-05,
+      "loss": 3.2439,
+      "step": 2430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.4777301599449e-05,
+      "loss": 3.2397,
+      "step": 2440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.477615366954924e-05,
+      "loss": 3.1739,
+      "step": 2450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.47750057396495e-05,
+      "loss": 3.0676,
+      "step": 2460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.477385780974974e-05,
+      "loss": 3.1125,
+      "step": 2470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.477270987985e-05,
+      "loss": 3.0561,
+      "step": 2480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.477156194995024e-05,
+      "loss": 3.0752,
+      "step": 2490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.47704140200505e-05,
+      "loss": 3.246,
+      "step": 2500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.476926609015076e-05,
+      "loss": 2.9042,
+      "step": 2510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.4768118160251e-05,
+      "loss": 3.0173,
+      "step": 2520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.476697023035126e-05,
+      "loss": 3.0124,
+      "step": 2530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.476582230045152e-05,
+      "loss": 3.0321,
+      "step": 2540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.476467437055177e-05,
+      "loss": 2.9113,
+      "step": 2550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.476352644065202e-05,
+      "loss": 3.1199,
+      "step": 2560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.476237851075227e-05,
+      "loss": 2.9343,
+      "step": 2570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.476123058085252e-05,
+      "loss": 3.1072,
+      "step": 2580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.476008265095278e-05,
+      "loss": 2.9417,
+      "step": 2590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.475893472105302e-05,
+      "loss": 3.2023,
+      "step": 2600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.475778679115328e-05,
+      "loss": 2.9171,
+      "step": 2610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.475663886125353e-05,
+      "loss": 3.0227,
+      "step": 2620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.475549093135378e-05,
+      "loss": 2.9307,
+      "step": 2630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.475434300145403e-05,
+      "loss": 2.9149,
+      "step": 2640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.47531950715543e-05,
+      "loss": 2.8354,
+      "step": 2650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.475204714165454e-05,
+      "loss": 3.0739,
+      "step": 2660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.47508992117548e-05,
+      "loss": 2.9342,
+      "step": 2670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.474975128185504e-05,
+      "loss": 3.0533,
+      "step": 2680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.47486033519553e-05,
+      "loss": 2.9246,
+      "step": 2690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.474745542205556e-05,
+      "loss": 3.0741,
+      "step": 2700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.474630749215581e-05,
+      "loss": 2.9495,
+      "step": 2710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.474515956225606e-05,
+      "loss": 2.9233,
+      "step": 2720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.474401163235631e-05,
+      "loss": 2.9415,
+      "step": 2730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.474286370245657e-05,
+      "loss": 2.8884,
+      "step": 2740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.474171577255681e-05,
+      "loss": 2.9905,
+      "step": 2750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.474056784265707e-05,
+      "loss": 2.973,
+      "step": 2760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.473941991275732e-05,
+      "loss": 2.9517,
+      "step": 2770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.473827198285757e-05,
+      "loss": 2.9529,
+      "step": 2780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.473712405295782e-05,
+      "loss": 2.8623,
+      "step": 2790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.473597612305808e-05,
+      "loss": 2.8216,
+      "step": 2800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.473482819315833e-05,
+      "loss": 3.0548,
+      "step": 2810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.473368026325858e-05,
+      "loss": 2.9289,
+      "step": 2820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.473253233335883e-05,
+      "loss": 2.953,
+      "step": 2830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.473138440345908e-05,
+      "loss": 2.9979,
+      "step": 2840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.473023647355935e-05,
+      "loss": 3.0157,
+      "step": 2850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.47290885436596e-05,
+      "loss": 2.8282,
+      "step": 2860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.472794061375985e-05,
+      "loss": 2.7961,
+      "step": 2870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.47267926838601e-05,
+      "loss": 2.8891,
+      "step": 2880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.472564475396036e-05,
+      "loss": 2.9442,
+      "step": 2890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.47244968240606e-05,
+      "loss": 2.6494,
+      "step": 2900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.472334889416086e-05,
+      "loss": 2.8148,
+      "step": 2910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.472220096426111e-05,
+      "loss": 2.9483,
+      "step": 2920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.472105303436136e-05,
+      "loss": 2.8929,
+      "step": 2930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.471990510446161e-05,
+      "loss": 2.9464,
+      "step": 2940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.471875717456187e-05,
+      "loss": 2.9213,
+      "step": 2950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.471760924466212e-05,
+      "loss": 3.0962,
+      "step": 2960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.471646131476237e-05,
+      "loss": 3.0758,
+      "step": 2970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.471531338486262e-05,
+      "loss": 2.7904,
+      "step": 2980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.471416545496287e-05,
+      "loss": 3.052,
+      "step": 2990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.471301752506313e-05,
+      "loss": 3.0937,
+      "step": 3000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.471186959516338e-05,
+      "loss": 2.9534,
+      "step": 3010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.471072166526363e-05,
+      "loss": 3.0523,
+      "step": 3020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.470957373536388e-05,
+      "loss": 2.7331,
+      "step": 3030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.470842580546415e-05,
+      "loss": 2.9624,
+      "step": 3040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.470727787556438e-05,
+      "loss": 2.6809,
+      "step": 3050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.470612994566465e-05,
+      "loss": 2.8155,
+      "step": 3060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.47049820157649e-05,
+      "loss": 2.7888,
+      "step": 3070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.470383408586515e-05,
+      "loss": 3.0145,
+      "step": 3080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.47026861559654e-05,
+      "loss": 2.7297,
+      "step": 3090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.470153822606566e-05,
+      "loss": 2.9185,
+      "step": 3100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.470039029616591e-05,
+      "loss": 2.7611,
+      "step": 3110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.469924236626616e-05,
+      "loss": 2.8027,
+      "step": 3120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.469809443636641e-05,
+      "loss": 2.9982,
+      "step": 3130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.469694650646666e-05,
+      "loss": 2.7913,
+      "step": 3140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.469579857656692e-05,
+      "loss": 2.8863,
+      "step": 3150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.469465064666717e-05,
+      "loss": 2.9006,
+      "step": 3160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.469350271676742e-05,
+      "loss": 2.9719,
+      "step": 3170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.469235478686767e-05,
+      "loss": 2.7888,
+      "step": 3180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.469120685696794e-05,
+      "loss": 2.7408,
+      "step": 3190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.469005892706817e-05,
+      "loss": 2.6998,
+      "step": 3200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.468891099716844e-05,
+      "loss": 3.0263,
+      "step": 3210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.468776306726869e-05,
+      "loss": 2.6031,
+      "step": 3220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.468661513736894e-05,
+      "loss": 2.8323,
+      "step": 3230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.468546720746919e-05,
+      "loss": 2.8291,
+      "step": 3240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.468431927756945e-05,
+      "loss": 2.6886,
+      "step": 3250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.46831713476697e-05,
+      "loss": 2.9708,
+      "step": 3260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.468202341776995e-05,
+      "loss": 2.8205,
+      "step": 3270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.46808754878702e-05,
+      "loss": 2.6771,
+      "step": 3280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.467972755797045e-05,
+      "loss": 2.7419,
+      "step": 3290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.467857962807071e-05,
+      "loss": 2.8625,
+      "step": 3300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.467743169817096e-05,
+      "loss": 2.7958,
+      "step": 3310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.467628376827121e-05,
+      "loss": 2.7471,
+      "step": 3320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.467513583837146e-05,
+      "loss": 2.9543,
+      "step": 3330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.467398790847172e-05,
+      "loss": 2.7834,
+      "step": 3340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.467283997857196e-05,
+      "loss": 2.7949,
+      "step": 3350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.467169204867222e-05,
+      "loss": 2.7033,
+      "step": 3360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.467054411877247e-05,
+      "loss": 2.8694,
+      "step": 3370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.466939618887272e-05,
+      "loss": 2.9514,
+      "step": 3380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.466824825897297e-05,
+      "loss": 2.7932,
+      "step": 3390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.466710032907324e-05,
+      "loss": 2.5889,
+      "step": 3400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.466595239917349e-05,
+      "loss": 2.6917,
+      "step": 3410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.466480446927374e-05,
+      "loss": 2.8022,
+      "step": 3420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.466365653937399e-05,
+      "loss": 2.6815,
+      "step": 3430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.466250860947424e-05,
+      "loss": 2.9107,
+      "step": 3440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.46613606795745e-05,
+      "loss": 2.8105,
+      "step": 3450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.466021274967475e-05,
+      "loss": 2.7439,
+      "step": 3460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.4659064819775e-05,
+      "loss": 2.9259,
+      "step": 3470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.465791688987525e-05,
+      "loss": 2.6213,
+      "step": 3480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.465676895997551e-05,
+      "loss": 2.8192,
+      "step": 3490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.465562103007575e-05,
+      "loss": 2.6801,
+      "step": 3500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.465447310017601e-05,
+      "loss": 2.6996,
+      "step": 3510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.465332517027626e-05,
+      "loss": 2.7099,
+      "step": 3520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.465217724037651e-05,
+      "loss": 2.9161,
+      "step": 3530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.465102931047676e-05,
+      "loss": 2.7603,
+      "step": 3540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.464988138057703e-05,
+      "loss": 2.7046,
+      "step": 3550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.464873345067728e-05,
+      "loss": 2.8045,
+      "step": 3560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.464758552077753e-05,
+      "loss": 2.6699,
+      "step": 3570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.464643759087778e-05,
+      "loss": 2.7366,
+      "step": 3580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.464528966097803e-05,
+      "loss": 2.7144,
+      "step": 3590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.464414173107829e-05,
+      "loss": 2.7739,
+      "step": 3600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.464299380117854e-05,
+      "loss": 2.7746,
+      "step": 3610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.464184587127879e-05,
+      "loss": 2.6927,
+      "step": 3620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.464069794137904e-05,
+      "loss": 2.7681,
+      "step": 3630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.46395500114793e-05,
+      "loss": 2.6716,
+      "step": 3640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.463840208157954e-05,
+      "loss": 2.6923,
+      "step": 3650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.46372541516798e-05,
+      "loss": 2.8407,
+      "step": 3660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.463610622178005e-05,
+      "loss": 2.6866,
+      "step": 3670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.46349582918803e-05,
+      "loss": 2.7622,
+      "step": 3680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.463381036198055e-05,
+      "loss": 2.6729,
+      "step": 3690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.463266243208081e-05,
+      "loss": 2.8791,
+      "step": 3700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.463151450218106e-05,
+      "loss": 2.6846,
+      "step": 3710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.463036657228131e-05,
+      "loss": 2.6969,
+      "step": 3720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.462921864238156e-05,
+      "loss": 2.8313,
+      "step": 3730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.462807071248181e-05,
+      "loss": 2.5576,
+      "step": 3740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.462692278258208e-05,
+      "loss": 2.7452,
+      "step": 3750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.462577485268233e-05,
+      "loss": 2.6983,
+      "step": 3760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.462462692278258e-05,
+      "loss": 2.8379,
+      "step": 3770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.462347899288283e-05,
+      "loss": 2.6709,
+      "step": 3780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.462233106298309e-05,
+      "loss": 2.6945,
+      "step": 3790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.462118313308333e-05,
+      "loss": 2.546,
+      "step": 3800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.462003520318359e-05,
+      "loss": 2.7138,
+      "step": 3810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.461888727328384e-05,
+      "loss": 2.8632,
+      "step": 3820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.461773934338409e-05,
+      "loss": 2.7705,
+      "step": 3830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.461659141348434e-05,
+      "loss": 2.8571,
+      "step": 3840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.46154434835846e-05,
+      "loss": 2.7711,
+      "step": 3850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.461429555368485e-05,
+      "loss": 2.7545,
+      "step": 3860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.46131476237851e-05,
+      "loss": 2.779,
+      "step": 3870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.461199969388535e-05,
+      "loss": 2.6158,
+      "step": 3880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.46108517639856e-05,
+      "loss": 2.7686,
+      "step": 3890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.460970383408587e-05,
+      "loss": 2.6452,
+      "step": 3900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.460855590418612e-05,
+      "loss": 2.8311,
+      "step": 3910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.460740797428637e-05,
+      "loss": 2.7475,
+      "step": 3920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.460626004438662e-05,
+      "loss": 2.6455,
+      "step": 3930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.460511211448686e-05,
+      "loss": 2.6214,
+      "step": 3940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.460396418458711e-05,
+      "loss": 2.6484,
+      "step": 3950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.460281625468738e-05,
+      "loss": 2.6851,
+      "step": 3960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.460166832478763e-05,
+      "loss": 2.6477,
+      "step": 3970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.460052039488788e-05,
+      "loss": 2.6404,
+      "step": 3980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.459937246498813e-05,
+      "loss": 2.5576,
+      "step": 3990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.459822453508838e-05,
+      "loss": 2.5932,
+      "step": 4000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.459707660518864e-05,
+      "loss": 2.7058,
+      "step": 4010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.459592867528889e-05,
+      "loss": 2.7952,
+      "step": 4020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.459478074538914e-05,
+      "loss": 2.7664,
+      "step": 4030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.459363281548939e-05,
+      "loss": 2.6005,
+      "step": 4040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.459248488558965e-05,
+      "loss": 2.6448,
+      "step": 4050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.459133695568989e-05,
+      "loss": 2.834,
+      "step": 4060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.459018902579015e-05,
+      "loss": 2.7995,
+      "step": 4070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.45890410958904e-05,
+      "loss": 2.7187,
+      "step": 4080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.458789316599065e-05,
+      "loss": 2.6999,
+      "step": 4090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.45867452360909e-05,
+      "loss": 2.7226,
+      "step": 4100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.458559730619117e-05,
+      "loss": 2.6152,
+      "step": 4110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.458444937629142e-05,
+      "loss": 2.7185,
+      "step": 4120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.458330144639167e-05,
+      "loss": 2.6778,
+      "step": 4130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.458215351649192e-05,
+      "loss": 2.6688,
+      "step": 4140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.458100558659217e-05,
+      "loss": 2.635,
+      "step": 4150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.457985765669243e-05,
+      "loss": 2.78,
+      "step": 4160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.457870972679268e-05,
+      "loss": 2.71,
+      "step": 4170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.457756179689293e-05,
+      "loss": 2.6013,
+      "step": 4180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.457641386699318e-05,
+      "loss": 2.788,
+      "step": 4190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.457526593709344e-05,
+      "loss": 2.8168,
+      "step": 4200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.457411800719368e-05,
+      "loss": 2.5982,
+      "step": 4210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.457297007729394e-05,
+      "loss": 2.4876,
+      "step": 4220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.457182214739419e-05,
+      "loss": 2.7644,
+      "step": 4230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.457067421749444e-05,
+      "loss": 2.7613,
+      "step": 4240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.456952628759469e-05,
+      "loss": 2.8393,
+      "step": 4250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.456837835769496e-05,
+      "loss": 2.8137,
+      "step": 4260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.45672304277952e-05,
+      "loss": 2.6791,
+      "step": 4270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.456608249789545e-05,
+      "loss": 2.4652,
+      "step": 4280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.45649345679957e-05,
+      "loss": 2.7958,
+      "step": 4290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.456378663809595e-05,
+      "loss": 2.6984,
+      "step": 4300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.456263870819622e-05,
+      "loss": 2.7622,
+      "step": 4310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.456149077829647e-05,
+      "loss": 2.7571,
+      "step": 4320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.456034284839672e-05,
+      "loss": 2.6325,
+      "step": 4330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.455919491849697e-05,
+      "loss": 2.5599,
+      "step": 4340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.455804698859723e-05,
+      "loss": 2.6531,
+      "step": 4350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.455689905869747e-05,
+      "loss": 2.5004,
+      "step": 4360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.455575112879773e-05,
+      "loss": 2.6754,
+      "step": 4370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.455460319889798e-05,
+      "loss": 2.56,
+      "step": 4380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.455345526899823e-05,
+      "loss": 2.8167,
+      "step": 4390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.455230733909848e-05,
+      "loss": 2.7,
+      "step": 4400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.455115940919874e-05,
+      "loss": 2.6957,
+      "step": 4410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.4550011479299e-05,
+      "loss": 2.7933,
+      "step": 4420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.454886354939924e-05,
+      "loss": 2.608,
+      "step": 4430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.454771561949949e-05,
+      "loss": 2.6778,
+      "step": 4440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.454656768959974e-05,
+      "loss": 2.8608,
+      "step": 4450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.45454197597e-05,
+      "loss": 2.6279,
+      "step": 4460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.454427182980026e-05,
+      "loss": 2.6663,
+      "step": 4470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.45431238999005e-05,
+      "loss": 2.7537,
+      "step": 4480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.454197597000076e-05,
+      "loss": 2.7609,
+      "step": 4490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.454082804010102e-05,
+      "loss": 2.6625,
+      "step": 4500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.453968011020126e-05,
+      "loss": 2.7404,
+      "step": 4510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.453853218030152e-05,
+      "loss": 2.6341,
+      "step": 4520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.453738425040177e-05,
+      "loss": 2.5057,
+      "step": 4530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.453623632050202e-05,
+      "loss": 2.6679,
+      "step": 4540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.453508839060227e-05,
+      "loss": 2.6484,
+      "step": 4550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.453394046070253e-05,
+      "loss": 2.7186,
+      "step": 4560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.453279253080278e-05,
+      "loss": 2.7935,
+      "step": 4570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.453164460090303e-05,
+      "loss": 2.5873,
+      "step": 4580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.453049667100328e-05,
+      "loss": 2.619,
+      "step": 4590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.452934874110353e-05,
+      "loss": 2.7257,
+      "step": 4600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.45282008112038e-05,
+      "loss": 2.6903,
+      "step": 4610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.452705288130404e-05,
+      "loss": 2.7274,
+      "step": 4620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.45259049514043e-05,
+      "loss": 2.6509,
+      "step": 4630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.452475702150454e-05,
+      "loss": 2.5339,
+      "step": 4640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.452360909160481e-05,
+      "loss": 2.6897,
+      "step": 4650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.452246116170504e-05,
+      "loss": 2.7581,
+      "step": 4660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.452131323180531e-05,
+      "loss": 2.6016,
+      "step": 4670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.452016530190556e-05,
+      "loss": 2.5829,
+      "step": 4680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.451901737200581e-05,
+      "loss": 2.6231,
+      "step": 4690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.451786944210606e-05,
+      "loss": 2.7234,
+      "step": 4700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.451672151220632e-05,
+      "loss": 2.5646,
+      "step": 4710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.451557358230657e-05,
+      "loss": 2.6269,
+      "step": 4720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.451442565240682e-05,
+      "loss": 2.5701,
+      "step": 4730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.451327772250707e-05,
+      "loss": 2.6957,
+      "step": 4740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.451212979260732e-05,
+      "loss": 2.6925,
+      "step": 4750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.451098186270758e-05,
+      "loss": 2.7241,
+      "step": 4760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.450983393280783e-05,
+      "loss": 2.5634,
+      "step": 4770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.450868600290808e-05,
+      "loss": 2.4808,
+      "step": 4780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.450753807300833e-05,
+      "loss": 2.7959,
+      "step": 4790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.45063901431086e-05,
+      "loss": 2.8308,
+      "step": 4800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.450524221320883e-05,
+      "loss": 2.5612,
+      "step": 4810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.45040942833091e-05,
+      "loss": 2.588,
+      "step": 4820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.450294635340935e-05,
+      "loss": 2.5292,
+      "step": 4830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.45017984235096e-05,
+      "loss": 2.65,
+      "step": 4840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.450065049360985e-05,
+      "loss": 2.8327,
+      "step": 4850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.449950256371011e-05,
+      "loss": 2.5807,
+      "step": 4860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.449835463381036e-05,
+      "loss": 2.5776,
+      "step": 4870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.449720670391061e-05,
+      "loss": 2.6397,
+      "step": 4880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.449605877401086e-05,
+      "loss": 2.5579,
+      "step": 4890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.449491084411111e-05,
+      "loss": 2.6205,
+      "step": 4900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.449376291421137e-05,
+      "loss": 2.5928,
+      "step": 4910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.449261498431162e-05,
+      "loss": 2.6368,
+      "step": 4920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.449146705441187e-05,
+      "loss": 2.6542,
+      "step": 4930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.449031912451212e-05,
+      "loss": 2.5073,
+      "step": 4940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.448917119461238e-05,
+      "loss": 2.634,
+      "step": 4950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.448802326471262e-05,
+      "loss": 2.7398,
+      "step": 4960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.448687533481288e-05,
+      "loss": 2.582,
+      "step": 4970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.448572740491313e-05,
+      "loss": 2.5201,
+      "step": 4980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.448457947501338e-05,
+      "loss": 2.6258,
+      "step": 4990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.448343154511363e-05,
+      "loss": 2.7648,
+      "step": 5000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.44822836152139e-05,
+      "loss": 2.6088,
+      "step": 5010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.448113568531415e-05,
+      "loss": 2.4955,
+      "step": 5020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.44799877554144e-05,
+      "loss": 2.5972,
+      "step": 5030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.447883982551465e-05,
+      "loss": 2.5607,
+      "step": 5040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.44776918956149e-05,
+      "loss": 2.908,
+      "step": 5050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.447654396571516e-05,
+      "loss": 2.6634,
+      "step": 5060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.447539603581541e-05,
+      "loss": 2.6533,
+      "step": 5070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.447424810591566e-05,
+      "loss": 2.5117,
+      "step": 5080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.447310017601591e-05,
+      "loss": 2.6334,
+      "step": 5090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.447195224611617e-05,
+      "loss": 2.5308,
+      "step": 5100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.447080431621641e-05,
+      "loss": 2.4802,
+      "step": 5110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.446965638631667e-05,
+      "loss": 2.5936,
+      "step": 5120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.446850845641692e-05,
+      "loss": 2.5952,
+      "step": 5130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.446736052651717e-05,
+      "loss": 2.5248,
+      "step": 5140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.446621259661742e-05,
+      "loss": 2.6126,
+      "step": 5150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.446506466671769e-05,
+      "loss": 2.5338,
+      "step": 5160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.446391673681794e-05,
+      "loss": 2.6727,
+      "step": 5170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.446276880691819e-05,
+      "loss": 2.6769,
+      "step": 5180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.446162087701844e-05,
+      "loss": 2.3136,
+      "step": 5190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.446047294711869e-05,
+      "loss": 2.4571,
+      "step": 5200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.445932501721895e-05,
+      "loss": 2.6469,
+      "step": 5210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.44581770873192e-05,
+      "loss": 2.8288,
+      "step": 5220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.445702915741945e-05,
+      "loss": 2.4396,
+      "step": 5230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.44558812275197e-05,
+      "loss": 2.6382,
+      "step": 5240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.445473329761996e-05,
+      "loss": 2.7628,
+      "step": 5250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.44535853677202e-05,
+      "loss": 2.6046,
+      "step": 5260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.445243743782046e-05,
+      "loss": 2.6822,
+      "step": 5270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.445128950792071e-05,
+      "loss": 2.567,
+      "step": 5280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.445014157802096e-05,
+      "loss": 2.6242,
+      "step": 5290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.444899364812121e-05,
+      "loss": 2.5819,
+      "step": 5300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.444784571822146e-05,
+      "loss": 2.6644,
+      "step": 5310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.444669778832172e-05,
+      "loss": 2.7282,
+      "step": 5320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.444554985842197e-05,
+      "loss": 2.673,
+      "step": 5330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.444440192852222e-05,
+      "loss": 2.7259,
+      "step": 5340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.444325399862247e-05,
+      "loss": 2.6308,
+      "step": 5350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.444210606872274e-05,
+      "loss": 2.5606,
+      "step": 5360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.444095813882297e-05,
+      "loss": 2.6316,
+      "step": 5370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.443981020892324e-05,
+      "loss": 2.4193,
+      "step": 5380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.443866227902349e-05,
+      "loss": 2.5797,
+      "step": 5390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.443751434912374e-05,
+      "loss": 2.4512,
+      "step": 5400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.443636641922399e-05,
+      "loss": 2.5457,
+      "step": 5410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.443521848932425e-05,
+      "loss": 2.5214,
+      "step": 5420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.44340705594245e-05,
+      "loss": 2.544,
+      "step": 5430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.443292262952475e-05,
+      "loss": 2.5645,
+      "step": 5440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.4431774699625e-05,
+      "loss": 2.6642,
+      "step": 5450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.443062676972525e-05,
+      "loss": 2.5516,
+      "step": 5460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.442947883982551e-05,
+      "loss": 2.4333,
+      "step": 5470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.442833090992576e-05,
+      "loss": 2.5505,
+      "step": 5480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.442718298002601e-05,
+      "loss": 2.5689,
+      "step": 5490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.442603505012626e-05,
+      "loss": 2.5314,
+      "step": 5500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.442488712022653e-05,
+      "loss": 2.4611,
+      "step": 5510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.442373919032676e-05,
+      "loss": 2.729,
+      "step": 5520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.442259126042703e-05,
+      "loss": 2.4271,
+      "step": 5530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.442144333052728e-05,
+      "loss": 2.4162,
+      "step": 5540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.442029540062753e-05,
+      "loss": 2.5993,
+      "step": 5550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.441914747072777e-05,
+      "loss": 2.5678,
+      "step": 5560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.441799954082804e-05,
+      "loss": 2.5595,
+      "step": 5570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.441685161092829e-05,
+      "loss": 2.6571,
+      "step": 5580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.441570368102854e-05,
+      "loss": 2.4359,
+      "step": 5590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.441455575112879e-05,
+      "loss": 2.3997,
+      "step": 5600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.441340782122904e-05,
+      "loss": 2.4481,
+      "step": 5610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.44122598913293e-05,
+      "loss": 2.6263,
+      "step": 5620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.441111196142955e-05,
+      "loss": 2.5327,
+      "step": 5630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.44099640315298e-05,
+      "loss": 2.5171,
+      "step": 5640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.440881610163005e-05,
+      "loss": 2.5972,
+      "step": 5650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.440766817173031e-05,
+      "loss": 2.5614,
+      "step": 5660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.440652024183055e-05,
+      "loss": 2.4796,
+      "step": 5670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.440537231193081e-05,
+      "loss": 2.618,
+      "step": 5680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.440422438203106e-05,
+      "loss": 2.4076,
+      "step": 5690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.440307645213131e-05,
+      "loss": 2.3988,
+      "step": 5700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.440192852223156e-05,
+      "loss": 2.5978,
+      "step": 5710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.440078059233183e-05,
+      "loss": 2.5632,
+      "step": 5720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.439963266243208e-05,
+      "loss": 2.6037,
+      "step": 5730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.439848473253233e-05,
+      "loss": 2.6822,
+      "step": 5740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.439733680263258e-05,
+      "loss": 2.6126,
+      "step": 5750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.439618887273283e-05,
+      "loss": 2.5006,
+      "step": 5760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.439504094283309e-05,
+      "loss": 2.5681,
+      "step": 5770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.439389301293334e-05,
+      "loss": 2.649,
+      "step": 5780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.439274508303359e-05,
+      "loss": 2.5249,
+      "step": 5790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.439159715313384e-05,
+      "loss": 2.5555,
+      "step": 5800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.43904492232341e-05,
+      "loss": 2.5555,
+      "step": 5810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.438930129333434e-05,
+      "loss": 2.6718,
+      "step": 5820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.43881533634346e-05,
+      "loss": 2.563,
+      "step": 5830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.438700543353485e-05,
+      "loss": 2.749,
+      "step": 5840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.43858575036351e-05,
+      "loss": 2.6454,
+      "step": 5850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.438470957373535e-05,
+      "loss": 2.6021,
+      "step": 5860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.438356164383562e-05,
+      "loss": 2.5549,
+      "step": 5870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.438241371393587e-05,
+      "loss": 2.5174,
+      "step": 5880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.438126578403612e-05,
+      "loss": 2.355,
+      "step": 5890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.438011785413636e-05,
+      "loss": 2.6059,
+      "step": 5900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.437896992423661e-05,
+      "loss": 2.4822,
+      "step": 5910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.437782199433688e-05,
+      "loss": 2.5986,
+      "step": 5920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.437667406443713e-05,
+      "loss": 2.5175,
+      "step": 5930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.437552613453738e-05,
+      "loss": 2.5759,
+      "step": 5940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.437437820463763e-05,
+      "loss": 2.5933,
+      "step": 5950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.437323027473789e-05,
+      "loss": 2.415,
+      "step": 5960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.437208234483813e-05,
+      "loss": 2.3885,
+      "step": 5970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.437093441493839e-05,
+      "loss": 2.6101,
+      "step": 5980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.436978648503864e-05,
+      "loss": 2.6701,
+      "step": 5990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.436863855513889e-05,
+      "loss": 2.6997,
+      "step": 6000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.436749062523914e-05,
+      "loss": 2.4212,
+      "step": 6010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.43663426953394e-05,
+      "loss": 2.2829,
+      "step": 6020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.436519476543965e-05,
+      "loss": 2.5493,
+      "step": 6030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.43640468355399e-05,
+      "loss": 2.4852,
+      "step": 6040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.436289890564015e-05,
+      "loss": 2.5323,
+      "step": 6050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.43617509757404e-05,
+      "loss": 2.5525,
+      "step": 6060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.436060304584067e-05,
+      "loss": 2.5107,
+      "step": 6070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.435945511594092e-05,
+      "loss": 2.6317,
+      "step": 6080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.435830718604117e-05,
+      "loss": 2.6064,
+      "step": 6090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.435715925614142e-05,
+      "loss": 2.5429,
+      "step": 6100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.435601132624168e-05,
+      "loss": 2.5296,
+      "step": 6110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.435486339634192e-05,
+      "loss": 2.5833,
+      "step": 6120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.435371546644218e-05,
+      "loss": 2.4318,
+      "step": 6130
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.435256753654243e-05,
+      "loss": 2.5217,
+      "step": 6140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.435141960664268e-05,
+      "loss": 2.5903,
+      "step": 6150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.435027167674293e-05,
+      "loss": 2.7374,
+      "step": 6160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.434912374684319e-05,
+      "loss": 2.463,
+      "step": 6170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.434797581694344e-05,
+      "loss": 2.5325,
+      "step": 6180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.434682788704369e-05,
+      "loss": 2.4613,
+      "step": 6190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.434567995714394e-05,
+      "loss": 2.3402,
+      "step": 6200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.434453202724419e-05,
+      "loss": 2.6366,
+      "step": 6210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.434338409734446e-05,
+      "loss": 2.4745,
+      "step": 6220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.43422361674447e-05,
+      "loss": 2.5084,
+      "step": 6230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.434108823754495e-05,
+      "loss": 2.4891,
+      "step": 6240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.43399403076452e-05,
+      "loss": 2.4879,
+      "step": 6250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.433879237774547e-05,
+      "loss": 2.4543,
+      "step": 6260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.43376444478457e-05,
+      "loss": 2.6375,
+      "step": 6270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.433649651794597e-05,
+      "loss": 2.5657,
+      "step": 6280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.433534858804622e-05,
+      "loss": 2.496,
+      "step": 6290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.433420065814647e-05,
+      "loss": 2.5125,
+      "step": 6300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.433305272824672e-05,
+      "loss": 2.6857,
+      "step": 6310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.433190479834698e-05,
+      "loss": 2.607,
+      "step": 6320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.433075686844723e-05,
+      "loss": 2.5522,
+      "step": 6330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.432960893854748e-05,
+      "loss": 2.5241,
+      "step": 6340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.432846100864773e-05,
+      "loss": 2.5488,
+      "step": 6350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.432731307874798e-05,
+      "loss": 2.5599,
+      "step": 6360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.432616514884824e-05,
+      "loss": 2.5845,
+      "step": 6370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.43250172189485e-05,
+      "loss": 2.581,
+      "step": 6380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.432386928904874e-05,
+      "loss": 2.4974,
+      "step": 6390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.432272135914899e-05,
+      "loss": 2.4294,
+      "step": 6400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.432157342924926e-05,
+      "loss": 2.4891,
+      "step": 6410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.432042549934949e-05,
+      "loss": 2.2961,
+      "step": 6420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.431927756944976e-05,
+      "loss": 2.5419,
+      "step": 6430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.431812963955e-05,
+      "loss": 2.5116,
+      "step": 6440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.431698170965026e-05,
+      "loss": 2.5693,
+      "step": 6450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.43158337797505e-05,
+      "loss": 2.562,
+      "step": 6460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.431468584985077e-05,
+      "loss": 2.4525,
+      "step": 6470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.431353791995102e-05,
+      "loss": 2.5194,
+      "step": 6480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.431238999005127e-05,
+      "loss": 2.47,
+      "step": 6490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.431124206015152e-05,
+      "loss": 2.4286,
+      "step": 6500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.431009413025177e-05,
+      "loss": 2.5252,
+      "step": 6510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.430894620035203e-05,
+      "loss": 2.592,
+      "step": 6520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.430779827045228e-05,
+      "loss": 2.5549,
+      "step": 6530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.430665034055253e-05,
+      "loss": 2.4486,
+      "step": 6540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.430550241065278e-05,
+      "loss": 2.3146,
+      "step": 6550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.430435448075304e-05,
+      "loss": 2.4864,
+      "step": 6560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.430320655085328e-05,
+      "loss": 2.6198,
+      "step": 6570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.430205862095354e-05,
+      "loss": 2.7657,
+      "step": 6580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.43009106910538e-05,
+      "loss": 2.5005,
+      "step": 6590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.429976276115404e-05,
+      "loss": 2.5396,
+      "step": 6600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.42986148312543e-05,
+      "loss": 2.5141,
+      "step": 6610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.429746690135456e-05,
+      "loss": 2.6054,
+      "step": 6620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.429631897145481e-05,
+      "loss": 2.3687,
+      "step": 6630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.429517104155506e-05,
+      "loss": 2.5646,
+      "step": 6640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.429402311165531e-05,
+      "loss": 2.3409,
+      "step": 6650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.429287518175556e-05,
+      "loss": 2.4455,
+      "step": 6660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.429172725185582e-05,
+      "loss": 2.4494,
+      "step": 6670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.429057932195606e-05,
+      "loss": 2.4822,
+      "step": 6680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.428943139205632e-05,
+      "loss": 2.4744,
+      "step": 6690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.428828346215657e-05,
+      "loss": 2.6134,
+      "step": 6700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.428713553225682e-05,
+      "loss": 2.5182,
+      "step": 6710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.428598760235707e-05,
+      "loss": 2.6903,
+      "step": 6720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.428483967245733e-05,
+      "loss": 2.4903,
+      "step": 6730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.428369174255758e-05,
+      "loss": 2.5823,
+      "step": 6740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.428254381265783e-05,
+      "loss": 2.4966,
+      "step": 6750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.428139588275808e-05,
+      "loss": 2.4935,
+      "step": 6760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.428024795285833e-05,
+      "loss": 2.5241,
+      "step": 6770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.42791000229586e-05,
+      "loss": 2.4723,
+      "step": 6780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.427795209305885e-05,
+      "loss": 2.7274,
+      "step": 6790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.42768041631591e-05,
+      "loss": 2.4842,
+      "step": 6800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.427565623325935e-05,
+      "loss": 2.3448,
+      "step": 6810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.427450830335961e-05,
+      "loss": 2.6728,
+      "step": 6820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.427336037345985e-05,
+      "loss": 2.3298,
+      "step": 6830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.427221244356011e-05,
+      "loss": 2.4578,
+      "step": 6840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.427106451366036e-05,
+      "loss": 2.6131,
+      "step": 6850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.426991658376061e-05,
+      "loss": 2.4032,
+      "step": 6860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.426876865386086e-05,
+      "loss": 2.6014,
+      "step": 6870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.426762072396112e-05,
+      "loss": 2.4355,
+      "step": 6880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.426647279406137e-05,
+      "loss": 2.4493,
+      "step": 6890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.426532486416162e-05,
+      "loss": 2.4602,
+      "step": 6900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.426417693426187e-05,
+      "loss": 2.4947,
+      "step": 6910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.426302900436212e-05,
+      "loss": 2.3583,
+      "step": 6920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.426188107446238e-05,
+      "loss": 2.5159,
+      "step": 6930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.426073314456263e-05,
+      "loss": 2.5174,
+      "step": 6940
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.425958521466288e-05,
+      "loss": 2.576,
+      "step": 6950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.425843728476313e-05,
+      "loss": 2.432,
+      "step": 6960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.42572893548634e-05,
+      "loss": 2.4902,
+      "step": 6970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.425614142496363e-05,
+      "loss": 2.2624,
+      "step": 6980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.42549934950639e-05,
+      "loss": 2.4385,
+      "step": 6990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.425384556516415e-05,
+      "loss": 2.5013,
+      "step": 7000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.42526976352644e-05,
+      "loss": 2.5408,
+      "step": 7010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.425154970536465e-05,
+      "loss": 2.474,
+      "step": 7020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.425040177546491e-05,
+      "loss": 2.4858,
+      "step": 7030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.424925384556516e-05,
+      "loss": 2.2509,
+      "step": 7040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.424810591566541e-05,
+      "loss": 2.359,
+      "step": 7050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.424695798576566e-05,
+      "loss": 2.5426,
+      "step": 7060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.424581005586591e-05,
+      "loss": 2.4264,
+      "step": 7070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.424466212596617e-05,
+      "loss": 2.5518,
+      "step": 7080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.424351419606642e-05,
+      "loss": 2.5255,
+      "step": 7090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.424236626616667e-05,
+      "loss": 2.4653,
+      "step": 7100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.424121833626692e-05,
+      "loss": 2.4592,
+      "step": 7110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.424007040636719e-05,
+      "loss": 2.4017,
+      "step": 7120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.423892247646742e-05,
+      "loss": 2.5006,
+      "step": 7130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.423777454656769e-05,
+      "loss": 2.3145,
+      "step": 7140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.423662661666794e-05,
+      "loss": 2.5234,
+      "step": 7150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.423547868676819e-05,
+      "loss": 2.4864,
+      "step": 7160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.423433075686844e-05,
+      "loss": 2.4368,
+      "step": 7170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.42331828269687e-05,
+      "loss": 2.4981,
+      "step": 7180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.423203489706895e-05,
+      "loss": 2.6454,
+      "step": 7190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.42308869671692e-05,
+      "loss": 2.375,
+      "step": 7200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.422973903726945e-05,
+      "loss": 2.4225,
+      "step": 7210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.42285911073697e-05,
+      "loss": 2.4261,
+      "step": 7220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.422744317746996e-05,
+      "loss": 2.3922,
+      "step": 7230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.422629524757021e-05,
+      "loss": 2.5142,
+      "step": 7240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.422514731767046e-05,
+      "loss": 2.3692,
+      "step": 7250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.422399938777071e-05,
+      "loss": 2.6403,
+      "step": 7260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.422285145787097e-05,
+      "loss": 2.2905,
+      "step": 7270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.422170352797121e-05,
+      "loss": 2.4056,
+      "step": 7280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.422055559807147e-05,
+      "loss": 2.3205,
+      "step": 7290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.421940766817172e-05,
+      "loss": 2.3754,
+      "step": 7300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.421825973827197e-05,
+      "loss": 2.5183,
+      "step": 7310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.421711180837222e-05,
+      "loss": 2.4578,
+      "step": 7320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.421596387847249e-05,
+      "loss": 2.5255,
+      "step": 7330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.421481594857274e-05,
+      "loss": 2.4513,
+      "step": 7340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.421366801867299e-05,
+      "loss": 2.249,
+      "step": 7350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.421252008877324e-05,
+      "loss": 2.3235,
+      "step": 7360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.421137215887349e-05,
+      "loss": 2.4313,
+      "step": 7370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.421022422897375e-05,
+      "loss": 2.4903,
+      "step": 7380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.4209076299074e-05,
+      "loss": 2.7509,
+      "step": 7390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.420792836917425e-05,
+      "loss": 2.4988,
+      "step": 7400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.42067804392745e-05,
+      "loss": 2.4295,
+      "step": 7410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.420563250937476e-05,
+      "loss": 2.3173,
+      "step": 7420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.4204484579475e-05,
+      "loss": 2.336,
+      "step": 7430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.420333664957526e-05,
+      "loss": 2.431,
+      "step": 7440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.420218871967551e-05,
+      "loss": 2.4829,
+      "step": 7450
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.420104078977576e-05,
+      "loss": 2.4922,
+      "step": 7460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.419989285987601e-05,
+      "loss": 2.5869,
+      "step": 7470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.419874492997628e-05,
+      "loss": 2.641,
+      "step": 7480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.419759700007653e-05,
+      "loss": 2.4379,
+      "step": 7490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.419644907017678e-05,
+      "loss": 2.7288,
+      "step": 7500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.419530114027703e-05,
+      "loss": 2.4314,
+      "step": 7510
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.419415321037727e-05,
+      "loss": 2.4849,
+      "step": 7520
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.419300528047754e-05,
+      "loss": 2.4184,
+      "step": 7530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.419185735057779e-05,
+      "loss": 2.5132,
+      "step": 7540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.419070942067804e-05,
+      "loss": 2.5325,
+      "step": 7550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.418956149077829e-05,
+      "loss": 2.6415,
+      "step": 7560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.418841356087855e-05,
+      "loss": 2.3421,
+      "step": 7570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.418726563097879e-05,
+      "loss": 2.4999,
+      "step": 7580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.418611770107905e-05,
+      "loss": 2.4517,
+      "step": 7590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.41849697711793e-05,
+      "loss": 2.7374,
+      "step": 7600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.418382184127955e-05,
+      "loss": 2.5988,
+      "step": 7610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.41826739113798e-05,
+      "loss": 2.4698,
+      "step": 7620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.418152598148006e-05,
+      "loss": 2.4015,
+      "step": 7630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.418037805158031e-05,
+      "loss": 2.5677,
+      "step": 7640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.417923012168056e-05,
+      "loss": 2.4629,
+      "step": 7650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.417808219178081e-05,
+      "loss": 2.3641,
+      "step": 7660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.417693426188106e-05,
+      "loss": 2.6262,
+      "step": 7670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.417578633198133e-05,
+      "loss": 2.4152,
+      "step": 7680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.417463840208158e-05,
+      "loss": 2.4886,
+      "step": 7690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.417349047218183e-05,
+      "loss": 2.4493,
+      "step": 7700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.417234254228208e-05,
+      "loss": 2.4002,
+      "step": 7710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.417119461238234e-05,
+      "loss": 2.4884,
+      "step": 7720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.417004668248258e-05,
+      "loss": 2.5483,
+      "step": 7730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.416889875258284e-05,
+      "loss": 2.4399,
+      "step": 7740
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.416775082268309e-05,
+      "loss": 2.5101,
+      "step": 7750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.416660289278334e-05,
+      "loss": 2.4663,
+      "step": 7760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.416545496288359e-05,
+      "loss": 2.4115,
+      "step": 7770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.416430703298385e-05,
+      "loss": 2.4903,
+      "step": 7780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.41631591030841e-05,
+      "loss": 2.4335,
+      "step": 7790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.416201117318435e-05,
+      "loss": 2.4449,
+      "step": 7800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.41608632432846e-05,
+      "loss": 2.4144,
+      "step": 7810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.415971531338485e-05,
+      "loss": 2.5128,
+      "step": 7820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.415856738348512e-05,
+      "loss": 2.4701,
+      "step": 7830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.415741945358537e-05,
+      "loss": 2.5814,
+      "step": 7840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.415627152368562e-05,
+      "loss": 2.4873,
+      "step": 7850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.415512359378586e-05,
+      "loss": 2.408,
+      "step": 7860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.415397566388613e-05,
+      "loss": 2.398,
+      "step": 7870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.415282773398636e-05,
+      "loss": 2.4964,
+      "step": 7880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.415167980408663e-05,
+      "loss": 2.4579,
+      "step": 7890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.415053187418688e-05,
+      "loss": 2.3346,
+      "step": 7900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.414938394428713e-05,
+      "loss": 2.4434,
+      "step": 7910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.414823601438738e-05,
+      "loss": 2.4743,
+      "step": 7920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.414708808448764e-05,
+      "loss": 2.4962,
+      "step": 7930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.414594015458789e-05,
+      "loss": 2.4578,
+      "step": 7940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.414479222468814e-05,
+      "loss": 2.5341,
+      "step": 7950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.414364429478839e-05,
+      "loss": 2.513,
+      "step": 7960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.414249636488864e-05,
+      "loss": 2.3374,
+      "step": 7970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.41413484349889e-05,
+      "loss": 2.3452,
+      "step": 7980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.414020050508915e-05,
+      "loss": 2.5131,
+      "step": 7990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.41390525751894e-05,
+      "loss": 2.381,
+      "step": 8000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.413790464528965e-05,
+      "loss": 2.4937,
+      "step": 8010
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.413675671538992e-05,
+      "loss": 2.3876,
+      "step": 8020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.413560878549015e-05,
+      "loss": 2.4398,
+      "step": 8030
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.413446085559042e-05,
+      "loss": 2.4354,
+      "step": 8040
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.413331292569067e-05,
+      "loss": 2.3806,
+      "step": 8050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.413216499579092e-05,
+      "loss": 2.3433,
+      "step": 8060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.413101706589117e-05,
+      "loss": 2.5044,
+      "step": 8070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.412986913599142e-05,
+      "loss": 2.4692,
+      "step": 8080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.412872120609168e-05,
+      "loss": 2.4451,
+      "step": 8090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.412757327619193e-05,
+      "loss": 2.4131,
+      "step": 8100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.412642534629218e-05,
+      "loss": 2.3799,
+      "step": 8110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.412527741639243e-05,
+      "loss": 2.5191,
+      "step": 8120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.412412948649269e-05,
+      "loss": 2.259,
+      "step": 8130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.412298155659293e-05,
+      "loss": 2.547,
+      "step": 8140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.412183362669319e-05,
+      "loss": 2.517,
+      "step": 8150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.412068569679344e-05,
+      "loss": 2.5557,
+      "step": 8160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.411953776689369e-05,
+      "loss": 2.351,
+      "step": 8170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.411838983699394e-05,
+      "loss": 2.3933,
+      "step": 8180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.41172419070942e-05,
+      "loss": 2.4597,
+      "step": 8190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.411609397719445e-05,
+      "loss": 2.5477,
+      "step": 8200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.41149460472947e-05,
+      "loss": 2.5544,
+      "step": 8210
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.411379811739495e-05,
+      "loss": 2.3719,
+      "step": 8220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.41126501874952e-05,
+      "loss": 2.3567,
+      "step": 8230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.411150225759547e-05,
+      "loss": 2.5424,
+      "step": 8240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.411035432769572e-05,
+      "loss": 2.4958,
+      "step": 8250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.410920639779597e-05,
+      "loss": 2.4586,
+      "step": 8260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.410805846789622e-05,
+      "loss": 2.3664,
+      "step": 8270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.410691053799648e-05,
+      "loss": 2.3118,
+      "step": 8280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.410576260809672e-05,
+      "loss": 2.337,
+      "step": 8290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.410461467819698e-05,
+      "loss": 2.4353,
+      "step": 8300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.410346674829723e-05,
+      "loss": 2.4504,
+      "step": 8310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.410231881839748e-05,
+      "loss": 2.613,
+      "step": 8320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.410117088849773e-05,
+      "loss": 2.3635,
+      "step": 8330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4100022958598e-05,
+      "loss": 2.4348,
+      "step": 8340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.409887502869824e-05,
+      "loss": 2.6183,
+      "step": 8350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.409772709879849e-05,
+      "loss": 2.3298,
+      "step": 8360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.409657916889874e-05,
+      "loss": 2.5216,
+      "step": 8370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.409543123899899e-05,
+      "loss": 2.4639,
+      "step": 8380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.409428330909926e-05,
+      "loss": 2.3936,
+      "step": 8390
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.40931353791995e-05,
+      "loss": 2.5363,
+      "step": 8400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.409198744929976e-05,
+      "loss": 2.7654,
+      "step": 8410
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.40908395194e-05,
+      "loss": 2.3895,
+      "step": 8420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.408969158950027e-05,
+      "loss": 2.4607,
+      "step": 8430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.40885436596005e-05,
+      "loss": 2.4244,
+      "step": 8440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.408739572970077e-05,
+      "loss": 2.3239,
+      "step": 8450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.408624779980102e-05,
+      "loss": 2.5026,
+      "step": 8460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.408509986990127e-05,
+      "loss": 2.5441,
+      "step": 8470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.408395194000152e-05,
+      "loss": 2.3849,
+      "step": 8480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.408280401010178e-05,
+      "loss": 2.397,
+      "step": 8490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.408165608020203e-05,
+      "loss": 2.4597,
+      "step": 8500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.408050815030228e-05,
+      "loss": 2.4396,
+      "step": 8510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.407936022040253e-05,
+      "loss": 2.4501,
+      "step": 8520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.407821229050278e-05,
+      "loss": 2.4649,
+      "step": 8530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.407706436060304e-05,
+      "loss": 2.6184,
+      "step": 8540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.40759164307033e-05,
+      "loss": 2.4982,
+      "step": 8550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.407476850080354e-05,
+      "loss": 2.596,
+      "step": 8560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.40736205709038e-05,
+      "loss": 2.3217,
+      "step": 8570
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.407247264100406e-05,
+      "loss": 2.2922,
+      "step": 8580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.40713247111043e-05,
+      "loss": 2.571,
+      "step": 8590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.407017678120456e-05,
+      "loss": 2.4764,
+      "step": 8600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.406902885130481e-05,
+      "loss": 2.478,
+      "step": 8610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.406788092140506e-05,
+      "loss": 2.6295,
+      "step": 8620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.406673299150531e-05,
+      "loss": 2.5292,
+      "step": 8630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.406558506160557e-05,
+      "loss": 2.4299,
+      "step": 8640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.406443713170582e-05,
+      "loss": 2.3595,
+      "step": 8650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.406328920180607e-05,
+      "loss": 2.3447,
+      "step": 8660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.406214127190632e-05,
+      "loss": 2.4882,
+      "step": 8670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.406099334200657e-05,
+      "loss": 2.4932,
+      "step": 8680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.405984541210683e-05,
+      "loss": 2.4043,
+      "step": 8690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.405869748220708e-05,
+      "loss": 2.3409,
+      "step": 8700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.405754955230733e-05,
+      "loss": 2.2872,
+      "step": 8710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.405640162240758e-05,
+      "loss": 2.5622,
+      "step": 8720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.405525369250785e-05,
+      "loss": 2.3997,
+      "step": 8730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.405410576260808e-05,
+      "loss": 2.5346,
+      "step": 8740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.405295783270835e-05,
+      "loss": 2.5783,
+      "step": 8750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.40518099028086e-05,
+      "loss": 2.4764,
+      "step": 8760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.405066197290885e-05,
+      "loss": 2.5222,
+      "step": 8770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.40495140430091e-05,
+      "loss": 2.4583,
+      "step": 8780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.404836611310936e-05,
+      "loss": 2.4888,
+      "step": 8790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.404721818320961e-05,
+      "loss": 2.5642,
+      "step": 8800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.404607025330986e-05,
+      "loss": 2.5271,
+      "step": 8810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.404492232341011e-05,
+      "loss": 2.2613,
+      "step": 8820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.404377439351036e-05,
+      "loss": 2.4599,
+      "step": 8830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.404262646361062e-05,
+      "loss": 2.3576,
+      "step": 8840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.404147853371087e-05,
+      "loss": 2.3699,
+      "step": 8850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.404033060381112e-05,
+      "loss": 2.4351,
+      "step": 8860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.403918267391137e-05,
+      "loss": 2.506,
+      "step": 8870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.403803474401163e-05,
+      "loss": 2.4608,
+      "step": 8880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.403688681411187e-05,
+      "loss": 2.4689,
+      "step": 8890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.403573888421213e-05,
+      "loss": 2.4956,
+      "step": 8900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.403459095431238e-05,
+      "loss": 2.5486,
+      "step": 8910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.403344302441263e-05,
+      "loss": 2.464,
+      "step": 8920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.403229509451288e-05,
+      "loss": 2.4612,
+      "step": 8930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.403114716461315e-05,
+      "loss": 2.5053,
+      "step": 8940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.40299992347134e-05,
+      "loss": 2.3801,
+      "step": 8950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.402885130481365e-05,
+      "loss": 2.3124,
+      "step": 8960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.40277033749139e-05,
+      "loss": 2.5126,
+      "step": 8970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.402655544501415e-05,
+      "loss": 2.2925,
+      "step": 8980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.402540751511441e-05,
+      "loss": 2.4248,
+      "step": 8990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.402425958521466e-05,
+      "loss": 2.419,
+      "step": 9000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.402311165531491e-05,
+      "loss": 2.5001,
+      "step": 9010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.402196372541516e-05,
+      "loss": 2.3678,
+      "step": 9020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.402081579551542e-05,
+      "loss": 2.3578,
+      "step": 9030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.401966786561566e-05,
+      "loss": 2.6891,
+      "step": 9040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.401851993571592e-05,
+      "loss": 2.4932,
+      "step": 9050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.401737200581617e-05,
+      "loss": 2.3296,
+      "step": 9060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.401622407591642e-05,
+      "loss": 2.4056,
+      "step": 9070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.401507614601667e-05,
+      "loss": 2.2845,
+      "step": 9080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.401392821611694e-05,
+      "loss": 2.355,
+      "step": 9090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.401278028621719e-05,
+      "loss": 2.3486,
+      "step": 9100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.401163235631744e-05,
+      "loss": 2.404,
+      "step": 9110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.401048442641769e-05,
+      "loss": 2.4653,
+      "step": 9120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.400933649651794e-05,
+      "loss": 2.4725,
+      "step": 9130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.40081885666182e-05,
+      "loss": 2.4857,
+      "step": 9140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.400704063671845e-05,
+      "loss": 2.3903,
+      "step": 9150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.40058927068187e-05,
+      "loss": 2.3422,
+      "step": 9160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.400474477691895e-05,
+      "loss": 2.4166,
+      "step": 9170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.400359684701921e-05,
+      "loss": 2.4943,
+      "step": 9180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.400244891711945e-05,
+      "loss": 2.5084,
+      "step": 9190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.400130098721971e-05,
+      "loss": 2.3879,
+      "step": 9200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.400015305731996e-05,
+      "loss": 2.3852,
+      "step": 9210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.399900512742021e-05,
+      "loss": 2.3985,
+      "step": 9220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.399785719752046e-05,
+      "loss": 2.367,
+      "step": 9230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.399670926762072e-05,
+      "loss": 2.3779,
+      "step": 9240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.399556133772097e-05,
+      "loss": 2.5811,
+      "step": 9250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.399441340782122e-05,
+      "loss": 2.2684,
+      "step": 9260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.399326547792147e-05,
+      "loss": 2.355,
+      "step": 9270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.399211754802172e-05,
+      "loss": 2.5137,
+      "step": 9280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.399096961812199e-05,
+      "loss": 2.2747,
+      "step": 9290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.398982168822224e-05,
+      "loss": 2.3249,
+      "step": 9300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.398867375832249e-05,
+      "loss": 2.3487,
+      "step": 9310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.398752582842274e-05,
+      "loss": 2.3265,
+      "step": 9320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.3986377898523e-05,
+      "loss": 2.3974,
+      "step": 9330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.398522996862324e-05,
+      "loss": 2.4213,
+      "step": 9340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.39840820387235e-05,
+      "loss": 2.6385,
+      "step": 9350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.398293410882375e-05,
+      "loss": 2.2452,
+      "step": 9360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.3981786178924e-05,
+      "loss": 2.4249,
+      "step": 9370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.398063824902425e-05,
+      "loss": 2.5345,
+      "step": 9380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.397949031912451e-05,
+      "loss": 2.4251,
+      "step": 9390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.397834238922476e-05,
+      "loss": 2.6291,
+      "step": 9400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.397719445932501e-05,
+      "loss": 2.3414,
+      "step": 9410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.397604652942526e-05,
+      "loss": 2.29,
+      "step": 9420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.397489859952551e-05,
+      "loss": 2.4312,
+      "step": 9430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.397375066962578e-05,
+      "loss": 2.4964,
+      "step": 9440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.397260273972601e-05,
+      "loss": 2.4861,
+      "step": 9450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.397145480982628e-05,
+      "loss": 2.3922,
+      "step": 9460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.397030687992653e-05,
+      "loss": 2.4208,
+      "step": 9470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.396915895002677e-05,
+      "loss": 2.4215,
+      "step": 9480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.396801102012702e-05,
+      "loss": 2.3285,
+      "step": 9490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.396686309022729e-05,
+      "loss": 2.4463,
+      "step": 9500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.396571516032754e-05,
+      "loss": 2.4225,
+      "step": 9510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.396456723042779e-05,
+      "loss": 2.4135,
+      "step": 9520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.396341930052804e-05,
+      "loss": 2.3217,
+      "step": 9530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.396227137062829e-05,
+      "loss": 2.2805,
+      "step": 9540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.396112344072855e-05,
+      "loss": 2.4706,
+      "step": 9550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.39599755108288e-05,
+      "loss": 2.4158,
+      "step": 9560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.395882758092905e-05,
+      "loss": 2.5718,
+      "step": 9570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.39576796510293e-05,
+      "loss": 2.3949,
+      "step": 9580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.395653172112956e-05,
+      "loss": 2.2696,
+      "step": 9590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.39553837912298e-05,
+      "loss": 2.39,
+      "step": 9600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.395423586133006e-05,
+      "loss": 2.2425,
+      "step": 9610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.395308793143031e-05,
+      "loss": 2.387,
+      "step": 9620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.395194000153056e-05,
+      "loss": 2.272,
+      "step": 9630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.395079207163081e-05,
+      "loss": 2.5719,
+      "step": 9640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.394964414173108e-05,
+      "loss": 2.3377,
+      "step": 9650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.394849621183133e-05,
+      "loss": 2.487,
+      "step": 9660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.394734828193158e-05,
+      "loss": 2.6558,
+      "step": 9670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.394620035203183e-05,
+      "loss": 2.2869,
+      "step": 9680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.394505242213208e-05,
+      "loss": 2.4923,
+      "step": 9690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.394390449223234e-05,
+      "loss": 2.5278,
+      "step": 9700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.394275656233259e-05,
+      "loss": 2.1439,
+      "step": 9710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.394160863243284e-05,
+      "loss": 2.475,
+      "step": 9720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.394046070253309e-05,
+      "loss": 2.5219,
+      "step": 9730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.393931277263335e-05,
+      "loss": 2.4108,
+      "step": 9740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.393816484273359e-05,
+      "loss": 2.3457,
+      "step": 9750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.393701691283385e-05,
+      "loss": 2.6334,
+      "step": 9760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.39358689829341e-05,
+      "loss": 2.2417,
+      "step": 9770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.393472105303435e-05,
+      "loss": 2.4549,
+      "step": 9780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.39335731231346e-05,
+      "loss": 2.4922,
+      "step": 9790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.393242519323487e-05,
+      "loss": 2.3684,
+      "step": 9800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.393127726333511e-05,
+      "loss": 2.3319,
+      "step": 9810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.393012933343536e-05,
+      "loss": 2.4571,
+      "step": 9820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.392898140353561e-05,
+      "loss": 2.425,
+      "step": 9830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.392783347363586e-05,
+      "loss": 2.413,
+      "step": 9840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.392668554373613e-05,
+      "loss": 2.4079,
+      "step": 9850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.392553761383638e-05,
+      "loss": 2.3138,
+      "step": 9860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.392438968393663e-05,
+      "loss": 2.2556,
+      "step": 9870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.392324175403688e-05,
+      "loss": 2.4765,
+      "step": 9880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.392209382413714e-05,
+      "loss": 2.4289,
+      "step": 9890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.392094589423738e-05,
+      "loss": 2.4863,
+      "step": 9900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.391979796433764e-05,
+      "loss": 2.5249,
+      "step": 9910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.391865003443789e-05,
+      "loss": 2.2344,
+      "step": 9920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.391750210453814e-05,
+      "loss": 2.4126,
+      "step": 9930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.391635417463839e-05,
+      "loss": 2.3149,
+      "step": 9940
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.391520624473865e-05,
+      "loss": 2.327,
+      "step": 9950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.39140583148389e-05,
+      "loss": 2.2465,
+      "step": 9960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.391291038493915e-05,
+      "loss": 2.4432,
+      "step": 9970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.39117624550394e-05,
+      "loss": 2.3712,
+      "step": 9980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.391061452513965e-05,
+      "loss": 2.3845,
+      "step": 9990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.390946659523992e-05,
+      "loss": 2.3821,
+      "step": 10000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.390831866534017e-05,
+      "loss": 2.271,
+      "step": 10010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.390717073544042e-05,
+      "loss": 2.1434,
+      "step": 10020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.390602280554067e-05,
+      "loss": 2.4505,
+      "step": 10030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.390487487564093e-05,
+      "loss": 2.5227,
+      "step": 10040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.390372694574117e-05,
+      "loss": 2.422,
+      "step": 10050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.390257901584143e-05,
+      "loss": 2.2901,
+      "step": 10060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.390143108594168e-05,
+      "loss": 2.4317,
+      "step": 10070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.390028315604193e-05,
+      "loss": 2.3955,
+      "step": 10080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.389913522614218e-05,
+      "loss": 2.1425,
+      "step": 10090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.389798729624244e-05,
+      "loss": 2.2226,
+      "step": 10100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.389683936634269e-05,
+      "loss": 2.4621,
+      "step": 10110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.389569143644294e-05,
+      "loss": 2.4293,
+      "step": 10120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.389454350654319e-05,
+      "loss": 2.2333,
+      "step": 10130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.389339557664344e-05,
+      "loss": 2.4262,
+      "step": 10140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.38922476467437e-05,
+      "loss": 2.356,
+      "step": 10150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.389109971684395e-05,
+      "loss": 2.275,
+      "step": 10160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.38899517869442e-05,
+      "loss": 2.4345,
+      "step": 10170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.388880385704445e-05,
+      "loss": 2.1931,
+      "step": 10180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.388765592714472e-05,
+      "loss": 2.2639,
+      "step": 10190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.388650799724495e-05,
+      "loss": 2.1628,
+      "step": 10200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.388536006734522e-05,
+      "loss": 2.2662,
+      "step": 10210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.388421213744547e-05,
+      "loss": 2.5224,
+      "step": 10220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.388306420754572e-05,
+      "loss": 2.3744,
+      "step": 10230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.388191627764597e-05,
+      "loss": 2.2251,
+      "step": 10240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.388076834774623e-05,
+      "loss": 2.3539,
+      "step": 10250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.387962041784648e-05,
+      "loss": 2.4698,
+      "step": 10260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.387847248794673e-05,
+      "loss": 2.494,
+      "step": 10270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.387732455804698e-05,
+      "loss": 2.2839,
+      "step": 10280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.387617662814723e-05,
+      "loss": 2.3071,
+      "step": 10290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.38750286982475e-05,
+      "loss": 2.4762,
+      "step": 10300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.387388076834774e-05,
+      "loss": 2.4067,
+      "step": 10310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.387273283844799e-05,
+      "loss": 2.3881,
+      "step": 10320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.387158490854824e-05,
+      "loss": 2.5002,
+      "step": 10330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.38704369786485e-05,
+      "loss": 2.267,
+      "step": 10340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.386928904874874e-05,
+      "loss": 2.3602,
+      "step": 10350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.3868141118849e-05,
+      "loss": 2.3916,
+      "step": 10360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.386699318894926e-05,
+      "loss": 2.4821,
+      "step": 10370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.38658452590495e-05,
+      "loss": 2.2948,
+      "step": 10380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.386469732914976e-05,
+      "loss": 2.4443,
+      "step": 10390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.386354939925002e-05,
+      "loss": 2.1842,
+      "step": 10400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.386240146935027e-05,
+      "loss": 2.429,
+      "step": 10410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.386125353945052e-05,
+      "loss": 2.3127,
+      "step": 10420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.386010560955077e-05,
+      "loss": 2.3859,
+      "step": 10430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.385895767965102e-05,
+      "loss": 2.3572,
+      "step": 10440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.385780974975128e-05,
+      "loss": 2.3231,
+      "step": 10450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.385666181985153e-05,
+      "loss": 2.4732,
+      "step": 10460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.385551388995178e-05,
+      "loss": 2.4238,
+      "step": 10470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.385436596005203e-05,
+      "loss": 2.4412,
+      "step": 10480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.38532180301523e-05,
+      "loss": 2.3499,
+      "step": 10490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.385207010025253e-05,
+      "loss": 2.2671,
+      "step": 10500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.38509221703528e-05,
+      "loss": 2.2533,
+      "step": 10510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.384977424045304e-05,
+      "loss": 2.3043,
+      "step": 10520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.38486263105533e-05,
+      "loss": 2.3881,
+      "step": 10530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.384747838065354e-05,
+      "loss": 2.357,
+      "step": 10540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.384633045075381e-05,
+      "loss": 2.2684,
+      "step": 10550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.384518252085406e-05,
+      "loss": 2.2792,
+      "step": 10560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.384403459095431e-05,
+      "loss": 2.4235,
+      "step": 10570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.384288666105456e-05,
+      "loss": 2.5161,
+      "step": 10580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.384173873115481e-05,
+      "loss": 2.3757,
+      "step": 10590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.384059080125507e-05,
+      "loss": 2.3657,
+      "step": 10600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.383944287135532e-05,
+      "loss": 2.4695,
+      "step": 10610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.383829494145557e-05,
+      "loss": 2.4115,
+      "step": 10620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.383714701155582e-05,
+      "loss": 2.3022,
+      "step": 10630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.383599908165608e-05,
+      "loss": 2.5163,
+      "step": 10640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.383485115175632e-05,
+      "loss": 2.2954,
+      "step": 10650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.383370322185658e-05,
+      "loss": 2.1751,
+      "step": 10660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.383255529195683e-05,
+      "loss": 2.4378,
+      "step": 10670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.383140736205708e-05,
+      "loss": 2.345,
+      "step": 10680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.383025943215733e-05,
+      "loss": 2.3188,
+      "step": 10690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.38291115022576e-05,
+      "loss": 2.3681,
+      "step": 10700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.382796357235785e-05,
+      "loss": 2.3591,
+      "step": 10710
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.38268156424581e-05,
+      "loss": 2.4835,
+      "step": 10720
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.382566771255835e-05,
+      "loss": 2.4497,
+      "step": 10730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.38245197826586e-05,
+      "loss": 2.5095,
+      "step": 10740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.382337185275886e-05,
+      "loss": 2.4533,
+      "step": 10750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.382222392285911e-05,
+      "loss": 2.418,
+      "step": 10760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.382107599295936e-05,
+      "loss": 2.333,
+      "step": 10770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.381992806305961e-05,
+      "loss": 2.2331,
+      "step": 10780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.381878013315987e-05,
+      "loss": 2.3604,
+      "step": 10790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.381763220326011e-05,
+      "loss": 2.2745,
+      "step": 10800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.381648427336037e-05,
+      "loss": 2.529,
+      "step": 10810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.381533634346062e-05,
+      "loss": 2.3826,
+      "step": 10820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.381418841356087e-05,
+      "loss": 2.4125,
+      "step": 10830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.381304048366112e-05,
+      "loss": 2.3584,
+      "step": 10840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.381189255376137e-05,
+      "loss": 2.5134,
+      "step": 10850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.381074462386163e-05,
+      "loss": 2.338,
+      "step": 10860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.380959669396188e-05,
+      "loss": 2.2527,
+      "step": 10870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.380844876406213e-05,
+      "loss": 2.4074,
+      "step": 10880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.380730083416238e-05,
+      "loss": 2.4478,
+      "step": 10890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.380615290426265e-05,
+      "loss": 2.3271,
+      "step": 10900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.380500497436288e-05,
+      "loss": 2.3356,
+      "step": 10910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.380385704446315e-05,
+      "loss": 2.4516,
+      "step": 10920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.38027091145634e-05,
+      "loss": 2.4043,
+      "step": 10930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.380156118466365e-05,
+      "loss": 2.3021,
+      "step": 10940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.38004132547639e-05,
+      "loss": 2.4477,
+      "step": 10950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.379926532486416e-05,
+      "loss": 2.3816,
+      "step": 10960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.379811739496441e-05,
+      "loss": 2.3474,
+      "step": 10970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.379696946506466e-05,
+      "loss": 2.4045,
+      "step": 10980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.379582153516491e-05,
+      "loss": 2.4954,
+      "step": 10990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.379467360526516e-05,
+      "loss": 2.1915,
+      "step": 11000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.379352567536542e-05,
+      "loss": 2.37,
+      "step": 11010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.379237774546567e-05,
+      "loss": 2.376,
+      "step": 11020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.379122981556592e-05,
+      "loss": 2.2351,
+      "step": 11030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.379008188566617e-05,
+      "loss": 2.5471,
+      "step": 11040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.378893395576644e-05,
+      "loss": 2.2648,
+      "step": 11050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.378778602586667e-05,
+      "loss": 2.3284,
+      "step": 11060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.378663809596694e-05,
+      "loss": 2.1715,
+      "step": 11070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.378549016606719e-05,
+      "loss": 2.3307,
+      "step": 11080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.378434223616744e-05,
+      "loss": 2.445,
+      "step": 11090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.378319430626769e-05,
+      "loss": 2.2966,
+      "step": 11100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.378204637636795e-05,
+      "loss": 2.2975,
+      "step": 11110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.37808984464682e-05,
+      "loss": 2.324,
+      "step": 11120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.377975051656845e-05,
+      "loss": 2.3363,
+      "step": 11130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.37786025866687e-05,
+      "loss": 2.2898,
+      "step": 11140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.377745465676895e-05,
+      "loss": 2.5369,
+      "step": 11150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.377630672686921e-05,
+      "loss": 2.3867,
+      "step": 11160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.377515879696946e-05,
+      "loss": 2.1834,
+      "step": 11170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.377401086706971e-05,
+      "loss": 2.4214,
+      "step": 11180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.377286293716996e-05,
+      "loss": 2.383,
+      "step": 11190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.377171500727022e-05,
+      "loss": 2.4393,
+      "step": 11200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.377056707737046e-05,
+      "loss": 2.2626,
+      "step": 11210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.376941914747072e-05,
+      "loss": 2.5723,
+      "step": 11220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.376827121757097e-05,
+      "loss": 2.3699,
+      "step": 11230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.376712328767122e-05,
+      "loss": 2.2889,
+      "step": 11240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.376597535777147e-05,
+      "loss": 2.3935,
+      "step": 11250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.376482742787174e-05,
+      "loss": 2.4298,
+      "step": 11260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.376367949797199e-05,
+      "loss": 2.4403,
+      "step": 11270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.376253156807224e-05,
+      "loss": 2.3397,
+      "step": 11280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.376138363817249e-05,
+      "loss": 2.3147,
+      "step": 11290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.376023570827274e-05,
+      "loss": 2.2914,
+      "step": 11300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.3759087778373e-05,
+      "loss": 2.2931,
+      "step": 11310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.375793984847325e-05,
+      "loss": 2.474,
+      "step": 11320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.37567919185735e-05,
+      "loss": 2.3722,
+      "step": 11330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.375564398867375e-05,
+      "loss": 2.3729,
+      "step": 11340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.375449605877401e-05,
+      "loss": 2.3508,
+      "step": 11350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.375334812887425e-05,
+      "loss": 2.4189,
+      "step": 11360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.375220019897451e-05,
+      "loss": 2.3625,
+      "step": 11370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.375105226907476e-05,
+      "loss": 2.2821,
+      "step": 11380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.374990433917501e-05,
+      "loss": 2.1243,
+      "step": 11390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.374875640927526e-05,
+      "loss": 2.5268,
+      "step": 11400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.374760847937553e-05,
+      "loss": 2.5364,
+      "step": 11410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.374646054947578e-05,
+      "loss": 2.3356,
+      "step": 11420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.374531261957603e-05,
+      "loss": 2.3403,
+      "step": 11430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.374416468967627e-05,
+      "loss": 2.1942,
+      "step": 11440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.374301675977652e-05,
+      "loss": 2.3442,
+      "step": 11450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.374186882987679e-05,
+      "loss": 2.5323,
+      "step": 11460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.374072089997704e-05,
+      "loss": 2.5004,
+      "step": 11470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.373957297007729e-05,
+      "loss": 2.4023,
+      "step": 11480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.373842504017754e-05,
+      "loss": 2.4,
+      "step": 11490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.37372771102778e-05,
+      "loss": 2.4025,
+      "step": 11500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.373612918037804e-05,
+      "loss": 2.3317,
+      "step": 11510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.37349812504783e-05,
+      "loss": 2.44,
+      "step": 11520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.373383332057855e-05,
+      "loss": 2.425,
+      "step": 11530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.37326853906788e-05,
+      "loss": 2.2812,
+      "step": 11540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.373153746077905e-05,
+      "loss": 2.385,
+      "step": 11550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.373038953087931e-05,
+      "loss": 2.2369,
+      "step": 11560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.372924160097956e-05,
+      "loss": 2.2852,
+      "step": 11570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.372809367107981e-05,
+      "loss": 2.2471,
+      "step": 11580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.372694574118006e-05,
+      "loss": 2.3977,
+      "step": 11590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.372579781128031e-05,
+      "loss": 2.3631,
+      "step": 11600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.372464988138058e-05,
+      "loss": 2.3651,
+      "step": 11610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.372350195148083e-05,
+      "loss": 2.3743,
+      "step": 11620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.372235402158108e-05,
+      "loss": 2.2272,
+      "step": 11630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.372120609168133e-05,
+      "loss": 2.3615,
+      "step": 11640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.372005816178159e-05,
+      "loss": 2.4234,
+      "step": 11650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.371891023188183e-05,
+      "loss": 2.2753,
+      "step": 11660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.371776230198209e-05,
+      "loss": 2.5549,
+      "step": 11670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.371661437208234e-05,
+      "loss": 2.35,
+      "step": 11680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.371546644218259e-05,
+      "loss": 2.3645,
+      "step": 11690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.371431851228284e-05,
+      "loss": 2.4055,
+      "step": 11700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.37131705823831e-05,
+      "loss": 2.3869,
+      "step": 11710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.371202265248335e-05,
+      "loss": 2.4044,
+      "step": 11720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.37108747225836e-05,
+      "loss": 2.3254,
+      "step": 11730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.370972679268385e-05,
+      "loss": 2.4501,
+      "step": 11740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.37085788627841e-05,
+      "loss": 2.4091,
+      "step": 11750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.370743093288437e-05,
+      "loss": 2.5114,
+      "step": 11760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.370628300298461e-05,
+      "loss": 2.4002,
+      "step": 11770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.370513507308486e-05,
+      "loss": 2.3165,
+      "step": 11780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.370398714318511e-05,
+      "loss": 2.2193,
+      "step": 11790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.370283921328538e-05,
+      "loss": 2.4262,
+      "step": 11800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.370169128338561e-05,
+      "loss": 2.2108,
+      "step": 11810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.370054335348588e-05,
+      "loss": 2.4794,
+      "step": 11820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.369939542358613e-05,
+      "loss": 2.269,
+      "step": 11830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.369824749368638e-05,
+      "loss": 2.326,
+      "step": 11840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.369709956378663e-05,
+      "loss": 2.2338,
+      "step": 11850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.369595163388689e-05,
+      "loss": 2.3256,
+      "step": 11860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.369480370398714e-05,
+      "loss": 2.3478,
+      "step": 11870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.369365577408739e-05,
+      "loss": 2.3314,
+      "step": 11880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.369250784418764e-05,
+      "loss": 2.4038,
+      "step": 11890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.369135991428789e-05,
+      "loss": 2.1864,
+      "step": 11900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.369021198438815e-05,
+      "loss": 2.4204,
+      "step": 11910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.36890640544884e-05,
+      "loss": 2.2147,
+      "step": 11920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.368791612458865e-05,
+      "loss": 2.5668,
+      "step": 11930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.36867681946889e-05,
+      "loss": 2.332,
+      "step": 11940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.368562026478917e-05,
+      "loss": 2.3703,
+      "step": 11950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.36844723348894e-05,
+      "loss": 2.4366,
+      "step": 11960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.368332440498967e-05,
+      "loss": 2.354,
+      "step": 11970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.368217647508992e-05,
+      "loss": 2.2652,
+      "step": 11980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.368102854519017e-05,
+      "loss": 2.6514,
+      "step": 11990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.367988061529042e-05,
+      "loss": 2.1994,
+      "step": 12000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.367873268539068e-05,
+      "loss": 2.4806,
+      "step": 12010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.367758475549093e-05,
+      "loss": 2.2287,
+      "step": 12020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.367643682559118e-05,
+      "loss": 2.5402,
+      "step": 12030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.367528889569143e-05,
+      "loss": 2.4361,
+      "step": 12040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.367414096579168e-05,
+      "loss": 2.3656,
+      "step": 12050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.367299303589194e-05,
+      "loss": 2.2664,
+      "step": 12060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.367184510599219e-05,
+      "loss": 2.2987,
+      "step": 12070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.367069717609244e-05,
+      "loss": 2.407,
+      "step": 12080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.366954924619269e-05,
+      "loss": 2.4681,
+      "step": 12090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.366840131629296e-05,
+      "loss": 2.2474,
+      "step": 12100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.366725338639319e-05,
+      "loss": 2.3242,
+      "step": 12110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.366610545649345e-05,
+      "loss": 2.3703,
+      "step": 12120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.36649575265937e-05,
+      "loss": 2.3122,
+      "step": 12130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.366380959669395e-05,
+      "loss": 2.2348,
+      "step": 12140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.36626616667942e-05,
+      "loss": 2.4153,
+      "step": 12150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.366151373689447e-05,
+      "loss": 2.353,
+      "step": 12160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.366036580699472e-05,
+      "loss": 2.3627,
+      "step": 12170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.365921787709497e-05,
+      "loss": 2.4828,
+      "step": 12180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.365806994719522e-05,
+      "loss": 2.4023,
+      "step": 12190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.365692201729547e-05,
+      "loss": 2.4205,
+      "step": 12200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.365577408739573e-05,
+      "loss": 2.1948,
+      "step": 12210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.365462615749597e-05,
+      "loss": 2.3802,
+      "step": 12220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.365347822759623e-05,
+      "loss": 2.5415,
+      "step": 12230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.365233029769648e-05,
+      "loss": 2.4706,
+      "step": 12240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.365118236779673e-05,
+      "loss": 2.2677,
+      "step": 12250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.365003443789698e-05,
+      "loss": 2.3108,
+      "step": 12260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.364888650799724e-05,
+      "loss": 2.3898,
+      "step": 12270
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.364773857809749e-05,
+      "loss": 2.425,
+      "step": 12280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.364659064819774e-05,
+      "loss": 2.2774,
+      "step": 12290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.364544271829799e-05,
+      "loss": 2.2725,
+      "step": 12300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.364429478839824e-05,
+      "loss": 2.2732,
+      "step": 12310
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.36431468584985e-05,
+      "loss": 2.3666,
+      "step": 12320
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.364199892859876e-05,
+      "loss": 2.4211,
+      "step": 12330
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.3640850998699e-05,
+      "loss": 2.3356,
+      "step": 12340
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.363970306879926e-05,
+      "loss": 2.3319,
+      "step": 12350
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.363855513889952e-05,
+      "loss": 2.5851,
+      "step": 12360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.363740720899976e-05,
+      "loss": 2.2406,
+      "step": 12370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.363625927910002e-05,
+      "loss": 2.3085,
+      "step": 12380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.363511134920027e-05,
+      "loss": 2.2373,
+      "step": 12390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.363396341930052e-05,
+      "loss": 2.3122,
+      "step": 12400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.363281548940077e-05,
+      "loss": 2.3204,
+      "step": 12410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.363166755950103e-05,
+      "loss": 2.479,
+      "step": 12420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.363051962960128e-05,
+      "loss": 2.2807,
+      "step": 12430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.362937169970153e-05,
+      "loss": 2.2451,
+      "step": 12440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.362822376980178e-05,
+      "loss": 2.3542,
+      "step": 12450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.362707583990203e-05,
+      "loss": 2.3743,
+      "step": 12460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.36259279100023e-05,
+      "loss": 2.3822,
+      "step": 12470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.362477998010254e-05,
+      "loss": 2.4121,
+      "step": 12480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.36236320502028e-05,
+      "loss": 2.4187,
+      "step": 12490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.362248412030304e-05,
+      "loss": 2.3057,
+      "step": 12500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.362133619040331e-05,
+      "loss": 2.3269,
+      "step": 12510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.362018826050354e-05,
+      "loss": 2.271,
+      "step": 12520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.361904033060381e-05,
+      "loss": 2.3977,
+      "step": 12530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.361789240070406e-05,
+      "loss": 2.2111,
+      "step": 12540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.361674447080431e-05,
+      "loss": 2.4026,
+      "step": 12550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.361559654090456e-05,
+      "loss": 2.3818,
+      "step": 12560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.361444861100482e-05,
+      "loss": 2.3616,
+      "step": 12570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.361330068110507e-05,
+      "loss": 2.4243,
+      "step": 12580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.361215275120532e-05,
+      "loss": 2.3006,
+      "step": 12590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.361100482130557e-05,
+      "loss": 2.5167,
+      "step": 12600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.360985689140582e-05,
+      "loss": 2.4123,
+      "step": 12610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.360870896150608e-05,
+      "loss": 2.1422,
+      "step": 12620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.360756103160633e-05,
+      "loss": 2.3564,
+      "step": 12630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.360641310170658e-05,
+      "loss": 2.4115,
+      "step": 12640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.360526517180683e-05,
+      "loss": 2.3314,
+      "step": 12650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.36041172419071e-05,
+      "loss": 2.3689,
+      "step": 12660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.360296931200733e-05,
+      "loss": 2.2997,
+      "step": 12670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.36018213821076e-05,
+      "loss": 2.3525,
+      "step": 12680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.360067345220785e-05,
+      "loss": 2.3043,
+      "step": 12690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.35995255223081e-05,
+      "loss": 2.3326,
+      "step": 12700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.359837759240835e-05,
+      "loss": 2.3074,
+      "step": 12710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.359722966250861e-05,
+      "loss": 2.4562,
+      "step": 12720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.359608173260886e-05,
+      "loss": 2.3826,
+      "step": 12730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.359493380270911e-05,
+      "loss": 2.3318,
+      "step": 12740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.359378587280936e-05,
+      "loss": 2.2759,
+      "step": 12750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.359263794290961e-05,
+      "loss": 2.1905,
+      "step": 12760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.359149001300987e-05,
+      "loss": 2.1773,
+      "step": 12770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.359034208311012e-05,
+      "loss": 2.3424,
+      "step": 12780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.358919415321037e-05,
+      "loss": 2.3363,
+      "step": 12790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.358804622331062e-05,
+      "loss": 2.3675,
+      "step": 12800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.358689829341088e-05,
+      "loss": 2.4018,
+      "step": 12810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.358575036351112e-05,
+      "loss": 2.3343,
+      "step": 12820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.358460243361138e-05,
+      "loss": 2.3259,
+      "step": 12830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.358345450371163e-05,
+      "loss": 2.3028,
+      "step": 12840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.358230657381188e-05,
+      "loss": 2.332,
+      "step": 12850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.358115864391213e-05,
+      "loss": 2.2914,
+      "step": 12860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.35800107140124e-05,
+      "loss": 2.4118,
+      "step": 12870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.357886278411265e-05,
+      "loss": 2.315,
+      "step": 12880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.35777148542129e-05,
+      "loss": 2.2677,
+      "step": 12890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.357656692431315e-05,
+      "loss": 2.103,
+      "step": 12900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.35754189944134e-05,
+      "loss": 2.2763,
+      "step": 12910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.357427106451366e-05,
+      "loss": 2.5329,
+      "step": 12920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.357312313461391e-05,
+      "loss": 2.1768,
+      "step": 12930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.357197520471416e-05,
+      "loss": 2.2341,
+      "step": 12940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.357082727481441e-05,
+      "loss": 2.2603,
+      "step": 12950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.356967934491467e-05,
+      "loss": 2.2989,
+      "step": 12960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.356853141501491e-05,
+      "loss": 2.3064,
+      "step": 12970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.356738348511517e-05,
+      "loss": 2.2096,
+      "step": 12980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.356623555521542e-05,
+      "loss": 2.2708,
+      "step": 12990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.356508762531567e-05,
+      "loss": 2.3146,
+      "step": 13000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.356393969541592e-05,
+      "loss": 2.4268,
+      "step": 13010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.356279176551619e-05,
+      "loss": 2.1831,
+      "step": 13020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.356164383561644e-05,
+      "loss": 2.2272,
+      "step": 13030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.356049590571669e-05,
+      "loss": 2.3226,
+      "step": 13040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.355934797581694e-05,
+      "loss": 2.3758,
+      "step": 13050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.355820004591719e-05,
+      "loss": 2.3052,
+      "step": 13060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.355705211601745e-05,
+      "loss": 2.3094,
+      "step": 13070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.35559041861177e-05,
+      "loss": 2.4553,
+      "step": 13080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.355475625621795e-05,
+      "loss": 2.292,
+      "step": 13090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.35536083263182e-05,
+      "loss": 2.31,
+      "step": 13100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.355246039641846e-05,
+      "loss": 2.3864,
+      "step": 13110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.35513124665187e-05,
+      "loss": 2.3587,
+      "step": 13120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.355016453661896e-05,
+      "loss": 2.5124,
+      "step": 13130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.354901660671921e-05,
+      "loss": 2.2575,
+      "step": 13140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.354786867681946e-05,
+      "loss": 2.3064,
+      "step": 13150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.354672074691971e-05,
+      "loss": 2.2703,
+      "step": 13160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.354557281701997e-05,
+      "loss": 2.3634,
+      "step": 13170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.354442488712022e-05,
+      "loss": 2.293,
+      "step": 13180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.354327695722047e-05,
+      "loss": 2.2826,
+      "step": 13190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.354212902732072e-05,
+      "loss": 2.4001,
+      "step": 13200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.354098109742097e-05,
+      "loss": 2.206,
+      "step": 13210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.353983316752124e-05,
+      "loss": 2.3091,
+      "step": 13220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.353868523762149e-05,
+      "loss": 2.3232,
+      "step": 13230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.353753730772174e-05,
+      "loss": 2.2874,
+      "step": 13240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.353638937782199e-05,
+      "loss": 2.3084,
+      "step": 13250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.353524144792225e-05,
+      "loss": 2.2454,
+      "step": 13260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.353409351802249e-05,
+      "loss": 2.4223,
+      "step": 13270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.353294558812275e-05,
+      "loss": 2.3552,
+      "step": 13280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.3531797658223e-05,
+      "loss": 2.1786,
+      "step": 13290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.353064972832325e-05,
+      "loss": 2.4732,
+      "step": 13300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.35295017984235e-05,
+      "loss": 2.2598,
+      "step": 13310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.352835386852376e-05,
+      "loss": 2.1984,
+      "step": 13320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.352720593862401e-05,
+      "loss": 2.2942,
+      "step": 13330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.352605800872426e-05,
+      "loss": 2.223,
+      "step": 13340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.352491007882451e-05,
+      "loss": 2.2408,
+      "step": 13350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.352376214892476e-05,
+      "loss": 2.4088,
+      "step": 13360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.352261421902503e-05,
+      "loss": 2.2642,
+      "step": 13370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.352146628912528e-05,
+      "loss": 2.4924,
+      "step": 13380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.352031835922553e-05,
+      "loss": 2.2789,
+      "step": 13390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.351917042932577e-05,
+      "loss": 2.2417,
+      "step": 13400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.351802249942604e-05,
+      "loss": 2.2101,
+      "step": 13410
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.351687456952627e-05,
+      "loss": 2.509,
+      "step": 13420
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.351572663962654e-05,
+      "loss": 2.1525,
+      "step": 13430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.351457870972679e-05,
+      "loss": 2.4141,
+      "step": 13440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.351343077982704e-05,
+      "loss": 2.266,
+      "step": 13450
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.351228284992729e-05,
+      "loss": 2.0951,
+      "step": 13460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.351113492002755e-05,
+      "loss": 2.2181,
+      "step": 13470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.35099869901278e-05,
+      "loss": 2.1146,
+      "step": 13480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.350883906022805e-05,
+      "loss": 2.4439,
+      "step": 13490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.35076911303283e-05,
+      "loss": 2.195,
+      "step": 13500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.350654320042855e-05,
+      "loss": 2.3261,
+      "step": 13510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.350539527052881e-05,
+      "loss": 2.2256,
+      "step": 13520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.350424734062906e-05,
+      "loss": 2.2752,
+      "step": 13530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.350309941072931e-05,
+      "loss": 2.2507,
+      "step": 13540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.350195148082956e-05,
+      "loss": 2.3102,
+      "step": 13550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.350080355092981e-05,
+      "loss": 2.3391,
+      "step": 13560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.349965562103006e-05,
+      "loss": 2.2573,
+      "step": 13570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.349850769113033e-05,
+      "loss": 2.1625,
+      "step": 13580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.349735976123058e-05,
+      "loss": 2.4131,
+      "step": 13590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.349621183133083e-05,
+      "loss": 2.4559,
+      "step": 13600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.349506390143108e-05,
+      "loss": 2.4044,
+      "step": 13610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.349391597153133e-05,
+      "loss": 2.3821,
+      "step": 13620
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.349276804163159e-05,
+      "loss": 2.2971,
+      "step": 13630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.349162011173184e-05,
+      "loss": 2.2937,
+      "step": 13640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.349047218183209e-05,
+      "loss": 2.0907,
+      "step": 13650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.348932425193234e-05,
+      "loss": 2.3585,
+      "step": 13660
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.34881763220326e-05,
+      "loss": 2.2481,
+      "step": 13670
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.348702839213284e-05,
+      "loss": 2.4779,
+      "step": 13680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.34858804622331e-05,
+      "loss": 2.3081,
+      "step": 13690
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.348473253233335e-05,
+      "loss": 2.3814,
+      "step": 13700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.34835846024336e-05,
+      "loss": 2.2353,
+      "step": 13710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.348243667253385e-05,
+      "loss": 2.306,
+      "step": 13720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.348128874263411e-05,
+      "loss": 2.4238,
+      "step": 13730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.348014081273436e-05,
+      "loss": 2.2383,
+      "step": 13740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.347899288283461e-05,
+      "loss": 2.2931,
+      "step": 13750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.347784495293486e-05,
+      "loss": 2.3193,
+      "step": 13760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.347669702303511e-05,
+      "loss": 2.2198,
+      "step": 13770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.347554909313538e-05,
+      "loss": 2.4402,
+      "step": 13780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.347440116323563e-05,
+      "loss": 2.4029,
+      "step": 13790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.347325323333588e-05,
+      "loss": 2.3742,
+      "step": 13800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.347210530343613e-05,
+      "loss": 2.4752,
+      "step": 13810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.347095737353639e-05,
+      "loss": 2.3004,
+      "step": 13820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.346980944363663e-05,
+      "loss": 2.3053,
+      "step": 13830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.346866151373689e-05,
+      "loss": 2.4795,
+      "step": 13840
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.346751358383714e-05,
+      "loss": 2.4155,
+      "step": 13850
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.346636565393739e-05,
+      "loss": 2.1328,
+      "step": 13860
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.346521772403764e-05,
+      "loss": 2.35,
+      "step": 13870
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.34640697941379e-05,
+      "loss": 2.2504,
+      "step": 13880
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.346292186423815e-05,
+      "loss": 2.4482,
+      "step": 13890
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.34617739343384e-05,
+      "loss": 2.3519,
+      "step": 13900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.346062600443865e-05,
+      "loss": 2.235,
+      "step": 13910
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.34594780745389e-05,
+      "loss": 2.4264,
+      "step": 13920
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.345833014463917e-05,
+      "loss": 2.2312,
+      "step": 13930
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.345718221473942e-05,
+      "loss": 2.2255,
+      "step": 13940
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.345603428483967e-05,
+      "loss": 2.1806,
+      "step": 13950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.345488635493992e-05,
+      "loss": 2.3928,
+      "step": 13960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.345373842504018e-05,
+      "loss": 2.2133,
+      "step": 13970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.345259049514042e-05,
+      "loss": 2.3201,
+      "step": 13980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.345144256524068e-05,
+      "loss": 2.397,
+      "step": 13990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.345029463534093e-05,
+      "loss": 2.6114,
+      "step": 14000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.344914670544118e-05,
+      "loss": 2.295,
+      "step": 14010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.344799877554143e-05,
+      "loss": 2.379,
+      "step": 14020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.344685084564169e-05,
+      "loss": 2.3791,
+      "step": 14030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.344570291574194e-05,
+      "loss": 2.3877,
+      "step": 14040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.344455498584219e-05,
+      "loss": 2.2528,
+      "step": 14050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.344340705594244e-05,
+      "loss": 2.2556,
+      "step": 14060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.344225912604269e-05,
+      "loss": 2.2157,
+      "step": 14070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.344111119614295e-05,
+      "loss": 2.2975,
+      "step": 14080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.34399632662432e-05,
+      "loss": 2.3822,
+      "step": 14090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.343881533634345e-05,
+      "loss": 2.3011,
+      "step": 14100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.34376674064437e-05,
+      "loss": 2.591,
+      "step": 14110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.343651947654397e-05,
+      "loss": 2.2273,
+      "step": 14120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.34353715466442e-05,
+      "loss": 2.4511,
+      "step": 14130
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.343422361674447e-05,
+      "loss": 2.2518,
+      "step": 14140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.343307568684472e-05,
+      "loss": 2.41,
+      "step": 14150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.343192775694497e-05,
+      "loss": 2.5384,
+      "step": 14160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.343077982704522e-05,
+      "loss": 2.3104,
+      "step": 14170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.342963189714548e-05,
+      "loss": 2.2071,
+      "step": 14180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.342848396724573e-05,
+      "loss": 2.3209,
+      "step": 14190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.342733603734598e-05,
+      "loss": 2.2609,
+      "step": 14200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.342618810744623e-05,
+      "loss": 2.2341,
+      "step": 14210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.342504017754648e-05,
+      "loss": 2.1809,
+      "step": 14220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.342389224764674e-05,
+      "loss": 2.272,
+      "step": 14230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.342274431774699e-05,
+      "loss": 2.2577,
+      "step": 14240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.342159638784724e-05,
+      "loss": 2.2964,
+      "step": 14250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.342044845794749e-05,
+      "loss": 2.3179,
+      "step": 14260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.341930052804776e-05,
+      "loss": 2.3136,
+      "step": 14270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.341815259814799e-05,
+      "loss": 2.3995,
+      "step": 14280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.341700466824826e-05,
+      "loss": 2.1885,
+      "step": 14290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.34158567383485e-05,
+      "loss": 2.2488,
+      "step": 14300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.341470880844876e-05,
+      "loss": 2.3358,
+      "step": 14310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.3413560878549e-05,
+      "loss": 2.1294,
+      "step": 14320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.341241294864927e-05,
+      "loss": 2.405,
+      "step": 14330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.341126501874952e-05,
+      "loss": 2.365,
+      "step": 14340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.341011708884977e-05,
+      "loss": 2.3698,
+      "step": 14350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.340896915895002e-05,
+      "loss": 2.3344,
+      "step": 14360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.340782122905027e-05,
+      "loss": 2.3121,
+      "step": 14370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.340667329915053e-05,
+      "loss": 2.3663,
+      "step": 14380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.340552536925078e-05,
+      "loss": 2.3921,
+      "step": 14390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.340437743935103e-05,
+      "loss": 2.1418,
+      "step": 14400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.340322950945128e-05,
+      "loss": 2.2467,
+      "step": 14410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.340208157955154e-05,
+      "loss": 2.2237,
+      "step": 14420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.340093364965178e-05,
+      "loss": 2.3937,
+      "step": 14430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.339978571975204e-05,
+      "loss": 2.216,
+      "step": 14440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.33986377898523e-05,
+      "loss": 2.1079,
+      "step": 14450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.339748985995254e-05,
+      "loss": 2.3949,
+      "step": 14460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.33963419300528e-05,
+      "loss": 2.3225,
+      "step": 14470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.339519400015306e-05,
+      "loss": 2.222,
+      "step": 14480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.339404607025331e-05,
+      "loss": 2.2862,
+      "step": 14490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.339289814035356e-05,
+      "loss": 2.1325,
+      "step": 14500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.339175021045381e-05,
+      "loss": 2.2182,
+      "step": 14510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.339060228055406e-05,
+      "loss": 2.3039,
+      "step": 14520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.338945435065432e-05,
+      "loss": 2.3359,
+      "step": 14530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.338830642075457e-05,
+      "loss": 2.2173,
+      "step": 14540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.338715849085482e-05,
+      "loss": 2.3169,
+      "step": 14550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.338601056095507e-05,
+      "loss": 2.4076,
+      "step": 14560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.338486263105533e-05,
+      "loss": 2.4774,
+      "step": 14570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.338371470115557e-05,
+      "loss": 2.1946,
+      "step": 14580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.338256677125583e-05,
+      "loss": 2.3241,
+      "step": 14590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.338141884135608e-05,
+      "loss": 2.2627,
+      "step": 14600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.338027091145633e-05,
+      "loss": 2.1874,
+      "step": 14610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.337912298155658e-05,
+      "loss": 2.3173,
+      "step": 14620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.337797505165685e-05,
+      "loss": 2.3252,
+      "step": 14630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.33768271217571e-05,
+      "loss": 2.2229,
+      "step": 14640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.337567919185735e-05,
+      "loss": 2.2875,
+      "step": 14650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.33745312619576e-05,
+      "loss": 2.4456,
+      "step": 14660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.337338333205785e-05,
+      "loss": 2.3431,
+      "step": 14670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.337223540215811e-05,
+      "loss": 2.2015,
+      "step": 14680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.337108747225836e-05,
+      "loss": 2.3863,
+      "step": 14690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.336993954235861e-05,
+      "loss": 2.3425,
+      "step": 14700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.336879161245886e-05,
+      "loss": 2.1134,
+      "step": 14710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.336764368255912e-05,
+      "loss": 2.2136,
+      "step": 14720
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.336649575265936e-05,
+      "loss": 2.3675,
+      "step": 14730
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.336534782275962e-05,
+      "loss": 2.3984,
+      "step": 14740
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.336419989285987e-05,
+      "loss": 2.2772,
+      "step": 14750
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.336305196296012e-05,
+      "loss": 2.245,
+      "step": 14760
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.336190403306037e-05,
+      "loss": 2.2958,
+      "step": 14770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.336075610316063e-05,
+      "loss": 2.3509,
+      "step": 14780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.335960817326088e-05,
+      "loss": 2.2847,
+      "step": 14790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.335846024336113e-05,
+      "loss": 2.2609,
+      "step": 14800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.335731231346138e-05,
+      "loss": 2.1676,
+      "step": 14810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.335616438356163e-05,
+      "loss": 2.2766,
+      "step": 14820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.33550164536619e-05,
+      "loss": 2.1233,
+      "step": 14830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.335386852376215e-05,
+      "loss": 2.3846,
+      "step": 14840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.33527205938624e-05,
+      "loss": 2.2972,
+      "step": 14850
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.335157266396265e-05,
+      "loss": 2.4486,
+      "step": 14860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.335042473406291e-05,
+      "loss": 2.1944,
+      "step": 14870
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.334927680416315e-05,
+      "loss": 2.4821,
+      "step": 14880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.334812887426341e-05,
+      "loss": 2.3942,
+      "step": 14890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.334698094436366e-05,
+      "loss": 2.1593,
+      "step": 14900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.334583301446391e-05,
+      "loss": 2.0722,
+      "step": 14910
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.334468508456416e-05,
+      "loss": 2.3226,
+      "step": 14920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.334353715466441e-05,
+      "loss": 2.4229,
+      "step": 14930
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.334238922476467e-05,
+      "loss": 2.3827,
+      "step": 14940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.334124129486492e-05,
+      "loss": 2.4316,
+      "step": 14950
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.334009336496517e-05,
+      "loss": 2.3625,
+      "step": 14960
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.333894543506542e-05,
+      "loss": 2.1722,
+      "step": 14970
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.333779750516569e-05,
+      "loss": 2.228,
+      "step": 14980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.333664957526592e-05,
+      "loss": 2.3808,
+      "step": 14990
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.333550164536619e-05,
+      "loss": 2.3626,
+      "step": 15000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.333435371546644e-05,
+      "loss": 2.3942,
+      "step": 15010
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.333320578556668e-05,
+      "loss": 2.1429,
+      "step": 15020
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.333205785566693e-05,
+      "loss": 2.1908,
+      "step": 15030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.33309099257672e-05,
+      "loss": 2.2315,
+      "step": 15040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.332976199586745e-05,
+      "loss": 2.2568,
+      "step": 15050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.33286140659677e-05,
+      "loss": 2.2699,
+      "step": 15060
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.332746613606795e-05,
+      "loss": 2.3914,
+      "step": 15070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.33263182061682e-05,
+      "loss": 2.3274,
+      "step": 15080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.332517027626846e-05,
+      "loss": 2.2163,
+      "step": 15090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.332402234636871e-05,
+      "loss": 2.2613,
+      "step": 15100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.332287441646896e-05,
+      "loss": 2.2476,
+      "step": 15110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.332172648656921e-05,
+      "loss": 2.1921,
+      "step": 15120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.332057855666947e-05,
+      "loss": 2.3215,
+      "step": 15130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.331943062676971e-05,
+      "loss": 2.2523,
+      "step": 15140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.331828269686997e-05,
+      "loss": 2.4381,
+      "step": 15150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.331713476697022e-05,
+      "loss": 2.4197,
+      "step": 15160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.331598683707047e-05,
+      "loss": 2.4066,
+      "step": 15170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.331483890717072e-05,
+      "loss": 2.2728,
+      "step": 15180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.331369097727099e-05,
+      "loss": 2.2645,
+      "step": 15190
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.331254304737124e-05,
+      "loss": 2.223,
+      "step": 15200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.331139511747149e-05,
+      "loss": 2.307,
+      "step": 15210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.331024718757174e-05,
+      "loss": 2.4303,
+      "step": 15220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.330909925767199e-05,
+      "loss": 2.319,
+      "step": 15230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.330795132777225e-05,
+      "loss": 2.2305,
+      "step": 15240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.33068033978725e-05,
+      "loss": 2.4156,
+      "step": 15250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.330565546797275e-05,
+      "loss": 2.1408,
+      "step": 15260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.3304507538073e-05,
+      "loss": 2.2634,
+      "step": 15270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.330335960817326e-05,
+      "loss": 2.279,
+      "step": 15280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.33022116782735e-05,
+      "loss": 2.3423,
+      "step": 15290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.330106374837376e-05,
+      "loss": 2.3457,
+      "step": 15300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.329991581847401e-05,
+      "loss": 2.3394,
+      "step": 15310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.329876788857426e-05,
+      "loss": 2.3248,
+      "step": 15320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.329761995867451e-05,
+      "loss": 2.2656,
+      "step": 15330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.329647202877478e-05,
+      "loss": 2.3563,
+      "step": 15340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.329532409887503e-05,
+      "loss": 2.2914,
+      "step": 15350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.329417616897527e-05,
+      "loss": 2.1784,
+      "step": 15360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.329302823907552e-05,
+      "loss": 2.4644,
+      "step": 15370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.329188030917577e-05,
+      "loss": 2.3576,
+      "step": 15380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.329073237927604e-05,
+      "loss": 2.2998,
+      "step": 15390
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.328958444937629e-05,
+      "loss": 2.2146,
+      "step": 15400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.328843651947654e-05,
+      "loss": 2.3585,
+      "step": 15410
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.328728858957679e-05,
+      "loss": 2.4503,
+      "step": 15420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.328614065967705e-05,
+      "loss": 2.2822,
+      "step": 15430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.328499272977729e-05,
+      "loss": 2.3026,
+      "step": 15440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.328384479987755e-05,
+      "loss": 2.4153,
+      "step": 15450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.32826968699778e-05,
+      "loss": 2.378,
+      "step": 15460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.328154894007805e-05,
+      "loss": 2.1821,
+      "step": 15470
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.32804010101783e-05,
+      "loss": 2.4306,
+      "step": 15480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.327925308027856e-05,
+      "loss": 2.3266,
+      "step": 15490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.327810515037881e-05,
+      "loss": 2.2769,
+      "step": 15500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.327695722047906e-05,
+      "loss": 2.1418,
+      "step": 15510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.327580929057931e-05,
+      "loss": 2.3107,
+      "step": 15520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.327466136067956e-05,
+      "loss": 2.2493,
+      "step": 15530
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.327351343077983e-05,
+      "loss": 2.0463,
+      "step": 15540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.327236550088008e-05,
+      "loss": 2.333,
+      "step": 15550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.327121757098033e-05,
+      "loss": 2.1686,
+      "step": 15560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.327006964108058e-05,
+      "loss": 2.2571,
+      "step": 15570
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.326892171118084e-05,
+      "loss": 2.4521,
+      "step": 15580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.326777378128108e-05,
+      "loss": 2.3552,
+      "step": 15590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.326662585138134e-05,
+      "loss": 2.3205,
+      "step": 15600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.326547792148159e-05,
+      "loss": 2.2821,
+      "step": 15610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.326432999158184e-05,
+      "loss": 2.3741,
+      "step": 15620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.326318206168209e-05,
+      "loss": 2.3859,
+      "step": 15630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.326203413178235e-05,
+      "loss": 2.148,
+      "step": 15640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.32608862018826e-05,
+      "loss": 2.3914,
+      "step": 15650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.325973827198285e-05,
+      "loss": 2.264,
+      "step": 15660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.32585903420831e-05,
+      "loss": 2.4421,
+      "step": 15670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.325744241218335e-05,
+      "loss": 2.1604,
+      "step": 15680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.325629448228361e-05,
+      "loss": 2.2579,
+      "step": 15690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.325514655238386e-05,
+      "loss": 2.284,
+      "step": 15700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.325399862248411e-05,
+      "loss": 2.2637,
+      "step": 15710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.325285069258436e-05,
+      "loss": 2.4107,
+      "step": 15720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.325170276268463e-05,
+      "loss": 2.2126,
+      "step": 15730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.325055483278486e-05,
+      "loss": 2.2364,
+      "step": 15740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.324940690288513e-05,
+      "loss": 2.412,
+      "step": 15750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.324825897298538e-05,
+      "loss": 2.2962,
+      "step": 15760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.324711104308563e-05,
+      "loss": 2.2913,
+      "step": 15770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.324596311318588e-05,
+      "loss": 2.2912,
+      "step": 15780
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.324481518328614e-05,
+      "loss": 2.2815,
+      "step": 15790
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.324366725338639e-05,
+      "loss": 2.2964,
+      "step": 15800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.324251932348664e-05,
+      "loss": 2.185,
+      "step": 15810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.324137139358689e-05,
+      "loss": 2.3656,
+      "step": 15820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.324022346368714e-05,
+      "loss": 2.4604,
+      "step": 15830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.32390755337874e-05,
+      "loss": 2.2057,
+      "step": 15840
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.323792760388765e-05,
+      "loss": 2.2621,
+      "step": 15850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.32367796739879e-05,
+      "loss": 2.306,
+      "step": 15860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.323563174408815e-05,
+      "loss": 2.3401,
+      "step": 15870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.323448381418842e-05,
+      "loss": 2.3248,
+      "step": 15880
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.323333588428865e-05,
+      "loss": 2.2819,
+      "step": 15890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.323218795438892e-05,
+      "loss": 2.295,
+      "step": 15900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.323104002448917e-05,
+      "loss": 2.163,
+      "step": 15910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.322989209458942e-05,
+      "loss": 2.2978,
+      "step": 15920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.322874416468967e-05,
+      "loss": 2.2107,
+      "step": 15930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.322759623478993e-05,
+      "loss": 2.1558,
+      "step": 15940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.322644830489018e-05,
+      "loss": 2.2587,
+      "step": 15950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.322530037499043e-05,
+      "loss": 2.3085,
+      "step": 15960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.322415244509068e-05,
+      "loss": 2.1927,
+      "step": 15970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.322300451519093e-05,
+      "loss": 2.2842,
+      "step": 15980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.322185658529119e-05,
+      "loss": 2.3327,
+      "step": 15990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.322070865539144e-05,
+      "loss": 2.2629,
+      "step": 16000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.321956072549169e-05,
+      "loss": 2.3612,
+      "step": 16010
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.321841279559194e-05,
+      "loss": 2.1775,
+      "step": 16020
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.32172648656922e-05,
+      "loss": 2.0319,
+      "step": 16030
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.321611693579244e-05,
+      "loss": 2.4107,
+      "step": 16040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.32149690058927e-05,
+      "loss": 2.3191,
+      "step": 16050
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.321382107599295e-05,
+      "loss": 2.2959,
+      "step": 16060
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.32126731460932e-05,
+      "loss": 2.2414,
+      "step": 16070
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.321152521619345e-05,
+      "loss": 2.3421,
+      "step": 16080
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.321037728629372e-05,
+      "loss": 2.3875,
+      "step": 16090
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.320922935639397e-05,
+      "loss": 2.2103,
+      "step": 16100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.320808142649422e-05,
+      "loss": 2.1911,
+      "step": 16110
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.320693349659447e-05,
+      "loss": 2.2895,
+      "step": 16120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.320578556669472e-05,
+      "loss": 2.2787,
+      "step": 16130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.320463763679498e-05,
+      "loss": 2.4706,
+      "step": 16140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.320348970689523e-05,
+      "loss": 2.4695,
+      "step": 16150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.320234177699548e-05,
+      "loss": 2.2866,
+      "step": 16160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.320119384709573e-05,
+      "loss": 2.39,
+      "step": 16170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.3200045917196e-05,
+      "loss": 2.2187,
+      "step": 16180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.319889798729623e-05,
+      "loss": 2.201,
+      "step": 16190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.319775005739649e-05,
+      "loss": 2.3729,
+      "step": 16200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.319660212749674e-05,
+      "loss": 2.3579,
+      "step": 16210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.319545419759699e-05,
+      "loss": 2.3777,
+      "step": 16220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.319430626769724e-05,
+      "loss": 2.1759,
+      "step": 16230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.31931583377975e-05,
+      "loss": 2.2291,
+      "step": 16240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.319201040789776e-05,
+      "loss": 2.2566,
+      "step": 16250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.3190862477998e-05,
+      "loss": 2.3888,
+      "step": 16260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.318971454809826e-05,
+      "loss": 2.2851,
+      "step": 16270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.31885666181985e-05,
+      "loss": 2.2656,
+      "step": 16280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.318741868829877e-05,
+      "loss": 2.3115,
+      "step": 16290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.3186270758399e-05,
+      "loss": 2.2484,
+      "step": 16300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.318512282849927e-05,
+      "loss": 2.3759,
+      "step": 16310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.318397489859952e-05,
+      "loss": 2.1175,
+      "step": 16320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.318282696869977e-05,
+      "loss": 2.4329,
+      "step": 16330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.318167903880002e-05,
+      "loss": 2.3511,
+      "step": 16340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.318053110890028e-05,
+      "loss": 2.3351,
+      "step": 16350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.317938317900053e-05,
+      "loss": 2.2861,
+      "step": 16360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.317823524910078e-05,
+      "loss": 2.2588,
+      "step": 16370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.317708731920103e-05,
+      "loss": 2.3293,
+      "step": 16380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.317593938930128e-05,
+      "loss": 2.3565,
+      "step": 16390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.317479145940154e-05,
+      "loss": 2.3083,
+      "step": 16400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.31736435295018e-05,
+      "loss": 2.1427,
+      "step": 16410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.317249559960204e-05,
+      "loss": 2.295,
+      "step": 16420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.31713476697023e-05,
+      "loss": 2.2885,
+      "step": 16430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.317019973980256e-05,
+      "loss": 2.3806,
+      "step": 16440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.31690518099028e-05,
+      "loss": 2.3592,
+      "step": 16450
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.316790388000306e-05,
+      "loss": 2.168,
+      "step": 16460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.316675595010331e-05,
+      "loss": 2.2131,
+      "step": 16470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.316560802020356e-05,
+      "loss": 2.2344,
+      "step": 16480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.31644600903038e-05,
+      "loss": 2.4098,
+      "step": 16490
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.316331216040407e-05,
+      "loss": 2.193,
+      "step": 16500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.316216423050432e-05,
+      "loss": 2.311,
+      "step": 16510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.316101630060457e-05,
+      "loss": 2.2645,
+      "step": 16520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.315986837070482e-05,
+      "loss": 2.4772,
+      "step": 16530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.315872044080507e-05,
+      "loss": 2.3807,
+      "step": 16540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.315757251090533e-05,
+      "loss": 2.3073,
+      "step": 16550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.315642458100558e-05,
+      "loss": 2.1269,
+      "step": 16560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.315527665110583e-05,
+      "loss": 2.2474,
+      "step": 16570
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.315412872120608e-05,
+      "loss": 2.1396,
+      "step": 16580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.315298079130635e-05,
+      "loss": 2.1524,
+      "step": 16590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.315183286140658e-05,
+      "loss": 2.2917,
+      "step": 16600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.315068493150685e-05,
+      "loss": 2.234,
+      "step": 16610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.31495370016071e-05,
+      "loss": 2.2011,
+      "step": 16620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.314838907170735e-05,
+      "loss": 2.1758,
+      "step": 16630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.31472411418076e-05,
+      "loss": 2.244,
+      "step": 16640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.314609321190786e-05,
+      "loss": 2.1724,
+      "step": 16650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.314494528200811e-05,
+      "loss": 2.2365,
+      "step": 16660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.314379735210836e-05,
+      "loss": 2.2324,
+      "step": 16670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.314264942220861e-05,
+      "loss": 2.0785,
+      "step": 16680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.314150149230886e-05,
+      "loss": 2.3523,
+      "step": 16690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.314035356240912e-05,
+      "loss": 2.1939,
+      "step": 16700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.313920563250937e-05,
+      "loss": 2.2547,
+      "step": 16710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.313805770260962e-05,
+      "loss": 2.2913,
+      "step": 16720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.313690977270987e-05,
+      "loss": 2.1337,
+      "step": 16730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.313576184281013e-05,
+      "loss": 2.3802,
+      "step": 16740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.313461391291037e-05,
+      "loss": 2.293,
+      "step": 16750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.313346598301063e-05,
+      "loss": 2.2514,
+      "step": 16760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.313231805311088e-05,
+      "loss": 2.2336,
+      "step": 16770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.313117012321113e-05,
+      "loss": 2.3622,
+      "step": 16780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.313002219331138e-05,
+      "loss": 2.2395,
+      "step": 16790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.312887426341165e-05,
+      "loss": 2.3468,
+      "step": 16800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.31277263335119e-05,
+      "loss": 2.1451,
+      "step": 16810
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.312657840361215e-05,
+      "loss": 2.1965,
+      "step": 16820
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.31254304737124e-05,
+      "loss": 2.3705,
+      "step": 16830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.312428254381265e-05,
+      "loss": 2.3266,
+      "step": 16840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.312313461391291e-05,
+      "loss": 2.1545,
+      "step": 16850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.312198668401316e-05,
+      "loss": 2.3367,
+      "step": 16860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.312083875411341e-05,
+      "loss": 2.3205,
+      "step": 16870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.311969082421366e-05,
+      "loss": 2.2351,
+      "step": 16880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.311854289431392e-05,
+      "loss": 2.416,
+      "step": 16890
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.311739496441416e-05,
+      "loss": 2.3016,
+      "step": 16900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.311624703451442e-05,
+      "loss": 2.3877,
+      "step": 16910
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.311509910461467e-05,
+      "loss": 2.4066,
+      "step": 16920
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.311395117471492e-05,
+      "loss": 2.2501,
+      "step": 16930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.311280324481517e-05,
+      "loss": 2.2941,
+      "step": 16940
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.311165531491544e-05,
+      "loss": 2.2306,
+      "step": 16950
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.311050738501569e-05,
+      "loss": 2.2855,
+      "step": 16960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.310935945511594e-05,
+      "loss": 2.2765,
+      "step": 16970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.310821152521618e-05,
+      "loss": 2.1754,
+      "step": 16980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.310706359531643e-05,
+      "loss": 2.1003,
+      "step": 16990
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.31059156654167e-05,
+      "loss": 2.2342,
+      "step": 17000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.310476773551695e-05,
+      "loss": 2.1303,
+      "step": 17010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.31036198056172e-05,
+      "loss": 2.1927,
+      "step": 17020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.310247187571745e-05,
+      "loss": 2.2265,
+      "step": 17030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.310132394581771e-05,
+      "loss": 2.2287,
+      "step": 17040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.310017601591795e-05,
+      "loss": 2.2932,
+      "step": 17050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.309902808601821e-05,
+      "loss": 2.1557,
+      "step": 17060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.309788015611846e-05,
+      "loss": 2.2273,
+      "step": 17070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.309673222621871e-05,
+      "loss": 2.2235,
+      "step": 17080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.309558429631896e-05,
+      "loss": 2.3022,
+      "step": 17090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.309443636641922e-05,
+      "loss": 2.1894,
+      "step": 17100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.309328843651947e-05,
+      "loss": 2.1746,
+      "step": 17110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.309214050661972e-05,
+      "loss": 2.2692,
+      "step": 17120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.309099257671997e-05,
+      "loss": 2.2649,
+      "step": 17130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.308984464682022e-05,
+      "loss": 2.1512,
+      "step": 17140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.308869671692049e-05,
+      "loss": 2.2452,
+      "step": 17150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.308754878702074e-05,
+      "loss": 2.2896,
+      "step": 17160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.308640085712099e-05,
+      "loss": 2.3671,
+      "step": 17170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.308525292722124e-05,
+      "loss": 2.2995,
+      "step": 17180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.30841049973215e-05,
+      "loss": 2.1578,
+      "step": 17190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.308295706742174e-05,
+      "loss": 2.2408,
+      "step": 17200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.3081809137522e-05,
+      "loss": 2.2052,
+      "step": 17210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.308066120762225e-05,
+      "loss": 2.3263,
+      "step": 17220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.30795132777225e-05,
+      "loss": 2.0939,
+      "step": 17230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.307836534782275e-05,
+      "loss": 2.2335,
+      "step": 17240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.307721741792301e-05,
+      "loss": 2.2544,
+      "step": 17250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.307606948802326e-05,
+      "loss": 2.2108,
+      "step": 17260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.307492155812351e-05,
+      "loss": 2.3519,
+      "step": 17270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.307377362822376e-05,
+      "loss": 2.0705,
+      "step": 17280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.307262569832401e-05,
+      "loss": 2.3298,
+      "step": 17290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.307147776842428e-05,
+      "loss": 2.2684,
+      "step": 17300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.307032983852453e-05,
+      "loss": 2.3119,
+      "step": 17310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.306918190862477e-05,
+      "loss": 2.1097,
+      "step": 17320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.306803397872502e-05,
+      "loss": 2.3098,
+      "step": 17330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.306688604882529e-05,
+      "loss": 2.2174,
+      "step": 17340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.306573811892552e-05,
+      "loss": 2.253,
+      "step": 17350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.306459018902579e-05,
+      "loss": 2.1467,
+      "step": 17360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.306344225912604e-05,
+      "loss": 2.2941,
+      "step": 17370
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.306229432922629e-05,
+      "loss": 2.1951,
+      "step": 17380
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.306114639932654e-05,
+      "loss": 2.1766,
+      "step": 17390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.30599984694268e-05,
+      "loss": 2.3114,
+      "step": 17400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.305885053952705e-05,
+      "loss": 2.1728,
+      "step": 17410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.30577026096273e-05,
+      "loss": 2.2314,
+      "step": 17420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.305655467972755e-05,
+      "loss": 2.3144,
+      "step": 17430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.30554067498278e-05,
+      "loss": 2.2649,
+      "step": 17440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.305425881992806e-05,
+      "loss": 2.3096,
+      "step": 17450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.305311089002831e-05,
+      "loss": 2.1801,
+      "step": 17460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.305196296012856e-05,
+      "loss": 2.3033,
+      "step": 17470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.305081503022881e-05,
+      "loss": 2.1239,
+      "step": 17480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.304966710032908e-05,
+      "loss": 2.0794,
+      "step": 17490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.304851917042931e-05,
+      "loss": 2.3641,
+      "step": 17500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.304737124052958e-05,
+      "loss": 2.276,
+      "step": 17510
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.304622331062983e-05,
+      "loss": 2.1028,
+      "step": 17520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.304507538073008e-05,
+      "loss": 2.1869,
+      "step": 17530
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.304392745083033e-05,
+      "loss": 2.256,
+      "step": 17540
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.304277952093059e-05,
+      "loss": 2.2127,
+      "step": 17550
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.304163159103084e-05,
+      "loss": 2.2148,
+      "step": 17560
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.304048366113109e-05,
+      "loss": 2.2831,
+      "step": 17570
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.303933573123134e-05,
+      "loss": 2.2642,
+      "step": 17580
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.303818780133159e-05,
+      "loss": 2.3169,
+      "step": 17590
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.303703987143185e-05,
+      "loss": 2.2904,
+      "step": 17600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.30358919415321e-05,
+      "loss": 2.0593,
+      "step": 17610
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.303474401163235e-05,
+      "loss": 2.2211,
+      "step": 17620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.30335960817326e-05,
+      "loss": 2.2007,
+      "step": 17630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.303244815183287e-05,
+      "loss": 2.1603,
+      "step": 17640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.30313002219331e-05,
+      "loss": 2.1813,
+      "step": 17650
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.303015229203336e-05,
+      "loss": 2.2312,
+      "step": 17660
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.302900436213361e-05,
+      "loss": 2.3556,
+      "step": 17670
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.302785643223386e-05,
+      "loss": 2.1816,
+      "step": 17680
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.302670850233411e-05,
+      "loss": 2.2071,
+      "step": 17690
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.302556057243436e-05,
+      "loss": 2.1292,
+      "step": 17700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.302441264253463e-05,
+      "loss": 2.2156,
+      "step": 17710
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.302326471263488e-05,
+      "loss": 2.1674,
+      "step": 17720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.302211678273513e-05,
+      "loss": 2.3084,
+      "step": 17730
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.302096885283538e-05,
+      "loss": 2.3207,
+      "step": 17740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.301982092293564e-05,
+      "loss": 2.1942,
+      "step": 17750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.301867299303588e-05,
+      "loss": 2.2554,
+      "step": 17760
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.301752506313614e-05,
+      "loss": 2.1281,
+      "step": 17770
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.301637713323639e-05,
+      "loss": 2.2325,
+      "step": 17780
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.301522920333664e-05,
+      "loss": 2.2771,
+      "step": 17790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.301408127343689e-05,
+      "loss": 2.2087,
+      "step": 17800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.301293334353715e-05,
+      "loss": 2.2251,
+      "step": 17810
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.30117854136374e-05,
+      "loss": 2.1282,
+      "step": 17820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.301063748373765e-05,
+      "loss": 2.2933,
+      "step": 17830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.30094895538379e-05,
+      "loss": 2.3277,
+      "step": 17840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.300834162393815e-05,
+      "loss": 2.281,
+      "step": 17850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.300719369403842e-05,
+      "loss": 2.0847,
+      "step": 17860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.300604576413867e-05,
+      "loss": 2.2264,
+      "step": 17870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.300489783423892e-05,
+      "loss": 2.1888,
+      "step": 17880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.300374990433917e-05,
+      "loss": 2.3695,
+      "step": 17890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.300260197443943e-05,
+      "loss": 2.0931,
+      "step": 17900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.300145404453967e-05,
+      "loss": 2.2363,
+      "step": 17910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.300030611463993e-05,
+      "loss": 2.2881,
+      "step": 17920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.299915818474018e-05,
+      "loss": 2.0732,
+      "step": 17930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.299801025484043e-05,
+      "loss": 2.1447,
+      "step": 17940
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.299686232494068e-05,
+      "loss": 2.2634,
+      "step": 17950
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.299571439504094e-05,
+      "loss": 2.2007,
+      "step": 17960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.299456646514119e-05,
+      "loss": 2.2903,
+      "step": 17970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.299341853524144e-05,
+      "loss": 2.1614,
+      "step": 17980
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.299227060534169e-05,
+      "loss": 2.0166,
+      "step": 17990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.299112267544194e-05,
+      "loss": 2.2268,
+      "step": 18000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.29899747455422e-05,
+      "loss": 2.1942,
+      "step": 18010
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.298882681564245e-05,
+      "loss": 2.1718,
+      "step": 18020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.29876788857427e-05,
+      "loss": 2.3629,
+      "step": 18030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.298653095584295e-05,
+      "loss": 2.2589,
+      "step": 18040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.298538302594322e-05,
+      "loss": 2.2103,
+      "step": 18050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.298423509604345e-05,
+      "loss": 2.3398,
+      "step": 18060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.298308716614372e-05,
+      "loss": 2.244,
+      "step": 18070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.298193923624397e-05,
+      "loss": 2.0738,
+      "step": 18080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.298079130634422e-05,
+      "loss": 2.1828,
+      "step": 18090
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.297964337644447e-05,
+      "loss": 2.1346,
+      "step": 18100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.297849544654473e-05,
+      "loss": 2.2229,
+      "step": 18110
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.297734751664498e-05,
+      "loss": 2.2092,
+      "step": 18120
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.297619958674523e-05,
+      "loss": 2.1871,
+      "step": 18130
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.297505165684548e-05,
+      "loss": 2.2733,
+      "step": 18140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.297390372694573e-05,
+      "loss": 2.0636,
+      "step": 18150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.297275579704599e-05,
+      "loss": 2.3042,
+      "step": 18160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.297160786714624e-05,
+      "loss": 2.2913,
+      "step": 18170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.297045993724649e-05,
+      "loss": 2.0483,
+      "step": 18180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.296931200734674e-05,
+      "loss": 2.2505,
+      "step": 18190
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.2968164077447e-05,
+      "loss": 2.2083,
+      "step": 18200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.296701614754724e-05,
+      "loss": 2.1107,
+      "step": 18210
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.29658682176475e-05,
+      "loss": 1.968,
+      "step": 18220
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.296472028774776e-05,
+      "loss": 2.1112,
+      "step": 18230
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.2963572357848e-05,
+      "loss": 2.3513,
+      "step": 18240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.296242442794826e-05,
+      "loss": 2.1443,
+      "step": 18250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.296127649804852e-05,
+      "loss": 2.1794,
+      "step": 18260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.296012856814877e-05,
+      "loss": 2.2204,
+      "step": 18270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.295898063824902e-05,
+      "loss": 2.0922,
+      "step": 18280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.295783270834927e-05,
+      "loss": 2.2849,
+      "step": 18290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.295668477844952e-05,
+      "loss": 2.1106,
+      "step": 18300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.295553684854978e-05,
+      "loss": 2.1416,
+      "step": 18310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.295438891865003e-05,
+      "loss": 2.2271,
+      "step": 18320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.295324098875028e-05,
+      "loss": 2.1659,
+      "step": 18330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.295209305885053e-05,
+      "loss": 2.2435,
+      "step": 18340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.29509451289508e-05,
+      "loss": 2.0014,
+      "step": 18350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.294979719905103e-05,
+      "loss": 2.2483,
+      "step": 18360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.29486492691513e-05,
+      "loss": 2.14,
+      "step": 18370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.294750133925154e-05,
+      "loss": 2.0989,
+      "step": 18380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.29463534093518e-05,
+      "loss": 2.2758,
+      "step": 18390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.294520547945204e-05,
+      "loss": 2.1909,
+      "step": 18400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.294405754955231e-05,
+      "loss": 2.0855,
+      "step": 18410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.294290961965256e-05,
+      "loss": 2.239,
+      "step": 18420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.294176168975281e-05,
+      "loss": 2.3609,
+      "step": 18430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.294061375985306e-05,
+      "loss": 2.116,
+      "step": 18440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.29394658299533e-05,
+      "loss": 2.1181,
+      "step": 18450
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.293831790005357e-05,
+      "loss": 2.1413,
+      "step": 18460
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.293716997015382e-05,
+      "loss": 2.3332,
+      "step": 18470
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.293602204025407e-05,
+      "loss": 2.4078,
+      "step": 18480
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.293487411035432e-05,
+      "loss": 2.118,
+      "step": 18490
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.293372618045458e-05,
+      "loss": 2.0906,
+      "step": 18500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.293257825055482e-05,
+      "loss": 2.1516,
+      "step": 18510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.293143032065508e-05,
+      "loss": 2.2682,
+      "step": 18520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.293028239075533e-05,
+      "loss": 2.2977,
+      "step": 18530
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.292913446085558e-05,
+      "loss": 2.2,
+      "step": 18540
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.292798653095583e-05,
+      "loss": 2.0786,
+      "step": 18550
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.29268386010561e-05,
+      "loss": 2.0866,
+      "step": 18560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.292569067115635e-05,
+      "loss": 2.1532,
+      "step": 18570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.29245427412566e-05,
+      "loss": 2.19,
+      "step": 18580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.292339481135685e-05,
+      "loss": 2.1441,
+      "step": 18590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.29222468814571e-05,
+      "loss": 2.1753,
+      "step": 18600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.292109895155736e-05,
+      "loss": 2.1426,
+      "step": 18610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.291995102165761e-05,
+      "loss": 2.0291,
+      "step": 18620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.291880309175786e-05,
+      "loss": 2.2777,
+      "step": 18630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.291765516185811e-05,
+      "loss": 2.1502,
+      "step": 18640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.291650723195837e-05,
+      "loss": 2.126,
+      "step": 18650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.291535930205861e-05,
+      "loss": 2.1571,
+      "step": 18660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.291421137215887e-05,
+      "loss": 2.1501,
+      "step": 18670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.291306344225912e-05,
+      "loss": 2.159,
+      "step": 18680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.291191551235937e-05,
+      "loss": 2.1959,
+      "step": 18690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.291076758245962e-05,
+      "loss": 2.046,
+      "step": 18700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.290961965255988e-05,
+      "loss": 2.1886,
+      "step": 18710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.290847172266013e-05,
+      "loss": 2.0992,
+      "step": 18720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.290732379276038e-05,
+      "loss": 1.9705,
+      "step": 18730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.290617586286063e-05,
+      "loss": 2.0625,
+      "step": 18740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.290502793296088e-05,
+      "loss": 2.1177,
+      "step": 18750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.290388000306115e-05,
+      "loss": 2.0367,
+      "step": 18760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.29027320731614e-05,
+      "loss": 2.1612,
+      "step": 18770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.290158414326165e-05,
+      "loss": 2.1699,
+      "step": 18780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.29004362133619e-05,
+      "loss": 2.2608,
+      "step": 18790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.289928828346216e-05,
+      "loss": 2.1199,
+      "step": 18800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.28981403535624e-05,
+      "loss": 2.0404,
+      "step": 18810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.289699242366266e-05,
+      "loss": 2.2272,
+      "step": 18820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.289584449376291e-05,
+      "loss": 2.1413,
+      "step": 18830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.289469656386316e-05,
+      "loss": 2.1551,
+      "step": 18840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.289354863396341e-05,
+      "loss": 1.9775,
+      "step": 18850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.289240070406367e-05,
+      "loss": 2.0251,
+      "step": 18860
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.289125277416392e-05,
+      "loss": 2.147,
+      "step": 18870
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.289010484426417e-05,
+      "loss": 2.0837,
+      "step": 18880
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.288895691436442e-05,
+      "loss": 2.0744,
+      "step": 18890
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.288780898446467e-05,
+      "loss": 1.9858,
+      "step": 18900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.288666105456494e-05,
+      "loss": 1.9599,
+      "step": 18910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.288551312466519e-05,
+      "loss": 1.9183,
+      "step": 18920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.288436519476544e-05,
+      "loss": 2.0502,
+      "step": 18930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.288321726486568e-05,
+      "loss": 2.102,
+      "step": 18940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.288206933496595e-05,
+      "loss": 1.9674,
+      "step": 18950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.288092140506618e-05,
+      "loss": 2.0162,
+      "step": 18960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.287977347516645e-05,
+      "loss": 2.1112,
+      "step": 18970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.28786255452667e-05,
+      "loss": 2.1593,
+      "step": 18980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.287747761536695e-05,
+      "loss": 2.1593,
+      "step": 18990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.28763296854672e-05,
+      "loss": 2.2071,
+      "step": 19000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.287518175556746e-05,
+      "loss": 2.1513,
+      "step": 19010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.287403382566771e-05,
+      "loss": 2.1671,
+      "step": 19020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.287288589576796e-05,
+      "loss": 2.158,
+      "step": 19030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.287173796586821e-05,
+      "loss": 2.1322,
+      "step": 19040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.287059003596846e-05,
+      "loss": 2.0793,
+      "step": 19050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.286944210606872e-05,
+      "loss": 2.091,
+      "step": 19060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.286829417616896e-05,
+      "loss": 2.1837,
+      "step": 19070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.286714624626922e-05,
+      "loss": 2.1276,
+      "step": 19080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.286599831636947e-05,
+      "loss": 2.1193,
+      "step": 19090
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.286485038646972e-05,
+      "loss": 2.2873,
+      "step": 19100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.286370245656997e-05,
+      "loss": 2.1961,
+      "step": 19110
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.286255452667024e-05,
+      "loss": 2.164,
+      "step": 19120
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.286140659677049e-05,
+      "loss": 2.0963,
+      "step": 19130
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.286025866687074e-05,
+      "loss": 2.1949,
+      "step": 19140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.285911073697099e-05,
+      "loss": 2.1534,
+      "step": 19150
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.285796280707124e-05,
+      "loss": 2.1291,
+      "step": 19160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.28568148771715e-05,
+      "loss": 2.0831,
+      "step": 19170
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.285566694727175e-05,
+      "loss": 2.0838,
+      "step": 19180
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.2854519017372e-05,
+      "loss": 2.0517,
+      "step": 19190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.285337108747225e-05,
+      "loss": 2.0558,
+      "step": 19200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.285222315757251e-05,
+      "loss": 2.1291,
+      "step": 19210
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.285107522767275e-05,
+      "loss": 2.0521,
+      "step": 19220
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.284992729777301e-05,
+      "loss": 1.9651,
+      "step": 19230
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.284877936787326e-05,
+      "loss": 1.992,
+      "step": 19240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.284763143797351e-05,
+      "loss": 1.937,
+      "step": 19250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.284648350807376e-05,
+      "loss": 1.9733,
+      "step": 19260
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.284533557817403e-05,
+      "loss": 1.9941,
+      "step": 19270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.284418764827427e-05,
+      "loss": 2.0672,
+      "step": 19280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.284303971837452e-05,
+      "loss": 1.9313,
+      "step": 19290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.284189178847477e-05,
+      "loss": 2.1625,
+      "step": 19300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.284074385857502e-05,
+      "loss": 1.9404,
+      "step": 19310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.283959592867529e-05,
+      "loss": 2.1855,
+      "step": 19320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.283844799877554e-05,
+      "loss": 2.1676,
+      "step": 19330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.283730006887579e-05,
+      "loss": 2.09,
+      "step": 19340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.283615213897604e-05,
+      "loss": 2.147,
+      "step": 19350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.28350042090763e-05,
+      "loss": 2.1848,
+      "step": 19360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.283385627917654e-05,
+      "loss": 2.0473,
+      "step": 19370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.28327083492768e-05,
+      "loss": 1.9712,
+      "step": 19380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.283156041937705e-05,
+      "loss": 1.8138,
+      "step": 19390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.28304124894773e-05,
+      "loss": 2.2014,
+      "step": 19400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.282926455957755e-05,
+      "loss": 2.1959,
+      "step": 19410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.282811662967781e-05,
+      "loss": 2.0354,
+      "step": 19420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.282696869977806e-05,
+      "loss": 2.2209,
+      "step": 19430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.282582076987831e-05,
+      "loss": 2.2447,
+      "step": 19440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.282467283997856e-05,
+      "loss": 2.0198,
+      "step": 19450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.282352491007881e-05,
+      "loss": 2.2642,
+      "step": 19460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.282237698017908e-05,
+      "loss": 2.1222,
+      "step": 19470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.282122905027933e-05,
+      "loss": 2.203,
+      "step": 19480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.282008112037958e-05,
+      "loss": 1.9619,
+      "step": 19490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.281893319047983e-05,
+      "loss": 2.1362,
+      "step": 19500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.281778526058009e-05,
+      "loss": 2.1801,
+      "step": 19510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.281663733068033e-05,
+      "loss": 1.9375,
+      "step": 19520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.281548940078059e-05,
+      "loss": 2.2308,
+      "step": 19530
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.281434147088084e-05,
+      "loss": 2.0549,
+      "step": 19540
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.281319354098109e-05,
+      "loss": 2.0557,
+      "step": 19550
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.281204561108134e-05,
+      "loss": 2.0398,
+      "step": 19560
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.28108976811816e-05,
+      "loss": 2.0712,
+      "step": 19570
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.280974975128185e-05,
+      "loss": 2.1118,
+      "step": 19580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.28086018213821e-05,
+      "loss": 1.9344,
+      "step": 19590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.280745389148235e-05,
+      "loss": 2.2428,
+      "step": 19600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.28063059615826e-05,
+      "loss": 1.9533,
+      "step": 19610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.280515803168286e-05,
+      "loss": 2.0087,
+      "step": 19620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.280401010178311e-05,
+      "loss": 2.0341,
+      "step": 19630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.280286217188336e-05,
+      "loss": 1.9284,
+      "step": 19640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.280171424198361e-05,
+      "loss": 2.0419,
+      "step": 19650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.280056631208388e-05,
+      "loss": 2.1146,
+      "step": 19660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.279941838218411e-05,
+      "loss": 2.145,
+      "step": 19670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.279827045228438e-05,
+      "loss": 2.0671,
+      "step": 19680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.279712252238463e-05,
+      "loss": 2.0572,
+      "step": 19690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.279597459248488e-05,
+      "loss": 2.1427,
+      "step": 19700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.279482666258513e-05,
+      "loss": 1.8885,
+      "step": 19710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.279367873268539e-05,
+      "loss": 1.9874,
+      "step": 19720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.279253080278564e-05,
+      "loss": 2.1012,
+      "step": 19730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.279138287288589e-05,
+      "loss": 2.0682,
+      "step": 19740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.279023494298614e-05,
+      "loss": 2.1275,
+      "step": 19750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.278908701308639e-05,
+      "loss": 1.9585,
+      "step": 19760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.278793908318665e-05,
+      "loss": 1.9213,
+      "step": 19770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.27867911532869e-05,
+      "loss": 2.1952,
+      "step": 19780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.278564322338715e-05,
+      "loss": 2.0494,
+      "step": 19790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.27844952934874e-05,
+      "loss": 2.1636,
+      "step": 19800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.278334736358767e-05,
+      "loss": 2.0222,
+      "step": 19810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.27821994336879e-05,
+      "loss": 2.0593,
+      "step": 19820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.278105150378817e-05,
+      "loss": 2.0985,
+      "step": 19830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.277990357388842e-05,
+      "loss": 2.1555,
+      "step": 19840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.277875564398867e-05,
+      "loss": 2.0434,
+      "step": 19850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.277760771408892e-05,
+      "loss": 1.9868,
+      "step": 19860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.277645978418918e-05,
+      "loss": 2.0601,
+      "step": 19870
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.277531185428943e-05,
+      "loss": 2.1747,
+      "step": 19880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.277416392438968e-05,
+      "loss": 2.2033,
+      "step": 19890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.277301599448993e-05,
+      "loss": 1.968,
+      "step": 19900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.277186806459018e-05,
+      "loss": 1.884,
+      "step": 19910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.277072013469044e-05,
+      "loss": 1.9946,
+      "step": 19920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.276957220479069e-05,
+      "loss": 1.9072,
+      "step": 19930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.276842427489094e-05,
+      "loss": 1.9486,
+      "step": 19940
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.276727634499119e-05,
+      "loss": 2.1695,
+      "step": 19950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.276612841509145e-05,
+      "loss": 1.9067,
+      "step": 19960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.276498048519169e-05,
+      "loss": 2.1862,
+      "step": 19970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.276383255529195e-05,
+      "loss": 2.2222,
+      "step": 19980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.27626846253922e-05,
+      "loss": 2.0742,
+      "step": 19990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.276153669549245e-05,
+      "loss": 2.0782,
+      "step": 20000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.27603887655927e-05,
+      "loss": 2.0586,
+      "step": 20010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.275924083569297e-05,
+      "loss": 2.0704,
+      "step": 20020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.275809290579322e-05,
+      "loss": 1.8974,
+      "step": 20030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.275694497589347e-05,
+      "loss": 1.7789,
+      "step": 20040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.275579704599372e-05,
+      "loss": 2.0329,
+      "step": 20050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.275464911609397e-05,
+      "loss": 2.0295,
+      "step": 20060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.275350118619423e-05,
+      "loss": 2.0021,
+      "step": 20070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.275235325629448e-05,
+      "loss": 2.0095,
+      "step": 20080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.275120532639473e-05,
+      "loss": 2.0138,
+      "step": 20090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.275005739649498e-05,
+      "loss": 2.0924,
+      "step": 20100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.274890946659524e-05,
+      "loss": 2.1784,
+      "step": 20110
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.274776153669548e-05,
+      "loss": 2.1769,
+      "step": 20120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.274661360679574e-05,
+      "loss": 1.9557,
+      "step": 20130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.274546567689599e-05,
+      "loss": 2.0405,
+      "step": 20140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.274431774699624e-05,
+      "loss": 1.9609,
+      "step": 20150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.274316981709649e-05,
+      "loss": 1.9207,
+      "step": 20160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.274202188719676e-05,
+      "loss": 1.9699,
+      "step": 20170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.2740873957297e-05,
+      "loss": 1.8862,
+      "step": 20180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.273972602739726e-05,
+      "loss": 2.2062,
+      "step": 20190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.27385780974975e-05,
+      "loss": 1.9379,
+      "step": 20200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.273743016759776e-05,
+      "loss": 2.0908,
+      "step": 20210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.273628223769802e-05,
+      "loss": 1.9133,
+      "step": 20220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.273513430779827e-05,
+      "loss": 2.0852,
+      "step": 20230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.273398637789852e-05,
+      "loss": 2.157,
+      "step": 20240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.273283844799877e-05,
+      "loss": 1.912,
+      "step": 20250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.273169051809903e-05,
+      "loss": 1.9344,
+      "step": 20260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.273054258819927e-05,
+      "loss": 2.0602,
+      "step": 20270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.272939465829953e-05,
+      "loss": 2.0266,
+      "step": 20280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.272824672839978e-05,
+      "loss": 1.9892,
+      "step": 20290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.272709879850003e-05,
+      "loss": 2.167,
+      "step": 20300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.272595086860028e-05,
+      "loss": 1.987,
+      "step": 20310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.272480293870054e-05,
+      "loss": 2.0503,
+      "step": 20320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.27236550088008e-05,
+      "loss": 1.8848,
+      "step": 20330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.272250707890104e-05,
+      "loss": 1.9808,
+      "step": 20340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.27213591490013e-05,
+      "loss": 2.0348,
+      "step": 20350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.272021121910154e-05,
+      "loss": 1.9616,
+      "step": 20360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.271906328920181e-05,
+      "loss": 2.0238,
+      "step": 20370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.271791535930206e-05,
+      "loss": 2.0097,
+      "step": 20380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.271676742940231e-05,
+      "loss": 2.2454,
+      "step": 20390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.271561949950256e-05,
+      "loss": 1.9998,
+      "step": 20400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.27144715696028e-05,
+      "loss": 2.0652,
+      "step": 20410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.271332363970306e-05,
+      "loss": 1.8432,
+      "step": 20420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.271217570980332e-05,
+      "loss": 2.1335,
+      "step": 20430
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.271102777990357e-05,
+      "loss": 1.9994,
+      "step": 20440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.270987985000382e-05,
+      "loss": 1.9815,
+      "step": 20450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.270873192010407e-05,
+      "loss": 1.8999,
+      "step": 20460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.270758399020432e-05,
+      "loss": 1.9092,
+      "step": 20470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.270643606030458e-05,
+      "loss": 2.0146,
+      "step": 20480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.270528813040483e-05,
+      "loss": 2.1275,
+      "step": 20490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.270414020050508e-05,
+      "loss": 1.9058,
+      "step": 20500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.270299227060533e-05,
+      "loss": 2.0214,
+      "step": 20510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.27018443407056e-05,
+      "loss": 2.06,
+      "step": 20520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.270069641080583e-05,
+      "loss": 1.9198,
+      "step": 20530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.26995484809061e-05,
+      "loss": 2.1717,
+      "step": 20540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.269840055100635e-05,
+      "loss": 2.0217,
+      "step": 20550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.26972526211066e-05,
+      "loss": 2.0118,
+      "step": 20560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.269610469120684e-05,
+      "loss": 1.9746,
+      "step": 20570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.269495676130711e-05,
+      "loss": 2.0362,
+      "step": 20580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.269380883140736e-05,
+      "loss": 1.9062,
+      "step": 20590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.269266090150761e-05,
+      "loss": 2.1264,
+      "step": 20600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.269151297160786e-05,
+      "loss": 2.0932,
+      "step": 20610
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.269036504170811e-05,
+      "loss": 1.9109,
+      "step": 20620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.268921711180837e-05,
+      "loss": 2.1419,
+      "step": 20630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.268806918190862e-05,
+      "loss": 2.0591,
+      "step": 20640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.268692125200887e-05,
+      "loss": 1.9284,
+      "step": 20650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.268577332210912e-05,
+      "loss": 2.0693,
+      "step": 20660
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.268462539220938e-05,
+      "loss": 1.9234,
+      "step": 20670
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.268347746230962e-05,
+      "loss": 2.0563,
+      "step": 20680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.268232953240988e-05,
+      "loss": 1.9556,
+      "step": 20690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.268118160251013e-05,
+      "loss": 2.1402,
+      "step": 20700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.268003367261038e-05,
+      "loss": 2.1079,
+      "step": 20710
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.267888574271063e-05,
+      "loss": 2.1315,
+      "step": 20720
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.26777378128109e-05,
+      "loss": 2.012,
+      "step": 20730
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.267658988291115e-05,
+      "loss": 2.0212,
+      "step": 20740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.26754419530114e-05,
+      "loss": 1.9184,
+      "step": 20750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.267429402311165e-05,
+      "loss": 1.8423,
+      "step": 20760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.26731460932119e-05,
+      "loss": 2.0014,
+      "step": 20770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.267199816331216e-05,
+      "loss": 2.1116,
+      "step": 20780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.267085023341241e-05,
+      "loss": 1.9149,
+      "step": 20790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.266970230351266e-05,
+      "loss": 2.0819,
+      "step": 20800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.266855437361291e-05,
+      "loss": 1.9222,
+      "step": 20810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.266740644371317e-05,
+      "loss": 1.8166,
+      "step": 20820
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.266625851381341e-05,
+      "loss": 2.0738,
+      "step": 20830
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.266511058391367e-05,
+      "loss": 2.0254,
+      "step": 20840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.266396265401392e-05,
+      "loss": 1.8892,
+      "step": 20850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.266281472411417e-05,
+      "loss": 2.0728,
+      "step": 20860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.266166679421442e-05,
+      "loss": 1.9574,
+      "step": 20870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.266051886431469e-05,
+      "loss": 2.0463,
+      "step": 20880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.265937093441494e-05,
+      "loss": 1.964,
+      "step": 20890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.265822300451518e-05,
+      "loss": 1.9986,
+      "step": 20900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.265707507461543e-05,
+      "loss": 2.1005,
+      "step": 20910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.265592714471568e-05,
+      "loss": 1.9026,
+      "step": 20920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.265477921481595e-05,
+      "loss": 2.1384,
+      "step": 20930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.26536312849162e-05,
+      "loss": 2.0379,
+      "step": 20940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.265248335501645e-05,
+      "loss": 2.1261,
+      "step": 20950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.26513354251167e-05,
+      "loss": 1.9934,
+      "step": 20960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.265018749521696e-05,
+      "loss": 1.8071,
+      "step": 20970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.26490395653172e-05,
+      "loss": 1.9474,
+      "step": 20980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.264789163541746e-05,
+      "loss": 2.045,
+      "step": 20990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.264674370551771e-05,
+      "loss": 1.9087,
+      "step": 21000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.264559577561796e-05,
+      "loss": 2.0427,
+      "step": 21010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.264444784571821e-05,
+      "loss": 1.9819,
+      "step": 21020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.264329991581847e-05,
+      "loss": 1.8572,
+      "step": 21030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.264215198591872e-05,
+      "loss": 2.184,
+      "step": 21040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.264100405601897e-05,
+      "loss": 1.9601,
+      "step": 21050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.263985612611922e-05,
+      "loss": 1.8872,
+      "step": 21060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.263870819621947e-05,
+      "loss": 1.9464,
+      "step": 21070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.263756026631974e-05,
+      "loss": 1.9701,
+      "step": 21080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.263641233641999e-05,
+      "loss": 1.873,
+      "step": 21090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.263526440652024e-05,
+      "loss": 1.9575,
+      "step": 21100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.263411647662049e-05,
+      "loss": 1.9833,
+      "step": 21110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.263296854672075e-05,
+      "loss": 1.7896,
+      "step": 21120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.263182061682099e-05,
+      "loss": 1.9968,
+      "step": 21130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.263067268692125e-05,
+      "loss": 1.9937,
+      "step": 21140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.26295247570215e-05,
+      "loss": 1.9328,
+      "step": 21150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.262837682712175e-05,
+      "loss": 1.8427,
+      "step": 21160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.2627228897222e-05,
+      "loss": 1.8947,
+      "step": 21170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.262608096732226e-05,
+      "loss": 1.9286,
+      "step": 21180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.262493303742251e-05,
+      "loss": 1.7842,
+      "step": 21190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.262378510752276e-05,
+      "loss": 2.0403,
+      "step": 21200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.262263717762301e-05,
+      "loss": 1.9647,
+      "step": 21210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.262148924772326e-05,
+      "loss": 2.0729,
+      "step": 21220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.262034131782353e-05,
+      "loss": 1.9207,
+      "step": 21230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.261919338792377e-05,
+      "loss": 1.9829,
+      "step": 21240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.261804545802402e-05,
+      "loss": 1.9564,
+      "step": 21250
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.261689752812427e-05,
+      "loss": 1.8565,
+      "step": 21260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.261574959822454e-05,
+      "loss": 1.9227,
+      "step": 21270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.261460166832477e-05,
+      "loss": 1.9451,
+      "step": 21280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.261345373842504e-05,
+      "loss": 1.8237,
+      "step": 21290
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.261230580852529e-05,
+      "loss": 1.9551,
+      "step": 21300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.261115787862554e-05,
+      "loss": 1.8844,
+      "step": 21310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.261000994872579e-05,
+      "loss": 1.8625,
+      "step": 21320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.260886201882605e-05,
+      "loss": 2.1321,
+      "step": 21330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.26077140889263e-05,
+      "loss": 1.967,
+      "step": 21340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.260656615902655e-05,
+      "loss": 1.9922,
+      "step": 21350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.26054182291268e-05,
+      "loss": 1.8799,
+      "step": 21360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.260427029922705e-05,
+      "loss": 2.0097,
+      "step": 21370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.260312236932731e-05,
+      "loss": 1.7675,
+      "step": 21380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.260197443942756e-05,
+      "loss": 1.8943,
+      "step": 21390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.260082650952781e-05,
+      "loss": 1.8381,
+      "step": 21400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.259967857962806e-05,
+      "loss": 2.0144,
+      "step": 21410
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.259853064972833e-05,
+      "loss": 2.1263,
+      "step": 21420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.259738271982856e-05,
+      "loss": 1.8786,
+      "step": 21430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.259623478992883e-05,
+      "loss": 1.8522,
+      "step": 21440
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.259508686002908e-05,
+      "loss": 1.7335,
+      "step": 21450
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.259393893012933e-05,
+      "loss": 2.0143,
+      "step": 21460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.259279100022958e-05,
+      "loss": 1.9854,
+      "step": 21470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.259164307032984e-05,
+      "loss": 1.8808,
+      "step": 21480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.259049514043009e-05,
+      "loss": 1.9026,
+      "step": 21490
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.258934721053034e-05,
+      "loss": 1.9204,
+      "step": 21500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.258819928063059e-05,
+      "loss": 2.1099,
+      "step": 21510
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.258705135073084e-05,
+      "loss": 1.8401,
+      "step": 21520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.25859034208311e-05,
+      "loss": 1.8062,
+      "step": 21530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.258475549093135e-05,
+      "loss": 1.9742,
+      "step": 21540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.25836075610316e-05,
+      "loss": 1.8376,
+      "step": 21550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.258245963113185e-05,
+      "loss": 2.2692,
+      "step": 21560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.258131170123211e-05,
+      "loss": 1.8975,
+      "step": 21570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.258016377133235e-05,
+      "loss": 1.9999,
+      "step": 21580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.257901584143261e-05,
+      "loss": 2.0603,
+      "step": 21590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.257786791153286e-05,
+      "loss": 2.0348,
+      "step": 21600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.257671998163311e-05,
+      "loss": 1.9044,
+      "step": 21610
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.257557205173336e-05,
+      "loss": 1.8118,
+      "step": 21620
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.257442412183363e-05,
+      "loss": 1.896,
+      "step": 21630
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.257327619193388e-05,
+      "loss": 2.0814,
+      "step": 21640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.257212826203413e-05,
+      "loss": 2.0197,
+      "step": 21650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.257098033213438e-05,
+      "loss": 1.9091,
+      "step": 21660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.256983240223463e-05,
+      "loss": 1.8149,
+      "step": 21670
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.256868447233489e-05,
+      "loss": 1.8914,
+      "step": 21680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.256753654243514e-05,
+      "loss": 1.944,
+      "step": 21690
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.256638861253539e-05,
+      "loss": 1.8263,
+      "step": 21700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.256524068263564e-05,
+      "loss": 1.923,
+      "step": 21710
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.25640927527359e-05,
+      "loss": 1.9711,
+      "step": 21720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.256294482283614e-05,
+      "loss": 1.8502,
+      "step": 21730
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.25617968929364e-05,
+      "loss": 2.046,
+      "step": 21740
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.256064896303665e-05,
+      "loss": 1.9151,
+      "step": 21750
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.25595010331369e-05,
+      "loss": 1.9037,
+      "step": 21760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.255835310323715e-05,
+      "loss": 1.8943,
+      "step": 21770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.25572051733374e-05,
+      "loss": 1.976,
+      "step": 21780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.255605724343767e-05,
+      "loss": 1.8367,
+      "step": 21790
+    },
+    {
+      "epoch": 1.0,
+      "eval_cer": 0.6809186446203842,
+      "eval_em": 0.007357212003872217,
+      "eval_f1": 0.007357212003872217,
+      "eval_loss": 1.6967871189117432,
+      "eval_runtime": 1613.3656,
+      "eval_samples_per_second": 6.403,
+      "eval_steps_per_second": 0.801,
+      "eval_wer": 0.9926427879961278,
+      "step": 21795
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.255490931353792e-05,
+      "loss": 1.9686,
+      "step": 21800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.255376138363817e-05,
+      "loss": 1.934,
+      "step": 21810
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.255261345373842e-05,
+      "loss": 1.9759,
+      "step": 21820
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.255146552383868e-05,
+      "loss": 1.9817,
+      "step": 21830
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.255031759393892e-05,
+      "loss": 2.0723,
+      "step": 21840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.254916966403918e-05,
+      "loss": 1.9923,
+      "step": 21850
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.254802173413943e-05,
+      "loss": 1.8454,
+      "step": 21860
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.254687380423968e-05,
+      "loss": 1.9951,
+      "step": 21870
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.254572587433993e-05,
+      "loss": 1.8319,
+      "step": 21880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.254457794444019e-05,
+      "loss": 1.7327,
+      "step": 21890
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.254343001454044e-05,
+      "loss": 1.9359,
+      "step": 21900
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.254228208464069e-05,
+      "loss": 1.7933,
+      "step": 21910
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.254113415474094e-05,
+      "loss": 1.8433,
+      "step": 21920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.253998622484119e-05,
+      "loss": 1.915,
+      "step": 21930
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.253883829494145e-05,
+      "loss": 1.9021,
+      "step": 21940
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.25376903650417e-05,
+      "loss": 1.9212,
+      "step": 21950
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.253654243514195e-05,
+      "loss": 1.9267,
+      "step": 21960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.25353945052422e-05,
+      "loss": 1.887,
+      "step": 21970
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.253424657534247e-05,
+      "loss": 1.9403,
+      "step": 21980
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.25330986454427e-05,
+      "loss": 2.0774,
+      "step": 21990
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.253195071554297e-05,
+      "loss": 1.7852,
+      "step": 22000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.253080278564322e-05,
+      "loss": 1.9055,
+      "step": 22010
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.252965485574347e-05,
+      "loss": 2.0133,
+      "step": 22020
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.252850692584372e-05,
+      "loss": 1.8704,
+      "step": 22030
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.252735899594398e-05,
+      "loss": 1.9248,
+      "step": 22040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.252621106604423e-05,
+      "loss": 1.9413,
+      "step": 22050
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.252506313614448e-05,
+      "loss": 1.8849,
+      "step": 22060
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.252391520624473e-05,
+      "loss": 1.8454,
+      "step": 22070
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.252276727634498e-05,
+      "loss": 1.7279,
+      "step": 22080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.252161934644524e-05,
+      "loss": 1.8899,
+      "step": 22090
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.252047141654549e-05,
+      "loss": 2.0092,
+      "step": 22100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.251932348664574e-05,
+      "loss": 1.5853,
+      "step": 22110
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.251817555674599e-05,
+      "loss": 1.9128,
+      "step": 22120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.251702762684626e-05,
+      "loss": 2.0123,
+      "step": 22130
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.251587969694649e-05,
+      "loss": 1.7468,
+      "step": 22140
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.251473176704676e-05,
+      "loss": 1.9968,
+      "step": 22150
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.2513583837147e-05,
+      "loss": 1.8703,
+      "step": 22160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.251243590724726e-05,
+      "loss": 1.9425,
+      "step": 22170
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.25112879773475e-05,
+      "loss": 1.9055,
+      "step": 22180
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.251014004744777e-05,
+      "loss": 1.8149,
+      "step": 22190
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.250899211754802e-05,
+      "loss": 1.9086,
+      "step": 22200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.250784418764827e-05,
+      "loss": 1.7575,
+      "step": 22210
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.250669625774852e-05,
+      "loss": 1.9251,
+      "step": 22220
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.250554832784877e-05,
+      "loss": 1.7077,
+      "step": 22230
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.250440039794903e-05,
+      "loss": 1.8194,
+      "step": 22240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.250325246804928e-05,
+      "loss": 1.7855,
+      "step": 22250
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.250210453814953e-05,
+      "loss": 2.0507,
+      "step": 22260
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.250095660824978e-05,
+      "loss": 2.092,
+      "step": 22270
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.249980867835004e-05,
+      "loss": 1.7923,
+      "step": 22280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.249866074845028e-05,
+      "loss": 1.7926,
+      "step": 22290
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.249751281855054e-05,
+      "loss": 1.9351,
+      "step": 22300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.24963648886508e-05,
+      "loss": 1.8147,
+      "step": 22310
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.249521695875104e-05,
+      "loss": 2.0077,
+      "step": 22320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.24940690288513e-05,
+      "loss": 2.0613,
+      "step": 22330
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.249292109895156e-05,
+      "loss": 2.0291,
+      "step": 22340
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.249177316905181e-05,
+      "loss": 1.9596,
+      "step": 22350
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.249062523915206e-05,
+      "loss": 1.7864,
+      "step": 22360
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.24894773092523e-05,
+      "loss": 1.7816,
+      "step": 22370
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.248832937935256e-05,
+      "loss": 1.9417,
+      "step": 22380
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.248718144945282e-05,
+      "loss": 1.7509,
+      "step": 22390
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.248603351955307e-05,
+      "loss": 1.7989,
+      "step": 22400
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.248488558965332e-05,
+      "loss": 1.7944,
+      "step": 22410
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.248373765975357e-05,
+      "loss": 1.7604,
+      "step": 22420
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.248258972985383e-05,
+      "loss": 1.8117,
+      "step": 22430
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.248144179995407e-05,
+      "loss": 1.8081,
+      "step": 22440
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.248029387005433e-05,
+      "loss": 1.8426,
+      "step": 22450
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.247914594015458e-05,
+      "loss": 1.7097,
+      "step": 22460
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.247799801025483e-05,
+      "loss": 1.9075,
+      "step": 22470
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.247685008035508e-05,
+      "loss": 1.8477,
+      "step": 22480
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.247570215045535e-05,
+      "loss": 1.7949,
+      "step": 22490
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.24745542205556e-05,
+      "loss": 1.8143,
+      "step": 22500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.247340629065585e-05,
+      "loss": 1.8253,
+      "step": 22510
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.24722583607561e-05,
+      "loss": 1.8424,
+      "step": 22520
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.247111043085634e-05,
+      "loss": 1.7062,
+      "step": 22530
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.246996250095661e-05,
+      "loss": 1.9765,
+      "step": 22540
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.246881457105686e-05,
+      "loss": 1.9567,
+      "step": 22550
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.246766664115711e-05,
+      "loss": 1.9597,
+      "step": 22560
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.246651871125736e-05,
+      "loss": 1.969,
+      "step": 22570
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.246537078135762e-05,
+      "loss": 1.7662,
+      "step": 22580
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.246422285145786e-05,
+      "loss": 1.9079,
+      "step": 22590
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.246307492155812e-05,
+      "loss": 1.921,
+      "step": 22600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.246192699165837e-05,
+      "loss": 1.9208,
+      "step": 22610
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.246077906175862e-05,
+      "loss": 1.7711,
+      "step": 22620
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.245963113185887e-05,
+      "loss": 1.8247,
+      "step": 22630
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.245848320195913e-05,
+      "loss": 1.8513,
+      "step": 22640
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.245733527205938e-05,
+      "loss": 1.9907,
+      "step": 22650
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.245618734215963e-05,
+      "loss": 1.777,
+      "step": 22660
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.245503941225988e-05,
+      "loss": 1.8565,
+      "step": 22670
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.245389148236013e-05,
+      "loss": 1.9214,
+      "step": 22680
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.24527435524604e-05,
+      "loss": 1.9221,
+      "step": 22690
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.245159562256065e-05,
+      "loss": 1.756,
+      "step": 22700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.24504476926609e-05,
+      "loss": 1.9282,
+      "step": 22710
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.244929976276115e-05,
+      "loss": 1.828,
+      "step": 22720
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.244815183286141e-05,
+      "loss": 1.693,
+      "step": 22730
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.244700390296165e-05,
+      "loss": 1.9082,
+      "step": 22740
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.244585597306191e-05,
+      "loss": 1.8161,
+      "step": 22750
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.244470804316216e-05,
+      "loss": 1.9991,
+      "step": 22760
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.244356011326241e-05,
+      "loss": 1.8303,
+      "step": 22770
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.244241218336266e-05,
+      "loss": 1.851,
+      "step": 22780
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.244126425346292e-05,
+      "loss": 1.8133,
+      "step": 22790
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.244011632356317e-05,
+      "loss": 1.8974,
+      "step": 22800
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.243896839366342e-05,
+      "loss": 1.9322,
+      "step": 22810
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.243782046376367e-05,
+      "loss": 1.9946,
+      "step": 22820
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.243667253386392e-05,
+      "loss": 1.9328,
+      "step": 22830
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.243552460396419e-05,
+      "loss": 2.0766,
+      "step": 22840
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.243437667406444e-05,
+      "loss": 1.7898,
+      "step": 22850
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.243322874416468e-05,
+      "loss": 1.8206,
+      "step": 22860
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.243208081426493e-05,
+      "loss": 1.9445,
+      "step": 22870
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.24309328843652e-05,
+      "loss": 2.0059,
+      "step": 22880
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.242978495446543e-05,
+      "loss": 1.9183,
+      "step": 22890
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.24286370245657e-05,
+      "loss": 1.7678,
+      "step": 22900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.242748909466595e-05,
+      "loss": 1.9057,
+      "step": 22910
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.24263411647662e-05,
+      "loss": 1.7187,
+      "step": 22920
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.242519323486645e-05,
+      "loss": 1.9671,
+      "step": 22930
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.242404530496671e-05,
+      "loss": 1.8492,
+      "step": 22940
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.242289737506696e-05,
+      "loss": 1.8372,
+      "step": 22950
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.242174944516721e-05,
+      "loss": 1.8674,
+      "step": 22960
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.242060151526746e-05,
+      "loss": 1.8686,
+      "step": 22970
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.241945358536771e-05,
+      "loss": 1.8903,
+      "step": 22980
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.241830565546797e-05,
+      "loss": 2.0926,
+      "step": 22990
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.241715772556822e-05,
+      "loss": 1.9018,
+      "step": 23000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.241600979566847e-05,
+      "loss": 1.7074,
+      "step": 23010
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.241486186576872e-05,
+      "loss": 1.9679,
+      "step": 23020
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.241371393586899e-05,
+      "loss": 1.8452,
+      "step": 23030
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.241256600596922e-05,
+      "loss": 1.9136,
+      "step": 23040
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.241141807606949e-05,
+      "loss": 1.764,
+      "step": 23050
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.241027014616974e-05,
+      "loss": 2.0487,
+      "step": 23060
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.240912221626999e-05,
+      "loss": 1.8063,
+      "step": 23070
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.240797428637024e-05,
+      "loss": 2.1105,
+      "step": 23080
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.24068263564705e-05,
+      "loss": 1.8059,
+      "step": 23090
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.240567842657075e-05,
+      "loss": 1.919,
+      "step": 23100
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.2404530496671e-05,
+      "loss": 1.8781,
+      "step": 23110
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.240338256677125e-05,
+      "loss": 1.9194,
+      "step": 23120
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.24022346368715e-05,
+      "loss": 1.9622,
+      "step": 23130
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.240108670697176e-05,
+      "loss": 1.8328,
+      "step": 23140
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.2399938777072e-05,
+      "loss": 1.9971,
+      "step": 23150
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.239879084717226e-05,
+      "loss": 2.12,
+      "step": 23160
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.239764291727251e-05,
+      "loss": 2.0939,
+      "step": 23170
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.239649498737276e-05,
+      "loss": 1.7553,
+      "step": 23180
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.239534705747301e-05,
+      "loss": 1.8409,
+      "step": 23190
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.239419912757327e-05,
+      "loss": 1.8073,
+      "step": 23200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.239305119767352e-05,
+      "loss": 1.9743,
+      "step": 23210
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.239190326777377e-05,
+      "loss": 1.8193,
+      "step": 23220
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.239075533787402e-05,
+      "loss": 1.8754,
+      "step": 23230
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.238960740797427e-05,
+      "loss": 1.6653,
+      "step": 23240
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.238845947807454e-05,
+      "loss": 1.7153,
+      "step": 23250
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.238731154817479e-05,
+      "loss": 1.8329,
+      "step": 23260
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.238616361827504e-05,
+      "loss": 1.8622,
+      "step": 23270
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.238501568837529e-05,
+      "loss": 2.0436,
+      "step": 23280
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.238386775847555e-05,
+      "loss": 1.5932,
+      "step": 23290
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.238271982857579e-05,
+      "loss": 2.0093,
+      "step": 23300
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.238157189867605e-05,
+      "loss": 1.8207,
+      "step": 23310
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.23804239687763e-05,
+      "loss": 1.7703,
+      "step": 23320
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.237927603887655e-05,
+      "loss": 1.8641,
+      "step": 23330
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.23781281089768e-05,
+      "loss": 2.0188,
+      "step": 23340
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.237698017907706e-05,
+      "loss": 1.9083,
+      "step": 23350
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.237583224917731e-05,
+      "loss": 1.9887,
+      "step": 23360
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.237468431927756e-05,
+      "loss": 1.8425,
+      "step": 23370
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.237353638937781e-05,
+      "loss": 1.9453,
+      "step": 23380
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.237238845947806e-05,
+      "loss": 1.8342,
+      "step": 23390
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.237124052957833e-05,
+      "loss": 1.8835,
+      "step": 23400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.237009259967858e-05,
+      "loss": 1.6404,
+      "step": 23410
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.236894466977883e-05,
+      "loss": 1.7042,
+      "step": 23420
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.236779673987908e-05,
+      "loss": 1.7686,
+      "step": 23430
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.236664880997934e-05,
+      "loss": 1.7073,
+      "step": 23440
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.236550088007958e-05,
+      "loss": 1.7056,
+      "step": 23450
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.236435295017984e-05,
+      "loss": 1.9394,
+      "step": 23460
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.236320502028009e-05,
+      "loss": 1.9515,
+      "step": 23470
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.236205709038034e-05,
+      "loss": 1.6554,
+      "step": 23480
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.236090916048059e-05,
+      "loss": 1.8281,
+      "step": 23490
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.235976123058085e-05,
+      "loss": 1.7949,
+      "step": 23500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.23586133006811e-05,
+      "loss": 2.0147,
+      "step": 23510
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.235746537078135e-05,
+      "loss": 1.687,
+      "step": 23520
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.23563174408816e-05,
+      "loss": 2.0143,
+      "step": 23530
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.235516951098185e-05,
+      "loss": 1.7218,
+      "step": 23540
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.235402158108211e-05,
+      "loss": 1.7901,
+      "step": 23550
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.235287365118236e-05,
+      "loss": 1.8049,
+      "step": 23560
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.235172572128261e-05,
+      "loss": 1.6671,
+      "step": 23570
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.235057779138286e-05,
+      "loss": 1.9665,
+      "step": 23580
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.234942986148313e-05,
+      "loss": 1.7157,
+      "step": 23590
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.234828193158336e-05,
+      "loss": 1.9211,
+      "step": 23600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.234713400168363e-05,
+      "loss": 1.8634,
+      "step": 23610
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.234598607178388e-05,
+      "loss": 1.7849,
+      "step": 23620
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.234483814188413e-05,
+      "loss": 1.9507,
+      "step": 23630
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.234369021198438e-05,
+      "loss": 1.8503,
+      "step": 23640
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.234254228208464e-05,
+      "loss": 1.9097,
+      "step": 23650
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.234139435218489e-05,
+      "loss": 1.9456,
+      "step": 23660
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.234024642228514e-05,
+      "loss": 1.7982,
+      "step": 23670
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.233909849238539e-05,
+      "loss": 2.15,
+      "step": 23680
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.233795056248564e-05,
+      "loss": 1.9132,
+      "step": 23690
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.23368026325859e-05,
+      "loss": 1.9006,
+      "step": 23700
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.233565470268615e-05,
+      "loss": 1.7136,
+      "step": 23710
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.23345067727864e-05,
+      "loss": 1.84,
+      "step": 23720
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.233335884288665e-05,
+      "loss": 1.8447,
+      "step": 23730
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.233221091298692e-05,
+      "loss": 1.883,
+      "step": 23740
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.233106298308715e-05,
+      "loss": 1.8483,
+      "step": 23750
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.232991505318742e-05,
+      "loss": 1.7493,
+      "step": 23760
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.232876712328767e-05,
+      "loss": 2.0091,
+      "step": 23770
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.232761919338792e-05,
+      "loss": 1.8131,
+      "step": 23780
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.232647126348817e-05,
+      "loss": 1.9086,
+      "step": 23790
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.232532333358843e-05,
+      "loss": 1.7573,
+      "step": 23800
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.232417540368868e-05,
+      "loss": 1.8109,
+      "step": 23810
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.232302747378893e-05,
+      "loss": 1.733,
+      "step": 23820
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.232187954388918e-05,
+      "loss": 1.939,
+      "step": 23830
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.232073161398943e-05,
+      "loss": 1.7217,
+      "step": 23840
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.231958368408969e-05,
+      "loss": 1.8617,
+      "step": 23850
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.231843575418994e-05,
+      "loss": 1.945,
+      "step": 23860
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.231728782429019e-05,
+      "loss": 1.8849,
+      "step": 23870
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.231613989439044e-05,
+      "loss": 1.7378,
+      "step": 23880
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.23149919644907e-05,
+      "loss": 1.8815,
+      "step": 23890
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.231384403459094e-05,
+      "loss": 1.7809,
+      "step": 23900
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.23126961046912e-05,
+      "loss": 1.9083,
+      "step": 23910
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.231154817479145e-05,
+      "loss": 1.8219,
+      "step": 23920
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.23104002448917e-05,
+      "loss": 1.7593,
+      "step": 23930
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.230925231499195e-05,
+      "loss": 1.8331,
+      "step": 23940
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.230810438509222e-05,
+      "loss": 1.6438,
+      "step": 23950
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.230695645519247e-05,
+      "loss": 1.8489,
+      "step": 23960
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.230580852529272e-05,
+      "loss": 1.7207,
+      "step": 23970
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.230466059539297e-05,
+      "loss": 1.6838,
+      "step": 23980
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.230351266549322e-05,
+      "loss": 1.7346,
+      "step": 23990
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.230236473559348e-05,
+      "loss": 1.7029,
+      "step": 24000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.230121680569373e-05,
+      "loss": 1.8844,
+      "step": 24010
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.230006887579398e-05,
+      "loss": 1.9011,
+      "step": 24020
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.229892094589423e-05,
+      "loss": 1.9955,
+      "step": 24030
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.229777301599449e-05,
+      "loss": 1.7937,
+      "step": 24040
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.229662508609473e-05,
+      "loss": 1.6272,
+      "step": 24050
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.229547715619499e-05,
+      "loss": 1.9501,
+      "step": 24060
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.229432922629524e-05,
+      "loss": 1.842,
+      "step": 24070
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.229318129639549e-05,
+      "loss": 1.7337,
+      "step": 24080
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.229203336649574e-05,
+      "loss": 1.8213,
+      "step": 24090
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.2290885436596e-05,
+      "loss": 1.7312,
+      "step": 24100
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.228973750669626e-05,
+      "loss": 1.8398,
+      "step": 24110
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.22885895767965e-05,
+      "loss": 2.0784,
+      "step": 24120
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.228744164689676e-05,
+      "loss": 2.0411,
+      "step": 24130
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.2286293716997e-05,
+      "loss": 1.8469,
+      "step": 24140
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.228514578709727e-05,
+      "loss": 1.9117,
+      "step": 24150
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.228399785719752e-05,
+      "loss": 1.8738,
+      "step": 24160
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.228284992729777e-05,
+      "loss": 1.9201,
+      "step": 24170
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.228170199739802e-05,
+      "loss": 1.6893,
+      "step": 24180
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.228055406749828e-05,
+      "loss": 1.7033,
+      "step": 24190
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.227940613759852e-05,
+      "loss": 1.7947,
+      "step": 24200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.227825820769878e-05,
+      "loss": 1.777,
+      "step": 24210
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.227711027779903e-05,
+      "loss": 1.9467,
+      "step": 24220
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.227596234789928e-05,
+      "loss": 1.6795,
+      "step": 24230
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.227481441799953e-05,
+      "loss": 1.8166,
+      "step": 24240
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.22736664880998e-05,
+      "loss": 1.7942,
+      "step": 24250
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.227251855820004e-05,
+      "loss": 1.7819,
+      "step": 24260
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.22713706283003e-05,
+      "loss": 1.759,
+      "step": 24270
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.227022269840054e-05,
+      "loss": 1.5781,
+      "step": 24280
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.22690747685008e-05,
+      "loss": 1.8253,
+      "step": 24290
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.226792683860106e-05,
+      "loss": 1.8581,
+      "step": 24300
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.226677890870131e-05,
+      "loss": 1.8407,
+      "step": 24310
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.226563097880156e-05,
+      "loss": 1.6021,
+      "step": 24320
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.22644830489018e-05,
+      "loss": 1.7847,
+      "step": 24330
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.226333511900207e-05,
+      "loss": 1.6773,
+      "step": 24340
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.22621871891023e-05,
+      "loss": 1.9947,
+      "step": 24350
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.226103925920257e-05,
+      "loss": 1.7344,
+      "step": 24360
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.225989132930282e-05,
+      "loss": 1.8209,
+      "step": 24370
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.225874339940307e-05,
+      "loss": 1.8664,
+      "step": 24380
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.225759546950332e-05,
+      "loss": 1.7639,
+      "step": 24390
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.225644753960358e-05,
+      "loss": 1.7295,
+      "step": 24400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.225529960970383e-05,
+      "loss": 1.9184,
+      "step": 24410
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.225415167980408e-05,
+      "loss": 1.7335,
+      "step": 24420
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.225300374990433e-05,
+      "loss": 1.7043,
+      "step": 24430
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.225185582000458e-05,
+      "loss": 1.7069,
+      "step": 24440
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.225070789010485e-05,
+      "loss": 1.8388,
+      "step": 24450
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.22495599602051e-05,
+      "loss": 1.534,
+      "step": 24460
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.224841203030535e-05,
+      "loss": 1.8394,
+      "step": 24470
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.22472641004056e-05,
+      "loss": 1.8032,
+      "step": 24480
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.224611617050586e-05,
+      "loss": 1.9147,
+      "step": 24490
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.22449682406061e-05,
+      "loss": 1.6866,
+      "step": 24500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.224382031070636e-05,
+      "loss": 1.8275,
+      "step": 24510
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.224267238080661e-05,
+      "loss": 1.6072,
+      "step": 24520
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.224152445090686e-05,
+      "loss": 1.647,
+      "step": 24530
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.224037652100711e-05,
+      "loss": 1.7287,
+      "step": 24540
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.223922859110736e-05,
+      "loss": 1.8913,
+      "step": 24550
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.223808066120762e-05,
+      "loss": 1.8344,
+      "step": 24560
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.223693273130787e-05,
+      "loss": 1.6316,
+      "step": 24570
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.223578480140812e-05,
+      "loss": 1.6081,
+      "step": 24580
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.223463687150837e-05,
+      "loss": 1.8015,
+      "step": 24590
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.223348894160863e-05,
+      "loss": 1.9029,
+      "step": 24600
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.223234101170887e-05,
+      "loss": 1.7171,
+      "step": 24610
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.223119308180913e-05,
+      "loss": 1.629,
+      "step": 24620
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.223004515190938e-05,
+      "loss": 1.8047,
+      "step": 24630
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.222889722200963e-05,
+      "loss": 1.7289,
+      "step": 24640
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.222774929210988e-05,
+      "loss": 1.6154,
+      "step": 24650
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.222660136221015e-05,
+      "loss": 1.9621,
+      "step": 24660
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.22254534323104e-05,
+      "loss": 1.771,
+      "step": 24670
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.222430550241065e-05,
+      "loss": 1.8836,
+      "step": 24680
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.22231575725109e-05,
+      "loss": 1.922,
+      "step": 24690
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.222200964261115e-05,
+      "loss": 1.767,
+      "step": 24700
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.222086171271141e-05,
+      "loss": 1.8683,
+      "step": 24710
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.221971378281166e-05,
+      "loss": 1.747,
+      "step": 24720
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.221856585291191e-05,
+      "loss": 1.6661,
+      "step": 24730
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.221741792301216e-05,
+      "loss": 1.8557,
+      "step": 24740
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.221626999311242e-05,
+      "loss": 1.7419,
+      "step": 24750
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.221512206321266e-05,
+      "loss": 1.9118,
+      "step": 24760
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.221397413331292e-05,
+      "loss": 1.9273,
+      "step": 24770
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.221282620341317e-05,
+      "loss": 1.8666,
+      "step": 24780
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.221167827351342e-05,
+      "loss": 1.7105,
+      "step": 24790
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.221053034361367e-05,
+      "loss": 1.8322,
+      "step": 24800
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.220938241371394e-05,
+      "loss": 1.7659,
+      "step": 24810
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.220823448381418e-05,
+      "loss": 1.7885,
+      "step": 24820
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.220708655391443e-05,
+      "loss": 1.7722,
+      "step": 24830
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.220593862401468e-05,
+      "loss": 2.0253,
+      "step": 24840
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.220479069411493e-05,
+      "loss": 1.7132,
+      "step": 24850
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.22036427642152e-05,
+      "loss": 1.6111,
+      "step": 24860
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.220249483431545e-05,
+      "loss": 1.7164,
+      "step": 24870
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.22013469044157e-05,
+      "loss": 1.8163,
+      "step": 24880
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.220019897451595e-05,
+      "loss": 1.7617,
+      "step": 24890
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.219905104461621e-05,
+      "loss": 1.8876,
+      "step": 24900
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.219790311471645e-05,
+      "loss": 1.894,
+      "step": 24910
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.219675518481671e-05,
+      "loss": 1.823,
+      "step": 24920
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.219560725491696e-05,
+      "loss": 1.7115,
+      "step": 24930
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.219445932501721e-05,
+      "loss": 1.8017,
+      "step": 24940
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.219331139511746e-05,
+      "loss": 1.8892,
+      "step": 24950
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.219216346521772e-05,
+      "loss": 1.8259,
+      "step": 24960
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.219101553531797e-05,
+      "loss": 1.8004,
+      "step": 24970
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.218986760541822e-05,
+      "loss": 1.7264,
+      "step": 24980
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.218871967551847e-05,
+      "loss": 1.9641,
+      "step": 24990
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.218757174561872e-05,
+      "loss": 1.698,
+      "step": 25000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.218642381571899e-05,
+      "loss": 1.6988,
+      "step": 25010
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.218527588581924e-05,
+      "loss": 1.7841,
+      "step": 25020
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.218412795591949e-05,
+      "loss": 1.7853,
+      "step": 25030
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.218298002601974e-05,
+      "loss": 1.7413,
+      "step": 25040
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.218183209612e-05,
+      "loss": 1.8072,
+      "step": 25050
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.218068416622024e-05,
+      "loss": 1.5969,
+      "step": 25060
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.21795362363205e-05,
+      "loss": 1.8559,
+      "step": 25070
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.217838830642075e-05,
+      "loss": 1.8429,
+      "step": 25080
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.2177240376521e-05,
+      "loss": 1.727,
+      "step": 25090
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.217609244662125e-05,
+      "loss": 1.7553,
+      "step": 25100
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.217494451672151e-05,
+      "loss": 1.8666,
+      "step": 25110
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.217379658682176e-05,
+      "loss": 1.7429,
+      "step": 25120
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.217264865692201e-05,
+      "loss": 1.7695,
+      "step": 25130
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.217150072702226e-05,
+      "loss": 1.9913,
+      "step": 25140
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.217035279712251e-05,
+      "loss": 1.8891,
+      "step": 25150
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.216920486722277e-05,
+      "loss": 1.6215,
+      "step": 25160
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.216805693732302e-05,
+      "loss": 1.8359,
+      "step": 25170
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.216690900742327e-05,
+      "loss": 1.9163,
+      "step": 25180
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.216576107752352e-05,
+      "loss": 1.841,
+      "step": 25190
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.216461314762379e-05,
+      "loss": 1.6633,
+      "step": 25200
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.216346521772402e-05,
+      "loss": 1.8654,
+      "step": 25210
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.216231728782429e-05,
+      "loss": 1.8689,
+      "step": 25220
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.216116935792454e-05,
+      "loss": 1.7808,
+      "step": 25230
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.216002142802479e-05,
+      "loss": 1.9029,
+      "step": 25240
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.215887349812504e-05,
+      "loss": 1.7967,
+      "step": 25250
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.21577255682253e-05,
+      "loss": 1.7343,
+      "step": 25260
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.215657763832555e-05,
+      "loss": 1.8518,
+      "step": 25270
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.21554297084258e-05,
+      "loss": 1.8155,
+      "step": 25280
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.215428177852605e-05,
+      "loss": 1.5454,
+      "step": 25290
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.21531338486263e-05,
+      "loss": 1.6308,
+      "step": 25300
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.215198591872656e-05,
+      "loss": 1.8552,
+      "step": 25310
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.215083798882681e-05,
+      "loss": 1.761,
+      "step": 25320
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.214969005892706e-05,
+      "loss": 1.6978,
+      "step": 25330
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.214854212902731e-05,
+      "loss": 1.7553,
+      "step": 25340
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.214739419912758e-05,
+      "loss": 1.6818,
+      "step": 25350
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.214624626922781e-05,
+      "loss": 1.8892,
+      "step": 25360
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.214509833932808e-05,
+      "loss": 1.8301,
+      "step": 25370
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.214395040942833e-05,
+      "loss": 1.7439,
+      "step": 25380
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.214280247952858e-05,
+      "loss": 1.7232,
+      "step": 25390
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.214165454962883e-05,
+      "loss": 1.7882,
+      "step": 25400
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.214050661972909e-05,
+      "loss": 1.884,
+      "step": 25410
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.213935868982934e-05,
+      "loss": 1.8405,
+      "step": 25420
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.213821075992959e-05,
+      "loss": 1.8591,
+      "step": 25430
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.213706283002984e-05,
+      "loss": 1.8535,
+      "step": 25440
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.213591490013009e-05,
+      "loss": 1.8013,
+      "step": 25450
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.213476697023035e-05,
+      "loss": 1.6786,
+      "step": 25460
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.21336190403306e-05,
+      "loss": 1.6113,
+      "step": 25470
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.213247111043085e-05,
+      "loss": 1.6315,
+      "step": 25480
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.21313231805311e-05,
+      "loss": 1.6292,
+      "step": 25490
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.213017525063136e-05,
+      "loss": 1.6252,
+      "step": 25500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.21290273207316e-05,
+      "loss": 1.669,
+      "step": 25510
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.212787939083186e-05,
+      "loss": 1.7902,
+      "step": 25520
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.212673146093211e-05,
+      "loss": 1.9244,
+      "step": 25530
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.212558353103236e-05,
+      "loss": 1.8232,
+      "step": 25540
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.212443560113261e-05,
+      "loss": 1.653,
+      "step": 25550
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.212328767123288e-05,
+      "loss": 1.951,
+      "step": 25560
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.212213974133313e-05,
+      "loss": 1.6676,
+      "step": 25570
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.212099181143338e-05,
+      "loss": 1.6556,
+      "step": 25580
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.211984388153363e-05,
+      "loss": 1.7376,
+      "step": 25590
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.211869595163388e-05,
+      "loss": 1.8957,
+      "step": 25600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.211754802173414e-05,
+      "loss": 1.803,
+      "step": 25610
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.211640009183439e-05,
+      "loss": 1.6839,
+      "step": 25620
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.211525216193464e-05,
+      "loss": 1.7382,
+      "step": 25630
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.211410423203489e-05,
+      "loss": 1.825,
+      "step": 25640
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.211295630213515e-05,
+      "loss": 1.7867,
+      "step": 25650
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.211180837223539e-05,
+      "loss": 1.6722,
+      "step": 25660
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.211066044233565e-05,
+      "loss": 1.8755,
+      "step": 25670
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.21095125124359e-05,
+      "loss": 1.6821,
+      "step": 25680
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.210836458253615e-05,
+      "loss": 1.8445,
+      "step": 25690
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.21072166526364e-05,
+      "loss": 1.7127,
+      "step": 25700
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.210606872273667e-05,
+      "loss": 1.7978,
+      "step": 25710
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.210492079283692e-05,
+      "loss": 1.6774,
+      "step": 25720
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.210377286293717e-05,
+      "loss": 1.6432,
+      "step": 25730
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.210262493303742e-05,
+      "loss": 1.6549,
+      "step": 25740
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.210147700313767e-05,
+      "loss": 1.7625,
+      "step": 25750
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.210032907323793e-05,
+      "loss": 1.7876,
+      "step": 25760
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.209918114333818e-05,
+      "loss": 1.7558,
+      "step": 25770
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.209803321343843e-05,
+      "loss": 1.731,
+      "step": 25780
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.209688528353868e-05,
+      "loss": 1.5703,
+      "step": 25790
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.209573735363894e-05,
+      "loss": 1.7462,
+      "step": 25800
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.209458942373918e-05,
+      "loss": 1.7294,
+      "step": 25810
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.209344149383944e-05,
+      "loss": 1.6713,
+      "step": 25820
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.209229356393969e-05,
+      "loss": 1.7951,
+      "step": 25830
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.209114563403994e-05,
+      "loss": 1.7283,
+      "step": 25840
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.208999770414019e-05,
+      "loss": 1.8084,
+      "step": 25850
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.208884977424045e-05,
+      "loss": 1.8978,
+      "step": 25860
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.20877018443407e-05,
+      "loss": 1.8984,
+      "step": 25870
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.208655391444095e-05,
+      "loss": 1.6792,
+      "step": 25880
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.20854059845412e-05,
+      "loss": 1.9929,
+      "step": 25890
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.208425805464145e-05,
+      "loss": 1.9381,
+      "step": 25900
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.208311012474172e-05,
+      "loss": 1.5616,
+      "step": 25910
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.208196219484195e-05,
+      "loss": 1.7871,
+      "step": 25920
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.208081426494222e-05,
+      "loss": 1.8034,
+      "step": 25930
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.207966633504247e-05,
+      "loss": 1.6757,
+      "step": 25940
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.207851840514272e-05,
+      "loss": 1.5556,
+      "step": 25950
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.207737047524297e-05,
+      "loss": 1.871,
+      "step": 25960
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.207622254534323e-05,
+      "loss": 1.8182,
+      "step": 25970
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.207507461544348e-05,
+      "loss": 1.7453,
+      "step": 25980
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.207392668554373e-05,
+      "loss": 1.861,
+      "step": 25990
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.207277875564398e-05,
+      "loss": 1.6674,
+      "step": 26000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.207163082574423e-05,
+      "loss": 1.8693,
+      "step": 26010
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.207048289584449e-05,
+      "loss": 1.8655,
+      "step": 26020
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.206933496594474e-05,
+      "loss": 1.811,
+      "step": 26030
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.206818703604499e-05,
+      "loss": 1.7803,
+      "step": 26040
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.206703910614524e-05,
+      "loss": 1.7987,
+      "step": 26050
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.20658911762455e-05,
+      "loss": 1.6842,
+      "step": 26060
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.206474324634574e-05,
+      "loss": 1.724,
+      "step": 26070
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2063595316446e-05,
+      "loss": 1.7101,
+      "step": 26080
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.206244738654626e-05,
+      "loss": 1.5574,
+      "step": 26090
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.20612994566465e-05,
+      "loss": 1.6339,
+      "step": 26100
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.206015152674675e-05,
+      "loss": 1.6725,
+      "step": 26110
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.205900359684702e-05,
+      "loss": 1.6615,
+      "step": 26120
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.205785566694727e-05,
+      "loss": 1.8006,
+      "step": 26130
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.205670773704752e-05,
+      "loss": 1.6566,
+      "step": 26140
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.205555980714777e-05,
+      "loss": 1.7103,
+      "step": 26150
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.205441187724802e-05,
+      "loss": 1.7666,
+      "step": 26160
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.205326394734828e-05,
+      "loss": 1.8017,
+      "step": 26170
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.205211601744853e-05,
+      "loss": 1.8629,
+      "step": 26180
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.205096808754878e-05,
+      "loss": 1.7045,
+      "step": 26190
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.204982015764903e-05,
+      "loss": 1.7574,
+      "step": 26200
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.20486722277493e-05,
+      "loss": 1.7417,
+      "step": 26210
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.204752429784953e-05,
+      "loss": 1.7083,
+      "step": 26220
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.20463763679498e-05,
+      "loss": 1.5426,
+      "step": 26230
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.204522843805004e-05,
+      "loss": 1.7566,
+      "step": 26240
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.20440805081503e-05,
+      "loss": 1.6434,
+      "step": 26250
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.204293257825054e-05,
+      "loss": 1.5807,
+      "step": 26260
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.204178464835081e-05,
+      "loss": 1.7318,
+      "step": 26270
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.204063671845106e-05,
+      "loss": 1.6452,
+      "step": 26280
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.20394887885513e-05,
+      "loss": 1.8224,
+      "step": 26290
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.203834085865156e-05,
+      "loss": 1.7872,
+      "step": 26300
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.20371929287518e-05,
+      "loss": 1.7631,
+      "step": 26310
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.203604499885207e-05,
+      "loss": 1.744,
+      "step": 26320
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.203489706895232e-05,
+      "loss": 1.8813,
+      "step": 26330
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.203374913905257e-05,
+      "loss": 1.7225,
+      "step": 26340
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.203260120915282e-05,
+      "loss": 1.7649,
+      "step": 26350
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.203145327925308e-05,
+      "loss": 1.7377,
+      "step": 26360
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.203030534935332e-05,
+      "loss": 1.8029,
+      "step": 26370
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.202915741945358e-05,
+      "loss": 1.8189,
+      "step": 26380
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.202800948955383e-05,
+      "loss": 1.5503,
+      "step": 26390
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.202686155965408e-05,
+      "loss": 1.8813,
+      "step": 26400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.202571362975433e-05,
+      "loss": 1.7654,
+      "step": 26410
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.20245656998546e-05,
+      "loss": 1.7215,
+      "step": 26420
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.202341776995485e-05,
+      "loss": 1.7748,
+      "step": 26430
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.20222698400551e-05,
+      "loss": 1.821,
+      "step": 26440
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.202112191015534e-05,
+      "loss": 1.7368,
+      "step": 26450
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.20199739802556e-05,
+      "loss": 1.7702,
+      "step": 26460
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.201882605035586e-05,
+      "loss": 1.6925,
+      "step": 26470
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.201767812045611e-05,
+      "loss": 1.7376,
+      "step": 26480
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.201653019055636e-05,
+      "loss": 1.6559,
+      "step": 26490
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.201538226065661e-05,
+      "loss": 1.6204,
+      "step": 26500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.201423433075687e-05,
+      "loss": 1.6593,
+      "step": 26510
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.201308640085711e-05,
+      "loss": 1.8477,
+      "step": 26520
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.201193847095737e-05,
+      "loss": 1.799,
+      "step": 26530
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.201079054105762e-05,
+      "loss": 1.8518,
+      "step": 26540
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.200964261115787e-05,
+      "loss": 1.9711,
+      "step": 26550
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.200849468125812e-05,
+      "loss": 1.7435,
+      "step": 26560
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.200734675135838e-05,
+      "loss": 1.6303,
+      "step": 26570
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.200619882145863e-05,
+      "loss": 1.6566,
+      "step": 26580
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.200505089155888e-05,
+      "loss": 1.6868,
+      "step": 26590
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.200390296165913e-05,
+      "loss": 1.6857,
+      "step": 26600
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.200275503175938e-05,
+      "loss": 1.7217,
+      "step": 26610
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.200160710185965e-05,
+      "loss": 1.8053,
+      "step": 26620
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.20004591719599e-05,
+      "loss": 1.8391,
+      "step": 26630
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.199931124206015e-05,
+      "loss": 1.5665,
+      "step": 26640
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.19981633121604e-05,
+      "loss": 1.9246,
+      "step": 26650
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.199701538226066e-05,
+      "loss": 1.7469,
+      "step": 26660
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.19958674523609e-05,
+      "loss": 1.9565,
+      "step": 26670
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.199471952246116e-05,
+      "loss": 1.6351,
+      "step": 26680
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.199357159256141e-05,
+      "loss": 1.7731,
+      "step": 26690
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.199242366266166e-05,
+      "loss": 1.5796,
+      "step": 26700
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.199127573276191e-05,
+      "loss": 1.6701,
+      "step": 26710
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.199012780286217e-05,
+      "loss": 1.5225,
+      "step": 26720
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.198897987296242e-05,
+      "loss": 1.9541,
+      "step": 26730
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.198783194306267e-05,
+      "loss": 1.7107,
+      "step": 26740
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.198668401316292e-05,
+      "loss": 1.7835,
+      "step": 26750
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.198553608326317e-05,
+      "loss": 1.7061,
+      "step": 26760
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.198438815336344e-05,
+      "loss": 1.7515,
+      "step": 26770
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.198324022346368e-05,
+      "loss": 1.7204,
+      "step": 26780
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.198209229356393e-05,
+      "loss": 1.6767,
+      "step": 26790
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.198094436366418e-05,
+      "loss": 1.5629,
+      "step": 26800
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.197979643376445e-05,
+      "loss": 1.5967,
+      "step": 26810
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.197864850386468e-05,
+      "loss": 1.6766,
+      "step": 26820
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.197750057396495e-05,
+      "loss": 1.721,
+      "step": 26830
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.19763526440652e-05,
+      "loss": 1.6415,
+      "step": 26840
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.197520471416545e-05,
+      "loss": 1.5124,
+      "step": 26850
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.19740567842657e-05,
+      "loss": 1.4531,
+      "step": 26860
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.197290885436596e-05,
+      "loss": 1.7265,
+      "step": 26870
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.197176092446621e-05,
+      "loss": 1.6166,
+      "step": 26880
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.197061299456646e-05,
+      "loss": 1.7642,
+      "step": 26890
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.196946506466671e-05,
+      "loss": 1.8535,
+      "step": 26900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.196831713476696e-05,
+      "loss": 1.6946,
+      "step": 26910
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.196716920486722e-05,
+      "loss": 1.6506,
+      "step": 26920
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.196602127496747e-05,
+      "loss": 1.7309,
+      "step": 26930
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.196487334506772e-05,
+      "loss": 1.8258,
+      "step": 26940
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.196372541516797e-05,
+      "loss": 1.781,
+      "step": 26950
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.196257748526824e-05,
+      "loss": 1.8444,
+      "step": 26960
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.196142955536847e-05,
+      "loss": 1.8525,
+      "step": 26970
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.196028162546874e-05,
+      "loss": 1.7711,
+      "step": 26980
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.195913369556899e-05,
+      "loss": 1.7302,
+      "step": 26990
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.195798576566924e-05,
+      "loss": 1.68,
+      "step": 27000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.195683783576949e-05,
+      "loss": 1.7452,
+      "step": 27010
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.195568990586975e-05,
+      "loss": 1.6456,
+      "step": 27020
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.195454197597e-05,
+      "loss": 1.6087,
+      "step": 27030
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.195339404607025e-05,
+      "loss": 1.6846,
+      "step": 27040
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.19522461161705e-05,
+      "loss": 1.9413,
+      "step": 27050
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.195109818627075e-05,
+      "loss": 1.818,
+      "step": 27060
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.194995025637101e-05,
+      "loss": 1.6719,
+      "step": 27070
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.194880232647126e-05,
+      "loss": 1.9237,
+      "step": 27080
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.194765439657151e-05,
+      "loss": 1.6829,
+      "step": 27090
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.194650646667176e-05,
+      "loss": 1.5875,
+      "step": 27100
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.194535853677202e-05,
+      "loss": 1.5732,
+      "step": 27110
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.194421060687226e-05,
+      "loss": 1.9326,
+      "step": 27120
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.194306267697252e-05,
+      "loss": 1.4966,
+      "step": 27130
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.194191474707277e-05,
+      "loss": 1.6786,
+      "step": 27140
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.194076681717302e-05,
+      "loss": 1.7617,
+      "step": 27150
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.193961888727327e-05,
+      "loss": 1.7965,
+      "step": 27160
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.193847095737354e-05,
+      "loss": 1.7407,
+      "step": 27170
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.193732302747379e-05,
+      "loss": 1.8857,
+      "step": 27180
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.193617509757404e-05,
+      "loss": 1.7825,
+      "step": 27190
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.193502716767429e-05,
+      "loss": 1.676,
+      "step": 27200
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.193387923777454e-05,
+      "loss": 1.7324,
+      "step": 27210
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.19327313078748e-05,
+      "loss": 1.7648,
+      "step": 27220
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.193158337797505e-05,
+      "loss": 1.6244,
+      "step": 27230
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.19304354480753e-05,
+      "loss": 1.8292,
+      "step": 27240
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.192928751817555e-05,
+      "loss": 1.7023,
+      "step": 27250
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.192813958827581e-05,
+      "loss": 1.5375,
+      "step": 27260
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.192699165837605e-05,
+      "loss": 1.8046,
+      "step": 27270
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.192584372847631e-05,
+      "loss": 1.656,
+      "step": 27280
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.192469579857656e-05,
+      "loss": 1.6816,
+      "step": 27290
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.192354786867681e-05,
+      "loss": 1.6201,
+      "step": 27300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.192239993877706e-05,
+      "loss": 1.7669,
+      "step": 27310
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.192125200887731e-05,
+      "loss": 1.7293,
+      "step": 27320
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.192010407897758e-05,
+      "loss": 1.7144,
+      "step": 27330
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.191895614907783e-05,
+      "loss": 1.5907,
+      "step": 27340
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.191780821917808e-05,
+      "loss": 1.8048,
+      "step": 27350
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.191666028927833e-05,
+      "loss": 1.6078,
+      "step": 27360
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.191551235937859e-05,
+      "loss": 1.7447,
+      "step": 27370
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.191436442947883e-05,
+      "loss": 1.7624,
+      "step": 27380
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.191321649957909e-05,
+      "loss": 1.6829,
+      "step": 27390
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.191206856967934e-05,
+      "loss": 1.6689,
+      "step": 27400
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.191092063977959e-05,
+      "loss": 1.6797,
+      "step": 27410
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.190977270987984e-05,
+      "loss": 1.7257,
+      "step": 27420
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.19086247799801e-05,
+      "loss": 1.6948,
+      "step": 27430
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.190747685008035e-05,
+      "loss": 1.812,
+      "step": 27440
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.19063289201806e-05,
+      "loss": 1.7934,
+      "step": 27450
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.190518099028085e-05,
+      "loss": 1.8195,
+      "step": 27460
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.19040330603811e-05,
+      "loss": 1.631,
+      "step": 27470
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.190288513048136e-05,
+      "loss": 1.6874,
+      "step": 27480
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.190173720058161e-05,
+      "loss": 1.6703,
+      "step": 27490
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.190058927068186e-05,
+      "loss": 1.5896,
+      "step": 27500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.189944134078211e-05,
+      "loss": 1.617,
+      "step": 27510
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.189829341088238e-05,
+      "loss": 1.8082,
+      "step": 27520
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.189714548098261e-05,
+      "loss": 1.617,
+      "step": 27530
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.189599755108288e-05,
+      "loss": 1.4966,
+      "step": 27540
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.189484962118313e-05,
+      "loss": 1.7048,
+      "step": 27550
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.189370169128338e-05,
+      "loss": 1.6654,
+      "step": 27560
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.189255376138363e-05,
+      "loss": 1.5123,
+      "step": 27570
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.189140583148389e-05,
+      "loss": 1.7642,
+      "step": 27580
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.189025790158414e-05,
+      "loss": 1.7783,
+      "step": 27590
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.188910997168439e-05,
+      "loss": 1.6234,
+      "step": 27600
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.188796204178464e-05,
+      "loss": 1.6141,
+      "step": 27610
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.188681411188489e-05,
+      "loss": 1.7877,
+      "step": 27620
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.188566618198515e-05,
+      "loss": 1.7603,
+      "step": 27630
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.18845182520854e-05,
+      "loss": 1.7813,
+      "step": 27640
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.188337032218565e-05,
+      "loss": 1.5839,
+      "step": 27650
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.18822223922859e-05,
+      "loss": 1.8346,
+      "step": 27660
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.188107446238617e-05,
+      "loss": 1.7225,
+      "step": 27670
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.18799265324864e-05,
+      "loss": 1.8475,
+      "step": 27680
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.187877860258667e-05,
+      "loss": 1.5368,
+      "step": 27690
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.187763067268692e-05,
+      "loss": 1.7478,
+      "step": 27700
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.187648274278717e-05,
+      "loss": 1.671,
+      "step": 27710
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.187533481288742e-05,
+      "loss": 1.8328,
+      "step": 27720
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.187418688298768e-05,
+      "loss": 1.5565,
+      "step": 27730
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.187303895308793e-05,
+      "loss": 1.6989,
+      "step": 27740
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.187189102318818e-05,
+      "loss": 1.8531,
+      "step": 27750
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.187074309328843e-05,
+      "loss": 1.543,
+      "step": 27760
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.186959516338868e-05,
+      "loss": 1.8334,
+      "step": 27770
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.186844723348894e-05,
+      "loss": 1.7125,
+      "step": 27780
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.186729930358919e-05,
+      "loss": 1.7481,
+      "step": 27790
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.186615137368944e-05,
+      "loss": 1.7026,
+      "step": 27800
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.186500344378969e-05,
+      "loss": 1.6206,
+      "step": 27810
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.186385551388995e-05,
+      "loss": 1.6961,
+      "step": 27820
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.186270758399019e-05,
+      "loss": 1.6335,
+      "step": 27830
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.186155965409045e-05,
+      "loss": 1.6409,
+      "step": 27840
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.18604117241907e-05,
+      "loss": 1.6902,
+      "step": 27850
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.185926379429095e-05,
+      "loss": 1.9642,
+      "step": 27860
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.18581158643912e-05,
+      "loss": 1.7802,
+      "step": 27870
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.185696793449147e-05,
+      "loss": 1.5728,
+      "step": 27880
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.185582000459172e-05,
+      "loss": 1.66,
+      "step": 27890
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.185467207469197e-05,
+      "loss": 1.696,
+      "step": 27900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.185352414479222e-05,
+      "loss": 1.6017,
+      "step": 27910
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.185237621489247e-05,
+      "loss": 1.779,
+      "step": 27920
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.185122828499273e-05,
+      "loss": 1.8186,
+      "step": 27930
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.185008035509298e-05,
+      "loss": 1.657,
+      "step": 27940
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.184893242519323e-05,
+      "loss": 1.7363,
+      "step": 27950
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.184778449529348e-05,
+      "loss": 1.6542,
+      "step": 27960
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.184663656539374e-05,
+      "loss": 1.7713,
+      "step": 27970
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.184548863549398e-05,
+      "loss": 1.822,
+      "step": 27980
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.184434070559424e-05,
+      "loss": 1.6389,
+      "step": 27990
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.184319277569449e-05,
+      "loss": 1.6023,
+      "step": 28000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.184204484579474e-05,
+      "loss": 1.806,
+      "step": 28010
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.184089691589499e-05,
+      "loss": 1.8083,
+      "step": 28020
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.183974898599526e-05,
+      "loss": 1.7924,
+      "step": 28030
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.18386010560955e-05,
+      "loss": 1.6261,
+      "step": 28040
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.183745312619576e-05,
+      "loss": 1.6665,
+      "step": 28050
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.1836305196296e-05,
+      "loss": 1.7557,
+      "step": 28060
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.183515726639625e-05,
+      "loss": 1.8625,
+      "step": 28070
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.183400933649652e-05,
+      "loss": 1.7231,
+      "step": 28080
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.183286140659677e-05,
+      "loss": 1.6332,
+      "step": 28090
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.183171347669702e-05,
+      "loss": 1.6826,
+      "step": 28100
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.183056554679727e-05,
+      "loss": 1.6782,
+      "step": 28110
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.182941761689753e-05,
+      "loss": 1.7969,
+      "step": 28120
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.182826968699777e-05,
+      "loss": 1.8588,
+      "step": 28130
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.182712175709803e-05,
+      "loss": 1.601,
+      "step": 28140
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.182597382719828e-05,
+      "loss": 1.6474,
+      "step": 28150
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.182482589729853e-05,
+      "loss": 1.4742,
+      "step": 28160
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.182367796739878e-05,
+      "loss": 1.618,
+      "step": 28170
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.182253003749904e-05,
+      "loss": 1.8565,
+      "step": 28180
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.18213821075993e-05,
+      "loss": 1.6814,
+      "step": 28190
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.182023417769954e-05,
+      "loss": 1.6945,
+      "step": 28200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.18190862477998e-05,
+      "loss": 1.7316,
+      "step": 28210
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.181793831790004e-05,
+      "loss": 1.75,
+      "step": 28220
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.181679038800031e-05,
+      "loss": 1.5976,
+      "step": 28230
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.181564245810056e-05,
+      "loss": 1.6613,
+      "step": 28240
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.18144945282008e-05,
+      "loss": 1.7627,
+      "step": 28250
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.181334659830106e-05,
+      "loss": 1.5581,
+      "step": 28260
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.181219866840132e-05,
+      "loss": 1.5508,
+      "step": 28270
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.181105073850156e-05,
+      "loss": 1.6551,
+      "step": 28280
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.180990280860182e-05,
+      "loss": 1.5962,
+      "step": 28290
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.180875487870207e-05,
+      "loss": 1.6453,
+      "step": 28300
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.180760694880232e-05,
+      "loss": 1.6403,
+      "step": 28310
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.180645901890257e-05,
+      "loss": 1.735,
+      "step": 28320
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.180531108900283e-05,
+      "loss": 1.6525,
+      "step": 28330
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.180416315910308e-05,
+      "loss": 1.887,
+      "step": 28340
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.180301522920333e-05,
+      "loss": 1.9131,
+      "step": 28350
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.180186729930358e-05,
+      "loss": 1.7148,
+      "step": 28360
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.180071936940383e-05,
+      "loss": 1.6136,
+      "step": 28370
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.17995714395041e-05,
+      "loss": 1.6147,
+      "step": 28380
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.179842350960435e-05,
+      "loss": 1.7571,
+      "step": 28390
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.17972755797046e-05,
+      "loss": 1.6498,
+      "step": 28400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.179612764980484e-05,
+      "loss": 1.73,
+      "step": 28410
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.179497971990511e-05,
+      "loss": 1.5858,
+      "step": 28420
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.179383179000534e-05,
+      "loss": 1.6415,
+      "step": 28430
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.179268386010561e-05,
+      "loss": 1.864,
+      "step": 28440
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.179153593020586e-05,
+      "loss": 1.6191,
+      "step": 28450
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.179038800030611e-05,
+      "loss": 1.731,
+      "step": 28460
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.178924007040636e-05,
+      "loss": 1.695,
+      "step": 28470
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.178809214050662e-05,
+      "loss": 1.6579,
+      "step": 28480
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.178694421060687e-05,
+      "loss": 1.7611,
+      "step": 28490
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.178579628070712e-05,
+      "loss": 1.5625,
+      "step": 28500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.178464835080737e-05,
+      "loss": 1.5914,
+      "step": 28510
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.178350042090762e-05,
+      "loss": 1.6631,
+      "step": 28520
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.178235249100788e-05,
+      "loss": 1.7493,
+      "step": 28530
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.178120456110813e-05,
+      "loss": 1.7485,
+      "step": 28540
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.178005663120838e-05,
+      "loss": 1.4974,
+      "step": 28550
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.177890870130863e-05,
+      "loss": 1.672,
+      "step": 28560
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.17777607714089e-05,
+      "loss": 1.5626,
+      "step": 28570
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.177661284150913e-05,
+      "loss": 1.7328,
+      "step": 28580
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.17754649116094e-05,
+      "loss": 1.6526,
+      "step": 28590
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.177431698170965e-05,
+      "loss": 1.7391,
+      "step": 28600
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.17731690518099e-05,
+      "loss": 1.7038,
+      "step": 28610
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.177202112191015e-05,
+      "loss": 1.7715,
+      "step": 28620
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.177087319201041e-05,
+      "loss": 1.6235,
+      "step": 28630
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.176972526211066e-05,
+      "loss": 1.7172,
+      "step": 28640
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.176857733221091e-05,
+      "loss": 1.8622,
+      "step": 28650
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.176742940231116e-05,
+      "loss": 1.7841,
+      "step": 28660
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.176628147241141e-05,
+      "loss": 1.7482,
+      "step": 28670
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.176513354251167e-05,
+      "loss": 1.5707,
+      "step": 28680
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.176398561261191e-05,
+      "loss": 1.6577,
+      "step": 28690
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.176283768271217e-05,
+      "loss": 1.8449,
+      "step": 28700
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.176168975281242e-05,
+      "loss": 1.6624,
+      "step": 28710
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.176054182291267e-05,
+      "loss": 1.558,
+      "step": 28720
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.175939389301292e-05,
+      "loss": 1.5914,
+      "step": 28730
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.175824596311318e-05,
+      "loss": 1.7125,
+      "step": 28740
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.175709803321343e-05,
+      "loss": 1.8393,
+      "step": 28750
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.175595010331368e-05,
+      "loss": 1.7763,
+      "step": 28760
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.175480217341393e-05,
+      "loss": 1.7878,
+      "step": 28770
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.175365424351418e-05,
+      "loss": 1.9319,
+      "step": 28780
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.175250631361445e-05,
+      "loss": 1.5909,
+      "step": 28790
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.17513583837147e-05,
+      "loss": 1.6788,
+      "step": 28800
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.175021045381495e-05,
+      "loss": 1.9155,
+      "step": 28810
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.17490625239152e-05,
+      "loss": 1.8382,
+      "step": 28820
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.174791459401546e-05,
+      "loss": 1.6078,
+      "step": 28830
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.17467666641157e-05,
+      "loss": 1.8384,
+      "step": 28840
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.174561873421596e-05,
+      "loss": 1.7854,
+      "step": 28850
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.174447080431621e-05,
+      "loss": 1.7076,
+      "step": 28860
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.174332287441646e-05,
+      "loss": 1.5982,
+      "step": 28870
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.174217494451671e-05,
+      "loss": 1.6739,
+      "step": 28880
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.174102701461697e-05,
+      "loss": 1.7341,
+      "step": 28890
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.173987908471722e-05,
+      "loss": 1.7197,
+      "step": 28900
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.173873115481747e-05,
+      "loss": 1.4897,
+      "step": 28910
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.173758322491772e-05,
+      "loss": 1.8193,
+      "step": 28920
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.173643529501797e-05,
+      "loss": 1.7524,
+      "step": 28930
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.173528736511824e-05,
+      "loss": 1.7356,
+      "step": 28940
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.173413943521849e-05,
+      "loss": 1.7174,
+      "step": 28950
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.173299150531874e-05,
+      "loss": 1.718,
+      "step": 28960
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.173184357541899e-05,
+      "loss": 1.7419,
+      "step": 28970
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.173069564551925e-05,
+      "loss": 1.7299,
+      "step": 28980
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.172954771561949e-05,
+      "loss": 1.8231,
+      "step": 28990
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.172839978571975e-05,
+      "loss": 1.6838,
+      "step": 29000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.172725185582e-05,
+      "loss": 1.7051,
+      "step": 29010
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.172610392592025e-05,
+      "loss": 1.5735,
+      "step": 29020
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.17249559960205e-05,
+      "loss": 1.7367,
+      "step": 29030
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.172380806612076e-05,
+      "loss": 1.5408,
+      "step": 29040
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.172266013622101e-05,
+      "loss": 1.7242,
+      "step": 29050
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.172151220632126e-05,
+      "loss": 1.5737,
+      "step": 29060
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.172036427642151e-05,
+      "loss": 1.7291,
+      "step": 29070
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.171921634652176e-05,
+      "loss": 1.6385,
+      "step": 29080
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.171806841662202e-05,
+      "loss": 1.7152,
+      "step": 29090
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.171692048672227e-05,
+      "loss": 1.7241,
+      "step": 29100
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.171577255682252e-05,
+      "loss": 1.6749,
+      "step": 29110
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.171462462692277e-05,
+      "loss": 1.6292,
+      "step": 29120
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.171347669702304e-05,
+      "loss": 1.6393,
+      "step": 29130
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.171232876712327e-05,
+      "loss": 1.7277,
+      "step": 29140
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.171118083722354e-05,
+      "loss": 1.5583,
+      "step": 29150
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.171003290732379e-05,
+      "loss": 1.6626,
+      "step": 29160
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.170888497742404e-05,
+      "loss": 1.4833,
+      "step": 29170
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.170773704752429e-05,
+      "loss": 1.751,
+      "step": 29180
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.170658911762455e-05,
+      "loss": 1.7768,
+      "step": 29190
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.17054411877248e-05,
+      "loss": 1.645,
+      "step": 29200
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.170429325782505e-05,
+      "loss": 1.6505,
+      "step": 29210
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.17031453279253e-05,
+      "loss": 1.6596,
+      "step": 29220
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.170199739802555e-05,
+      "loss": 1.961,
+      "step": 29230
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.170084946812581e-05,
+      "loss": 1.6289,
+      "step": 29240
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.169970153822606e-05,
+      "loss": 1.4775,
+      "step": 29250
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.169855360832631e-05,
+      "loss": 1.781,
+      "step": 29260
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.169740567842656e-05,
+      "loss": 1.7261,
+      "step": 29270
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.169625774852683e-05,
+      "loss": 1.6084,
+      "step": 29280
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.169510981862706e-05,
+      "loss": 1.5469,
+      "step": 29290
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.169396188872733e-05,
+      "loss": 1.8622,
+      "step": 29300
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.169281395882758e-05,
+      "loss": 1.6433,
+      "step": 29310
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.169166602892783e-05,
+      "loss": 1.4562,
+      "step": 29320
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.169051809902808e-05,
+      "loss": 1.5929,
+      "step": 29330
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.168937016912834e-05,
+      "loss": 1.7027,
+      "step": 29340
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.168822223922859e-05,
+      "loss": 1.581,
+      "step": 29350
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.168707430932884e-05,
+      "loss": 1.7434,
+      "step": 29360
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.168592637942909e-05,
+      "loss": 1.5611,
+      "step": 29370
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.168477844952934e-05,
+      "loss": 1.6673,
+      "step": 29380
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.16836305196296e-05,
+      "loss": 1.9114,
+      "step": 29390
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.168248258972985e-05,
+      "loss": 1.5296,
+      "step": 29400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.16813346598301e-05,
+      "loss": 1.7731,
+      "step": 29410
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.168018672993035e-05,
+      "loss": 1.6783,
+      "step": 29420
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.167903880003061e-05,
+      "loss": 1.9307,
+      "step": 29430
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.167789087013085e-05,
+      "loss": 1.6385,
+      "step": 29440
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.167674294023111e-05,
+      "loss": 1.8201,
+      "step": 29450
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.167559501033136e-05,
+      "loss": 1.7093,
+      "step": 29460
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.167444708043161e-05,
+      "loss": 1.7325,
+      "step": 29470
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.167329915053186e-05,
+      "loss": 1.8222,
+      "step": 29480
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.167215122063213e-05,
+      "loss": 1.8272,
+      "step": 29490
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.167100329073238e-05,
+      "loss": 1.8338,
+      "step": 29500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.166985536083263e-05,
+      "loss": 1.8381,
+      "step": 29510
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.166870743093288e-05,
+      "loss": 1.6227,
+      "step": 29520
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.166755950103313e-05,
+      "loss": 1.5509,
+      "step": 29530
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.166641157113339e-05,
+      "loss": 1.7491,
+      "step": 29540
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.166526364123364e-05,
+      "loss": 1.6649,
+      "step": 29550
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.166411571133389e-05,
+      "loss": 1.5193,
+      "step": 29560
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.166296778143414e-05,
+      "loss": 1.7728,
+      "step": 29570
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.16618198515344e-05,
+      "loss": 1.6473,
+      "step": 29580
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.166067192163464e-05,
+      "loss": 1.6047,
+      "step": 29590
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.16595239917349e-05,
+      "loss": 1.6374,
+      "step": 29600
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.165837606183515e-05,
+      "loss": 1.7244,
+      "step": 29610
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.16572281319354e-05,
+      "loss": 1.4751,
+      "step": 29620
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.165608020203565e-05,
+      "loss": 1.7414,
+      "step": 29630
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.165493227213592e-05,
+      "loss": 1.7404,
+      "step": 29640
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.165378434223617e-05,
+      "loss": 1.707,
+      "step": 29650
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.165263641233642e-05,
+      "loss": 1.7871,
+      "step": 29660
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.165148848243667e-05,
+      "loss": 1.6612,
+      "step": 29670
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.165034055253692e-05,
+      "loss": 1.5383,
+      "step": 29680
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.164919262263718e-05,
+      "loss": 1.7711,
+      "step": 29690
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.164804469273743e-05,
+      "loss": 1.7076,
+      "step": 29700
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.164689676283768e-05,
+      "loss": 1.6383,
+      "step": 29710
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.164574883293793e-05,
+      "loss": 1.5948,
+      "step": 29720
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.164460090303819e-05,
+      "loss": 1.5079,
+      "step": 29730
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.164345297313843e-05,
+      "loss": 1.5974,
+      "step": 29740
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.164230504323869e-05,
+      "loss": 1.8618,
+      "step": 29750
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.164115711333894e-05,
+      "loss": 1.5317,
+      "step": 29760
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.164000918343919e-05,
+      "loss": 1.8154,
+      "step": 29770
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.163886125353944e-05,
+      "loss": 1.6833,
+      "step": 29780
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.16377133236397e-05,
+      "loss": 1.5598,
+      "step": 29790
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.163656539373995e-05,
+      "loss": 1.6283,
+      "step": 29800
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.16354174638402e-05,
+      "loss": 1.6343,
+      "step": 29810
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.163426953394045e-05,
+      "loss": 1.8091,
+      "step": 29820
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.16331216040407e-05,
+      "loss": 1.6871,
+      "step": 29830
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.163197367414097e-05,
+      "loss": 1.541,
+      "step": 29840
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.163082574424122e-05,
+      "loss": 1.5119,
+      "step": 29850
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.162967781434147e-05,
+      "loss": 1.7766,
+      "step": 29860
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.162852988444172e-05,
+      "loss": 1.4752,
+      "step": 29870
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.162738195454198e-05,
+      "loss": 1.6976,
+      "step": 29880
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.162623402464222e-05,
+      "loss": 1.6189,
+      "step": 29890
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.162508609474248e-05,
+      "loss": 1.6144,
+      "step": 29900
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.162393816484273e-05,
+      "loss": 1.5806,
+      "step": 29910
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.162279023494298e-05,
+      "loss": 1.5447,
+      "step": 29920
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.162164230504323e-05,
+      "loss": 1.7286,
+      "step": 29930
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.162049437514349e-05,
+      "loss": 1.8954,
+      "step": 29940
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.161934644524374e-05,
+      "loss": 1.8929,
+      "step": 29950
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.161819851534399e-05,
+      "loss": 1.5495,
+      "step": 29960
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.161705058544424e-05,
+      "loss": 1.5774,
+      "step": 29970
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.161590265554449e-05,
+      "loss": 1.7271,
+      "step": 29980
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.161475472564476e-05,
+      "loss": 1.6238,
+      "step": 29990
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.1613606795745e-05,
+      "loss": 1.7376,
+      "step": 30000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.161245886584526e-05,
+      "loss": 1.5599,
+      "step": 30010
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.16113109359455e-05,
+      "loss": 1.7459,
+      "step": 30020
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.161016300604575e-05,
+      "loss": 1.6748,
+      "step": 30030
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.1609015076146e-05,
+      "loss": 1.6762,
+      "step": 30040
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.160786714624627e-05,
+      "loss": 1.798,
+      "step": 30050
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.160671921634652e-05,
+      "loss": 1.6256,
+      "step": 30060
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.160557128644677e-05,
+      "loss": 1.6561,
+      "step": 30070
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.160442335654702e-05,
+      "loss": 1.7527,
+      "step": 30080
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.160327542664727e-05,
+      "loss": 1.5708,
+      "step": 30090
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.160212749674753e-05,
+      "loss": 1.6828,
+      "step": 30100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.160097956684778e-05,
+      "loss": 1.5059,
+      "step": 30110
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.159983163694803e-05,
+      "loss": 1.8011,
+      "step": 30120
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.159868370704828e-05,
+      "loss": 1.6049,
+      "step": 30130
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.159753577714854e-05,
+      "loss": 1.6185,
+      "step": 30140
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.159638784724878e-05,
+      "loss": 1.694,
+      "step": 30150
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.159523991734904e-05,
+      "loss": 1.4446,
+      "step": 30160
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.15940919874493e-05,
+      "loss": 1.6979,
+      "step": 30170
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.159294405754954e-05,
+      "loss": 1.707,
+      "step": 30180
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.15917961276498e-05,
+      "loss": 1.8113,
+      "step": 30190
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.159064819775006e-05,
+      "loss": 1.6193,
+      "step": 30200
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.15895002678503e-05,
+      "loss": 1.6555,
+      "step": 30210
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.158835233795056e-05,
+      "loss": 1.5542,
+      "step": 30220
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.15872044080508e-05,
+      "loss": 1.732,
+      "step": 30230
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.158605647815106e-05,
+      "loss": 1.5982,
+      "step": 30240
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.158490854825132e-05,
+      "loss": 1.5504,
+      "step": 30250
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.158376061835157e-05,
+      "loss": 1.5305,
+      "step": 30260
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.158261268845182e-05,
+      "loss": 1.6818,
+      "step": 30270
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.158146475855207e-05,
+      "loss": 1.6038,
+      "step": 30280
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.158031682865233e-05,
+      "loss": 1.5762,
+      "step": 30290
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.157916889875257e-05,
+      "loss": 1.5594,
+      "step": 30300
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.157802096885283e-05,
+      "loss": 1.6875,
+      "step": 30310
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.157687303895308e-05,
+      "loss": 1.6312,
+      "step": 30320
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.157572510905333e-05,
+      "loss": 1.8426,
+      "step": 30330
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.157457717915358e-05,
+      "loss": 1.7041,
+      "step": 30340
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.157342924925385e-05,
+      "loss": 1.623,
+      "step": 30350
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.15722813193541e-05,
+      "loss": 1.6185,
+      "step": 30360
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.157113338945434e-05,
+      "loss": 1.6719,
+      "step": 30370
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.15699854595546e-05,
+      "loss": 1.626,
+      "step": 30380
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.156883752965484e-05,
+      "loss": 1.7165,
+      "step": 30390
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.156768959975511e-05,
+      "loss": 1.6552,
+      "step": 30400
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.156654166985536e-05,
+      "loss": 1.8953,
+      "step": 30410
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.156539373995561e-05,
+      "loss": 1.7917,
+      "step": 30420
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.156424581005586e-05,
+      "loss": 1.6832,
+      "step": 30430
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.156309788015612e-05,
+      "loss": 1.6342,
+      "step": 30440
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.156194995025636e-05,
+      "loss": 1.4769,
+      "step": 30450
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.156080202035662e-05,
+      "loss": 1.6693,
+      "step": 30460
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.155965409045687e-05,
+      "loss": 1.7266,
+      "step": 30470
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.155850616055712e-05,
+      "loss": 1.7314,
+      "step": 30480
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.155735823065737e-05,
+      "loss": 1.7,
+      "step": 30490
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.155621030075763e-05,
+      "loss": 1.6201,
+      "step": 30500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.155506237085788e-05,
+      "loss": 1.6646,
+      "step": 30510
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.155391444095813e-05,
+      "loss": 1.5529,
+      "step": 30520
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.155276651105838e-05,
+      "loss": 1.7494,
+      "step": 30530
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.155161858115863e-05,
+      "loss": 1.5203,
+      "step": 30540
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.15504706512589e-05,
+      "loss": 1.749,
+      "step": 30550
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.154932272135915e-05,
+      "loss": 1.5094,
+      "step": 30560
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.15481747914594e-05,
+      "loss": 1.7297,
+      "step": 30570
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.154702686155965e-05,
+      "loss": 1.6311,
+      "step": 30580
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.154587893165991e-05,
+      "loss": 1.5454,
+      "step": 30590
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.154473100176015e-05,
+      "loss": 1.3899,
+      "step": 30600
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.154358307186041e-05,
+      "loss": 1.8404,
+      "step": 30610
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.154243514196066e-05,
+      "loss": 1.6156,
+      "step": 30620
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.154128721206091e-05,
+      "loss": 1.518,
+      "step": 30630
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.154013928216116e-05,
+      "loss": 1.7884,
+      "step": 30640
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.153899135226142e-05,
+      "loss": 1.5136,
+      "step": 30650
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.153784342236167e-05,
+      "loss": 1.6897,
+      "step": 30660
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.153669549246192e-05,
+      "loss": 1.5768,
+      "step": 30670
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.153554756256217e-05,
+      "loss": 1.6454,
+      "step": 30680
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.153439963266242e-05,
+      "loss": 1.5994,
+      "step": 30690
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.153325170276268e-05,
+      "loss": 1.6314,
+      "step": 30700
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.153210377286293e-05,
+      "loss": 1.7212,
+      "step": 30710
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.153095584296318e-05,
+      "loss": 1.5593,
+      "step": 30720
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.152980791306343e-05,
+      "loss": 1.7284,
+      "step": 30730
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.15286599831637e-05,
+      "loss": 1.6126,
+      "step": 30740
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.152751205326393e-05,
+      "loss": 1.4686,
+      "step": 30750
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.15263641233642e-05,
+      "loss": 1.7337,
+      "step": 30760
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.152521619346445e-05,
+      "loss": 1.8224,
+      "step": 30770
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.15240682635647e-05,
+      "loss": 1.7828,
+      "step": 30780
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.152292033366495e-05,
+      "loss": 1.6947,
+      "step": 30790
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.152177240376521e-05,
+      "loss": 1.4799,
+      "step": 30800
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.152062447386546e-05,
+      "loss": 1.6566,
+      "step": 30810
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.151947654396571e-05,
+      "loss": 1.8069,
+      "step": 30820
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.151832861406596e-05,
+      "loss": 1.5764,
+      "step": 30830
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.151718068416621e-05,
+      "loss": 1.6822,
+      "step": 30840
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.151603275426647e-05,
+      "loss": 1.6923,
+      "step": 30850
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.151488482436672e-05,
+      "loss": 1.85,
+      "step": 30860
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.151373689446697e-05,
+      "loss": 1.6024,
+      "step": 30870
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.151258896456722e-05,
+      "loss": 1.7491,
+      "step": 30880
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.151144103466749e-05,
+      "loss": 1.5697,
+      "step": 30890
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.151029310476772e-05,
+      "loss": 1.8215,
+      "step": 30900
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.150914517486799e-05,
+      "loss": 1.5942,
+      "step": 30910
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.150799724496824e-05,
+      "loss": 1.6798,
+      "step": 30920
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.150684931506849e-05,
+      "loss": 1.7445,
+      "step": 30930
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.150570138516874e-05,
+      "loss": 1.6774,
+      "step": 30940
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.1504553455269e-05,
+      "loss": 1.8219,
+      "step": 30950
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.150340552536925e-05,
+      "loss": 1.687,
+      "step": 30960
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.15022575954695e-05,
+      "loss": 1.9059,
+      "step": 30970
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.150110966556975e-05,
+      "loss": 1.6801,
+      "step": 30980
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.149996173567e-05,
+      "loss": 1.7882,
+      "step": 30990
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.149881380577026e-05,
+      "loss": 1.6131,
+      "step": 31000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.149766587587051e-05,
+      "loss": 1.8189,
+      "step": 31010
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.149651794597076e-05,
+      "loss": 1.6817,
+      "step": 31020
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.149537001607101e-05,
+      "loss": 1.7948,
+      "step": 31030
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.149422208617127e-05,
+      "loss": 1.7671,
+      "step": 31040
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.149307415627151e-05,
+      "loss": 1.6627,
+      "step": 31050
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.149192622637177e-05,
+      "loss": 1.5941,
+      "step": 31060
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.149077829647202e-05,
+      "loss": 1.8713,
+      "step": 31070
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.148963036657227e-05,
+      "loss": 1.5807,
+      "step": 31080
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.148848243667252e-05,
+      "loss": 1.5402,
+      "step": 31090
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.148733450677279e-05,
+      "loss": 1.5557,
+      "step": 31100
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.148618657687304e-05,
+      "loss": 1.5455,
+      "step": 31110
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.148503864697329e-05,
+      "loss": 1.7316,
+      "step": 31120
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.148389071707354e-05,
+      "loss": 1.5077,
+      "step": 31130
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.148274278717379e-05,
+      "loss": 1.5297,
+      "step": 31140
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.148159485727405e-05,
+      "loss": 1.6998,
+      "step": 31150
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.14804469273743e-05,
+      "loss": 1.7204,
+      "step": 31160
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.147929899747455e-05,
+      "loss": 1.807,
+      "step": 31170
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.14781510675748e-05,
+      "loss": 1.5754,
+      "step": 31180
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.147700313767506e-05,
+      "loss": 1.6747,
+      "step": 31190
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.14758552077753e-05,
+      "loss": 1.7329,
+      "step": 31200
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.147470727787556e-05,
+      "loss": 1.6807,
+      "step": 31210
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.147355934797581e-05,
+      "loss": 1.6102,
+      "step": 31220
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.147241141807606e-05,
+      "loss": 1.6597,
+      "step": 31230
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.147126348817631e-05,
+      "loss": 1.6951,
+      "step": 31240
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.147011555827658e-05,
+      "loss": 1.5983,
+      "step": 31250
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.146896762837683e-05,
+      "loss": 1.6145,
+      "step": 31260
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.146781969847708e-05,
+      "loss": 1.6158,
+      "step": 31270
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.146667176857733e-05,
+      "loss": 1.776,
+      "step": 31280
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.146552383867758e-05,
+      "loss": 1.5145,
+      "step": 31290
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.146437590877784e-05,
+      "loss": 1.3503,
+      "step": 31300
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.146322797887809e-05,
+      "loss": 1.662,
+      "step": 31310
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.146208004897834e-05,
+      "loss": 1.5609,
+      "step": 31320
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.146093211907859e-05,
+      "loss": 1.7897,
+      "step": 31330
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.145978418917885e-05,
+      "loss": 1.7345,
+      "step": 31340
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.145863625927909e-05,
+      "loss": 1.6541,
+      "step": 31350
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.145748832937935e-05,
+      "loss": 1.5791,
+      "step": 31360
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.14563403994796e-05,
+      "loss": 1.4982,
+      "step": 31370
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.145519246957985e-05,
+      "loss": 1.4718,
+      "step": 31380
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.14540445396801e-05,
+      "loss": 1.5594,
+      "step": 31390
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.145289660978035e-05,
+      "loss": 1.6326,
+      "step": 31400
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.145174867988061e-05,
+      "loss": 1.6541,
+      "step": 31410
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.145060074998086e-05,
+      "loss": 1.6665,
+      "step": 31420
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.144945282008111e-05,
+      "loss": 1.7475,
+      "step": 31430
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.144830489018136e-05,
+      "loss": 1.7475,
+      "step": 31440
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.144715696028163e-05,
+      "loss": 1.6457,
+      "step": 31450
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.144600903038186e-05,
+      "loss": 1.6183,
+      "step": 31460
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.144486110048213e-05,
+      "loss": 1.6654,
+      "step": 31470
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.144371317058238e-05,
+      "loss": 1.6337,
+      "step": 31480
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.144256524068263e-05,
+      "loss": 1.693,
+      "step": 31490
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.144141731078288e-05,
+      "loss": 1.7698,
+      "step": 31500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.144026938088314e-05,
+      "loss": 1.6522,
+      "step": 31510
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.143912145098339e-05,
+      "loss": 1.6327,
+      "step": 31520
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.143797352108364e-05,
+      "loss": 1.6137,
+      "step": 31530
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.143682559118389e-05,
+      "loss": 1.7786,
+      "step": 31540
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.143567766128414e-05,
+      "loss": 1.6544,
+      "step": 31550
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.14345297313844e-05,
+      "loss": 1.701,
+      "step": 31560
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.143338180148465e-05,
+      "loss": 1.6187,
+      "step": 31570
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.14322338715849e-05,
+      "loss": 1.4989,
+      "step": 31580
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.143108594168515e-05,
+      "loss": 1.4888,
+      "step": 31590
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.142993801178542e-05,
+      "loss": 1.7661,
+      "step": 31600
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.142879008188565e-05,
+      "loss": 1.5894,
+      "step": 31610
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.142764215198592e-05,
+      "loss": 1.5513,
+      "step": 31620
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.142649422208617e-05,
+      "loss": 1.657,
+      "step": 31630
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.142534629218642e-05,
+      "loss": 1.6335,
+      "step": 31640
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.142419836228667e-05,
+      "loss": 1.5377,
+      "step": 31650
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.142305043238693e-05,
+      "loss": 1.6562,
+      "step": 31660
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.142190250248718e-05,
+      "loss": 1.6059,
+      "step": 31670
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.142075457258743e-05,
+      "loss": 1.6641,
+      "step": 31680
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.141960664268768e-05,
+      "loss": 1.802,
+      "step": 31690
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.141845871278793e-05,
+      "loss": 1.4848,
+      "step": 31700
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.141731078288819e-05,
+      "loss": 1.5668,
+      "step": 31710
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.141616285298844e-05,
+      "loss": 1.64,
+      "step": 31720
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.141501492308869e-05,
+      "loss": 1.8219,
+      "step": 31730
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.141386699318894e-05,
+      "loss": 1.5172,
+      "step": 31740
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.14127190632892e-05,
+      "loss": 1.5614,
+      "step": 31750
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.141157113338944e-05,
+      "loss": 1.6526,
+      "step": 31760
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.14104232034897e-05,
+      "loss": 1.52,
+      "step": 31770
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.140927527358995e-05,
+      "loss": 1.432,
+      "step": 31780
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.14081273436902e-05,
+      "loss": 1.4419,
+      "step": 31790
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.140697941379045e-05,
+      "loss": 1.6927,
+      "step": 31800
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.140583148389072e-05,
+      "loss": 1.5922,
+      "step": 31810
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.140468355399097e-05,
+      "loss": 1.8625,
+      "step": 31820
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.140353562409122e-05,
+      "loss": 1.6607,
+      "step": 31830
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.140238769419147e-05,
+      "loss": 1.7179,
+      "step": 31840
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.140123976429172e-05,
+      "loss": 1.6887,
+      "step": 31850
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.140009183439198e-05,
+      "loss": 1.6267,
+      "step": 31860
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.139894390449223e-05,
+      "loss": 1.5799,
+      "step": 31870
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.139779597459248e-05,
+      "loss": 1.6267,
+      "step": 31880
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.139664804469273e-05,
+      "loss": 1.6495,
+      "step": 31890
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.139550011479299e-05,
+      "loss": 1.6716,
+      "step": 31900
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.139435218489323e-05,
+      "loss": 1.49,
+      "step": 31910
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.139320425499349e-05,
+      "loss": 1.7872,
+      "step": 31920
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.139205632509374e-05,
+      "loss": 1.8457,
+      "step": 31930
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.139090839519399e-05,
+      "loss": 1.5505,
+      "step": 31940
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.138976046529424e-05,
+      "loss": 1.7133,
+      "step": 31950
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.13886125353945e-05,
+      "loss": 1.6828,
+      "step": 31960
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.138746460549476e-05,
+      "loss": 1.8357,
+      "step": 31970
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.1386316675595e-05,
+      "loss": 1.5904,
+      "step": 31980
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.138516874569525e-05,
+      "loss": 1.5701,
+      "step": 31990
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.13840208157955e-05,
+      "loss": 1.7282,
+      "step": 32000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.138287288589577e-05,
+      "loss": 1.6414,
+      "step": 32010
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.138172495599602e-05,
+      "loss": 1.651,
+      "step": 32020
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.138057702609627e-05,
+      "loss": 1.5805,
+      "step": 32030
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.137942909619652e-05,
+      "loss": 1.3968,
+      "step": 32040
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.137828116629678e-05,
+      "loss": 1.6465,
+      "step": 32050
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.137713323639702e-05,
+      "loss": 1.4138,
+      "step": 32060
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.137598530649728e-05,
+      "loss": 1.644,
+      "step": 32070
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.137483737659753e-05,
+      "loss": 1.8728,
+      "step": 32080
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.137368944669778e-05,
+      "loss": 1.5608,
+      "step": 32090
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.137254151679803e-05,
+      "loss": 1.56,
+      "step": 32100
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.13713935868983e-05,
+      "loss": 1.6661,
+      "step": 32110
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.137024565699854e-05,
+      "loss": 1.6293,
+      "step": 32120
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.13690977270988e-05,
+      "loss": 1.7332,
+      "step": 32130
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.136794979719904e-05,
+      "loss": 1.7266,
+      "step": 32140
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.13668018672993e-05,
+      "loss": 1.6778,
+      "step": 32150
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.136565393739956e-05,
+      "loss": 1.7466,
+      "step": 32160
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.13645060074998e-05,
+      "loss": 1.5653,
+      "step": 32170
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.136335807760006e-05,
+      "loss": 1.5697,
+      "step": 32180
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.13622101477003e-05,
+      "loss": 1.7221,
+      "step": 32190
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.136106221780057e-05,
+      "loss": 1.6379,
+      "step": 32200
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.13599142879008e-05,
+      "loss": 1.6812,
+      "step": 32210
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.135876635800107e-05,
+      "loss": 1.7181,
+      "step": 32220
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.135761842810132e-05,
+      "loss": 1.5602,
+      "step": 32230
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.135647049820157e-05,
+      "loss": 1.577,
+      "step": 32240
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.135532256830182e-05,
+      "loss": 1.5889,
+      "step": 32250
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.135417463840208e-05,
+      "loss": 1.5674,
+      "step": 32260
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.135302670850233e-05,
+      "loss": 1.5238,
+      "step": 32270
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.135187877860258e-05,
+      "loss": 1.7627,
+      "step": 32280
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.135073084870283e-05,
+      "loss": 1.7731,
+      "step": 32290
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.134958291880308e-05,
+      "loss": 1.7652,
+      "step": 32300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.134843498890335e-05,
+      "loss": 1.9462,
+      "step": 32310
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.13472870590036e-05,
+      "loss": 1.6535,
+      "step": 32320
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.134613912910384e-05,
+      "loss": 1.4856,
+      "step": 32330
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.13449911992041e-05,
+      "loss": 1.6156,
+      "step": 32340
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.134384326930436e-05,
+      "loss": 1.6448,
+      "step": 32350
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.13426953394046e-05,
+      "loss": 1.5416,
+      "step": 32360
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.134154740950486e-05,
+      "loss": 1.7043,
+      "step": 32370
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.134039947960511e-05,
+      "loss": 1.639,
+      "step": 32380
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.133925154970536e-05,
+      "loss": 1.6751,
+      "step": 32390
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.133810361980561e-05,
+      "loss": 1.5598,
+      "step": 32400
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.133695568990587e-05,
+      "loss": 1.7473,
+      "step": 32410
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.133580776000612e-05,
+      "loss": 1.6559,
+      "step": 32420
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.133465983010637e-05,
+      "loss": 1.544,
+      "step": 32430
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.133351190020662e-05,
+      "loss": 1.7831,
+      "step": 32440
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.133236397030687e-05,
+      "loss": 1.615,
+      "step": 32450
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.133121604040713e-05,
+      "loss": 1.6571,
+      "step": 32460
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.133006811050738e-05,
+      "loss": 1.7647,
+      "step": 32470
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.132892018060763e-05,
+      "loss": 1.7839,
+      "step": 32480
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.132777225070788e-05,
+      "loss": 1.682,
+      "step": 32490
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.132662432080815e-05,
+      "loss": 1.8066,
+      "step": 32500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.132547639090838e-05,
+      "loss": 1.7624,
+      "step": 32510
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.132432846100865e-05,
+      "loss": 1.7222,
+      "step": 32520
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.13231805311089e-05,
+      "loss": 1.8306,
+      "step": 32530
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.132203260120915e-05,
+      "loss": 1.6035,
+      "step": 32540
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.13208846713094e-05,
+      "loss": 1.6149,
+      "step": 32550
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.131973674140966e-05,
+      "loss": 1.6406,
+      "step": 32560
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.131858881150991e-05,
+      "loss": 1.6482,
+      "step": 32570
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.131744088161016e-05,
+      "loss": 1.7508,
+      "step": 32580
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.131629295171041e-05,
+      "loss": 1.6239,
+      "step": 32590
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.131514502181066e-05,
+      "loss": 1.5567,
+      "step": 32600
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.131399709191092e-05,
+      "loss": 1.6279,
+      "step": 32610
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.131284916201117e-05,
+      "loss": 1.6957,
+      "step": 32620
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.131170123211142e-05,
+      "loss": 1.6234,
+      "step": 32630
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.131055330221167e-05,
+      "loss": 1.623,
+      "step": 32640
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.130940537231194e-05,
+      "loss": 1.6242,
+      "step": 32650
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.130825744241217e-05,
+      "loss": 1.5564,
+      "step": 32660
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.130710951251243e-05,
+      "loss": 1.4828,
+      "step": 32670
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.130596158261268e-05,
+      "loss": 1.5787,
+      "step": 32680
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.130481365271293e-05,
+      "loss": 1.7666,
+      "step": 32690
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.130366572281318e-05,
+      "loss": 1.6694,
+      "step": 32700
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.130251779291345e-05,
+      "loss": 1.6608,
+      "step": 32710
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.13013698630137e-05,
+      "loss": 1.5462,
+      "step": 32720
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.130022193311395e-05,
+      "loss": 1.5954,
+      "step": 32730
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.12990740032142e-05,
+      "loss": 1.7091,
+      "step": 32740
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.129792607331445e-05,
+      "loss": 1.7618,
+      "step": 32750
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.129677814341471e-05,
+      "loss": 1.6084,
+      "step": 32760
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.129563021351495e-05,
+      "loss": 1.7948,
+      "step": 32770
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.129448228361521e-05,
+      "loss": 1.5566,
+      "step": 32780
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.129333435371546e-05,
+      "loss": 1.5412,
+      "step": 32790
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.129218642381571e-05,
+      "loss": 1.6301,
+      "step": 32800
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.129103849391596e-05,
+      "loss": 1.7156,
+      "step": 32810
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.128989056401622e-05,
+      "loss": 1.5351,
+      "step": 32820
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.128874263411647e-05,
+      "loss": 1.5894,
+      "step": 32830
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.128759470421672e-05,
+      "loss": 1.5234,
+      "step": 32840
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.128644677431697e-05,
+      "loss": 1.6364,
+      "step": 32850
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.128529884441722e-05,
+      "loss": 1.7232,
+      "step": 32860
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.128415091451749e-05,
+      "loss": 1.5413,
+      "step": 32870
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.128300298461774e-05,
+      "loss": 1.5686,
+      "step": 32880
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.128185505471799e-05,
+      "loss": 1.5938,
+      "step": 32890
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.128070712481824e-05,
+      "loss": 1.6271,
+      "step": 32900
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.12795591949185e-05,
+      "loss": 1.5883,
+      "step": 32910
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.127841126501874e-05,
+      "loss": 1.5095,
+      "step": 32920
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.1277263335119e-05,
+      "loss": 1.6999,
+      "step": 32930
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.127611540521925e-05,
+      "loss": 1.605,
+      "step": 32940
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.12749674753195e-05,
+      "loss": 1.616,
+      "step": 32950
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.127381954541975e-05,
+      "loss": 1.7732,
+      "step": 32960
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.127267161552001e-05,
+      "loss": 1.6238,
+      "step": 32970
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.127152368562026e-05,
+      "loss": 1.6673,
+      "step": 32980
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.127037575572051e-05,
+      "loss": 1.7748,
+      "step": 32990
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.126922782582076e-05,
+      "loss": 1.51,
+      "step": 33000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.126807989592101e-05,
+      "loss": 1.6195,
+      "step": 33010
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.126693196602127e-05,
+      "loss": 1.6135,
+      "step": 33020
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.126578403612152e-05,
+      "loss": 1.6637,
+      "step": 33030
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.126463610622177e-05,
+      "loss": 1.6668,
+      "step": 33040
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.126348817632202e-05,
+      "loss": 1.5832,
+      "step": 33050
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.126234024642229e-05,
+      "loss": 1.9808,
+      "step": 33060
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.126119231652252e-05,
+      "loss": 1.613,
+      "step": 33070
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.126004438662279e-05,
+      "loss": 1.8166,
+      "step": 33080
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.125889645672304e-05,
+      "loss": 1.5319,
+      "step": 33090
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.125774852682329e-05,
+      "loss": 1.6753,
+      "step": 33100
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.125660059692354e-05,
+      "loss": 1.6337,
+      "step": 33110
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.12554526670238e-05,
+      "loss": 1.5646,
+      "step": 33120
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.125430473712405e-05,
+      "loss": 1.911,
+      "step": 33130
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.12531568072243e-05,
+      "loss": 1.5406,
+      "step": 33140
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.125200887732455e-05,
+      "loss": 1.5354,
+      "step": 33150
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.12508609474248e-05,
+      "loss": 1.7196,
+      "step": 33160
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.124971301752506e-05,
+      "loss": 1.6493,
+      "step": 33170
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.124856508762531e-05,
+      "loss": 1.5729,
+      "step": 33180
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.124741715772556e-05,
+      "loss": 1.6328,
+      "step": 33190
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.124626922782581e-05,
+      "loss": 1.5788,
+      "step": 33200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.124512129792608e-05,
+      "loss": 1.6227,
+      "step": 33210
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.124397336802631e-05,
+      "loss": 1.4578,
+      "step": 33220
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.124282543812658e-05,
+      "loss": 1.6702,
+      "step": 33230
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.124167750822683e-05,
+      "loss": 1.5609,
+      "step": 33240
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.124052957832708e-05,
+      "loss": 1.6657,
+      "step": 33250
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.123938164842733e-05,
+      "loss": 1.5855,
+      "step": 33260
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.123823371852759e-05,
+      "loss": 1.7732,
+      "step": 33270
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.123708578862784e-05,
+      "loss": 1.7943,
+      "step": 33280
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.123593785872809e-05,
+      "loss": 1.5744,
+      "step": 33290
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.123478992882834e-05,
+      "loss": 1.5653,
+      "step": 33300
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.123364199892859e-05,
+      "loss": 1.7441,
+      "step": 33310
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.123249406902885e-05,
+      "loss": 1.5847,
+      "step": 33320
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.12313461391291e-05,
+      "loss": 1.7674,
+      "step": 33330
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.123019820922935e-05,
+      "loss": 1.6724,
+      "step": 33340
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.12290502793296e-05,
+      "loss": 1.4739,
+      "step": 33350
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.122790234942986e-05,
+      "loss": 1.6461,
+      "step": 33360
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.12267544195301e-05,
+      "loss": 1.4741,
+      "step": 33370
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.122560648963036e-05,
+      "loss": 1.756,
+      "step": 33380
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.122445855973061e-05,
+      "loss": 1.7719,
+      "step": 33390
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.122331062983086e-05,
+      "loss": 1.7492,
+      "step": 33400
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.122216269993111e-05,
+      "loss": 1.8131,
+      "step": 33410
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.122101477003138e-05,
+      "loss": 1.6089,
+      "step": 33420
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.121986684013163e-05,
+      "loss": 1.6144,
+      "step": 33430
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.121871891023188e-05,
+      "loss": 1.7575,
+      "step": 33440
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.121757098033213e-05,
+      "loss": 1.7782,
+      "step": 33450
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.121642305043238e-05,
+      "loss": 1.8059,
+      "step": 33460
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.121527512053264e-05,
+      "loss": 1.7845,
+      "step": 33470
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.121412719063289e-05,
+      "loss": 1.3339,
+      "step": 33480
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.121297926073314e-05,
+      "loss": 1.5949,
+      "step": 33490
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.121183133083339e-05,
+      "loss": 1.4933,
+      "step": 33500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.121068340093365e-05,
+      "loss": 1.5897,
+      "step": 33510
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.120953547103389e-05,
+      "loss": 1.593,
+      "step": 33520
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.120838754113415e-05,
+      "loss": 1.4385,
+      "step": 33530
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.12072396112344e-05,
+      "loss": 1.7575,
+      "step": 33540
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.120609168133465e-05,
+      "loss": 1.5721,
+      "step": 33550
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.12049437514349e-05,
+      "loss": 1.6791,
+      "step": 33560
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.120379582153517e-05,
+      "loss": 1.5032,
+      "step": 33570
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.120264789163542e-05,
+      "loss": 1.5385,
+      "step": 33580
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.120149996173567e-05,
+      "loss": 1.7834,
+      "step": 33590
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.120035203183592e-05,
+      "loss": 1.583,
+      "step": 33600
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.119920410193617e-05,
+      "loss": 1.661,
+      "step": 33610
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.119805617203643e-05,
+      "loss": 1.5394,
+      "step": 33620
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.119690824213668e-05,
+      "loss": 1.6022,
+      "step": 33630
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.119576031223693e-05,
+      "loss": 1.7478,
+      "step": 33640
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.119461238233718e-05,
+      "loss": 1.5973,
+      "step": 33650
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.119346445243744e-05,
+      "loss": 1.6382,
+      "step": 33660
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.119231652253768e-05,
+      "loss": 1.7365,
+      "step": 33670
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.119116859263794e-05,
+      "loss": 1.491,
+      "step": 33680
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.119002066273819e-05,
+      "loss": 1.5616,
+      "step": 33690
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.118887273283844e-05,
+      "loss": 1.5655,
+      "step": 33700
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.118772480293869e-05,
+      "loss": 1.6678,
+      "step": 33710
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.118657687303895e-05,
+      "loss": 1.6423,
+      "step": 33720
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.11854289431392e-05,
+      "loss": 1.697,
+      "step": 33730
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.118428101323945e-05,
+      "loss": 1.5883,
+      "step": 33740
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.11831330833397e-05,
+      "loss": 1.6674,
+      "step": 33750
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.118198515343995e-05,
+      "loss": 1.6095,
+      "step": 33760
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.118083722354022e-05,
+      "loss": 1.6301,
+      "step": 33770
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.117968929364047e-05,
+      "loss": 1.5253,
+      "step": 33780
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.117854136374072e-05,
+      "loss": 1.5668,
+      "step": 33790
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.117739343384097e-05,
+      "loss": 1.8764,
+      "step": 33800
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.117624550394123e-05,
+      "loss": 1.6325,
+      "step": 33810
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.117509757404147e-05,
+      "loss": 1.7767,
+      "step": 33820
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.117394964414173e-05,
+      "loss": 1.6608,
+      "step": 33830
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.117280171424198e-05,
+      "loss": 1.6015,
+      "step": 33840
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.117165378434223e-05,
+      "loss": 1.6139,
+      "step": 33850
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.117050585444248e-05,
+      "loss": 1.4551,
+      "step": 33860
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.116935792454274e-05,
+      "loss": 1.7232,
+      "step": 33870
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.116820999464299e-05,
+      "loss": 1.7057,
+      "step": 33880
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.116706206474324e-05,
+      "loss": 1.5436,
+      "step": 33890
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.116591413484349e-05,
+      "loss": 1.64,
+      "step": 33900
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.116476620494374e-05,
+      "loss": 1.8013,
+      "step": 33910
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.1163618275044e-05,
+      "loss": 1.6234,
+      "step": 33920
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.116247034514426e-05,
+      "loss": 1.5787,
+      "step": 33930
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.11613224152445e-05,
+      "loss": 1.666,
+      "step": 33940
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.116017448534475e-05,
+      "loss": 1.6718,
+      "step": 33950
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.115902655544502e-05,
+      "loss": 1.6309,
+      "step": 33960
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.115787862554525e-05,
+      "loss": 1.5891,
+      "step": 33970
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.115673069564552e-05,
+      "loss": 1.7488,
+      "step": 33980
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.115558276574577e-05,
+      "loss": 1.9937,
+      "step": 33990
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.115443483584602e-05,
+      "loss": 1.8476,
+      "step": 34000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.115328690594627e-05,
+      "loss": 1.5153,
+      "step": 34010
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.115213897604653e-05,
+      "loss": 1.6153,
+      "step": 34020
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.115099104614678e-05,
+      "loss": 1.6496,
+      "step": 34030
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.114984311624703e-05,
+      "loss": 1.5999,
+      "step": 34040
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.114869518634728e-05,
+      "loss": 1.6764,
+      "step": 34050
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.114754725644753e-05,
+      "loss": 1.7442,
+      "step": 34060
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.11463993265478e-05,
+      "loss": 1.6012,
+      "step": 34070
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.114525139664804e-05,
+      "loss": 1.6345,
+      "step": 34080
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.11441034667483e-05,
+      "loss": 1.547,
+      "step": 34090
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.114295553684854e-05,
+      "loss": 1.7415,
+      "step": 34100
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.114180760694881e-05,
+      "loss": 1.5023,
+      "step": 34110
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.114065967704904e-05,
+      "loss": 1.8191,
+      "step": 34120
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.11395117471493e-05,
+      "loss": 1.7237,
+      "step": 34130
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.113836381724956e-05,
+      "loss": 1.4958,
+      "step": 34140
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.11372158873498e-05,
+      "loss": 1.5855,
+      "step": 34150
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.113606795745006e-05,
+      "loss": 1.3591,
+      "step": 34160
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.11349200275503e-05,
+      "loss": 1.8671,
+      "step": 34170
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.113377209765057e-05,
+      "loss": 1.8108,
+      "step": 34180
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.113262416775082e-05,
+      "loss": 1.5149,
+      "step": 34190
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.113147623785107e-05,
+      "loss": 1.5408,
+      "step": 34200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.113032830795132e-05,
+      "loss": 1.6239,
+      "step": 34210
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.112918037805158e-05,
+      "loss": 1.6771,
+      "step": 34220
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.112803244815182e-05,
+      "loss": 1.8167,
+      "step": 34230
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.112688451825208e-05,
+      "loss": 1.5026,
+      "step": 34240
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.112573658835233e-05,
+      "loss": 1.5471,
+      "step": 34250
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.112458865845258e-05,
+      "loss": 1.8419,
+      "step": 34260
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.112344072855283e-05,
+      "loss": 1.4838,
+      "step": 34270
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.11222927986531e-05,
+      "loss": 1.4179,
+      "step": 34280
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.112114486875334e-05,
+      "loss": 1.6519,
+      "step": 34290
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.11199969388536e-05,
+      "loss": 1.6269,
+      "step": 34300
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.111884900895384e-05,
+      "loss": 1.4763,
+      "step": 34310
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.11177010790541e-05,
+      "loss": 1.7262,
+      "step": 34320
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.111655314915436e-05,
+      "loss": 1.7059,
+      "step": 34330
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.111540521925461e-05,
+      "loss": 1.7944,
+      "step": 34340
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.111425728935486e-05,
+      "loss": 1.6086,
+      "step": 34350
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.111310935945511e-05,
+      "loss": 1.6756,
+      "step": 34360
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.111196142955537e-05,
+      "loss": 1.6545,
+      "step": 34370
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.111081349965561e-05,
+      "loss": 1.5537,
+      "step": 34380
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.110966556975587e-05,
+      "loss": 1.7587,
+      "step": 34390
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.110851763985612e-05,
+      "loss": 1.5493,
+      "step": 34400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.110736970995637e-05,
+      "loss": 1.5593,
+      "step": 34410
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.110622178005662e-05,
+      "loss": 1.699,
+      "step": 34420
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.110507385015688e-05,
+      "loss": 1.4328,
+      "step": 34430
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.110392592025713e-05,
+      "loss": 1.6164,
+      "step": 34440
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.110277799035738e-05,
+      "loss": 1.5704,
+      "step": 34450
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.110163006045763e-05,
+      "loss": 1.6845,
+      "step": 34460
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.110048213055788e-05,
+      "loss": 1.7476,
+      "step": 34470
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.109933420065815e-05,
+      "loss": 1.5741,
+      "step": 34480
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.10981862707584e-05,
+      "loss": 1.5955,
+      "step": 34490
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.109703834085865e-05,
+      "loss": 1.8047,
+      "step": 34500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.10958904109589e-05,
+      "loss": 1.5164,
+      "step": 34510
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.109474248105916e-05,
+      "loss": 1.7563,
+      "step": 34520
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.10935945511594e-05,
+      "loss": 1.7722,
+      "step": 34530
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.109244662125966e-05,
+      "loss": 1.5634,
+      "step": 34540
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.109129869135991e-05,
+      "loss": 1.5352,
+      "step": 34550
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.109015076146016e-05,
+      "loss": 1.8178,
+      "step": 34560
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.108900283156041e-05,
+      "loss": 1.5856,
+      "step": 34570
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.108785490166067e-05,
+      "loss": 1.7543,
+      "step": 34580
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.108670697176092e-05,
+      "loss": 1.6409,
+      "step": 34590
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.108555904186117e-05,
+      "loss": 1.478,
+      "step": 34600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.108441111196142e-05,
+      "loss": 1.6505,
+      "step": 34610
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.108326318206167e-05,
+      "loss": 1.5425,
+      "step": 34620
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.108211525216193e-05,
+      "loss": 1.6999,
+      "step": 34630
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.108096732226218e-05,
+      "loss": 1.5555,
+      "step": 34640
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.107981939236243e-05,
+      "loss": 1.6994,
+      "step": 34650
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.107867146246268e-05,
+      "loss": 1.7332,
+      "step": 34660
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.107752353256295e-05,
+      "loss": 1.7216,
+      "step": 34670
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.107637560266318e-05,
+      "loss": 1.6742,
+      "step": 34680
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.107522767276345e-05,
+      "loss": 1.4947,
+      "step": 34690
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.10740797428637e-05,
+      "loss": 1.7416,
+      "step": 34700
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.107293181296395e-05,
+      "loss": 1.5951,
+      "step": 34710
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.10717838830642e-05,
+      "loss": 1.6862,
+      "step": 34720
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.107063595316446e-05,
+      "loss": 1.5597,
+      "step": 34730
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.106948802326471e-05,
+      "loss": 1.6271,
+      "step": 34740
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.106834009336496e-05,
+      "loss": 1.656,
+      "step": 34750
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.106719216346521e-05,
+      "loss": 1.6151,
+      "step": 34760
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.106604423356546e-05,
+      "loss": 1.6295,
+      "step": 34770
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.106489630366572e-05,
+      "loss": 1.5458,
+      "step": 34780
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.106374837376597e-05,
+      "loss": 1.5508,
+      "step": 34790
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.106260044386622e-05,
+      "loss": 1.6979,
+      "step": 34800
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.106145251396647e-05,
+      "loss": 1.468,
+      "step": 34810
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.106030458406674e-05,
+      "loss": 1.6236,
+      "step": 34820
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.105915665416697e-05,
+      "loss": 1.58,
+      "step": 34830
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.105800872426724e-05,
+      "loss": 1.7781,
+      "step": 34840
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.105686079436749e-05,
+      "loss": 1.5411,
+      "step": 34850
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.105571286446774e-05,
+      "loss": 1.6306,
+      "step": 34860
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.105456493456799e-05,
+      "loss": 1.5209,
+      "step": 34870
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.105341700466825e-05,
+      "loss": 1.6003,
+      "step": 34880
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.10522690747685e-05,
+      "loss": 1.4881,
+      "step": 34890
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.105112114486875e-05,
+      "loss": 1.5858,
+      "step": 34900
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.1049973214969e-05,
+      "loss": 1.7294,
+      "step": 34910
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.104882528506925e-05,
+      "loss": 1.6984,
+      "step": 34920
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.104767735516951e-05,
+      "loss": 1.6916,
+      "step": 34930
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.104652942526976e-05,
+      "loss": 1.617,
+      "step": 34940
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.104538149537001e-05,
+      "loss": 1.5928,
+      "step": 34950
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.104423356547026e-05,
+      "loss": 1.648,
+      "step": 34960
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.104308563557052e-05,
+      "loss": 1.6534,
+      "step": 34970
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.104193770567076e-05,
+      "loss": 1.5183,
+      "step": 34980
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.104078977577102e-05,
+      "loss": 1.5207,
+      "step": 34990
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.103964184587127e-05,
+      "loss": 1.595,
+      "step": 35000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.103849391597152e-05,
+      "loss": 1.5382,
+      "step": 35010
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.103734598607177e-05,
+      "loss": 1.8473,
+      "step": 35020
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.103619805617204e-05,
+      "loss": 1.4968,
+      "step": 35030
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.103505012627229e-05,
+      "loss": 1.6312,
+      "step": 35040
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.103390219637254e-05,
+      "loss": 1.8109,
+      "step": 35050
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.103275426647279e-05,
+      "loss": 1.7097,
+      "step": 35060
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.103160633657304e-05,
+      "loss": 1.6945,
+      "step": 35070
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.10304584066733e-05,
+      "loss": 1.7455,
+      "step": 35080
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.102931047677355e-05,
+      "loss": 1.4927,
+      "step": 35090
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.10281625468738e-05,
+      "loss": 1.5646,
+      "step": 35100
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.102701461697405e-05,
+      "loss": 1.4919,
+      "step": 35110
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.102586668707431e-05,
+      "loss": 1.6375,
+      "step": 35120
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.102471875717455e-05,
+      "loss": 1.7641,
+      "step": 35130
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.102357082727481e-05,
+      "loss": 1.6888,
+      "step": 35140
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.102242289737506e-05,
+      "loss": 1.6011,
+      "step": 35150
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.102127496747531e-05,
+      "loss": 1.4923,
+      "step": 35160
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.102012703757556e-05,
+      "loss": 1.7908,
+      "step": 35170
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.101897910767583e-05,
+      "loss": 1.9118,
+      "step": 35180
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.101783117777608e-05,
+      "loss": 1.6221,
+      "step": 35190
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.101668324787633e-05,
+      "loss": 1.606,
+      "step": 35200
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.101553531797658e-05,
+      "loss": 1.6244,
+      "step": 35210
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.101438738807683e-05,
+      "loss": 1.6009,
+      "step": 35220
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.101323945817709e-05,
+      "loss": 1.6629,
+      "step": 35230
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.101209152827734e-05,
+      "loss": 1.6676,
+      "step": 35240
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.101094359837759e-05,
+      "loss": 1.472,
+      "step": 35250
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.100979566847784e-05,
+      "loss": 1.5704,
+      "step": 35260
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.10086477385781e-05,
+      "loss": 1.4924,
+      "step": 35270
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.100749980867834e-05,
+      "loss": 1.7656,
+      "step": 35280
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.10063518787786e-05,
+      "loss": 1.7305,
+      "step": 35290
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.100520394887885e-05,
+      "loss": 1.514,
+      "step": 35300
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.10040560189791e-05,
+      "loss": 1.8518,
+      "step": 35310
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.100290808907935e-05,
+      "loss": 1.568,
+      "step": 35320
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.100176015917961e-05,
+      "loss": 1.5504,
+      "step": 35330
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.100061222927986e-05,
+      "loss": 1.5853,
+      "step": 35340
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.099946429938011e-05,
+      "loss": 1.5003,
+      "step": 35350
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.099831636948036e-05,
+      "loss": 1.5175,
+      "step": 35360
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.099716843958061e-05,
+      "loss": 1.7648,
+      "step": 35370
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.099602050968088e-05,
+      "loss": 1.6576,
+      "step": 35380
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.099487257978113e-05,
+      "loss": 1.5674,
+      "step": 35390
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.099372464988138e-05,
+      "loss": 1.6307,
+      "step": 35400
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.099257671998163e-05,
+      "loss": 1.4934,
+      "step": 35410
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.099142879008189e-05,
+      "loss": 1.795,
+      "step": 35420
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.099028086018213e-05,
+      "loss": 1.6758,
+      "step": 35430
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.098913293028239e-05,
+      "loss": 1.6257,
+      "step": 35440
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.098798500038264e-05,
+      "loss": 1.4872,
+      "step": 35450
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.098683707048289e-05,
+      "loss": 1.5669,
+      "step": 35460
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.098568914058314e-05,
+      "loss": 1.8047,
+      "step": 35470
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.09845412106834e-05,
+      "loss": 1.7988,
+      "step": 35480
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.098339328078365e-05,
+      "loss": 1.5984,
+      "step": 35490
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.09822453508839e-05,
+      "loss": 1.5152,
+      "step": 35500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.098109742098415e-05,
+      "loss": 1.8737,
+      "step": 35510
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.09799494910844e-05,
+      "loss": 1.6483,
+      "step": 35520
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.097880156118467e-05,
+      "loss": 1.5449,
+      "step": 35530
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.09776536312849e-05,
+      "loss": 1.508,
+      "step": 35540
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.097650570138517e-05,
+      "loss": 1.8204,
+      "step": 35550
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.097535777148542e-05,
+      "loss": 1.558,
+      "step": 35560
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.097420984158566e-05,
+      "loss": 1.6447,
+      "step": 35570
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.097306191168591e-05,
+      "loss": 1.6317,
+      "step": 35580
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.097191398178618e-05,
+      "loss": 1.5717,
+      "step": 35590
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.097076605188641e-05,
+      "loss": 1.7396,
+      "step": 35600
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.096961812198668e-05,
+      "loss": 1.652,
+      "step": 35610
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.096847019208693e-05,
+      "loss": 1.6489,
+      "step": 35620
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.096732226218718e-05,
+      "loss": 1.5094,
+      "step": 35630
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.096617433228744e-05,
+      "loss": 1.6934,
+      "step": 35640
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.096502640238769e-05,
+      "loss": 1.5354,
+      "step": 35650
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.096387847248794e-05,
+      "loss": 1.7028,
+      "step": 35660
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.096273054258819e-05,
+      "loss": 1.5686,
+      "step": 35670
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.096158261268845e-05,
+      "loss": 1.5244,
+      "step": 35680
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.096043468278869e-05,
+      "loss": 1.5716,
+      "step": 35690
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.095928675288895e-05,
+      "loss": 1.5609,
+      "step": 35700
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.09581388229892e-05,
+      "loss": 1.4365,
+      "step": 35710
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.095699089308945e-05,
+      "loss": 1.7719,
+      "step": 35720
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.09558429631897e-05,
+      "loss": 1.666,
+      "step": 35730
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.095469503328997e-05,
+      "loss": 1.5854,
+      "step": 35740
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.09535471033902e-05,
+      "loss": 1.5237,
+      "step": 35750
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.095239917349047e-05,
+      "loss": 1.7081,
+      "step": 35760
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.095125124359072e-05,
+      "loss": 1.702,
+      "step": 35770
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.095010331369097e-05,
+      "loss": 1.6725,
+      "step": 35780
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.094895538379123e-05,
+      "loss": 1.5228,
+      "step": 35790
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.094780745389148e-05,
+      "loss": 1.5301,
+      "step": 35800
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.094665952399173e-05,
+      "loss": 1.5519,
+      "step": 35810
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.094551159409198e-05,
+      "loss": 1.6078,
+      "step": 35820
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.094436366419224e-05,
+      "loss": 1.6398,
+      "step": 35830
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.094321573429248e-05,
+      "loss": 1.4446,
+      "step": 35840
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.094206780439274e-05,
+      "loss": 1.5206,
+      "step": 35850
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.094091987449299e-05,
+      "loss": 1.6557,
+      "step": 35860
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.093977194459324e-05,
+      "loss": 1.6674,
+      "step": 35870
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.093862401469349e-05,
+      "loss": 1.6172,
+      "step": 35880
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.093747608479376e-05,
+      "loss": 1.4007,
+      "step": 35890
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.093632815489399e-05,
+      "loss": 1.4672,
+      "step": 35900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.093518022499425e-05,
+      "loss": 1.6687,
+      "step": 35910
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.09340322950945e-05,
+      "loss": 1.5769,
+      "step": 35920
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.093288436519475e-05,
+      "loss": 1.5892,
+      "step": 35930
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.093173643529502e-05,
+      "loss": 1.6953,
+      "step": 35940
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.093058850539527e-05,
+      "loss": 1.7043,
+      "step": 35950
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.092944057549552e-05,
+      "loss": 1.614,
+      "step": 35960
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.092829264559577e-05,
+      "loss": 1.5213,
+      "step": 35970
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.092714471569603e-05,
+      "loss": 1.6077,
+      "step": 35980
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.092599678579627e-05,
+      "loss": 1.4832,
+      "step": 35990
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.092484885589653e-05,
+      "loss": 1.6439,
+      "step": 36000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.092370092599678e-05,
+      "loss": 1.6293,
+      "step": 36010
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.092255299609703e-05,
+      "loss": 1.4075,
+      "step": 36020
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.092140506619728e-05,
+      "loss": 1.7454,
+      "step": 36030
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.092025713629754e-05,
+      "loss": 1.6582,
+      "step": 36040
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.091910920639778e-05,
+      "loss": 1.7324,
+      "step": 36050
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.091796127649804e-05,
+      "loss": 1.597,
+      "step": 36060
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.09168133465983e-05,
+      "loss": 1.4861,
+      "step": 36070
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.091566541669854e-05,
+      "loss": 1.7839,
+      "step": 36080
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.09145174867988e-05,
+      "loss": 1.9108,
+      "step": 36090
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.091336955689906e-05,
+      "loss": 1.6302,
+      "step": 36100
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.09122216269993e-05,
+      "loss": 1.4782,
+      "step": 36110
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.091107369709956e-05,
+      "loss": 1.517,
+      "step": 36120
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.090992576719982e-05,
+      "loss": 1.6791,
+      "step": 36130
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.090877783730006e-05,
+      "loss": 1.7123,
+      "step": 36140
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.090762990740032e-05,
+      "loss": 1.6001,
+      "step": 36150
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.090648197750057e-05,
+      "loss": 1.7113,
+      "step": 36160
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.090533404760082e-05,
+      "loss": 1.6203,
+      "step": 36170
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.090418611770107e-05,
+      "loss": 1.4064,
+      "step": 36180
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.090303818780133e-05,
+      "loss": 1.6044,
+      "step": 36190
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.090189025790157e-05,
+      "loss": 1.6234,
+      "step": 36200
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.090074232800183e-05,
+      "loss": 1.7274,
+      "step": 36210
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.089959439810208e-05,
+      "loss": 1.6826,
+      "step": 36220
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.089844646820233e-05,
+      "loss": 1.638,
+      "step": 36230
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.08972985383026e-05,
+      "loss": 1.7217,
+      "step": 36240
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.089615060840284e-05,
+      "loss": 1.6659,
+      "step": 36250
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.08950026785031e-05,
+      "loss": 1.5878,
+      "step": 36260
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.089385474860334e-05,
+      "loss": 1.5637,
+      "step": 36270
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.089270681870361e-05,
+      "loss": 1.8441,
+      "step": 36280
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.089155888880384e-05,
+      "loss": 1.5772,
+      "step": 36290
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.089041095890411e-05,
+      "loss": 1.4971,
+      "step": 36300
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.088926302900436e-05,
+      "loss": 1.3881,
+      "step": 36310
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.088811509910461e-05,
+      "loss": 1.5486,
+      "step": 36320
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.088696716920486e-05,
+      "loss": 1.6677,
+      "step": 36330
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.088581923930512e-05,
+      "loss": 1.5321,
+      "step": 36340
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.088467130940536e-05,
+      "loss": 1.7149,
+      "step": 36350
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.088352337950562e-05,
+      "loss": 1.4283,
+      "step": 36360
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.088237544960587e-05,
+      "loss": 1.6005,
+      "step": 36370
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.088122751970612e-05,
+      "loss": 1.5054,
+      "step": 36380
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.088007958980638e-05,
+      "loss": 1.5242,
+      "step": 36390
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.087893165990663e-05,
+      "loss": 1.4981,
+      "step": 36400
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.087778373000688e-05,
+      "loss": 1.679,
+      "step": 36410
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.087663580010713e-05,
+      "loss": 1.6712,
+      "step": 36420
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.08754878702074e-05,
+      "loss": 1.4544,
+      "step": 36430
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.087433994030763e-05,
+      "loss": 1.7463,
+      "step": 36440
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.08731920104079e-05,
+      "loss": 1.7877,
+      "step": 36450
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.087204408050815e-05,
+      "loss": 1.6544,
+      "step": 36460
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.08708961506084e-05,
+      "loss": 1.6217,
+      "step": 36470
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.086974822070865e-05,
+      "loss": 1.6413,
+      "step": 36480
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.086860029080891e-05,
+      "loss": 1.5231,
+      "step": 36490
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.086745236090915e-05,
+      "loss": 1.5961,
+      "step": 36500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.086630443100941e-05,
+      "loss": 1.6123,
+      "step": 36510
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.086515650110966e-05,
+      "loss": 1.5776,
+      "step": 36520
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.086400857120991e-05,
+      "loss": 1.7064,
+      "step": 36530
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.086286064131017e-05,
+      "loss": 1.4436,
+      "step": 36540
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.086171271141042e-05,
+      "loss": 1.6155,
+      "step": 36550
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.086056478151067e-05,
+      "loss": 1.7527,
+      "step": 36560
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.085941685161092e-05,
+      "loss": 1.6662,
+      "step": 36570
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.085826892171118e-05,
+      "loss": 1.5591,
+      "step": 36580
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.085712099181142e-05,
+      "loss": 1.4321,
+      "step": 36590
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.085597306191168e-05,
+      "loss": 1.5321,
+      "step": 36600
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.085482513201193e-05,
+      "loss": 1.608,
+      "step": 36610
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.085367720211218e-05,
+      "loss": 1.5614,
+      "step": 36620
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.085252927221243e-05,
+      "loss": 1.3454,
+      "step": 36630
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.08513813423127e-05,
+      "loss": 1.668,
+      "step": 36640
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.085023341241293e-05,
+      "loss": 1.8036,
+      "step": 36650
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.08490854825132e-05,
+      "loss": 1.807,
+      "step": 36660
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.084793755261345e-05,
+      "loss": 1.6439,
+      "step": 36670
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.08467896227137e-05,
+      "loss": 1.5659,
+      "step": 36680
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.084564169281396e-05,
+      "loss": 1.5158,
+      "step": 36690
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.084449376291421e-05,
+      "loss": 1.626,
+      "step": 36700
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.084334583301446e-05,
+      "loss": 1.6546,
+      "step": 36710
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.084219790311471e-05,
+      "loss": 1.4848,
+      "step": 36720
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.084104997321497e-05,
+      "loss": 1.6084,
+      "step": 36730
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.083990204331521e-05,
+      "loss": 1.6768,
+      "step": 36740
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.083875411341547e-05,
+      "loss": 1.4316,
+      "step": 36750
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.083760618351572e-05,
+      "loss": 1.5048,
+      "step": 36760
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.083645825361597e-05,
+      "loss": 1.665,
+      "step": 36770
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.083531032371622e-05,
+      "loss": 1.7329,
+      "step": 36780
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.083416239381649e-05,
+      "loss": 1.6322,
+      "step": 36790
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.083301446391672e-05,
+      "loss": 1.4938,
+      "step": 36800
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.083186653401699e-05,
+      "loss": 1.5465,
+      "step": 36810
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.083071860411724e-05,
+      "loss": 1.3828,
+      "step": 36820
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.082957067421749e-05,
+      "loss": 1.5281,
+      "step": 36830
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.082842274431775e-05,
+      "loss": 1.554,
+      "step": 36840
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.0827274814418e-05,
+      "loss": 1.5386,
+      "step": 36850
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.082612688451825e-05,
+      "loss": 1.4277,
+      "step": 36860
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.08249789546185e-05,
+      "loss": 1.6481,
+      "step": 36870
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.082383102471875e-05,
+      "loss": 1.6798,
+      "step": 36880
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.0822683094819e-05,
+      "loss": 1.6743,
+      "step": 36890
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.082153516491926e-05,
+      "loss": 1.5998,
+      "step": 36900
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.08203872350195e-05,
+      "loss": 1.5217,
+      "step": 36910
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.081923930511976e-05,
+      "loss": 1.5788,
+      "step": 36920
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.081809137522001e-05,
+      "loss": 1.5432,
+      "step": 36930
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.081694344532026e-05,
+      "loss": 1.6435,
+      "step": 36940
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.081579551542051e-05,
+      "loss": 1.6339,
+      "step": 36950
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.081464758552077e-05,
+      "loss": 1.6042,
+      "step": 36960
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.081349965562102e-05,
+      "loss": 1.5928,
+      "step": 36970
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.081235172572127e-05,
+      "loss": 1.5948,
+      "step": 36980
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.081120379582154e-05,
+      "loss": 1.807,
+      "step": 36990
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.081005586592177e-05,
+      "loss": 1.5808,
+      "step": 37000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.080890793602204e-05,
+      "loss": 1.638,
+      "step": 37010
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.080776000612229e-05,
+      "loss": 1.6771,
+      "step": 37020
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.080661207622254e-05,
+      "loss": 1.7028,
+      "step": 37030
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.080546414632279e-05,
+      "loss": 1.5918,
+      "step": 37040
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.080431621642305e-05,
+      "loss": 1.5246,
+      "step": 37050
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.080316828652329e-05,
+      "loss": 1.5963,
+      "step": 37060
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.080202035662355e-05,
+      "loss": 1.6624,
+      "step": 37070
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.08008724267238e-05,
+      "loss": 1.6387,
+      "step": 37080
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.079972449682405e-05,
+      "loss": 1.7257,
+      "step": 37090
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.07985765669243e-05,
+      "loss": 1.4647,
+      "step": 37100
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.079742863702456e-05,
+      "loss": 1.5596,
+      "step": 37110
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.079628070712481e-05,
+      "loss": 1.5395,
+      "step": 37120
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.079513277722506e-05,
+      "loss": 1.5445,
+      "step": 37130
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.079398484732533e-05,
+      "loss": 1.3937,
+      "step": 37140
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.079283691742556e-05,
+      "loss": 1.5001,
+      "step": 37150
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.079168898752583e-05,
+      "loss": 1.5469,
+      "step": 37160
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.079054105762608e-05,
+      "loss": 1.7096,
+      "step": 37170
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.078939312772633e-05,
+      "loss": 1.6745,
+      "step": 37180
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.078824519782658e-05,
+      "loss": 1.6375,
+      "step": 37190
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.078709726792684e-05,
+      "loss": 1.6754,
+      "step": 37200
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.078594933802707e-05,
+      "loss": 1.6149,
+      "step": 37210
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.078480140812734e-05,
+      "loss": 1.7287,
+      "step": 37220
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.078365347822759e-05,
+      "loss": 1.6116,
+      "step": 37230
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.078250554832784e-05,
+      "loss": 1.6133,
+      "step": 37240
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.078135761842809e-05,
+      "loss": 1.4665,
+      "step": 37250
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.078020968852835e-05,
+      "loss": 1.5431,
+      "step": 37260
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.07790617586286e-05,
+      "loss": 1.6263,
+      "step": 37270
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.077791382872885e-05,
+      "loss": 1.568,
+      "step": 37280
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.077676589882911e-05,
+      "loss": 1.6078,
+      "step": 37290
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.077561796892935e-05,
+      "loss": 1.498,
+      "step": 37300
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.077447003902961e-05,
+      "loss": 1.7553,
+      "step": 37310
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.077332210912986e-05,
+      "loss": 1.4907,
+      "step": 37320
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.077217417923011e-05,
+      "loss": 1.4775,
+      "step": 37330
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.077102624933036e-05,
+      "loss": 1.7794,
+      "step": 37340
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.076987831943063e-05,
+      "loss": 1.6372,
+      "step": 37350
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.076873038953086e-05,
+      "loss": 1.6772,
+      "step": 37360
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.076758245963113e-05,
+      "loss": 1.701,
+      "step": 37370
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.076643452973138e-05,
+      "loss": 1.4183,
+      "step": 37380
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.076528659983163e-05,
+      "loss": 1.5313,
+      "step": 37390
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.076413866993188e-05,
+      "loss": 1.6767,
+      "step": 37400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.076299074003214e-05,
+      "loss": 1.5338,
+      "step": 37410
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.076184281013239e-05,
+      "loss": 1.4235,
+      "step": 37420
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.076069488023264e-05,
+      "loss": 1.4601,
+      "step": 37430
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.07595469503329e-05,
+      "loss": 1.7721,
+      "step": 37440
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.075839902043314e-05,
+      "loss": 1.4808,
+      "step": 37450
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.07572510905334e-05,
+      "loss": 1.5398,
+      "step": 37460
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.075610316063365e-05,
+      "loss": 1.7171,
+      "step": 37470
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.07549552307339e-05,
+      "loss": 1.4355,
+      "step": 37480
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.075380730083415e-05,
+      "loss": 1.6334,
+      "step": 37490
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.075265937093442e-05,
+      "loss": 1.6541,
+      "step": 37500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.075151144103465e-05,
+      "loss": 1.6485,
+      "step": 37510
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.075036351113492e-05,
+      "loss": 1.4931,
+      "step": 37520
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.074921558123516e-05,
+      "loss": 1.3963,
+      "step": 37530
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.074806765133541e-05,
+      "loss": 1.7676,
+      "step": 37540
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.074691972143566e-05,
+      "loss": 1.5377,
+      "step": 37550
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.074577179153593e-05,
+      "loss": 1.5594,
+      "step": 37560
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.074462386163618e-05,
+      "loss": 1.5426,
+      "step": 37570
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.074347593173643e-05,
+      "loss": 1.7161,
+      "step": 37580
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.074232800183669e-05,
+      "loss": 1.594,
+      "step": 37590
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.074118007193693e-05,
+      "loss": 1.5228,
+      "step": 37600
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.074003214203719e-05,
+      "loss": 1.4155,
+      "step": 37610
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.073888421213744e-05,
+      "loss": 1.4619,
+      "step": 37620
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.073773628223769e-05,
+      "loss": 1.5426,
+      "step": 37630
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.073658835233794e-05,
+      "loss": 1.7399,
+      "step": 37640
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.07354404224382e-05,
+      "loss": 1.7232,
+      "step": 37650
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.073429249253844e-05,
+      "loss": 1.5594,
+      "step": 37660
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.07331445626387e-05,
+      "loss": 1.5159,
+      "step": 37670
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.073199663273895e-05,
+      "loss": 1.6029,
+      "step": 37680
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.07308487028392e-05,
+      "loss": 1.623,
+      "step": 37690
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.072970077293945e-05,
+      "loss": 1.6482,
+      "step": 37700
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.072855284303972e-05,
+      "loss": 1.631,
+      "step": 37710
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.072740491313997e-05,
+      "loss": 1.6467,
+      "step": 37720
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.072625698324022e-05,
+      "loss": 1.6094,
+      "step": 37730
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.072510905334048e-05,
+      "loss": 1.4503,
+      "step": 37740
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.072396112344072e-05,
+      "loss": 1.6857,
+      "step": 37750
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.072281319354098e-05,
+      "loss": 1.5491,
+      "step": 37760
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.072166526364123e-05,
+      "loss": 1.5253,
+      "step": 37770
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.072051733374148e-05,
+      "loss": 1.5896,
+      "step": 37780
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.071936940384173e-05,
+      "loss": 1.5711,
+      "step": 37790
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.071822147394199e-05,
+      "loss": 1.5939,
+      "step": 37800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.071707354404223e-05,
+      "loss": 1.6929,
+      "step": 37810
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.071592561414249e-05,
+      "loss": 1.6531,
+      "step": 37820
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.071477768424274e-05,
+      "loss": 1.53,
+      "step": 37830
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.071362975434299e-05,
+      "loss": 1.5882,
+      "step": 37840
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.071248182444324e-05,
+      "loss": 1.5207,
+      "step": 37850
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.07113338945435e-05,
+      "loss": 1.581,
+      "step": 37860
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.071018596464375e-05,
+      "loss": 1.6294,
+      "step": 37870
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.0709038034744e-05,
+      "loss": 1.6726,
+      "step": 37880
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.070789010484427e-05,
+      "loss": 1.6601,
+      "step": 37890
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.07067421749445e-05,
+      "loss": 1.5996,
+      "step": 37900
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.070559424504477e-05,
+      "loss": 1.7116,
+      "step": 37910
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.070444631514502e-05,
+      "loss": 1.6231,
+      "step": 37920
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.070329838524527e-05,
+      "loss": 1.5893,
+      "step": 37930
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.070215045534552e-05,
+      "loss": 1.4943,
+      "step": 37940
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.070100252544578e-05,
+      "loss": 1.5008,
+      "step": 37950
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.069985459554602e-05,
+      "loss": 1.495,
+      "step": 37960
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.069870666564628e-05,
+      "loss": 1.5295,
+      "step": 37970
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.069755873574653e-05,
+      "loss": 1.6021,
+      "step": 37980
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.069641080584678e-05,
+      "loss": 1.526,
+      "step": 37990
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.069526287594703e-05,
+      "loss": 1.591,
+      "step": 38000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.06941149460473e-05,
+      "loss": 1.7289,
+      "step": 38010
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.069296701614754e-05,
+      "loss": 1.5225,
+      "step": 38020
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.06918190862478e-05,
+      "loss": 1.5612,
+      "step": 38030
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.069067115634806e-05,
+      "loss": 1.6937,
+      "step": 38040
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.068952322644829e-05,
+      "loss": 1.6288,
+      "step": 38050
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.068837529654856e-05,
+      "loss": 1.48,
+      "step": 38060
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.06872273666488e-05,
+      "loss": 1.6339,
+      "step": 38070
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.068607943674906e-05,
+      "loss": 1.703,
+      "step": 38080
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.06849315068493e-05,
+      "loss": 1.7255,
+      "step": 38090
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.068378357694957e-05,
+      "loss": 1.5939,
+      "step": 38100
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.06826356470498e-05,
+      "loss": 1.5295,
+      "step": 38110
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.068148771715007e-05,
+      "loss": 1.5753,
+      "step": 38120
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.068033978725032e-05,
+      "loss": 1.5824,
+      "step": 38130
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.067919185735057e-05,
+      "loss": 1.6192,
+      "step": 38140
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.067804392745082e-05,
+      "loss": 1.3596,
+      "step": 38150
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.067689599755108e-05,
+      "loss": 1.478,
+      "step": 38160
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.067574806765133e-05,
+      "loss": 1.5832,
+      "step": 38170
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.067460013775158e-05,
+      "loss": 1.6099,
+      "step": 38180
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.067345220785185e-05,
+      "loss": 1.7055,
+      "step": 38190
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.067230427795208e-05,
+      "loss": 1.5496,
+      "step": 38200
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.067115634805234e-05,
+      "loss": 1.509,
+      "step": 38210
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.06700084181526e-05,
+      "loss": 1.6018,
+      "step": 38220
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.066886048825284e-05,
+      "loss": 1.5255,
+      "step": 38230
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.06677125583531e-05,
+      "loss": 1.6684,
+      "step": 38240
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.066656462845334e-05,
+      "loss": 1.6178,
+      "step": 38250
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.06654166985536e-05,
+      "loss": 1.6389,
+      "step": 38260
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.066426876865386e-05,
+      "loss": 1.6114,
+      "step": 38270
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.066312083875411e-05,
+      "loss": 1.6053,
+      "step": 38280
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.066197290885436e-05,
+      "loss": 1.4963,
+      "step": 38290
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.066082497895461e-05,
+      "loss": 1.4325,
+      "step": 38300
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.065967704905486e-05,
+      "loss": 1.604,
+      "step": 38310
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.065852911915512e-05,
+      "loss": 1.6402,
+      "step": 38320
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.065738118925537e-05,
+      "loss": 1.5513,
+      "step": 38330
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.065623325935562e-05,
+      "loss": 1.5583,
+      "step": 38340
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.065508532945587e-05,
+      "loss": 1.5339,
+      "step": 38350
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.065393739955613e-05,
+      "loss": 1.6657,
+      "step": 38360
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.065278946965637e-05,
+      "loss": 1.5059,
+      "step": 38370
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.065164153975663e-05,
+      "loss": 1.5231,
+      "step": 38380
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.065049360985688e-05,
+      "loss": 1.5954,
+      "step": 38390
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.064934567995713e-05,
+      "loss": 1.6153,
+      "step": 38400
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.064819775005738e-05,
+      "loss": 1.6157,
+      "step": 38410
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.064704982015765e-05,
+      "loss": 1.6106,
+      "step": 38420
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.06459018902579e-05,
+      "loss": 1.4553,
+      "step": 38430
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.064475396035815e-05,
+      "loss": 1.6571,
+      "step": 38440
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.06436060304584e-05,
+      "loss": 1.5677,
+      "step": 38450
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.064245810055865e-05,
+      "loss": 1.6707,
+      "step": 38460
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.064131017065891e-05,
+      "loss": 1.6078,
+      "step": 38470
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.064016224075916e-05,
+      "loss": 1.5635,
+      "step": 38480
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.063901431085941e-05,
+      "loss": 1.8004,
+      "step": 38490
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.063786638095966e-05,
+      "loss": 1.587,
+      "step": 38500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.063671845105992e-05,
+      "loss": 1.671,
+      "step": 38510
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.063557052116016e-05,
+      "loss": 1.6067,
+      "step": 38520
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.063442259126042e-05,
+      "loss": 1.6834,
+      "step": 38530
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.063327466136067e-05,
+      "loss": 1.4986,
+      "step": 38540
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.063212673146092e-05,
+      "loss": 1.4954,
+      "step": 38550
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.063097880156117e-05,
+      "loss": 1.5453,
+      "step": 38560
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.062983087166143e-05,
+      "loss": 1.5208,
+      "step": 38570
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.062868294176168e-05,
+      "loss": 1.6643,
+      "step": 38580
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.062753501186193e-05,
+      "loss": 1.586,
+      "step": 38590
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.062638708196218e-05,
+      "loss": 1.5939,
+      "step": 38600
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.062523915206243e-05,
+      "loss": 1.5623,
+      "step": 38610
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.06240912221627e-05,
+      "loss": 1.5409,
+      "step": 38620
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.062294329226295e-05,
+      "loss": 1.469,
+      "step": 38630
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.06217953623632e-05,
+      "loss": 1.6128,
+      "step": 38640
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.062064743246345e-05,
+      "loss": 1.4778,
+      "step": 38650
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.061949950256371e-05,
+      "loss": 1.6577,
+      "step": 38660
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.061835157266395e-05,
+      "loss": 1.4531,
+      "step": 38670
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.061720364276421e-05,
+      "loss": 1.6218,
+      "step": 38680
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.061605571286446e-05,
+      "loss": 1.5084,
+      "step": 38690
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.061490778296471e-05,
+      "loss": 1.5663,
+      "step": 38700
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.061375985306496e-05,
+      "loss": 1.6835,
+      "step": 38710
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.061261192316522e-05,
+      "loss": 1.5801,
+      "step": 38720
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.061146399326547e-05,
+      "loss": 1.4886,
+      "step": 38730
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.061031606336572e-05,
+      "loss": 1.6308,
+      "step": 38740
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.060916813346597e-05,
+      "loss": 1.7451,
+      "step": 38750
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.060802020356622e-05,
+      "loss": 1.5606,
+      "step": 38760
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.060687227366649e-05,
+      "loss": 1.6187,
+      "step": 38770
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.060572434376674e-05,
+      "loss": 1.4749,
+      "step": 38780
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.060457641386699e-05,
+      "loss": 1.5114,
+      "step": 38790
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.060342848396724e-05,
+      "loss": 1.6287,
+      "step": 38800
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.06022805540675e-05,
+      "loss": 1.4932,
+      "step": 38810
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.060113262416774e-05,
+      "loss": 1.6384,
+      "step": 38820
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.0599984694268e-05,
+      "loss": 1.5651,
+      "step": 38830
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.059883676436825e-05,
+      "loss": 1.5422,
+      "step": 38840
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.05976888344685e-05,
+      "loss": 1.5637,
+      "step": 38850
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.059654090456875e-05,
+      "loss": 1.4926,
+      "step": 38860
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.059539297466901e-05,
+      "loss": 1.7078,
+      "step": 38870
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.059424504476926e-05,
+      "loss": 1.5286,
+      "step": 38880
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.059309711486951e-05,
+      "loss": 1.57,
+      "step": 38890
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.059194918496976e-05,
+      "loss": 1.7029,
+      "step": 38900
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.059080125507001e-05,
+      "loss": 1.4943,
+      "step": 38910
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.058965332517027e-05,
+      "loss": 1.7416,
+      "step": 38920
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.058850539527052e-05,
+      "loss": 1.6461,
+      "step": 38930
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.058735746537077e-05,
+      "loss": 1.5446,
+      "step": 38940
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.058620953547102e-05,
+      "loss": 1.6173,
+      "step": 38950
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.058506160557129e-05,
+      "loss": 1.321,
+      "step": 38960
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.058391367567152e-05,
+      "loss": 1.6945,
+      "step": 38970
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.058276574577179e-05,
+      "loss": 1.3773,
+      "step": 38980
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.058161781587204e-05,
+      "loss": 1.6326,
+      "step": 38990
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.058046988597229e-05,
+      "loss": 1.6877,
+      "step": 39000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.057932195607254e-05,
+      "loss": 1.653,
+      "step": 39010
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.05781740261728e-05,
+      "loss": 1.6972,
+      "step": 39020
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.057702609627305e-05,
+      "loss": 1.4883,
+      "step": 39030
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.05758781663733e-05,
+      "loss": 1.6257,
+      "step": 39040
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.057473023647355e-05,
+      "loss": 1.6402,
+      "step": 39050
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.05735823065738e-05,
+      "loss": 1.5882,
+      "step": 39060
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.057243437667406e-05,
+      "loss": 1.5626,
+      "step": 39070
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.057128644677431e-05,
+      "loss": 1.4625,
+      "step": 39080
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.057013851687456e-05,
+      "loss": 1.5663,
+      "step": 39090
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.056899058697481e-05,
+      "loss": 1.7463,
+      "step": 39100
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.056784265707508e-05,
+      "loss": 1.4363,
+      "step": 39110
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.056669472717531e-05,
+      "loss": 1.4616,
+      "step": 39120
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.056554679727558e-05,
+      "loss": 1.4737,
+      "step": 39130
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.056439886737583e-05,
+      "loss": 1.5952,
+      "step": 39140
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.056325093747608e-05,
+      "loss": 1.6149,
+      "step": 39150
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.056210300757632e-05,
+      "loss": 1.6059,
+      "step": 39160
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.056095507767659e-05,
+      "loss": 1.4339,
+      "step": 39170
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.055980714777684e-05,
+      "loss": 1.4034,
+      "step": 39180
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.055865921787709e-05,
+      "loss": 1.6184,
+      "step": 39190
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.055751128797734e-05,
+      "loss": 1.7124,
+      "step": 39200
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.055636335807759e-05,
+      "loss": 1.5876,
+      "step": 39210
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.055521542817785e-05,
+      "loss": 1.6532,
+      "step": 39220
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.05540674982781e-05,
+      "loss": 1.8254,
+      "step": 39230
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.055291956837835e-05,
+      "loss": 1.5845,
+      "step": 39240
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.05517716384786e-05,
+      "loss": 1.495,
+      "step": 39250
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.055062370857886e-05,
+      "loss": 1.6606,
+      "step": 39260
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.05494757786791e-05,
+      "loss": 1.5283,
+      "step": 39270
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.054832784877936e-05,
+      "loss": 1.7108,
+      "step": 39280
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.054717991887961e-05,
+      "loss": 1.5241,
+      "step": 39290
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.054603198897986e-05,
+      "loss": 1.6415,
+      "step": 39300
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.054488405908011e-05,
+      "loss": 1.6799,
+      "step": 39310
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.054373612918038e-05,
+      "loss": 1.54,
+      "step": 39320
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.054258819928063e-05,
+      "loss": 1.7394,
+      "step": 39330
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.054144026938088e-05,
+      "loss": 1.4706,
+      "step": 39340
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.054029233948113e-05,
+      "loss": 1.5258,
+      "step": 39350
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.053914440958138e-05,
+      "loss": 1.635,
+      "step": 39360
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.053799647968164e-05,
+      "loss": 1.4521,
+      "step": 39370
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.053684854978189e-05,
+      "loss": 1.5689,
+      "step": 39380
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.053570061988214e-05,
+      "loss": 1.4613,
+      "step": 39390
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.053455268998239e-05,
+      "loss": 1.5025,
+      "step": 39400
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.053340476008265e-05,
+      "loss": 1.4633,
+      "step": 39410
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.053225683018289e-05,
+      "loss": 1.4847,
+      "step": 39420
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.053110890028315e-05,
+      "loss": 1.6089,
+      "step": 39430
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.05299609703834e-05,
+      "loss": 1.4846,
+      "step": 39440
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.052881304048365e-05,
+      "loss": 1.5796,
+      "step": 39450
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.05276651105839e-05,
+      "loss": 1.5246,
+      "step": 39460
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.052651718068417e-05,
+      "loss": 1.6767,
+      "step": 39470
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.052536925078442e-05,
+      "loss": 1.3117,
+      "step": 39480
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.052422132088466e-05,
+      "loss": 1.5254,
+      "step": 39490
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.052307339098491e-05,
+      "loss": 1.5444,
+      "step": 39500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.052192546108516e-05,
+      "loss": 1.6069,
+      "step": 39510
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.052077753118543e-05,
+      "loss": 1.5698,
+      "step": 39520
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.051962960128568e-05,
+      "loss": 1.4596,
+      "step": 39530
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.051848167138593e-05,
+      "loss": 1.6767,
+      "step": 39540
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.051733374148618e-05,
+      "loss": 1.6839,
+      "step": 39550
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.051618581158644e-05,
+      "loss": 1.6813,
+      "step": 39560
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.051503788168668e-05,
+      "loss": 1.7714,
+      "step": 39570
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.051388995178694e-05,
+      "loss": 1.4573,
+      "step": 39580
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.051274202188719e-05,
+      "loss": 1.7371,
+      "step": 39590
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.051159409198744e-05,
+      "loss": 1.6091,
+      "step": 39600
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.051044616208769e-05,
+      "loss": 1.6104,
+      "step": 39610
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.050929823218794e-05,
+      "loss": 1.5467,
+      "step": 39620
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.05081503022882e-05,
+      "loss": 1.6148,
+      "step": 39630
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.050700237238845e-05,
+      "loss": 1.5608,
+      "step": 39640
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.05058544424887e-05,
+      "loss": 1.5565,
+      "step": 39650
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.050470651258895e-05,
+      "loss": 1.4216,
+      "step": 39660
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.050355858268922e-05,
+      "loss": 1.7272,
+      "step": 39670
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.050241065278945e-05,
+      "loss": 1.7954,
+      "step": 39680
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.050126272288972e-05,
+      "loss": 1.3681,
+      "step": 39690
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.050011479298997e-05,
+      "loss": 1.6539,
+      "step": 39700
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.049896686309022e-05,
+      "loss": 1.4828,
+      "step": 39710
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.049781893319047e-05,
+      "loss": 1.529,
+      "step": 39720
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.049667100329073e-05,
+      "loss": 1.4399,
+      "step": 39730
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.049552307339098e-05,
+      "loss": 1.5846,
+      "step": 39740
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.049437514349123e-05,
+      "loss": 1.5582,
+      "step": 39750
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.049322721359148e-05,
+      "loss": 1.6566,
+      "step": 39760
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.049207928369173e-05,
+      "loss": 1.6007,
+      "step": 39770
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.049093135379199e-05,
+      "loss": 1.5492,
+      "step": 39780
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.048978342389224e-05,
+      "loss": 1.5271,
+      "step": 39790
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.048863549399249e-05,
+      "loss": 1.6956,
+      "step": 39800
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.048748756409274e-05,
+      "loss": 1.7404,
+      "step": 39810
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0486339634193e-05,
+      "loss": 1.5111,
+      "step": 39820
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.048519170429324e-05,
+      "loss": 1.6342,
+      "step": 39830
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.04840437743935e-05,
+      "loss": 1.5113,
+      "step": 39840
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.048289584449375e-05,
+      "loss": 1.5133,
+      "step": 39850
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0481747914594e-05,
+      "loss": 1.7427,
+      "step": 39860
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.048059998469425e-05,
+      "loss": 1.5194,
+      "step": 39870
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.047945205479452e-05,
+      "loss": 1.5085,
+      "step": 39880
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.047830412489477e-05,
+      "loss": 1.471,
+      "step": 39890
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.047715619499502e-05,
+      "loss": 1.6189,
+      "step": 39900
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.047600826509527e-05,
+      "loss": 1.4899,
+      "step": 39910
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.047486033519552e-05,
+      "loss": 1.5985,
+      "step": 39920
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.047371240529578e-05,
+      "loss": 1.5944,
+      "step": 39930
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.047256447539603e-05,
+      "loss": 1.7271,
+      "step": 39940
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.047141654549628e-05,
+      "loss": 1.5647,
+      "step": 39950
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.047026861559653e-05,
+      "loss": 1.6064,
+      "step": 39960
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.04691206856968e-05,
+      "loss": 1.4357,
+      "step": 39970
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.046797275579703e-05,
+      "loss": 1.5836,
+      "step": 39980
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.046682482589729e-05,
+      "loss": 1.4041,
+      "step": 39990
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.046567689599754e-05,
+      "loss": 1.6322,
+      "step": 40000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.046452896609779e-05,
+      "loss": 1.7753,
+      "step": 40010
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.046338103619804e-05,
+      "loss": 1.414,
+      "step": 40020
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.04622331062983e-05,
+      "loss": 1.6634,
+      "step": 40030
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.046108517639856e-05,
+      "loss": 1.5827,
+      "step": 40040
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.04599372464988e-05,
+      "loss": 1.4887,
+      "step": 40050
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.045878931659906e-05,
+      "loss": 1.6141,
+      "step": 40060
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.04576413866993e-05,
+      "loss": 1.6178,
+      "step": 40070
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.045649345679957e-05,
+      "loss": 1.3214,
+      "step": 40080
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.045534552689982e-05,
+      "loss": 1.7103,
+      "step": 40090
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.045419759700007e-05,
+      "loss": 1.4902,
+      "step": 40100
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.045304966710032e-05,
+      "loss": 1.6224,
+      "step": 40110
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.045190173720058e-05,
+      "loss": 1.6967,
+      "step": 40120
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.045075380730082e-05,
+      "loss": 1.5034,
+      "step": 40130
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.044960587740108e-05,
+      "loss": 1.7206,
+      "step": 40140
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.044845794750133e-05,
+      "loss": 1.6236,
+      "step": 40150
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.044731001760158e-05,
+      "loss": 1.5349,
+      "step": 40160
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.044616208770183e-05,
+      "loss": 1.6271,
+      "step": 40170
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.04450141578021e-05,
+      "loss": 1.6461,
+      "step": 40180
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.044386622790234e-05,
+      "loss": 1.6717,
+      "step": 40190
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.04427182980026e-05,
+      "loss": 1.4559,
+      "step": 40200
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.044157036810284e-05,
+      "loss": 1.6203,
+      "step": 40210
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.04404224382031e-05,
+      "loss": 1.9062,
+      "step": 40220
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.043927450830336e-05,
+      "loss": 1.5361,
+      "step": 40230
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.043812657840361e-05,
+      "loss": 1.5963,
+      "step": 40240
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.043697864850386e-05,
+      "loss": 1.5358,
+      "step": 40250
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.043583071860411e-05,
+      "loss": 1.4401,
+      "step": 40260
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.043468278870437e-05,
+      "loss": 1.4401,
+      "step": 40270
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.04335348588046e-05,
+      "loss": 1.6798,
+      "step": 40280
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.043238692890487e-05,
+      "loss": 1.6521,
+      "step": 40290
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.043123899900512e-05,
+      "loss": 1.4689,
+      "step": 40300
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.043009106910537e-05,
+      "loss": 1.5772,
+      "step": 40310
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.042894313920562e-05,
+      "loss": 1.7164,
+      "step": 40320
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.042779520930588e-05,
+      "loss": 1.3034,
+      "step": 40330
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.042664727940613e-05,
+      "loss": 1.7327,
+      "step": 40340
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.042549934950638e-05,
+      "loss": 1.5747,
+      "step": 40350
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.042435141960663e-05,
+      "loss": 1.7173,
+      "step": 40360
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.042320348970688e-05,
+      "loss": 1.4399,
+      "step": 40370
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.042205555980715e-05,
+      "loss": 1.5559,
+      "step": 40380
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.04209076299074e-05,
+      "loss": 1.7947,
+      "step": 40390
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.041975970000765e-05,
+      "loss": 1.5933,
+      "step": 40400
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.04186117701079e-05,
+      "loss": 1.3721,
+      "step": 40410
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.041746384020816e-05,
+      "loss": 1.4561,
+      "step": 40420
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.04163159103084e-05,
+      "loss": 1.4794,
+      "step": 40430
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.041516798040866e-05,
+      "loss": 1.6379,
+      "step": 40440
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.041402005050891e-05,
+      "loss": 1.5691,
+      "step": 40450
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.041287212060916e-05,
+      "loss": 1.5197,
+      "step": 40460
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.041172419070941e-05,
+      "loss": 1.5345,
+      "step": 40470
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.041057626080967e-05,
+      "loss": 1.4875,
+      "step": 40480
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.040942833090992e-05,
+      "loss": 1.6914,
+      "step": 40490
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.040828040101017e-05,
+      "loss": 1.5756,
+      "step": 40500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.040713247111042e-05,
+      "loss": 1.653,
+      "step": 40510
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.040598454121067e-05,
+      "loss": 1.6611,
+      "step": 40520
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.040483661131093e-05,
+      "loss": 1.6838,
+      "step": 40530
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.040368868141118e-05,
+      "loss": 1.426,
+      "step": 40540
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.040254075151143e-05,
+      "loss": 1.7055,
+      "step": 40550
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.040139282161168e-05,
+      "loss": 1.6641,
+      "step": 40560
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.040024489171195e-05,
+      "loss": 1.5474,
+      "step": 40570
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.039909696181218e-05,
+      "loss": 1.4432,
+      "step": 40580
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.039794903191245e-05,
+      "loss": 1.5333,
+      "step": 40590
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.03968011020127e-05,
+      "loss": 1.4227,
+      "step": 40600
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.039565317211295e-05,
+      "loss": 1.5236,
+      "step": 40610
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.03945052422132e-05,
+      "loss": 1.4844,
+      "step": 40620
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.039335731231346e-05,
+      "loss": 1.5123,
+      "step": 40630
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.039220938241371e-05,
+      "loss": 1.5009,
+      "step": 40640
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.039106145251396e-05,
+      "loss": 1.4723,
+      "step": 40650
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.038991352261421e-05,
+      "loss": 1.6133,
+      "step": 40660
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.038876559271446e-05,
+      "loss": 1.5679,
+      "step": 40670
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.038761766281472e-05,
+      "loss": 1.3871,
+      "step": 40680
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.038646973291497e-05,
+      "loss": 1.5083,
+      "step": 40690
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.038532180301522e-05,
+      "loss": 1.6343,
+      "step": 40700
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.038417387311547e-05,
+      "loss": 1.4955,
+      "step": 40710
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.038302594321574e-05,
+      "loss": 1.5852,
+      "step": 40720
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.038187801331597e-05,
+      "loss": 1.7248,
+      "step": 40730
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.038073008341624e-05,
+      "loss": 1.6212,
+      "step": 40740
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.037958215351649e-05,
+      "loss": 1.5452,
+      "step": 40750
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.037843422361674e-05,
+      "loss": 1.6316,
+      "step": 40760
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.037728629371699e-05,
+      "loss": 1.6936,
+      "step": 40770
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.037613836381725e-05,
+      "loss": 1.5253,
+      "step": 40780
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.03749904339175e-05,
+      "loss": 1.5537,
+      "step": 40790
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.037384250401775e-05,
+      "loss": 1.4979,
+      "step": 40800
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.0372694574118e-05,
+      "loss": 1.5755,
+      "step": 40810
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.037154664421825e-05,
+      "loss": 1.5179,
+      "step": 40820
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.037039871431851e-05,
+      "loss": 1.5716,
+      "step": 40830
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.036925078441876e-05,
+      "loss": 1.6279,
+      "step": 40840
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.036810285451901e-05,
+      "loss": 1.4804,
+      "step": 40850
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.036695492461926e-05,
+      "loss": 1.5021,
+      "step": 40860
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.036580699471952e-05,
+      "loss": 1.5003,
+      "step": 40870
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.036465906481976e-05,
+      "loss": 1.752,
+      "step": 40880
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.036351113492002e-05,
+      "loss": 1.2907,
+      "step": 40890
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.036236320502027e-05,
+      "loss": 1.6634,
+      "step": 40900
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.036121527512052e-05,
+      "loss": 1.4779,
+      "step": 40910
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.036006734522077e-05,
+      "loss": 1.6203,
+      "step": 40920
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.035891941532104e-05,
+      "loss": 1.5552,
+      "step": 40930
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.035777148542129e-05,
+      "loss": 1.5414,
+      "step": 40940
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.035662355552154e-05,
+      "loss": 1.4233,
+      "step": 40950
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.035547562562179e-05,
+      "loss": 1.4919,
+      "step": 40960
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.035432769572204e-05,
+      "loss": 1.4187,
+      "step": 40970
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.03531797658223e-05,
+      "loss": 1.5929,
+      "step": 40980
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.035203183592255e-05,
+      "loss": 1.6222,
+      "step": 40990
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.03508839060228e-05,
+      "loss": 1.4136,
+      "step": 41000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.034973597612305e-05,
+      "loss": 1.488,
+      "step": 41010
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.03485880462233e-05,
+      "loss": 1.5111,
+      "step": 41020
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.034744011632355e-05,
+      "loss": 1.6413,
+      "step": 41030
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.034629218642381e-05,
+      "loss": 1.3856,
+      "step": 41040
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.034514425652406e-05,
+      "loss": 1.5713,
+      "step": 41050
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.034399632662431e-05,
+      "loss": 1.5127,
+      "step": 41060
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.034284839672456e-05,
+      "loss": 1.549,
+      "step": 41070
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.034170046682481e-05,
+      "loss": 1.6843,
+      "step": 41080
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.034055253692508e-05,
+      "loss": 1.6293,
+      "step": 41090
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.033940460702533e-05,
+      "loss": 1.3912,
+      "step": 41100
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.033825667712558e-05,
+      "loss": 1.5906,
+      "step": 41110
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.033710874722582e-05,
+      "loss": 1.3895,
+      "step": 41120
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.033596081732609e-05,
+      "loss": 1.5649,
+      "step": 41130
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.033481288742632e-05,
+      "loss": 1.457,
+      "step": 41140
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.033366495752659e-05,
+      "loss": 1.6461,
+      "step": 41150
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.033251702762684e-05,
+      "loss": 1.5947,
+      "step": 41160
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.033136909772709e-05,
+      "loss": 1.5757,
+      "step": 41170
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.033022116782734e-05,
+      "loss": 1.5144,
+      "step": 41180
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.03290732379276e-05,
+      "loss": 1.6217,
+      "step": 41190
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.032792530802785e-05,
+      "loss": 1.5335,
+      "step": 41200
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.03267773781281e-05,
+      "loss": 1.6179,
+      "step": 41210
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.032562944822835e-05,
+      "loss": 1.6507,
+      "step": 41220
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.03244815183286e-05,
+      "loss": 1.894,
+      "step": 41230
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.032333358842886e-05,
+      "loss": 1.6332,
+      "step": 41240
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.032218565852911e-05,
+      "loss": 1.5262,
+      "step": 41250
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.032103772862936e-05,
+      "loss": 1.3866,
+      "step": 41260
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.031988979872961e-05,
+      "loss": 1.4093,
+      "step": 41270
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.031874186882988e-05,
+      "loss": 1.4937,
+      "step": 41280
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.031759393893011e-05,
+      "loss": 1.5707,
+      "step": 41290
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.031644600903038e-05,
+      "loss": 1.718,
+      "step": 41300
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.031529807913063e-05,
+      "loss": 1.5834,
+      "step": 41310
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.031415014923088e-05,
+      "loss": 1.4034,
+      "step": 41320
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.031300221933113e-05,
+      "loss": 1.422,
+      "step": 41330
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.031185428943139e-05,
+      "loss": 1.6077,
+      "step": 41340
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.031070635953164e-05,
+      "loss": 1.566,
+      "step": 41350
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.030955842963189e-05,
+      "loss": 1.6618,
+      "step": 41360
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.030841049973214e-05,
+      "loss": 1.6113,
+      "step": 41370
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.030726256983239e-05,
+      "loss": 1.5389,
+      "step": 41380
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.030611463993265e-05,
+      "loss": 1.7131,
+      "step": 41390
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.03049667100329e-05,
+      "loss": 1.4968,
+      "step": 41400
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.030381878013315e-05,
+      "loss": 1.5161,
+      "step": 41410
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.03026708502334e-05,
+      "loss": 1.5803,
+      "step": 41420
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.030152292033367e-05,
+      "loss": 1.5935,
+      "step": 41430
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.03003749904339e-05,
+      "loss": 1.7717,
+      "step": 41440
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.029922706053416e-05,
+      "loss": 1.5566,
+      "step": 41450
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.029807913063441e-05,
+      "loss": 1.5456,
+      "step": 41460
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.029693120073466e-05,
+      "loss": 1.5759,
+      "step": 41470
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.029578327083491e-05,
+      "loss": 1.4467,
+      "step": 41480
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.029463534093518e-05,
+      "loss": 1.5004,
+      "step": 41490
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.029348741103543e-05,
+      "loss": 1.605,
+      "step": 41500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.029233948113568e-05,
+      "loss": 1.6168,
+      "step": 41510
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.029119155123593e-05,
+      "loss": 1.5219,
+      "step": 41520
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.029004362133618e-05,
+      "loss": 1.5787,
+      "step": 41530
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.028889569143644e-05,
+      "loss": 1.5274,
+      "step": 41540
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.028774776153669e-05,
+      "loss": 1.4387,
+      "step": 41550
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.028659983163694e-05,
+      "loss": 1.5701,
+      "step": 41560
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.028545190173719e-05,
+      "loss": 1.5206,
+      "step": 41570
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.028430397183745e-05,
+      "loss": 1.5685,
+      "step": 41580
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.028315604193769e-05,
+      "loss": 1.503,
+      "step": 41590
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.028200811203795e-05,
+      "loss": 1.601,
+      "step": 41600
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.02808601821382e-05,
+      "loss": 1.5842,
+      "step": 41610
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.027971225223845e-05,
+      "loss": 1.5532,
+      "step": 41620
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.02785643223387e-05,
+      "loss": 1.4462,
+      "step": 41630
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.027741639243897e-05,
+      "loss": 1.5567,
+      "step": 41640
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.027626846253922e-05,
+      "loss": 1.6252,
+      "step": 41650
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.027512053263947e-05,
+      "loss": 1.5163,
+      "step": 41660
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.027397260273972e-05,
+      "loss": 1.6418,
+      "step": 41670
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.027282467283997e-05,
+      "loss": 1.4653,
+      "step": 41680
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.027167674294023e-05,
+      "loss": 1.5356,
+      "step": 41690
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.027052881304048e-05,
+      "loss": 1.5065,
+      "step": 41700
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.026938088314073e-05,
+      "loss": 1.703,
+      "step": 41710
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.026823295324098e-05,
+      "loss": 1.4489,
+      "step": 41720
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.026708502334124e-05,
+      "loss": 1.7937,
+      "step": 41730
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.026593709344148e-05,
+      "loss": 1.7051,
+      "step": 41740
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.026478916354174e-05,
+      "loss": 1.6192,
+      "step": 41750
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.026364123364199e-05,
+      "loss": 1.566,
+      "step": 41760
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.026249330374224e-05,
+      "loss": 1.5551,
+      "step": 41770
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.026134537384249e-05,
+      "loss": 1.6971,
+      "step": 41780
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.026019744394275e-05,
+      "loss": 1.6007,
+      "step": 41790
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.0259049514043e-05,
+      "loss": 1.7219,
+      "step": 41800
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.025790158414325e-05,
+      "loss": 1.47,
+      "step": 41810
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.02567536542435e-05,
+      "loss": 1.458,
+      "step": 41820
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.025560572434375e-05,
+      "loss": 1.6171,
+      "step": 41830
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.025445779444402e-05,
+      "loss": 1.5011,
+      "step": 41840
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.025330986454427e-05,
+      "loss": 1.5277,
+      "step": 41850
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.025216193464452e-05,
+      "loss": 1.6547,
+      "step": 41860
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.025101400474477e-05,
+      "loss": 1.7179,
+      "step": 41870
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.024986607484503e-05,
+      "loss": 1.4325,
+      "step": 41880
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.024871814494527e-05,
+      "loss": 1.4485,
+      "step": 41890
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.024757021504553e-05,
+      "loss": 1.3547,
+      "step": 41900
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.024642228514578e-05,
+      "loss": 1.4694,
+      "step": 41910
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.024527435524603e-05,
+      "loss": 1.4927,
+      "step": 41920
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.024412642534628e-05,
+      "loss": 1.5471,
+      "step": 41930
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.024297849544654e-05,
+      "loss": 1.4117,
+      "step": 41940
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.024183056554679e-05,
+      "loss": 1.6791,
+      "step": 41950
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.024068263564704e-05,
+      "loss": 1.4732,
+      "step": 41960
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.023953470574729e-05,
+      "loss": 1.6193,
+      "step": 41970
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.023838677584754e-05,
+      "loss": 1.4126,
+      "step": 41980
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.02372388459478e-05,
+      "loss": 1.4237,
+      "step": 41990
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.023609091604806e-05,
+      "loss": 1.7089,
+      "step": 42000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.02349429861483e-05,
+      "loss": 1.5217,
+      "step": 42010
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.023379505624856e-05,
+      "loss": 1.6546,
+      "step": 42020
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.023264712634882e-05,
+      "loss": 1.5437,
+      "step": 42030
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.023149919644906e-05,
+      "loss": 1.4076,
+      "step": 42040
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.023035126654932e-05,
+      "loss": 1.6435,
+      "step": 42050
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.022920333664957e-05,
+      "loss": 1.5135,
+      "step": 42060
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.022805540674982e-05,
+      "loss": 1.6548,
+      "step": 42070
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.022690747685007e-05,
+      "loss": 1.6144,
+      "step": 42080
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.022575954695033e-05,
+      "loss": 1.3807,
+      "step": 42090
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.022461161705058e-05,
+      "loss": 1.6023,
+      "step": 42100
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.022346368715083e-05,
+      "loss": 1.5629,
+      "step": 42110
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.022231575725108e-05,
+      "loss": 1.7745,
+      "step": 42120
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.022116782735133e-05,
+      "loss": 1.58,
+      "step": 42130
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.02200198974516e-05,
+      "loss": 1.5376,
+      "step": 42140
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.021887196755184e-05,
+      "loss": 1.5841,
+      "step": 42150
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.02177240376521e-05,
+      "loss": 1.5413,
+      "step": 42160
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.021657610775234e-05,
+      "loss": 1.5908,
+      "step": 42170
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.021542817785261e-05,
+      "loss": 1.5787,
+      "step": 42180
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.021428024795284e-05,
+      "loss": 1.7164,
+      "step": 42190
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.021313231805311e-05,
+      "loss": 1.5544,
+      "step": 42200
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.021198438815336e-05,
+      "loss": 1.5308,
+      "step": 42210
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.021083645825361e-05,
+      "loss": 1.5074,
+      "step": 42220
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.020968852835386e-05,
+      "loss": 1.7115,
+      "step": 42230
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.020854059845412e-05,
+      "loss": 1.8193,
+      "step": 42240
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.020739266855437e-05,
+      "loss": 1.4625,
+      "step": 42250
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.020624473865462e-05,
+      "loss": 1.6494,
+      "step": 42260
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.020509680875487e-05,
+      "loss": 1.5116,
+      "step": 42270
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.020394887885512e-05,
+      "loss": 1.4575,
+      "step": 42280
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.020280094895538e-05,
+      "loss": 1.7299,
+      "step": 42290
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.020165301905563e-05,
+      "loss": 1.6599,
+      "step": 42300
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.020050508915588e-05,
+      "loss": 1.682,
+      "step": 42310
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.019935715925613e-05,
+      "loss": 1.7634,
+      "step": 42320
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.01982092293564e-05,
+      "loss": 1.7692,
+      "step": 42330
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.019706129945663e-05,
+      "loss": 1.6526,
+      "step": 42340
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.01959133695569e-05,
+      "loss": 1.5733,
+      "step": 42350
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.019476543965715e-05,
+      "loss": 1.5698,
+      "step": 42360
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.01936175097574e-05,
+      "loss": 1.4982,
+      "step": 42370
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.019246957985765e-05,
+      "loss": 1.5975,
+      "step": 42380
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.01913216499579e-05,
+      "loss": 1.4073,
+      "step": 42390
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.019017372005816e-05,
+      "loss": 1.4039,
+      "step": 42400
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.018902579015841e-05,
+      "loss": 1.6246,
+      "step": 42410
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.018787786025866e-05,
+      "loss": 1.595,
+      "step": 42420
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.018672993035891e-05,
+      "loss": 1.4715,
+      "step": 42430
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.018558200045917e-05,
+      "loss": 1.5032,
+      "step": 42440
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.018443407055941e-05,
+      "loss": 1.5543,
+      "step": 42450
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.018328614065967e-05,
+      "loss": 1.4719,
+      "step": 42460
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.018213821075992e-05,
+      "loss": 1.5585,
+      "step": 42470
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.018099028086017e-05,
+      "loss": 1.7135,
+      "step": 42480
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.017984235096042e-05,
+      "loss": 1.6335,
+      "step": 42490
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.017869442106068e-05,
+      "loss": 1.4629,
+      "step": 42500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.017754649116093e-05,
+      "loss": 1.6395,
+      "step": 42510
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.017639856126118e-05,
+      "loss": 1.6042,
+      "step": 42520
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.017525063136143e-05,
+      "loss": 1.7543,
+      "step": 42530
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.017410270146168e-05,
+      "loss": 1.4062,
+      "step": 42540
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.017295477156195e-05,
+      "loss": 1.4358,
+      "step": 42550
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.01718068416622e-05,
+      "loss": 1.4728,
+      "step": 42560
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.017065891176245e-05,
+      "loss": 1.6522,
+      "step": 42570
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.01695109818627e-05,
+      "loss": 1.6007,
+      "step": 42580
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.016836305196296e-05,
+      "loss": 1.5082,
+      "step": 42590
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.01672151220632e-05,
+      "loss": 1.6672,
+      "step": 42600
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.016606719216346e-05,
+      "loss": 1.3461,
+      "step": 42610
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.016491926226371e-05,
+      "loss": 1.4132,
+      "step": 42620
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.016377133236396e-05,
+      "loss": 1.4169,
+      "step": 42630
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.016262340246421e-05,
+      "loss": 1.4898,
+      "step": 42640
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.016147547256447e-05,
+      "loss": 1.5527,
+      "step": 42650
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.016032754266472e-05,
+      "loss": 1.6073,
+      "step": 42660
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.015917961276497e-05,
+      "loss": 1.6943,
+      "step": 42670
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.015803168286522e-05,
+      "loss": 1.5766,
+      "step": 42680
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.015688375296547e-05,
+      "loss": 1.4892,
+      "step": 42690
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.015573582306574e-05,
+      "loss": 1.6064,
+      "step": 42700
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.015458789316599e-05,
+      "loss": 1.7606,
+      "step": 42710
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.015343996326624e-05,
+      "loss": 1.5724,
+      "step": 42720
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.015229203336649e-05,
+      "loss": 1.4919,
+      "step": 42730
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.015114410346675e-05,
+      "loss": 1.5459,
+      "step": 42740
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.014999617356698e-05,
+      "loss": 1.7671,
+      "step": 42750
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.014884824366725e-05,
+      "loss": 1.5529,
+      "step": 42760
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.01477003137675e-05,
+      "loss": 1.5452,
+      "step": 42770
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.014655238386775e-05,
+      "loss": 1.4181,
+      "step": 42780
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.0145404453968e-05,
+      "loss": 1.4857,
+      "step": 42790
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.014425652406826e-05,
+      "loss": 1.8625,
+      "step": 42800
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.014310859416851e-05,
+      "loss": 1.4969,
+      "step": 42810
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.014196066426876e-05,
+      "loss": 1.6837,
+      "step": 42820
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.014081273436901e-05,
+      "loss": 1.5138,
+      "step": 42830
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.013966480446926e-05,
+      "loss": 1.5844,
+      "step": 42840
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.013851687456952e-05,
+      "loss": 1.5688,
+      "step": 42850
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.013736894466977e-05,
+      "loss": 1.4189,
+      "step": 42860
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.013622101477002e-05,
+      "loss": 1.5461,
+      "step": 42870
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.013507308487027e-05,
+      "loss": 1.5785,
+      "step": 42880
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.013392515497054e-05,
+      "loss": 1.5022,
+      "step": 42890
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.013277722507077e-05,
+      "loss": 1.7825,
+      "step": 42900
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.013162929517104e-05,
+      "loss": 1.4555,
+      "step": 42910
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.013048136527129e-05,
+      "loss": 1.6024,
+      "step": 42920
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.012933343537154e-05,
+      "loss": 1.6613,
+      "step": 42930
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.012818550547179e-05,
+      "loss": 1.573,
+      "step": 42940
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.012703757557205e-05,
+      "loss": 1.5831,
+      "step": 42950
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.01258896456723e-05,
+      "loss": 1.5283,
+      "step": 42960
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.012474171577255e-05,
+      "loss": 1.5686,
+      "step": 42970
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.01235937858728e-05,
+      "loss": 1.5192,
+      "step": 42980
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.012244585597305e-05,
+      "loss": 1.573,
+      "step": 42990
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.012129792607331e-05,
+      "loss": 1.6002,
+      "step": 43000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.012014999617356e-05,
+      "loss": 1.6692,
+      "step": 43010
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.011900206627381e-05,
+      "loss": 1.5971,
+      "step": 43020
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.011785413637406e-05,
+      "loss": 1.4196,
+      "step": 43030
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.011670620647433e-05,
+      "loss": 1.4805,
+      "step": 43040
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.011555827657456e-05,
+      "loss": 1.4031,
+      "step": 43050
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.011441034667483e-05,
+      "loss": 1.3888,
+      "step": 43060
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.011326241677508e-05,
+      "loss": 1.5795,
+      "step": 43070
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.011211448687532e-05,
+      "loss": 1.5971,
+      "step": 43080
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.011096655697557e-05,
+      "loss": 1.7061,
+      "step": 43090
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.010981862707584e-05,
+      "loss": 1.4697,
+      "step": 43100
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.010867069717609e-05,
+      "loss": 1.6092,
+      "step": 43110
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.010752276727634e-05,
+      "loss": 1.3899,
+      "step": 43120
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.010637483737659e-05,
+      "loss": 1.6047,
+      "step": 43130
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.010522690747684e-05,
+      "loss": 1.5155,
+      "step": 43140
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.01040789775771e-05,
+      "loss": 1.6893,
+      "step": 43150
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.010293104767735e-05,
+      "loss": 1.425,
+      "step": 43160
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.01017831177776e-05,
+      "loss": 1.6787,
+      "step": 43170
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.010063518787785e-05,
+      "loss": 1.5941,
+      "step": 43180
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.009948725797811e-05,
+      "loss": 1.7944,
+      "step": 43190
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.009833932807835e-05,
+      "loss": 1.4682,
+      "step": 43200
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.009719139817861e-05,
+      "loss": 1.5883,
+      "step": 43210
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.009604346827886e-05,
+      "loss": 1.44,
+      "step": 43220
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.009489553837911e-05,
+      "loss": 1.6651,
+      "step": 43230
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.009374760847936e-05,
+      "loss": 1.6738,
+      "step": 43240
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.009259967857963e-05,
+      "loss": 1.4552,
+      "step": 43250
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.009145174867988e-05,
+      "loss": 1.5894,
+      "step": 43260
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.009030381878013e-05,
+      "loss": 1.5399,
+      "step": 43270
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.008915588888038e-05,
+      "loss": 1.5493,
+      "step": 43280
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.008800795898063e-05,
+      "loss": 1.3251,
+      "step": 43290
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.008686002908089e-05,
+      "loss": 1.4838,
+      "step": 43300
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.008571209918114e-05,
+      "loss": 1.5887,
+      "step": 43310
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.008456416928139e-05,
+      "loss": 1.5609,
+      "step": 43320
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.008341623938164e-05,
+      "loss": 1.6838,
+      "step": 43330
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.00822683094819e-05,
+      "loss": 1.4685,
+      "step": 43340
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.008112037958214e-05,
+      "loss": 1.3612,
+      "step": 43350
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.00799724496824e-05,
+      "loss": 1.5349,
+      "step": 43360
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.007882451978265e-05,
+      "loss": 1.5909,
+      "step": 43370
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.00776765898829e-05,
+      "loss": 1.5364,
+      "step": 43380
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.007652865998315e-05,
+      "loss": 1.4457,
+      "step": 43390
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.007538073008342e-05,
+      "loss": 1.7035,
+      "step": 43400
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.007423280018366e-05,
+      "loss": 1.5445,
+      "step": 43410
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.007308487028391e-05,
+      "loss": 1.3913,
+      "step": 43420
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.007193694038416e-05,
+      "loss": 1.7133,
+      "step": 43430
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.007078901048441e-05,
+      "loss": 1.5278,
+      "step": 43440
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.006964108058468e-05,
+      "loss": 1.5404,
+      "step": 43450
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.006849315068493e-05,
+      "loss": 1.6034,
+      "step": 43460
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.006734522078518e-05,
+      "loss": 1.5559,
+      "step": 43470
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.006619729088543e-05,
+      "loss": 1.5126,
+      "step": 43480
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.006504936098569e-05,
+      "loss": 1.5239,
+      "step": 43490
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.006390143108593e-05,
+      "loss": 1.6339,
+      "step": 43500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.006275350118619e-05,
+      "loss": 1.7427,
+      "step": 43510
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.006160557128644e-05,
+      "loss": 1.5754,
+      "step": 43520
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.006045764138669e-05,
+      "loss": 1.5496,
+      "step": 43530
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.005930971148694e-05,
+      "loss": 1.5608,
+      "step": 43540
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.00581617815872e-05,
+      "loss": 1.73,
+      "step": 43550
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.005701385168745e-05,
+      "loss": 1.4883,
+      "step": 43560
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.00558659217877e-05,
+      "loss": 1.505,
+      "step": 43570
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.005471799188795e-05,
+      "loss": 1.6848,
+      "step": 43580
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.00535700619882e-05,
+      "loss": 1.6,
+      "step": 43590
+    },
+    {
+      "epoch": 2.0,
+      "eval_cer": 0.6759632739175152,
+      "eval_em": 0.007357212003872217,
+      "eval_f1": 0.007357212003872217,
+      "eval_loss": 1.3662258386611938,
+      "eval_runtime": 1597.2728,
+      "eval_samples_per_second": 6.467,
+      "eval_steps_per_second": 0.809,
+      "eval_wer": 0.9926427879961278,
+      "step": 43591
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.005242213208847e-05,
+      "loss": 1.3539,
+      "step": 43600
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.005127420218872e-05,
+      "loss": 1.3967,
+      "step": 43610
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.005012627228897e-05,
+      "loss": 1.395,
+      "step": 43620
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.004897834238922e-05,
+      "loss": 1.3328,
+      "step": 43630
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.004783041248948e-05,
+      "loss": 1.6421,
+      "step": 43640
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.004668248258972e-05,
+      "loss": 1.4279,
+      "step": 43650
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.004553455268998e-05,
+      "loss": 1.593,
+      "step": 43660
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.004438662279023e-05,
+      "loss": 1.5539,
+      "step": 43670
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.004323869289048e-05,
+      "loss": 1.425,
+      "step": 43680
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.004209076299073e-05,
+      "loss": 1.716,
+      "step": 43690
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.004094283309099e-05,
+      "loss": 1.7074,
+      "step": 43700
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.003979490319124e-05,
+      "loss": 1.4454,
+      "step": 43710
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.003864697329149e-05,
+      "loss": 1.5878,
+      "step": 43720
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.003749904339174e-05,
+      "loss": 1.4326,
+      "step": 43730
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.003635111349199e-05,
+      "loss": 1.5875,
+      "step": 43740
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.003520318359225e-05,
+      "loss": 1.3423,
+      "step": 43750
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.003405525369249e-05,
+      "loss": 1.5028,
+      "step": 43760
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.003290732379275e-05,
+      "loss": 1.3354,
+      "step": 43770
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.0031759393893e-05,
+      "loss": 1.4614,
+      "step": 43780
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.003061146399325e-05,
+      "loss": 1.5344,
+      "step": 43790
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.00294635340935e-05,
+      "loss": 1.4577,
+      "step": 43800
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.002831560419377e-05,
+      "loss": 1.4971,
+      "step": 43810
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.002716767429402e-05,
+      "loss": 1.7659,
+      "step": 43820
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.002601974439427e-05,
+      "loss": 1.3368,
+      "step": 43830
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.002487181449452e-05,
+      "loss": 1.4385,
+      "step": 43840
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.002372388459477e-05,
+      "loss": 1.5726,
+      "step": 43850
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.002257595469503e-05,
+      "loss": 1.4449,
+      "step": 43860
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.002142802479528e-05,
+      "loss": 1.5708,
+      "step": 43870
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.002028009489553e-05,
+      "loss": 1.532,
+      "step": 43880
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.001913216499578e-05,
+      "loss": 1.3469,
+      "step": 43890
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.001798423509604e-05,
+      "loss": 1.4767,
+      "step": 43900
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.001683630519628e-05,
+      "loss": 1.6318,
+      "step": 43910
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.001568837529654e-05,
+      "loss": 1.6562,
+      "step": 43920
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.001454044539679e-05,
+      "loss": 1.6546,
+      "step": 43930
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.001339251549704e-05,
+      "loss": 1.3882,
+      "step": 43940
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.001224458559729e-05,
+      "loss": 1.6826,
+      "step": 43950
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.001109665569756e-05,
+      "loss": 1.4653,
+      "step": 43960
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.00099487257978e-05,
+      "loss": 1.3271,
+      "step": 43970
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.000880079589806e-05,
+      "loss": 1.6414,
+      "step": 43980
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.00076528659983e-05,
+      "loss": 1.5411,
+      "step": 43990
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.000650493609856e-05,
+      "loss": 1.6065,
+      "step": 44000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.000535700619882e-05,
+      "loss": 1.5951,
+      "step": 44010
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.000420907629907e-05,
+      "loss": 1.6011,
+      "step": 44020
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.000306114639932e-05,
+      "loss": 1.694,
+      "step": 44030
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.000191321649957e-05,
+      "loss": 1.5059,
+      "step": 44040
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.000076528659983e-05,
+      "loss": 1.2977,
+      "step": 44050
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.999961735670007e-05,
+      "loss": 1.519,
+      "step": 44060
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.999846942680033e-05,
+      "loss": 1.4868,
+      "step": 44070
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.999732149690058e-05,
+      "loss": 1.661,
+      "step": 44080
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.999617356700083e-05,
+      "loss": 1.3873,
+      "step": 44090
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.999502563710108e-05,
+      "loss": 1.4833,
+      "step": 44100
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.999387770720134e-05,
+      "loss": 1.6247,
+      "step": 44110
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.99927297773016e-05,
+      "loss": 1.3426,
+      "step": 44120
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.999158184740184e-05,
+      "loss": 1.5173,
+      "step": 44130
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.99904339175021e-05,
+      "loss": 1.4733,
+      "step": 44140
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.998928598760234e-05,
+      "loss": 1.5012,
+      "step": 44150
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.998813805770261e-05,
+      "loss": 1.5319,
+      "step": 44160
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.998699012780286e-05,
+      "loss": 1.5384,
+      "step": 44170
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.998584219790311e-05,
+      "loss": 1.4845,
+      "step": 44180
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.998469426800336e-05,
+      "loss": 1.5829,
+      "step": 44190
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.998354633810362e-05,
+      "loss": 1.4729,
+      "step": 44200
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.998239840820386e-05,
+      "loss": 1.6066,
+      "step": 44210
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.998125047830412e-05,
+      "loss": 1.3849,
+      "step": 44220
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.998010254840437e-05,
+      "loss": 1.4449,
+      "step": 44230
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.997895461850462e-05,
+      "loss": 1.6594,
+      "step": 44240
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.997780668860487e-05,
+      "loss": 1.5519,
+      "step": 44250
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.997665875870513e-05,
+      "loss": 1.4462,
+      "step": 44260
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.997551082880538e-05,
+      "loss": 1.4138,
+      "step": 44270
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.997436289890563e-05,
+      "loss": 1.404,
+      "step": 44280
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.997321496900588e-05,
+      "loss": 1.5597,
+      "step": 44290
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.997206703910613e-05,
+      "loss": 1.3481,
+      "step": 44300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.99709191092064e-05,
+      "loss": 1.4826,
+      "step": 44310
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.996977117930665e-05,
+      "loss": 1.5287,
+      "step": 44320
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.99686232494069e-05,
+      "loss": 1.4548,
+      "step": 44330
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.996747531950715e-05,
+      "loss": 1.5383,
+      "step": 44340
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.996632738960741e-05,
+      "loss": 1.4027,
+      "step": 44350
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.996517945970765e-05,
+      "loss": 1.3518,
+      "step": 44360
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.996403152980791e-05,
+      "loss": 1.5913,
+      "step": 44370
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.996288359990816e-05,
+      "loss": 1.5741,
+      "step": 44380
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.996173567000841e-05,
+      "loss": 1.5755,
+      "step": 44390
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.996058774010866e-05,
+      "loss": 1.5079,
+      "step": 44400
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.995943981020892e-05,
+      "loss": 1.4252,
+      "step": 44410
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.995829188030917e-05,
+      "loss": 1.4009,
+      "step": 44420
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.995714395040942e-05,
+      "loss": 1.4753,
+      "step": 44430
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.995599602050967e-05,
+      "loss": 1.5435,
+      "step": 44440
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.995484809060992e-05,
+      "loss": 1.5272,
+      "step": 44450
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.995370016071018e-05,
+      "loss": 1.406,
+      "step": 44460
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.995255223081043e-05,
+      "loss": 1.508,
+      "step": 44470
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.995140430091068e-05,
+      "loss": 1.6142,
+      "step": 44480
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.995025637101093e-05,
+      "loss": 1.5321,
+      "step": 44490
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.99491084411112e-05,
+      "loss": 1.6069,
+      "step": 44500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.994796051121143e-05,
+      "loss": 1.5803,
+      "step": 44510
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.99468125813117e-05,
+      "loss": 1.5209,
+      "step": 44520
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.994566465141195e-05,
+      "loss": 1.4025,
+      "step": 44530
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.99445167215122e-05,
+      "loss": 1.5605,
+      "step": 44540
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.994336879161245e-05,
+      "loss": 1.496,
+      "step": 44550
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.994222086171271e-05,
+      "loss": 1.5074,
+      "step": 44560
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.994107293181296e-05,
+      "loss": 1.5839,
+      "step": 44570
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.993992500191321e-05,
+      "loss": 1.4419,
+      "step": 44580
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.993877707201346e-05,
+      "loss": 1.5299,
+      "step": 44590
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.993762914211371e-05,
+      "loss": 1.6026,
+      "step": 44600
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.993648121221397e-05,
+      "loss": 1.4962,
+      "step": 44610
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.993533328231422e-05,
+      "loss": 1.6235,
+      "step": 44620
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.993418535241447e-05,
+      "loss": 1.4117,
+      "step": 44630
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.993303742251472e-05,
+      "loss": 1.5916,
+      "step": 44640
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.993188949261499e-05,
+      "loss": 1.5644,
+      "step": 44650
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.993074156271522e-05,
+      "loss": 1.5294,
+      "step": 44660
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.992959363281549e-05,
+      "loss": 1.6062,
+      "step": 44670
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.992844570291574e-05,
+      "loss": 1.4792,
+      "step": 44680
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.992729777301599e-05,
+      "loss": 1.4072,
+      "step": 44690
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.992614984311623e-05,
+      "loss": 1.4575,
+      "step": 44700
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.99250019132165e-05,
+      "loss": 1.5296,
+      "step": 44710
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.992385398331675e-05,
+      "loss": 1.5118,
+      "step": 44720
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.9922706053417e-05,
+      "loss": 1.4981,
+      "step": 44730
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.992155812351725e-05,
+      "loss": 1.6948,
+      "step": 44740
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.99204101936175e-05,
+      "loss": 1.451,
+      "step": 44750
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.991926226371776e-05,
+      "loss": 1.6101,
+      "step": 44760
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.991811433381801e-05,
+      "loss": 1.4189,
+      "step": 44770
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.991696640391826e-05,
+      "loss": 1.4257,
+      "step": 44780
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.991581847401851e-05,
+      "loss": 1.5364,
+      "step": 44790
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.991467054411877e-05,
+      "loss": 1.4319,
+      "step": 44800
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.991352261421901e-05,
+      "loss": 1.3113,
+      "step": 44810
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.991237468431927e-05,
+      "loss": 1.701,
+      "step": 44820
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.991122675441952e-05,
+      "loss": 1.505,
+      "step": 44830
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.991007882451977e-05,
+      "loss": 1.4464,
+      "step": 44840
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.990893089462002e-05,
+      "loss": 1.5923,
+      "step": 44850
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.990778296472029e-05,
+      "loss": 1.6808,
+      "step": 44860
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.990663503482054e-05,
+      "loss": 1.434,
+      "step": 44870
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.990548710492079e-05,
+      "loss": 1.6947,
+      "step": 44880
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.990433917502104e-05,
+      "loss": 1.5405,
+      "step": 44890
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.990319124512129e-05,
+      "loss": 1.6391,
+      "step": 44900
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.990204331522155e-05,
+      "loss": 1.4778,
+      "step": 44910
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.99008953853218e-05,
+      "loss": 1.459,
+      "step": 44920
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.989974745542205e-05,
+      "loss": 1.4667,
+      "step": 44930
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.98985995255223e-05,
+      "loss": 1.3658,
+      "step": 44940
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.989745159562256e-05,
+      "loss": 1.514,
+      "step": 44950
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.98963036657228e-05,
+      "loss": 1.4371,
+      "step": 44960
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.989515573582306e-05,
+      "loss": 1.5031,
+      "step": 44970
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.989400780592331e-05,
+      "loss": 1.4584,
+      "step": 44980
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.989285987602356e-05,
+      "loss": 1.4013,
+      "step": 44990
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.989171194612381e-05,
+      "loss": 1.4687,
+      "step": 45000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.989056401622408e-05,
+      "loss": 1.5379,
+      "step": 45010
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.988941608632433e-05,
+      "loss": 1.4839,
+      "step": 45020
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.988826815642458e-05,
+      "loss": 1.4666,
+      "step": 45030
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.988712022652482e-05,
+      "loss": 1.6228,
+      "step": 45040
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.988597229662507e-05,
+      "loss": 1.4648,
+      "step": 45050
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.988482436672534e-05,
+      "loss": 1.5539,
+      "step": 45060
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.988367643682559e-05,
+      "loss": 1.4561,
+      "step": 45070
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.988252850692584e-05,
+      "loss": 1.4059,
+      "step": 45080
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.988138057702609e-05,
+      "loss": 1.3893,
+      "step": 45090
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.988023264712635e-05,
+      "loss": 1.4296,
+      "step": 45100
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.987908471722659e-05,
+      "loss": 1.4863,
+      "step": 45110
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.987793678732685e-05,
+      "loss": 1.4095,
+      "step": 45120
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.98767888574271e-05,
+      "loss": 1.5083,
+      "step": 45130
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.987564092752735e-05,
+      "loss": 1.6294,
+      "step": 45140
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.98744929976276e-05,
+      "loss": 1.3713,
+      "step": 45150
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.987334506772785e-05,
+      "loss": 1.5026,
+      "step": 45160
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.987219713782811e-05,
+      "loss": 1.4658,
+      "step": 45170
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.987104920792836e-05,
+      "loss": 1.6561,
+      "step": 45180
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.986990127802861e-05,
+      "loss": 1.6627,
+      "step": 45190
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.986875334812886e-05,
+      "loss": 1.6141,
+      "step": 45200
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.986760541822913e-05,
+      "loss": 1.4118,
+      "step": 45210
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.986645748832936e-05,
+      "loss": 1.5076,
+      "step": 45220
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.986530955842963e-05,
+      "loss": 1.4761,
+      "step": 45230
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.986416162852988e-05,
+      "loss": 1.318,
+      "step": 45240
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.986301369863013e-05,
+      "loss": 1.3477,
+      "step": 45250
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.986186576873038e-05,
+      "loss": 1.4983,
+      "step": 45260
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.986071783883064e-05,
+      "loss": 1.4271,
+      "step": 45270
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.985956990893089e-05,
+      "loss": 1.5548,
+      "step": 45280
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.985842197903114e-05,
+      "loss": 1.5247,
+      "step": 45290
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.985727404913139e-05,
+      "loss": 1.4922,
+      "step": 45300
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.985612611923164e-05,
+      "loss": 1.639,
+      "step": 45310
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.98549781893319e-05,
+      "loss": 1.5826,
+      "step": 45320
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.985383025943215e-05,
+      "loss": 1.5059,
+      "step": 45330
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.98526823295324e-05,
+      "loss": 1.5889,
+      "step": 45340
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.985153439963265e-05,
+      "loss": 1.4868,
+      "step": 45350
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.985038646973292e-05,
+      "loss": 1.4205,
+      "step": 45360
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.984923853983315e-05,
+      "loss": 1.4349,
+      "step": 45370
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.984809060993341e-05,
+      "loss": 1.5123,
+      "step": 45380
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.984694268003366e-05,
+      "loss": 1.5212,
+      "step": 45390
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.984579475013391e-05,
+      "loss": 1.5038,
+      "step": 45400
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.984464682023416e-05,
+      "loss": 1.4911,
+      "step": 45410
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.984349889033443e-05,
+      "loss": 1.6029,
+      "step": 45420
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.984235096043468e-05,
+      "loss": 1.6203,
+      "step": 45430
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.984120303053493e-05,
+      "loss": 1.5155,
+      "step": 45440
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.984005510063518e-05,
+      "loss": 1.4956,
+      "step": 45450
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.983890717073543e-05,
+      "loss": 1.4937,
+      "step": 45460
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.983775924083569e-05,
+      "loss": 1.5379,
+      "step": 45470
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.983661131093594e-05,
+      "loss": 1.5568,
+      "step": 45480
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.983546338103619e-05,
+      "loss": 1.6076,
+      "step": 45490
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.983431545113644e-05,
+      "loss": 1.548,
+      "step": 45500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.98331675212367e-05,
+      "loss": 1.5093,
+      "step": 45510
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.983201959133694e-05,
+      "loss": 1.4089,
+      "step": 45520
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.98308716614372e-05,
+      "loss": 1.4479,
+      "step": 45530
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.982972373153745e-05,
+      "loss": 1.5965,
+      "step": 45540
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.98285758016377e-05,
+      "loss": 1.5172,
+      "step": 45550
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.982742787173795e-05,
+      "loss": 1.5941,
+      "step": 45560
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.982627994183822e-05,
+      "loss": 1.573,
+      "step": 45570
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.982513201193847e-05,
+      "loss": 1.4167,
+      "step": 45580
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.982398408203872e-05,
+      "loss": 1.5124,
+      "step": 45590
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.982283615213897e-05,
+      "loss": 1.5772,
+      "step": 45600
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.982168822223922e-05,
+      "loss": 1.6438,
+      "step": 45610
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.982054029233948e-05,
+      "loss": 1.5118,
+      "step": 45620
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.981939236243973e-05,
+      "loss": 1.5602,
+      "step": 45630
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.981824443253998e-05,
+      "loss": 1.4399,
+      "step": 45640
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.981709650264023e-05,
+      "loss": 1.4972,
+      "step": 45650
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.981594857274049e-05,
+      "loss": 1.6351,
+      "step": 45660
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.981480064284073e-05,
+      "loss": 1.5241,
+      "step": 45670
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.981365271294099e-05,
+      "loss": 1.538,
+      "step": 45680
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.981250478304124e-05,
+      "loss": 1.4526,
+      "step": 45690
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.981135685314149e-05,
+      "loss": 1.5579,
+      "step": 45700
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.981020892324174e-05,
+      "loss": 1.45,
+      "step": 45710
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.9809060993342e-05,
+      "loss": 1.5091,
+      "step": 45720
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.980791306344225e-05,
+      "loss": 1.476,
+      "step": 45730
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.98067651335425e-05,
+      "loss": 1.393,
+      "step": 45740
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.980561720364275e-05,
+      "loss": 1.4975,
+      "step": 45750
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.9804469273743e-05,
+      "loss": 1.533,
+      "step": 45760
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.980332134384327e-05,
+      "loss": 1.5256,
+      "step": 45770
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.980217341394352e-05,
+      "loss": 1.4759,
+      "step": 45780
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.980102548404377e-05,
+      "loss": 1.4745,
+      "step": 45790
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.979987755414402e-05,
+      "loss": 1.5952,
+      "step": 45800
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.979872962424428e-05,
+      "loss": 1.4787,
+      "step": 45810
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.979758169434452e-05,
+      "loss": 1.7151,
+      "step": 45820
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.979643376444478e-05,
+      "loss": 1.6596,
+      "step": 45830
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.979528583454503e-05,
+      "loss": 1.592,
+      "step": 45840
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.979413790464528e-05,
+      "loss": 1.4302,
+      "step": 45850
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.979298997474553e-05,
+      "loss": 1.4337,
+      "step": 45860
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.979184204484579e-05,
+      "loss": 1.4082,
+      "step": 45870
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.979069411494604e-05,
+      "loss": 1.497,
+      "step": 45880
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.978954618504629e-05,
+      "loss": 1.4731,
+      "step": 45890
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.978839825514654e-05,
+      "loss": 1.6786,
+      "step": 45900
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.978725032524679e-05,
+      "loss": 1.5608,
+      "step": 45910
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.978610239534706e-05,
+      "loss": 1.4852,
+      "step": 45920
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.97849544654473e-05,
+      "loss": 1.5031,
+      "step": 45930
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.978380653554756e-05,
+      "loss": 1.5021,
+      "step": 45940
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.97826586056478e-05,
+      "loss": 1.4673,
+      "step": 45950
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.978151067574807e-05,
+      "loss": 1.4606,
+      "step": 45960
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.97803627458483e-05,
+      "loss": 1.5011,
+      "step": 45970
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.977921481594857e-05,
+      "loss": 1.5688,
+      "step": 45980
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.977806688604882e-05,
+      "loss": 1.5644,
+      "step": 45990
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.977691895614907e-05,
+      "loss": 1.4855,
+      "step": 46000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.977577102624932e-05,
+      "loss": 1.6815,
+      "step": 46010
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.977462309634958e-05,
+      "loss": 1.5898,
+      "step": 46020
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.977347516644983e-05,
+      "loss": 1.342,
+      "step": 46030
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.977232723655008e-05,
+      "loss": 1.7503,
+      "step": 46040
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.977117930665033e-05,
+      "loss": 1.5352,
+      "step": 46050
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.977003137675058e-05,
+      "loss": 1.4181,
+      "step": 46060
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.976888344685084e-05,
+      "loss": 1.4512,
+      "step": 46070
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.97677355169511e-05,
+      "loss": 1.483,
+      "step": 46080
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.976658758705134e-05,
+      "loss": 1.4348,
+      "step": 46090
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.97654396571516e-05,
+      "loss": 1.536,
+      "step": 46100
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.976429172725186e-05,
+      "loss": 1.4072,
+      "step": 46110
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.97631437973521e-05,
+      "loss": 1.391,
+      "step": 46120
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.976199586745236e-05,
+      "loss": 1.5241,
+      "step": 46130
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.976084793755261e-05,
+      "loss": 1.503,
+      "step": 46140
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.975970000765286e-05,
+      "loss": 1.4319,
+      "step": 46150
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.97585520777531e-05,
+      "loss": 1.367,
+      "step": 46160
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.975740414785337e-05,
+      "loss": 1.4074,
+      "step": 46170
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.975625621795362e-05,
+      "loss": 1.6665,
+      "step": 46180
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.975510828805387e-05,
+      "loss": 1.3578,
+      "step": 46190
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.975396035815412e-05,
+      "loss": 1.4786,
+      "step": 46200
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.975281242825437e-05,
+      "loss": 1.545,
+      "step": 46210
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.975166449835463e-05,
+      "loss": 1.4502,
+      "step": 46220
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.975051656845488e-05,
+      "loss": 1.4959,
+      "step": 46230
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.974936863855513e-05,
+      "loss": 1.5245,
+      "step": 46240
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.974822070865538e-05,
+      "loss": 1.4505,
+      "step": 46250
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.974707277875565e-05,
+      "loss": 1.4756,
+      "step": 46260
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.974592484885588e-05,
+      "loss": 1.4568,
+      "step": 46270
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.974477691895615e-05,
+      "loss": 1.5546,
+      "step": 46280
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.97436289890564e-05,
+      "loss": 1.3837,
+      "step": 46290
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.974248105915665e-05,
+      "loss": 1.7114,
+      "step": 46300
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.97413331292569e-05,
+      "loss": 1.397,
+      "step": 46310
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.974018519935716e-05,
+      "loss": 1.5536,
+      "step": 46320
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.973903726945741e-05,
+      "loss": 1.5318,
+      "step": 46330
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.973788933955766e-05,
+      "loss": 1.6034,
+      "step": 46340
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.973674140965791e-05,
+      "loss": 1.403,
+      "step": 46350
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.973559347975816e-05,
+      "loss": 1.4711,
+      "step": 46360
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.973444554985842e-05,
+      "loss": 1.6046,
+      "step": 46370
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.973329761995867e-05,
+      "loss": 1.3889,
+      "step": 46380
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.973214969005892e-05,
+      "loss": 1.5593,
+      "step": 46390
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.973100176015917e-05,
+      "loss": 1.6702,
+      "step": 46400
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.972985383025943e-05,
+      "loss": 1.4997,
+      "step": 46410
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.972870590035967e-05,
+      "loss": 1.4793,
+      "step": 46420
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.972755797045993e-05,
+      "loss": 1.4856,
+      "step": 46430
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.972641004056018e-05,
+      "loss": 1.4083,
+      "step": 46440
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.972526211066043e-05,
+      "loss": 1.6255,
+      "step": 46450
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.972411418076068e-05,
+      "loss": 1.3795,
+      "step": 46460
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.972296625086095e-05,
+      "loss": 1.4736,
+      "step": 46470
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.97218183209612e-05,
+      "loss": 1.6881,
+      "step": 46480
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.972067039106145e-05,
+      "loss": 1.5383,
+      "step": 46490
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.97195224611617e-05,
+      "loss": 1.5631,
+      "step": 46500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.971837453126195e-05,
+      "loss": 1.3759,
+      "step": 46510
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.971722660136221e-05,
+      "loss": 1.4919,
+      "step": 46520
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.971607867146245e-05,
+      "loss": 1.4873,
+      "step": 46530
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.971493074156271e-05,
+      "loss": 1.4449,
+      "step": 46540
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.971378281166296e-05,
+      "loss": 1.3978,
+      "step": 46550
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.971263488176321e-05,
+      "loss": 1.4919,
+      "step": 46560
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.971148695186346e-05,
+      "loss": 1.5786,
+      "step": 46570
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.971033902196372e-05,
+      "loss": 1.4772,
+      "step": 46580
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.970919109206397e-05,
+      "loss": 1.4677,
+      "step": 46590
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.970804316216422e-05,
+      "loss": 1.6833,
+      "step": 46600
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.970689523226447e-05,
+      "loss": 1.5529,
+      "step": 46610
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.970574730236472e-05,
+      "loss": 1.3888,
+      "step": 46620
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.970459937246499e-05,
+      "loss": 1.3925,
+      "step": 46630
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.970345144256524e-05,
+      "loss": 1.602,
+      "step": 46640
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.970230351266549e-05,
+      "loss": 1.593,
+      "step": 46650
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.970115558276573e-05,
+      "loss": 1.3897,
+      "step": 46660
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.9700007652866e-05,
+      "loss": 1.4863,
+      "step": 46670
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.969885972296623e-05,
+      "loss": 1.6452,
+      "step": 46680
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.96977117930665e-05,
+      "loss": 1.571,
+      "step": 46690
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.969656386316675e-05,
+      "loss": 1.484,
+      "step": 46700
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.9695415933267e-05,
+      "loss": 1.5189,
+      "step": 46710
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.969426800336725e-05,
+      "loss": 1.6161,
+      "step": 46720
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.969312007346751e-05,
+      "loss": 1.4773,
+      "step": 46730
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.969197214356776e-05,
+      "loss": 1.3764,
+      "step": 46740
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.969082421366801e-05,
+      "loss": 1.5857,
+      "step": 46750
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.968967628376826e-05,
+      "loss": 1.476,
+      "step": 46760
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.968852835386851e-05,
+      "loss": 1.4632,
+      "step": 46770
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.968738042396877e-05,
+      "loss": 1.3218,
+      "step": 46780
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.968623249406902e-05,
+      "loss": 1.5934,
+      "step": 46790
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.968508456416927e-05,
+      "loss": 1.4317,
+      "step": 46800
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.968393663426952e-05,
+      "loss": 1.3649,
+      "step": 46810
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.968278870436979e-05,
+      "loss": 1.6732,
+      "step": 46820
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.968164077447002e-05,
+      "loss": 1.5798,
+      "step": 46830
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.968049284457029e-05,
+      "loss": 1.3944,
+      "step": 46840
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.967934491467054e-05,
+      "loss": 1.4722,
+      "step": 46850
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.967819698477079e-05,
+      "loss": 1.6145,
+      "step": 46860
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.967704905487104e-05,
+      "loss": 1.6654,
+      "step": 46870
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.96759011249713e-05,
+      "loss": 1.3737,
+      "step": 46880
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.967475319507155e-05,
+      "loss": 1.412,
+      "step": 46890
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.96736052651718e-05,
+      "loss": 1.6133,
+      "step": 46900
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.967245733527205e-05,
+      "loss": 1.6304,
+      "step": 46910
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.96713094053723e-05,
+      "loss": 1.6173,
+      "step": 46920
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.967016147547256e-05,
+      "loss": 1.4824,
+      "step": 46930
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.966901354557281e-05,
+      "loss": 1.5334,
+      "step": 46940
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.966786561567306e-05,
+      "loss": 1.5099,
+      "step": 46950
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.966671768577331e-05,
+      "loss": 1.7246,
+      "step": 46960
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.966556975587358e-05,
+      "loss": 1.5941,
+      "step": 46970
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.966442182597381e-05,
+      "loss": 1.4714,
+      "step": 46980
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.966327389607408e-05,
+      "loss": 1.458,
+      "step": 46990
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.966212596617432e-05,
+      "loss": 1.4206,
+      "step": 47000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.966097803627457e-05,
+      "loss": 1.5573,
+      "step": 47010
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.965983010637482e-05,
+      "loss": 1.6275,
+      "step": 47020
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.965868217647509e-05,
+      "loss": 1.4037,
+      "step": 47030
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.965753424657534e-05,
+      "loss": 1.6974,
+      "step": 47040
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.965638631667559e-05,
+      "loss": 1.4449,
+      "step": 47050
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.965523838677584e-05,
+      "loss": 1.3958,
+      "step": 47060
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.965409045687609e-05,
+      "loss": 1.5052,
+      "step": 47070
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.965294252697635e-05,
+      "loss": 1.3052,
+      "step": 47080
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.96517945970766e-05,
+      "loss": 1.3836,
+      "step": 47090
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.965064666717685e-05,
+      "loss": 1.5497,
+      "step": 47100
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.96494987372771e-05,
+      "loss": 1.4918,
+      "step": 47110
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.964835080737736e-05,
+      "loss": 1.524,
+      "step": 47120
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.96472028774776e-05,
+      "loss": 1.4386,
+      "step": 47130
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.964605494757786e-05,
+      "loss": 1.6821,
+      "step": 47140
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.964490701767811e-05,
+      "loss": 1.4696,
+      "step": 47150
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.964375908777836e-05,
+      "loss": 1.6011,
+      "step": 47160
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.964261115787861e-05,
+      "loss": 1.4329,
+      "step": 47170
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.964146322797888e-05,
+      "loss": 1.5334,
+      "step": 47180
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.964031529807913e-05,
+      "loss": 1.5399,
+      "step": 47190
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.963916736817938e-05,
+      "loss": 1.4969,
+      "step": 47200
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.963801943827963e-05,
+      "loss": 1.4829,
+      "step": 47210
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.963687150837988e-05,
+      "loss": 1.5399,
+      "step": 47220
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.963572357848014e-05,
+      "loss": 1.4563,
+      "step": 47230
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.963457564858039e-05,
+      "loss": 1.2891,
+      "step": 47240
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.963342771868064e-05,
+      "loss": 1.4491,
+      "step": 47250
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.963227978878089e-05,
+      "loss": 1.2325,
+      "step": 47260
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.963113185888115e-05,
+      "loss": 1.4039,
+      "step": 47270
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.962998392898139e-05,
+      "loss": 1.4914,
+      "step": 47280
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.962883599908165e-05,
+      "loss": 1.3777,
+      "step": 47290
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.96276880691819e-05,
+      "loss": 1.6142,
+      "step": 47300
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.962654013928215e-05,
+      "loss": 1.4821,
+      "step": 47310
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.96253922093824e-05,
+      "loss": 1.5191,
+      "step": 47320
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.962424427948266e-05,
+      "loss": 1.4272,
+      "step": 47330
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.962309634958291e-05,
+      "loss": 1.5229,
+      "step": 47340
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.962194841968316e-05,
+      "loss": 1.5093,
+      "step": 47350
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.962080048978341e-05,
+      "loss": 1.5312,
+      "step": 47360
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.961965255988366e-05,
+      "loss": 1.4557,
+      "step": 47370
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.961850462998393e-05,
+      "loss": 1.5565,
+      "step": 47380
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.961735670008418e-05,
+      "loss": 1.5322,
+      "step": 47390
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.961620877018443e-05,
+      "loss": 1.342,
+      "step": 47400
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.961506084028468e-05,
+      "loss": 1.487,
+      "step": 47410
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.961391291038494e-05,
+      "loss": 1.3696,
+      "step": 47420
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.961276498048518e-05,
+      "loss": 1.5095,
+      "step": 47430
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.961161705058544e-05,
+      "loss": 1.5645,
+      "step": 47440
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.961046912068569e-05,
+      "loss": 1.5948,
+      "step": 47450
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.960932119078594e-05,
+      "loss": 1.3072,
+      "step": 47460
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.960817326088619e-05,
+      "loss": 1.5378,
+      "step": 47470
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.960702533098645e-05,
+      "loss": 1.3764,
+      "step": 47480
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.96058774010867e-05,
+      "loss": 1.6386,
+      "step": 47490
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.960472947118695e-05,
+      "loss": 1.603,
+      "step": 47500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.96035815412872e-05,
+      "loss": 1.5189,
+      "step": 47510
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.960243361138745e-05,
+      "loss": 1.4059,
+      "step": 47520
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.960128568148772e-05,
+      "loss": 1.3684,
+      "step": 47530
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.960013775158797e-05,
+      "loss": 1.5222,
+      "step": 47540
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.959898982168822e-05,
+      "loss": 1.4276,
+      "step": 47550
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.959784189178847e-05,
+      "loss": 1.4128,
+      "step": 47560
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.959669396188873e-05,
+      "loss": 1.5524,
+      "step": 47570
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.959554603198897e-05,
+      "loss": 1.5099,
+      "step": 47580
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.959439810208923e-05,
+      "loss": 1.5735,
+      "step": 47590
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.959325017218948e-05,
+      "loss": 1.6053,
+      "step": 47600
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.959210224228973e-05,
+      "loss": 1.3746,
+      "step": 47610
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.959095431238998e-05,
+      "loss": 1.4785,
+      "step": 47620
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.958980638249024e-05,
+      "loss": 1.4216,
+      "step": 47630
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.958865845259049e-05,
+      "loss": 1.5037,
+      "step": 47640
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.958751052269074e-05,
+      "loss": 1.5913,
+      "step": 47650
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.958636259279099e-05,
+      "loss": 1.3409,
+      "step": 47660
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.958521466289124e-05,
+      "loss": 1.4454,
+      "step": 47670
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.95840667329915e-05,
+      "loss": 1.4415,
+      "step": 47680
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.958291880309175e-05,
+      "loss": 1.5572,
+      "step": 47690
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.9581770873192e-05,
+      "loss": 1.3993,
+      "step": 47700
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.958062294329225e-05,
+      "loss": 1.5266,
+      "step": 47710
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.957947501339252e-05,
+      "loss": 1.661,
+      "step": 47720
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.957832708349275e-05,
+      "loss": 1.4683,
+      "step": 47730
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.957717915359302e-05,
+      "loss": 1.6146,
+      "step": 47740
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.957603122369327e-05,
+      "loss": 1.3673,
+      "step": 47750
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.957488329379352e-05,
+      "loss": 1.3989,
+      "step": 47760
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.957373536389377e-05,
+      "loss": 1.7446,
+      "step": 47770
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.957258743399403e-05,
+      "loss": 1.4207,
+      "step": 47780
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.957143950409428e-05,
+      "loss": 1.6065,
+      "step": 47790
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.957029157419453e-05,
+      "loss": 1.4857,
+      "step": 47800
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.956914364429478e-05,
+      "loss": 1.4056,
+      "step": 47810
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.956799571439503e-05,
+      "loss": 1.5378,
+      "step": 47820
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.956684778449529e-05,
+      "loss": 1.3878,
+      "step": 47830
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.956569985459554e-05,
+      "loss": 1.7521,
+      "step": 47840
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.956455192469579e-05,
+      "loss": 1.3666,
+      "step": 47850
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.956340399479604e-05,
+      "loss": 1.4231,
+      "step": 47860
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.956225606489629e-05,
+      "loss": 1.4679,
+      "step": 47870
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.956110813499654e-05,
+      "loss": 1.562,
+      "step": 47880
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.95599602050968e-05,
+      "loss": 1.5482,
+      "step": 47890
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.955881227519706e-05,
+      "loss": 1.2137,
+      "step": 47900
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.95576643452973e-05,
+      "loss": 1.4571,
+      "step": 47910
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.955651641539756e-05,
+      "loss": 1.5721,
+      "step": 47920
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.95553684854978e-05,
+      "loss": 1.4994,
+      "step": 47930
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.955422055559807e-05,
+      "loss": 1.4947,
+      "step": 47940
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.955307262569832e-05,
+      "loss": 1.5703,
+      "step": 47950
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.955192469579857e-05,
+      "loss": 1.5898,
+      "step": 47960
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.955077676589882e-05,
+      "loss": 1.4154,
+      "step": 47970
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.954962883599908e-05,
+      "loss": 1.4356,
+      "step": 47980
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.954848090609932e-05,
+      "loss": 1.6556,
+      "step": 47990
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.954733297619958e-05,
+      "loss": 1.5996,
+      "step": 48000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.954618504629983e-05,
+      "loss": 1.4728,
+      "step": 48010
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.954503711640008e-05,
+      "loss": 1.3592,
+      "step": 48020
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.954388918650033e-05,
+      "loss": 1.4566,
+      "step": 48030
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.95427412566006e-05,
+      "loss": 1.6031,
+      "step": 48040
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.954159332670084e-05,
+      "loss": 1.4039,
+      "step": 48050
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.95404453968011e-05,
+      "loss": 1.48,
+      "step": 48060
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.953929746690134e-05,
+      "loss": 1.5256,
+      "step": 48070
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.95381495370016e-05,
+      "loss": 1.474,
+      "step": 48080
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.953700160710186e-05,
+      "loss": 1.3899,
+      "step": 48090
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.953585367720211e-05,
+      "loss": 1.5359,
+      "step": 48100
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.953470574730236e-05,
+      "loss": 1.4179,
+      "step": 48110
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.95335578174026e-05,
+      "loss": 1.3703,
+      "step": 48120
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.953240988750287e-05,
+      "loss": 1.4377,
+      "step": 48130
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.95312619576031e-05,
+      "loss": 1.5149,
+      "step": 48140
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.953011402770337e-05,
+      "loss": 1.4243,
+      "step": 48150
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.952896609780362e-05,
+      "loss": 1.4782,
+      "step": 48160
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.952781816790387e-05,
+      "loss": 1.4009,
+      "step": 48170
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.952667023800412e-05,
+      "loss": 1.7113,
+      "step": 48180
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.952552230810438e-05,
+      "loss": 1.4146,
+      "step": 48190
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.952437437820463e-05,
+      "loss": 1.5323,
+      "step": 48200
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.952322644830488e-05,
+      "loss": 1.5947,
+      "step": 48210
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.952207851840513e-05,
+      "loss": 1.5776,
+      "step": 48220
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.952093058850538e-05,
+      "loss": 1.4436,
+      "step": 48230
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.951978265860565e-05,
+      "loss": 1.4295,
+      "step": 48240
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.95186347287059e-05,
+      "loss": 1.5702,
+      "step": 48250
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.951748679880615e-05,
+      "loss": 1.5354,
+      "step": 48260
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.95163388689064e-05,
+      "loss": 1.4749,
+      "step": 48270
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.951519093900666e-05,
+      "loss": 1.4626,
+      "step": 48280
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.95140430091069e-05,
+      "loss": 1.4395,
+      "step": 48290
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.951289507920716e-05,
+      "loss": 1.5259,
+      "step": 48300
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.951174714930741e-05,
+      "loss": 1.5808,
+      "step": 48310
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.951059921940766e-05,
+      "loss": 1.5339,
+      "step": 48320
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.950945128950791e-05,
+      "loss": 1.7441,
+      "step": 48330
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.950830335960817e-05,
+      "loss": 1.4508,
+      "step": 48340
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.950715542970842e-05,
+      "loss": 1.576,
+      "step": 48350
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.950600749980867e-05,
+      "loss": 1.5515,
+      "step": 48360
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.950485956990892e-05,
+      "loss": 1.4723,
+      "step": 48370
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.950371164000917e-05,
+      "loss": 1.5225,
+      "step": 48380
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.950256371010943e-05,
+      "loss": 1.5567,
+      "step": 48390
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.950141578020968e-05,
+      "loss": 1.5068,
+      "step": 48400
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.950026785030993e-05,
+      "loss": 1.4125,
+      "step": 48410
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.949911992041018e-05,
+      "loss": 1.5073,
+      "step": 48420
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.949797199051045e-05,
+      "loss": 1.4414,
+      "step": 48430
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.949682406061068e-05,
+      "loss": 1.4201,
+      "step": 48440
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.949567613071095e-05,
+      "loss": 1.4767,
+      "step": 48450
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.94945282008112e-05,
+      "loss": 1.3627,
+      "step": 48460
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.949338027091145e-05,
+      "loss": 1.4824,
+      "step": 48470
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.94922323410117e-05,
+      "loss": 1.4533,
+      "step": 48480
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.949108441111196e-05,
+      "loss": 1.3722,
+      "step": 48490
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.948993648121221e-05,
+      "loss": 1.5827,
+      "step": 48500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.948878855131246e-05,
+      "loss": 1.5309,
+      "step": 48510
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.948764062141271e-05,
+      "loss": 1.5489,
+      "step": 48520
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.948649269151296e-05,
+      "loss": 1.5266,
+      "step": 48530
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.948534476161322e-05,
+      "loss": 1.3267,
+      "step": 48540
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.948419683171347e-05,
+      "loss": 1.389,
+      "step": 48550
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.948304890181372e-05,
+      "loss": 1.7098,
+      "step": 48560
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.948190097191397e-05,
+      "loss": 1.3633,
+      "step": 48570
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.948075304201424e-05,
+      "loss": 1.4493,
+      "step": 48580
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.947960511211447e-05,
+      "loss": 1.418,
+      "step": 48590
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.947845718221474e-05,
+      "loss": 1.4654,
+      "step": 48600
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.947730925231499e-05,
+      "loss": 1.4003,
+      "step": 48610
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.947616132241523e-05,
+      "loss": 1.4656,
+      "step": 48620
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.947501339251548e-05,
+      "loss": 1.3385,
+      "step": 48630
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.947386546261575e-05,
+      "loss": 1.5732,
+      "step": 48640
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.9472717532716e-05,
+      "loss": 1.5568,
+      "step": 48650
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.947156960281625e-05,
+      "loss": 1.6411,
+      "step": 48660
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.94704216729165e-05,
+      "loss": 1.2941,
+      "step": 48670
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.946927374301675e-05,
+      "loss": 1.4894,
+      "step": 48680
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.946812581311701e-05,
+      "loss": 1.5288,
+      "step": 48690
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.946697788321726e-05,
+      "loss": 1.399,
+      "step": 48700
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.946582995331751e-05,
+      "loss": 1.5308,
+      "step": 48710
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.946468202341776e-05,
+      "loss": 1.2843,
+      "step": 48720
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.946353409351802e-05,
+      "loss": 1.5241,
+      "step": 48730
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.946238616361826e-05,
+      "loss": 1.5309,
+      "step": 48740
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.946123823371852e-05,
+      "loss": 1.4693,
+      "step": 48750
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.946009030381877e-05,
+      "loss": 1.4937,
+      "step": 48760
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.945894237391902e-05,
+      "loss": 1.3183,
+      "step": 48770
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.945779444401927e-05,
+      "loss": 1.5719,
+      "step": 48780
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.945664651411954e-05,
+      "loss": 1.5955,
+      "step": 48790
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.945549858421979e-05,
+      "loss": 1.3577,
+      "step": 48800
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.945435065432004e-05,
+      "loss": 1.7604,
+      "step": 48810
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.945320272442029e-05,
+      "loss": 1.5998,
+      "step": 48820
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.945205479452054e-05,
+      "loss": 1.4908,
+      "step": 48830
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.94509068646208e-05,
+      "loss": 1.5171,
+      "step": 48840
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.944975893472105e-05,
+      "loss": 1.4699,
+      "step": 48850
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.94486110048213e-05,
+      "loss": 1.4228,
+      "step": 48860
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.944746307492155e-05,
+      "loss": 1.4624,
+      "step": 48870
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.944631514502181e-05,
+      "loss": 1.5544,
+      "step": 48880
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.944516721512205e-05,
+      "loss": 1.3989,
+      "step": 48890
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.944401928522231e-05,
+      "loss": 1.584,
+      "step": 48900
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.944287135532256e-05,
+      "loss": 1.4555,
+      "step": 48910
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.944172342542281e-05,
+      "loss": 1.6165,
+      "step": 48920
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.944057549552306e-05,
+      "loss": 1.5626,
+      "step": 48930
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.943942756562333e-05,
+      "loss": 1.4992,
+      "step": 48940
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.943827963572358e-05,
+      "loss": 1.4625,
+      "step": 48950
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.943713170582382e-05,
+      "loss": 1.4802,
+      "step": 48960
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.943598377592407e-05,
+      "loss": 1.5935,
+      "step": 48970
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.943483584602432e-05,
+      "loss": 1.4178,
+      "step": 48980
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.943368791612459e-05,
+      "loss": 1.6215,
+      "step": 48990
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.943253998622484e-05,
+      "loss": 1.5515,
+      "step": 49000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.943139205632509e-05,
+      "loss": 1.607,
+      "step": 49010
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.943024412642534e-05,
+      "loss": 1.4525,
+      "step": 49020
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.94290961965256e-05,
+      "loss": 1.4949,
+      "step": 49030
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.942794826662584e-05,
+      "loss": 1.5578,
+      "step": 49040
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.94268003367261e-05,
+      "loss": 1.4572,
+      "step": 49050
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.942565240682635e-05,
+      "loss": 1.4565,
+      "step": 49060
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.94245044769266e-05,
+      "loss": 1.3969,
+      "step": 49070
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.942335654702685e-05,
+      "loss": 1.6506,
+      "step": 49080
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.942220861712711e-05,
+      "loss": 1.492,
+      "step": 49090
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.942106068722736e-05,
+      "loss": 1.3511,
+      "step": 49100
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.941991275732761e-05,
+      "loss": 1.4226,
+      "step": 49110
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.941876482742786e-05,
+      "loss": 1.5683,
+      "step": 49120
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.941761689752811e-05,
+      "loss": 1.3516,
+      "step": 49130
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.941646896762838e-05,
+      "loss": 1.4947,
+      "step": 49140
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.941532103772863e-05,
+      "loss": 1.3052,
+      "step": 49150
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.941417310782888e-05,
+      "loss": 1.2999,
+      "step": 49160
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.941302517792913e-05,
+      "loss": 1.4013,
+      "step": 49170
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.941187724802939e-05,
+      "loss": 1.4595,
+      "step": 49180
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.941072931812963e-05,
+      "loss": 1.4245,
+      "step": 49190
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.940958138822989e-05,
+      "loss": 1.2648,
+      "step": 49200
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.940843345833014e-05,
+      "loss": 1.5364,
+      "step": 49210
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.940728552843039e-05,
+      "loss": 1.4847,
+      "step": 49220
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.940613759853064e-05,
+      "loss": 1.3523,
+      "step": 49230
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.940498966863089e-05,
+      "loss": 1.4125,
+      "step": 49240
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.940384173873115e-05,
+      "loss": 1.4654,
+      "step": 49250
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.94026938088314e-05,
+      "loss": 1.4967,
+      "step": 49260
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.940154587893165e-05,
+      "loss": 1.4214,
+      "step": 49270
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.94003979490319e-05,
+      "loss": 1.4112,
+      "step": 49280
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.939925001913216e-05,
+      "loss": 1.4829,
+      "step": 49290
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.93981020892324e-05,
+      "loss": 1.5818,
+      "step": 49300
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.939695415933266e-05,
+      "loss": 1.5762,
+      "step": 49310
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.939580622943291e-05,
+      "loss": 1.3578,
+      "step": 49320
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.939465829953316e-05,
+      "loss": 1.6279,
+      "step": 49330
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.939351036963341e-05,
+      "loss": 1.4587,
+      "step": 49340
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.939236243973368e-05,
+      "loss": 1.5178,
+      "step": 49350
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.939121450983393e-05,
+      "loss": 1.4473,
+      "step": 49360
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.939006657993418e-05,
+      "loss": 1.4281,
+      "step": 49370
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.938891865003443e-05,
+      "loss": 1.4523,
+      "step": 49380
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.938777072013468e-05,
+      "loss": 1.5054,
+      "step": 49390
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.938662279023494e-05,
+      "loss": 1.3953,
+      "step": 49400
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.938547486033519e-05,
+      "loss": 1.4966,
+      "step": 49410
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.938432693043544e-05,
+      "loss": 1.5713,
+      "step": 49420
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.938317900053569e-05,
+      "loss": 1.3717,
+      "step": 49430
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.938203107063595e-05,
+      "loss": 1.6064,
+      "step": 49440
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.938088314073619e-05,
+      "loss": 1.5045,
+      "step": 49450
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.937973521083645e-05,
+      "loss": 1.548,
+      "step": 49460
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.93785872809367e-05,
+      "loss": 1.515,
+      "step": 49470
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.937743935103695e-05,
+      "loss": 1.5421,
+      "step": 49480
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.93762914211372e-05,
+      "loss": 1.5372,
+      "step": 49490
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.937514349123747e-05,
+      "loss": 1.5893,
+      "step": 49500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.937399556133772e-05,
+      "loss": 1.5604,
+      "step": 49510
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.937284763143797e-05,
+      "loss": 1.2721,
+      "step": 49520
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.937169970153822e-05,
+      "loss": 1.4148,
+      "step": 49530
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.937055177163847e-05,
+      "loss": 1.5271,
+      "step": 49540
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.936940384173873e-05,
+      "loss": 1.6969,
+      "step": 49550
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.936825591183898e-05,
+      "loss": 1.5414,
+      "step": 49560
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.936710798193923e-05,
+      "loss": 1.4505,
+      "step": 49570
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.936596005203948e-05,
+      "loss": 1.6065,
+      "step": 49580
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.936481212213974e-05,
+      "loss": 1.5272,
+      "step": 49590
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.936366419223998e-05,
+      "loss": 1.4509,
+      "step": 49600
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.936251626234024e-05,
+      "loss": 1.652,
+      "step": 49610
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.936136833244049e-05,
+      "loss": 1.5003,
+      "step": 49620
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.936022040254074e-05,
+      "loss": 1.374,
+      "step": 49630
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.935907247264099e-05,
+      "loss": 1.4098,
+      "step": 49640
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.935792454274125e-05,
+      "loss": 1.5677,
+      "step": 49650
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.93567766128415e-05,
+      "loss": 1.5597,
+      "step": 49660
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.935562868294175e-05,
+      "loss": 1.4411,
+      "step": 49670
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.9354480753042e-05,
+      "loss": 1.4904,
+      "step": 49680
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.935333282314225e-05,
+      "loss": 1.58,
+      "step": 49690
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.935218489324252e-05,
+      "loss": 1.3395,
+      "step": 49700
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.935103696334277e-05,
+      "loss": 1.5342,
+      "step": 49710
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.934988903344302e-05,
+      "loss": 1.4303,
+      "step": 49720
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.934874110354327e-05,
+      "loss": 1.4901,
+      "step": 49730
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.934759317364353e-05,
+      "loss": 1.5339,
+      "step": 49740
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.934644524374377e-05,
+      "loss": 1.3781,
+      "step": 49750
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.934529731384403e-05,
+      "loss": 1.4043,
+      "step": 49760
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.934414938394428e-05,
+      "loss": 1.5492,
+      "step": 49770
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.934300145404453e-05,
+      "loss": 1.3646,
+      "step": 49780
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.934185352414478e-05,
+      "loss": 1.4916,
+      "step": 49790
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.934070559424504e-05,
+      "loss": 1.3766,
+      "step": 49800
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.933955766434529e-05,
+      "loss": 1.5123,
+      "step": 49810
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.933840973444554e-05,
+      "loss": 1.4537,
+      "step": 49820
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.933726180454579e-05,
+      "loss": 1.5374,
+      "step": 49830
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.933611387464604e-05,
+      "loss": 1.3993,
+      "step": 49840
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.93349659447463e-05,
+      "loss": 1.4275,
+      "step": 49850
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.933381801484656e-05,
+      "loss": 1.3672,
+      "step": 49860
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.93326700849468e-05,
+      "loss": 1.4818,
+      "step": 49870
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.933152215504706e-05,
+      "loss": 1.2645,
+      "step": 49880
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.933037422514732e-05,
+      "loss": 1.4577,
+      "step": 49890
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.932922629524756e-05,
+      "loss": 1.5544,
+      "step": 49900
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.932807836534782e-05,
+      "loss": 1.4035,
+      "step": 49910
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.932693043544807e-05,
+      "loss": 1.5032,
+      "step": 49920
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.932578250554832e-05,
+      "loss": 1.563,
+      "step": 49930
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.932463457564857e-05,
+      "loss": 1.2872,
+      "step": 49940
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.932348664574883e-05,
+      "loss": 1.478,
+      "step": 49950
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.932233871584908e-05,
+      "loss": 1.5527,
+      "step": 49960
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.932119078594933e-05,
+      "loss": 1.4395,
+      "step": 49970
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.932004285604958e-05,
+      "loss": 1.3583,
+      "step": 49980
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.931889492614983e-05,
+      "loss": 1.4028,
+      "step": 49990
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.93177469962501e-05,
+      "loss": 1.5004,
+      "step": 50000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.931659906635034e-05,
+      "loss": 1.525,
+      "step": 50010
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.93154511364506e-05,
+      "loss": 1.4961,
+      "step": 50020
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.931430320655084e-05,
+      "loss": 1.3174,
+      "step": 50030
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.931315527665111e-05,
+      "loss": 1.4632,
+      "step": 50040
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.931200734675134e-05,
+      "loss": 1.5471,
+      "step": 50050
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.931085941685161e-05,
+      "loss": 1.6334,
+      "step": 50060
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.930971148695186e-05,
+      "loss": 1.4519,
+      "step": 50070
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.93085635570521e-05,
+      "loss": 1.539,
+      "step": 50080
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.930741562715236e-05,
+      "loss": 1.367,
+      "step": 50090
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.930626769725262e-05,
+      "loss": 1.4978,
+      "step": 50100
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.930511976735287e-05,
+      "loss": 1.6578,
+      "step": 50110
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.930397183745312e-05,
+      "loss": 1.4487,
+      "step": 50120
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.930282390755337e-05,
+      "loss": 1.4036,
+      "step": 50130
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.930167597765362e-05,
+      "loss": 1.565,
+      "step": 50140
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.930052804775388e-05,
+      "loss": 1.5616,
+      "step": 50150
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.929938011785413e-05,
+      "loss": 1.3847,
+      "step": 50160
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.929823218795438e-05,
+      "loss": 1.3489,
+      "step": 50170
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.929708425805463e-05,
+      "loss": 1.5186,
+      "step": 50180
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.92959363281549e-05,
+      "loss": 1.4518,
+      "step": 50190
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.929478839825513e-05,
+      "loss": 1.4631,
+      "step": 50200
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.92936404683554e-05,
+      "loss": 1.5274,
+      "step": 50210
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.929249253845565e-05,
+      "loss": 1.4251,
+      "step": 50220
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.92913446085559e-05,
+      "loss": 1.4565,
+      "step": 50230
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.929019667865615e-05,
+      "loss": 1.5873,
+      "step": 50240
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.928904874875641e-05,
+      "loss": 1.3933,
+      "step": 50250
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.928790081885666e-05,
+      "loss": 1.535,
+      "step": 50260
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.928675288895691e-05,
+      "loss": 1.3068,
+      "step": 50270
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.928560495905716e-05,
+      "loss": 1.4165,
+      "step": 50280
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.928445702915741e-05,
+      "loss": 1.5244,
+      "step": 50290
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.928330909925767e-05,
+      "loss": 1.5665,
+      "step": 50300
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.928216116935792e-05,
+      "loss": 1.5444,
+      "step": 50310
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.928101323945817e-05,
+      "loss": 1.3509,
+      "step": 50320
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.927986530955842e-05,
+      "loss": 1.5268,
+      "step": 50330
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.927871737965868e-05,
+      "loss": 1.5447,
+      "step": 50340
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.927756944975892e-05,
+      "loss": 1.5389,
+      "step": 50350
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.927642151985918e-05,
+      "loss": 1.4522,
+      "step": 50360
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.927527358995943e-05,
+      "loss": 1.5724,
+      "step": 50370
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.927412566005968e-05,
+      "loss": 1.4405,
+      "step": 50380
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.927297773015993e-05,
+      "loss": 1.4647,
+      "step": 50390
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.92718298002602e-05,
+      "loss": 1.5382,
+      "step": 50400
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.927068187036045e-05,
+      "loss": 1.4471,
+      "step": 50410
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.92695339404607e-05,
+      "loss": 1.4137,
+      "step": 50420
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.926838601056095e-05,
+      "loss": 1.4111,
+      "step": 50430
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.92672380806612e-05,
+      "loss": 1.4728,
+      "step": 50440
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.926609015076146e-05,
+      "loss": 1.5032,
+      "step": 50450
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.926494222086171e-05,
+      "loss": 1.2668,
+      "step": 50460
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.926379429096196e-05,
+      "loss": 1.4592,
+      "step": 50470
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.926264636106221e-05,
+      "loss": 1.4408,
+      "step": 50480
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.926149843116247e-05,
+      "loss": 1.6125,
+      "step": 50490
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.926035050126271e-05,
+      "loss": 1.6429,
+      "step": 50500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.925920257136297e-05,
+      "loss": 1.499,
+      "step": 50510
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.925805464146322e-05,
+      "loss": 1.6966,
+      "step": 50520
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.925690671156347e-05,
+      "loss": 1.4226,
+      "step": 50530
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.925575878166372e-05,
+      "loss": 1.6261,
+      "step": 50540
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.925461085176399e-05,
+      "loss": 1.3735,
+      "step": 50550
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.925346292186424e-05,
+      "loss": 1.3337,
+      "step": 50560
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.925231499196449e-05,
+      "loss": 1.5212,
+      "step": 50570
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.925116706206473e-05,
+      "loss": 1.3744,
+      "step": 50580
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.925001913216498e-05,
+      "loss": 1.5697,
+      "step": 50590
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.924887120226525e-05,
+      "loss": 1.5353,
+      "step": 50600
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.924772327236548e-05,
+      "loss": 1.4663,
+      "step": 50610
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.924657534246575e-05,
+      "loss": 1.4614,
+      "step": 50620
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.9245427412566e-05,
+      "loss": 1.4298,
+      "step": 50630
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.924427948266625e-05,
+      "loss": 1.65,
+      "step": 50640
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.92431315527665e-05,
+      "loss": 1.4836,
+      "step": 50650
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.924198362286676e-05,
+      "loss": 1.638,
+      "step": 50660
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.924083569296701e-05,
+      "loss": 1.4427,
+      "step": 50670
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.923968776306726e-05,
+      "loss": 1.6017,
+      "step": 50680
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.923853983316751e-05,
+      "loss": 1.6371,
+      "step": 50690
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.923739190326776e-05,
+      "loss": 1.6294,
+      "step": 50700
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.923624397336802e-05,
+      "loss": 1.5328,
+      "step": 50710
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.923509604346827e-05,
+      "loss": 1.5087,
+      "step": 50720
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.923394811356852e-05,
+      "loss": 1.5208,
+      "step": 50730
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.923280018366877e-05,
+      "loss": 1.4636,
+      "step": 50740
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.923165225376904e-05,
+      "loss": 1.4487,
+      "step": 50750
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.923050432386927e-05,
+      "loss": 1.5748,
+      "step": 50760
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.922935639396954e-05,
+      "loss": 1.487,
+      "step": 50770
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.922820846406979e-05,
+      "loss": 1.4397,
+      "step": 50780
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.922706053417004e-05,
+      "loss": 1.4694,
+      "step": 50790
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.922591260427029e-05,
+      "loss": 1.527,
+      "step": 50800
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.922476467437055e-05,
+      "loss": 1.421,
+      "step": 50810
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.92236167444708e-05,
+      "loss": 1.3829,
+      "step": 50820
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.922246881457105e-05,
+      "loss": 1.4318,
+      "step": 50830
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.92213208846713e-05,
+      "loss": 1.5119,
+      "step": 50840
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.922017295477155e-05,
+      "loss": 1.3518,
+      "step": 50850
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.921902502487181e-05,
+      "loss": 1.4255,
+      "step": 50860
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.921787709497206e-05,
+      "loss": 1.4499,
+      "step": 50870
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.921672916507231e-05,
+      "loss": 1.4771,
+      "step": 50880
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.921558123517256e-05,
+      "loss": 1.6163,
+      "step": 50890
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.921443330527283e-05,
+      "loss": 1.5103,
+      "step": 50900
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.921328537537306e-05,
+      "loss": 1.5288,
+      "step": 50910
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.921213744547332e-05,
+      "loss": 1.5468,
+      "step": 50920
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.921098951557357e-05,
+      "loss": 1.3923,
+      "step": 50930
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.920984158567382e-05,
+      "loss": 1.5364,
+      "step": 50940
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.920869365577407e-05,
+      "loss": 1.5911,
+      "step": 50950
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.920754572587434e-05,
+      "loss": 1.6006,
+      "step": 50960
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.920639779597459e-05,
+      "loss": 1.5627,
+      "step": 50970
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.920524986607484e-05,
+      "loss": 1.5057,
+      "step": 50980
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.920410193617509e-05,
+      "loss": 1.4979,
+      "step": 50990
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.920295400627534e-05,
+      "loss": 1.4824,
+      "step": 51000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.92018060763756e-05,
+      "loss": 1.5442,
+      "step": 51010
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.920065814647585e-05,
+      "loss": 1.5545,
+      "step": 51020
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.91995102165761e-05,
+      "loss": 1.3256,
+      "step": 51030
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.919836228667635e-05,
+      "loss": 1.4378,
+      "step": 51040
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.919721435677661e-05,
+      "loss": 1.539,
+      "step": 51050
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.919606642687685e-05,
+      "loss": 1.5837,
+      "step": 51060
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.919491849697711e-05,
+      "loss": 1.7017,
+      "step": 51070
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.919377056707736e-05,
+      "loss": 1.4574,
+      "step": 51080
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.919262263717761e-05,
+      "loss": 1.4851,
+      "step": 51090
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.919147470727786e-05,
+      "loss": 1.4275,
+      "step": 51100
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.919032677737813e-05,
+      "loss": 1.4987,
+      "step": 51110
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.918917884747838e-05,
+      "loss": 1.588,
+      "step": 51120
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.918803091757863e-05,
+      "loss": 1.2723,
+      "step": 51130
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.918688298767888e-05,
+      "loss": 1.484,
+      "step": 51140
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.918573505777913e-05,
+      "loss": 1.4213,
+      "step": 51150
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.918458712787939e-05,
+      "loss": 1.5589,
+      "step": 51160
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.918343919797964e-05,
+      "loss": 1.5168,
+      "step": 51170
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.918229126807989e-05,
+      "loss": 1.3963,
+      "step": 51180
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.918114333818014e-05,
+      "loss": 1.5275,
+      "step": 51190
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.91799954082804e-05,
+      "loss": 1.6555,
+      "step": 51200
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.917884747838064e-05,
+      "loss": 1.5758,
+      "step": 51210
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.91776995484809e-05,
+      "loss": 1.3858,
+      "step": 51220
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.917655161858115e-05,
+      "loss": 1.4321,
+      "step": 51230
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.91754036886814e-05,
+      "loss": 1.412,
+      "step": 51240
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.917425575878165e-05,
+      "loss": 1.4579,
+      "step": 51250
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.917310782888191e-05,
+      "loss": 1.3553,
+      "step": 51260
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.917195989898216e-05,
+      "loss": 1.3682,
+      "step": 51270
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.917081196908241e-05,
+      "loss": 1.4565,
+      "step": 51280
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.916966403918266e-05,
+      "loss": 1.3016,
+      "step": 51290
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.916851610928291e-05,
+      "loss": 1.322,
+      "step": 51300
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.916736817938318e-05,
+      "loss": 1.6025,
+      "step": 51310
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.916622024948343e-05,
+      "loss": 1.4997,
+      "step": 51320
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.916507231958368e-05,
+      "loss": 1.3885,
+      "step": 51330
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.916392438968393e-05,
+      "loss": 1.6167,
+      "step": 51340
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.916277645978419e-05,
+      "loss": 1.3516,
+      "step": 51350
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.916162852988443e-05,
+      "loss": 1.3329,
+      "step": 51360
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.916048059998469e-05,
+      "loss": 1.5636,
+      "step": 51370
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.915933267008494e-05,
+      "loss": 1.5486,
+      "step": 51380
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.915818474018519e-05,
+      "loss": 1.4994,
+      "step": 51390
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.915703681028544e-05,
+      "loss": 1.335,
+      "step": 51400
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.91558888803857e-05,
+      "loss": 1.4518,
+      "step": 51410
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.915474095048595e-05,
+      "loss": 1.4164,
+      "step": 51420
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.91535930205862e-05,
+      "loss": 1.5301,
+      "step": 51430
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.915244509068645e-05,
+      "loss": 1.4824,
+      "step": 51440
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.91512971607867e-05,
+      "loss": 1.3781,
+      "step": 51450
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.915014923088697e-05,
+      "loss": 1.5797,
+      "step": 51460
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.914900130098722e-05,
+      "loss": 1.6345,
+      "step": 51470
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.914785337108747e-05,
+      "loss": 1.3499,
+      "step": 51480
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.914670544118772e-05,
+      "loss": 1.4692,
+      "step": 51490
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.914555751128798e-05,
+      "loss": 1.5151,
+      "step": 51500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.914440958138822e-05,
+      "loss": 1.4592,
+      "step": 51510
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.914326165148848e-05,
+      "loss": 1.7116,
+      "step": 51520
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.914211372158873e-05,
+      "loss": 1.5322,
+      "step": 51530
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.914096579168898e-05,
+      "loss": 1.4695,
+      "step": 51540
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.913981786178923e-05,
+      "loss": 1.5364,
+      "step": 51550
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.913866993188949e-05,
+      "loss": 1.2762,
+      "step": 51560
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.913752200198974e-05,
+      "loss": 1.6781,
+      "step": 51570
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.913637407208999e-05,
+      "loss": 1.5751,
+      "step": 51580
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.913522614219024e-05,
+      "loss": 1.5572,
+      "step": 51590
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.913407821229049e-05,
+      "loss": 1.5323,
+      "step": 51600
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.913293028239075e-05,
+      "loss": 1.3951,
+      "step": 51610
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.9131782352491e-05,
+      "loss": 1.5211,
+      "step": 51620
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.913063442259125e-05,
+      "loss": 1.5326,
+      "step": 51630
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.91294864926915e-05,
+      "loss": 1.5346,
+      "step": 51640
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.912833856279177e-05,
+      "loss": 1.5556,
+      "step": 51650
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.9127190632892e-05,
+      "loss": 1.4954,
+      "step": 51660
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.912604270299227e-05,
+      "loss": 1.5159,
+      "step": 51670
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.912489477309252e-05,
+      "loss": 1.3573,
+      "step": 51680
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.912374684319277e-05,
+      "loss": 1.4231,
+      "step": 51690
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.912259891329302e-05,
+      "loss": 1.6049,
+      "step": 51700
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.912145098339328e-05,
+      "loss": 1.5094,
+      "step": 51710
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.912030305349353e-05,
+      "loss": 1.5144,
+      "step": 51720
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.911915512359378e-05,
+      "loss": 1.4966,
+      "step": 51730
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.911800719369403e-05,
+      "loss": 1.6418,
+      "step": 51740
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.911685926379428e-05,
+      "loss": 1.6205,
+      "step": 51750
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.911571133389454e-05,
+      "loss": 1.4515,
+      "step": 51760
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.911456340399479e-05,
+      "loss": 1.5781,
+      "step": 51770
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.911341547409504e-05,
+      "loss": 1.418,
+      "step": 51780
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.911226754419529e-05,
+      "loss": 1.5834,
+      "step": 51790
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.911111961429556e-05,
+      "loss": 1.4337,
+      "step": 51800
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.910997168439579e-05,
+      "loss": 1.5277,
+      "step": 51810
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.910882375449606e-05,
+      "loss": 1.4944,
+      "step": 51820
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.91076758245963e-05,
+      "loss": 1.5869,
+      "step": 51830
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.910652789469656e-05,
+      "loss": 1.5616,
+      "step": 51840
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.91053799647968e-05,
+      "loss": 1.4019,
+      "step": 51850
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.910423203489707e-05,
+      "loss": 1.349,
+      "step": 51860
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.910308410499732e-05,
+      "loss": 1.5926,
+      "step": 51870
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.910193617509757e-05,
+      "loss": 1.287,
+      "step": 51880
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.910078824519782e-05,
+      "loss": 1.4751,
+      "step": 51890
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.909964031529807e-05,
+      "loss": 1.5701,
+      "step": 51900
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.909849238539833e-05,
+      "loss": 1.4829,
+      "step": 51910
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.909734445549858e-05,
+      "loss": 1.6192,
+      "step": 51920
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.909619652559883e-05,
+      "loss": 1.4411,
+      "step": 51930
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.909504859569908e-05,
+      "loss": 1.5452,
+      "step": 51940
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.909390066579934e-05,
+      "loss": 1.3263,
+      "step": 51950
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.909275273589958e-05,
+      "loss": 1.5584,
+      "step": 51960
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.909160480599984e-05,
+      "loss": 1.6416,
+      "step": 51970
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.90904568761001e-05,
+      "loss": 1.6021,
+      "step": 51980
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.908930894620034e-05,
+      "loss": 1.3781,
+      "step": 51990
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.90881610163006e-05,
+      "loss": 1.4637,
+      "step": 52000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.908701308640084e-05,
+      "loss": 1.5511,
+      "step": 52010
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.908586515650111e-05,
+      "loss": 1.5138,
+      "step": 52020
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.908471722660136e-05,
+      "loss": 1.5668,
+      "step": 52030
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.90835692967016e-05,
+      "loss": 1.6958,
+      "step": 52040
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.908242136680186e-05,
+      "loss": 1.3975,
+      "step": 52050
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.908127343690212e-05,
+      "loss": 1.5176,
+      "step": 52060
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.908012550700236e-05,
+      "loss": 1.5946,
+      "step": 52070
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.907897757710262e-05,
+      "loss": 1.4424,
+      "step": 52080
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.907782964720287e-05,
+      "loss": 1.376,
+      "step": 52090
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.907668171730312e-05,
+      "loss": 1.4787,
+      "step": 52100
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.907553378740337e-05,
+      "loss": 1.5456,
+      "step": 52110
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.907438585750363e-05,
+      "loss": 1.6129,
+      "step": 52120
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.907323792760388e-05,
+      "loss": 1.4866,
+      "step": 52130
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.907208999770413e-05,
+      "loss": 1.4695,
+      "step": 52140
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.907094206780438e-05,
+      "loss": 1.5692,
+      "step": 52150
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.906979413790463e-05,
+      "loss": 1.5819,
+      "step": 52160
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.90686462080049e-05,
+      "loss": 1.5294,
+      "step": 52170
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.906749827810515e-05,
+      "loss": 1.5176,
+      "step": 52180
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.90663503482054e-05,
+      "loss": 1.4756,
+      "step": 52190
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.906520241830565e-05,
+      "loss": 1.382,
+      "step": 52200
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.906405448840591e-05,
+      "loss": 1.4589,
+      "step": 52210
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.906290655850614e-05,
+      "loss": 1.3842,
+      "step": 52220
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.906175862860641e-05,
+      "loss": 1.4066,
+      "step": 52230
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.906061069870666e-05,
+      "loss": 1.3821,
+      "step": 52240
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.905946276880691e-05,
+      "loss": 1.7087,
+      "step": 52250
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.905831483890716e-05,
+      "loss": 1.3215,
+      "step": 52260
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.905716690900742e-05,
+      "loss": 1.4152,
+      "step": 52270
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.905601897910767e-05,
+      "loss": 1.6606,
+      "step": 52280
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.905487104920792e-05,
+      "loss": 1.6391,
+      "step": 52290
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.905372311930817e-05,
+      "loss": 1.4502,
+      "step": 52300
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.905257518940842e-05,
+      "loss": 1.5817,
+      "step": 52310
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.905142725950868e-05,
+      "loss": 1.4224,
+      "step": 52320
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.905027932960893e-05,
+      "loss": 1.3932,
+      "step": 52330
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.904913139970918e-05,
+      "loss": 1.4711,
+      "step": 52340
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.904798346980943e-05,
+      "loss": 1.3898,
+      "step": 52350
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.90468355399097e-05,
+      "loss": 1.5014,
+      "step": 52360
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.904568761000993e-05,
+      "loss": 1.5984,
+      "step": 52370
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.90445396801102e-05,
+      "loss": 1.4858,
+      "step": 52380
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.904339175021045e-05,
+      "loss": 1.6121,
+      "step": 52390
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.90422438203107e-05,
+      "loss": 1.3854,
+      "step": 52400
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.904109589041095e-05,
+      "loss": 1.4974,
+      "step": 52410
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.903994796051121e-05,
+      "loss": 1.5199,
+      "step": 52420
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.903880003061146e-05,
+      "loss": 1.4728,
+      "step": 52430
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.903765210071171e-05,
+      "loss": 1.5087,
+      "step": 52440
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.903650417081196e-05,
+      "loss": 1.3952,
+      "step": 52450
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.903535624091221e-05,
+      "loss": 1.428,
+      "step": 52460
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.903420831101247e-05,
+      "loss": 1.7163,
+      "step": 52470
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.903306038111272e-05,
+      "loss": 1.6221,
+      "step": 52480
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.903191245121297e-05,
+      "loss": 1.4598,
+      "step": 52490
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.903076452131322e-05,
+      "loss": 1.5211,
+      "step": 52500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.902961659141349e-05,
+      "loss": 1.3883,
+      "step": 52510
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.902846866151372e-05,
+      "loss": 1.5044,
+      "step": 52520
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.902732073161399e-05,
+      "loss": 1.4617,
+      "step": 52530
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.902617280171423e-05,
+      "loss": 1.5452,
+      "step": 52540
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.902502487181448e-05,
+      "loss": 1.5324,
+      "step": 52550
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.902387694191473e-05,
+      "loss": 1.5368,
+      "step": 52560
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.9022729012015e-05,
+      "loss": 1.6263,
+      "step": 52570
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.902158108211525e-05,
+      "loss": 1.4947,
+      "step": 52580
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.90204331522155e-05,
+      "loss": 1.5877,
+      "step": 52590
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.901928522231575e-05,
+      "loss": 1.4938,
+      "step": 52600
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.9018137292416e-05,
+      "loss": 1.625,
+      "step": 52610
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.901698936251626e-05,
+      "loss": 1.4983,
+      "step": 52620
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.901584143261651e-05,
+      "loss": 1.5193,
+      "step": 52630
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.901469350271676e-05,
+      "loss": 1.367,
+      "step": 52640
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.901354557281701e-05,
+      "loss": 1.3382,
+      "step": 52650
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.901239764291727e-05,
+      "loss": 1.5475,
+      "step": 52660
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.901124971301751e-05,
+      "loss": 1.6382,
+      "step": 52670
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.901010178311777e-05,
+      "loss": 1.3848,
+      "step": 52680
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.900895385321802e-05,
+      "loss": 1.5069,
+      "step": 52690
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.900780592331827e-05,
+      "loss": 1.5603,
+      "step": 52700
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.900665799341852e-05,
+      "loss": 1.5497,
+      "step": 52710
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.900551006351879e-05,
+      "loss": 1.4856,
+      "step": 52720
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.900436213361904e-05,
+      "loss": 1.498,
+      "step": 52730
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.900321420371929e-05,
+      "loss": 1.503,
+      "step": 52740
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.900206627381954e-05,
+      "loss": 1.305,
+      "step": 52750
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.900091834391979e-05,
+      "loss": 1.3572,
+      "step": 52760
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.899977041402005e-05,
+      "loss": 1.3787,
+      "step": 52770
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.89986224841203e-05,
+      "loss": 1.5399,
+      "step": 52780
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.899747455422055e-05,
+      "loss": 1.4227,
+      "step": 52790
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.89963266243208e-05,
+      "loss": 1.4582,
+      "step": 52800
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.899517869442106e-05,
+      "loss": 1.5747,
+      "step": 52810
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.89940307645213e-05,
+      "loss": 1.4531,
+      "step": 52820
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.899288283462156e-05,
+      "loss": 1.6844,
+      "step": 52830
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.899173490472181e-05,
+      "loss": 1.5934,
+      "step": 52840
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.899058697482206e-05,
+      "loss": 1.6284,
+      "step": 52850
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.898943904492231e-05,
+      "loss": 1.4333,
+      "step": 52860
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.898829111502257e-05,
+      "loss": 1.393,
+      "step": 52870
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.898714318512282e-05,
+      "loss": 1.5762,
+      "step": 52880
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.898599525522307e-05,
+      "loss": 1.5188,
+      "step": 52890
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.898484732532332e-05,
+      "loss": 1.4559,
+      "step": 52900
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.898369939542357e-05,
+      "loss": 1.351,
+      "step": 52910
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.898255146552384e-05,
+      "loss": 1.5389,
+      "step": 52920
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.898140353562409e-05,
+      "loss": 1.5211,
+      "step": 52930
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.898025560572434e-05,
+      "loss": 1.2923,
+      "step": 52940
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.897910767582459e-05,
+      "loss": 1.5317,
+      "step": 52950
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.897795974592485e-05,
+      "loss": 1.3682,
+      "step": 52960
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.897681181602509e-05,
+      "loss": 1.5912,
+      "step": 52970
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.897566388612535e-05,
+      "loss": 1.6731,
+      "step": 52980
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.89745159562256e-05,
+      "loss": 1.4115,
+      "step": 52990
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.897336802632585e-05,
+      "loss": 1.3684,
+      "step": 53000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.89722200964261e-05,
+      "loss": 1.3964,
+      "step": 53010
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.897107216652636e-05,
+      "loss": 1.3067,
+      "step": 53020
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.896992423662661e-05,
+      "loss": 1.4338,
+      "step": 53030
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.896877630672686e-05,
+      "loss": 1.4449,
+      "step": 53040
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.896762837682711e-05,
+      "loss": 1.4767,
+      "step": 53050
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.896648044692736e-05,
+      "loss": 1.5514,
+      "step": 53060
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.896533251702763e-05,
+      "loss": 1.611,
+      "step": 53070
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.896418458712788e-05,
+      "loss": 1.4338,
+      "step": 53080
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.896303665722813e-05,
+      "loss": 1.5792,
+      "step": 53090
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.896188872732838e-05,
+      "loss": 1.3327,
+      "step": 53100
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.896074079742864e-05,
+      "loss": 1.4403,
+      "step": 53110
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.895959286752888e-05,
+      "loss": 1.5366,
+      "step": 53120
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.895844493762914e-05,
+      "loss": 1.3325,
+      "step": 53130
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.895729700772939e-05,
+      "loss": 1.6461,
+      "step": 53140
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.895614907782964e-05,
+      "loss": 1.4527,
+      "step": 53150
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.895500114792989e-05,
+      "loss": 1.3942,
+      "step": 53160
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.895385321803015e-05,
+      "loss": 1.45,
+      "step": 53170
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.89527052881304e-05,
+      "loss": 1.2702,
+      "step": 53180
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.895155735823065e-05,
+      "loss": 1.4824,
+      "step": 53190
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.89504094283309e-05,
+      "loss": 1.3708,
+      "step": 53200
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.894926149843115e-05,
+      "loss": 1.4543,
+      "step": 53210
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.894811356853141e-05,
+      "loss": 1.3584,
+      "step": 53220
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.894696563863166e-05,
+      "loss": 1.4668,
+      "step": 53230
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.894581770873191e-05,
+      "loss": 1.4424,
+      "step": 53240
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.894466977883216e-05,
+      "loss": 1.4163,
+      "step": 53250
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.894352184893243e-05,
+      "loss": 1.4366,
+      "step": 53260
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.894237391903266e-05,
+      "loss": 1.4093,
+      "step": 53270
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.894122598913293e-05,
+      "loss": 1.5219,
+      "step": 53280
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.894007805923318e-05,
+      "loss": 1.3321,
+      "step": 53290
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.893893012933343e-05,
+      "loss": 1.7371,
+      "step": 53300
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.893778219943368e-05,
+      "loss": 1.3493,
+      "step": 53310
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.893663426953394e-05,
+      "loss": 1.4673,
+      "step": 53320
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.893548633963419e-05,
+      "loss": 1.6515,
+      "step": 53330
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.893433840973444e-05,
+      "loss": 1.5431,
+      "step": 53340
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.893319047983469e-05,
+      "loss": 1.3678,
+      "step": 53350
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.893204254993494e-05,
+      "loss": 1.4147,
+      "step": 53360
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.89308946200352e-05,
+      "loss": 1.5262,
+      "step": 53370
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.892974669013544e-05,
+      "loss": 1.3285,
+      "step": 53380
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.89285987602357e-05,
+      "loss": 1.6141,
+      "step": 53390
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.892745083033595e-05,
+      "loss": 1.4936,
+      "step": 53400
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.89263029004362e-05,
+      "loss": 1.1505,
+      "step": 53410
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.892515497053645e-05,
+      "loss": 1.4776,
+      "step": 53420
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.892400704063672e-05,
+      "loss": 1.522,
+      "step": 53430
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.892285911073697e-05,
+      "loss": 1.4249,
+      "step": 53440
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.892171118083722e-05,
+      "loss": 1.5042,
+      "step": 53450
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.892056325093747e-05,
+      "loss": 1.4928,
+      "step": 53460
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.891941532103772e-05,
+      "loss": 1.5947,
+      "step": 53470
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.891826739113798e-05,
+      "loss": 1.5899,
+      "step": 53480
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.891711946123823e-05,
+      "loss": 1.4627,
+      "step": 53490
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.891597153133848e-05,
+      "loss": 1.5184,
+      "step": 53500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.891482360143873e-05,
+      "loss": 1.379,
+      "step": 53510
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.891367567153899e-05,
+      "loss": 1.6069,
+      "step": 53520
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.891252774163923e-05,
+      "loss": 1.454,
+      "step": 53530
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.891137981173949e-05,
+      "loss": 1.5118,
+      "step": 53540
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.891023188183974e-05,
+      "loss": 1.3697,
+      "step": 53550
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.890908395193999e-05,
+      "loss": 1.5992,
+      "step": 53560
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.890793602204024e-05,
+      "loss": 1.4845,
+      "step": 53570
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.89067880921405e-05,
+      "loss": 1.4026,
+      "step": 53580
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.890564016224075e-05,
+      "loss": 1.404,
+      "step": 53590
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.8904492232341e-05,
+      "loss": 1.5148,
+      "step": 53600
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.890334430244125e-05,
+      "loss": 1.3662,
+      "step": 53610
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.89021963725415e-05,
+      "loss": 1.4623,
+      "step": 53620
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.890104844264177e-05,
+      "loss": 1.6018,
+      "step": 53630
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.889990051274202e-05,
+      "loss": 1.6612,
+      "step": 53640
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.889875258284227e-05,
+      "loss": 1.4677,
+      "step": 53650
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.889760465294252e-05,
+      "loss": 1.4954,
+      "step": 53660
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.889645672304278e-05,
+      "loss": 1.7599,
+      "step": 53670
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.889530879314302e-05,
+      "loss": 1.3921,
+      "step": 53680
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.889416086324328e-05,
+      "loss": 1.3352,
+      "step": 53690
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.889301293334353e-05,
+      "loss": 1.7387,
+      "step": 53700
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.889186500344378e-05,
+      "loss": 1.6173,
+      "step": 53710
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.889071707354403e-05,
+      "loss": 1.4975,
+      "step": 53720
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.888956914364429e-05,
+      "loss": 1.4664,
+      "step": 53730
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.888842121374454e-05,
+      "loss": 1.4181,
+      "step": 53740
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.888727328384479e-05,
+      "loss": 1.3854,
+      "step": 53750
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.888612535394504e-05,
+      "loss": 1.4124,
+      "step": 53760
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.888497742404529e-05,
+      "loss": 1.7298,
+      "step": 53770
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.888382949414556e-05,
+      "loss": 1.4937,
+      "step": 53780
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.88826815642458e-05,
+      "loss": 1.4721,
+      "step": 53790
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.888153363434606e-05,
+      "loss": 1.4089,
+      "step": 53800
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.88803857044463e-05,
+      "loss": 1.2848,
+      "step": 53810
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.887923777454657e-05,
+      "loss": 1.4728,
+      "step": 53820
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.88780898446468e-05,
+      "loss": 1.5453,
+      "step": 53830
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.887694191474707e-05,
+      "loss": 1.5078,
+      "step": 53840
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.887579398484732e-05,
+      "loss": 1.3637,
+      "step": 53850
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.887464605494757e-05,
+      "loss": 1.4322,
+      "step": 53860
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.887349812504782e-05,
+      "loss": 1.6156,
+      "step": 53870
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.887235019514808e-05,
+      "loss": 1.4288,
+      "step": 53880
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.887120226524833e-05,
+      "loss": 1.3938,
+      "step": 53890
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.887005433534858e-05,
+      "loss": 1.3764,
+      "step": 53900
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.886890640544883e-05,
+      "loss": 1.5958,
+      "step": 53910
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.886775847554908e-05,
+      "loss": 1.4991,
+      "step": 53920
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.886661054564934e-05,
+      "loss": 1.5436,
+      "step": 53930
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.88654626157496e-05,
+      "loss": 1.3549,
+      "step": 53940
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.886431468584984e-05,
+      "loss": 1.5558,
+      "step": 53950
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.88631667559501e-05,
+      "loss": 1.5646,
+      "step": 53960
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.886201882605036e-05,
+      "loss": 1.5428,
+      "step": 53970
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.88608708961506e-05,
+      "loss": 1.4154,
+      "step": 53980
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.885972296625086e-05,
+      "loss": 1.3638,
+      "step": 53990
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.88585750363511e-05,
+      "loss": 1.5667,
+      "step": 54000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.885742710645136e-05,
+      "loss": 1.4825,
+      "step": 54010
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.88562791765516e-05,
+      "loss": 1.566,
+      "step": 54020
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.885513124665187e-05,
+      "loss": 1.2695,
+      "step": 54030
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.885398331675212e-05,
+      "loss": 1.2822,
+      "step": 54040
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.885283538685237e-05,
+      "loss": 1.4525,
+      "step": 54050
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.885168745695262e-05,
+      "loss": 1.4354,
+      "step": 54060
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.885053952705287e-05,
+      "loss": 1.3333,
+      "step": 54070
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.884939159715313e-05,
+      "loss": 1.5365,
+      "step": 54080
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.884824366725338e-05,
+      "loss": 1.4504,
+      "step": 54090
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.884709573735363e-05,
+      "loss": 1.4353,
+      "step": 54100
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.884594780745388e-05,
+      "loss": 1.3881,
+      "step": 54110
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.884479987755415e-05,
+      "loss": 1.448,
+      "step": 54120
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.884365194765438e-05,
+      "loss": 1.435,
+      "step": 54130
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.884250401775465e-05,
+      "loss": 1.458,
+      "step": 54140
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.88413560878549e-05,
+      "loss": 1.4007,
+      "step": 54150
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.884020815795515e-05,
+      "loss": 1.4979,
+      "step": 54160
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.88390602280554e-05,
+      "loss": 1.4866,
+      "step": 54170
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.883791229815566e-05,
+      "loss": 1.399,
+      "step": 54180
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.883676436825591e-05,
+      "loss": 1.4814,
+      "step": 54190
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.883561643835616e-05,
+      "loss": 1.6047,
+      "step": 54200
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.883446850845641e-05,
+      "loss": 1.4449,
+      "step": 54210
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.883332057855666e-05,
+      "loss": 1.504,
+      "step": 54220
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.883217264865692e-05,
+      "loss": 1.4971,
+      "step": 54230
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.883102471875717e-05,
+      "loss": 1.4774,
+      "step": 54240
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.882987678885742e-05,
+      "loss": 1.6014,
+      "step": 54250
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.882872885895767e-05,
+      "loss": 1.4834,
+      "step": 54260
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.882758092905793e-05,
+      "loss": 1.3747,
+      "step": 54270
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.882643299915817e-05,
+      "loss": 1.4076,
+      "step": 54280
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.882528506925843e-05,
+      "loss": 1.3399,
+      "step": 54290
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.882413713935868e-05,
+      "loss": 1.6707,
+      "step": 54300
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.882298920945893e-05,
+      "loss": 1.5672,
+      "step": 54310
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.882184127955918e-05,
+      "loss": 1.3965,
+      "step": 54320
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.882069334965945e-05,
+      "loss": 1.551,
+      "step": 54330
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.88195454197597e-05,
+      "loss": 1.5021,
+      "step": 54340
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.881839748985995e-05,
+      "loss": 1.4204,
+      "step": 54350
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.88172495599602e-05,
+      "loss": 1.5144,
+      "step": 54360
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.881610163006045e-05,
+      "loss": 1.6936,
+      "step": 54370
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.881495370016071e-05,
+      "loss": 1.475,
+      "step": 54380
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.881380577026096e-05,
+      "loss": 1.431,
+      "step": 54390
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.881265784036121e-05,
+      "loss": 1.5303,
+      "step": 54400
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.881150991046146e-05,
+      "loss": 1.4659,
+      "step": 54410
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.881036198056172e-05,
+      "loss": 1.3364,
+      "step": 54420
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.880921405066196e-05,
+      "loss": 1.3173,
+      "step": 54430
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.880806612076222e-05,
+      "loss": 1.5421,
+      "step": 54440
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.880691819086247e-05,
+      "loss": 1.3333,
+      "step": 54450
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.880577026096272e-05,
+      "loss": 1.4833,
+      "step": 54460
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.880462233106297e-05,
+      "loss": 1.4031,
+      "step": 54470
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.880347440116324e-05,
+      "loss": 1.4453,
+      "step": 54480
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.880232647126349e-05,
+      "loss": 1.5875,
+      "step": 54490
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.880117854136373e-05,
+      "loss": 1.6734,
+      "step": 54500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.880003061146398e-05,
+      "loss": 1.3855,
+      "step": 54510
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.879888268156423e-05,
+      "loss": 1.7339,
+      "step": 54520
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.87977347516645e-05,
+      "loss": 1.3852,
+      "step": 54530
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.879658682176475e-05,
+      "loss": 1.334,
+      "step": 54540
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.8795438891865e-05,
+      "loss": 1.5877,
+      "step": 54550
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.879429096196525e-05,
+      "loss": 1.6696,
+      "step": 54560
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.879314303206551e-05,
+      "loss": 1.6296,
+      "step": 54570
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.879199510216575e-05,
+      "loss": 1.4319,
+      "step": 54580
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.879084717226601e-05,
+      "loss": 1.4133,
+      "step": 54590
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.878969924236626e-05,
+      "loss": 1.3004,
+      "step": 54600
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.878855131246651e-05,
+      "loss": 1.4737,
+      "step": 54610
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.878740338256676e-05,
+      "loss": 1.3661,
+      "step": 54620
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.878625545266702e-05,
+      "loss": 1.3908,
+      "step": 54630
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.878510752276727e-05,
+      "loss": 1.3981,
+      "step": 54640
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.878395959286752e-05,
+      "loss": 1.5,
+      "step": 54650
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.878281166296777e-05,
+      "loss": 1.5463,
+      "step": 54660
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.878166373306802e-05,
+      "loss": 1.6593,
+      "step": 54670
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.878051580316829e-05,
+      "loss": 1.2552,
+      "step": 54680
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.877936787326854e-05,
+      "loss": 1.6081,
+      "step": 54690
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.877821994336879e-05,
+      "loss": 1.4284,
+      "step": 54700
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.877707201346904e-05,
+      "loss": 1.4417,
+      "step": 54710
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.877592408356929e-05,
+      "loss": 1.4524,
+      "step": 54720
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.877477615366954e-05,
+      "loss": 1.4573,
+      "step": 54730
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.87736282237698e-05,
+      "loss": 1.5508,
+      "step": 54740
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.877248029387005e-05,
+      "loss": 1.2775,
+      "step": 54750
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.87713323639703e-05,
+      "loss": 1.4036,
+      "step": 54760
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.877018443407055e-05,
+      "loss": 1.4726,
+      "step": 54770
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.87690365041708e-05,
+      "loss": 1.6099,
+      "step": 54780
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.876788857427106e-05,
+      "loss": 1.5413,
+      "step": 54790
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.876674064437131e-05,
+      "loss": 1.7406,
+      "step": 54800
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.876559271447156e-05,
+      "loss": 1.3727,
+      "step": 54810
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.876444478457181e-05,
+      "loss": 1.4866,
+      "step": 54820
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.876329685467207e-05,
+      "loss": 1.432,
+      "step": 54830
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.876214892477231e-05,
+      "loss": 1.5674,
+      "step": 54840
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.876100099487257e-05,
+      "loss": 1.3943,
+      "step": 54850
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.875985306497282e-05,
+      "loss": 1.5304,
+      "step": 54860
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.875870513507307e-05,
+      "loss": 1.4254,
+      "step": 54870
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.875755720517332e-05,
+      "loss": 1.472,
+      "step": 54880
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.875640927527359e-05,
+      "loss": 1.3051,
+      "step": 54890
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.875526134537384e-05,
+      "loss": 1.6175,
+      "step": 54900
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.875411341547409e-05,
+      "loss": 1.3878,
+      "step": 54910
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.875296548557434e-05,
+      "loss": 1.4961,
+      "step": 54920
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.875181755567459e-05,
+      "loss": 1.6045,
+      "step": 54930
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.875066962577485e-05,
+      "loss": 1.3566,
+      "step": 54940
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.87495216958751e-05,
+      "loss": 1.4023,
+      "step": 54950
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.874837376597535e-05,
+      "loss": 1.4232,
+      "step": 54960
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.87472258360756e-05,
+      "loss": 1.5044,
+      "step": 54970
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.874607790617586e-05,
+      "loss": 1.405,
+      "step": 54980
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.87449299762761e-05,
+      "loss": 1.4756,
+      "step": 54990
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.874378204637636e-05,
+      "loss": 1.6143,
+      "step": 55000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.874263411647661e-05,
+      "loss": 1.3297,
+      "step": 55010
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.874148618657686e-05,
+      "loss": 1.3324,
+      "step": 55020
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.874033825667711e-05,
+      "loss": 1.4269,
+      "step": 55030
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.873919032677738e-05,
+      "loss": 1.5311,
+      "step": 55040
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.873804239687763e-05,
+      "loss": 1.4556,
+      "step": 55050
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.873689446697788e-05,
+      "loss": 1.4652,
+      "step": 55060
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.873574653707813e-05,
+      "loss": 1.7047,
+      "step": 55070
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.873459860717838e-05,
+      "loss": 1.6366,
+      "step": 55080
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.873345067727864e-05,
+      "loss": 1.5438,
+      "step": 55090
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.873230274737889e-05,
+      "loss": 1.4717,
+      "step": 55100
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.873115481747914e-05,
+      "loss": 1.4014,
+      "step": 55110
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.873000688757939e-05,
+      "loss": 1.4657,
+      "step": 55120
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.872885895767965e-05,
+      "loss": 1.5083,
+      "step": 55130
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.872771102777989e-05,
+      "loss": 1.5215,
+      "step": 55140
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.872656309788015e-05,
+      "loss": 1.3695,
+      "step": 55150
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.87254151679804e-05,
+      "loss": 1.3621,
+      "step": 55160
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.872426723808065e-05,
+      "loss": 1.292,
+      "step": 55170
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.87231193081809e-05,
+      "loss": 1.47,
+      "step": 55180
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.872197137828116e-05,
+      "loss": 1.3948,
+      "step": 55190
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.872082344838141e-05,
+      "loss": 1.5888,
+      "step": 55200
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.871967551848166e-05,
+      "loss": 1.6316,
+      "step": 55210
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.871852758858191e-05,
+      "loss": 1.4758,
+      "step": 55220
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.871737965868216e-05,
+      "loss": 1.5219,
+      "step": 55230
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.871623172878243e-05,
+      "loss": 1.3734,
+      "step": 55240
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.871508379888268e-05,
+      "loss": 1.4427,
+      "step": 55250
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.871393586898293e-05,
+      "loss": 1.4541,
+      "step": 55260
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.871278793908318e-05,
+      "loss": 1.629,
+      "step": 55270
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.871164000918344e-05,
+      "loss": 1.4477,
+      "step": 55280
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.871049207928368e-05,
+      "loss": 1.5451,
+      "step": 55290
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.870934414938394e-05,
+      "loss": 1.5667,
+      "step": 55300
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.870819621948419e-05,
+      "loss": 1.5214,
+      "step": 55310
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.870704828958444e-05,
+      "loss": 1.5672,
+      "step": 55320
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.870590035968469e-05,
+      "loss": 1.4353,
+      "step": 55330
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.870475242978495e-05,
+      "loss": 1.3901,
+      "step": 55340
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.87036044998852e-05,
+      "loss": 1.3985,
+      "step": 55350
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.870245656998545e-05,
+      "loss": 1.4264,
+      "step": 55360
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.87013086400857e-05,
+      "loss": 1.4669,
+      "step": 55370
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.870016071018595e-05,
+      "loss": 1.4166,
+      "step": 55380
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.869901278028622e-05,
+      "loss": 1.6329,
+      "step": 55390
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.869786485038647e-05,
+      "loss": 1.5638,
+      "step": 55400
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.869671692048672e-05,
+      "loss": 1.3603,
+      "step": 55410
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.869556899058697e-05,
+      "loss": 1.5599,
+      "step": 55420
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.869442106068723e-05,
+      "loss": 1.4377,
+      "step": 55430
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.869327313078747e-05,
+      "loss": 1.3454,
+      "step": 55440
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.869212520088773e-05,
+      "loss": 1.6072,
+      "step": 55450
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.869097727098798e-05,
+      "loss": 1.3871,
+      "step": 55460
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.868982934108823e-05,
+      "loss": 1.4609,
+      "step": 55470
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.868868141118848e-05,
+      "loss": 1.6138,
+      "step": 55480
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.868753348128874e-05,
+      "loss": 1.4618,
+      "step": 55490
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.868638555138899e-05,
+      "loss": 1.4169,
+      "step": 55500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.868523762148924e-05,
+      "loss": 1.4742,
+      "step": 55510
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.868408969158949e-05,
+      "loss": 1.2507,
+      "step": 55520
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.868294176168974e-05,
+      "loss": 1.325,
+      "step": 55530
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.868179383179e-05,
+      "loss": 1.5446,
+      "step": 55540
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.868064590189025e-05,
+      "loss": 1.4257,
+      "step": 55550
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.86794979719905e-05,
+      "loss": 1.3997,
+      "step": 55560
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.867835004209075e-05,
+      "loss": 1.2696,
+      "step": 55570
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.867720211219102e-05,
+      "loss": 1.4666,
+      "step": 55580
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.867605418229125e-05,
+      "loss": 1.5099,
+      "step": 55590
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.867490625239152e-05,
+      "loss": 1.5751,
+      "step": 55600
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.867375832249177e-05,
+      "loss": 1.4435,
+      "step": 55610
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.867261039259202e-05,
+      "loss": 1.3295,
+      "step": 55620
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.867146246269227e-05,
+      "loss": 1.5516,
+      "step": 55630
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.867031453279253e-05,
+      "loss": 1.3903,
+      "step": 55640
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.866916660289278e-05,
+      "loss": 1.4851,
+      "step": 55650
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.866801867299303e-05,
+      "loss": 1.4977,
+      "step": 55660
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.866687074309328e-05,
+      "loss": 1.5626,
+      "step": 55670
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.866572281319353e-05,
+      "loss": 1.6476,
+      "step": 55680
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.866457488329379e-05,
+      "loss": 1.5061,
+      "step": 55690
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.866342695339404e-05,
+      "loss": 1.5004,
+      "step": 55700
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.866227902349429e-05,
+      "loss": 1.4444,
+      "step": 55710
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.866113109359454e-05,
+      "loss": 1.3999,
+      "step": 55720
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.86599831636948e-05,
+      "loss": 1.4749,
+      "step": 55730
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.865883523379504e-05,
+      "loss": 1.6868,
+      "step": 55740
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.86576873038953e-05,
+      "loss": 1.3429,
+      "step": 55750
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.865653937399556e-05,
+      "loss": 1.6672,
+      "step": 55760
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.86553914440958e-05,
+      "loss": 1.5407,
+      "step": 55770
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.865424351419606e-05,
+      "loss": 1.5358,
+      "step": 55780
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.865309558429632e-05,
+      "loss": 1.4967,
+      "step": 55790
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.865194765439657e-05,
+      "loss": 1.5771,
+      "step": 55800
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.865079972449682e-05,
+      "loss": 1.4551,
+      "step": 55810
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.864965179459707e-05,
+      "loss": 1.3834,
+      "step": 55820
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.864850386469732e-05,
+      "loss": 1.2764,
+      "step": 55830
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.864735593479758e-05,
+      "loss": 1.3273,
+      "step": 55840
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.864620800489783e-05,
+      "loss": 1.5002,
+      "step": 55850
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.864506007499808e-05,
+      "loss": 1.4547,
+      "step": 55860
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.864391214509833e-05,
+      "loss": 1.5793,
+      "step": 55870
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.86427642151986e-05,
+      "loss": 1.4079,
+      "step": 55880
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.864161628529883e-05,
+      "loss": 1.4238,
+      "step": 55890
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.86404683553991e-05,
+      "loss": 1.4668,
+      "step": 55900
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.863932042549934e-05,
+      "loss": 1.4339,
+      "step": 55910
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.86381724955996e-05,
+      "loss": 1.4839,
+      "step": 55920
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.863702456569984e-05,
+      "loss": 1.4488,
+      "step": 55930
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.863587663580011e-05,
+      "loss": 1.4013,
+      "step": 55940
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.863472870590036e-05,
+      "loss": 1.3864,
+      "step": 55950
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.86335807760006e-05,
+      "loss": 1.3892,
+      "step": 55960
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.863243284610086e-05,
+      "loss": 1.3425,
+      "step": 55970
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.86312849162011e-05,
+      "loss": 1.5731,
+      "step": 55980
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.863013698630137e-05,
+      "loss": 1.4948,
+      "step": 55990
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.862898905640162e-05,
+      "loss": 1.5796,
+      "step": 56000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.862784112650187e-05,
+      "loss": 1.4631,
+      "step": 56010
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.862669319660212e-05,
+      "loss": 1.4244,
+      "step": 56020
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.862554526670238e-05,
+      "loss": 1.5419,
+      "step": 56030
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.862439733680262e-05,
+      "loss": 1.5643,
+      "step": 56040
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.862324940690288e-05,
+      "loss": 1.2758,
+      "step": 56050
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.862210147700313e-05,
+      "loss": 1.4921,
+      "step": 56060
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.862095354710338e-05,
+      "loss": 1.2425,
+      "step": 56070
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.861980561720363e-05,
+      "loss": 1.3391,
+      "step": 56080
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.861865768730388e-05,
+      "loss": 1.4553,
+      "step": 56090
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.861750975740415e-05,
+      "loss": 1.4805,
+      "step": 56100
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.86163618275044e-05,
+      "loss": 1.5594,
+      "step": 56110
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.861521389760465e-05,
+      "loss": 1.5183,
+      "step": 56120
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.86140659677049e-05,
+      "loss": 1.4937,
+      "step": 56130
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.861291803780516e-05,
+      "loss": 1.4604,
+      "step": 56140
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.86117701079054e-05,
+      "loss": 1.4397,
+      "step": 56150
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.861062217800566e-05,
+      "loss": 1.5169,
+      "step": 56160
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.860947424810591e-05,
+      "loss": 1.5069,
+      "step": 56170
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.860832631820616e-05,
+      "loss": 1.3078,
+      "step": 56180
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.860717838830641e-05,
+      "loss": 1.5068,
+      "step": 56190
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.860603045840667e-05,
+      "loss": 1.3483,
+      "step": 56200
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.860488252850692e-05,
+      "loss": 1.4856,
+      "step": 56210
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.860373459860717e-05,
+      "loss": 1.5496,
+      "step": 56220
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.860258666870742e-05,
+      "loss": 1.5614,
+      "step": 56230
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.860143873880767e-05,
+      "loss": 1.4325,
+      "step": 56240
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.860029080890793e-05,
+      "loss": 1.5715,
+      "step": 56250
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.859914287900818e-05,
+      "loss": 1.5171,
+      "step": 56260
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.859799494910843e-05,
+      "loss": 1.3095,
+      "step": 56270
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.859684701920868e-05,
+      "loss": 1.4737,
+      "step": 56280
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.859569908930895e-05,
+      "loss": 1.4475,
+      "step": 56290
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.859455115940918e-05,
+      "loss": 1.507,
+      "step": 56300
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.859340322950945e-05,
+      "loss": 1.2991,
+      "step": 56310
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.85922552996097e-05,
+      "loss": 1.4024,
+      "step": 56320
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.859110736970995e-05,
+      "loss": 1.3213,
+      "step": 56330
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.85899594398102e-05,
+      "loss": 1.4392,
+      "step": 56340
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.858881150991046e-05,
+      "loss": 1.5008,
+      "step": 56350
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.858766358001071e-05,
+      "loss": 1.5457,
+      "step": 56360
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.858651565011096e-05,
+      "loss": 1.4712,
+      "step": 56370
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.858536772021121e-05,
+      "loss": 1.5562,
+      "step": 56380
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.858421979031146e-05,
+      "loss": 1.3325,
+      "step": 56390
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.858307186041172e-05,
+      "loss": 1.3953,
+      "step": 56400
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.858192393051197e-05,
+      "loss": 1.3786,
+      "step": 56410
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.858077600061222e-05,
+      "loss": 1.4593,
+      "step": 56420
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.857962807071247e-05,
+      "loss": 1.3931,
+      "step": 56430
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.857848014081274e-05,
+      "loss": 1.5284,
+      "step": 56440
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.857733221091297e-05,
+      "loss": 1.4768,
+      "step": 56450
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.857618428101323e-05,
+      "loss": 1.4902,
+      "step": 56460
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.857503635111348e-05,
+      "loss": 1.3748,
+      "step": 56470
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.857388842121373e-05,
+      "loss": 1.5463,
+      "step": 56480
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.857274049131398e-05,
+      "loss": 1.4625,
+      "step": 56490
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.857159256141425e-05,
+      "loss": 1.4813,
+      "step": 56500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.85704446315145e-05,
+      "loss": 1.4694,
+      "step": 56510
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.856929670161475e-05,
+      "loss": 1.5962,
+      "step": 56520
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.8568148771715e-05,
+      "loss": 1.4254,
+      "step": 56530
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.856700084181525e-05,
+      "loss": 1.5185,
+      "step": 56540
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.856585291191551e-05,
+      "loss": 1.3883,
+      "step": 56550
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.856470498201576e-05,
+      "loss": 1.5527,
+      "step": 56560
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.856355705211601e-05,
+      "loss": 1.4552,
+      "step": 56570
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.856240912221626e-05,
+      "loss": 1.5385,
+      "step": 56580
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.856126119231652e-05,
+      "loss": 1.3816,
+      "step": 56590
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.856011326241676e-05,
+      "loss": 1.5,
+      "step": 56600
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.855896533251702e-05,
+      "loss": 1.4955,
+      "step": 56610
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.855781740261727e-05,
+      "loss": 1.3838,
+      "step": 56620
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.855666947271752e-05,
+      "loss": 1.3892,
+      "step": 56630
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.855552154281777e-05,
+      "loss": 1.3581,
+      "step": 56640
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.855437361291804e-05,
+      "loss": 1.2572,
+      "step": 56650
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.855322568301829e-05,
+      "loss": 1.5358,
+      "step": 56660
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.855207775311854e-05,
+      "loss": 1.4091,
+      "step": 56670
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.855092982321879e-05,
+      "loss": 1.6087,
+      "step": 56680
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.854978189331904e-05,
+      "loss": 1.5375,
+      "step": 56690
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.85486339634193e-05,
+      "loss": 1.536,
+      "step": 56700
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.854748603351955e-05,
+      "loss": 1.4765,
+      "step": 56710
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.85463381036198e-05,
+      "loss": 1.4769,
+      "step": 56720
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.854519017372005e-05,
+      "loss": 1.4401,
+      "step": 56730
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.854404224382031e-05,
+      "loss": 1.3344,
+      "step": 56740
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.854289431392055e-05,
+      "loss": 1.5614,
+      "step": 56750
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.854174638402081e-05,
+      "loss": 1.4419,
+      "step": 56760
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.854059845412106e-05,
+      "loss": 1.5774,
+      "step": 56770
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.853945052422131e-05,
+      "loss": 1.251,
+      "step": 56780
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.853830259432156e-05,
+      "loss": 1.2979,
+      "step": 56790
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.853715466442182e-05,
+      "loss": 1.4693,
+      "step": 56800
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.853600673452207e-05,
+      "loss": 1.4495,
+      "step": 56810
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.853485880462232e-05,
+      "loss": 1.5016,
+      "step": 56820
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.853371087472257e-05,
+      "loss": 1.628,
+      "step": 56830
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.853256294482282e-05,
+      "loss": 1.5179,
+      "step": 56840
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.853141501492309e-05,
+      "loss": 1.5831,
+      "step": 56850
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.853026708502334e-05,
+      "loss": 1.3156,
+      "step": 56860
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.852911915512359e-05,
+      "loss": 1.5651,
+      "step": 56870
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.852797122522384e-05,
+      "loss": 1.4678,
+      "step": 56880
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.85268232953241e-05,
+      "loss": 1.4432,
+      "step": 56890
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.852567536542434e-05,
+      "loss": 1.4826,
+      "step": 56900
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.85245274355246e-05,
+      "loss": 1.5223,
+      "step": 56910
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.852337950562485e-05,
+      "loss": 1.6519,
+      "step": 56920
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.85222315757251e-05,
+      "loss": 1.4313,
+      "step": 56930
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.852108364582535e-05,
+      "loss": 1.3185,
+      "step": 56940
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.851993571592561e-05,
+      "loss": 1.5728,
+      "step": 56950
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.851878778602586e-05,
+      "loss": 1.2944,
+      "step": 56960
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.851763985612611e-05,
+      "loss": 1.4526,
+      "step": 56970
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.851649192622636e-05,
+      "loss": 1.3898,
+      "step": 56980
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.851534399632661e-05,
+      "loss": 1.4569,
+      "step": 56990
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.851419606642688e-05,
+      "loss": 1.4492,
+      "step": 57000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.851304813652713e-05,
+      "loss": 1.5717,
+      "step": 57010
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.851190020662738e-05,
+      "loss": 1.2872,
+      "step": 57020
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.851075227672763e-05,
+      "loss": 1.5542,
+      "step": 57030
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.850960434682789e-05,
+      "loss": 1.2861,
+      "step": 57040
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.850845641692813e-05,
+      "loss": 1.5495,
+      "step": 57050
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.850730848702839e-05,
+      "loss": 1.5561,
+      "step": 57060
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.850616055712864e-05,
+      "loss": 1.495,
+      "step": 57070
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.850501262722889e-05,
+      "loss": 1.4031,
+      "step": 57080
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.850386469732914e-05,
+      "loss": 1.5082,
+      "step": 57090
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.85027167674294e-05,
+      "loss": 1.4481,
+      "step": 57100
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.850156883752965e-05,
+      "loss": 1.459,
+      "step": 57110
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.85004209076299e-05,
+      "loss": 1.3835,
+      "step": 57120
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.849927297773015e-05,
+      "loss": 1.3595,
+      "step": 57130
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.84981250478304e-05,
+      "loss": 1.5513,
+      "step": 57140
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.849697711793066e-05,
+      "loss": 1.6942,
+      "step": 57150
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.849582918803091e-05,
+      "loss": 1.5743,
+      "step": 57160
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.849468125813116e-05,
+      "loss": 1.4155,
+      "step": 57170
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.849353332823141e-05,
+      "loss": 1.5416,
+      "step": 57180
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.849238539833168e-05,
+      "loss": 1.5551,
+      "step": 57190
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.849123746843191e-05,
+      "loss": 1.449,
+      "step": 57200
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.849008953853218e-05,
+      "loss": 1.4189,
+      "step": 57210
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.848894160863243e-05,
+      "loss": 1.4335,
+      "step": 57220
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.848779367873268e-05,
+      "loss": 1.6824,
+      "step": 57230
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.848664574883293e-05,
+      "loss": 1.4799,
+      "step": 57240
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.848549781893319e-05,
+      "loss": 1.6431,
+      "step": 57250
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.848434988903344e-05,
+      "loss": 1.3629,
+      "step": 57260
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.848320195913369e-05,
+      "loss": 1.5015,
+      "step": 57270
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.848205402923394e-05,
+      "loss": 1.5054,
+      "step": 57280
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.848090609933419e-05,
+      "loss": 1.3669,
+      "step": 57290
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.847975816943445e-05,
+      "loss": 1.2989,
+      "step": 57300
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.84786102395347e-05,
+      "loss": 1.4517,
+      "step": 57310
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.847746230963495e-05,
+      "loss": 1.5138,
+      "step": 57320
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.84763143797352e-05,
+      "loss": 1.6206,
+      "step": 57330
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.847516644983547e-05,
+      "loss": 1.3356,
+      "step": 57340
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.84740185199357e-05,
+      "loss": 1.4655,
+      "step": 57350
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.847287059003597e-05,
+      "loss": 1.4165,
+      "step": 57360
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.847172266013622e-05,
+      "loss": 1.5402,
+      "step": 57370
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.847057473023647e-05,
+      "loss": 1.2735,
+      "step": 57380
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.846942680033672e-05,
+      "loss": 1.3226,
+      "step": 57390
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.846827887043698e-05,
+      "loss": 1.5173,
+      "step": 57400
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.846713094053723e-05,
+      "loss": 1.4793,
+      "step": 57410
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.846598301063748e-05,
+      "loss": 1.4214,
+      "step": 57420
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.846483508073773e-05,
+      "loss": 1.4448,
+      "step": 57430
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.846368715083798e-05,
+      "loss": 1.5067,
+      "step": 57440
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.846253922093824e-05,
+      "loss": 1.5216,
+      "step": 57450
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.846139129103849e-05,
+      "loss": 1.5139,
+      "step": 57460
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.846024336113874e-05,
+      "loss": 1.4498,
+      "step": 57470
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.845909543123899e-05,
+      "loss": 1.4624,
+      "step": 57480
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.845794750133924e-05,
+      "loss": 1.4831,
+      "step": 57490
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.845679957143949e-05,
+      "loss": 1.3914,
+      "step": 57500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.845565164153975e-05,
+      "loss": 1.3879,
+      "step": 57510
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.845450371164e-05,
+      "loss": 1.4929,
+      "step": 57520
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.845335578174025e-05,
+      "loss": 1.4228,
+      "step": 57530
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.84522078518405e-05,
+      "loss": 1.3598,
+      "step": 57540
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.845105992194075e-05,
+      "loss": 1.3511,
+      "step": 57550
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.844991199204102e-05,
+      "loss": 1.3701,
+      "step": 57560
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.844876406214127e-05,
+      "loss": 1.4486,
+      "step": 57570
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.844761613224152e-05,
+      "loss": 1.3848,
+      "step": 57580
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.844646820234177e-05,
+      "loss": 1.4498,
+      "step": 57590
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.844532027244203e-05,
+      "loss": 1.5955,
+      "step": 57600
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.844417234254227e-05,
+      "loss": 1.4492,
+      "step": 57610
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.844302441264253e-05,
+      "loss": 1.4872,
+      "step": 57620
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.844187648274278e-05,
+      "loss": 1.6161,
+      "step": 57630
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.844072855284303e-05,
+      "loss": 1.5282,
+      "step": 57640
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.843958062294328e-05,
+      "loss": 1.521,
+      "step": 57650
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.843843269304354e-05,
+      "loss": 1.5481,
+      "step": 57660
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.843728476314379e-05,
+      "loss": 1.4394,
+      "step": 57670
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.843613683324404e-05,
+      "loss": 1.6223,
+      "step": 57680
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.843498890334429e-05,
+      "loss": 1.5234,
+      "step": 57690
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.843384097344454e-05,
+      "loss": 1.6008,
+      "step": 57700
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.84326930435448e-05,
+      "loss": 1.4702,
+      "step": 57710
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.843154511364506e-05,
+      "loss": 1.4979,
+      "step": 57720
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.84303971837453e-05,
+      "loss": 1.2851,
+      "step": 57730
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.842924925384556e-05,
+      "loss": 1.395,
+      "step": 57740
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.842810132394582e-05,
+      "loss": 1.4798,
+      "step": 57750
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.842695339404605e-05,
+      "loss": 1.4199,
+      "step": 57760
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.842580546414632e-05,
+      "loss": 1.3551,
+      "step": 57770
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.842465753424657e-05,
+      "loss": 1.5869,
+      "step": 57780
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.842350960434682e-05,
+      "loss": 1.5707,
+      "step": 57790
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.842236167444707e-05,
+      "loss": 1.3604,
+      "step": 57800
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.842121374454733e-05,
+      "loss": 1.6456,
+      "step": 57810
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.842006581464758e-05,
+      "loss": 1.4552,
+      "step": 57820
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.841891788474783e-05,
+      "loss": 1.4421,
+      "step": 57830
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.841776995484808e-05,
+      "loss": 1.3195,
+      "step": 57840
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.841662202494833e-05,
+      "loss": 1.3957,
+      "step": 57850
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.84154740950486e-05,
+      "loss": 1.6577,
+      "step": 57860
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.841432616514884e-05,
+      "loss": 1.6569,
+      "step": 57870
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.84131782352491e-05,
+      "loss": 1.4943,
+      "step": 57880
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.841203030534934e-05,
+      "loss": 1.3904,
+      "step": 57890
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.841088237544961e-05,
+      "loss": 1.2641,
+      "step": 57900
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.840973444554984e-05,
+      "loss": 1.3741,
+      "step": 57910
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.84085865156501e-05,
+      "loss": 1.4234,
+      "step": 57920
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.840743858575036e-05,
+      "loss": 1.5213,
+      "step": 57930
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.84062906558506e-05,
+      "loss": 1.2232,
+      "step": 57940
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.840514272595086e-05,
+      "loss": 1.5018,
+      "step": 57950
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.840399479605112e-05,
+      "loss": 1.5938,
+      "step": 57960
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.840284686615137e-05,
+      "loss": 1.3957,
+      "step": 57970
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.840169893625162e-05,
+      "loss": 1.5634,
+      "step": 57980
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.840055100635187e-05,
+      "loss": 1.4358,
+      "step": 57990
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.839940307645212e-05,
+      "loss": 1.5469,
+      "step": 58000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.839825514655238e-05,
+      "loss": 1.5,
+      "step": 58010
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.839710721665263e-05,
+      "loss": 1.4316,
+      "step": 58020
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.839595928675288e-05,
+      "loss": 1.5306,
+      "step": 58030
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.839481135685313e-05,
+      "loss": 1.4349,
+      "step": 58040
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.83936634269534e-05,
+      "loss": 1.5217,
+      "step": 58050
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.839251549705363e-05,
+      "loss": 1.4652,
+      "step": 58060
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.83913675671539e-05,
+      "loss": 1.4232,
+      "step": 58070
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.839021963725414e-05,
+      "loss": 1.3158,
+      "step": 58080
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.83890717073544e-05,
+      "loss": 1.5489,
+      "step": 58090
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.838792377745464e-05,
+      "loss": 1.2251,
+      "step": 58100
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.838677584755491e-05,
+      "loss": 1.5259,
+      "step": 58110
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.838562791765516e-05,
+      "loss": 1.4247,
+      "step": 58120
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.838447998775541e-05,
+      "loss": 1.3798,
+      "step": 58130
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.838333205785566e-05,
+      "loss": 1.5338,
+      "step": 58140
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.838218412795591e-05,
+      "loss": 1.4687,
+      "step": 58150
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.838103619805617e-05,
+      "loss": 1.4441,
+      "step": 58160
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.837988826815642e-05,
+      "loss": 1.5241,
+      "step": 58170
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.837874033825667e-05,
+      "loss": 1.5285,
+      "step": 58180
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.837759240835692e-05,
+      "loss": 1.5448,
+      "step": 58190
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.837644447845718e-05,
+      "loss": 1.5198,
+      "step": 58200
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.837529654855742e-05,
+      "loss": 1.579,
+      "step": 58210
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.837414861865768e-05,
+      "loss": 1.6192,
+      "step": 58220
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.837300068875793e-05,
+      "loss": 1.3736,
+      "step": 58230
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.837185275885818e-05,
+      "loss": 1.4333,
+      "step": 58240
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.837070482895843e-05,
+      "loss": 1.6623,
+      "step": 58250
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.83695568990587e-05,
+      "loss": 1.4949,
+      "step": 58260
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.836840896915895e-05,
+      "loss": 1.5,
+      "step": 58270
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.83672610392592e-05,
+      "loss": 1.6146,
+      "step": 58280
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.836611310935945e-05,
+      "loss": 1.418,
+      "step": 58290
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.83649651794597e-05,
+      "loss": 1.4379,
+      "step": 58300
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.836381724955996e-05,
+      "loss": 1.3635,
+      "step": 58310
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.836266931966021e-05,
+      "loss": 1.4683,
+      "step": 58320
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.836152138976046e-05,
+      "loss": 1.4804,
+      "step": 58330
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.836037345986071e-05,
+      "loss": 1.4764,
+      "step": 58340
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.835922552996097e-05,
+      "loss": 1.6682,
+      "step": 58350
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.835807760006121e-05,
+      "loss": 1.5088,
+      "step": 58360
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.835692967016147e-05,
+      "loss": 1.4077,
+      "step": 58370
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.835578174026172e-05,
+      "loss": 1.572,
+      "step": 58380
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.835463381036197e-05,
+      "loss": 1.5182,
+      "step": 58390
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.835348588046222e-05,
+      "loss": 1.347,
+      "step": 58400
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.835233795056249e-05,
+      "loss": 1.4105,
+      "step": 58410
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.835119002066273e-05,
+      "loss": 1.4282,
+      "step": 58420
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.835004209076298e-05,
+      "loss": 1.583,
+      "step": 58430
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.834889416086323e-05,
+      "loss": 1.4589,
+      "step": 58440
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.834774623096348e-05,
+      "loss": 1.4487,
+      "step": 58450
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.834659830106375e-05,
+      "loss": 1.654,
+      "step": 58460
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.8345450371164e-05,
+      "loss": 1.4241,
+      "step": 58470
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.834430244126425e-05,
+      "loss": 1.5382,
+      "step": 58480
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.83431545113645e-05,
+      "loss": 1.3224,
+      "step": 58490
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.834200658146476e-05,
+      "loss": 1.4228,
+      "step": 58500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.8340858651565e-05,
+      "loss": 1.5794,
+      "step": 58510
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.833971072166526e-05,
+      "loss": 1.3571,
+      "step": 58520
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.833856279176551e-05,
+      "loss": 1.4746,
+      "step": 58530
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.833741486186576e-05,
+      "loss": 1.3707,
+      "step": 58540
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.833626693196601e-05,
+      "loss": 1.4199,
+      "step": 58550
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.833511900206627e-05,
+      "loss": 1.2687,
+      "step": 58560
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.833397107216652e-05,
+      "loss": 1.4093,
+      "step": 58570
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.833282314226677e-05,
+      "loss": 1.4457,
+      "step": 58580
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.833167521236702e-05,
+      "loss": 1.4972,
+      "step": 58590
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.833052728246727e-05,
+      "loss": 1.5483,
+      "step": 58600
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.832937935256754e-05,
+      "loss": 1.4344,
+      "step": 58610
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.832823142266779e-05,
+      "loss": 1.3161,
+      "step": 58620
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.832708349276804e-05,
+      "loss": 1.4285,
+      "step": 58630
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.832593556286829e-05,
+      "loss": 1.4863,
+      "step": 58640
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.832478763296855e-05,
+      "loss": 1.3698,
+      "step": 58650
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.832363970306879e-05,
+      "loss": 1.4055,
+      "step": 58660
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.832249177316905e-05,
+      "loss": 1.3035,
+      "step": 58670
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.83213438432693e-05,
+      "loss": 1.6947,
+      "step": 58680
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.832019591336955e-05,
+      "loss": 1.4017,
+      "step": 58690
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.83190479834698e-05,
+      "loss": 1.4891,
+      "step": 58700
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.831790005357006e-05,
+      "loss": 1.3856,
+      "step": 58710
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.831675212367031e-05,
+      "loss": 1.3962,
+      "step": 58720
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.831560419377056e-05,
+      "loss": 1.4252,
+      "step": 58730
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.831445626387081e-05,
+      "loss": 1.2978,
+      "step": 58740
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.831330833397106e-05,
+      "loss": 1.5233,
+      "step": 58750
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.831216040407132e-05,
+      "loss": 1.3745,
+      "step": 58760
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.831101247417157e-05,
+      "loss": 1.3606,
+      "step": 58770
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.830986454427182e-05,
+      "loss": 1.5347,
+      "step": 58780
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.830871661437207e-05,
+      "loss": 1.3783,
+      "step": 58790
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.830756868447234e-05,
+      "loss": 1.3618,
+      "step": 58800
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.830642075457257e-05,
+      "loss": 1.4987,
+      "step": 58810
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.830527282467284e-05,
+      "loss": 1.6621,
+      "step": 58820
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.830412489477309e-05,
+      "loss": 1.4321,
+      "step": 58830
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.830297696487334e-05,
+      "loss": 1.4799,
+      "step": 58840
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.830182903497359e-05,
+      "loss": 1.4628,
+      "step": 58850
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.830068110507384e-05,
+      "loss": 1.4942,
+      "step": 58860
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.82995331751741e-05,
+      "loss": 1.3437,
+      "step": 58870
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.829838524527435e-05,
+      "loss": 1.4122,
+      "step": 58880
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.82972373153746e-05,
+      "loss": 1.3834,
+      "step": 58890
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.829608938547485e-05,
+      "loss": 1.3628,
+      "step": 58900
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.829494145557511e-05,
+      "loss": 1.3918,
+      "step": 58910
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.829379352567535e-05,
+      "loss": 1.4094,
+      "step": 58920
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.829264559577561e-05,
+      "loss": 1.6323,
+      "step": 58930
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.829149766587586e-05,
+      "loss": 1.5757,
+      "step": 58940
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.829034973597611e-05,
+      "loss": 1.332,
+      "step": 58950
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.828920180607636e-05,
+      "loss": 1.4247,
+      "step": 58960
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.828805387617663e-05,
+      "loss": 1.6441,
+      "step": 58970
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.828690594627688e-05,
+      "loss": 1.4154,
+      "step": 58980
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.828575801637713e-05,
+      "loss": 1.4452,
+      "step": 58990
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.828461008647738e-05,
+      "loss": 1.4025,
+      "step": 59000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.828346215657763e-05,
+      "loss": 1.3181,
+      "step": 59010
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.828231422667789e-05,
+      "loss": 1.4125,
+      "step": 59020
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.828116629677814e-05,
+      "loss": 1.4141,
+      "step": 59030
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.828001836687839e-05,
+      "loss": 1.5654,
+      "step": 59040
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.827887043697864e-05,
+      "loss": 1.412,
+      "step": 59050
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.82777225070789e-05,
+      "loss": 1.309,
+      "step": 59060
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.827657457717914e-05,
+      "loss": 1.3214,
+      "step": 59070
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.82754266472794e-05,
+      "loss": 1.3874,
+      "step": 59080
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.827427871737965e-05,
+      "loss": 1.59,
+      "step": 59090
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.82731307874799e-05,
+      "loss": 1.3935,
+      "step": 59100
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.827198285758015e-05,
+      "loss": 1.4589,
+      "step": 59110
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.827083492768041e-05,
+      "loss": 1.3829,
+      "step": 59120
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.826968699778066e-05,
+      "loss": 1.6643,
+      "step": 59130
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.826853906788091e-05,
+      "loss": 1.4109,
+      "step": 59140
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.826739113798116e-05,
+      "loss": 1.4389,
+      "step": 59150
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.826624320808141e-05,
+      "loss": 1.5855,
+      "step": 59160
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.826509527818168e-05,
+      "loss": 1.4336,
+      "step": 59170
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.826394734828193e-05,
+      "loss": 1.4602,
+      "step": 59180
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.826279941838218e-05,
+      "loss": 1.499,
+      "step": 59190
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.826165148848243e-05,
+      "loss": 1.5645,
+      "step": 59200
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.826050355858269e-05,
+      "loss": 1.4538,
+      "step": 59210
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.825935562868293e-05,
+      "loss": 1.3832,
+      "step": 59220
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.825820769878319e-05,
+      "loss": 1.4399,
+      "step": 59230
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.825705976888344e-05,
+      "loss": 1.2856,
+      "step": 59240
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.825591183898369e-05,
+      "loss": 1.6629,
+      "step": 59250
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.825476390908394e-05,
+      "loss": 1.405,
+      "step": 59260
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.82536159791842e-05,
+      "loss": 1.5819,
+      "step": 59270
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.825246804928445e-05,
+      "loss": 1.3057,
+      "step": 59280
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.82513201193847e-05,
+      "loss": 1.4984,
+      "step": 59290
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.825017218948495e-05,
+      "loss": 1.3525,
+      "step": 59300
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.82490242595852e-05,
+      "loss": 1.3047,
+      "step": 59310
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.824787632968547e-05,
+      "loss": 1.449,
+      "step": 59320
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.824672839978572e-05,
+      "loss": 1.5105,
+      "step": 59330
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.824558046988597e-05,
+      "loss": 1.5475,
+      "step": 59340
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.824443253998622e-05,
+      "loss": 1.4928,
+      "step": 59350
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.824328461008648e-05,
+      "loss": 1.4651,
+      "step": 59360
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.824213668018672e-05,
+      "loss": 1.4405,
+      "step": 59370
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.824098875028698e-05,
+      "loss": 1.3991,
+      "step": 59380
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.823984082038723e-05,
+      "loss": 1.432,
+      "step": 59390
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.823869289048748e-05,
+      "loss": 1.4363,
+      "step": 59400
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.823754496058773e-05,
+      "loss": 1.3601,
+      "step": 59410
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.823639703068799e-05,
+      "loss": 1.3125,
+      "step": 59420
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.823524910078824e-05,
+      "loss": 1.4323,
+      "step": 59430
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.823410117088849e-05,
+      "loss": 1.2458,
+      "step": 59440
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.823295324098874e-05,
+      "loss": 1.4757,
+      "step": 59450
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.823180531108899e-05,
+      "loss": 1.537,
+      "step": 59460
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.823065738118925e-05,
+      "loss": 1.5256,
+      "step": 59470
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.82295094512895e-05,
+      "loss": 1.5064,
+      "step": 59480
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.822836152138975e-05,
+      "loss": 1.4527,
+      "step": 59490
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.822721359149e-05,
+      "loss": 1.2502,
+      "step": 59500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.822606566159027e-05,
+      "loss": 1.4786,
+      "step": 59510
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.82249177316905e-05,
+      "loss": 1.5088,
+      "step": 59520
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.822376980179077e-05,
+      "loss": 1.4669,
+      "step": 59530
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.822262187189102e-05,
+      "loss": 1.4059,
+      "step": 59540
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.822147394199127e-05,
+      "loss": 1.3519,
+      "step": 59550
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.822032601209152e-05,
+      "loss": 1.4339,
+      "step": 59560
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.821917808219178e-05,
+      "loss": 1.3876,
+      "step": 59570
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.821803015229203e-05,
+      "loss": 1.4173,
+      "step": 59580
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.821688222239228e-05,
+      "loss": 1.5017,
+      "step": 59590
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.821573429249253e-05,
+      "loss": 1.3973,
+      "step": 59600
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.821458636259278e-05,
+      "loss": 1.4421,
+      "step": 59610
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.821343843269304e-05,
+      "loss": 1.4487,
+      "step": 59620
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.821229050279329e-05,
+      "loss": 1.5659,
+      "step": 59630
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.821114257289354e-05,
+      "loss": 1.3849,
+      "step": 59640
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.820999464299379e-05,
+      "loss": 1.5789,
+      "step": 59650
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.820884671309406e-05,
+      "loss": 1.4952,
+      "step": 59660
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.820769878319429e-05,
+      "loss": 1.6233,
+      "step": 59670
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.820655085329456e-05,
+      "loss": 1.6485,
+      "step": 59680
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.82054029233948e-05,
+      "loss": 1.2942,
+      "step": 59690
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.820425499349506e-05,
+      "loss": 1.372,
+      "step": 59700
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.82031070635953e-05,
+      "loss": 1.3807,
+      "step": 59710
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.820195913369557e-05,
+      "loss": 1.5534,
+      "step": 59720
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.820081120379582e-05,
+      "loss": 1.4115,
+      "step": 59730
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.819966327389607e-05,
+      "loss": 1.469,
+      "step": 59740
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.819851534399632e-05,
+      "loss": 1.4122,
+      "step": 59750
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.819736741409657e-05,
+      "loss": 1.4914,
+      "step": 59760
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.819621948419683e-05,
+      "loss": 1.3019,
+      "step": 59770
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.819507155429708e-05,
+      "loss": 1.4923,
+      "step": 59780
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.819392362439733e-05,
+      "loss": 1.4025,
+      "step": 59790
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.819277569449758e-05,
+      "loss": 1.4099,
+      "step": 59800
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.819162776459784e-05,
+      "loss": 1.514,
+      "step": 59810
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.819047983469808e-05,
+      "loss": 1.3212,
+      "step": 59820
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.818933190479834e-05,
+      "loss": 1.5389,
+      "step": 59830
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.81881839748986e-05,
+      "loss": 1.2965,
+      "step": 59840
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.818703604499884e-05,
+      "loss": 1.532,
+      "step": 59850
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.81858881150991e-05,
+      "loss": 1.5097,
+      "step": 59860
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.818474018519936e-05,
+      "loss": 1.531,
+      "step": 59870
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.81835922552996e-05,
+      "loss": 1.5899,
+      "step": 59880
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.818244432539986e-05,
+      "loss": 1.4216,
+      "step": 59890
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.81812963955001e-05,
+      "loss": 1.4922,
+      "step": 59900
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.818014846560036e-05,
+      "loss": 1.5454,
+      "step": 59910
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.817900053570062e-05,
+      "loss": 1.4694,
+      "step": 59920
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.817785260580087e-05,
+      "loss": 1.5313,
+      "step": 59930
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.817670467590112e-05,
+      "loss": 1.5858,
+      "step": 59940
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.817555674600137e-05,
+      "loss": 1.4357,
+      "step": 59950
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.817440881610163e-05,
+      "loss": 1.4881,
+      "step": 59960
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.817326088620187e-05,
+      "loss": 1.4319,
+      "step": 59970
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.817211295630213e-05,
+      "loss": 1.4258,
+      "step": 59980
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.817096502640238e-05,
+      "loss": 1.5459,
+      "step": 59990
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.816981709650263e-05,
+      "loss": 1.2968,
+      "step": 60000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.816866916660288e-05,
+      "loss": 1.5054,
+      "step": 60010
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.816752123670315e-05,
+      "loss": 1.533,
+      "step": 60020
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.81663733068034e-05,
+      "loss": 1.5193,
+      "step": 60030
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.816522537690364e-05,
+      "loss": 1.5291,
+      "step": 60040
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.81640774470039e-05,
+      "loss": 1.4061,
+      "step": 60050
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.816292951710414e-05,
+      "loss": 1.428,
+      "step": 60060
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.816178158720441e-05,
+      "loss": 1.3995,
+      "step": 60070
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.816063365730466e-05,
+      "loss": 1.2725,
+      "step": 60080
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.815948572740491e-05,
+      "loss": 1.6096,
+      "step": 60090
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.815833779750516e-05,
+      "loss": 1.5829,
+      "step": 60100
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.815718986760542e-05,
+      "loss": 1.2879,
+      "step": 60110
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.815604193770566e-05,
+      "loss": 1.5614,
+      "step": 60120
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.815489400780592e-05,
+      "loss": 1.5125,
+      "step": 60130
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.815374607790617e-05,
+      "loss": 1.4403,
+      "step": 60140
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.815259814800642e-05,
+      "loss": 1.3066,
+      "step": 60150
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.815145021810667e-05,
+      "loss": 1.3488,
+      "step": 60160
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.815030228820693e-05,
+      "loss": 1.5563,
+      "step": 60170
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.814915435830718e-05,
+      "loss": 1.539,
+      "step": 60180
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.814800642840743e-05,
+      "loss": 1.5087,
+      "step": 60190
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.814685849850768e-05,
+      "loss": 1.3924,
+      "step": 60200
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.814571056860793e-05,
+      "loss": 1.3713,
+      "step": 60210
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.81445626387082e-05,
+      "loss": 1.4259,
+      "step": 60220
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.814341470880843e-05,
+      "loss": 1.5546,
+      "step": 60230
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.81422667789087e-05,
+      "loss": 1.4535,
+      "step": 60240
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.814111884900895e-05,
+      "loss": 1.5956,
+      "step": 60250
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.81399709191092e-05,
+      "loss": 1.4518,
+      "step": 60260
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.813882298920945e-05,
+      "loss": 1.4181,
+      "step": 60270
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.813767505930971e-05,
+      "loss": 1.5117,
+      "step": 60280
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.813652712940996e-05,
+      "loss": 1.4318,
+      "step": 60290
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.813537919951021e-05,
+      "loss": 1.4679,
+      "step": 60300
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.813423126961046e-05,
+      "loss": 1.3563,
+      "step": 60310
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.813308333971071e-05,
+      "loss": 1.5551,
+      "step": 60320
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.813193540981097e-05,
+      "loss": 1.4778,
+      "step": 60330
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.813078747991122e-05,
+      "loss": 1.5653,
+      "step": 60340
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.812963955001147e-05,
+      "loss": 1.4246,
+      "step": 60350
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.812849162011172e-05,
+      "loss": 1.2246,
+      "step": 60360
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.812734369021199e-05,
+      "loss": 1.321,
+      "step": 60370
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.812619576031222e-05,
+      "loss": 1.6741,
+      "step": 60380
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.812504783041248e-05,
+      "loss": 1.412,
+      "step": 60390
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.812389990051273e-05,
+      "loss": 1.4544,
+      "step": 60400
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.812275197061298e-05,
+      "loss": 1.3144,
+      "step": 60410
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.812160404071323e-05,
+      "loss": 1.3543,
+      "step": 60420
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.81204561108135e-05,
+      "loss": 1.5267,
+      "step": 60430
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.811930818091375e-05,
+      "loss": 1.3764,
+      "step": 60440
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.8118160251014e-05,
+      "loss": 1.4026,
+      "step": 60450
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.811701232111425e-05,
+      "loss": 1.3822,
+      "step": 60460
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.81158643912145e-05,
+      "loss": 1.3358,
+      "step": 60470
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.811471646131476e-05,
+      "loss": 1.5269,
+      "step": 60480
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.811356853141501e-05,
+      "loss": 1.5238,
+      "step": 60490
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.811242060151526e-05,
+      "loss": 1.476,
+      "step": 60500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.811127267161551e-05,
+      "loss": 1.3656,
+      "step": 60510
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.811012474171577e-05,
+      "loss": 1.3842,
+      "step": 60520
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.810897681181601e-05,
+      "loss": 1.598,
+      "step": 60530
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.810782888191627e-05,
+      "loss": 1.3761,
+      "step": 60540
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.810668095201652e-05,
+      "loss": 1.3401,
+      "step": 60550
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.810553302211677e-05,
+      "loss": 1.5165,
+      "step": 60560
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.810438509221702e-05,
+      "loss": 1.4003,
+      "step": 60570
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.810323716231729e-05,
+      "loss": 1.4069,
+      "step": 60580
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.810208923241754e-05,
+      "loss": 1.3773,
+      "step": 60590
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.810094130251779e-05,
+      "loss": 1.4806,
+      "step": 60600
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.809979337261804e-05,
+      "loss": 1.3423,
+      "step": 60610
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.809864544271829e-05,
+      "loss": 1.3466,
+      "step": 60620
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.809749751281855e-05,
+      "loss": 1.3953,
+      "step": 60630
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.80963495829188e-05,
+      "loss": 1.5211,
+      "step": 60640
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.809520165301905e-05,
+      "loss": 1.5443,
+      "step": 60650
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.80940537231193e-05,
+      "loss": 1.6213,
+      "step": 60660
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.809290579321956e-05,
+      "loss": 1.5303,
+      "step": 60670
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.80917578633198e-05,
+      "loss": 1.4193,
+      "step": 60680
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.809060993342006e-05,
+      "loss": 1.4398,
+      "step": 60690
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.808946200352031e-05,
+      "loss": 1.5258,
+      "step": 60700
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.808831407362056e-05,
+      "loss": 1.3983,
+      "step": 60710
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.808716614372081e-05,
+      "loss": 1.3604,
+      "step": 60720
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.808601821382107e-05,
+      "loss": 1.4329,
+      "step": 60730
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.808487028392132e-05,
+      "loss": 1.4038,
+      "step": 60740
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.808372235402157e-05,
+      "loss": 1.5247,
+      "step": 60750
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.808257442412182e-05,
+      "loss": 1.4739,
+      "step": 60760
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.808142649422207e-05,
+      "loss": 1.5516,
+      "step": 60770
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.808027856432234e-05,
+      "loss": 1.3956,
+      "step": 60780
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.807913063442259e-05,
+      "loss": 1.5563,
+      "step": 60790
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.807798270452284e-05,
+      "loss": 1.4089,
+      "step": 60800
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.807683477462309e-05,
+      "loss": 1.4582,
+      "step": 60810
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.807568684472335e-05,
+      "loss": 1.552,
+      "step": 60820
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.807453891482359e-05,
+      "loss": 1.2697,
+      "step": 60830
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.807339098492385e-05,
+      "loss": 1.4652,
+      "step": 60840
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.80722430550241e-05,
+      "loss": 1.5803,
+      "step": 60850
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.807109512512435e-05,
+      "loss": 1.5108,
+      "step": 60860
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.80699471952246e-05,
+      "loss": 1.5002,
+      "step": 60870
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.806879926532486e-05,
+      "loss": 1.4038,
+      "step": 60880
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.806765133542511e-05,
+      "loss": 1.5799,
+      "step": 60890
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.806650340552536e-05,
+      "loss": 1.5568,
+      "step": 60900
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.806535547562561e-05,
+      "loss": 1.3705,
+      "step": 60910
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.806420754572586e-05,
+      "loss": 1.5848,
+      "step": 60920
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.806305961582613e-05,
+      "loss": 1.4349,
+      "step": 60930
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.806191168592638e-05,
+      "loss": 1.4464,
+      "step": 60940
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.806076375602663e-05,
+      "loss": 1.4045,
+      "step": 60950
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.805961582612688e-05,
+      "loss": 1.4408,
+      "step": 60960
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.805846789622714e-05,
+      "loss": 1.4663,
+      "step": 60970
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.805731996632738e-05,
+      "loss": 1.2301,
+      "step": 60980
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.805617203642764e-05,
+      "loss": 1.2631,
+      "step": 60990
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.805502410652789e-05,
+      "loss": 1.4968,
+      "step": 61000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.805387617662814e-05,
+      "loss": 1.3969,
+      "step": 61010
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.805272824672839e-05,
+      "loss": 1.3492,
+      "step": 61020
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.805158031682865e-05,
+      "loss": 1.4268,
+      "step": 61030
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.80504323869289e-05,
+      "loss": 1.5474,
+      "step": 61040
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.804928445702915e-05,
+      "loss": 1.4744,
+      "step": 61050
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.80481365271294e-05,
+      "loss": 1.4095,
+      "step": 61060
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.804698859722965e-05,
+      "loss": 1.5265,
+      "step": 61070
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.804584066732991e-05,
+      "loss": 1.4447,
+      "step": 61080
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.804469273743016e-05,
+      "loss": 1.5364,
+      "step": 61090
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.804354480753041e-05,
+      "loss": 1.3168,
+      "step": 61100
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.804239687763066e-05,
+      "loss": 1.449,
+      "step": 61110
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.804124894773093e-05,
+      "loss": 1.469,
+      "step": 61120
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.804010101783116e-05,
+      "loss": 1.6179,
+      "step": 61130
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.803895308793143e-05,
+      "loss": 1.393,
+      "step": 61140
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.803780515803168e-05,
+      "loss": 1.3599,
+      "step": 61150
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.803665722813193e-05,
+      "loss": 1.5257,
+      "step": 61160
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.803550929823218e-05,
+      "loss": 1.3909,
+      "step": 61170
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.803436136833244e-05,
+      "loss": 1.2719,
+      "step": 61180
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.803321343843269e-05,
+      "loss": 1.3564,
+      "step": 61190
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.803206550853294e-05,
+      "loss": 1.3667,
+      "step": 61200
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.803091757863319e-05,
+      "loss": 1.3938,
+      "step": 61210
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.802976964873344e-05,
+      "loss": 1.4572,
+      "step": 61220
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.80286217188337e-05,
+      "loss": 1.318,
+      "step": 61230
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.802747378893395e-05,
+      "loss": 1.3688,
+      "step": 61240
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.80263258590342e-05,
+      "loss": 1.4985,
+      "step": 61250
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.802517792913445e-05,
+      "loss": 1.5991,
+      "step": 61260
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.802402999923472e-05,
+      "loss": 1.4545,
+      "step": 61270
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.802288206933495e-05,
+      "loss": 1.3904,
+      "step": 61280
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.802173413943522e-05,
+      "loss": 1.5221,
+      "step": 61290
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.802058620953547e-05,
+      "loss": 1.3882,
+      "step": 61300
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.801943827963572e-05,
+      "loss": 1.4174,
+      "step": 61310
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.801829034973597e-05,
+      "loss": 1.5055,
+      "step": 61320
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.801714241983623e-05,
+      "loss": 1.5506,
+      "step": 61330
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.801599448993648e-05,
+      "loss": 1.5145,
+      "step": 61340
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.801484656003673e-05,
+      "loss": 1.3565,
+      "step": 61350
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.801369863013698e-05,
+      "loss": 1.3653,
+      "step": 61360
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.801255070023723e-05,
+      "loss": 1.1724,
+      "step": 61370
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.801140277033749e-05,
+      "loss": 1.532,
+      "step": 61380
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.801025484043774e-05,
+      "loss": 1.4464,
+      "step": 61390
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.800910691053799e-05,
+      "loss": 1.4243,
+      "step": 61400
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.800795898063824e-05,
+      "loss": 1.443,
+      "step": 61410
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.80068110507385e-05,
+      "loss": 1.4794,
+      "step": 61420
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.800566312083874e-05,
+      "loss": 1.5305,
+      "step": 61430
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.8004515190939e-05,
+      "loss": 1.3895,
+      "step": 61440
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.800336726103925e-05,
+      "loss": 1.5007,
+      "step": 61450
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.80022193311395e-05,
+      "loss": 1.3514,
+      "step": 61460
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.800107140123975e-05,
+      "loss": 1.413,
+      "step": 61470
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.799992347134002e-05,
+      "loss": 1.4593,
+      "step": 61480
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.799877554144027e-05,
+      "loss": 1.3789,
+      "step": 61490
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.799762761154052e-05,
+      "loss": 1.4773,
+      "step": 61500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.799647968164077e-05,
+      "loss": 1.6913,
+      "step": 61510
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.799533175174102e-05,
+      "loss": 1.4337,
+      "step": 61520
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.799418382184128e-05,
+      "loss": 1.5871,
+      "step": 61530
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.799303589194153e-05,
+      "loss": 1.3818,
+      "step": 61540
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.799188796204178e-05,
+      "loss": 1.3203,
+      "step": 61550
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.799074003214203e-05,
+      "loss": 1.4006,
+      "step": 61560
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.798959210224229e-05,
+      "loss": 1.383,
+      "step": 61570
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.798844417234253e-05,
+      "loss": 1.5712,
+      "step": 61580
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.798729624244279e-05,
+      "loss": 1.426,
+      "step": 61590
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.798614831254304e-05,
+      "loss": 1.3929,
+      "step": 61600
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.798500038264329e-05,
+      "loss": 1.4209,
+      "step": 61610
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.798385245274354e-05,
+      "loss": 1.4925,
+      "step": 61620
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.798270452284379e-05,
+      "loss": 1.3731,
+      "step": 61630
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.798155659294406e-05,
+      "loss": 1.3694,
+      "step": 61640
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.79804086630443e-05,
+      "loss": 1.3963,
+      "step": 61650
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.797926073314456e-05,
+      "loss": 1.4573,
+      "step": 61660
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.79781128032448e-05,
+      "loss": 1.5595,
+      "step": 61670
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.797696487334507e-05,
+      "loss": 1.3662,
+      "step": 61680
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.79758169434453e-05,
+      "loss": 1.451,
+      "step": 61690
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.797466901354557e-05,
+      "loss": 1.3233,
+      "step": 61700
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.797352108364582e-05,
+      "loss": 1.5494,
+      "step": 61710
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.797237315374607e-05,
+      "loss": 1.337,
+      "step": 61720
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.797122522384632e-05,
+      "loss": 1.3523,
+      "step": 61730
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.797007729394658e-05,
+      "loss": 1.261,
+      "step": 61740
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.796892936404683e-05,
+      "loss": 1.4444,
+      "step": 61750
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.796778143414708e-05,
+      "loss": 1.4632,
+      "step": 61760
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.796663350424733e-05,
+      "loss": 1.5246,
+      "step": 61770
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.796548557434758e-05,
+      "loss": 1.501,
+      "step": 61780
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.796433764444784e-05,
+      "loss": 1.5783,
+      "step": 61790
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.79631897145481e-05,
+      "loss": 1.3874,
+      "step": 61800
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.796204178464834e-05,
+      "loss": 1.5563,
+      "step": 61810
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.79608938547486e-05,
+      "loss": 1.4503,
+      "step": 61820
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.795974592484886e-05,
+      "loss": 1.4917,
+      "step": 61830
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.795859799494909e-05,
+      "loss": 1.3984,
+      "step": 61840
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.795745006504936e-05,
+      "loss": 1.4066,
+      "step": 61850
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.79563021351496e-05,
+      "loss": 1.3597,
+      "step": 61860
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.795515420524986e-05,
+      "loss": 1.3355,
+      "step": 61870
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.79540062753501e-05,
+      "loss": 1.2981,
+      "step": 61880
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.795285834545037e-05,
+      "loss": 1.3257,
+      "step": 61890
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.795171041555062e-05,
+      "loss": 1.4691,
+      "step": 61900
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.795056248565087e-05,
+      "loss": 1.636,
+      "step": 61910
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.794941455575112e-05,
+      "loss": 1.5301,
+      "step": 61920
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.794826662585137e-05,
+      "loss": 1.4165,
+      "step": 61930
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.794711869595163e-05,
+      "loss": 1.5184,
+      "step": 61940
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.794597076605188e-05,
+      "loss": 1.4404,
+      "step": 61950
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.794482283615213e-05,
+      "loss": 1.3229,
+      "step": 61960
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.794367490625238e-05,
+      "loss": 1.4036,
+      "step": 61970
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.794252697635265e-05,
+      "loss": 1.4295,
+      "step": 61980
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.794137904645288e-05,
+      "loss": 1.4934,
+      "step": 61990
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.794023111655314e-05,
+      "loss": 1.3351,
+      "step": 62000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.79390831866534e-05,
+      "loss": 1.4285,
+      "step": 62010
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.793793525675364e-05,
+      "loss": 1.499,
+      "step": 62020
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.79367873268539e-05,
+      "loss": 1.525,
+      "step": 62030
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.793563939695416e-05,
+      "loss": 1.5094,
+      "step": 62040
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.793449146705441e-05,
+      "loss": 1.3634,
+      "step": 62050
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.793334353715466e-05,
+      "loss": 1.4939,
+      "step": 62060
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.793219560725491e-05,
+      "loss": 1.4976,
+      "step": 62070
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.793104767735516e-05,
+      "loss": 1.4315,
+      "step": 62080
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.792989974745542e-05,
+      "loss": 1.4801,
+      "step": 62090
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.792875181755567e-05,
+      "loss": 1.4104,
+      "step": 62100
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.792760388765592e-05,
+      "loss": 1.3727,
+      "step": 62110
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.792645595775617e-05,
+      "loss": 1.2053,
+      "step": 62120
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.792530802785643e-05,
+      "loss": 1.6884,
+      "step": 62130
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.792416009795667e-05,
+      "loss": 1.3636,
+      "step": 62140
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.792301216805693e-05,
+      "loss": 1.4481,
+      "step": 62150
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.792186423815718e-05,
+      "loss": 1.6235,
+      "step": 62160
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.792071630825743e-05,
+      "loss": 1.3481,
+      "step": 62170
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.791956837835768e-05,
+      "loss": 1.2794,
+      "step": 62180
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.791842044845795e-05,
+      "loss": 1.2339,
+      "step": 62190
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.79172725185582e-05,
+      "loss": 1.4573,
+      "step": 62200
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.791612458865845e-05,
+      "loss": 1.4476,
+      "step": 62210
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.79149766587587e-05,
+      "loss": 1.6146,
+      "step": 62220
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.791382872885895e-05,
+      "loss": 1.562,
+      "step": 62230
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.791268079895921e-05,
+      "loss": 1.5576,
+      "step": 62240
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.791153286905946e-05,
+      "loss": 1.486,
+      "step": 62250
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.791038493915971e-05,
+      "loss": 1.3259,
+      "step": 62260
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.790923700925996e-05,
+      "loss": 1.5927,
+      "step": 62270
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.790808907936022e-05,
+      "loss": 1.2729,
+      "step": 62280
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.790694114946046e-05,
+      "loss": 1.332,
+      "step": 62290
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.790579321956072e-05,
+      "loss": 1.3454,
+      "step": 62300
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.790464528966097e-05,
+      "loss": 1.408,
+      "step": 62310
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.790349735976122e-05,
+      "loss": 1.4228,
+      "step": 62320
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.790234942986147e-05,
+      "loss": 1.5794,
+      "step": 62330
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.790120149996173e-05,
+      "loss": 1.3443,
+      "step": 62340
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.790005357006198e-05,
+      "loss": 1.4003,
+      "step": 62350
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.789890564016223e-05,
+      "loss": 1.4592,
+      "step": 62360
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.789775771026248e-05,
+      "loss": 1.6743,
+      "step": 62370
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.789660978036273e-05,
+      "loss": 1.5165,
+      "step": 62380
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.7895461850463e-05,
+      "loss": 1.5828,
+      "step": 62390
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.789431392056325e-05,
+      "loss": 1.5069,
+      "step": 62400
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.78931659906635e-05,
+      "loss": 1.5099,
+      "step": 62410
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.789201806076375e-05,
+      "loss": 1.4833,
+      "step": 62420
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.789087013086401e-05,
+      "loss": 1.3681,
+      "step": 62430
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.788972220096425e-05,
+      "loss": 1.2828,
+      "step": 62440
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.788857427106451e-05,
+      "loss": 1.3767,
+      "step": 62450
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.788742634116476e-05,
+      "loss": 1.4859,
+      "step": 62460
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.788627841126501e-05,
+      "loss": 1.5077,
+      "step": 62470
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.788513048136526e-05,
+      "loss": 1.312,
+      "step": 62480
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.788398255146552e-05,
+      "loss": 1.3807,
+      "step": 62490
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.788283462156577e-05,
+      "loss": 1.4107,
+      "step": 62500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.788168669166602e-05,
+      "loss": 1.3875,
+      "step": 62510
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.788053876176627e-05,
+      "loss": 1.5139,
+      "step": 62520
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.787939083186652e-05,
+      "loss": 1.4408,
+      "step": 62530
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.787824290196679e-05,
+      "loss": 1.3861,
+      "step": 62540
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.787709497206704e-05,
+      "loss": 1.5434,
+      "step": 62550
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.787594704216729e-05,
+      "loss": 1.412,
+      "step": 62560
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.787479911226754e-05,
+      "loss": 1.4614,
+      "step": 62570
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.78736511823678e-05,
+      "loss": 1.4387,
+      "step": 62580
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.787250325246804e-05,
+      "loss": 1.3413,
+      "step": 62590
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.78713553225683e-05,
+      "loss": 1.3599,
+      "step": 62600
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.787020739266855e-05,
+      "loss": 1.4026,
+      "step": 62610
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.78690594627688e-05,
+      "loss": 1.5299,
+      "step": 62620
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.786791153286905e-05,
+      "loss": 1.3854,
+      "step": 62630
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.786676360296931e-05,
+      "loss": 1.3961,
+      "step": 62640
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.786561567306956e-05,
+      "loss": 1.3798,
+      "step": 62650
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.786446774316981e-05,
+      "loss": 1.4709,
+      "step": 62660
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.786331981327006e-05,
+      "loss": 1.4906,
+      "step": 62670
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.786217188337031e-05,
+      "loss": 1.4127,
+      "step": 62680
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.786102395347057e-05,
+      "loss": 1.509,
+      "step": 62690
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.785987602357082e-05,
+      "loss": 1.3679,
+      "step": 62700
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.785872809367107e-05,
+      "loss": 1.3572,
+      "step": 62710
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.785758016377132e-05,
+      "loss": 1.4711,
+      "step": 62720
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.785643223387159e-05,
+      "loss": 1.4842,
+      "step": 62730
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.785528430397182e-05,
+      "loss": 1.4564,
+      "step": 62740
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.785413637407209e-05,
+      "loss": 1.3384,
+      "step": 62750
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.785298844417234e-05,
+      "loss": 1.5248,
+      "step": 62760
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.785184051427259e-05,
+      "loss": 1.4732,
+      "step": 62770
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.785069258437284e-05,
+      "loss": 1.4145,
+      "step": 62780
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.78495446544731e-05,
+      "loss": 1.3991,
+      "step": 62790
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.784839672457335e-05,
+      "loss": 1.5015,
+      "step": 62800
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.78472487946736e-05,
+      "loss": 1.4418,
+      "step": 62810
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.784610086477385e-05,
+      "loss": 1.4449,
+      "step": 62820
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.78449529348741e-05,
+      "loss": 1.3096,
+      "step": 62830
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.784380500497436e-05,
+      "loss": 1.5894,
+      "step": 62840
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.784265707507461e-05,
+      "loss": 1.3029,
+      "step": 62850
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.784150914517486e-05,
+      "loss": 1.4207,
+      "step": 62860
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.784036121527511e-05,
+      "loss": 1.5372,
+      "step": 62870
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.783921328537538e-05,
+      "loss": 1.3891,
+      "step": 62880
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.783806535547561e-05,
+      "loss": 1.3837,
+      "step": 62890
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.783691742557588e-05,
+      "loss": 1.5257,
+      "step": 62900
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.783576949567613e-05,
+      "loss": 1.3213,
+      "step": 62910
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.783462156577638e-05,
+      "loss": 1.7529,
+      "step": 62920
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.783347363587663e-05,
+      "loss": 1.603,
+      "step": 62930
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.783232570597689e-05,
+      "loss": 1.507,
+      "step": 62940
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.783117777607714e-05,
+      "loss": 1.3977,
+      "step": 62950
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.783002984617739e-05,
+      "loss": 1.5143,
+      "step": 62960
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.782888191627764e-05,
+      "loss": 1.3556,
+      "step": 62970
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.782773398637789e-05,
+      "loss": 1.5128,
+      "step": 62980
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.782658605647815e-05,
+      "loss": 1.3908,
+      "step": 62990
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.782543812657839e-05,
+      "loss": 1.3054,
+      "step": 63000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.782429019667865e-05,
+      "loss": 1.354,
+      "step": 63010
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.78231422667789e-05,
+      "loss": 1.3882,
+      "step": 63020
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.782199433687915e-05,
+      "loss": 1.3836,
+      "step": 63030
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.78208464069794e-05,
+      "loss": 1.5646,
+      "step": 63040
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.781969847707966e-05,
+      "loss": 1.4661,
+      "step": 63050
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.781855054717991e-05,
+      "loss": 1.3207,
+      "step": 63060
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.781740261728016e-05,
+      "loss": 1.5365,
+      "step": 63070
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.781625468738041e-05,
+      "loss": 1.3685,
+      "step": 63080
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.781510675748066e-05,
+      "loss": 1.3906,
+      "step": 63090
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.781395882758093e-05,
+      "loss": 1.4417,
+      "step": 63100
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.781281089768118e-05,
+      "loss": 1.4003,
+      "step": 63110
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.781166296778143e-05,
+      "loss": 1.4218,
+      "step": 63120
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.781051503788168e-05,
+      "loss": 1.5425,
+      "step": 63130
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.780936710798194e-05,
+      "loss": 1.3796,
+      "step": 63140
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.780821917808218e-05,
+      "loss": 1.4081,
+      "step": 63150
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.780707124818244e-05,
+      "loss": 1.3882,
+      "step": 63160
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.780592331828269e-05,
+      "loss": 1.5408,
+      "step": 63170
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.780477538838294e-05,
+      "loss": 1.3917,
+      "step": 63180
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.780362745848319e-05,
+      "loss": 1.4087,
+      "step": 63190
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.780247952858345e-05,
+      "loss": 1.4059,
+      "step": 63200
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.78013315986837e-05,
+      "loss": 1.3585,
+      "step": 63210
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.780018366878395e-05,
+      "loss": 1.4718,
+      "step": 63220
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.77990357388842e-05,
+      "loss": 1.381,
+      "step": 63230
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.779788780898445e-05,
+      "loss": 1.4534,
+      "step": 63240
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.779673987908472e-05,
+      "loss": 1.4951,
+      "step": 63250
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.779559194918497e-05,
+      "loss": 1.3635,
+      "step": 63260
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.779444401928522e-05,
+      "loss": 1.3923,
+      "step": 63270
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.779329608938547e-05,
+      "loss": 1.6553,
+      "step": 63280
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.779214815948573e-05,
+      "loss": 1.3273,
+      "step": 63290
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.779100022958596e-05,
+      "loss": 1.489,
+      "step": 63300
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.778985229968623e-05,
+      "loss": 1.4686,
+      "step": 63310
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.778870436978648e-05,
+      "loss": 1.313,
+      "step": 63320
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.778755643988673e-05,
+      "loss": 1.2914,
+      "step": 63330
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.778640850998698e-05,
+      "loss": 1.4919,
+      "step": 63340
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.778526058008724e-05,
+      "loss": 1.2602,
+      "step": 63350
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.778411265018749e-05,
+      "loss": 1.3868,
+      "step": 63360
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.778296472028774e-05,
+      "loss": 1.4679,
+      "step": 63370
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.778181679038799e-05,
+      "loss": 1.4283,
+      "step": 63380
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.778066886048824e-05,
+      "loss": 1.416,
+      "step": 63390
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.77795209305885e-05,
+      "loss": 1.4757,
+      "step": 63400
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.777837300068875e-05,
+      "loss": 1.5222,
+      "step": 63410
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.7777225070789e-05,
+      "loss": 1.4852,
+      "step": 63420
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.777607714088925e-05,
+      "loss": 1.5116,
+      "step": 63430
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.777492921098952e-05,
+      "loss": 1.4397,
+      "step": 63440
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.777378128108975e-05,
+      "loss": 1.3825,
+      "step": 63450
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.777263335119002e-05,
+      "loss": 1.4313,
+      "step": 63460
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.777148542129027e-05,
+      "loss": 1.5257,
+      "step": 63470
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.777033749139052e-05,
+      "loss": 1.6325,
+      "step": 63480
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.776918956149077e-05,
+      "loss": 1.4413,
+      "step": 63490
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.776804163159103e-05,
+      "loss": 1.5191,
+      "step": 63500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.776689370169128e-05,
+      "loss": 1.3176,
+      "step": 63510
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.776574577179153e-05,
+      "loss": 1.5515,
+      "step": 63520
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.776459784189178e-05,
+      "loss": 1.3796,
+      "step": 63530
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.776344991199203e-05,
+      "loss": 1.6978,
+      "step": 63540
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.776230198209229e-05,
+      "loss": 1.4638,
+      "step": 63550
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.776115405219254e-05,
+      "loss": 1.4626,
+      "step": 63560
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.776000612229279e-05,
+      "loss": 1.6532,
+      "step": 63570
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.775885819239304e-05,
+      "loss": 1.4244,
+      "step": 63580
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.77577102624933e-05,
+      "loss": 1.3031,
+      "step": 63590
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.775656233259354e-05,
+      "loss": 1.4953,
+      "step": 63600
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.77554144026938e-05,
+      "loss": 1.441,
+      "step": 63610
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.775426647279406e-05,
+      "loss": 1.3945,
+      "step": 63620
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.77531185428943e-05,
+      "loss": 1.6565,
+      "step": 63630
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.775197061299455e-05,
+      "loss": 1.6469,
+      "step": 63640
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.775082268309482e-05,
+      "loss": 1.177,
+      "step": 63650
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.774967475319507e-05,
+      "loss": 1.4304,
+      "step": 63660
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.774852682329532e-05,
+      "loss": 1.4031,
+      "step": 63670
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.774737889339557e-05,
+      "loss": 1.5218,
+      "step": 63680
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.774623096349582e-05,
+      "loss": 1.48,
+      "step": 63690
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.774508303359608e-05,
+      "loss": 1.4539,
+      "step": 63700
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.774393510369633e-05,
+      "loss": 1.4957,
+      "step": 63710
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.774278717379658e-05,
+      "loss": 1.5234,
+      "step": 63720
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.774163924389683e-05,
+      "loss": 1.4668,
+      "step": 63730
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.77404913139971e-05,
+      "loss": 1.534,
+      "step": 63740
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.773934338409733e-05,
+      "loss": 1.6449,
+      "step": 63750
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.77381954541976e-05,
+      "loss": 1.2778,
+      "step": 63760
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.773704752429784e-05,
+      "loss": 1.5589,
+      "step": 63770
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.77358995943981e-05,
+      "loss": 1.4031,
+      "step": 63780
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.773475166449834e-05,
+      "loss": 1.4402,
+      "step": 63790
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.77336037345986e-05,
+      "loss": 1.3591,
+      "step": 63800
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.773245580469886e-05,
+      "loss": 1.3101,
+      "step": 63810
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.77313078747991e-05,
+      "loss": 1.3227,
+      "step": 63820
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.773015994489936e-05,
+      "loss": 1.4969,
+      "step": 63830
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.77290120149996e-05,
+      "loss": 1.4054,
+      "step": 63840
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.772786408509987e-05,
+      "loss": 1.4248,
+      "step": 63850
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.772671615520012e-05,
+      "loss": 1.5013,
+      "step": 63860
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.772556822530037e-05,
+      "loss": 1.3535,
+      "step": 63870
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.772442029540062e-05,
+      "loss": 1.5653,
+      "step": 63880
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.772327236550088e-05,
+      "loss": 1.4915,
+      "step": 63890
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.772212443560112e-05,
+      "loss": 1.3215,
+      "step": 63900
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.772097650570138e-05,
+      "loss": 1.5375,
+      "step": 63910
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.771982857580163e-05,
+      "loss": 1.4758,
+      "step": 63920
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.771868064590188e-05,
+      "loss": 1.5698,
+      "step": 63930
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.771753271600213e-05,
+      "loss": 1.2828,
+      "step": 63940
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.77163847861024e-05,
+      "loss": 1.5762,
+      "step": 63950
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.771523685620264e-05,
+      "loss": 1.2642,
+      "step": 63960
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.77140889263029e-05,
+      "loss": 1.3987,
+      "step": 63970
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.771294099640314e-05,
+      "loss": 1.4021,
+      "step": 63980
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.77117930665034e-05,
+      "loss": 1.3779,
+      "step": 63990
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.771064513660366e-05,
+      "loss": 1.4871,
+      "step": 64000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.770949720670391e-05,
+      "loss": 1.4302,
+      "step": 64010
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.770834927680416e-05,
+      "loss": 1.322,
+      "step": 64020
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.770720134690441e-05,
+      "loss": 1.4227,
+      "step": 64030
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.770605341700467e-05,
+      "loss": 1.492,
+      "step": 64040
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.770490548710491e-05,
+      "loss": 1.477,
+      "step": 64050
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.770375755720517e-05,
+      "loss": 1.3124,
+      "step": 64060
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.770260962730542e-05,
+      "loss": 1.4302,
+      "step": 64070
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.770146169740567e-05,
+      "loss": 1.4534,
+      "step": 64080
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.770031376750592e-05,
+      "loss": 1.3068,
+      "step": 64090
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.769916583760618e-05,
+      "loss": 1.4318,
+      "step": 64100
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.769801790770643e-05,
+      "loss": 1.5164,
+      "step": 64110
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.769686997780668e-05,
+      "loss": 1.3837,
+      "step": 64120
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.769572204790693e-05,
+      "loss": 1.5531,
+      "step": 64130
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.769457411800718e-05,
+      "loss": 1.4969,
+      "step": 64140
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.769342618810745e-05,
+      "loss": 1.4053,
+      "step": 64150
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.76922782582077e-05,
+      "loss": 1.3875,
+      "step": 64160
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.769113032830795e-05,
+      "loss": 1.392,
+      "step": 64170
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.76899823984082e-05,
+      "loss": 1.3919,
+      "step": 64180
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.768883446850846e-05,
+      "loss": 1.3377,
+      "step": 64190
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.76876865386087e-05,
+      "loss": 1.4664,
+      "step": 64200
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.768653860870896e-05,
+      "loss": 1.3853,
+      "step": 64210
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.768539067880921e-05,
+      "loss": 1.3315,
+      "step": 64220
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.768424274890946e-05,
+      "loss": 1.4163,
+      "step": 64230
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.768309481900971e-05,
+      "loss": 1.3498,
+      "step": 64240
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.768194688910997e-05,
+      "loss": 1.4457,
+      "step": 64250
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.768079895921022e-05,
+      "loss": 1.461,
+      "step": 64260
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.767965102931047e-05,
+      "loss": 1.4407,
+      "step": 64270
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.767850309941072e-05,
+      "loss": 1.3628,
+      "step": 64280
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.767735516951097e-05,
+      "loss": 1.2975,
+      "step": 64290
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.767620723961123e-05,
+      "loss": 1.4453,
+      "step": 64300
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.767505930971148e-05,
+      "loss": 1.4297,
+      "step": 64310
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.767391137981173e-05,
+      "loss": 1.3389,
+      "step": 64320
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.767276344991198e-05,
+      "loss": 1.3494,
+      "step": 64330
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.767161552001223e-05,
+      "loss": 1.245,
+      "step": 64340
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.767046759011248e-05,
+      "loss": 1.458,
+      "step": 64350
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.766931966021275e-05,
+      "loss": 1.4048,
+      "step": 64360
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.7668171730313e-05,
+      "loss": 1.4132,
+      "step": 64370
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.766702380041325e-05,
+      "loss": 1.5352,
+      "step": 64380
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.76658758705135e-05,
+      "loss": 1.5697,
+      "step": 64390
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.766472794061375e-05,
+      "loss": 1.2475,
+      "step": 64400
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.766358001071401e-05,
+      "loss": 1.3333,
+      "step": 64410
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.766243208081426e-05,
+      "loss": 1.316,
+      "step": 64420
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.766128415091451e-05,
+      "loss": 1.4794,
+      "step": 64430
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.766013622101476e-05,
+      "loss": 1.4415,
+      "step": 64440
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.765898829111502e-05,
+      "loss": 1.3488,
+      "step": 64450
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.765784036121526e-05,
+      "loss": 1.3609,
+      "step": 64460
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.765669243131552e-05,
+      "loss": 1.4941,
+      "step": 64470
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.765554450141577e-05,
+      "loss": 1.3808,
+      "step": 64480
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.765439657151602e-05,
+      "loss": 1.3773,
+      "step": 64490
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.765324864161627e-05,
+      "loss": 1.4978,
+      "step": 64500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.765210071171654e-05,
+      "loss": 1.3889,
+      "step": 64510
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.765095278181679e-05,
+      "loss": 1.2549,
+      "step": 64520
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.764980485191704e-05,
+      "loss": 1.5057,
+      "step": 64530
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.764865692201729e-05,
+      "loss": 1.4004,
+      "step": 64540
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.764750899211754e-05,
+      "loss": 1.3751,
+      "step": 64550
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.76463610622178e-05,
+      "loss": 1.5222,
+      "step": 64560
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.764521313231805e-05,
+      "loss": 1.4891,
+      "step": 64570
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.76440652024183e-05,
+      "loss": 1.4311,
+      "step": 64580
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.764291727251855e-05,
+      "loss": 1.5561,
+      "step": 64590
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.764176934261881e-05,
+      "loss": 1.3723,
+      "step": 64600
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.764062141271905e-05,
+      "loss": 1.5061,
+      "step": 64610
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.763947348281931e-05,
+      "loss": 1.3593,
+      "step": 64620
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.763832555291956e-05,
+      "loss": 1.4471,
+      "step": 64630
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.763717762301981e-05,
+      "loss": 1.301,
+      "step": 64640
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.763602969312006e-05,
+      "loss": 1.3655,
+      "step": 64650
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.763488176322032e-05,
+      "loss": 1.4252,
+      "step": 64660
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.763373383332057e-05,
+      "loss": 1.4345,
+      "step": 64670
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.763258590342082e-05,
+      "loss": 1.3668,
+      "step": 64680
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.763143797352107e-05,
+      "loss": 1.324,
+      "step": 64690
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.763029004362132e-05,
+      "loss": 1.5148,
+      "step": 64700
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.762914211372159e-05,
+      "loss": 1.6436,
+      "step": 64710
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.762799418382184e-05,
+      "loss": 1.2779,
+      "step": 64720
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.762684625392209e-05,
+      "loss": 1.3143,
+      "step": 64730
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.762569832402234e-05,
+      "loss": 1.4739,
+      "step": 64740
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.76245503941226e-05,
+      "loss": 1.2766,
+      "step": 64750
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.762340246422284e-05,
+      "loss": 1.3103,
+      "step": 64760
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.76222545343231e-05,
+      "loss": 1.4613,
+      "step": 64770
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.762110660442335e-05,
+      "loss": 1.4824,
+      "step": 64780
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.76199586745236e-05,
+      "loss": 1.3297,
+      "step": 64790
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.761881074462385e-05,
+      "loss": 1.403,
+      "step": 64800
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.761766281472411e-05,
+      "loss": 1.3571,
+      "step": 64810
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.761651488482436e-05,
+      "loss": 1.4149,
+      "step": 64820
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.761536695492461e-05,
+      "loss": 1.6229,
+      "step": 64830
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.761421902502486e-05,
+      "loss": 1.3761,
+      "step": 64840
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.761307109512511e-05,
+      "loss": 1.5387,
+      "step": 64850
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.761192316522538e-05,
+      "loss": 1.5805,
+      "step": 64860
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.761077523532563e-05,
+      "loss": 1.2236,
+      "step": 64870
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.760962730542588e-05,
+      "loss": 1.3195,
+      "step": 64880
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.760847937552613e-05,
+      "loss": 1.3505,
+      "step": 64890
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.760733144562639e-05,
+      "loss": 1.3907,
+      "step": 64900
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.760618351572663e-05,
+      "loss": 1.5276,
+      "step": 64910
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.760503558582689e-05,
+      "loss": 1.3415,
+      "step": 64920
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.760388765592714e-05,
+      "loss": 1.312,
+      "step": 64930
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.760273972602739e-05,
+      "loss": 1.3941,
+      "step": 64940
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.760159179612764e-05,
+      "loss": 1.5314,
+      "step": 64950
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.76004438662279e-05,
+      "loss": 1.4187,
+      "step": 64960
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.759929593632815e-05,
+      "loss": 1.351,
+      "step": 64970
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.75981480064284e-05,
+      "loss": 1.3413,
+      "step": 64980
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.759700007652865e-05,
+      "loss": 1.4351,
+      "step": 64990
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.75958521466289e-05,
+      "loss": 1.628,
+      "step": 65000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.759470421672916e-05,
+      "loss": 1.6029,
+      "step": 65010
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.759355628682941e-05,
+      "loss": 1.329,
+      "step": 65020
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.759240835692966e-05,
+      "loss": 1.5173,
+      "step": 65030
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.759126042702991e-05,
+      "loss": 1.3655,
+      "step": 65040
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.759011249713018e-05,
+      "loss": 1.3573,
+      "step": 65050
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.758896456723041e-05,
+      "loss": 1.4631,
+      "step": 65060
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.758781663733068e-05,
+      "loss": 1.4352,
+      "step": 65070
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.758666870743093e-05,
+      "loss": 1.3972,
+      "step": 65080
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.758552077753118e-05,
+      "loss": 1.4378,
+      "step": 65090
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.758437284763143e-05,
+      "loss": 1.28,
+      "step": 65100
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.758322491773169e-05,
+      "loss": 1.45,
+      "step": 65110
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.758207698783194e-05,
+      "loss": 1.3625,
+      "step": 65120
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.758092905793219e-05,
+      "loss": 1.4408,
+      "step": 65130
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.757978112803244e-05,
+      "loss": 1.4547,
+      "step": 65140
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.757863319813269e-05,
+      "loss": 1.4568,
+      "step": 65150
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.757748526823295e-05,
+      "loss": 1.3876,
+      "step": 65160
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.75763373383332e-05,
+      "loss": 1.4529,
+      "step": 65170
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.757518940843345e-05,
+      "loss": 1.3886,
+      "step": 65180
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.75740414785337e-05,
+      "loss": 1.4734,
+      "step": 65190
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.757289354863397e-05,
+      "loss": 1.4217,
+      "step": 65200
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.75717456187342e-05,
+      "loss": 1.4476,
+      "step": 65210
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.757059768883447e-05,
+      "loss": 1.3502,
+      "step": 65220
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.756944975893472e-05,
+      "loss": 1.3918,
+      "step": 65230
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.756830182903497e-05,
+      "loss": 1.4656,
+      "step": 65240
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.756715389913521e-05,
+      "loss": 1.2977,
+      "step": 65250
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.756600596923548e-05,
+      "loss": 1.484,
+      "step": 65260
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.756485803933573e-05,
+      "loss": 1.3588,
+      "step": 65270
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.756371010943598e-05,
+      "loss": 1.3687,
+      "step": 65280
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.756256217953623e-05,
+      "loss": 1.4671,
+      "step": 65290
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.756141424963648e-05,
+      "loss": 1.4903,
+      "step": 65300
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.756026631973674e-05,
+      "loss": 1.5955,
+      "step": 65310
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.755911838983699e-05,
+      "loss": 1.4422,
+      "step": 65320
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.755797045993724e-05,
+      "loss": 1.4034,
+      "step": 65330
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.755682253003749e-05,
+      "loss": 1.2193,
+      "step": 65340
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.755567460013775e-05,
+      "loss": 1.433,
+      "step": 65350
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.755452667023799e-05,
+      "loss": 1.5668,
+      "step": 65360
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.755337874033825e-05,
+      "loss": 1.3281,
+      "step": 65370
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.75522308104385e-05,
+      "loss": 1.4673,
+      "step": 65380
+    },
+    {
+      "epoch": 3.0,
+      "eval_cer": 0.6724827633926006,
+      "eval_em": 0.007357212003872217,
+      "eval_f1": 0.007357212003872217,
+      "eval_loss": 1.2895028591156006,
+      "eval_runtime": 1603.0047,
+      "eval_samples_per_second": 6.444,
+      "eval_steps_per_second": 0.806,
+      "eval_wer": 0.9926427879961278,
+      "step": 65386
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.755108288053875e-05,
+      "loss": 1.2448,
+      "step": 65390
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.7549934950639e-05,
+      "loss": 1.2836,
+      "step": 65400
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.754878702073927e-05,
+      "loss": 1.3496,
+      "step": 65410
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.754763909083952e-05,
+      "loss": 1.226,
+      "step": 65420
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.754649116093977e-05,
+      "loss": 1.4369,
+      "step": 65430
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.754534323104002e-05,
+      "loss": 1.3679,
+      "step": 65440
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.754419530114027e-05,
+      "loss": 1.4648,
+      "step": 65450
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.754304737124053e-05,
+      "loss": 1.498,
+      "step": 65460
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.754189944134078e-05,
+      "loss": 1.362,
+      "step": 65470
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.754075151144103e-05,
+      "loss": 1.5391,
+      "step": 65480
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.753960358154128e-05,
+      "loss": 1.6444,
+      "step": 65490
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.753845565164154e-05,
+      "loss": 1.3563,
+      "step": 65500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.753730772174178e-05,
+      "loss": 1.515,
+      "step": 65510
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.753615979184204e-05,
+      "loss": 1.3921,
+      "step": 65520
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.753501186194229e-05,
+      "loss": 1.3769,
+      "step": 65530
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.753386393204254e-05,
+      "loss": 1.4322,
+      "step": 65540
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.753271600214279e-05,
+      "loss": 1.3423,
+      "step": 65550
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.753156807224306e-05,
+      "loss": 1.2718,
+      "step": 65560
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.75304201423433e-05,
+      "loss": 1.3486,
+      "step": 65570
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.752927221244356e-05,
+      "loss": 1.3913,
+      "step": 65580
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.75281242825438e-05,
+      "loss": 1.4431,
+      "step": 65590
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.752697635264405e-05,
+      "loss": 1.2263,
+      "step": 65600
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.752582842274432e-05,
+      "loss": 1.6518,
+      "step": 65610
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.752468049284457e-05,
+      "loss": 1.3645,
+      "step": 65620
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.752353256294482e-05,
+      "loss": 1.3842,
+      "step": 65630
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.752238463304507e-05,
+      "loss": 1.3009,
+      "step": 65640
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.752123670314533e-05,
+      "loss": 1.4812,
+      "step": 65650
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.752008877324557e-05,
+      "loss": 1.3699,
+      "step": 65660
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.751894084334583e-05,
+      "loss": 1.4889,
+      "step": 65670
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.751779291344608e-05,
+      "loss": 1.3586,
+      "step": 65680
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.751664498354633e-05,
+      "loss": 1.4547,
+      "step": 65690
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.751549705364658e-05,
+      "loss": 1.3613,
+      "step": 65700
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.751434912374683e-05,
+      "loss": 1.5721,
+      "step": 65710
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.75132011938471e-05,
+      "loss": 1.5738,
+      "step": 65720
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.751205326394734e-05,
+      "loss": 1.3421,
+      "step": 65730
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.75109053340476e-05,
+      "loss": 1.5103,
+      "step": 65740
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.750975740414784e-05,
+      "loss": 1.4397,
+      "step": 65750
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.75086094742481e-05,
+      "loss": 1.2982,
+      "step": 65760
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.750746154434834e-05,
+      "loss": 1.3984,
+      "step": 65770
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.75063136144486e-05,
+      "loss": 1.531,
+      "step": 65780
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.750516568454886e-05,
+      "loss": 1.3137,
+      "step": 65790
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.75040177546491e-05,
+      "loss": 1.5549,
+      "step": 65800
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.750286982474936e-05,
+      "loss": 1.5426,
+      "step": 65810
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.750172189484962e-05,
+      "loss": 1.519,
+      "step": 65820
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.750057396494987e-05,
+      "loss": 1.4239,
+      "step": 65830
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.749942603505012e-05,
+      "loss": 1.3185,
+      "step": 65840
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.749827810515037e-05,
+      "loss": 1.3394,
+      "step": 65850
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.749713017525062e-05,
+      "loss": 1.4665,
+      "step": 65860
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.749598224535088e-05,
+      "loss": 1.4738,
+      "step": 65870
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.749483431545113e-05,
+      "loss": 1.3314,
+      "step": 65880
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.749368638555138e-05,
+      "loss": 1.3207,
+      "step": 65890
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.749253845565163e-05,
+      "loss": 1.4542,
+      "step": 65900
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.74913905257519e-05,
+      "loss": 1.4468,
+      "step": 65910
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.749024259585213e-05,
+      "loss": 1.2767,
+      "step": 65920
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.74890946659524e-05,
+      "loss": 1.3194,
+      "step": 65930
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.748794673605264e-05,
+      "loss": 1.4515,
+      "step": 65940
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.74867988061529e-05,
+      "loss": 1.4103,
+      "step": 65950
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.748565087625314e-05,
+      "loss": 1.4386,
+      "step": 65960
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.748450294635341e-05,
+      "loss": 1.4215,
+      "step": 65970
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.748335501645366e-05,
+      "loss": 1.4036,
+      "step": 65980
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.748220708655391e-05,
+      "loss": 1.4033,
+      "step": 65990
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.748105915665416e-05,
+      "loss": 1.505,
+      "step": 66000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.747991122675441e-05,
+      "loss": 1.3238,
+      "step": 66010
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.747876329685467e-05,
+      "loss": 1.4164,
+      "step": 66020
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.747761536695492e-05,
+      "loss": 1.4109,
+      "step": 66030
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.747646743705517e-05,
+      "loss": 1.4624,
+      "step": 66040
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.747531950715542e-05,
+      "loss": 1.5819,
+      "step": 66050
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.747417157725568e-05,
+      "loss": 1.2678,
+      "step": 66060
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.747302364735592e-05,
+      "loss": 1.2876,
+      "step": 66070
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.747187571745618e-05,
+      "loss": 1.437,
+      "step": 66080
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.747072778755643e-05,
+      "loss": 1.2719,
+      "step": 66090
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.746957985765668e-05,
+      "loss": 1.3156,
+      "step": 66100
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.746843192775693e-05,
+      "loss": 1.4289,
+      "step": 66110
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.74672839978572e-05,
+      "loss": 1.3708,
+      "step": 66120
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.746613606795745e-05,
+      "loss": 1.3507,
+      "step": 66130
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.74649881380577e-05,
+      "loss": 1.4079,
+      "step": 66140
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.746384020815795e-05,
+      "loss": 1.2993,
+      "step": 66150
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.74626922782582e-05,
+      "loss": 1.355,
+      "step": 66160
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.746154434835846e-05,
+      "loss": 1.4872,
+      "step": 66170
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.746039641845871e-05,
+      "loss": 1.5324,
+      "step": 66180
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.745924848855896e-05,
+      "loss": 1.4905,
+      "step": 66190
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.745810055865921e-05,
+      "loss": 1.3013,
+      "step": 66200
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.745695262875947e-05,
+      "loss": 1.3327,
+      "step": 66210
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.745580469885971e-05,
+      "loss": 1.3716,
+      "step": 66220
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.745465676895997e-05,
+      "loss": 1.36,
+      "step": 66230
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.745350883906022e-05,
+      "loss": 1.4905,
+      "step": 66240
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.745236090916047e-05,
+      "loss": 1.2785,
+      "step": 66250
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.745121297926072e-05,
+      "loss": 1.3375,
+      "step": 66260
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.745006504936098e-05,
+      "loss": 1.4561,
+      "step": 66270
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.744891711946123e-05,
+      "loss": 1.475,
+      "step": 66280
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.744776918956148e-05,
+      "loss": 1.5146,
+      "step": 66290
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.744662125966173e-05,
+      "loss": 1.4539,
+      "step": 66300
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.744547332976198e-05,
+      "loss": 1.4878,
+      "step": 66310
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.744432539986225e-05,
+      "loss": 1.2838,
+      "step": 66320
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.74431774699625e-05,
+      "loss": 1.433,
+      "step": 66330
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.744202954006275e-05,
+      "loss": 1.432,
+      "step": 66340
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.7440881610163e-05,
+      "loss": 1.4161,
+      "step": 66350
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.743973368026326e-05,
+      "loss": 1.3445,
+      "step": 66360
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.74385857503635e-05,
+      "loss": 1.3931,
+      "step": 66370
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.743743782046376e-05,
+      "loss": 1.4227,
+      "step": 66380
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.743628989056401e-05,
+      "loss": 1.5155,
+      "step": 66390
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.743514196066426e-05,
+      "loss": 1.4364,
+      "step": 66400
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.743399403076451e-05,
+      "loss": 1.4579,
+      "step": 66410
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.743284610086477e-05,
+      "loss": 1.3101,
+      "step": 66420
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.743169817096502e-05,
+      "loss": 1.4585,
+      "step": 66430
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.743055024106527e-05,
+      "loss": 1.4465,
+      "step": 66440
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.742940231116552e-05,
+      "loss": 1.4903,
+      "step": 66450
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.742825438126577e-05,
+      "loss": 1.4397,
+      "step": 66460
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.742710645136604e-05,
+      "loss": 1.3489,
+      "step": 66470
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.742595852146629e-05,
+      "loss": 1.3527,
+      "step": 66480
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.742481059156654e-05,
+      "loss": 1.1766,
+      "step": 66490
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.742366266166679e-05,
+      "loss": 1.5746,
+      "step": 66500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.742251473176705e-05,
+      "loss": 1.3702,
+      "step": 66510
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.742136680186729e-05,
+      "loss": 1.4212,
+      "step": 66520
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.742021887196755e-05,
+      "loss": 1.5969,
+      "step": 66530
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.74190709420678e-05,
+      "loss": 1.4566,
+      "step": 66540
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.741792301216805e-05,
+      "loss": 1.3174,
+      "step": 66550
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.74167750822683e-05,
+      "loss": 1.5426,
+      "step": 66560
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.741562715236856e-05,
+      "loss": 1.2256,
+      "step": 66570
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.741447922246881e-05,
+      "loss": 1.4435,
+      "step": 66580
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.741333129256906e-05,
+      "loss": 1.4141,
+      "step": 66590
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.741218336266931e-05,
+      "loss": 1.233,
+      "step": 66600
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.741103543276956e-05,
+      "loss": 1.3918,
+      "step": 66610
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.740988750286982e-05,
+      "loss": 1.5103,
+      "step": 66620
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.740873957297007e-05,
+      "loss": 1.3392,
+      "step": 66630
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.740759164307032e-05,
+      "loss": 1.4511,
+      "step": 66640
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.740644371317057e-05,
+      "loss": 1.6466,
+      "step": 66650
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.740529578327084e-05,
+      "loss": 1.4982,
+      "step": 66660
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.740414785337107e-05,
+      "loss": 1.3881,
+      "step": 66670
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.740299992347134e-05,
+      "loss": 1.5461,
+      "step": 66680
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.740185199357159e-05,
+      "loss": 1.459,
+      "step": 66690
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.740070406367184e-05,
+      "loss": 1.3869,
+      "step": 66700
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.739955613377209e-05,
+      "loss": 1.3327,
+      "step": 66710
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.739840820387235e-05,
+      "loss": 1.4524,
+      "step": 66720
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.73972602739726e-05,
+      "loss": 1.2816,
+      "step": 66730
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.739611234407285e-05,
+      "loss": 1.4407,
+      "step": 66740
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.73949644141731e-05,
+      "loss": 1.3228,
+      "step": 66750
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.739381648427335e-05,
+      "loss": 1.3609,
+      "step": 66760
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.739266855437361e-05,
+      "loss": 1.414,
+      "step": 66770
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.739152062447386e-05,
+      "loss": 1.3513,
+      "step": 66780
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.739037269457411e-05,
+      "loss": 1.3168,
+      "step": 66790
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.738922476467436e-05,
+      "loss": 1.3825,
+      "step": 66800
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.738807683477463e-05,
+      "loss": 1.4038,
+      "step": 66810
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.738692890487486e-05,
+      "loss": 1.3925,
+      "step": 66820
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.738578097497513e-05,
+      "loss": 1.5023,
+      "step": 66830
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.738463304507538e-05,
+      "loss": 1.474,
+      "step": 66840
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.738348511517563e-05,
+      "loss": 1.402,
+      "step": 66850
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.738233718527588e-05,
+      "loss": 1.3974,
+      "step": 66860
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.738118925537614e-05,
+      "loss": 1.4359,
+      "step": 66870
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.738004132547639e-05,
+      "loss": 1.2279,
+      "step": 66880
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.737889339557664e-05,
+      "loss": 1.3178,
+      "step": 66890
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.737774546567689e-05,
+      "loss": 1.3959,
+      "step": 66900
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.737659753577714e-05,
+      "loss": 1.3126,
+      "step": 66910
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.73754496058774e-05,
+      "loss": 1.3407,
+      "step": 66920
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.737430167597765e-05,
+      "loss": 1.5314,
+      "step": 66930
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.73731537460779e-05,
+      "loss": 1.3152,
+      "step": 66940
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.737200581617815e-05,
+      "loss": 1.3725,
+      "step": 66950
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.737085788627841e-05,
+      "loss": 1.4118,
+      "step": 66960
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.736970995637865e-05,
+      "loss": 1.4947,
+      "step": 66970
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.736856202647891e-05,
+      "loss": 1.4686,
+      "step": 66980
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.736741409657916e-05,
+      "loss": 1.57,
+      "step": 66990
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.736626616667941e-05,
+      "loss": 1.409,
+      "step": 67000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.736511823677966e-05,
+      "loss": 1.3499,
+      "step": 67010
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.736397030687993e-05,
+      "loss": 1.4411,
+      "step": 67020
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.736282237698018e-05,
+      "loss": 1.2935,
+      "step": 67030
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.736167444708043e-05,
+      "loss": 1.162,
+      "step": 67040
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.736052651718068e-05,
+      "loss": 1.3842,
+      "step": 67050
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.735937858728093e-05,
+      "loss": 1.3492,
+      "step": 67060
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.735823065738119e-05,
+      "loss": 1.3716,
+      "step": 67070
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.735708272748143e-05,
+      "loss": 1.4565,
+      "step": 67080
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.735593479758169e-05,
+      "loss": 1.4376,
+      "step": 67090
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.735478686768194e-05,
+      "loss": 1.5368,
+      "step": 67100
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.735363893778219e-05,
+      "loss": 1.4594,
+      "step": 67110
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.735249100788244e-05,
+      "loss": 1.4002,
+      "step": 67120
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.73513430779827e-05,
+      "loss": 1.4128,
+      "step": 67130
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.735019514808295e-05,
+      "loss": 1.433,
+      "step": 67140
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.73490472181832e-05,
+      "loss": 1.3735,
+      "step": 67150
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.734789928828345e-05,
+      "loss": 1.3195,
+      "step": 67160
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.73467513583837e-05,
+      "loss": 1.323,
+      "step": 67170
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.734560342848397e-05,
+      "loss": 1.4387,
+      "step": 67180
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.734445549858422e-05,
+      "loss": 1.4412,
+      "step": 67190
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.734330756868447e-05,
+      "loss": 1.4818,
+      "step": 67200
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.734215963878471e-05,
+      "loss": 1.3674,
+      "step": 67210
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.734101170888498e-05,
+      "loss": 1.4595,
+      "step": 67220
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.733986377898521e-05,
+      "loss": 1.4534,
+      "step": 67230
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.733871584908548e-05,
+      "loss": 1.4545,
+      "step": 67240
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.733756791918573e-05,
+      "loss": 1.3572,
+      "step": 67250
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.733641998928598e-05,
+      "loss": 1.3559,
+      "step": 67260
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.733527205938623e-05,
+      "loss": 1.4108,
+      "step": 67270
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.733412412948649e-05,
+      "loss": 1.5713,
+      "step": 67280
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.733297619958674e-05,
+      "loss": 1.536,
+      "step": 67290
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.733182826968699e-05,
+      "loss": 1.3621,
+      "step": 67300
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.733068033978724e-05,
+      "loss": 1.4044,
+      "step": 67310
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.732953240988749e-05,
+      "loss": 1.2756,
+      "step": 67320
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.732838447998775e-05,
+      "loss": 1.4177,
+      "step": 67330
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7327236550088e-05,
+      "loss": 1.475,
+      "step": 67340
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.732608862018825e-05,
+      "loss": 1.4444,
+      "step": 67350
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.73249406902885e-05,
+      "loss": 1.5658,
+      "step": 67360
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.732379276038877e-05,
+      "loss": 1.3633,
+      "step": 67370
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7322644830489e-05,
+      "loss": 1.2764,
+      "step": 67380
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.732149690058927e-05,
+      "loss": 1.4735,
+      "step": 67390
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.732034897068952e-05,
+      "loss": 1.5227,
+      "step": 67400
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.731920104078977e-05,
+      "loss": 1.4258,
+      "step": 67410
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.731805311089002e-05,
+      "loss": 1.4465,
+      "step": 67420
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.731690518099028e-05,
+      "loss": 1.4456,
+      "step": 67430
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.731575725109053e-05,
+      "loss": 1.3806,
+      "step": 67440
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.731460932119078e-05,
+      "loss": 1.4732,
+      "step": 67450
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.731346139129103e-05,
+      "loss": 1.3994,
+      "step": 67460
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.731231346139128e-05,
+      "loss": 1.4503,
+      "step": 67470
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.731116553149154e-05,
+      "loss": 1.4261,
+      "step": 67480
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.731001760159179e-05,
+      "loss": 1.3918,
+      "step": 67490
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.730886967169204e-05,
+      "loss": 1.3773,
+      "step": 67500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.730772174179229e-05,
+      "loss": 1.4512,
+      "step": 67510
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.730657381189256e-05,
+      "loss": 1.4323,
+      "step": 67520
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.730542588199279e-05,
+      "loss": 1.3822,
+      "step": 67530
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.730427795209306e-05,
+      "loss": 1.3256,
+      "step": 67540
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.73031300221933e-05,
+      "loss": 1.3563,
+      "step": 67550
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.730198209229355e-05,
+      "loss": 1.4671,
+      "step": 67560
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.73008341623938e-05,
+      "loss": 1.4689,
+      "step": 67570
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.729968623249407e-05,
+      "loss": 1.34,
+      "step": 67580
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.729853830259432e-05,
+      "loss": 1.3781,
+      "step": 67590
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.729739037269457e-05,
+      "loss": 1.4132,
+      "step": 67600
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.729624244279482e-05,
+      "loss": 1.5269,
+      "step": 67610
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.729509451289507e-05,
+      "loss": 1.5414,
+      "step": 67620
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.729394658299533e-05,
+      "loss": 1.5111,
+      "step": 67630
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.729279865309558e-05,
+      "loss": 1.4188,
+      "step": 67640
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.729165072319583e-05,
+      "loss": 1.3334,
+      "step": 67650
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.729050279329608e-05,
+      "loss": 1.3603,
+      "step": 67660
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.728935486339634e-05,
+      "loss": 1.4225,
+      "step": 67670
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.728820693349658e-05,
+      "loss": 1.351,
+      "step": 67680
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.728705900359684e-05,
+      "loss": 1.3995,
+      "step": 67690
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.72859110736971e-05,
+      "loss": 1.5429,
+      "step": 67700
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.728476314379734e-05,
+      "loss": 1.2445,
+      "step": 67710
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.728361521389759e-05,
+      "loss": 1.3769,
+      "step": 67720
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.728246728399786e-05,
+      "loss": 1.3963,
+      "step": 67730
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.72813193540981e-05,
+      "loss": 1.4512,
+      "step": 67740
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.728017142419836e-05,
+      "loss": 1.3621,
+      "step": 67750
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.72790234942986e-05,
+      "loss": 1.3552,
+      "step": 67760
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.727787556439886e-05,
+      "loss": 1.4055,
+      "step": 67770
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.727672763449912e-05,
+      "loss": 1.523,
+      "step": 67780
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.727557970459937e-05,
+      "loss": 1.4992,
+      "step": 67790
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.727443177469962e-05,
+      "loss": 1.3642,
+      "step": 67800
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.727328384479987e-05,
+      "loss": 1.5209,
+      "step": 67810
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.727213591490013e-05,
+      "loss": 1.3974,
+      "step": 67820
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.727098798500037e-05,
+      "loss": 1.4926,
+      "step": 67830
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.726984005510063e-05,
+      "loss": 1.4301,
+      "step": 67840
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.726869212520088e-05,
+      "loss": 1.3538,
+      "step": 67850
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.726754419530113e-05,
+      "loss": 1.3829,
+      "step": 67860
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.726639626540138e-05,
+      "loss": 1.3282,
+      "step": 67870
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.726524833550164e-05,
+      "loss": 1.3242,
+      "step": 67880
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.72641004056019e-05,
+      "loss": 1.4436,
+      "step": 67890
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.726295247570214e-05,
+      "loss": 1.2942,
+      "step": 67900
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.72618045458024e-05,
+      "loss": 1.363,
+      "step": 67910
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.726065661590264e-05,
+      "loss": 1.3578,
+      "step": 67920
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.725950868600291e-05,
+      "loss": 1.394,
+      "step": 67930
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.725836075610316e-05,
+      "loss": 1.3678,
+      "step": 67940
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.725721282620341e-05,
+      "loss": 1.2588,
+      "step": 67950
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.725606489630366e-05,
+      "loss": 1.2452,
+      "step": 67960
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.725491696640392e-05,
+      "loss": 1.452,
+      "step": 67970
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.725376903650416e-05,
+      "loss": 1.4002,
+      "step": 67980
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.725262110660442e-05,
+      "loss": 1.4011,
+      "step": 67990
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.725147317670467e-05,
+      "loss": 1.4367,
+      "step": 68000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.725032524680492e-05,
+      "loss": 1.3331,
+      "step": 68010
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.724917731690517e-05,
+      "loss": 1.3124,
+      "step": 68020
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.724802938700543e-05,
+      "loss": 1.4566,
+      "step": 68030
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.724688145710568e-05,
+      "loss": 1.3796,
+      "step": 68040
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.724573352720593e-05,
+      "loss": 1.3898,
+      "step": 68050
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.724458559730618e-05,
+      "loss": 1.426,
+      "step": 68060
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.724343766740643e-05,
+      "loss": 1.4581,
+      "step": 68070
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.72422897375067e-05,
+      "loss": 1.2304,
+      "step": 68080
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.724114180760695e-05,
+      "loss": 1.5421,
+      "step": 68090
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.72399938777072e-05,
+      "loss": 1.4088,
+      "step": 68100
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.723884594780745e-05,
+      "loss": 1.4117,
+      "step": 68110
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.723769801790771e-05,
+      "loss": 1.5159,
+      "step": 68120
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.723655008800795e-05,
+      "loss": 1.3665,
+      "step": 68130
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.723540215810821e-05,
+      "loss": 1.4492,
+      "step": 68140
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.723425422820846e-05,
+      "loss": 1.3041,
+      "step": 68150
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.723310629830871e-05,
+      "loss": 1.5257,
+      "step": 68160
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.723195836840896e-05,
+      "loss": 1.3695,
+      "step": 68170
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.723081043850922e-05,
+      "loss": 1.2331,
+      "step": 68180
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.722966250860947e-05,
+      "loss": 1.7153,
+      "step": 68190
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.722851457870972e-05,
+      "loss": 1.3892,
+      "step": 68200
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.722736664880997e-05,
+      "loss": 1.4307,
+      "step": 68210
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.722621871891022e-05,
+      "loss": 1.3102,
+      "step": 68220
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.722507078901048e-05,
+      "loss": 1.3398,
+      "step": 68230
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.722392285911073e-05,
+      "loss": 1.4323,
+      "step": 68240
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.722277492921098e-05,
+      "loss": 1.325,
+      "step": 68250
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.722162699931123e-05,
+      "loss": 1.2471,
+      "step": 68260
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.72204790694115e-05,
+      "loss": 1.5794,
+      "step": 68270
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.721933113951173e-05,
+      "loss": 1.4321,
+      "step": 68280
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.7218183209612e-05,
+      "loss": 1.419,
+      "step": 68290
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.721703527971225e-05,
+      "loss": 1.3591,
+      "step": 68300
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.72158873498125e-05,
+      "loss": 1.2975,
+      "step": 68310
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.721473941991275e-05,
+      "loss": 1.4835,
+      "step": 68320
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.721359149001301e-05,
+      "loss": 1.3594,
+      "step": 68330
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.721244356011326e-05,
+      "loss": 1.3661,
+      "step": 68340
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.721129563021351e-05,
+      "loss": 1.4385,
+      "step": 68350
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.721014770031376e-05,
+      "loss": 1.3552,
+      "step": 68360
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.720899977041401e-05,
+      "loss": 1.4146,
+      "step": 68370
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.720785184051427e-05,
+      "loss": 1.3432,
+      "step": 68380
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.720670391061452e-05,
+      "loss": 1.5357,
+      "step": 68390
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.720555598071477e-05,
+      "loss": 1.4332,
+      "step": 68400
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.720440805081502e-05,
+      "loss": 1.3493,
+      "step": 68410
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.720326012091529e-05,
+      "loss": 1.3271,
+      "step": 68420
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.720211219101552e-05,
+      "loss": 1.337,
+      "step": 68430
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.720096426111579e-05,
+      "loss": 1.5509,
+      "step": 68440
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.719981633121604e-05,
+      "loss": 1.4132,
+      "step": 68450
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.719866840131629e-05,
+      "loss": 1.2787,
+      "step": 68460
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.719752047141654e-05,
+      "loss": 1.452,
+      "step": 68470
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.719637254151679e-05,
+      "loss": 1.41,
+      "step": 68480
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.719522461161705e-05,
+      "loss": 1.493,
+      "step": 68490
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.71940766817173e-05,
+      "loss": 1.4335,
+      "step": 68500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.719292875181755e-05,
+      "loss": 1.4166,
+      "step": 68510
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.71917808219178e-05,
+      "loss": 1.5055,
+      "step": 68520
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.719063289201806e-05,
+      "loss": 1.3134,
+      "step": 68530
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.71894849621183e-05,
+      "loss": 1.4498,
+      "step": 68540
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.718833703221856e-05,
+      "loss": 1.3297,
+      "step": 68550
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.718718910231881e-05,
+      "loss": 1.4062,
+      "step": 68560
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.718604117241906e-05,
+      "loss": 1.23,
+      "step": 68570
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.718489324251931e-05,
+      "loss": 1.3916,
+      "step": 68580
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.718374531261957e-05,
+      "loss": 1.3712,
+      "step": 68590
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.718259738271982e-05,
+      "loss": 1.216,
+      "step": 68600
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.718144945282007e-05,
+      "loss": 1.6309,
+      "step": 68610
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.718030152292032e-05,
+      "loss": 1.4466,
+      "step": 68620
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.717915359302057e-05,
+      "loss": 1.309,
+      "step": 68630
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.717800566312084e-05,
+      "loss": 1.3234,
+      "step": 68640
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.717685773322109e-05,
+      "loss": 1.5579,
+      "step": 68650
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.717570980332134e-05,
+      "loss": 1.3856,
+      "step": 68660
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.717456187342159e-05,
+      "loss": 1.4612,
+      "step": 68670
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.717341394352185e-05,
+      "loss": 1.199,
+      "step": 68680
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.717226601362209e-05,
+      "loss": 1.4679,
+      "step": 68690
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.717111808372235e-05,
+      "loss": 1.5175,
+      "step": 68700
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.71699701538226e-05,
+      "loss": 1.6088,
+      "step": 68710
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.716882222392285e-05,
+      "loss": 1.4646,
+      "step": 68720
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.71676742940231e-05,
+      "loss": 1.348,
+      "step": 68730
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.716652636412336e-05,
+      "loss": 1.4422,
+      "step": 68740
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.716537843422361e-05,
+      "loss": 1.5377,
+      "step": 68750
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.716423050432386e-05,
+      "loss": 1.4044,
+      "step": 68760
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.716308257442411e-05,
+      "loss": 1.4952,
+      "step": 68770
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.716193464452436e-05,
+      "loss": 1.3782,
+      "step": 68780
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.716078671462463e-05,
+      "loss": 1.3388,
+      "step": 68790
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.715963878472488e-05,
+      "loss": 1.3109,
+      "step": 68800
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.715849085482513e-05,
+      "loss": 1.5251,
+      "step": 68810
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.715734292492538e-05,
+      "loss": 1.3711,
+      "step": 68820
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.715619499502564e-05,
+      "loss": 1.4351,
+      "step": 68830
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.715504706512587e-05,
+      "loss": 1.4385,
+      "step": 68840
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.715389913522614e-05,
+      "loss": 1.3414,
+      "step": 68850
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.715275120532639e-05,
+      "loss": 1.5015,
+      "step": 68860
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.715160327542664e-05,
+      "loss": 1.2241,
+      "step": 68870
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.715045534552689e-05,
+      "loss": 1.2776,
+      "step": 68880
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.714930741562715e-05,
+      "loss": 1.3228,
+      "step": 68890
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.71481594857274e-05,
+      "loss": 1.4675,
+      "step": 68900
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.714701155582765e-05,
+      "loss": 1.4398,
+      "step": 68910
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.71458636259279e-05,
+      "loss": 1.3511,
+      "step": 68920
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.714471569602815e-05,
+      "loss": 1.5591,
+      "step": 68930
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.714356776612841e-05,
+      "loss": 1.3474,
+      "step": 68940
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.714241983622866e-05,
+      "loss": 1.5232,
+      "step": 68950
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.714127190632891e-05,
+      "loss": 1.3902,
+      "step": 68960
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.714012397642916e-05,
+      "loss": 1.3785,
+      "step": 68970
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.713897604652943e-05,
+      "loss": 1.4144,
+      "step": 68980
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.713782811662966e-05,
+      "loss": 1.4427,
+      "step": 68990
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.713668018672993e-05,
+      "loss": 1.4511,
+      "step": 69000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.713553225683018e-05,
+      "loss": 1.3618,
+      "step": 69010
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.713438432693043e-05,
+      "loss": 1.3891,
+      "step": 69020
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.713323639703068e-05,
+      "loss": 1.2894,
+      "step": 69030
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.713208846713094e-05,
+      "loss": 1.3179,
+      "step": 69040
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.713094053723119e-05,
+      "loss": 1.2653,
+      "step": 69050
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.712979260733144e-05,
+      "loss": 1.1856,
+      "step": 69060
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.712864467743169e-05,
+      "loss": 1.3729,
+      "step": 69070
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.712749674753194e-05,
+      "loss": 1.3086,
+      "step": 69080
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.71263488176322e-05,
+      "loss": 1.5073,
+      "step": 69090
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.712520088773245e-05,
+      "loss": 1.392,
+      "step": 69100
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.71240529578327e-05,
+      "loss": 1.35,
+      "step": 69110
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.712290502793295e-05,
+      "loss": 1.3838,
+      "step": 69120
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.712175709803322e-05,
+      "loss": 1.4752,
+      "step": 69130
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.712060916813345e-05,
+      "loss": 1.3259,
+      "step": 69140
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.711946123823372e-05,
+      "loss": 1.5308,
+      "step": 69150
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.711831330833397e-05,
+      "loss": 1.2878,
+      "step": 69160
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.711716537843421e-05,
+      "loss": 1.4622,
+      "step": 69170
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.711601744853446e-05,
+      "loss": 1.4811,
+      "step": 69180
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.711486951863473e-05,
+      "loss": 1.1966,
+      "step": 69190
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.711372158873498e-05,
+      "loss": 1.3339,
+      "step": 69200
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.711257365883523e-05,
+      "loss": 1.3374,
+      "step": 69210
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.711142572893548e-05,
+      "loss": 1.3621,
+      "step": 69220
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.711027779903573e-05,
+      "loss": 1.4166,
+      "step": 69230
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.710912986913599e-05,
+      "loss": 1.4898,
+      "step": 69240
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.710798193923624e-05,
+      "loss": 1.423,
+      "step": 69250
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.710683400933649e-05,
+      "loss": 1.3999,
+      "step": 69260
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.710568607943674e-05,
+      "loss": 1.3714,
+      "step": 69270
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.7104538149537e-05,
+      "loss": 1.3995,
+      "step": 69280
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.710339021963724e-05,
+      "loss": 1.3789,
+      "step": 69290
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.71022422897375e-05,
+      "loss": 1.494,
+      "step": 69300
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.710109435983775e-05,
+      "loss": 1.4023,
+      "step": 69310
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.7099946429938e-05,
+      "loss": 1.2025,
+      "step": 69320
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.709879850003825e-05,
+      "loss": 1.4641,
+      "step": 69330
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.709765057013852e-05,
+      "loss": 1.2544,
+      "step": 69340
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.709650264023877e-05,
+      "loss": 1.3924,
+      "step": 69350
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.709535471033902e-05,
+      "loss": 1.415,
+      "step": 69360
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.709420678043927e-05,
+      "loss": 1.3618,
+      "step": 69370
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.709305885053952e-05,
+      "loss": 1.4593,
+      "step": 69380
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.709191092063978e-05,
+      "loss": 1.5272,
+      "step": 69390
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.709076299074003e-05,
+      "loss": 1.3829,
+      "step": 69400
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.708961506084028e-05,
+      "loss": 1.3166,
+      "step": 69410
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.708846713094053e-05,
+      "loss": 1.3406,
+      "step": 69420
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.708731920104079e-05,
+      "loss": 1.4092,
+      "step": 69430
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.708617127114103e-05,
+      "loss": 1.532,
+      "step": 69440
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.708502334124129e-05,
+      "loss": 1.2781,
+      "step": 69450
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.708387541134154e-05,
+      "loss": 1.2728,
+      "step": 69460
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.708272748144179e-05,
+      "loss": 1.3925,
+      "step": 69470
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.708157955154204e-05,
+      "loss": 1.4731,
+      "step": 69480
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.70804316216423e-05,
+      "loss": 1.3378,
+      "step": 69490
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.707928369174256e-05,
+      "loss": 1.367,
+      "step": 69500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.70781357618428e-05,
+      "loss": 1.5434,
+      "step": 69510
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.707698783194305e-05,
+      "loss": 1.4403,
+      "step": 69520
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.70758399020433e-05,
+      "loss": 1.3984,
+      "step": 69530
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.707469197214357e-05,
+      "loss": 1.5385,
+      "step": 69540
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.707354404224382e-05,
+      "loss": 1.2504,
+      "step": 69550
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.707239611234407e-05,
+      "loss": 1.4207,
+      "step": 69560
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.707124818244432e-05,
+      "loss": 1.5491,
+      "step": 69570
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.707010025254458e-05,
+      "loss": 1.4328,
+      "step": 69580
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.706895232264482e-05,
+      "loss": 1.3707,
+      "step": 69590
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.706780439274508e-05,
+      "loss": 1.2461,
+      "step": 69600
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.706665646284533e-05,
+      "loss": 1.449,
+      "step": 69610
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.706550853294558e-05,
+      "loss": 1.2544,
+      "step": 69620
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.706436060304583e-05,
+      "loss": 1.6143,
+      "step": 69630
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.70632126731461e-05,
+      "loss": 1.354,
+      "step": 69640
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.706206474324634e-05,
+      "loss": 1.2657,
+      "step": 69650
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.70609168133466e-05,
+      "loss": 1.3541,
+      "step": 69660
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.705976888344684e-05,
+      "loss": 1.4012,
+      "step": 69670
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.705862095354709e-05,
+      "loss": 1.4092,
+      "step": 69680
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.705747302364736e-05,
+      "loss": 1.3754,
+      "step": 69690
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.70563250937476e-05,
+      "loss": 1.247,
+      "step": 69700
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.705517716384786e-05,
+      "loss": 1.4979,
+      "step": 69710
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.70540292339481e-05,
+      "loss": 1.2595,
+      "step": 69720
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.705288130404837e-05,
+      "loss": 1.4335,
+      "step": 69730
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.70517333741486e-05,
+      "loss": 1.4152,
+      "step": 69740
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.705058544424887e-05,
+      "loss": 1.5381,
+      "step": 69750
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.704943751434912e-05,
+      "loss": 1.3807,
+      "step": 69760
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.704828958444937e-05,
+      "loss": 1.3464,
+      "step": 69770
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.704714165454962e-05,
+      "loss": 1.3642,
+      "step": 69780
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.704599372464988e-05,
+      "loss": 1.6087,
+      "step": 69790
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.704484579475013e-05,
+      "loss": 1.386,
+      "step": 69800
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.704369786485038e-05,
+      "loss": 1.3189,
+      "step": 69810
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.704254993495063e-05,
+      "loss": 1.3246,
+      "step": 69820
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.704140200505088e-05,
+      "loss": 1.5124,
+      "step": 69830
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.704025407515114e-05,
+      "loss": 1.3968,
+      "step": 69840
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.703910614525138e-05,
+      "loss": 1.2452,
+      "step": 69850
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.703795821535164e-05,
+      "loss": 1.4184,
+      "step": 69860
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.70368102854519e-05,
+      "loss": 1.4206,
+      "step": 69870
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.703566235555214e-05,
+      "loss": 1.3331,
+      "step": 69880
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.70345144256524e-05,
+      "loss": 1.4106,
+      "step": 69890
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.703336649575266e-05,
+      "loss": 1.3216,
+      "step": 69900
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.703221856585291e-05,
+      "loss": 1.3357,
+      "step": 69910
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.703107063595316e-05,
+      "loss": 1.2729,
+      "step": 69920
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.702992270605341e-05,
+      "loss": 1.3589,
+      "step": 69930
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.702877477615366e-05,
+      "loss": 1.402,
+      "step": 69940
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.702762684625392e-05,
+      "loss": 1.3207,
+      "step": 69950
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.702647891635417e-05,
+      "loss": 1.4119,
+      "step": 69960
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.702533098645442e-05,
+      "loss": 1.3857,
+      "step": 69970
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.702418305655467e-05,
+      "loss": 1.5677,
+      "step": 69980
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.702303512665493e-05,
+      "loss": 1.4204,
+      "step": 69990
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.702188719675517e-05,
+      "loss": 1.5252,
+      "step": 70000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.702073926685543e-05,
+      "loss": 1.2625,
+      "step": 70010
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.701959133695568e-05,
+      "loss": 1.4432,
+      "step": 70020
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.701844340705593e-05,
+      "loss": 1.4247,
+      "step": 70030
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.701729547715618e-05,
+      "loss": 1.2689,
+      "step": 70040
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.701614754725645e-05,
+      "loss": 1.5912,
+      "step": 70050
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.70149996173567e-05,
+      "loss": 1.3867,
+      "step": 70060
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.701385168745695e-05,
+      "loss": 1.3797,
+      "step": 70070
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.70127037575572e-05,
+      "loss": 1.3867,
+      "step": 70080
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.701155582765745e-05,
+      "loss": 1.3322,
+      "step": 70090
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.701040789775771e-05,
+      "loss": 1.5111,
+      "step": 70100
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.700925996785796e-05,
+      "loss": 1.3689,
+      "step": 70110
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.700811203795821e-05,
+      "loss": 1.6423,
+      "step": 70120
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.700696410805846e-05,
+      "loss": 1.3214,
+      "step": 70130
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.700581617815872e-05,
+      "loss": 1.5238,
+      "step": 70140
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.700466824825896e-05,
+      "loss": 1.4992,
+      "step": 70150
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.700352031835922e-05,
+      "loss": 1.2973,
+      "step": 70160
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.700237238845947e-05,
+      "loss": 1.4384,
+      "step": 70170
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.700122445855972e-05,
+      "loss": 1.5366,
+      "step": 70180
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.700007652865997e-05,
+      "loss": 1.3302,
+      "step": 70190
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.699892859876023e-05,
+      "loss": 1.401,
+      "step": 70200
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.699778066886048e-05,
+      "loss": 1.3855,
+      "step": 70210
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.699663273896073e-05,
+      "loss": 1.3662,
+      "step": 70220
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.699548480906098e-05,
+      "loss": 1.4016,
+      "step": 70230
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.699433687916123e-05,
+      "loss": 1.2829,
+      "step": 70240
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.69931889492615e-05,
+      "loss": 1.3793,
+      "step": 70250
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.699204101936175e-05,
+      "loss": 1.3276,
+      "step": 70260
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.6990893089462e-05,
+      "loss": 1.4157,
+      "step": 70270
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.698974515956225e-05,
+      "loss": 1.3576,
+      "step": 70280
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.698859722966251e-05,
+      "loss": 1.4116,
+      "step": 70290
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.698744929976275e-05,
+      "loss": 1.4929,
+      "step": 70300
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.698630136986301e-05,
+      "loss": 1.359,
+      "step": 70310
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.698515343996326e-05,
+      "loss": 1.3744,
+      "step": 70320
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.698400551006351e-05,
+      "loss": 1.3262,
+      "step": 70330
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.698285758016376e-05,
+      "loss": 1.278,
+      "step": 70340
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.698170965026402e-05,
+      "loss": 1.5682,
+      "step": 70350
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.698056172036427e-05,
+      "loss": 1.357,
+      "step": 70360
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.697941379046452e-05,
+      "loss": 1.2507,
+      "step": 70370
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.697826586056477e-05,
+      "loss": 1.3733,
+      "step": 70380
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.697711793066502e-05,
+      "loss": 1.3882,
+      "step": 70390
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.697597000076529e-05,
+      "loss": 1.2727,
+      "step": 70400
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.697482207086554e-05,
+      "loss": 1.2618,
+      "step": 70410
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.697367414096579e-05,
+      "loss": 1.3631,
+      "step": 70420
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.697252621106604e-05,
+      "loss": 1.4426,
+      "step": 70430
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.69713782811663e-05,
+      "loss": 1.4985,
+      "step": 70440
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.697023035126654e-05,
+      "loss": 1.3758,
+      "step": 70450
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.69690824213668e-05,
+      "loss": 1.3432,
+      "step": 70460
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.696793449146705e-05,
+      "loss": 1.3791,
+      "step": 70470
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.69667865615673e-05,
+      "loss": 1.5248,
+      "step": 70480
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.696563863166755e-05,
+      "loss": 1.2861,
+      "step": 70490
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.696449070176781e-05,
+      "loss": 1.424,
+      "step": 70500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.696334277186806e-05,
+      "loss": 1.3654,
+      "step": 70510
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.696219484196831e-05,
+      "loss": 1.2341,
+      "step": 70520
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.696104691206856e-05,
+      "loss": 1.4028,
+      "step": 70530
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.695989898216881e-05,
+      "loss": 1.343,
+      "step": 70540
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.695875105226907e-05,
+      "loss": 1.5454,
+      "step": 70550
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.695760312236932e-05,
+      "loss": 1.2487,
+      "step": 70560
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.695645519246957e-05,
+      "loss": 1.4386,
+      "step": 70570
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.695530726256982e-05,
+      "loss": 1.479,
+      "step": 70580
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.695415933267009e-05,
+      "loss": 1.3002,
+      "step": 70590
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.695301140277032e-05,
+      "loss": 1.4916,
+      "step": 70600
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.695186347287059e-05,
+      "loss": 1.5621,
+      "step": 70610
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.695071554297084e-05,
+      "loss": 1.3806,
+      "step": 70620
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.694956761307109e-05,
+      "loss": 1.5362,
+      "step": 70630
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.694841968317134e-05,
+      "loss": 1.3676,
+      "step": 70640
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.69472717532716e-05,
+      "loss": 1.3997,
+      "step": 70650
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.694612382337185e-05,
+      "loss": 1.3728,
+      "step": 70660
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.69449758934721e-05,
+      "loss": 1.3891,
+      "step": 70670
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.694382796357235e-05,
+      "loss": 1.4421,
+      "step": 70680
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.69426800336726e-05,
+      "loss": 1.5019,
+      "step": 70690
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.694153210377286e-05,
+      "loss": 1.3168,
+      "step": 70700
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.694038417387311e-05,
+      "loss": 1.4025,
+      "step": 70710
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.693923624397336e-05,
+      "loss": 1.4327,
+      "step": 70720
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.693808831407361e-05,
+      "loss": 1.4795,
+      "step": 70730
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.693694038417388e-05,
+      "loss": 1.3378,
+      "step": 70740
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.693579245427411e-05,
+      "loss": 1.4487,
+      "step": 70750
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.693464452437438e-05,
+      "loss": 1.4638,
+      "step": 70760
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.693349659447463e-05,
+      "loss": 1.3148,
+      "step": 70770
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.693234866457488e-05,
+      "loss": 1.441,
+      "step": 70780
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.693120073467513e-05,
+      "loss": 1.4345,
+      "step": 70790
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.693005280477539e-05,
+      "loss": 1.6042,
+      "step": 70800
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.692890487487564e-05,
+      "loss": 1.3302,
+      "step": 70810
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.692775694497589e-05,
+      "loss": 1.4224,
+      "step": 70820
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.692660901507614e-05,
+      "loss": 1.4604,
+      "step": 70830
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.692546108517639e-05,
+      "loss": 1.2275,
+      "step": 70840
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.692431315527665e-05,
+      "loss": 1.3991,
+      "step": 70850
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.69231652253769e-05,
+      "loss": 1.3335,
+      "step": 70860
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.692201729547715e-05,
+      "loss": 1.5544,
+      "step": 70870
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.69208693655774e-05,
+      "loss": 1.3405,
+      "step": 70880
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.691972143567766e-05,
+      "loss": 1.3669,
+      "step": 70890
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.69185735057779e-05,
+      "loss": 1.2981,
+      "step": 70900
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.691742557587816e-05,
+      "loss": 1.332,
+      "step": 70910
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.691627764597841e-05,
+      "loss": 1.3747,
+      "step": 70920
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.691512971607866e-05,
+      "loss": 1.3437,
+      "step": 70930
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.691398178617891e-05,
+      "loss": 1.2987,
+      "step": 70940
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.691283385627918e-05,
+      "loss": 1.2216,
+      "step": 70950
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.691168592637943e-05,
+      "loss": 1.295,
+      "step": 70960
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.691053799647968e-05,
+      "loss": 1.3325,
+      "step": 70970
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.690939006657993e-05,
+      "loss": 1.4474,
+      "step": 70980
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.690824213668018e-05,
+      "loss": 1.1688,
+      "step": 70990
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.690709420678044e-05,
+      "loss": 1.2641,
+      "step": 71000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.690594627688069e-05,
+      "loss": 1.4603,
+      "step": 71010
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.690479834698094e-05,
+      "loss": 1.2779,
+      "step": 71020
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.690365041708119e-05,
+      "loss": 1.3852,
+      "step": 71030
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.690250248718145e-05,
+      "loss": 1.279,
+      "step": 71040
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.690135455728169e-05,
+      "loss": 1.3959,
+      "step": 71050
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.690020662738195e-05,
+      "loss": 1.3214,
+      "step": 71060
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.68990586974822e-05,
+      "loss": 1.3904,
+      "step": 71070
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.689791076758245e-05,
+      "loss": 1.3436,
+      "step": 71080
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.68967628376827e-05,
+      "loss": 1.401,
+      "step": 71090
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.689561490778297e-05,
+      "loss": 1.4734,
+      "step": 71100
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.689446697788322e-05,
+      "loss": 1.3844,
+      "step": 71110
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.689331904798347e-05,
+      "loss": 1.3197,
+      "step": 71120
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.689217111808371e-05,
+      "loss": 1.4937,
+      "step": 71130
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.689102318818396e-05,
+      "loss": 1.4406,
+      "step": 71140
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.688987525828423e-05,
+      "loss": 1.4168,
+      "step": 71150
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.688872732838448e-05,
+      "loss": 1.3876,
+      "step": 71160
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.688757939848473e-05,
+      "loss": 1.2781,
+      "step": 71170
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.688643146858498e-05,
+      "loss": 1.3539,
+      "step": 71180
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.688528353868523e-05,
+      "loss": 1.4055,
+      "step": 71190
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.688413560878548e-05,
+      "loss": 1.2746,
+      "step": 71200
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.688298767888574e-05,
+      "loss": 1.5101,
+      "step": 71210
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.688183974898599e-05,
+      "loss": 1.3557,
+      "step": 71220
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.688069181908624e-05,
+      "loss": 1.3902,
+      "step": 71230
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.687954388918649e-05,
+      "loss": 1.501,
+      "step": 71240
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.687839595928674e-05,
+      "loss": 1.4686,
+      "step": 71250
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.6877248029387e-05,
+      "loss": 1.4114,
+      "step": 71260
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.687610009948725e-05,
+      "loss": 1.3363,
+      "step": 71270
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.68749521695875e-05,
+      "loss": 1.534,
+      "step": 71280
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.687380423968775e-05,
+      "loss": 1.3615,
+      "step": 71290
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.687265630978802e-05,
+      "loss": 1.558,
+      "step": 71300
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.687150837988825e-05,
+      "loss": 1.2507,
+      "step": 71310
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.687036044998852e-05,
+      "loss": 1.3568,
+      "step": 71320
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.686921252008877e-05,
+      "loss": 1.3364,
+      "step": 71330
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.686806459018902e-05,
+      "loss": 1.4421,
+      "step": 71340
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.686691666028927e-05,
+      "loss": 1.5592,
+      "step": 71350
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.686576873038953e-05,
+      "loss": 1.407,
+      "step": 71360
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.686462080048978e-05,
+      "loss": 1.3743,
+      "step": 71370
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.686347287059003e-05,
+      "loss": 1.5273,
+      "step": 71380
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.686232494069028e-05,
+      "loss": 1.2894,
+      "step": 71390
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.686117701079053e-05,
+      "loss": 1.5307,
+      "step": 71400
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.686002908089079e-05,
+      "loss": 1.4539,
+      "step": 71410
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.685888115099104e-05,
+      "loss": 1.2595,
+      "step": 71420
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.685773322109129e-05,
+      "loss": 1.2926,
+      "step": 71430
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.685658529119154e-05,
+      "loss": 1.4037,
+      "step": 71440
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.68554373612918e-05,
+      "loss": 1.4857,
+      "step": 71450
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.685428943139204e-05,
+      "loss": 1.4985,
+      "step": 71460
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.68531415014923e-05,
+      "loss": 1.3581,
+      "step": 71470
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.685199357159255e-05,
+      "loss": 1.4236,
+      "step": 71480
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.68508456416928e-05,
+      "loss": 1.3528,
+      "step": 71490
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.684969771179305e-05,
+      "loss": 1.4043,
+      "step": 71500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.684854978189332e-05,
+      "loss": 1.3194,
+      "step": 71510
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.684740185199357e-05,
+      "loss": 1.4906,
+      "step": 71520
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.684625392209382e-05,
+      "loss": 1.3484,
+      "step": 71530
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.684510599219407e-05,
+      "loss": 1.4325,
+      "step": 71540
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.684395806229432e-05,
+      "loss": 1.3339,
+      "step": 71550
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.684281013239458e-05,
+      "loss": 1.3928,
+      "step": 71560
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.684166220249483e-05,
+      "loss": 1.3155,
+      "step": 71570
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.684051427259508e-05,
+      "loss": 1.3223,
+      "step": 71580
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.683936634269533e-05,
+      "loss": 1.3481,
+      "step": 71590
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.68382184127956e-05,
+      "loss": 1.4056,
+      "step": 71600
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.683707048289583e-05,
+      "loss": 1.3357,
+      "step": 71610
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.68359225529961e-05,
+      "loss": 1.4375,
+      "step": 71620
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.683477462309634e-05,
+      "loss": 1.455,
+      "step": 71630
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.683362669319659e-05,
+      "loss": 1.1858,
+      "step": 71640
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.683247876329684e-05,
+      "loss": 1.3421,
+      "step": 71650
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.68313308333971e-05,
+      "loss": 1.3568,
+      "step": 71660
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.683018290349736e-05,
+      "loss": 1.2604,
+      "step": 71670
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.68290349735976e-05,
+      "loss": 1.3457,
+      "step": 71680
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.682788704369786e-05,
+      "loss": 1.4026,
+      "step": 71690
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.68267391137981e-05,
+      "loss": 1.3811,
+      "step": 71700
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.682559118389837e-05,
+      "loss": 1.369,
+      "step": 71710
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.682444325399862e-05,
+      "loss": 1.4083,
+      "step": 71720
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.682329532409887e-05,
+      "loss": 1.318,
+      "step": 71730
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.682214739419912e-05,
+      "loss": 1.3946,
+      "step": 71740
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.682099946429938e-05,
+      "loss": 1.3339,
+      "step": 71750
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.681985153439962e-05,
+      "loss": 1.3477,
+      "step": 71760
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.681870360449988e-05,
+      "loss": 1.2314,
+      "step": 71770
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.681755567460013e-05,
+      "loss": 1.3578,
+      "step": 71780
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.681640774470038e-05,
+      "loss": 1.3798,
+      "step": 71790
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.681525981480063e-05,
+      "loss": 1.3833,
+      "step": 71800
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.68141118849009e-05,
+      "loss": 1.443,
+      "step": 71810
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.681296395500114e-05,
+      "loss": 1.3393,
+      "step": 71820
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.68118160251014e-05,
+      "loss": 1.3681,
+      "step": 71830
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.681066809520164e-05,
+      "loss": 1.3783,
+      "step": 71840
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.68095201653019e-05,
+      "loss": 1.4669,
+      "step": 71850
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.680837223540216e-05,
+      "loss": 1.4769,
+      "step": 71860
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.680722430550241e-05,
+      "loss": 1.4091,
+      "step": 71870
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.680607637560266e-05,
+      "loss": 1.4077,
+      "step": 71880
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.680492844570291e-05,
+      "loss": 1.245,
+      "step": 71890
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.680378051580317e-05,
+      "loss": 1.4261,
+      "step": 71900
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.680263258590341e-05,
+      "loss": 1.451,
+      "step": 71910
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.680148465600367e-05,
+      "loss": 1.3224,
+      "step": 71920
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.680033672610392e-05,
+      "loss": 1.5175,
+      "step": 71930
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.679918879620417e-05,
+      "loss": 1.3329,
+      "step": 71940
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.679804086630442e-05,
+      "loss": 1.3737,
+      "step": 71950
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.679689293640468e-05,
+      "loss": 1.2651,
+      "step": 71960
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.679574500650493e-05,
+      "loss": 1.3699,
+      "step": 71970
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.679459707660518e-05,
+      "loss": 1.4187,
+      "step": 71980
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.679344914670543e-05,
+      "loss": 1.2948,
+      "step": 71990
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.679230121680568e-05,
+      "loss": 1.4608,
+      "step": 72000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.679115328690595e-05,
+      "loss": 1.3328,
+      "step": 72010
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.67900053570062e-05,
+      "loss": 1.4073,
+      "step": 72020
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.678885742710645e-05,
+      "loss": 1.3318,
+      "step": 72030
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.67877094972067e-05,
+      "loss": 1.4425,
+      "step": 72040
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.678656156730696e-05,
+      "loss": 1.3848,
+      "step": 72050
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.67854136374072e-05,
+      "loss": 1.2815,
+      "step": 72060
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.678426570750746e-05,
+      "loss": 1.259,
+      "step": 72070
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.678311777760771e-05,
+      "loss": 1.4172,
+      "step": 72080
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.678196984770796e-05,
+      "loss": 1.4079,
+      "step": 72090
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.678082191780821e-05,
+      "loss": 1.3695,
+      "step": 72100
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.677967398790847e-05,
+      "loss": 1.4363,
+      "step": 72110
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.677852605800872e-05,
+      "loss": 1.3495,
+      "step": 72120
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.677737812810897e-05,
+      "loss": 1.4251,
+      "step": 72130
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.677623019820922e-05,
+      "loss": 1.4853,
+      "step": 72140
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.677508226830947e-05,
+      "loss": 1.3108,
+      "step": 72150
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.677393433840973e-05,
+      "loss": 1.4483,
+      "step": 72160
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.677278640850998e-05,
+      "loss": 1.4099,
+      "step": 72170
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.677163847861023e-05,
+      "loss": 1.38,
+      "step": 72180
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.677049054871048e-05,
+      "loss": 1.5047,
+      "step": 72190
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.676934261881075e-05,
+      "loss": 1.3343,
+      "step": 72200
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.676819468891098e-05,
+      "loss": 1.3418,
+      "step": 72210
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.676704675901125e-05,
+      "loss": 1.2924,
+      "step": 72220
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.67658988291115e-05,
+      "loss": 1.3706,
+      "step": 72230
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.676475089921175e-05,
+      "loss": 1.4243,
+      "step": 72240
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.6763602969312e-05,
+      "loss": 1.2151,
+      "step": 72250
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.676245503941226e-05,
+      "loss": 1.25,
+      "step": 72260
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.676130710951251e-05,
+      "loss": 1.3743,
+      "step": 72270
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.676015917961276e-05,
+      "loss": 1.4676,
+      "step": 72280
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.675901124971301e-05,
+      "loss": 1.5667,
+      "step": 72290
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.675786331981326e-05,
+      "loss": 1.5199,
+      "step": 72300
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.675671538991352e-05,
+      "loss": 1.4428,
+      "step": 72310
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.675556746001377e-05,
+      "loss": 1.418,
+      "step": 72320
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.675441953011402e-05,
+      "loss": 1.4243,
+      "step": 72330
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.675327160021427e-05,
+      "loss": 1.4119,
+      "step": 72340
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.675212367031454e-05,
+      "loss": 1.2354,
+      "step": 72350
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.675097574041477e-05,
+      "loss": 1.4273,
+      "step": 72360
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.674982781051504e-05,
+      "loss": 1.3051,
+      "step": 72370
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.674867988061529e-05,
+      "loss": 1.3322,
+      "step": 72380
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.674753195071554e-05,
+      "loss": 1.5266,
+      "step": 72390
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.674638402081579e-05,
+      "loss": 1.3859,
+      "step": 72400
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.674523609091605e-05,
+      "loss": 1.2854,
+      "step": 72410
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.67440881610163e-05,
+      "loss": 1.3783,
+      "step": 72420
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.674294023111655e-05,
+      "loss": 1.4196,
+      "step": 72430
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.67417923012168e-05,
+      "loss": 1.5124,
+      "step": 72440
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.674064437131705e-05,
+      "loss": 1.4629,
+      "step": 72450
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.673949644141731e-05,
+      "loss": 1.4798,
+      "step": 72460
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.673834851151756e-05,
+      "loss": 1.3736,
+      "step": 72470
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.673720058161781e-05,
+      "loss": 1.5591,
+      "step": 72480
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.673605265171806e-05,
+      "loss": 1.4699,
+      "step": 72490
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.673490472181832e-05,
+      "loss": 1.4162,
+      "step": 72500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.673375679191856e-05,
+      "loss": 1.5169,
+      "step": 72510
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.673260886201882e-05,
+      "loss": 1.3535,
+      "step": 72520
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.673146093211907e-05,
+      "loss": 1.4488,
+      "step": 72530
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.673031300221932e-05,
+      "loss": 1.3793,
+      "step": 72540
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.672916507231957e-05,
+      "loss": 1.3131,
+      "step": 72550
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.672801714241984e-05,
+      "loss": 1.4926,
+      "step": 72560
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.672686921252009e-05,
+      "loss": 1.3299,
+      "step": 72570
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.672572128262034e-05,
+      "loss": 1.444,
+      "step": 72580
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.672457335272059e-05,
+      "loss": 1.3078,
+      "step": 72590
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.672342542282084e-05,
+      "loss": 1.3468,
+      "step": 72600
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.67222774929211e-05,
+      "loss": 1.421,
+      "step": 72610
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.672112956302134e-05,
+      "loss": 1.3056,
+      "step": 72620
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.67199816331216e-05,
+      "loss": 1.3411,
+      "step": 72630
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.671883370322185e-05,
+      "loss": 1.3368,
+      "step": 72640
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.67176857733221e-05,
+      "loss": 1.2592,
+      "step": 72650
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.671653784342235e-05,
+      "loss": 1.3325,
+      "step": 72660
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.671538991352261e-05,
+      "loss": 1.5185,
+      "step": 72670
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.671424198362286e-05,
+      "loss": 1.3915,
+      "step": 72680
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.671309405372311e-05,
+      "loss": 1.4588,
+      "step": 72690
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.671194612382336e-05,
+      "loss": 1.4322,
+      "step": 72700
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.671079819392361e-05,
+      "loss": 1.4764,
+      "step": 72710
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.670965026402388e-05,
+      "loss": 1.2744,
+      "step": 72720
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.670850233412413e-05,
+      "loss": 1.489,
+      "step": 72730
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.670735440422438e-05,
+      "loss": 1.3648,
+      "step": 72740
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.670620647432463e-05,
+      "loss": 1.4853,
+      "step": 72750
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.670505854442489e-05,
+      "loss": 1.5365,
+      "step": 72760
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.670391061452512e-05,
+      "loss": 1.3964,
+      "step": 72770
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.670276268462539e-05,
+      "loss": 1.4452,
+      "step": 72780
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.670161475472564e-05,
+      "loss": 1.2784,
+      "step": 72790
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.670046682482589e-05,
+      "loss": 1.524,
+      "step": 72800
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.669931889492614e-05,
+      "loss": 1.3725,
+      "step": 72810
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.66981709650264e-05,
+      "loss": 1.4066,
+      "step": 72820
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.669702303512665e-05,
+      "loss": 1.2539,
+      "step": 72830
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.66958751052269e-05,
+      "loss": 1.3268,
+      "step": 72840
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.669472717532715e-05,
+      "loss": 1.5468,
+      "step": 72850
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.66935792454274e-05,
+      "loss": 1.5355,
+      "step": 72860
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.669243131552766e-05,
+      "loss": 1.5337,
+      "step": 72870
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.669128338562791e-05,
+      "loss": 1.3086,
+      "step": 72880
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.669013545572816e-05,
+      "loss": 1.3869,
+      "step": 72890
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.668898752582841e-05,
+      "loss": 1.348,
+      "step": 72900
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.668783959592868e-05,
+      "loss": 1.4226,
+      "step": 72910
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.668669166602891e-05,
+      "loss": 1.3288,
+      "step": 72920
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.668554373612918e-05,
+      "loss": 1.3377,
+      "step": 72930
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.668439580622943e-05,
+      "loss": 1.4185,
+      "step": 72940
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.668324787632968e-05,
+      "loss": 1.4308,
+      "step": 72950
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.668209994642993e-05,
+      "loss": 1.4839,
+      "step": 72960
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.668095201653019e-05,
+      "loss": 1.3462,
+      "step": 72970
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.667980408663044e-05,
+      "loss": 1.3802,
+      "step": 72980
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.667865615673069e-05,
+      "loss": 1.2921,
+      "step": 72990
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.667750822683094e-05,
+      "loss": 1.7213,
+      "step": 73000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.667636029693119e-05,
+      "loss": 1.4068,
+      "step": 73010
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.667521236703145e-05,
+      "loss": 1.2548,
+      "step": 73020
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.66740644371317e-05,
+      "loss": 1.369,
+      "step": 73030
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.667291650723195e-05,
+      "loss": 1.3207,
+      "step": 73040
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.66717685773322e-05,
+      "loss": 1.3661,
+      "step": 73050
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.667062064743247e-05,
+      "loss": 1.3026,
+      "step": 73060
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.66694727175327e-05,
+      "loss": 1.1872,
+      "step": 73070
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.666832478763297e-05,
+      "loss": 1.3242,
+      "step": 73080
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.666717685773321e-05,
+      "loss": 1.2606,
+      "step": 73090
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.666602892783346e-05,
+      "loss": 1.3639,
+      "step": 73100
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.666488099793371e-05,
+      "loss": 1.4242,
+      "step": 73110
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.666373306803398e-05,
+      "loss": 1.2974,
+      "step": 73120
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.666258513813423e-05,
+      "loss": 1.5151,
+      "step": 73130
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.666143720823448e-05,
+      "loss": 1.3778,
+      "step": 73140
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.666028927833473e-05,
+      "loss": 1.2445,
+      "step": 73150
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.665914134843498e-05,
+      "loss": 1.3149,
+      "step": 73160
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.665799341853524e-05,
+      "loss": 1.433,
+      "step": 73170
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.665684548863549e-05,
+      "loss": 1.4739,
+      "step": 73180
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.665569755873574e-05,
+      "loss": 1.1851,
+      "step": 73190
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.665454962883599e-05,
+      "loss": 1.3088,
+      "step": 73200
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.665340169893625e-05,
+      "loss": 1.362,
+      "step": 73210
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.665225376903649e-05,
+      "loss": 1.463,
+      "step": 73220
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.665110583913675e-05,
+      "loss": 1.3696,
+      "step": 73230
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.6649957909237e-05,
+      "loss": 1.3183,
+      "step": 73240
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.664880997933725e-05,
+      "loss": 1.3237,
+      "step": 73250
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.66476620494375e-05,
+      "loss": 1.497,
+      "step": 73260
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.664651411953777e-05,
+      "loss": 1.4752,
+      "step": 73270
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.664536618963802e-05,
+      "loss": 1.2618,
+      "step": 73280
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.664421825973827e-05,
+      "loss": 1.4947,
+      "step": 73290
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.664307032983852e-05,
+      "loss": 1.3668,
+      "step": 73300
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.664192239993877e-05,
+      "loss": 1.5463,
+      "step": 73310
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.664077447003903e-05,
+      "loss": 1.5147,
+      "step": 73320
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.663962654013928e-05,
+      "loss": 1.3626,
+      "step": 73330
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.663847861023953e-05,
+      "loss": 1.3165,
+      "step": 73340
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.663733068033978e-05,
+      "loss": 1.3491,
+      "step": 73350
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.663618275044004e-05,
+      "loss": 1.3692,
+      "step": 73360
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.663503482054028e-05,
+      "loss": 1.5683,
+      "step": 73370
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.663388689064054e-05,
+      "loss": 1.5472,
+      "step": 73380
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.663273896074079e-05,
+      "loss": 1.3454,
+      "step": 73390
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.663159103084104e-05,
+      "loss": 1.388,
+      "step": 73400
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.663044310094129e-05,
+      "loss": 1.3923,
+      "step": 73410
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.662929517104156e-05,
+      "loss": 1.4412,
+      "step": 73420
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.66281472411418e-05,
+      "loss": 1.3908,
+      "step": 73430
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.662699931124205e-05,
+      "loss": 1.496,
+      "step": 73440
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.66258513813423e-05,
+      "loss": 1.3564,
+      "step": 73450
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.662470345144255e-05,
+      "loss": 1.4063,
+      "step": 73460
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.662355552154282e-05,
+      "loss": 1.3283,
+      "step": 73470
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.662240759164307e-05,
+      "loss": 1.2976,
+      "step": 73480
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.662125966174332e-05,
+      "loss": 1.4842,
+      "step": 73490
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.662011173184357e-05,
+      "loss": 1.4885,
+      "step": 73500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.661896380194383e-05,
+      "loss": 1.394,
+      "step": 73510
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.661781587204407e-05,
+      "loss": 1.3981,
+      "step": 73520
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.661666794214433e-05,
+      "loss": 1.5552,
+      "step": 73530
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.661552001224458e-05,
+      "loss": 1.4352,
+      "step": 73540
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.661437208234483e-05,
+      "loss": 1.4902,
+      "step": 73550
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.661322415244508e-05,
+      "loss": 1.488,
+      "step": 73560
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.661207622254534e-05,
+      "loss": 1.3463,
+      "step": 73570
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.66109282926456e-05,
+      "loss": 1.4062,
+      "step": 73580
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.660978036274584e-05,
+      "loss": 1.4211,
+      "step": 73590
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.660863243284609e-05,
+      "loss": 1.4061,
+      "step": 73600
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.660748450294634e-05,
+      "loss": 1.4429,
+      "step": 73610
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.66063365730466e-05,
+      "loss": 1.4784,
+      "step": 73620
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.660518864314686e-05,
+      "loss": 1.3417,
+      "step": 73630
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.66040407132471e-05,
+      "loss": 1.4388,
+      "step": 73640
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.660289278334736e-05,
+      "loss": 1.2513,
+      "step": 73650
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.660174485344762e-05,
+      "loss": 1.3582,
+      "step": 73660
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.660059692354786e-05,
+      "loss": 1.4005,
+      "step": 73670
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.659944899364812e-05,
+      "loss": 1.3699,
+      "step": 73680
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.659830106374837e-05,
+      "loss": 1.3437,
+      "step": 73690
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.659715313384862e-05,
+      "loss": 1.4045,
+      "step": 73700
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.659600520394887e-05,
+      "loss": 1.598,
+      "step": 73710
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.659485727404913e-05,
+      "loss": 1.2871,
+      "step": 73720
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.659370934414938e-05,
+      "loss": 1.4048,
+      "step": 73730
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.659256141424963e-05,
+      "loss": 1.345,
+      "step": 73740
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.659141348434988e-05,
+      "loss": 1.3877,
+      "step": 73750
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.659026555445013e-05,
+      "loss": 1.4493,
+      "step": 73760
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.65891176245504e-05,
+      "loss": 1.5977,
+      "step": 73770
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.658796969465064e-05,
+      "loss": 1.4356,
+      "step": 73780
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.65868217647509e-05,
+      "loss": 1.225,
+      "step": 73790
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.658567383485114e-05,
+      "loss": 1.537,
+      "step": 73800
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.658452590495141e-05,
+      "loss": 1.38,
+      "step": 73810
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.658337797505164e-05,
+      "loss": 1.4517,
+      "step": 73820
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.658223004515191e-05,
+      "loss": 1.5191,
+      "step": 73830
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.658108211525216e-05,
+      "loss": 1.442,
+      "step": 73840
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.657993418535241e-05,
+      "loss": 1.3318,
+      "step": 73850
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.657878625545266e-05,
+      "loss": 1.4934,
+      "step": 73860
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.657763832555292e-05,
+      "loss": 1.473,
+      "step": 73870
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.657649039565317e-05,
+      "loss": 1.2539,
+      "step": 73880
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.657534246575342e-05,
+      "loss": 1.3042,
+      "step": 73890
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.657419453585367e-05,
+      "loss": 1.4286,
+      "step": 73900
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.657304660595392e-05,
+      "loss": 1.4406,
+      "step": 73910
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.657189867605418e-05,
+      "loss": 1.4809,
+      "step": 73920
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.657075074615443e-05,
+      "loss": 1.356,
+      "step": 73930
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.656960281625468e-05,
+      "loss": 1.5226,
+      "step": 73940
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.656845488635493e-05,
+      "loss": 1.5202,
+      "step": 73950
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.656730695645518e-05,
+      "loss": 1.3829,
+      "step": 73960
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.656615902655543e-05,
+      "loss": 1.4637,
+      "step": 73970
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.65650110966557e-05,
+      "loss": 1.4241,
+      "step": 73980
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.656386316675595e-05,
+      "loss": 1.3272,
+      "step": 73990
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.65627152368562e-05,
+      "loss": 1.3446,
+      "step": 74000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.656156730695645e-05,
+      "loss": 1.2511,
+      "step": 74010
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.65604193770567e-05,
+      "loss": 1.3453,
+      "step": 74020
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.655927144715696e-05,
+      "loss": 1.3377,
+      "step": 74030
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.655812351725721e-05,
+      "loss": 1.3545,
+      "step": 74040
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.655697558735746e-05,
+      "loss": 1.5295,
+      "step": 74050
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.655582765745771e-05,
+      "loss": 1.235,
+      "step": 74060
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.655467972755797e-05,
+      "loss": 1.4859,
+      "step": 74070
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.655353179765821e-05,
+      "loss": 1.5371,
+      "step": 74080
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.655238386775847e-05,
+      "loss": 1.4822,
+      "step": 74090
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.655123593785872e-05,
+      "loss": 1.4604,
+      "step": 74100
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.655008800795897e-05,
+      "loss": 1.4061,
+      "step": 74110
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.654894007805922e-05,
+      "loss": 1.2721,
+      "step": 74120
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.654779214815948e-05,
+      "loss": 1.4045,
+      "step": 74130
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.654664421825973e-05,
+      "loss": 1.2581,
+      "step": 74140
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.654549628835998e-05,
+      "loss": 1.3275,
+      "step": 74150
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.654434835846023e-05,
+      "loss": 1.3976,
+      "step": 74160
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.654320042856048e-05,
+      "loss": 1.5946,
+      "step": 74170
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.654205249866075e-05,
+      "loss": 1.4979,
+      "step": 74180
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.6540904568761e-05,
+      "loss": 1.2964,
+      "step": 74190
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.653975663886125e-05,
+      "loss": 1.3729,
+      "step": 74200
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.65386087089615e-05,
+      "loss": 1.4631,
+      "step": 74210
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.653746077906176e-05,
+      "loss": 1.2858,
+      "step": 74220
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.6536312849162e-05,
+      "loss": 1.4274,
+      "step": 74230
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.653516491926226e-05,
+      "loss": 1.4097,
+      "step": 74240
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.653401698936251e-05,
+      "loss": 1.3137,
+      "step": 74250
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.653286905946276e-05,
+      "loss": 1.4822,
+      "step": 74260
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.653172112956301e-05,
+      "loss": 1.6589,
+      "step": 74270
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.653057319966327e-05,
+      "loss": 1.391,
+      "step": 74280
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.652942526976352e-05,
+      "loss": 1.3044,
+      "step": 74290
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.652827733986377e-05,
+      "loss": 1.4292,
+      "step": 74300
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.652712940996402e-05,
+      "loss": 1.3908,
+      "step": 74310
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.652598148006427e-05,
+      "loss": 1.3819,
+      "step": 74320
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.652483355016454e-05,
+      "loss": 1.4404,
+      "step": 74330
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.652368562026479e-05,
+      "loss": 1.4586,
+      "step": 74340
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.652253769036504e-05,
+      "loss": 1.4553,
+      "step": 74350
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.652138976046529e-05,
+      "loss": 1.485,
+      "step": 74360
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.652024183056555e-05,
+      "loss": 1.4297,
+      "step": 74370
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.651909390066578e-05,
+      "loss": 1.3123,
+      "step": 74380
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.651794597076605e-05,
+      "loss": 1.5652,
+      "step": 74390
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.65167980408663e-05,
+      "loss": 1.4616,
+      "step": 74400
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.651565011096655e-05,
+      "loss": 1.3978,
+      "step": 74410
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.65145021810668e-05,
+      "loss": 1.3838,
+      "step": 74420
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.651335425116706e-05,
+      "loss": 1.2895,
+      "step": 74430
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.651220632126731e-05,
+      "loss": 1.2619,
+      "step": 74440
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.651105839136756e-05,
+      "loss": 1.4636,
+      "step": 74450
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.650991046146781e-05,
+      "loss": 1.4319,
+      "step": 74460
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.650876253156806e-05,
+      "loss": 1.3262,
+      "step": 74470
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.650761460166832e-05,
+      "loss": 1.4436,
+      "step": 74480
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.650646667176857e-05,
+      "loss": 1.4004,
+      "step": 74490
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.650531874186882e-05,
+      "loss": 1.4977,
+      "step": 74500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.650417081196907e-05,
+      "loss": 1.3152,
+      "step": 74510
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.650302288206934e-05,
+      "loss": 1.5127,
+      "step": 74520
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.650187495216957e-05,
+      "loss": 1.3515,
+      "step": 74530
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.650072702226984e-05,
+      "loss": 1.2891,
+      "step": 74540
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.649957909237009e-05,
+      "loss": 1.225,
+      "step": 74550
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.649843116247034e-05,
+      "loss": 1.2017,
+      "step": 74560
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.649728323257059e-05,
+      "loss": 1.5477,
+      "step": 74570
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.649613530267085e-05,
+      "loss": 1.3458,
+      "step": 74580
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.64949873727711e-05,
+      "loss": 1.4125,
+      "step": 74590
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.649383944287135e-05,
+      "loss": 1.3148,
+      "step": 74600
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.64926915129716e-05,
+      "loss": 1.3666,
+      "step": 74610
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.649154358307185e-05,
+      "loss": 1.5163,
+      "step": 74620
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.649039565317211e-05,
+      "loss": 1.4923,
+      "step": 74630
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.648924772327236e-05,
+      "loss": 1.541,
+      "step": 74640
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.648809979337261e-05,
+      "loss": 1.3449,
+      "step": 74650
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.648695186347286e-05,
+      "loss": 1.4264,
+      "step": 74660
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.648580393357313e-05,
+      "loss": 1.3854,
+      "step": 74670
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.648465600367336e-05,
+      "loss": 1.4995,
+      "step": 74680
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.648350807377363e-05,
+      "loss": 1.2504,
+      "step": 74690
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.648236014387388e-05,
+      "loss": 1.4529,
+      "step": 74700
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.648121221397413e-05,
+      "loss": 1.3656,
+      "step": 74710
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.648006428407437e-05,
+      "loss": 1.3945,
+      "step": 74720
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.647891635417464e-05,
+      "loss": 1.3072,
+      "step": 74730
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.647776842427489e-05,
+      "loss": 1.2776,
+      "step": 74740
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.647662049437514e-05,
+      "loss": 1.3534,
+      "step": 74750
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.647547256447539e-05,
+      "loss": 1.4693,
+      "step": 74760
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.647432463457564e-05,
+      "loss": 1.5008,
+      "step": 74770
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.64731767046759e-05,
+      "loss": 1.4697,
+      "step": 74780
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.647202877477615e-05,
+      "loss": 1.209,
+      "step": 74790
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.64708808448764e-05,
+      "loss": 1.3616,
+      "step": 74800
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.646973291497665e-05,
+      "loss": 1.2967,
+      "step": 74810
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.646858498507691e-05,
+      "loss": 1.2466,
+      "step": 74820
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.646743705517715e-05,
+      "loss": 1.3415,
+      "step": 74830
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.646628912527741e-05,
+      "loss": 1.5142,
+      "step": 74840
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.646514119537766e-05,
+      "loss": 1.3532,
+      "step": 74850
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.646399326547791e-05,
+      "loss": 1.5125,
+      "step": 74860
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.646284533557816e-05,
+      "loss": 1.4115,
+      "step": 74870
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.646169740567843e-05,
+      "loss": 1.3766,
+      "step": 74880
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.646054947577868e-05,
+      "loss": 1.3855,
+      "step": 74890
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.645940154587893e-05,
+      "loss": 1.2967,
+      "step": 74900
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.645825361597918e-05,
+      "loss": 1.3516,
+      "step": 74910
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.645710568607943e-05,
+      "loss": 1.3381,
+      "step": 74920
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.645595775617969e-05,
+      "loss": 1.3589,
+      "step": 74930
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.645480982627994e-05,
+      "loss": 1.5386,
+      "step": 74940
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.645366189638019e-05,
+      "loss": 1.3706,
+      "step": 74950
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.645251396648044e-05,
+      "loss": 1.3501,
+      "step": 74960
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.64513660365807e-05,
+      "loss": 1.2597,
+      "step": 74970
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.645021810668094e-05,
+      "loss": 1.2257,
+      "step": 74980
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.64490701767812e-05,
+      "loss": 1.3791,
+      "step": 74990
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.644792224688145e-05,
+      "loss": 1.3529,
+      "step": 75000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.64467743169817e-05,
+      "loss": 1.2561,
+      "step": 75010
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.644562638708195e-05,
+      "loss": 1.4043,
+      "step": 75020
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.644447845718222e-05,
+      "loss": 1.32,
+      "step": 75030
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.644333052728247e-05,
+      "loss": 1.3527,
+      "step": 75040
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.644218259738271e-05,
+      "loss": 1.3745,
+      "step": 75050
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.644103466748296e-05,
+      "loss": 1.271,
+      "step": 75060
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.643988673758321e-05,
+      "loss": 1.3952,
+      "step": 75070
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.643873880768348e-05,
+      "loss": 1.4285,
+      "step": 75080
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.643759087778373e-05,
+      "loss": 1.4037,
+      "step": 75090
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.643644294788398e-05,
+      "loss": 1.3689,
+      "step": 75100
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.643529501798423e-05,
+      "loss": 1.4514,
+      "step": 75110
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.643414708808449e-05,
+      "loss": 1.4397,
+      "step": 75120
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.643299915818473e-05,
+      "loss": 1.5344,
+      "step": 75130
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.643185122828499e-05,
+      "loss": 1.3424,
+      "step": 75140
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.643070329838524e-05,
+      "loss": 1.3841,
+      "step": 75150
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.642955536848549e-05,
+      "loss": 1.2022,
+      "step": 75160
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.642840743858574e-05,
+      "loss": 1.5353,
+      "step": 75170
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.6427259508686e-05,
+      "loss": 1.4418,
+      "step": 75180
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.642611157878625e-05,
+      "loss": 1.3881,
+      "step": 75190
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.64249636488865e-05,
+      "loss": 1.2573,
+      "step": 75200
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.642381571898675e-05,
+      "loss": 1.2303,
+      "step": 75210
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.6422667789087e-05,
+      "loss": 1.4055,
+      "step": 75220
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.642151985918727e-05,
+      "loss": 1.4152,
+      "step": 75230
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.642037192928752e-05,
+      "loss": 1.4034,
+      "step": 75240
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.641922399938777e-05,
+      "loss": 1.3059,
+      "step": 75250
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.641807606948802e-05,
+      "loss": 1.477,
+      "step": 75260
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.641692813958828e-05,
+      "loss": 1.5507,
+      "step": 75270
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.641578020968852e-05,
+      "loss": 1.3691,
+      "step": 75280
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.641463227978878e-05,
+      "loss": 1.3099,
+      "step": 75290
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.641348434988903e-05,
+      "loss": 1.4133,
+      "step": 75300
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.641233641998928e-05,
+      "loss": 1.3823,
+      "step": 75310
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.641118849008953e-05,
+      "loss": 1.4442,
+      "step": 75320
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.641004056018978e-05,
+      "loss": 1.3829,
+      "step": 75330
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.640889263029004e-05,
+      "loss": 1.2828,
+      "step": 75340
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.640774470039029e-05,
+      "loss": 1.471,
+      "step": 75350
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.640659677049054e-05,
+      "loss": 1.384,
+      "step": 75360
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.640544884059079e-05,
+      "loss": 1.4409,
+      "step": 75370
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.640430091069105e-05,
+      "loss": 1.267,
+      "step": 75380
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.640315298079129e-05,
+      "loss": 1.3828,
+      "step": 75390
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.640200505089155e-05,
+      "loss": 1.3558,
+      "step": 75400
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.64008571209918e-05,
+      "loss": 1.331,
+      "step": 75410
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.639970919109205e-05,
+      "loss": 1.505,
+      "step": 75420
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.63985612611923e-05,
+      "loss": 1.554,
+      "step": 75430
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.639741333129257e-05,
+      "loss": 1.4041,
+      "step": 75440
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.639626540139282e-05,
+      "loss": 1.3954,
+      "step": 75450
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.639511747149307e-05,
+      "loss": 1.5829,
+      "step": 75460
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.639396954159332e-05,
+      "loss": 1.46,
+      "step": 75470
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.639282161169357e-05,
+      "loss": 1.2267,
+      "step": 75480
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.639167368179383e-05,
+      "loss": 1.5136,
+      "step": 75490
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.639052575189408e-05,
+      "loss": 1.499,
+      "step": 75500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.638937782199433e-05,
+      "loss": 1.3992,
+      "step": 75510
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.638822989209458e-05,
+      "loss": 1.4591,
+      "step": 75520
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.638708196219484e-05,
+      "loss": 1.386,
+      "step": 75530
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.638593403229508e-05,
+      "loss": 1.2891,
+      "step": 75540
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.638478610239534e-05,
+      "loss": 1.3051,
+      "step": 75550
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.638363817249559e-05,
+      "loss": 1.5295,
+      "step": 75560
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.638249024259584e-05,
+      "loss": 1.4862,
+      "step": 75570
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.638134231269609e-05,
+      "loss": 1.3719,
+      "step": 75580
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.638019438279636e-05,
+      "loss": 1.3314,
+      "step": 75590
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.63790464528966e-05,
+      "loss": 1.2378,
+      "step": 75600
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.637789852299686e-05,
+      "loss": 1.2489,
+      "step": 75610
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.63767505930971e-05,
+      "loss": 1.4816,
+      "step": 75620
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.637560266319736e-05,
+      "loss": 1.4617,
+      "step": 75630
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.637445473329762e-05,
+      "loss": 1.2777,
+      "step": 75640
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.637330680339787e-05,
+      "loss": 1.2814,
+      "step": 75650
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.637215887349812e-05,
+      "loss": 1.5214,
+      "step": 75660
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.637101094359837e-05,
+      "loss": 1.4618,
+      "step": 75670
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.636986301369863e-05,
+      "loss": 1.2578,
+      "step": 75680
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.636871508379887e-05,
+      "loss": 1.3256,
+      "step": 75690
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.636756715389913e-05,
+      "loss": 1.3215,
+      "step": 75700
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.636641922399938e-05,
+      "loss": 1.4748,
+      "step": 75710
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.636527129409963e-05,
+      "loss": 1.4584,
+      "step": 75720
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.636412336419988e-05,
+      "loss": 1.255,
+      "step": 75730
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.636297543430014e-05,
+      "loss": 1.4545,
+      "step": 75740
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.63618275044004e-05,
+      "loss": 1.3685,
+      "step": 75750
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.636067957450064e-05,
+      "loss": 1.5102,
+      "step": 75760
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.63595316446009e-05,
+      "loss": 1.4325,
+      "step": 75770
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.635838371470114e-05,
+      "loss": 1.2576,
+      "step": 75780
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.635723578480141e-05,
+      "loss": 1.2846,
+      "step": 75790
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.635608785490166e-05,
+      "loss": 1.5271,
+      "step": 75800
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.635493992500191e-05,
+      "loss": 1.3549,
+      "step": 75810
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.635379199510216e-05,
+      "loss": 1.4089,
+      "step": 75820
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.635264406520242e-05,
+      "loss": 1.1267,
+      "step": 75830
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.635149613530266e-05,
+      "loss": 1.3276,
+      "step": 75840
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.635034820540292e-05,
+      "loss": 1.3416,
+      "step": 75850
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.634920027550317e-05,
+      "loss": 1.2684,
+      "step": 75860
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.634805234560342e-05,
+      "loss": 1.4785,
+      "step": 75870
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.634690441570367e-05,
+      "loss": 1.3077,
+      "step": 75880
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.634575648580393e-05,
+      "loss": 1.3134,
+      "step": 75890
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.634460855590418e-05,
+      "loss": 1.3526,
+      "step": 75900
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.634346062600443e-05,
+      "loss": 1.4048,
+      "step": 75910
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.634231269610468e-05,
+      "loss": 1.307,
+      "step": 75920
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.634116476620493e-05,
+      "loss": 1.3359,
+      "step": 75930
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.63400168363052e-05,
+      "loss": 1.314,
+      "step": 75940
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.633886890640545e-05,
+      "loss": 1.3961,
+      "step": 75950
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.63377209765057e-05,
+      "loss": 1.3969,
+      "step": 75960
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.633657304660595e-05,
+      "loss": 1.3521,
+      "step": 75970
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.633542511670621e-05,
+      "loss": 1.3188,
+      "step": 75980
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.633427718680645e-05,
+      "loss": 1.5064,
+      "step": 75990
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.633312925690671e-05,
+      "loss": 1.5107,
+      "step": 76000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.633198132700696e-05,
+      "loss": 1.4018,
+      "step": 76010
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.633083339710721e-05,
+      "loss": 1.4105,
+      "step": 76020
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.632968546720746e-05,
+      "loss": 1.3439,
+      "step": 76030
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.632853753730772e-05,
+      "loss": 1.4932,
+      "step": 76040
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.632738960740797e-05,
+      "loss": 1.4261,
+      "step": 76050
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.632624167750822e-05,
+      "loss": 1.2462,
+      "step": 76060
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.632509374760847e-05,
+      "loss": 1.3411,
+      "step": 76070
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.632394581770872e-05,
+      "loss": 1.2769,
+      "step": 76080
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.632279788780898e-05,
+      "loss": 1.3833,
+      "step": 76090
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.632164995790923e-05,
+      "loss": 1.6086,
+      "step": 76100
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.632050202800948e-05,
+      "loss": 1.358,
+      "step": 76110
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.631935409810973e-05,
+      "loss": 1.3374,
+      "step": 76120
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.631820616821e-05,
+      "loss": 1.4984,
+      "step": 76130
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.631705823831023e-05,
+      "loss": 1.3512,
+      "step": 76140
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.63159103084105e-05,
+      "loss": 1.3885,
+      "step": 76150
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.631476237851075e-05,
+      "loss": 1.449,
+      "step": 76160
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.6313614448611e-05,
+      "loss": 1.4925,
+      "step": 76170
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.631246651871125e-05,
+      "loss": 1.3105,
+      "step": 76180
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.631131858881151e-05,
+      "loss": 1.438,
+      "step": 76190
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.631017065891176e-05,
+      "loss": 1.3522,
+      "step": 76200
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.630902272901201e-05,
+      "loss": 1.4014,
+      "step": 76210
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.630787479911226e-05,
+      "loss": 1.1467,
+      "step": 76220
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.630672686921251e-05,
+      "loss": 1.3651,
+      "step": 76230
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.630557893931277e-05,
+      "loss": 1.4243,
+      "step": 76240
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.630443100941302e-05,
+      "loss": 1.2866,
+      "step": 76250
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.630328307951327e-05,
+      "loss": 1.3072,
+      "step": 76260
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.630213514961352e-05,
+      "loss": 1.4437,
+      "step": 76270
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.630098721971379e-05,
+      "loss": 1.3497,
+      "step": 76280
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.629983928981402e-05,
+      "loss": 1.5991,
+      "step": 76290
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.629869135991429e-05,
+      "loss": 1.4304,
+      "step": 76300
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.629754343001454e-05,
+      "loss": 1.3927,
+      "step": 76310
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.629639550011479e-05,
+      "loss": 1.5176,
+      "step": 76320
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.629524757021504e-05,
+      "loss": 1.2963,
+      "step": 76330
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.62940996403153e-05,
+      "loss": 1.3294,
+      "step": 76340
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.629295171041555e-05,
+      "loss": 1.6118,
+      "step": 76350
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.62918037805158e-05,
+      "loss": 1.4471,
+      "step": 76360
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.629065585061605e-05,
+      "loss": 1.4171,
+      "step": 76370
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.62895079207163e-05,
+      "loss": 1.3721,
+      "step": 76380
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.628835999081656e-05,
+      "loss": 1.2959,
+      "step": 76390
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.628721206091681e-05,
+      "loss": 1.3224,
+      "step": 76400
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.628606413101706e-05,
+      "loss": 1.2832,
+      "step": 76410
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.628491620111731e-05,
+      "loss": 1.3266,
+      "step": 76420
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.628376827121757e-05,
+      "loss": 1.2632,
+      "step": 76430
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.628262034131781e-05,
+      "loss": 1.4079,
+      "step": 76440
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.628147241141807e-05,
+      "loss": 1.4024,
+      "step": 76450
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.628032448151832e-05,
+      "loss": 1.5838,
+      "step": 76460
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.627917655161857e-05,
+      "loss": 1.3042,
+      "step": 76470
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.627802862171882e-05,
+      "loss": 1.3607,
+      "step": 76480
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.627688069181909e-05,
+      "loss": 1.4888,
+      "step": 76490
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.627573276191934e-05,
+      "loss": 1.3698,
+      "step": 76500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.627458483201959e-05,
+      "loss": 1.2669,
+      "step": 76510
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.627343690211984e-05,
+      "loss": 1.4531,
+      "step": 76520
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.627228897222009e-05,
+      "loss": 1.3643,
+      "step": 76530
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.627114104232035e-05,
+      "loss": 1.3862,
+      "step": 76540
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.62699931124206e-05,
+      "loss": 1.2954,
+      "step": 76550
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.626884518252085e-05,
+      "loss": 1.381,
+      "step": 76560
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.62676972526211e-05,
+      "loss": 1.3998,
+      "step": 76570
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.626654932272136e-05,
+      "loss": 1.4737,
+      "step": 76580
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.62654013928216e-05,
+      "loss": 1.5994,
+      "step": 76590
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.626425346292186e-05,
+      "loss": 1.4098,
+      "step": 76600
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.626310553302211e-05,
+      "loss": 1.371,
+      "step": 76610
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.626195760312236e-05,
+      "loss": 1.3471,
+      "step": 76620
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.626080967322261e-05,
+      "loss": 1.444,
+      "step": 76630
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.625966174332288e-05,
+      "loss": 1.385,
+      "step": 76640
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.625851381342313e-05,
+      "loss": 1.3696,
+      "step": 76650
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.625736588352338e-05,
+      "loss": 1.5366,
+      "step": 76660
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.625621795362363e-05,
+      "loss": 1.2542,
+      "step": 76670
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.625507002372387e-05,
+      "loss": 1.2072,
+      "step": 76680
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.625392209382414e-05,
+      "loss": 1.5489,
+      "step": 76690
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.625277416392437e-05,
+      "loss": 1.3929,
+      "step": 76700
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.625162623402464e-05,
+      "loss": 1.2459,
+      "step": 76710
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.625047830412489e-05,
+      "loss": 1.5796,
+      "step": 76720
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.624933037422514e-05,
+      "loss": 1.3549,
+      "step": 76730
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.624818244432539e-05,
+      "loss": 1.2867,
+      "step": 76740
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.624703451442565e-05,
+      "loss": 1.3825,
+      "step": 76750
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.62458865845259e-05,
+      "loss": 1.3211,
+      "step": 76760
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.624473865462615e-05,
+      "loss": 1.406,
+      "step": 76770
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.62435907247264e-05,
+      "loss": 1.3259,
+      "step": 76780
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.624244279482665e-05,
+      "loss": 1.3651,
+      "step": 76790
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.624129486492691e-05,
+      "loss": 1.4175,
+      "step": 76800
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.624014693502716e-05,
+      "loss": 1.2772,
+      "step": 76810
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.623899900512741e-05,
+      "loss": 1.3245,
+      "step": 76820
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.623785107522766e-05,
+      "loss": 1.3604,
+      "step": 76830
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.623670314532793e-05,
+      "loss": 1.3927,
+      "step": 76840
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.623555521542816e-05,
+      "loss": 1.4077,
+      "step": 76850
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.623440728552843e-05,
+      "loss": 1.401,
+      "step": 76860
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.623325935562868e-05,
+      "loss": 1.6682,
+      "step": 76870
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.623211142572893e-05,
+      "loss": 1.4803,
+      "step": 76880
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.623096349582918e-05,
+      "loss": 1.2361,
+      "step": 76890
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.622981556592944e-05,
+      "loss": 1.4049,
+      "step": 76900
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.622866763602969e-05,
+      "loss": 1.3655,
+      "step": 76910
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.622751970612994e-05,
+      "loss": 1.4893,
+      "step": 76920
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.622637177623019e-05,
+      "loss": 1.242,
+      "step": 76930
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.622522384633044e-05,
+      "loss": 1.3903,
+      "step": 76940
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.62240759164307e-05,
+      "loss": 1.387,
+      "step": 76950
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.622292798653095e-05,
+      "loss": 1.2018,
+      "step": 76960
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.62217800566312e-05,
+      "loss": 1.3442,
+      "step": 76970
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.622063212673145e-05,
+      "loss": 1.3221,
+      "step": 76980
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.621948419683172e-05,
+      "loss": 1.413,
+      "step": 76990
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.621833626693195e-05,
+      "loss": 1.5033,
+      "step": 77000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.621718833703221e-05,
+      "loss": 1.3705,
+      "step": 77010
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.621604040713246e-05,
+      "loss": 1.4932,
+      "step": 77020
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.621489247723271e-05,
+      "loss": 1.3146,
+      "step": 77030
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.621374454733296e-05,
+      "loss": 1.2529,
+      "step": 77040
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.621259661743323e-05,
+      "loss": 1.3729,
+      "step": 77050
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.621144868753348e-05,
+      "loss": 1.515,
+      "step": 77060
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.621030075763373e-05,
+      "loss": 1.4653,
+      "step": 77070
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.620915282773398e-05,
+      "loss": 1.3563,
+      "step": 77080
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.620800489783423e-05,
+      "loss": 1.4375,
+      "step": 77090
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.620685696793449e-05,
+      "loss": 1.4833,
+      "step": 77100
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.620570903803474e-05,
+      "loss": 1.4455,
+      "step": 77110
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.620456110813499e-05,
+      "loss": 1.314,
+      "step": 77120
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.620341317823524e-05,
+      "loss": 1.2853,
+      "step": 77130
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.62022652483355e-05,
+      "loss": 1.4642,
+      "step": 77140
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.620111731843574e-05,
+      "loss": 1.2269,
+      "step": 77150
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.6199969388536e-05,
+      "loss": 1.3823,
+      "step": 77160
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.619882145863625e-05,
+      "loss": 1.283,
+      "step": 77170
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.61976735287365e-05,
+      "loss": 1.527,
+      "step": 77180
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.619652559883675e-05,
+      "loss": 1.4831,
+      "step": 77190
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.619537766893702e-05,
+      "loss": 1.3594,
+      "step": 77200
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.619422973903727e-05,
+      "loss": 1.3354,
+      "step": 77210
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.619308180913752e-05,
+      "loss": 1.3796,
+      "step": 77220
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.619193387923777e-05,
+      "loss": 1.2869,
+      "step": 77230
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.619078594933802e-05,
+      "loss": 1.4425,
+      "step": 77240
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.618963801943828e-05,
+      "loss": 1.3098,
+      "step": 77250
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.618849008953853e-05,
+      "loss": 1.4618,
+      "step": 77260
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.618734215963878e-05,
+      "loss": 1.5037,
+      "step": 77270
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.618619422973903e-05,
+      "loss": 1.3798,
+      "step": 77280
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.618504629983929e-05,
+      "loss": 1.2595,
+      "step": 77290
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.618389836993953e-05,
+      "loss": 1.3628,
+      "step": 77300
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.618275044003979e-05,
+      "loss": 1.2834,
+      "step": 77310
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.618160251014004e-05,
+      "loss": 1.2387,
+      "step": 77320
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.618045458024029e-05,
+      "loss": 1.4025,
+      "step": 77330
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.617930665034054e-05,
+      "loss": 1.3451,
+      "step": 77340
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.61781587204408e-05,
+      "loss": 1.3818,
+      "step": 77350
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.617701079054105e-05,
+      "loss": 1.1741,
+      "step": 77360
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.61758628606413e-05,
+      "loss": 1.3401,
+      "step": 77370
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.617471493074155e-05,
+      "loss": 1.4927,
+      "step": 77380
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.61735670008418e-05,
+      "loss": 1.3813,
+      "step": 77390
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.617241907094207e-05,
+      "loss": 1.4434,
+      "step": 77400
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.617127114104232e-05,
+      "loss": 1.2437,
+      "step": 77410
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.617012321114257e-05,
+      "loss": 1.3484,
+      "step": 77420
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.616897528124282e-05,
+      "loss": 1.4491,
+      "step": 77430
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.616782735134308e-05,
+      "loss": 1.2896,
+      "step": 77440
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.616667942144332e-05,
+      "loss": 1.4233,
+      "step": 77450
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.616553149154358e-05,
+      "loss": 1.5176,
+      "step": 77460
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.616438356164383e-05,
+      "loss": 1.4436,
+      "step": 77470
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.616323563174408e-05,
+      "loss": 1.4322,
+      "step": 77480
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.616208770184433e-05,
+      "loss": 1.3849,
+      "step": 77490
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.61609397719446e-05,
+      "loss": 1.4586,
+      "step": 77500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.615979184204484e-05,
+      "loss": 1.2159,
+      "step": 77510
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.615864391214509e-05,
+      "loss": 1.455,
+      "step": 77520
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.615749598224534e-05,
+      "loss": 1.4738,
+      "step": 77530
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.615634805234559e-05,
+      "loss": 1.4222,
+      "step": 77540
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.615520012244586e-05,
+      "loss": 1.3611,
+      "step": 77550
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.61540521925461e-05,
+      "loss": 1.6237,
+      "step": 77560
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.615290426264636e-05,
+      "loss": 1.374,
+      "step": 77570
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.61517563327466e-05,
+      "loss": 1.4718,
+      "step": 77580
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.615060840284687e-05,
+      "loss": 1.3533,
+      "step": 77590
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.61494604729471e-05,
+      "loss": 1.412,
+      "step": 77600
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.614831254304737e-05,
+      "loss": 1.3102,
+      "step": 77610
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.614716461314762e-05,
+      "loss": 1.2644,
+      "step": 77620
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.614601668324787e-05,
+      "loss": 1.1486,
+      "step": 77630
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.614486875334812e-05,
+      "loss": 1.4479,
+      "step": 77640
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.614372082344838e-05,
+      "loss": 1.4223,
+      "step": 77650
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.614257289354863e-05,
+      "loss": 1.2973,
+      "step": 77660
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.614142496364888e-05,
+      "loss": 1.5255,
+      "step": 77670
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.614027703374913e-05,
+      "loss": 1.2208,
+      "step": 77680
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.613912910384938e-05,
+      "loss": 1.4276,
+      "step": 77690
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.613798117394964e-05,
+      "loss": 1.311,
+      "step": 77700
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.61368332440499e-05,
+      "loss": 1.3575,
+      "step": 77710
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.613568531415014e-05,
+      "loss": 1.3854,
+      "step": 77720
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.61345373842504e-05,
+      "loss": 1.2999,
+      "step": 77730
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.613338945435066e-05,
+      "loss": 1.3905,
+      "step": 77740
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.61322415244509e-05,
+      "loss": 1.1427,
+      "step": 77750
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.613109359455116e-05,
+      "loss": 1.3652,
+      "step": 77760
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.612994566465141e-05,
+      "loss": 1.4749,
+      "step": 77770
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.612879773475166e-05,
+      "loss": 1.3463,
+      "step": 77780
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.612764980485191e-05,
+      "loss": 1.4833,
+      "step": 77790
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.612650187495217e-05,
+      "loss": 1.34,
+      "step": 77800
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.612535394505242e-05,
+      "loss": 1.394,
+      "step": 77810
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.612420601515267e-05,
+      "loss": 1.3719,
+      "step": 77820
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.612305808525292e-05,
+      "loss": 1.5048,
+      "step": 77830
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.612191015535317e-05,
+      "loss": 1.2359,
+      "step": 77840
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.612076222545343e-05,
+      "loss": 1.3719,
+      "step": 77850
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.611961429555368e-05,
+      "loss": 1.2919,
+      "step": 77860
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.611846636565393e-05,
+      "loss": 1.2566,
+      "step": 77870
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.611731843575418e-05,
+      "loss": 1.2385,
+      "step": 77880
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.611617050585445e-05,
+      "loss": 1.4705,
+      "step": 77890
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.611502257595468e-05,
+      "loss": 1.4434,
+      "step": 77900
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.611387464605495e-05,
+      "loss": 1.4469,
+      "step": 77910
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.61127267161552e-05,
+      "loss": 1.3212,
+      "step": 77920
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.611157878625545e-05,
+      "loss": 1.3568,
+      "step": 77930
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.61104308563557e-05,
+      "loss": 1.343,
+      "step": 77940
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.610928292645596e-05,
+      "loss": 1.4393,
+      "step": 77950
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.610813499655621e-05,
+      "loss": 1.3358,
+      "step": 77960
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.610698706665646e-05,
+      "loss": 1.3477,
+      "step": 77970
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.610583913675671e-05,
+      "loss": 1.3996,
+      "step": 77980
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.610469120685696e-05,
+      "loss": 1.3789,
+      "step": 77990
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.610354327695722e-05,
+      "loss": 1.2045,
+      "step": 78000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.610239534705747e-05,
+      "loss": 1.568,
+      "step": 78010
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.610124741715772e-05,
+      "loss": 1.3697,
+      "step": 78020
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.610009948725797e-05,
+      "loss": 1.3903,
+      "step": 78030
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.609895155735823e-05,
+      "loss": 1.3479,
+      "step": 78040
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.609780362745847e-05,
+      "loss": 1.5298,
+      "step": 78050
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.609665569755873e-05,
+      "loss": 1.3588,
+      "step": 78060
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.609550776765898e-05,
+      "loss": 1.2992,
+      "step": 78070
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.609435983775923e-05,
+      "loss": 1.4625,
+      "step": 78080
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.609321190785948e-05,
+      "loss": 1.3328,
+      "step": 78090
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.609206397795973e-05,
+      "loss": 1.1779,
+      "step": 78100
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.609091604806e-05,
+      "loss": 1.3443,
+      "step": 78110
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.608976811816025e-05,
+      "loss": 1.2686,
+      "step": 78120
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.60886201882605e-05,
+      "loss": 1.3185,
+      "step": 78130
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.608747225836075e-05,
+      "loss": 1.3158,
+      "step": 78140
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.608632432846101e-05,
+      "loss": 1.3819,
+      "step": 78150
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.608517639856125e-05,
+      "loss": 1.5197,
+      "step": 78160
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.608402846866151e-05,
+      "loss": 1.3481,
+      "step": 78170
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.608288053876176e-05,
+      "loss": 1.3354,
+      "step": 78180
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.608173260886201e-05,
+      "loss": 1.265,
+      "step": 78190
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.608058467896226e-05,
+      "loss": 1.3531,
+      "step": 78200
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.607943674906252e-05,
+      "loss": 1.3604,
+      "step": 78210
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.607828881916277e-05,
+      "loss": 1.2656,
+      "step": 78220
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.607714088926302e-05,
+      "loss": 1.4265,
+      "step": 78230
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.607599295936327e-05,
+      "loss": 1.4075,
+      "step": 78240
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.607484502946352e-05,
+      "loss": 1.4092,
+      "step": 78250
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.607369709956379e-05,
+      "loss": 1.3831,
+      "step": 78260
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.607254916966404e-05,
+      "loss": 1.4131,
+      "step": 78270
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.607140123976429e-05,
+      "loss": 1.3307,
+      "step": 78280
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.607025330986454e-05,
+      "loss": 1.3915,
+      "step": 78290
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.60691053799648e-05,
+      "loss": 1.348,
+      "step": 78300
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.606795745006503e-05,
+      "loss": 1.488,
+      "step": 78310
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.60668095201653e-05,
+      "loss": 1.3067,
+      "step": 78320
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.606566159026555e-05,
+      "loss": 1.404,
+      "step": 78330
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.60645136603658e-05,
+      "loss": 1.3138,
+      "step": 78340
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.606336573046605e-05,
+      "loss": 1.4476,
+      "step": 78350
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.606221780056631e-05,
+      "loss": 1.4065,
+      "step": 78360
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.606106987066656e-05,
+      "loss": 1.3935,
+      "step": 78370
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.605992194076681e-05,
+      "loss": 1.3942,
+      "step": 78380
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.605877401086706e-05,
+      "loss": 1.3894,
+      "step": 78390
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.605762608096731e-05,
+      "loss": 1.4105,
+      "step": 78400
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.605647815106757e-05,
+      "loss": 1.3424,
+      "step": 78410
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.605533022116782e-05,
+      "loss": 1.2276,
+      "step": 78420
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.605418229126807e-05,
+      "loss": 1.3089,
+      "step": 78430
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.605303436136832e-05,
+      "loss": 1.1661,
+      "step": 78440
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.605188643146859e-05,
+      "loss": 1.288,
+      "step": 78450
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.605073850156882e-05,
+      "loss": 1.4676,
+      "step": 78460
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.604959057166909e-05,
+      "loss": 1.3632,
+      "step": 78470
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.604844264176934e-05,
+      "loss": 1.4205,
+      "step": 78480
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.604729471186959e-05,
+      "loss": 1.4626,
+      "step": 78490
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.604614678196984e-05,
+      "loss": 1.3825,
+      "step": 78500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.60449988520701e-05,
+      "loss": 1.4761,
+      "step": 78510
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.604385092217035e-05,
+      "loss": 1.3065,
+      "step": 78520
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.60427029922706e-05,
+      "loss": 1.3555,
+      "step": 78530
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.604155506237085e-05,
+      "loss": 1.4042,
+      "step": 78540
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.60404071324711e-05,
+      "loss": 1.4831,
+      "step": 78550
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.603925920257136e-05,
+      "loss": 1.4393,
+      "step": 78560
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.603811127267161e-05,
+      "loss": 1.316,
+      "step": 78570
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.603696334277186e-05,
+      "loss": 1.2471,
+      "step": 78580
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.603581541287211e-05,
+      "loss": 1.2815,
+      "step": 78590
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.603466748297238e-05,
+      "loss": 1.3471,
+      "step": 78600
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.603351955307261e-05,
+      "loss": 1.3716,
+      "step": 78610
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.603237162317288e-05,
+      "loss": 1.6593,
+      "step": 78620
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.603122369327312e-05,
+      "loss": 1.4512,
+      "step": 78630
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.603007576337337e-05,
+      "loss": 1.3933,
+      "step": 78640
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.602892783347362e-05,
+      "loss": 1.3562,
+      "step": 78650
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.602777990357389e-05,
+      "loss": 1.3674,
+      "step": 78660
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.602663197367414e-05,
+      "loss": 1.3259,
+      "step": 78670
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.602548404377439e-05,
+      "loss": 1.4529,
+      "step": 78680
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.602433611387464e-05,
+      "loss": 1.3163,
+      "step": 78690
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.602318818397489e-05,
+      "loss": 1.5043,
+      "step": 78700
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.602204025407515e-05,
+      "loss": 1.4801,
+      "step": 78710
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.60208923241754e-05,
+      "loss": 1.3613,
+      "step": 78720
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.601974439427565e-05,
+      "loss": 1.3031,
+      "step": 78730
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.60185964643759e-05,
+      "loss": 1.4234,
+      "step": 78740
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.601744853447616e-05,
+      "loss": 1.3035,
+      "step": 78750
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.60163006045764e-05,
+      "loss": 1.3421,
+      "step": 78760
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.601515267467666e-05,
+      "loss": 1.268,
+      "step": 78770
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.601400474477691e-05,
+      "loss": 1.3739,
+      "step": 78780
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.601285681487716e-05,
+      "loss": 1.3306,
+      "step": 78790
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.601170888497741e-05,
+      "loss": 1.4318,
+      "step": 78800
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.601056095507768e-05,
+      "loss": 1.2216,
+      "step": 78810
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.600941302517793e-05,
+      "loss": 1.401,
+      "step": 78820
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.600826509527818e-05,
+      "loss": 1.3708,
+      "step": 78830
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.600711716537843e-05,
+      "loss": 1.3431,
+      "step": 78840
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.600596923547868e-05,
+      "loss": 1.5034,
+      "step": 78850
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.600482130557894e-05,
+      "loss": 1.4479,
+      "step": 78860
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.600367337567919e-05,
+      "loss": 1.3346,
+      "step": 78870
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.600252544577944e-05,
+      "loss": 1.4248,
+      "step": 78880
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.600137751587969e-05,
+      "loss": 1.3686,
+      "step": 78890
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.600022958597995e-05,
+      "loss": 1.3758,
+      "step": 78900
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.599908165608019e-05,
+      "loss": 1.2776,
+      "step": 78910
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.599793372618045e-05,
+      "loss": 1.3977,
+      "step": 78920
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.59967857962807e-05,
+      "loss": 1.2551,
+      "step": 78930
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.599563786638095e-05,
+      "loss": 1.58,
+      "step": 78940
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.59944899364812e-05,
+      "loss": 1.5001,
+      "step": 78950
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.599334200658147e-05,
+      "loss": 1.43,
+      "step": 78960
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.599219407668171e-05,
+      "loss": 1.3174,
+      "step": 78970
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.599104614678196e-05,
+      "loss": 1.3699,
+      "step": 78980
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.598989821688221e-05,
+      "loss": 1.4486,
+      "step": 78990
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.598875028698246e-05,
+      "loss": 1.3444,
+      "step": 79000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.598760235708273e-05,
+      "loss": 1.3449,
+      "step": 79010
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.598645442718298e-05,
+      "loss": 1.5511,
+      "step": 79020
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.598530649728323e-05,
+      "loss": 1.3561,
+      "step": 79030
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.598415856738348e-05,
+      "loss": 1.5045,
+      "step": 79040
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.598301063748374e-05,
+      "loss": 1.434,
+      "step": 79050
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.598186270758398e-05,
+      "loss": 1.3442,
+      "step": 79060
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.598071477768424e-05,
+      "loss": 1.3632,
+      "step": 79070
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.597956684778449e-05,
+      "loss": 1.3115,
+      "step": 79080
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.597841891788474e-05,
+      "loss": 1.3618,
+      "step": 79090
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.597727098798499e-05,
+      "loss": 1.2865,
+      "step": 79100
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.597612305808525e-05,
+      "loss": 1.4313,
+      "step": 79110
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.59749751281855e-05,
+      "loss": 1.5495,
+      "step": 79120
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.597382719828575e-05,
+      "loss": 1.2206,
+      "step": 79130
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.5972679268386e-05,
+      "loss": 1.409,
+      "step": 79140
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.597153133848625e-05,
+      "loss": 1.3049,
+      "step": 79150
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.597038340858652e-05,
+      "loss": 1.4042,
+      "step": 79160
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.596923547868677e-05,
+      "loss": 1.3436,
+      "step": 79170
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.596808754878702e-05,
+      "loss": 1.2193,
+      "step": 79180
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.596693961888727e-05,
+      "loss": 1.235,
+      "step": 79190
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.596579168898753e-05,
+      "loss": 1.5324,
+      "step": 79200
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.596464375908777e-05,
+      "loss": 1.285,
+      "step": 79210
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.596349582918803e-05,
+      "loss": 1.3513,
+      "step": 79220
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.596234789928828e-05,
+      "loss": 1.4273,
+      "step": 79230
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.596119996938853e-05,
+      "loss": 1.4275,
+      "step": 79240
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.596005203948878e-05,
+      "loss": 1.3836,
+      "step": 79250
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.595890410958904e-05,
+      "loss": 1.405,
+      "step": 79260
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.595775617968929e-05,
+      "loss": 1.4321,
+      "step": 79270
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.595660824978954e-05,
+      "loss": 1.3049,
+      "step": 79280
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.595546031988979e-05,
+      "loss": 1.4182,
+      "step": 79290
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.595431238999004e-05,
+      "loss": 1.2824,
+      "step": 79300
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.59531644600903e-05,
+      "loss": 1.3375,
+      "step": 79310
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.595201653019055e-05,
+      "loss": 1.467,
+      "step": 79320
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.59508686002908e-05,
+      "loss": 1.3211,
+      "step": 79330
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.594972067039105e-05,
+      "loss": 1.1725,
+      "step": 79340
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.594857274049132e-05,
+      "loss": 1.2856,
+      "step": 79350
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.594742481059155e-05,
+      "loss": 1.3682,
+      "step": 79360
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.594627688069182e-05,
+      "loss": 1.3121,
+      "step": 79370
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.594512895079207e-05,
+      "loss": 1.3607,
+      "step": 79380
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.594398102089232e-05,
+      "loss": 1.4286,
+      "step": 79390
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.594283309099257e-05,
+      "loss": 1.4591,
+      "step": 79400
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.594168516109283e-05,
+      "loss": 1.3606,
+      "step": 79410
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.594053723119308e-05,
+      "loss": 1.4982,
+      "step": 79420
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.593938930129333e-05,
+      "loss": 1.4188,
+      "step": 79430
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.593824137139358e-05,
+      "loss": 1.4652,
+      "step": 79440
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.593709344149383e-05,
+      "loss": 1.469,
+      "step": 79450
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.59359455115941e-05,
+      "loss": 1.3425,
+      "step": 79460
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.593479758169433e-05,
+      "loss": 1.5645,
+      "step": 79470
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.593364965179459e-05,
+      "loss": 1.4341,
+      "step": 79480
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.593250172189484e-05,
+      "loss": 1.5345,
+      "step": 79490
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.593135379199509e-05,
+      "loss": 1.3036,
+      "step": 79500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.593020586209534e-05,
+      "loss": 1.4134,
+      "step": 79510
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.59290579321956e-05,
+      "loss": 1.3155,
+      "step": 79520
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.592791000229586e-05,
+      "loss": 1.3053,
+      "step": 79530
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.59267620723961e-05,
+      "loss": 1.3032,
+      "step": 79540
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.592561414249636e-05,
+      "loss": 1.3907,
+      "step": 79550
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.59244662125966e-05,
+      "loss": 1.3024,
+      "step": 79560
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.592331828269687e-05,
+      "loss": 1.404,
+      "step": 79570
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.592217035279712e-05,
+      "loss": 1.487,
+      "step": 79580
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.592102242289737e-05,
+      "loss": 1.4384,
+      "step": 79590
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.591987449299762e-05,
+      "loss": 1.3307,
+      "step": 79600
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.591872656309788e-05,
+      "loss": 1.437,
+      "step": 79610
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.591757863319812e-05,
+      "loss": 1.3609,
+      "step": 79620
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.591643070329838e-05,
+      "loss": 1.2813,
+      "step": 79630
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.591528277339863e-05,
+      "loss": 1.2763,
+      "step": 79640
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.591413484349888e-05,
+      "loss": 1.3761,
+      "step": 79650
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.591298691359913e-05,
+      "loss": 1.6134,
+      "step": 79660
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.59118389836994e-05,
+      "loss": 1.4182,
+      "step": 79670
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.591069105379964e-05,
+      "loss": 1.4087,
+      "step": 79680
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.59095431238999e-05,
+      "loss": 1.2057,
+      "step": 79690
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.590839519400014e-05,
+      "loss": 1.2771,
+      "step": 79700
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.59072472641004e-05,
+      "loss": 1.3173,
+      "step": 79710
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.590609933420066e-05,
+      "loss": 1.3357,
+      "step": 79720
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.590495140430091e-05,
+      "loss": 1.2698,
+      "step": 79730
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.590380347440116e-05,
+      "loss": 1.3086,
+      "step": 79740
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.590265554450141e-05,
+      "loss": 1.4935,
+      "step": 79750
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.590150761460167e-05,
+      "loss": 1.3445,
+      "step": 79760
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.59003596847019e-05,
+      "loss": 1.3374,
+      "step": 79770
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.589921175480217e-05,
+      "loss": 1.4462,
+      "step": 79780
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.589806382490242e-05,
+      "loss": 1.4167,
+      "step": 79790
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.589691589500267e-05,
+      "loss": 1.3207,
+      "step": 79800
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.589576796510292e-05,
+      "loss": 1.4881,
+      "step": 79810
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.589462003520318e-05,
+      "loss": 1.2884,
+      "step": 79820
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.589347210530343e-05,
+      "loss": 1.4385,
+      "step": 79830
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.589232417540368e-05,
+      "loss": 1.4247,
+      "step": 79840
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.589117624550393e-05,
+      "loss": 1.3864,
+      "step": 79850
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.589002831560418e-05,
+      "loss": 1.2702,
+      "step": 79860
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.588888038570445e-05,
+      "loss": 1.2239,
+      "step": 79870
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.58877324558047e-05,
+      "loss": 1.528,
+      "step": 79880
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.588658452590495e-05,
+      "loss": 1.1916,
+      "step": 79890
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.58854365960052e-05,
+      "loss": 1.4144,
+      "step": 79900
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.588428866610546e-05,
+      "loss": 1.2973,
+      "step": 79910
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.58831407362057e-05,
+      "loss": 1.2627,
+      "step": 79920
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.588199280630596e-05,
+      "loss": 1.4542,
+      "step": 79930
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.588084487640621e-05,
+      "loss": 1.4074,
+      "step": 79940
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.587969694650646e-05,
+      "loss": 1.402,
+      "step": 79950
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.587854901660671e-05,
+      "loss": 1.3667,
+      "step": 79960
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.587740108670697e-05,
+      "loss": 1.381,
+      "step": 79970
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.587625315680722e-05,
+      "loss": 1.474,
+      "step": 79980
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.587510522690747e-05,
+      "loss": 1.4182,
+      "step": 79990
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.587395729700772e-05,
+      "loss": 1.4588,
+      "step": 80000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.587280936710797e-05,
+      "loss": 1.5832,
+      "step": 80010
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.587166143720823e-05,
+      "loss": 1.4217,
+      "step": 80020
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.587051350730848e-05,
+      "loss": 1.3494,
+      "step": 80030
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.586936557740873e-05,
+      "loss": 1.3808,
+      "step": 80040
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.586821764750898e-05,
+      "loss": 1.5211,
+      "step": 80050
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.586706971760925e-05,
+      "loss": 1.4551,
+      "step": 80060
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.586592178770948e-05,
+      "loss": 1.4461,
+      "step": 80070
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.586477385780975e-05,
+      "loss": 1.3837,
+      "step": 80080
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.586362592791e-05,
+      "loss": 1.3643,
+      "step": 80090
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.586247799801025e-05,
+      "loss": 1.3294,
+      "step": 80100
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.58613300681105e-05,
+      "loss": 1.2696,
+      "step": 80110
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.586018213821076e-05,
+      "loss": 1.3857,
+      "step": 80120
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.585903420831101e-05,
+      "loss": 1.4904,
+      "step": 80130
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.585788627841126e-05,
+      "loss": 1.4469,
+      "step": 80140
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.585673834851151e-05,
+      "loss": 1.5575,
+      "step": 80150
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.585559041861176e-05,
+      "loss": 1.3633,
+      "step": 80160
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.585444248871202e-05,
+      "loss": 1.4601,
+      "step": 80170
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.585329455881227e-05,
+      "loss": 1.3567,
+      "step": 80180
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.585214662891252e-05,
+      "loss": 1.4321,
+      "step": 80190
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.585099869901277e-05,
+      "loss": 1.1926,
+      "step": 80200
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.584985076911304e-05,
+      "loss": 1.3063,
+      "step": 80210
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.584870283921327e-05,
+      "loss": 1.4941,
+      "step": 80220
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.584755490931354e-05,
+      "loss": 1.4647,
+      "step": 80230
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.584640697941379e-05,
+      "loss": 1.3377,
+      "step": 80240
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.584525904951404e-05,
+      "loss": 1.6171,
+      "step": 80250
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.584411111961428e-05,
+      "loss": 1.3328,
+      "step": 80260
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.584296318971455e-05,
+      "loss": 1.4397,
+      "step": 80270
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.58418152598148e-05,
+      "loss": 1.2426,
+      "step": 80280
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.584066732991505e-05,
+      "loss": 1.3716,
+      "step": 80290
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.58395194000153e-05,
+      "loss": 1.3363,
+      "step": 80300
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.583837147011555e-05,
+      "loss": 1.4184,
+      "step": 80310
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.583722354021581e-05,
+      "loss": 1.3561,
+      "step": 80320
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.583607561031606e-05,
+      "loss": 1.3642,
+      "step": 80330
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.583492768041631e-05,
+      "loss": 1.3315,
+      "step": 80340
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.583377975051656e-05,
+      "loss": 1.1478,
+      "step": 80350
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.583263182061682e-05,
+      "loss": 1.3304,
+      "step": 80360
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.583148389071706e-05,
+      "loss": 1.3753,
+      "step": 80370
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.583033596081732e-05,
+      "loss": 1.3156,
+      "step": 80380
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.582918803091757e-05,
+      "loss": 1.5085,
+      "step": 80390
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.582804010101782e-05,
+      "loss": 1.2895,
+      "step": 80400
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.582689217111807e-05,
+      "loss": 1.3282,
+      "step": 80410
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.582574424121834e-05,
+      "loss": 1.2861,
+      "step": 80420
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.582459631131859e-05,
+      "loss": 1.3072,
+      "step": 80430
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.582344838141884e-05,
+      "loss": 1.4303,
+      "step": 80440
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.582230045151909e-05,
+      "loss": 1.2325,
+      "step": 80450
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.582115252161934e-05,
+      "loss": 1.4038,
+      "step": 80460
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.58200045917196e-05,
+      "loss": 1.364,
+      "step": 80470
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.581885666181985e-05,
+      "loss": 1.3877,
+      "step": 80480
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.58177087319201e-05,
+      "loss": 1.4123,
+      "step": 80490
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.581656080202035e-05,
+      "loss": 1.3255,
+      "step": 80500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.581541287212061e-05,
+      "loss": 1.29,
+      "step": 80510
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.581426494222085e-05,
+      "loss": 1.2253,
+      "step": 80520
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.581311701232111e-05,
+      "loss": 1.3798,
+      "step": 80530
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.581196908242136e-05,
+      "loss": 1.3502,
+      "step": 80540
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.581082115252161e-05,
+      "loss": 1.4064,
+      "step": 80550
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.580967322262186e-05,
+      "loss": 1.2406,
+      "step": 80560
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.580852529272213e-05,
+      "loss": 1.2688,
+      "step": 80570
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.580737736282238e-05,
+      "loss": 1.4021,
+      "step": 80580
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.580622943292262e-05,
+      "loss": 1.3277,
+      "step": 80590
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.580508150302287e-05,
+      "loss": 1.4292,
+      "step": 80600
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.580393357312312e-05,
+      "loss": 1.4478,
+      "step": 80610
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.580278564322339e-05,
+      "loss": 1.4221,
+      "step": 80620
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.580163771332364e-05,
+      "loss": 1.3378,
+      "step": 80630
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.580048978342389e-05,
+      "loss": 1.4771,
+      "step": 80640
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.579934185352414e-05,
+      "loss": 1.4159,
+      "step": 80650
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.57981939236244e-05,
+      "loss": 1.2477,
+      "step": 80660
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.579704599372464e-05,
+      "loss": 1.331,
+      "step": 80670
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.57958980638249e-05,
+      "loss": 1.2876,
+      "step": 80680
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.579475013392515e-05,
+      "loss": 1.2541,
+      "step": 80690
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.57936022040254e-05,
+      "loss": 1.2647,
+      "step": 80700
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.579245427412565e-05,
+      "loss": 1.3798,
+      "step": 80710
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.579130634422591e-05,
+      "loss": 1.3868,
+      "step": 80720
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.579015841432616e-05,
+      "loss": 1.5019,
+      "step": 80730
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.578901048442641e-05,
+      "loss": 1.3713,
+      "step": 80740
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.578786255452666e-05,
+      "loss": 1.3988,
+      "step": 80750
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.578671462462691e-05,
+      "loss": 1.4086,
+      "step": 80760
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.578556669472718e-05,
+      "loss": 1.4571,
+      "step": 80770
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.578441876482743e-05,
+      "loss": 1.3355,
+      "step": 80780
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.578327083492768e-05,
+      "loss": 1.3018,
+      "step": 80790
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.578212290502793e-05,
+      "loss": 1.2857,
+      "step": 80800
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.578097497512818e-05,
+      "loss": 1.3014,
+      "step": 80810
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.577982704522843e-05,
+      "loss": 1.2924,
+      "step": 80820
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.577867911532869e-05,
+      "loss": 1.4885,
+      "step": 80830
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.577753118542894e-05,
+      "loss": 1.3027,
+      "step": 80840
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.577638325552919e-05,
+      "loss": 1.3014,
+      "step": 80850
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.577523532562944e-05,
+      "loss": 1.1796,
+      "step": 80860
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.577408739572969e-05,
+      "loss": 1.2885,
+      "step": 80870
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.577293946582995e-05,
+      "loss": 1.4101,
+      "step": 80880
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.57717915359302e-05,
+      "loss": 1.4446,
+      "step": 80890
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.577064360603045e-05,
+      "loss": 1.3228,
+      "step": 80900
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.57694956761307e-05,
+      "loss": 1.2943,
+      "step": 80910
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.576834774623097e-05,
+      "loss": 1.4258,
+      "step": 80920
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.57671998163312e-05,
+      "loss": 1.4819,
+      "step": 80930
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.576605188643146e-05,
+      "loss": 1.3006,
+      "step": 80940
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.576490395653171e-05,
+      "loss": 1.5268,
+      "step": 80950
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.576375602663196e-05,
+      "loss": 1.3166,
+      "step": 80960
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.576260809673221e-05,
+      "loss": 1.3974,
+      "step": 80970
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.576146016683248e-05,
+      "loss": 1.3855,
+      "step": 80980
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.576031223693273e-05,
+      "loss": 1.4044,
+      "step": 80990
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.575916430703298e-05,
+      "loss": 1.4601,
+      "step": 81000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.575801637713323e-05,
+      "loss": 1.2833,
+      "step": 81010
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.575686844723348e-05,
+      "loss": 1.3721,
+      "step": 81020
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.575572051733374e-05,
+      "loss": 1.2465,
+      "step": 81030
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.575457258743399e-05,
+      "loss": 1.3329,
+      "step": 81040
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.575342465753424e-05,
+      "loss": 1.5865,
+      "step": 81050
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.575227672763449e-05,
+      "loss": 1.3852,
+      "step": 81060
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.575112879773475e-05,
+      "loss": 1.3534,
+      "step": 81070
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.574998086783499e-05,
+      "loss": 1.2037,
+      "step": 81080
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.574883293793525e-05,
+      "loss": 1.4219,
+      "step": 81090
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.57476850080355e-05,
+      "loss": 1.2485,
+      "step": 81100
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.574653707813575e-05,
+      "loss": 1.3743,
+      "step": 81110
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.5745389148236e-05,
+      "loss": 1.3156,
+      "step": 81120
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.574424121833627e-05,
+      "loss": 1.5589,
+      "step": 81130
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.574309328843652e-05,
+      "loss": 1.3434,
+      "step": 81140
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.574194535853677e-05,
+      "loss": 1.4177,
+      "step": 81150
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.574079742863702e-05,
+      "loss": 1.3191,
+      "step": 81160
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.573964949873727e-05,
+      "loss": 1.3889,
+      "step": 81170
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.573850156883753e-05,
+      "loss": 1.2979,
+      "step": 81180
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.573735363893778e-05,
+      "loss": 1.3645,
+      "step": 81190
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.573620570903803e-05,
+      "loss": 1.4066,
+      "step": 81200
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.573505777913828e-05,
+      "loss": 1.2161,
+      "step": 81210
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.573390984923854e-05,
+      "loss": 1.2266,
+      "step": 81220
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.573276191933878e-05,
+      "loss": 1.1791,
+      "step": 81230
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.573161398943904e-05,
+      "loss": 1.3066,
+      "step": 81240
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.573046605953929e-05,
+      "loss": 1.4988,
+      "step": 81250
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.572931812963954e-05,
+      "loss": 1.4137,
+      "step": 81260
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.572817019973979e-05,
+      "loss": 1.454,
+      "step": 81270
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.572702226984005e-05,
+      "loss": 1.3734,
+      "step": 81280
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.57258743399403e-05,
+      "loss": 1.2448,
+      "step": 81290
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.572472641004055e-05,
+      "loss": 1.3209,
+      "step": 81300
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.57235784801408e-05,
+      "loss": 1.3732,
+      "step": 81310
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.572243055024105e-05,
+      "loss": 1.3243,
+      "step": 81320
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.572128262034132e-05,
+      "loss": 1.4006,
+      "step": 81330
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.572013469044157e-05,
+      "loss": 1.3481,
+      "step": 81340
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.571898676054182e-05,
+      "loss": 1.3016,
+      "step": 81350
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.571783883064207e-05,
+      "loss": 1.3023,
+      "step": 81360
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.571669090074233e-05,
+      "loss": 1.3461,
+      "step": 81370
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.571554297084257e-05,
+      "loss": 1.3782,
+      "step": 81380
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.571439504094283e-05,
+      "loss": 1.3693,
+      "step": 81390
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.571324711104308e-05,
+      "loss": 1.317,
+      "step": 81400
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.571209918114333e-05,
+      "loss": 1.3004,
+      "step": 81410
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.571095125124358e-05,
+      "loss": 1.4979,
+      "step": 81420
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.570980332134384e-05,
+      "loss": 1.3629,
+      "step": 81430
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.570865539144409e-05,
+      "loss": 1.3968,
+      "step": 81440
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.570750746154434e-05,
+      "loss": 1.4106,
+      "step": 81450
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.570635953164459e-05,
+      "loss": 1.4492,
+      "step": 81460
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.570521160174484e-05,
+      "loss": 1.5436,
+      "step": 81470
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.57040636718451e-05,
+      "loss": 1.3248,
+      "step": 81480
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.570291574194536e-05,
+      "loss": 1.3148,
+      "step": 81490
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.57017678120456e-05,
+      "loss": 1.2096,
+      "step": 81500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.570061988214586e-05,
+      "loss": 1.5113,
+      "step": 81510
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.569947195224612e-05,
+      "loss": 1.2761,
+      "step": 81520
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.569832402234636e-05,
+      "loss": 1.3414,
+      "step": 81530
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.569717609244662e-05,
+      "loss": 1.4891,
+      "step": 81540
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.569602816254687e-05,
+      "loss": 1.3217,
+      "step": 81550
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.569488023264712e-05,
+      "loss": 1.3085,
+      "step": 81560
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.569373230274737e-05,
+      "loss": 1.3295,
+      "step": 81570
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.569258437284763e-05,
+      "loss": 1.3265,
+      "step": 81580
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.569143644294788e-05,
+      "loss": 1.2185,
+      "step": 81590
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.569028851304813e-05,
+      "loss": 1.4282,
+      "step": 81600
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.568914058314838e-05,
+      "loss": 1.3614,
+      "step": 81610
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.568799265324863e-05,
+      "loss": 1.3413,
+      "step": 81620
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.56868447233489e-05,
+      "loss": 1.2098,
+      "step": 81630
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.568569679344914e-05,
+      "loss": 1.5093,
+      "step": 81640
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.56845488635494e-05,
+      "loss": 1.3384,
+      "step": 81650
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.568340093364964e-05,
+      "loss": 1.5133,
+      "step": 81660
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.568225300374991e-05,
+      "loss": 1.4254,
+      "step": 81670
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.568110507385014e-05,
+      "loss": 1.3857,
+      "step": 81680
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.567995714395041e-05,
+      "loss": 1.3412,
+      "step": 81690
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.567880921405066e-05,
+      "loss": 1.3593,
+      "step": 81700
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.567766128415091e-05,
+      "loss": 1.4172,
+      "step": 81710
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.567651335425116e-05,
+      "loss": 1.492,
+      "step": 81720
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.567536542435142e-05,
+      "loss": 1.4242,
+      "step": 81730
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.567421749445167e-05,
+      "loss": 1.4048,
+      "step": 81740
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.567306956455192e-05,
+      "loss": 1.3308,
+      "step": 81750
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.567192163465217e-05,
+      "loss": 1.3637,
+      "step": 81760
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.567077370475242e-05,
+      "loss": 1.3881,
+      "step": 81770
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.566962577485268e-05,
+      "loss": 1.3821,
+      "step": 81780
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.566847784495293e-05,
+      "loss": 1.3493,
+      "step": 81790
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.566732991505318e-05,
+      "loss": 1.3048,
+      "step": 81800
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.566618198515343e-05,
+      "loss": 1.4578,
+      "step": 81810
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.56650340552537e-05,
+      "loss": 1.4924,
+      "step": 81820
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.566388612535393e-05,
+      "loss": 1.3739,
+      "step": 81830
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.56627381954542e-05,
+      "loss": 1.4344,
+      "step": 81840
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.566159026555445e-05,
+      "loss": 1.3778,
+      "step": 81850
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.56604423356547e-05,
+      "loss": 1.3219,
+      "step": 81860
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.565929440575495e-05,
+      "loss": 1.2605,
+      "step": 81870
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.565814647585521e-05,
+      "loss": 1.2696,
+      "step": 81880
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.565699854595546e-05,
+      "loss": 1.5466,
+      "step": 81890
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.565585061605571e-05,
+      "loss": 1.275,
+      "step": 81900
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.565470268615596e-05,
+      "loss": 1.4269,
+      "step": 81910
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.565355475625621e-05,
+      "loss": 1.4119,
+      "step": 81920
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.565240682635647e-05,
+      "loss": 1.3298,
+      "step": 81930
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.565125889645672e-05,
+      "loss": 1.272,
+      "step": 81940
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.565011096655697e-05,
+      "loss": 1.2806,
+      "step": 81950
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.564896303665722e-05,
+      "loss": 1.3554,
+      "step": 81960
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.564781510675748e-05,
+      "loss": 1.3671,
+      "step": 81970
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.564666717685772e-05,
+      "loss": 1.3881,
+      "step": 81980
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.564551924695798e-05,
+      "loss": 1.4472,
+      "step": 81990
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.564437131705823e-05,
+      "loss": 1.2545,
+      "step": 82000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.564322338715848e-05,
+      "loss": 1.358,
+      "step": 82010
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.564207545725873e-05,
+      "loss": 1.3877,
+      "step": 82020
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.5640927527359e-05,
+      "loss": 1.3823,
+      "step": 82030
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.563977959745925e-05,
+      "loss": 1.3677,
+      "step": 82040
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.56386316675595e-05,
+      "loss": 1.393,
+      "step": 82050
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.563748373765975e-05,
+      "loss": 1.3582,
+      "step": 82060
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.563633580776e-05,
+      "loss": 1.4349,
+      "step": 82070
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.563518787786026e-05,
+      "loss": 1.4226,
+      "step": 82080
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.563403994796051e-05,
+      "loss": 1.3235,
+      "step": 82090
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.563289201806076e-05,
+      "loss": 1.334,
+      "step": 82100
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.563174408816101e-05,
+      "loss": 1.3516,
+      "step": 82110
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.563059615826127e-05,
+      "loss": 1.3512,
+      "step": 82120
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.562944822836151e-05,
+      "loss": 1.4457,
+      "step": 82130
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.562830029846177e-05,
+      "loss": 1.3796,
+      "step": 82140
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.562715236856202e-05,
+      "loss": 1.2532,
+      "step": 82150
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.562600443866227e-05,
+      "loss": 1.1786,
+      "step": 82160
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.562485650876252e-05,
+      "loss": 1.3952,
+      "step": 82170
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.562370857886277e-05,
+      "loss": 1.4902,
+      "step": 82180
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.562256064896304e-05,
+      "loss": 1.2505,
+      "step": 82190
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.562141271906329e-05,
+      "loss": 1.3432,
+      "step": 82200
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.562026478916354e-05,
+      "loss": 1.2954,
+      "step": 82210
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.561911685926378e-05,
+      "loss": 1.4127,
+      "step": 82220
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.561796892936405e-05,
+      "loss": 1.3079,
+      "step": 82230
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.561682099946428e-05,
+      "loss": 1.2117,
+      "step": 82240
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.561567306956455e-05,
+      "loss": 1.3798,
+      "step": 82250
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.56145251396648e-05,
+      "loss": 1.2081,
+      "step": 82260
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.561337720976505e-05,
+      "loss": 1.3436,
+      "step": 82270
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.56122292798653e-05,
+      "loss": 1.622,
+      "step": 82280
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.561108134996556e-05,
+      "loss": 1.3162,
+      "step": 82290
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.560993342006581e-05,
+      "loss": 1.443,
+      "step": 82300
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.560878549016606e-05,
+      "loss": 1.2465,
+      "step": 82310
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.560763756026631e-05,
+      "loss": 1.515,
+      "step": 82320
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.560648963036656e-05,
+      "loss": 1.2847,
+      "step": 82330
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.560534170046682e-05,
+      "loss": 1.2888,
+      "step": 82340
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.560419377056707e-05,
+      "loss": 1.3467,
+      "step": 82350
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.560304584066732e-05,
+      "loss": 1.4413,
+      "step": 82360
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.560189791076757e-05,
+      "loss": 1.2791,
+      "step": 82370
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.560074998086784e-05,
+      "loss": 1.2996,
+      "step": 82380
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.559960205096807e-05,
+      "loss": 1.2972,
+      "step": 82390
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.559845412106834e-05,
+      "loss": 1.3024,
+      "step": 82400
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.559730619116859e-05,
+      "loss": 1.2517,
+      "step": 82410
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.559615826126884e-05,
+      "loss": 1.3057,
+      "step": 82420
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.559501033136909e-05,
+      "loss": 1.3624,
+      "step": 82430
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.559386240146935e-05,
+      "loss": 1.53,
+      "step": 82440
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.55927144715696e-05,
+      "loss": 1.4261,
+      "step": 82450
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.559156654166985e-05,
+      "loss": 1.4135,
+      "step": 82460
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.55904186117701e-05,
+      "loss": 1.5022,
+      "step": 82470
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.558927068187035e-05,
+      "loss": 1.3191,
+      "step": 82480
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.558812275197061e-05,
+      "loss": 1.3721,
+      "step": 82490
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.558697482207086e-05,
+      "loss": 1.3252,
+      "step": 82500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.558582689217111e-05,
+      "loss": 1.4215,
+      "step": 82510
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.558467896227136e-05,
+      "loss": 1.2197,
+      "step": 82520
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.558353103237163e-05,
+      "loss": 1.3985,
+      "step": 82530
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.558238310247186e-05,
+      "loss": 1.4328,
+      "step": 82540
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.558123517257212e-05,
+      "loss": 1.3146,
+      "step": 82550
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.558008724267237e-05,
+      "loss": 1.42,
+      "step": 82560
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.557893931277262e-05,
+      "loss": 1.4233,
+      "step": 82570
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.557779138287287e-05,
+      "loss": 1.3652,
+      "step": 82580
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.557664345297314e-05,
+      "loss": 1.3281,
+      "step": 82590
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.557549552307339e-05,
+      "loss": 1.4129,
+      "step": 82600
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.557434759317364e-05,
+      "loss": 1.4634,
+      "step": 82610
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.557319966327389e-05,
+      "loss": 1.2769,
+      "step": 82620
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.557205173337414e-05,
+      "loss": 1.2668,
+      "step": 82630
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.55709038034744e-05,
+      "loss": 1.4982,
+      "step": 82640
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.556975587357465e-05,
+      "loss": 1.4852,
+      "step": 82650
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.55686079436749e-05,
+      "loss": 1.3638,
+      "step": 82660
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.556746001377515e-05,
+      "loss": 1.4327,
+      "step": 82670
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.556631208387541e-05,
+      "loss": 1.3516,
+      "step": 82680
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.556516415397565e-05,
+      "loss": 1.5716,
+      "step": 82690
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.556401622407591e-05,
+      "loss": 1.3827,
+      "step": 82700
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.556286829417616e-05,
+      "loss": 1.3805,
+      "step": 82710
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.556172036427641e-05,
+      "loss": 1.3553,
+      "step": 82720
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.556057243437666e-05,
+      "loss": 1.4665,
+      "step": 82730
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.555942450447693e-05,
+      "loss": 1.2252,
+      "step": 82740
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.555827657457718e-05,
+      "loss": 1.3704,
+      "step": 82750
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.555712864467743e-05,
+      "loss": 1.3759,
+      "step": 82760
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.555598071477768e-05,
+      "loss": 1.2797,
+      "step": 82770
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.555483278487793e-05,
+      "loss": 1.0523,
+      "step": 82780
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.555368485497819e-05,
+      "loss": 1.3245,
+      "step": 82790
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.555253692507844e-05,
+      "loss": 1.4074,
+      "step": 82800
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.555138899517869e-05,
+      "loss": 1.301,
+      "step": 82810
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.555024106527894e-05,
+      "loss": 1.3273,
+      "step": 82820
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.55490931353792e-05,
+      "loss": 1.3221,
+      "step": 82830
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.554794520547944e-05,
+      "loss": 1.4434,
+      "step": 82840
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.55467972755797e-05,
+      "loss": 1.3592,
+      "step": 82850
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.554564934567995e-05,
+      "loss": 1.4214,
+      "step": 82860
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.55445014157802e-05,
+      "loss": 1.4266,
+      "step": 82870
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.554335348588045e-05,
+      "loss": 1.3403,
+      "step": 82880
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.554220555598071e-05,
+      "loss": 1.3485,
+      "step": 82890
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.554105762608096e-05,
+      "loss": 1.3474,
+      "step": 82900
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.553990969618121e-05,
+      "loss": 1.4512,
+      "step": 82910
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.553876176628146e-05,
+      "loss": 1.3717,
+      "step": 82920
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.553761383638171e-05,
+      "loss": 1.4311,
+      "step": 82930
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.553646590648198e-05,
+      "loss": 1.3465,
+      "step": 82940
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.553531797658223e-05,
+      "loss": 1.3782,
+      "step": 82950
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.553417004668248e-05,
+      "loss": 1.3651,
+      "step": 82960
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.553302211678273e-05,
+      "loss": 1.1826,
+      "step": 82970
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.553187418688299e-05,
+      "loss": 1.3223,
+      "step": 82980
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.553072625698323e-05,
+      "loss": 1.2759,
+      "step": 82990
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.552957832708349e-05,
+      "loss": 1.2616,
+      "step": 83000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.552843039718374e-05,
+      "loss": 1.437,
+      "step": 83010
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.552728246728399e-05,
+      "loss": 1.2443,
+      "step": 83020
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.552613453738424e-05,
+      "loss": 1.2693,
+      "step": 83030
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.55249866074845e-05,
+      "loss": 1.4196,
+      "step": 83040
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.552383867758475e-05,
+      "loss": 1.2626,
+      "step": 83050
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.5522690747685e-05,
+      "loss": 1.6568,
+      "step": 83060
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.552154281778525e-05,
+      "loss": 1.248,
+      "step": 83070
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.55203948878855e-05,
+      "loss": 1.4981,
+      "step": 83080
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.551924695798577e-05,
+      "loss": 1.2837,
+      "step": 83090
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.551809902808602e-05,
+      "loss": 1.4739,
+      "step": 83100
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.551695109818627e-05,
+      "loss": 1.2276,
+      "step": 83110
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.551580316828652e-05,
+      "loss": 1.5189,
+      "step": 83120
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.551465523838678e-05,
+      "loss": 1.4893,
+      "step": 83130
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.551350730848702e-05,
+      "loss": 1.2057,
+      "step": 83140
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.551235937858728e-05,
+      "loss": 1.3439,
+      "step": 83150
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.551121144868753e-05,
+      "loss": 1.1919,
+      "step": 83160
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.551006351878778e-05,
+      "loss": 1.304,
+      "step": 83170
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.550891558888803e-05,
+      "loss": 1.3945,
+      "step": 83180
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.550776765898829e-05,
+      "loss": 1.3439,
+      "step": 83190
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.550661972908854e-05,
+      "loss": 1.3106,
+      "step": 83200
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.550547179918879e-05,
+      "loss": 1.4249,
+      "step": 83210
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.550432386928904e-05,
+      "loss": 1.4345,
+      "step": 83220
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.550317593938929e-05,
+      "loss": 1.3541,
+      "step": 83230
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.550202800948955e-05,
+      "loss": 1.441,
+      "step": 83240
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.55008800795898e-05,
+      "loss": 1.2795,
+      "step": 83250
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.549973214969005e-05,
+      "loss": 1.323,
+      "step": 83260
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.54985842197903e-05,
+      "loss": 1.3347,
+      "step": 83270
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.549743628989057e-05,
+      "loss": 1.3779,
+      "step": 83280
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.54962883599908e-05,
+      "loss": 1.2992,
+      "step": 83290
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.549514043009107e-05,
+      "loss": 1.664,
+      "step": 83300
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.549399250019132e-05,
+      "loss": 1.3635,
+      "step": 83310
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.549284457029157e-05,
+      "loss": 1.6223,
+      "step": 83320
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.549169664039182e-05,
+      "loss": 1.1758,
+      "step": 83330
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.549054871049208e-05,
+      "loss": 1.2893,
+      "step": 83340
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.548940078059233e-05,
+      "loss": 1.3336,
+      "step": 83350
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.548825285069258e-05,
+      "loss": 1.2578,
+      "step": 83360
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.548710492079283e-05,
+      "loss": 1.3579,
+      "step": 83370
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.548595699089308e-05,
+      "loss": 1.3695,
+      "step": 83380
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.548480906099334e-05,
+      "loss": 1.248,
+      "step": 83390
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.548366113109359e-05,
+      "loss": 1.4265,
+      "step": 83400
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.548251320119384e-05,
+      "loss": 1.4529,
+      "step": 83410
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.548136527129409e-05,
+      "loss": 1.2508,
+      "step": 83420
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.548021734139436e-05,
+      "loss": 1.2809,
+      "step": 83430
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.547906941149459e-05,
+      "loss": 1.3021,
+      "step": 83440
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.547792148159486e-05,
+      "loss": 1.3308,
+      "step": 83450
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.54767735516951e-05,
+      "loss": 1.4289,
+      "step": 83460
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.547562562179536e-05,
+      "loss": 1.3848,
+      "step": 83470
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.54744776918956e-05,
+      "loss": 1.3224,
+      "step": 83480
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.547332976199587e-05,
+      "loss": 1.3181,
+      "step": 83490
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.547218183209612e-05,
+      "loss": 1.3704,
+      "step": 83500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.547103390219637e-05,
+      "loss": 1.378,
+      "step": 83510
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.546988597229662e-05,
+      "loss": 1.1544,
+      "step": 83520
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.546873804239687e-05,
+      "loss": 1.3146,
+      "step": 83530
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.546759011249713e-05,
+      "loss": 1.202,
+      "step": 83540
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.546644218259737e-05,
+      "loss": 1.3908,
+      "step": 83550
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.546529425269763e-05,
+      "loss": 1.427,
+      "step": 83560
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.546414632279788e-05,
+      "loss": 1.4224,
+      "step": 83570
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.546299839289813e-05,
+      "loss": 1.5207,
+      "step": 83580
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.546185046299838e-05,
+      "loss": 1.354,
+      "step": 83590
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.546070253309864e-05,
+      "loss": 1.5344,
+      "step": 83600
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.54595546031989e-05,
+      "loss": 1.3146,
+      "step": 83610
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.545840667329914e-05,
+      "loss": 1.4452,
+      "step": 83620
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.54572587433994e-05,
+      "loss": 1.2411,
+      "step": 83630
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.545611081349964e-05,
+      "loss": 1.4271,
+      "step": 83640
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.545496288359991e-05,
+      "loss": 1.255,
+      "step": 83650
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.545381495370016e-05,
+      "loss": 1.2786,
+      "step": 83660
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.545266702380041e-05,
+      "loss": 1.247,
+      "step": 83670
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.545151909390066e-05,
+      "loss": 1.2025,
+      "step": 83680
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.545037116400092e-05,
+      "loss": 1.3032,
+      "step": 83690
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.544922323410116e-05,
+      "loss": 1.5493,
+      "step": 83700
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.544807530420142e-05,
+      "loss": 1.4995,
+      "step": 83710
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.544692737430167e-05,
+      "loss": 1.3621,
+      "step": 83720
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.544577944440192e-05,
+      "loss": 1.301,
+      "step": 83730
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.544463151450217e-05,
+      "loss": 1.3783,
+      "step": 83740
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.544348358460243e-05,
+      "loss": 1.2437,
+      "step": 83750
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.544233565470268e-05,
+      "loss": 1.3837,
+      "step": 83760
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.544118772480293e-05,
+      "loss": 1.3579,
+      "step": 83770
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.544003979490318e-05,
+      "loss": 1.3797,
+      "step": 83780
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.543889186500343e-05,
+      "loss": 1.2182,
+      "step": 83790
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.54377439351037e-05,
+      "loss": 1.3131,
+      "step": 83800
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.543659600520395e-05,
+      "loss": 1.411,
+      "step": 83810
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.54354480753042e-05,
+      "loss": 1.4658,
+      "step": 83820
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.543430014540445e-05,
+      "loss": 1.3557,
+      "step": 83830
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.543315221550471e-05,
+      "loss": 1.3872,
+      "step": 83840
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.543200428560494e-05,
+      "loss": 1.2888,
+      "step": 83850
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.543085635570521e-05,
+      "loss": 1.4415,
+      "step": 83860
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.542970842580546e-05,
+      "loss": 1.3834,
+      "step": 83870
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.542856049590571e-05,
+      "loss": 1.4755,
+      "step": 83880
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.542741256600596e-05,
+      "loss": 1.3812,
+      "step": 83890
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.542626463610622e-05,
+      "loss": 1.187,
+      "step": 83900
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.542511670620647e-05,
+      "loss": 1.231,
+      "step": 83910
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.542396877630672e-05,
+      "loss": 1.2597,
+      "step": 83920
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.542282084640697e-05,
+      "loss": 1.5712,
+      "step": 83930
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.542167291650722e-05,
+      "loss": 1.3222,
+      "step": 83940
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.542052498660748e-05,
+      "loss": 1.4966,
+      "step": 83950
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.541937705670773e-05,
+      "loss": 1.3777,
+      "step": 83960
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.541822912680798e-05,
+      "loss": 1.1696,
+      "step": 83970
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.541708119690823e-05,
+      "loss": 1.2927,
+      "step": 83980
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.54159332670085e-05,
+      "loss": 1.1893,
+      "step": 83990
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.541478533710873e-05,
+      "loss": 1.4149,
+      "step": 84000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.5413637407209e-05,
+      "loss": 1.407,
+      "step": 84010
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.541248947730925e-05,
+      "loss": 1.643,
+      "step": 84020
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.54113415474095e-05,
+      "loss": 1.3647,
+      "step": 84030
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.541019361750975e-05,
+      "loss": 1.3727,
+      "step": 84040
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.540904568761001e-05,
+      "loss": 1.4482,
+      "step": 84050
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.540789775771026e-05,
+      "loss": 1.3818,
+      "step": 84060
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.540674982781051e-05,
+      "loss": 1.2939,
+      "step": 84070
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.540560189791076e-05,
+      "loss": 1.1568,
+      "step": 84080
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.540445396801101e-05,
+      "loss": 1.3842,
+      "step": 84090
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.540330603811127e-05,
+      "loss": 1.3251,
+      "step": 84100
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.540215810821152e-05,
+      "loss": 1.3688,
+      "step": 84110
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.540101017831177e-05,
+      "loss": 1.4487,
+      "step": 84120
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.539986224841202e-05,
+      "loss": 1.3477,
+      "step": 84130
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.539871431851229e-05,
+      "loss": 1.3691,
+      "step": 84140
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.539756638861252e-05,
+      "loss": 1.227,
+      "step": 84150
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.539641845871279e-05,
+      "loss": 1.631,
+      "step": 84160
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.539527052881304e-05,
+      "loss": 1.5402,
+      "step": 84170
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.539412259891328e-05,
+      "loss": 1.3373,
+      "step": 84180
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.539297466901353e-05,
+      "loss": 1.4832,
+      "step": 84190
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.53918267391138e-05,
+      "loss": 1.4693,
+      "step": 84200
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.539067880921405e-05,
+      "loss": 1.3287,
+      "step": 84210
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.53895308793143e-05,
+      "loss": 1.3532,
+      "step": 84220
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.538838294941455e-05,
+      "loss": 1.2429,
+      "step": 84230
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.53872350195148e-05,
+      "loss": 1.2507,
+      "step": 84240
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.538608708961506e-05,
+      "loss": 1.3354,
+      "step": 84250
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.538493915971531e-05,
+      "loss": 1.3278,
+      "step": 84260
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.538379122981556e-05,
+      "loss": 1.3697,
+      "step": 84270
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.538264329991581e-05,
+      "loss": 1.3615,
+      "step": 84280
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.538149537001607e-05,
+      "loss": 1.2418,
+      "step": 84290
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.538034744011631e-05,
+      "loss": 1.3473,
+      "step": 84300
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.537919951021657e-05,
+      "loss": 1.3752,
+      "step": 84310
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.537805158031682e-05,
+      "loss": 1.3835,
+      "step": 84320
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.537690365041707e-05,
+      "loss": 1.2248,
+      "step": 84330
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.537575572051732e-05,
+      "loss": 1.4461,
+      "step": 84340
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.537460779061759e-05,
+      "loss": 1.4119,
+      "step": 84350
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.537345986071784e-05,
+      "loss": 1.4803,
+      "step": 84360
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.537231193081809e-05,
+      "loss": 1.3229,
+      "step": 84370
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.537116400091834e-05,
+      "loss": 1.2119,
+      "step": 84380
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.537001607101859e-05,
+      "loss": 1.3918,
+      "step": 84390
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.536886814111885e-05,
+      "loss": 1.2147,
+      "step": 84400
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.53677202112191e-05,
+      "loss": 1.3754,
+      "step": 84410
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.536657228131935e-05,
+      "loss": 1.3756,
+      "step": 84420
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.53654243514196e-05,
+      "loss": 1.4009,
+      "step": 84430
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.536427642151986e-05,
+      "loss": 1.2736,
+      "step": 84440
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.53631284916201e-05,
+      "loss": 1.2896,
+      "step": 84450
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.536198056172036e-05,
+      "loss": 1.3616,
+      "step": 84460
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.536083263182061e-05,
+      "loss": 1.3976,
+      "step": 84470
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.535968470192086e-05,
+      "loss": 1.4173,
+      "step": 84480
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.535853677202111e-05,
+      "loss": 1.3305,
+      "step": 84490
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.535738884212138e-05,
+      "loss": 1.3156,
+      "step": 84500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.535624091222162e-05,
+      "loss": 1.3438,
+      "step": 84510
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.535509298232187e-05,
+      "loss": 1.4075,
+      "step": 84520
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.535394505242212e-05,
+      "loss": 1.4078,
+      "step": 84530
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.535279712252237e-05,
+      "loss": 1.127,
+      "step": 84540
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.535164919262264e-05,
+      "loss": 1.5521,
+      "step": 84550
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.535050126272289e-05,
+      "loss": 1.3735,
+      "step": 84560
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.534935333282314e-05,
+      "loss": 1.3788,
+      "step": 84570
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.534820540292339e-05,
+      "loss": 1.3107,
+      "step": 84580
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.534705747302365e-05,
+      "loss": 1.2181,
+      "step": 84590
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.534590954312389e-05,
+      "loss": 1.4887,
+      "step": 84600
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.534476161322415e-05,
+      "loss": 1.3342,
+      "step": 84610
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.53436136833244e-05,
+      "loss": 1.3854,
+      "step": 84620
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.534246575342465e-05,
+      "loss": 1.308,
+      "step": 84630
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.53413178235249e-05,
+      "loss": 1.3449,
+      "step": 84640
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.534016989362516e-05,
+      "loss": 1.2561,
+      "step": 84650
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.533902196372541e-05,
+      "loss": 1.3714,
+      "step": 84660
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.533787403382566e-05,
+      "loss": 1.4647,
+      "step": 84670
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.533672610392591e-05,
+      "loss": 1.2888,
+      "step": 84680
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.533557817402616e-05,
+      "loss": 1.3703,
+      "step": 84690
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.533443024412643e-05,
+      "loss": 1.3517,
+      "step": 84700
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.533328231422668e-05,
+      "loss": 1.4516,
+      "step": 84710
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.533213438432693e-05,
+      "loss": 1.6176,
+      "step": 84720
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.533098645442718e-05,
+      "loss": 1.4607,
+      "step": 84730
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.532983852452744e-05,
+      "loss": 1.3507,
+      "step": 84740
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.532869059462768e-05,
+      "loss": 1.3901,
+      "step": 84750
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.532754266472794e-05,
+      "loss": 1.2876,
+      "step": 84760
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.532639473482819e-05,
+      "loss": 1.3913,
+      "step": 84770
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.532524680492844e-05,
+      "loss": 1.2967,
+      "step": 84780
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.532409887502869e-05,
+      "loss": 1.2473,
+      "step": 84790
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.532295094512895e-05,
+      "loss": 1.3297,
+      "step": 84800
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.53218030152292e-05,
+      "loss": 1.2543,
+      "step": 84810
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.532065508532945e-05,
+      "loss": 1.3702,
+      "step": 84820
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.53195071554297e-05,
+      "loss": 1.4182,
+      "step": 84830
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.531835922552995e-05,
+      "loss": 1.3685,
+      "step": 84840
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.531721129563021e-05,
+      "loss": 1.2829,
+      "step": 84850
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.531606336573046e-05,
+      "loss": 1.4311,
+      "step": 84860
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.531491543583071e-05,
+      "loss": 1.2927,
+      "step": 84870
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.531376750593096e-05,
+      "loss": 1.3371,
+      "step": 84880
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.531261957603123e-05,
+      "loss": 1.4652,
+      "step": 84890
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.531147164613146e-05,
+      "loss": 1.2168,
+      "step": 84900
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.531032371623173e-05,
+      "loss": 1.3431,
+      "step": 84910
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.530917578633198e-05,
+      "loss": 1.5262,
+      "step": 84920
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.530802785643223e-05,
+      "loss": 1.2795,
+      "step": 84930
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.530687992653248e-05,
+      "loss": 1.3564,
+      "step": 84940
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.530573199663273e-05,
+      "loss": 1.304,
+      "step": 84950
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.530458406673299e-05,
+      "loss": 1.3797,
+      "step": 84960
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.530343613683324e-05,
+      "loss": 1.3772,
+      "step": 84970
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.530228820693349e-05,
+      "loss": 1.3169,
+      "step": 84980
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.530114027703374e-05,
+      "loss": 1.2589,
+      "step": 84990
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.5299992347134e-05,
+      "loss": 1.433,
+      "step": 85000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.529884441723424e-05,
+      "loss": 1.2035,
+      "step": 85010
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.52976964873345e-05,
+      "loss": 1.3782,
+      "step": 85020
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.529654855743475e-05,
+      "loss": 1.3497,
+      "step": 85030
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.5295400627535e-05,
+      "loss": 1.3589,
+      "step": 85040
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.529425269763525e-05,
+      "loss": 1.3399,
+      "step": 85050
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.529310476773552e-05,
+      "loss": 1.3585,
+      "step": 85060
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.529195683783577e-05,
+      "loss": 1.4118,
+      "step": 85070
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.529080890793602e-05,
+      "loss": 1.3276,
+      "step": 85080
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.528966097803627e-05,
+      "loss": 1.3272,
+      "step": 85090
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.528851304813652e-05,
+      "loss": 1.4657,
+      "step": 85100
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.528736511823678e-05,
+      "loss": 1.2193,
+      "step": 85110
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.528621718833703e-05,
+      "loss": 1.3106,
+      "step": 85120
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.528506925843728e-05,
+      "loss": 1.2928,
+      "step": 85130
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.528392132853753e-05,
+      "loss": 1.272,
+      "step": 85140
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.528277339863779e-05,
+      "loss": 1.2835,
+      "step": 85150
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.528162546873803e-05,
+      "loss": 1.303,
+      "step": 85160
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.528047753883829e-05,
+      "loss": 1.3738,
+      "step": 85170
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.527932960893854e-05,
+      "loss": 1.2119,
+      "step": 85180
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.527818167903879e-05,
+      "loss": 1.4446,
+      "step": 85190
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.527703374913904e-05,
+      "loss": 1.3635,
+      "step": 85200
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.52758858192393e-05,
+      "loss": 1.4305,
+      "step": 85210
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.527473788933955e-05,
+      "loss": 1.4322,
+      "step": 85220
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.52735899594398e-05,
+      "loss": 1.3086,
+      "step": 85230
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.527244202954005e-05,
+      "loss": 1.328,
+      "step": 85240
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.52712940996403e-05,
+      "loss": 1.2112,
+      "step": 85250
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.527014616974057e-05,
+      "loss": 1.4598,
+      "step": 85260
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.526899823984082e-05,
+      "loss": 1.5592,
+      "step": 85270
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.526785030994107e-05,
+      "loss": 1.4559,
+      "step": 85280
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.526670238004132e-05,
+      "loss": 1.3486,
+      "step": 85290
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.526555445014158e-05,
+      "loss": 1.3351,
+      "step": 85300
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.526440652024182e-05,
+      "loss": 1.4377,
+      "step": 85310
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.526325859034208e-05,
+      "loss": 1.3667,
+      "step": 85320
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.526211066044233e-05,
+      "loss": 1.4211,
+      "step": 85330
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.526096273054258e-05,
+      "loss": 1.5185,
+      "step": 85340
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.525981480064283e-05,
+      "loss": 1.2964,
+      "step": 85350
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.525866687074309e-05,
+      "loss": 1.6112,
+      "step": 85360
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.525751894084334e-05,
+      "loss": 1.4246,
+      "step": 85370
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.525637101094359e-05,
+      "loss": 1.2081,
+      "step": 85380
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.525522308104384e-05,
+      "loss": 1.2864,
+      "step": 85390
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.525407515114409e-05,
+      "loss": 1.553,
+      "step": 85400
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.525292722124436e-05,
+      "loss": 1.1993,
+      "step": 85410
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.52517792913446e-05,
+      "loss": 1.5178,
+      "step": 85420
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.525063136144486e-05,
+      "loss": 1.5381,
+      "step": 85430
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.52494834315451e-05,
+      "loss": 1.2017,
+      "step": 85440
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.524833550164537e-05,
+      "loss": 1.2881,
+      "step": 85450
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.52471875717456e-05,
+      "loss": 1.3776,
+      "step": 85460
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.524603964184587e-05,
+      "loss": 1.3174,
+      "step": 85470
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.524489171194612e-05,
+      "loss": 1.5033,
+      "step": 85480
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.524374378204637e-05,
+      "loss": 1.2901,
+      "step": 85490
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.524259585214662e-05,
+      "loss": 1.4088,
+      "step": 85500
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.524144792224688e-05,
+      "loss": 1.4959,
+      "step": 85510
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.524029999234713e-05,
+      "loss": 1.4501,
+      "step": 85520
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.523915206244738e-05,
+      "loss": 1.3323,
+      "step": 85530
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.523800413254763e-05,
+      "loss": 1.471,
+      "step": 85540
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.523685620264788e-05,
+      "loss": 1.3885,
+      "step": 85550
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.523570827274814e-05,
+      "loss": 1.3648,
+      "step": 85560
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.52345603428484e-05,
+      "loss": 1.3549,
+      "step": 85570
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.523341241294864e-05,
+      "loss": 1.3316,
+      "step": 85580
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.52322644830489e-05,
+      "loss": 1.3197,
+      "step": 85590
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.523111655314916e-05,
+      "loss": 1.3037,
+      "step": 85600
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.52299686232494e-05,
+      "loss": 1.1907,
+      "step": 85610
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.522882069334966e-05,
+      "loss": 1.4099,
+      "step": 85620
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.522767276344991e-05,
+      "loss": 1.2855,
+      "step": 85630
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.522652483355016e-05,
+      "loss": 1.3627,
+      "step": 85640
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.52253769036504e-05,
+      "loss": 1.3534,
+      "step": 85650
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.522422897375067e-05,
+      "loss": 1.3684,
+      "step": 85660
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.522308104385092e-05,
+      "loss": 1.4329,
+      "step": 85670
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.522193311395117e-05,
+      "loss": 1.4896,
+      "step": 85680
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.522078518405142e-05,
+      "loss": 1.272,
+      "step": 85690
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.521963725415167e-05,
+      "loss": 1.3584,
+      "step": 85700
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.521848932425193e-05,
+      "loss": 1.4175,
+      "step": 85710
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.521734139435218e-05,
+      "loss": 1.4055,
+      "step": 85720
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.521619346445243e-05,
+      "loss": 1.3399,
+      "step": 85730
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.521504553455268e-05,
+      "loss": 1.3331,
+      "step": 85740
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.521389760465295e-05,
+      "loss": 1.3036,
+      "step": 85750
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.521274967475318e-05,
+      "loss": 1.2382,
+      "step": 85760
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.521160174485345e-05,
+      "loss": 1.319,
+      "step": 85770
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.52104538149537e-05,
+      "loss": 1.2993,
+      "step": 85780
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.520930588505395e-05,
+      "loss": 1.4379,
+      "step": 85790
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.52081579551542e-05,
+      "loss": 1.2852,
+      "step": 85800
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.520701002525446e-05,
+      "loss": 1.3735,
+      "step": 85810
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.520586209535471e-05,
+      "loss": 1.2067,
+      "step": 85820
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.520471416545496e-05,
+      "loss": 1.4518,
+      "step": 85830
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.520356623555521e-05,
+      "loss": 1.3886,
+      "step": 85840
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.520241830565546e-05,
+      "loss": 1.2547,
+      "step": 85850
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.520127037575572e-05,
+      "loss": 1.2671,
+      "step": 85860
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.520012244585597e-05,
+      "loss": 1.3903,
+      "step": 85870
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.519897451595622e-05,
+      "loss": 1.2666,
+      "step": 85880
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.519782658605647e-05,
+      "loss": 1.3127,
+      "step": 85890
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.519667865615673e-05,
+      "loss": 1.4553,
+      "step": 85900
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.519553072625697e-05,
+      "loss": 1.3424,
+      "step": 85910
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.519438279635723e-05,
+      "loss": 1.3512,
+      "step": 85920
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.519323486645748e-05,
+      "loss": 1.4386,
+      "step": 85930
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.519208693655773e-05,
+      "loss": 1.3509,
+      "step": 85940
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.519093900665798e-05,
+      "loss": 1.3293,
+      "step": 85950
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.518979107675825e-05,
+      "loss": 1.3691,
+      "step": 85960
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.51886431468585e-05,
+      "loss": 1.2457,
+      "step": 85970
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.518749521695875e-05,
+      "loss": 1.2831,
+      "step": 85980
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.5186347287059e-05,
+      "loss": 1.3637,
+      "step": 85990
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.518519935715925e-05,
+      "loss": 1.2654,
+      "step": 86000
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.518405142725951e-05,
+      "loss": 1.3872,
+      "step": 86010
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.518290349735976e-05,
+      "loss": 1.3111,
+      "step": 86020
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.518175556746001e-05,
+      "loss": 1.2204,
+      "step": 86030
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.518060763756026e-05,
+      "loss": 1.4288,
+      "step": 86040
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.517945970766052e-05,
+      "loss": 1.3071,
+      "step": 86050
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.517831177776076e-05,
+      "loss": 1.2828,
+      "step": 86060
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.517716384786102e-05,
+      "loss": 1.3497,
+      "step": 86070
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.517601591796127e-05,
+      "loss": 1.3164,
+      "step": 86080
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.517486798806152e-05,
+      "loss": 1.2126,
+      "step": 86090
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.517372005816177e-05,
+      "loss": 1.2948,
+      "step": 86100
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.517257212826204e-05,
+      "loss": 1.3703,
+      "step": 86110
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.517142419836229e-05,
+      "loss": 1.2759,
+      "step": 86120
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.517027626846254e-05,
+      "loss": 1.2743,
+      "step": 86130
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.516912833856278e-05,
+      "loss": 1.2344,
+      "step": 86140
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.516798040866303e-05,
+      "loss": 1.3544,
+      "step": 86150
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.51668324787633e-05,
+      "loss": 1.2837,
+      "step": 86160
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.516568454886355e-05,
+      "loss": 1.3176,
+      "step": 86170
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.51645366189638e-05,
+      "loss": 1.6283,
+      "step": 86180
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.516338868906405e-05,
+      "loss": 1.2687,
+      "step": 86190
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.516224075916431e-05,
+      "loss": 1.2939,
+      "step": 86200
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.516109282926455e-05,
+      "loss": 1.2413,
+      "step": 86210
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.515994489936481e-05,
+      "loss": 1.3194,
+      "step": 86220
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.515879696946506e-05,
+      "loss": 1.421,
+      "step": 86230
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.515764903956531e-05,
+      "loss": 1.2247,
+      "step": 86240
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.515650110966556e-05,
+      "loss": 1.3282,
+      "step": 86250
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.515535317976582e-05,
+      "loss": 1.3511,
+      "step": 86260
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.515420524986607e-05,
+      "loss": 1.3203,
+      "step": 86270
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.515305731996632e-05,
+      "loss": 1.3589,
+      "step": 86280
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.515190939006657e-05,
+      "loss": 1.2996,
+      "step": 86290
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.515076146016682e-05,
+      "loss": 1.4094,
+      "step": 86300
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.514961353026709e-05,
+      "loss": 1.1755,
+      "step": 86310
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.514846560036732e-05,
+      "loss": 1.4674,
+      "step": 86320
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.514731767046759e-05,
+      "loss": 1.2631,
+      "step": 86330
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.514616974056784e-05,
+      "loss": 1.2769,
+      "step": 86340
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.514502181066809e-05,
+      "loss": 1.2842,
+      "step": 86350
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.514387388076834e-05,
+      "loss": 1.384,
+      "step": 86360
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.51427259508686e-05,
+      "loss": 1.547,
+      "step": 86370
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.514157802096885e-05,
+      "loss": 1.3606,
+      "step": 86380
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.51404300910691e-05,
+      "loss": 1.4162,
+      "step": 86390
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.513928216116935e-05,
+      "loss": 1.2616,
+      "step": 86400
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.51381342312696e-05,
+      "loss": 1.4037,
+      "step": 86410
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.513698630136986e-05,
+      "loss": 1.387,
+      "step": 86420
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.513583837147011e-05,
+      "loss": 1.3409,
+      "step": 86430
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.513469044157036e-05,
+      "loss": 1.2036,
+      "step": 86440
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.513354251167061e-05,
+      "loss": 1.4356,
+      "step": 86450
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.513239458177088e-05,
+      "loss": 1.4229,
+      "step": 86460
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.513124665187111e-05,
+      "loss": 1.2448,
+      "step": 86470
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.513009872197137e-05,
+      "loss": 1.2691,
+      "step": 86480
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.512895079207162e-05,
+      "loss": 1.2753,
+      "step": 86490
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.512780286217187e-05,
+      "loss": 1.5345,
+      "step": 86500
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.512665493227212e-05,
+      "loss": 1.348,
+      "step": 86510
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.512550700237239e-05,
+      "loss": 1.332,
+      "step": 86520
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.512435907247264e-05,
+      "loss": 1.2957,
+      "step": 86530
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.512321114257289e-05,
+      "loss": 1.2959,
+      "step": 86540
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.512206321267314e-05,
+      "loss": 1.2121,
+      "step": 86550
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.512091528277339e-05,
+      "loss": 1.2635,
+      "step": 86560
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.511976735287365e-05,
+      "loss": 1.4926,
+      "step": 86570
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.51186194229739e-05,
+      "loss": 1.2696,
+      "step": 86580
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.511747149307415e-05,
+      "loss": 1.3212,
+      "step": 86590
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.51163235631744e-05,
+      "loss": 1.2879,
+      "step": 86600
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.511517563327466e-05,
+      "loss": 1.2717,
+      "step": 86610
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.51140277033749e-05,
+      "loss": 1.4772,
+      "step": 86620
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.511287977347516e-05,
+      "loss": 1.4877,
+      "step": 86630
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.511173184357541e-05,
+      "loss": 1.2924,
+      "step": 86640
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.511058391367566e-05,
+      "loss": 1.5321,
+      "step": 86650
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.510943598377591e-05,
+      "loss": 1.3306,
+      "step": 86660
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.510828805387618e-05,
+      "loss": 1.2769,
+      "step": 86670
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.510714012397643e-05,
+      "loss": 1.278,
+      "step": 86680
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.510599219407668e-05,
+      "loss": 1.1643,
+      "step": 86690
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.510484426417693e-05,
+      "loss": 1.4772,
+      "step": 86700
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.510369633427718e-05,
+      "loss": 1.2201,
+      "step": 86710
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.510254840437744e-05,
+      "loss": 1.3809,
+      "step": 86720
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.510140047447769e-05,
+      "loss": 1.2568,
+      "step": 86730
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.510025254457794e-05,
+      "loss": 1.3646,
+      "step": 86740
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.509910461467819e-05,
+      "loss": 1.4094,
+      "step": 86750
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.509795668477845e-05,
+      "loss": 1.3432,
+      "step": 86760
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.509680875487869e-05,
+      "loss": 1.278,
+      "step": 86770
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.509566082497895e-05,
+      "loss": 1.2847,
+      "step": 86780
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.50945128950792e-05,
+      "loss": 1.4653,
+      "step": 86790
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.509336496517945e-05,
+      "loss": 1.4863,
+      "step": 86800
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.50922170352797e-05,
+      "loss": 1.2917,
+      "step": 86810
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.509106910537996e-05,
+      "loss": 1.2827,
+      "step": 86820
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.508992117548021e-05,
+      "loss": 1.3131,
+      "step": 86830
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.508877324558046e-05,
+      "loss": 1.3388,
+      "step": 86840
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.508762531568071e-05,
+      "loss": 1.362,
+      "step": 86850
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.508647738578096e-05,
+      "loss": 1.2474,
+      "step": 86860
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.508532945588123e-05,
+      "loss": 1.4241,
+      "step": 86870
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.508418152598148e-05,
+      "loss": 1.3836,
+      "step": 86880
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.508303359608173e-05,
+      "loss": 1.1723,
+      "step": 86890
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.508188566618198e-05,
+      "loss": 1.3471,
+      "step": 86900
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.508073773628224e-05,
+      "loss": 1.2904,
+      "step": 86910
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.507958980638248e-05,
+      "loss": 1.3079,
+      "step": 86920
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.507844187648274e-05,
+      "loss": 1.4485,
+      "step": 86930
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.507729394658299e-05,
+      "loss": 1.3778,
+      "step": 86940
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.507614601668324e-05,
+      "loss": 1.3698,
+      "step": 86950
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.507499808678349e-05,
+      "loss": 1.2759,
+      "step": 86960
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.507385015688375e-05,
+      "loss": 1.4666,
+      "step": 86970
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.5072702226984e-05,
+      "loss": 1.327,
+      "step": 86980
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.507155429708425e-05,
+      "loss": 1.2605,
+      "step": 86990
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.50704063671845e-05,
+      "loss": 1.3337,
+      "step": 87000
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.506925843728475e-05,
+      "loss": 1.3249,
+      "step": 87010
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.506811050738502e-05,
+      "loss": 1.3063,
+      "step": 87020
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.506696257748527e-05,
+      "loss": 1.4413,
+      "step": 87030
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.506581464758552e-05,
+      "loss": 1.2335,
+      "step": 87040
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.506466671768577e-05,
+      "loss": 1.3239,
+      "step": 87050
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.506351878778603e-05,
+      "loss": 1.4174,
+      "step": 87060
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.506237085788627e-05,
+      "loss": 1.2675,
+      "step": 87070
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.506122292798653e-05,
+      "loss": 1.2775,
+      "step": 87080
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.506007499808678e-05,
+      "loss": 1.4267,
+      "step": 87090
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.505892706818703e-05,
+      "loss": 1.4895,
+      "step": 87100
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.505777913828728e-05,
+      "loss": 1.3157,
+      "step": 87110
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.505663120838754e-05,
+      "loss": 1.3873,
+      "step": 87120
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.505548327848779e-05,
+      "loss": 1.2276,
+      "step": 87130
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.505433534858804e-05,
+      "loss": 1.1911,
+      "step": 87140
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.505318741868829e-05,
+      "loss": 1.5425,
+      "step": 87150
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.505203948878854e-05,
+      "loss": 1.2819,
+      "step": 87160
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.50508915588888e-05,
+      "loss": 1.3939,
+      "step": 87170
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.504974362898905e-05,
+      "loss": 1.2011,
+      "step": 87180
+    },
+    {
+      "epoch": 4.0,
+      "eval_cer": 0.671274588332702,
+      "eval_em": 0.007357212003872217,
+      "eval_f1": 0.007357212003872217,
+      "eval_loss": 1.2519690990447998,
+      "eval_runtime": 2213.4388,
+      "eval_samples_per_second": 4.667,
+      "eval_steps_per_second": 0.584,
+      "eval_wer": 0.9926427879961278,
+      "step": 87180
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.50485956990893e-05,
+      "loss": 1.4035,
+      "step": 87190
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.504744776918955e-05,
+      "loss": 1.3092,
+      "step": 87200
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.504629983928982e-05,
+      "loss": 1.1963,
+      "step": 87210
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.504515190939005e-05,
+      "loss": 1.44,
+      "step": 87220
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.504400397949032e-05,
+      "loss": 1.3867,
+      "step": 87230
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.504285604959057e-05,
+      "loss": 1.3175,
+      "step": 87240
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.504170811969082e-05,
+      "loss": 1.4491,
+      "step": 87250
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.504056018979107e-05,
+      "loss": 1.2776,
+      "step": 87260
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.503941225989133e-05,
+      "loss": 1.2797,
+      "step": 87270
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.503826432999158e-05,
+      "loss": 1.2738,
+      "step": 87280
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.503711640009183e-05,
+      "loss": 1.3575,
+      "step": 87290
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.503596847019208e-05,
+      "loss": 1.2012,
+      "step": 87300
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.503482054029233e-05,
+      "loss": 1.3299,
+      "step": 87310
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.503367261039259e-05,
+      "loss": 1.3477,
+      "step": 87320
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.503252468049284e-05,
+      "loss": 1.395,
+      "step": 87330
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.503137675059309e-05,
+      "loss": 1.2226,
+      "step": 87340
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.503022882069334e-05,
+      "loss": 1.2083,
+      "step": 87350
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.50290808907936e-05,
+      "loss": 1.4433,
+      "step": 87360
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.502793296089384e-05,
+      "loss": 1.263,
+      "step": 87370
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.50267850309941e-05,
+      "loss": 1.3756,
+      "step": 87380
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.502563710109436e-05,
+      "loss": 1.3761,
+      "step": 87390
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.50244891711946e-05,
+      "loss": 1.5534,
+      "step": 87400
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.502334124129486e-05,
+      "loss": 1.2882,
+      "step": 87410
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.502219331139512e-05,
+      "loss": 1.5829,
+      "step": 87420
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.502104538149537e-05,
+      "loss": 1.2882,
+      "step": 87430
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.501989745159562e-05,
+      "loss": 1.3306,
+      "step": 87440
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.501874952169587e-05,
+      "loss": 1.3861,
+      "step": 87450
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.501760159179612e-05,
+      "loss": 1.3333,
+      "step": 87460
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.501645366189638e-05,
+      "loss": 1.2733,
+      "step": 87470
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.501530573199663e-05,
+      "loss": 1.3608,
+      "step": 87480
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.501415780209688e-05,
+      "loss": 1.2823,
+      "step": 87490
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.501300987219713e-05,
+      "loss": 1.3738,
+      "step": 87500
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.50118619422974e-05,
+      "loss": 1.4585,
+      "step": 87510
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.501071401239763e-05,
+      "loss": 1.1724,
+      "step": 87520
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.50095660824979e-05,
+      "loss": 1.0986,
+      "step": 87530
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.500841815259814e-05,
+      "loss": 1.3742,
+      "step": 87540
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.50072702226984e-05,
+      "loss": 1.2982,
+      "step": 87550
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.500612229279864e-05,
+      "loss": 1.3994,
+      "step": 87560
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.500497436289891e-05,
+      "loss": 1.3961,
+      "step": 87570
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.500382643299916e-05,
+      "loss": 1.3271,
+      "step": 87580
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.500267850309941e-05,
+      "loss": 1.284,
+      "step": 87590
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.500153057319966e-05,
+      "loss": 1.2666,
+      "step": 87600
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.50003826432999e-05,
+      "loss": 1.5081,
+      "step": 87610
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.499923471340017e-05,
+      "loss": 1.3851,
+      "step": 87620
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.499808678350042e-05,
+      "loss": 1.3789,
+      "step": 87630
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.499693885360067e-05,
+      "loss": 1.3548,
+      "step": 87640
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.499579092370092e-05,
+      "loss": 1.1475,
+      "step": 87650
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.499464299380117e-05,
+      "loss": 1.3259,
+      "step": 87660
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.499349506390142e-05,
+      "loss": 1.2303,
+      "step": 87670
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.499234713400168e-05,
+      "loss": 1.1958,
+      "step": 87680
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.499119920410193e-05,
+      "loss": 1.3684,
+      "step": 87690
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.499005127420218e-05,
+      "loss": 1.3985,
+      "step": 87700
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.498890334430243e-05,
+      "loss": 1.438,
+      "step": 87710
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.498775541440268e-05,
+      "loss": 1.2704,
+      "step": 87720
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.498660748450295e-05,
+      "loss": 1.2988,
+      "step": 87730
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.49854595546032e-05,
+      "loss": 1.3111,
+      "step": 87740
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.498431162470345e-05,
+      "loss": 1.3368,
+      "step": 87750
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.49831636948037e-05,
+      "loss": 1.3513,
+      "step": 87760
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.498201576490396e-05,
+      "loss": 1.3699,
+      "step": 87770
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.49808678350042e-05,
+      "loss": 1.3403,
+      "step": 87780
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.497971990510446e-05,
+      "loss": 1.2512,
+      "step": 87790
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.497857197520471e-05,
+      "loss": 1.237,
+      "step": 87800
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.497742404530496e-05,
+      "loss": 1.3651,
+      "step": 87810
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.497627611540521e-05,
+      "loss": 1.2272,
+      "step": 87820
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.497512818550547e-05,
+      "loss": 1.2659,
+      "step": 87830
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.497398025560572e-05,
+      "loss": 1.2397,
+      "step": 87840
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.497283232570597e-05,
+      "loss": 1.3046,
+      "step": 87850
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.497168439580622e-05,
+      "loss": 1.1875,
+      "step": 87860
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.497053646590647e-05,
+      "loss": 1.5004,
+      "step": 87870
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.496938853600673e-05,
+      "loss": 1.2966,
+      "step": 87880
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.496824060610698e-05,
+      "loss": 1.3344,
+      "step": 87890
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.496709267620723e-05,
+      "loss": 1.2278,
+      "step": 87900
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.496594474630748e-05,
+      "loss": 1.272,
+      "step": 87910
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.496479681640775e-05,
+      "loss": 1.2154,
+      "step": 87920
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.496364888650798e-05,
+      "loss": 1.4654,
+      "step": 87930
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.496250095660825e-05,
+      "loss": 1.2718,
+      "step": 87940
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.49613530267085e-05,
+      "loss": 1.2419,
+      "step": 87950
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.496020509680875e-05,
+      "loss": 1.2107,
+      "step": 87960
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.4959057166909e-05,
+      "loss": 1.3935,
+      "step": 87970
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.495790923700926e-05,
+      "loss": 1.2291,
+      "step": 87980
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.495676130710951e-05,
+      "loss": 1.1448,
+      "step": 87990
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.495561337720976e-05,
+      "loss": 1.2527,
+      "step": 88000
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.495446544731001e-05,
+      "loss": 1.2485,
+      "step": 88010
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.495331751741026e-05,
+      "loss": 1.3711,
+      "step": 88020
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.495216958751052e-05,
+      "loss": 1.3987,
+      "step": 88030
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.495102165761077e-05,
+      "loss": 1.3274,
+      "step": 88040
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.494987372771102e-05,
+      "loss": 1.1464,
+      "step": 88050
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.494872579781127e-05,
+      "loss": 1.4549,
+      "step": 88060
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.494757786791154e-05,
+      "loss": 1.3781,
+      "step": 88070
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.494642993801177e-05,
+      "loss": 1.2128,
+      "step": 88080
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.494528200811204e-05,
+      "loss": 1.2359,
+      "step": 88090
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.494413407821228e-05,
+      "loss": 1.2634,
+      "step": 88100
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.494298614831253e-05,
+      "loss": 1.2938,
+      "step": 88110
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.494183821841278e-05,
+      "loss": 1.3556,
+      "step": 88120
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.494069028851305e-05,
+      "loss": 1.1423,
+      "step": 88130
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.49395423586133e-05,
+      "loss": 1.3563,
+      "step": 88140
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.493839442871355e-05,
+      "loss": 1.3915,
+      "step": 88150
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.49372464988138e-05,
+      "loss": 1.3538,
+      "step": 88160
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.493609856891405e-05,
+      "loss": 1.3049,
+      "step": 88170
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.493495063901431e-05,
+      "loss": 1.0926,
+      "step": 88180
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.493380270911456e-05,
+      "loss": 1.2114,
+      "step": 88190
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.493265477921481e-05,
+      "loss": 1.3465,
+      "step": 88200
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.493150684931506e-05,
+      "loss": 1.3637,
+      "step": 88210
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.493035891941532e-05,
+      "loss": 1.4429,
+      "step": 88220
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.492921098951556e-05,
+      "loss": 1.21,
+      "step": 88230
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.492806305961582e-05,
+      "loss": 1.3799,
+      "step": 88240
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.492691512971607e-05,
+      "loss": 1.4629,
+      "step": 88250
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.492576719981632e-05,
+      "loss": 1.2125,
+      "step": 88260
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.492461926991657e-05,
+      "loss": 1.2968,
+      "step": 88270
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.492347134001684e-05,
+      "loss": 1.3691,
+      "step": 88280
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.492232341011709e-05,
+      "loss": 1.1929,
+      "step": 88290
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.492117548021734e-05,
+      "loss": 1.4259,
+      "step": 88300
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.492002755031759e-05,
+      "loss": 1.4162,
+      "step": 88310
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.491887962041784e-05,
+      "loss": 1.3076,
+      "step": 88320
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.49177316905181e-05,
+      "loss": 1.2705,
+      "step": 88330
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.491658376061835e-05,
+      "loss": 1.2147,
+      "step": 88340
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.49154358307186e-05,
+      "loss": 1.4201,
+      "step": 88350
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.491428790081885e-05,
+      "loss": 1.2667,
+      "step": 88360
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.491313997091911e-05,
+      "loss": 1.2802,
+      "step": 88370
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.491199204101935e-05,
+      "loss": 1.2584,
+      "step": 88380
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.491084411111961e-05,
+      "loss": 1.3242,
+      "step": 88390
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.490969618121986e-05,
+      "loss": 1.2345,
+      "step": 88400
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.490854825132011e-05,
+      "loss": 1.1342,
+      "step": 88410
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.490740032142036e-05,
+      "loss": 1.2159,
+      "step": 88420
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.490625239152062e-05,
+      "loss": 1.3935,
+      "step": 88430
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.490510446162087e-05,
+      "loss": 1.2182,
+      "step": 88440
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.490395653172112e-05,
+      "loss": 1.4519,
+      "step": 88450
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.490280860182137e-05,
+      "loss": 1.2083,
+      "step": 88460
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.490166067192162e-05,
+      "loss": 1.3513,
+      "step": 88470
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.490051274202189e-05,
+      "loss": 1.296,
+      "step": 88480
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.489936481212214e-05,
+      "loss": 1.3974,
+      "step": 88490
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.489821688222239e-05,
+      "loss": 1.2954,
+      "step": 88500
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.489706895232264e-05,
+      "loss": 1.2909,
+      "step": 88510
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.48959210224229e-05,
+      "loss": 1.3242,
+      "step": 88520
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.489477309252314e-05,
+      "loss": 1.4661,
+      "step": 88530
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.48936251626234e-05,
+      "loss": 1.2796,
+      "step": 88540
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.489247723272365e-05,
+      "loss": 1.3629,
+      "step": 88550
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.48913293028239e-05,
+      "loss": 1.2835,
+      "step": 88560
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.489018137292415e-05,
+      "loss": 1.4554,
+      "step": 88570
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.488903344302441e-05,
+      "loss": 1.2336,
+      "step": 88580
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.488788551312466e-05,
+      "loss": 1.3827,
+      "step": 88590
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.488673758322491e-05,
+      "loss": 1.2706,
+      "step": 88600
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.488558965332516e-05,
+      "loss": 1.3626,
+      "step": 88610
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.488444172342541e-05,
+      "loss": 1.3337,
+      "step": 88620
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.488329379352568e-05,
+      "loss": 1.3203,
+      "step": 88630
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.488214586362593e-05,
+      "loss": 1.3688,
+      "step": 88640
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.488099793372618e-05,
+      "loss": 1.3068,
+      "step": 88650
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.487985000382643e-05,
+      "loss": 1.3151,
+      "step": 88660
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.487870207392669e-05,
+      "loss": 1.1995,
+      "step": 88670
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.487755414402693e-05,
+      "loss": 1.2432,
+      "step": 88680
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.487640621412719e-05,
+      "loss": 1.4079,
+      "step": 88690
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.487525828422744e-05,
+      "loss": 1.2755,
+      "step": 88700
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.487411035432769e-05,
+      "loss": 1.1614,
+      "step": 88710
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.487296242442794e-05,
+      "loss": 1.2893,
+      "step": 88720
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.48718144945282e-05,
+      "loss": 1.2681,
+      "step": 88730
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.487066656462845e-05,
+      "loss": 1.4133,
+      "step": 88740
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.48695186347287e-05,
+      "loss": 1.3992,
+      "step": 88750
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.486837070482895e-05,
+      "loss": 1.2732,
+      "step": 88760
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.48672227749292e-05,
+      "loss": 1.2782,
+      "step": 88770
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.486607484502946e-05,
+      "loss": 1.3379,
+      "step": 88780
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.486492691512971e-05,
+      "loss": 1.2055,
+      "step": 88790
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.486377898522996e-05,
+      "loss": 1.3586,
+      "step": 88800
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.486263105533021e-05,
+      "loss": 1.343,
+      "step": 88810
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.486148312543048e-05,
+      "loss": 1.1689,
+      "step": 88820
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.486033519553071e-05,
+      "loss": 1.3571,
+      "step": 88830
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.485918726563098e-05,
+      "loss": 1.4438,
+      "step": 88840
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.485803933573123e-05,
+      "loss": 1.2983,
+      "step": 88850
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.485689140583148e-05,
+      "loss": 1.1632,
+      "step": 88860
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.485574347593173e-05,
+      "loss": 1.2476,
+      "step": 88870
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.485459554603199e-05,
+      "loss": 1.1607,
+      "step": 88880
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.485344761613224e-05,
+      "loss": 1.1397,
+      "step": 88890
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.485229968623249e-05,
+      "loss": 1.3031,
+      "step": 88900
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.485115175633274e-05,
+      "loss": 1.4479,
+      "step": 88910
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.485000382643299e-05,
+      "loss": 1.2282,
+      "step": 88920
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.484885589653325e-05,
+      "loss": 1.2611,
+      "step": 88930
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.48477079666335e-05,
+      "loss": 1.2653,
+      "step": 88940
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.484656003673375e-05,
+      "loss": 1.2623,
+      "step": 88950
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.4845412106834e-05,
+      "loss": 1.3364,
+      "step": 88960
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.484426417693427e-05,
+      "loss": 1.2322,
+      "step": 88970
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.48431162470345e-05,
+      "loss": 1.3333,
+      "step": 88980
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.484196831713477e-05,
+      "loss": 1.4001,
+      "step": 88990
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.484082038723502e-05,
+      "loss": 1.2731,
+      "step": 89000
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.483967245733527e-05,
+      "loss": 1.2765,
+      "step": 89010
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.483852452743552e-05,
+      "loss": 1.2469,
+      "step": 89020
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.483737659753578e-05,
+      "loss": 1.2916,
+      "step": 89030
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.483622866763603e-05,
+      "loss": 1.2901,
+      "step": 89040
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.483508073773628e-05,
+      "loss": 1.4209,
+      "step": 89050
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.483393280783653e-05,
+      "loss": 1.6937,
+      "step": 89060
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.483278487793678e-05,
+      "loss": 1.3918,
+      "step": 89070
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.483163694803704e-05,
+      "loss": 1.3956,
+      "step": 89080
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.483048901813728e-05,
+      "loss": 1.3521,
+      "step": 89090
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.482934108823754e-05,
+      "loss": 1.3017,
+      "step": 89100
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.482819315833779e-05,
+      "loss": 1.3321,
+      "step": 89110
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.482704522843804e-05,
+      "loss": 1.4419,
+      "step": 89120
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.482589729853829e-05,
+      "loss": 1.314,
+      "step": 89130
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.482474936863855e-05,
+      "loss": 1.2053,
+      "step": 89140
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.48236014387388e-05,
+      "loss": 1.1405,
+      "step": 89150
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.482245350883905e-05,
+      "loss": 1.2829,
+      "step": 89160
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.48213055789393e-05,
+      "loss": 1.1953,
+      "step": 89170
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.482015764903955e-05,
+      "loss": 1.2036,
+      "step": 89180
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.481900971913982e-05,
+      "loss": 1.481,
+      "step": 89190
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.481786178924007e-05,
+      "loss": 1.2258,
+      "step": 89200
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.481671385934032e-05,
+      "loss": 1.4165,
+      "step": 89210
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.481556592944057e-05,
+      "loss": 1.4659,
+      "step": 89220
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.481441799954083e-05,
+      "loss": 1.2653,
+      "step": 89230
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.481327006964107e-05,
+      "loss": 1.2873,
+      "step": 89240
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.481212213974133e-05,
+      "loss": 1.3467,
+      "step": 89250
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.481097420984158e-05,
+      "loss": 1.3159,
+      "step": 89260
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.480982627994183e-05,
+      "loss": 1.233,
+      "step": 89270
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.480867835004208e-05,
+      "loss": 1.2724,
+      "step": 89280
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.480753042014234e-05,
+      "loss": 1.3382,
+      "step": 89290
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.480638249024259e-05,
+      "loss": 1.4139,
+      "step": 89300
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.480523456034284e-05,
+      "loss": 1.2475,
+      "step": 89310
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.480408663044309e-05,
+      "loss": 1.3901,
+      "step": 89320
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.480293870054334e-05,
+      "loss": 1.3074,
+      "step": 89330
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.48017907706436e-05,
+      "loss": 1.363,
+      "step": 89340
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.480064284074386e-05,
+      "loss": 1.2748,
+      "step": 89350
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.47994949108441e-05,
+      "loss": 1.2152,
+      "step": 89360
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.479834698094436e-05,
+      "loss": 1.2523,
+      "step": 89370
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.479719905104462e-05,
+      "loss": 1.3022,
+      "step": 89380
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.479605112114485e-05,
+      "loss": 1.2697,
+      "step": 89390
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.479490319124512e-05,
+      "loss": 1.2983,
+      "step": 89400
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.479375526134537e-05,
+      "loss": 1.5371,
+      "step": 89410
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.479260733144562e-05,
+      "loss": 1.4363,
+      "step": 89420
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.479145940154587e-05,
+      "loss": 1.2186,
+      "step": 89430
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.479031147164613e-05,
+      "loss": 1.3931,
+      "step": 89440
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.478916354174638e-05,
+      "loss": 1.3525,
+      "step": 89450
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.478801561184663e-05,
+      "loss": 1.4046,
+      "step": 89460
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.478686768194688e-05,
+      "loss": 1.4661,
+      "step": 89470
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.478571975204713e-05,
+      "loss": 1.2009,
+      "step": 89480
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.47845718221474e-05,
+      "loss": 1.3634,
+      "step": 89490
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.478342389224764e-05,
+      "loss": 1.4726,
+      "step": 89500
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.47822759623479e-05,
+      "loss": 1.1934,
+      "step": 89510
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.478112803244814e-05,
+      "loss": 1.2893,
+      "step": 89520
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.477998010254841e-05,
+      "loss": 1.4432,
+      "step": 89530
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.477883217264864e-05,
+      "loss": 1.3973,
+      "step": 89540
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.477768424274891e-05,
+      "loss": 1.3358,
+      "step": 89550
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.477653631284916e-05,
+      "loss": 1.4202,
+      "step": 89560
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.47753883829494e-05,
+      "loss": 1.2429,
+      "step": 89570
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.477424045304966e-05,
+      "loss": 1.2358,
+      "step": 89580
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.477309252314992e-05,
+      "loss": 1.3561,
+      "step": 89590
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.477194459325017e-05,
+      "loss": 1.294,
+      "step": 89600
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.477079666335042e-05,
+      "loss": 1.3789,
+      "step": 89610
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.476964873345067e-05,
+      "loss": 1.5051,
+      "step": 89620
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.476850080355092e-05,
+      "loss": 1.2974,
+      "step": 89630
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.476735287365118e-05,
+      "loss": 1.369,
+      "step": 89640
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.476620494375143e-05,
+      "loss": 1.3141,
+      "step": 89650
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.476505701385168e-05,
+      "loss": 1.2854,
+      "step": 89660
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.476390908395193e-05,
+      "loss": 1.5218,
+      "step": 89670
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.47627611540522e-05,
+      "loss": 1.2929,
+      "step": 89680
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.476161322415243e-05,
+      "loss": 1.3271,
+      "step": 89690
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.47604652942527e-05,
+      "loss": 1.334,
+      "step": 89700
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.475931736435295e-05,
+      "loss": 1.1987,
+      "step": 89710
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.47581694344532e-05,
+      "loss": 1.4156,
+      "step": 89720
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.475702150455344e-05,
+      "loss": 1.2938,
+      "step": 89730
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.475587357465371e-05,
+      "loss": 1.4165,
+      "step": 89740
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.475472564475396e-05,
+      "loss": 1.3213,
+      "step": 89750
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.475357771485421e-05,
+      "loss": 1.2865,
+      "step": 89760
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.475242978495446e-05,
+      "loss": 1.3568,
+      "step": 89770
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.475128185505471e-05,
+      "loss": 1.3037,
+      "step": 89780
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.475013392515497e-05,
+      "loss": 1.3517,
+      "step": 89790
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.474898599525522e-05,
+      "loss": 1.341,
+      "step": 89800
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.474783806535547e-05,
+      "loss": 1.2959,
+      "step": 89810
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.474669013545572e-05,
+      "loss": 1.2525,
+      "step": 89820
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.474554220555598e-05,
+      "loss": 1.3294,
+      "step": 89830
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.474439427565622e-05,
+      "loss": 1.297,
+      "step": 89840
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.474324634575648e-05,
+      "loss": 1.3344,
+      "step": 89850
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.474209841585673e-05,
+      "loss": 1.394,
+      "step": 89860
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.474095048595698e-05,
+      "loss": 1.2883,
+      "step": 89870
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.473980255605723e-05,
+      "loss": 1.3636,
+      "step": 89880
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.47386546261575e-05,
+      "loss": 1.2763,
+      "step": 89890
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.473750669625775e-05,
+      "loss": 1.3074,
+      "step": 89900
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.4736358766358e-05,
+      "loss": 1.2214,
+      "step": 89910
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.473521083645825e-05,
+      "loss": 1.2813,
+      "step": 89920
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.47340629065585e-05,
+      "loss": 1.3113,
+      "step": 89930
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.473291497665876e-05,
+      "loss": 1.1724,
+      "step": 89940
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.473176704675901e-05,
+      "loss": 1.3593,
+      "step": 89950
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.473061911685926e-05,
+      "loss": 1.2624,
+      "step": 89960
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.472947118695951e-05,
+      "loss": 1.2846,
+      "step": 89970
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.472832325705977e-05,
+      "loss": 1.3238,
+      "step": 89980
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.472717532716001e-05,
+      "loss": 1.2684,
+      "step": 89990
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.472602739726027e-05,
+      "loss": 1.3693,
+      "step": 90000
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.472487946736052e-05,
+      "loss": 1.1286,
+      "step": 90010
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.472373153746077e-05,
+      "loss": 1.3208,
+      "step": 90020
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.472258360756102e-05,
+      "loss": 1.2672,
+      "step": 90030
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.472143567766129e-05,
+      "loss": 1.24,
+      "step": 90040
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.472028774776154e-05,
+      "loss": 1.3886,
+      "step": 90050
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.471913981786178e-05,
+      "loss": 1.2486,
+      "step": 90060
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.471799188796203e-05,
+      "loss": 1.2794,
+      "step": 90070
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.471684395806228e-05,
+      "loss": 1.22,
+      "step": 90080
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.471569602816255e-05,
+      "loss": 1.3118,
+      "step": 90090
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.47145480982628e-05,
+      "loss": 1.4275,
+      "step": 90100
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.471340016836305e-05,
+      "loss": 1.2076,
+      "step": 90110
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.47122522384633e-05,
+      "loss": 1.2567,
+      "step": 90120
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.471110430856356e-05,
+      "loss": 1.4126,
+      "step": 90130
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.47099563786638e-05,
+      "loss": 1.2301,
+      "step": 90140
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.470880844876406e-05,
+      "loss": 1.4204,
+      "step": 90150
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.470766051886431e-05,
+      "loss": 1.1456,
+      "step": 90160
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.470651258896456e-05,
+      "loss": 1.1458,
+      "step": 90170
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.470536465906481e-05,
+      "loss": 1.3119,
+      "step": 90180
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.470421672916507e-05,
+      "loss": 1.3908,
+      "step": 90190
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.470306879926532e-05,
+      "loss": 1.252,
+      "step": 90200
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.470192086936557e-05,
+      "loss": 1.3503,
+      "step": 90210
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.470077293946582e-05,
+      "loss": 1.1897,
+      "step": 90220
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.469962500956607e-05,
+      "loss": 1.3116,
+      "step": 90230
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.469847707966634e-05,
+      "loss": 1.2851,
+      "step": 90240
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.469732914976659e-05,
+      "loss": 1.2396,
+      "step": 90250
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.469618121986684e-05,
+      "loss": 1.3108,
+      "step": 90260
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.469503328996709e-05,
+      "loss": 1.194,
+      "step": 90270
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.469388536006735e-05,
+      "loss": 1.203,
+      "step": 90280
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.469273743016759e-05,
+      "loss": 1.2143,
+      "step": 90290
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.469158950026785e-05,
+      "loss": 1.304,
+      "step": 90300
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.46904415703681e-05,
+      "loss": 1.3109,
+      "step": 90310
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.468929364046835e-05,
+      "loss": 1.2507,
+      "step": 90320
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.46881457105686e-05,
+      "loss": 1.5387,
+      "step": 90330
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.468699778066886e-05,
+      "loss": 1.2797,
+      "step": 90340
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.468584985076911e-05,
+      "loss": 1.4562,
+      "step": 90350
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.468470192086936e-05,
+      "loss": 1.3696,
+      "step": 90360
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.468355399096961e-05,
+      "loss": 1.2245,
+      "step": 90370
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.468240606106986e-05,
+      "loss": 1.3653,
+      "step": 90380
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.468125813117012e-05,
+      "loss": 1.205,
+      "step": 90390
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.468011020127037e-05,
+      "loss": 1.301,
+      "step": 90400
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.467896227137062e-05,
+      "loss": 1.3399,
+      "step": 90410
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.467781434147087e-05,
+      "loss": 1.1966,
+      "step": 90420
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.467666641157112e-05,
+      "loss": 1.3619,
+      "step": 90430
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.467551848167137e-05,
+      "loss": 1.3892,
+      "step": 90440
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.467437055177164e-05,
+      "loss": 1.3778,
+      "step": 90450
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.467322262187189e-05,
+      "loss": 1.2606,
+      "step": 90460
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.467207469197214e-05,
+      "loss": 1.3911,
+      "step": 90470
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.467092676207239e-05,
+      "loss": 1.3175,
+      "step": 90480
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.466977883217264e-05,
+      "loss": 1.3445,
+      "step": 90490
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.46686309022729e-05,
+      "loss": 1.3563,
+      "step": 90500
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.466748297237315e-05,
+      "loss": 1.3602,
+      "step": 90510
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.46663350424734e-05,
+      "loss": 1.3981,
+      "step": 90520
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.466518711257365e-05,
+      "loss": 1.2534,
+      "step": 90530
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.466403918267391e-05,
+      "loss": 1.4499,
+      "step": 90540
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.466289125277415e-05,
+      "loss": 1.2905,
+      "step": 90550
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.466174332287441e-05,
+      "loss": 1.227,
+      "step": 90560
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.466059539297466e-05,
+      "loss": 1.222,
+      "step": 90570
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.465944746307491e-05,
+      "loss": 1.1844,
+      "step": 90580
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.465829953317516e-05,
+      "loss": 1.4223,
+      "step": 90590
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.465715160327543e-05,
+      "loss": 1.2509,
+      "step": 90600
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.465600367337568e-05,
+      "loss": 1.2332,
+      "step": 90610
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.465485574347593e-05,
+      "loss": 1.3048,
+      "step": 90620
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.465370781357618e-05,
+      "loss": 1.3065,
+      "step": 90630
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.465255988367643e-05,
+      "loss": 1.2908,
+      "step": 90640
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.465141195377669e-05,
+      "loss": 1.2296,
+      "step": 90650
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.465026402387694e-05,
+      "loss": 1.2475,
+      "step": 90660
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.464911609397719e-05,
+      "loss": 1.253,
+      "step": 90670
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.464796816407744e-05,
+      "loss": 1.3333,
+      "step": 90680
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.46468202341777e-05,
+      "loss": 1.2215,
+      "step": 90690
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.464567230427794e-05,
+      "loss": 1.3444,
+      "step": 90700
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.46445243743782e-05,
+      "loss": 1.4946,
+      "step": 90710
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.464337644447845e-05,
+      "loss": 1.34,
+      "step": 90720
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.46422285145787e-05,
+      "loss": 1.2944,
+      "step": 90730
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.464108058467895e-05,
+      "loss": 1.1336,
+      "step": 90740
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.463993265477921e-05,
+      "loss": 1.3202,
+      "step": 90750
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.463878472487946e-05,
+      "loss": 1.2397,
+      "step": 90760
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.463763679497971e-05,
+      "loss": 1.3702,
+      "step": 90770
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.463648886507996e-05,
+      "loss": 1.2712,
+      "step": 90780
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.463534093518021e-05,
+      "loss": 1.1632,
+      "step": 90790
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.463419300528048e-05,
+      "loss": 1.3015,
+      "step": 90800
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.463304507538073e-05,
+      "loss": 1.2814,
+      "step": 90810
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.463189714548098e-05,
+      "loss": 1.4189,
+      "step": 90820
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.463074921558123e-05,
+      "loss": 1.3786,
+      "step": 90830
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.462960128568149e-05,
+      "loss": 1.2394,
+      "step": 90840
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.462845335578173e-05,
+      "loss": 1.2115,
+      "step": 90850
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.462730542588199e-05,
+      "loss": 1.3425,
+      "step": 90860
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.462615749598224e-05,
+      "loss": 1.2711,
+      "step": 90870
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.462500956608249e-05,
+      "loss": 1.3887,
+      "step": 90880
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.462386163618274e-05,
+      "loss": 1.3117,
+      "step": 90890
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.4622713706283e-05,
+      "loss": 1.263,
+      "step": 90900
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.462156577638325e-05,
+      "loss": 1.4858,
+      "step": 90910
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.46204178464835e-05,
+      "loss": 1.2451,
+      "step": 90920
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.461926991658375e-05,
+      "loss": 1.3276,
+      "step": 90930
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.4618121986684e-05,
+      "loss": 1.3292,
+      "step": 90940
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.461697405678427e-05,
+      "loss": 1.2685,
+      "step": 90950
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.461582612688452e-05,
+      "loss": 1.4931,
+      "step": 90960
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.461467819698477e-05,
+      "loss": 1.3219,
+      "step": 90970
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.461353026708502e-05,
+      "loss": 1.2719,
+      "step": 90980
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.461238233718528e-05,
+      "loss": 1.4034,
+      "step": 90990
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.461123440728552e-05,
+      "loss": 1.2453,
+      "step": 91000
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.461008647738578e-05,
+      "loss": 1.1563,
+      "step": 91010
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.460893854748603e-05,
+      "loss": 1.2238,
+      "step": 91020
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.460779061758628e-05,
+      "loss": 1.1974,
+      "step": 91030
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.460664268768653e-05,
+      "loss": 1.3685,
+      "step": 91040
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.460549475778679e-05,
+      "loss": 1.2158,
+      "step": 91050
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.460434682788704e-05,
+      "loss": 1.5261,
+      "step": 91060
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.460319889798729e-05,
+      "loss": 1.1676,
+      "step": 91070
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.460205096808754e-05,
+      "loss": 1.3371,
+      "step": 91080
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.460090303818779e-05,
+      "loss": 1.3004,
+      "step": 91090
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.459975510828805e-05,
+      "loss": 1.1099,
+      "step": 91100
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.45986071783883e-05,
+      "loss": 1.2684,
+      "step": 91110
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.459745924848855e-05,
+      "loss": 1.2333,
+      "step": 91120
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.45963113185888e-05,
+      "loss": 1.3891,
+      "step": 91130
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.459516338868907e-05,
+      "loss": 1.3412,
+      "step": 91140
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.45940154587893e-05,
+      "loss": 1.3112,
+      "step": 91150
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.459286752888957e-05,
+      "loss": 1.2675,
+      "step": 91160
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.459171959898982e-05,
+      "loss": 1.2234,
+      "step": 91170
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.459057166909007e-05,
+      "loss": 1.4093,
+      "step": 91180
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.458942373919032e-05,
+      "loss": 1.1871,
+      "step": 91190
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.458827580929058e-05,
+      "loss": 1.5535,
+      "step": 91200
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.458712787939083e-05,
+      "loss": 1.2352,
+      "step": 91210
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.458597994949108e-05,
+      "loss": 1.2697,
+      "step": 91220
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.458483201959133e-05,
+      "loss": 1.4248,
+      "step": 91230
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.458368408969158e-05,
+      "loss": 1.2477,
+      "step": 91240
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.458253615979184e-05,
+      "loss": 1.1858,
+      "step": 91250
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.458138822989209e-05,
+      "loss": 1.2043,
+      "step": 91260
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.458024029999234e-05,
+      "loss": 1.3951,
+      "step": 91270
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.457909237009259e-05,
+      "loss": 1.2525,
+      "step": 91280
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.457794444019286e-05,
+      "loss": 1.3097,
+      "step": 91290
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.457679651029309e-05,
+      "loss": 1.1278,
+      "step": 91300
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.457564858039336e-05,
+      "loss": 1.1373,
+      "step": 91310
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.45745006504936e-05,
+      "loss": 1.3475,
+      "step": 91320
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.457335272059386e-05,
+      "loss": 1.3865,
+      "step": 91330
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.45722047906941e-05,
+      "loss": 1.2832,
+      "step": 91340
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.457105686079437e-05,
+      "loss": 1.2339,
+      "step": 91350
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.456990893089462e-05,
+      "loss": 1.4169,
+      "step": 91360
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.456876100099487e-05,
+      "loss": 1.2426,
+      "step": 91370
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.456761307109512e-05,
+      "loss": 1.2658,
+      "step": 91380
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.456646514119537e-05,
+      "loss": 1.2189,
+      "step": 91390
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.456531721129563e-05,
+      "loss": 1.4088,
+      "step": 91400
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.456416928139588e-05,
+      "loss": 1.2887,
+      "step": 91410
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.456302135149613e-05,
+      "loss": 1.3337,
+      "step": 91420
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.456187342159638e-05,
+      "loss": 1.3155,
+      "step": 91430
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.456072549169664e-05,
+      "loss": 1.2464,
+      "step": 91440
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.455957756179688e-05,
+      "loss": 1.23,
+      "step": 91450
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.455842963189714e-05,
+      "loss": 1.5386,
+      "step": 91460
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.45572817019974e-05,
+      "loss": 1.2617,
+      "step": 91470
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.455613377209764e-05,
+      "loss": 1.3387,
+      "step": 91480
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.45549858421979e-05,
+      "loss": 1.3309,
+      "step": 91490
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.455383791229816e-05,
+      "loss": 1.2982,
+      "step": 91500
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.45526899823984e-05,
+      "loss": 1.5044,
+      "step": 91510
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.455154205249866e-05,
+      "loss": 1.1833,
+      "step": 91520
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.45503941225989e-05,
+      "loss": 1.3059,
+      "step": 91530
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.454924619269916e-05,
+      "loss": 1.2236,
+      "step": 91540
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.454809826279942e-05,
+      "loss": 1.3325,
+      "step": 91550
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.454695033289967e-05,
+      "loss": 1.4269,
+      "step": 91560
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.454580240299992e-05,
+      "loss": 1.3153,
+      "step": 91570
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.454465447310017e-05,
+      "loss": 1.303,
+      "step": 91580
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.454350654320043e-05,
+      "loss": 1.3489,
+      "step": 91590
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.454235861330067e-05,
+      "loss": 1.245,
+      "step": 91600
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.454121068340093e-05,
+      "loss": 1.1641,
+      "step": 91610
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.454006275350118e-05,
+      "loss": 1.3319,
+      "step": 91620
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.453891482360143e-05,
+      "loss": 1.2814,
+      "step": 91630
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.453776689370168e-05,
+      "loss": 1.2787,
+      "step": 91640
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.453661896380195e-05,
+      "loss": 1.4119,
+      "step": 91650
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.45354710339022e-05,
+      "loss": 1.2326,
+      "step": 91660
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.453432310400245e-05,
+      "loss": 1.202,
+      "step": 91670
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.45331751741027e-05,
+      "loss": 1.2779,
+      "step": 91680
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.453202724420294e-05,
+      "loss": 1.2821,
+      "step": 91690
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.453087931430321e-05,
+      "loss": 1.276,
+      "step": 91700
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.452973138440346e-05,
+      "loss": 1.4173,
+      "step": 91710
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.452858345450371e-05,
+      "loss": 1.3093,
+      "step": 91720
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.452743552460396e-05,
+      "loss": 1.2925,
+      "step": 91730
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.452628759470422e-05,
+      "loss": 1.3085,
+      "step": 91740
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.452513966480446e-05,
+      "loss": 1.3226,
+      "step": 91750
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.452399173490472e-05,
+      "loss": 1.1783,
+      "step": 91760
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.452284380500497e-05,
+      "loss": 1.1885,
+      "step": 91770
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.452169587510522e-05,
+      "loss": 1.2972,
+      "step": 91780
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.452054794520547e-05,
+      "loss": 1.3513,
+      "step": 91790
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.451940001530572e-05,
+      "loss": 1.4019,
+      "step": 91800
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.451825208540598e-05,
+      "loss": 1.262,
+      "step": 91810
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.451710415550623e-05,
+      "loss": 1.4545,
+      "step": 91820
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.451595622560648e-05,
+      "loss": 1.3546,
+      "step": 91830
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.451480829570673e-05,
+      "loss": 1.2628,
+      "step": 91840
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.4513660365807e-05,
+      "loss": 1.2704,
+      "step": 91850
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.451251243590723e-05,
+      "loss": 1.2985,
+      "step": 91860
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.45113645060075e-05,
+      "loss": 1.3532,
+      "step": 91870
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.451021657610775e-05,
+      "loss": 1.216,
+      "step": 91880
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.4509068646208e-05,
+      "loss": 1.1806,
+      "step": 91890
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.450792071630825e-05,
+      "loss": 1.425,
+      "step": 91900
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.450677278640851e-05,
+      "loss": 1.3296,
+      "step": 91910
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.450562485650876e-05,
+      "loss": 1.3531,
+      "step": 91920
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.450447692660901e-05,
+      "loss": 1.2284,
+      "step": 91930
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.450332899670926e-05,
+      "loss": 1.4402,
+      "step": 91940
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.450218106680951e-05,
+      "loss": 1.2502,
+      "step": 91950
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.450103313690977e-05,
+      "loss": 1.3597,
+      "step": 91960
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.449988520701002e-05,
+      "loss": 1.3701,
+      "step": 91970
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.449873727711027e-05,
+      "loss": 1.3179,
+      "step": 91980
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.449758934721052e-05,
+      "loss": 1.4024,
+      "step": 91990
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.449644141731079e-05,
+      "loss": 1.3459,
+      "step": 92000
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.449529348741102e-05,
+      "loss": 1.1437,
+      "step": 92010
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.449414555751128e-05,
+      "loss": 1.3267,
+      "step": 92020
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.449299762761153e-05,
+      "loss": 1.2729,
+      "step": 92030
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.449184969771178e-05,
+      "loss": 1.2327,
+      "step": 92040
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.449070176781203e-05,
+      "loss": 1.2653,
+      "step": 92050
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.44895538379123e-05,
+      "loss": 1.3755,
+      "step": 92060
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.448840590801255e-05,
+      "loss": 1.2838,
+      "step": 92070
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.44872579781128e-05,
+      "loss": 1.4102,
+      "step": 92080
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.448611004821305e-05,
+      "loss": 1.2135,
+      "step": 92090
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.44849621183133e-05,
+      "loss": 1.4175,
+      "step": 92100
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.448381418841356e-05,
+      "loss": 1.4349,
+      "step": 92110
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.448266625851381e-05,
+      "loss": 1.2228,
+      "step": 92120
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.448151832861406e-05,
+      "loss": 1.3658,
+      "step": 92130
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.448037039871431e-05,
+      "loss": 1.2861,
+      "step": 92140
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.447922246881457e-05,
+      "loss": 1.3601,
+      "step": 92150
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.447807453891481e-05,
+      "loss": 1.3005,
+      "step": 92160
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.447692660901507e-05,
+      "loss": 1.4324,
+      "step": 92170
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.447577867911532e-05,
+      "loss": 1.3782,
+      "step": 92180
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.447463074921557e-05,
+      "loss": 1.3412,
+      "step": 92190
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.447348281931582e-05,
+      "loss": 1.3179,
+      "step": 92200
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.447233488941609e-05,
+      "loss": 1.2086,
+      "step": 92210
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.447118695951634e-05,
+      "loss": 1.3173,
+      "step": 92220
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.447003902961659e-05,
+      "loss": 1.1831,
+      "step": 92230
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.446889109971684e-05,
+      "loss": 1.3839,
+      "step": 92240
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.446774316981709e-05,
+      "loss": 1.3526,
+      "step": 92250
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.446659523991735e-05,
+      "loss": 1.4208,
+      "step": 92260
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.44654473100176e-05,
+      "loss": 1.3016,
+      "step": 92270
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.446429938011785e-05,
+      "loss": 1.3209,
+      "step": 92280
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.44631514502181e-05,
+      "loss": 1.0939,
+      "step": 92290
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.446200352031836e-05,
+      "loss": 1.3627,
+      "step": 92300
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.44608555904186e-05,
+      "loss": 1.514,
+      "step": 92310
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.445970766051886e-05,
+      "loss": 1.2279,
+      "step": 92320
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.445855973061911e-05,
+      "loss": 1.2581,
+      "step": 92330
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.445741180071936e-05,
+      "loss": 1.427,
+      "step": 92340
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.445626387081961e-05,
+      "loss": 1.2975,
+      "step": 92350
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.445511594091987e-05,
+      "loss": 1.3111,
+      "step": 92360
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.445396801102012e-05,
+      "loss": 1.2037,
+      "step": 92370
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.445282008112037e-05,
+      "loss": 1.3078,
+      "step": 92380
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.445167215122062e-05,
+      "loss": 1.3128,
+      "step": 92390
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.445052422132087e-05,
+      "loss": 1.3642,
+      "step": 92400
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.444937629142114e-05,
+      "loss": 1.2941,
+      "step": 92410
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.444822836152139e-05,
+      "loss": 1.3481,
+      "step": 92420
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.444708043162164e-05,
+      "loss": 1.2098,
+      "step": 92430
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.444593250172189e-05,
+      "loss": 1.5344,
+      "step": 92440
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.444478457182215e-05,
+      "loss": 1.366,
+      "step": 92450
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.444363664192239e-05,
+      "loss": 1.2995,
+      "step": 92460
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.444248871202265e-05,
+      "loss": 1.213,
+      "step": 92470
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.44413407821229e-05,
+      "loss": 1.4192,
+      "step": 92480
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.444019285222315e-05,
+      "loss": 1.3273,
+      "step": 92490
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.44390449223234e-05,
+      "loss": 1.4254,
+      "step": 92500
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.443789699242366e-05,
+      "loss": 1.3627,
+      "step": 92510
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 6.443674906252391e-05,
+      "loss": 1.3301,
+      "step": 92520
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.443560113262416e-05,
+      "loss": 1.3344,
+      "step": 92530
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.443445320272441e-05,
+      "loss": 1.3413,
+      "step": 92540
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.443330527282466e-05,
+      "loss": 1.2193,
+      "step": 92550
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.443215734292493e-05,
+      "loss": 1.4186,
+      "step": 92560
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.443100941302518e-05,
+      "loss": 1.173,
+      "step": 92570
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.442986148312543e-05,
+      "loss": 1.3668,
+      "step": 92580
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.442871355322568e-05,
+      "loss": 1.4136,
+      "step": 92590
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.442756562332594e-05,
+      "loss": 1.2126,
+      "step": 92600
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.442641769342618e-05,
+      "loss": 1.197,
+      "step": 92610
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.442526976352644e-05,
+      "loss": 1.2827,
+      "step": 92620
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.442412183362669e-05,
+      "loss": 1.3091,
+      "step": 92630
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.442297390372694e-05,
+      "loss": 1.2088,
+      "step": 92640
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.442182597382719e-05,
+      "loss": 1.4171,
+      "step": 92650
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.442067804392745e-05,
+      "loss": 1.2806,
+      "step": 92660
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.44195301140277e-05,
+      "loss": 1.3285,
+      "step": 92670
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.441838218412795e-05,
+      "loss": 1.2859,
+      "step": 92680
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.44172342542282e-05,
+      "loss": 1.2467,
+      "step": 92690
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.441608632432845e-05,
+      "loss": 1.3386,
+      "step": 92700
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.441493839442871e-05,
+      "loss": 1.2461,
+      "step": 92710
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.441379046452896e-05,
+      "loss": 1.2944,
+      "step": 92720
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 6.441264253462921e-05,
+      "loss": 1.4151,
+      "step": 92730
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.441149460472946e-05,
+      "loss": 1.2686,
+      "step": 92740
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.441034667482973e-05,
+      "loss": 1.4446,
+      "step": 92750
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.440919874492996e-05,
+      "loss": 1.3595,
+      "step": 92760
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.440805081503023e-05,
+      "loss": 1.3309,
+      "step": 92770
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.440690288513048e-05,
+      "loss": 1.3006,
+      "step": 92780
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.440575495523073e-05,
+      "loss": 1.2332,
+      "step": 92790
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.440460702533098e-05,
+      "loss": 1.3617,
+      "step": 92800
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.440345909543124e-05,
+      "loss": 1.2258,
+      "step": 92810
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.440231116553149e-05,
+      "loss": 1.3319,
+      "step": 92820
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.440116323563174e-05,
+      "loss": 1.3974,
+      "step": 92830
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.440001530573199e-05,
+      "loss": 1.368,
+      "step": 92840
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.439886737583224e-05,
+      "loss": 1.2657,
+      "step": 92850
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.43977194459325e-05,
+      "loss": 1.4063,
+      "step": 92860
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.439657151603275e-05,
+      "loss": 1.2795,
+      "step": 92870
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.4395423586133e-05,
+      "loss": 1.1918,
+      "step": 92880
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.439427565623325e-05,
+      "loss": 1.299,
+      "step": 92890
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.439312772633352e-05,
+      "loss": 1.4114,
+      "step": 92900
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.439197979643375e-05,
+      "loss": 1.5188,
+      "step": 92910
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.439083186653402e-05,
+      "loss": 1.3911,
+      "step": 92920
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.438968393663427e-05,
+      "loss": 1.2922,
+      "step": 92930
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.438853600673452e-05,
+      "loss": 1.298,
+      "step": 92940
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 6.438738807683477e-05,
+      "loss": 1.3483,
+      "step": 92950
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.438624014693503e-05,
+      "loss": 1.1589,
+      "step": 92960
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.438509221703528e-05,
+      "loss": 1.2167,
+      "step": 92970
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.438394428713553e-05,
+      "loss": 1.3678,
+      "step": 92980
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.438279635723578e-05,
+      "loss": 1.2798,
+      "step": 92990
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.438164842733603e-05,
+      "loss": 1.3277,
+      "step": 93000
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.438050049743629e-05,
+      "loss": 1.3353,
+      "step": 93010
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.437935256753654e-05,
+      "loss": 1.3375,
+      "step": 93020
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.437820463763679e-05,
+      "loss": 1.4302,
+      "step": 93030
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.437705670773704e-05,
+      "loss": 1.364,
+      "step": 93040
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.43759087778373e-05,
+      "loss": 1.2509,
+      "step": 93050
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.437476084793754e-05,
+      "loss": 1.4915,
+      "step": 93060
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.43736129180378e-05,
+      "loss": 1.2343,
+      "step": 93070
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.437246498813805e-05,
+      "loss": 1.3975,
+      "step": 93080
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.43713170582383e-05,
+      "loss": 1.2974,
+      "step": 93090
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.437016912833855e-05,
+      "loss": 1.2506,
+      "step": 93100
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.436902119843882e-05,
+      "loss": 1.4025,
+      "step": 93110
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.436787326853907e-05,
+      "loss": 1.3611,
+      "step": 93120
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.436672533863932e-05,
+      "loss": 1.3342,
+      "step": 93130
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.436557740873957e-05,
+      "loss": 1.1302,
+      "step": 93140
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.436442947883982e-05,
+      "loss": 1.2362,
+      "step": 93150
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.436328154894008e-05,
+      "loss": 1.3547,
+      "step": 93160
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 6.436213361904032e-05,
+      "loss": 1.2487,
+      "step": 93170
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.436098568914058e-05,
+      "loss": 1.308,
+      "step": 93180
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.435983775924083e-05,
+      "loss": 1.3812,
+      "step": 93190
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.435868982934108e-05,
+      "loss": 1.2775,
+      "step": 93200
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.435754189944133e-05,
+      "loss": 1.3662,
+      "step": 93210
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.435639396954159e-05,
+      "loss": 1.252,
+      "step": 93220
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.435524603964184e-05,
+      "loss": 1.2325,
+      "step": 93230
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.435409810974209e-05,
+      "loss": 1.4244,
+      "step": 93240
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.435295017984234e-05,
+      "loss": 1.309,
+      "step": 93250
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.435180224994259e-05,
+      "loss": 1.2969,
+      "step": 93260
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.435065432004286e-05,
+      "loss": 1.281,
+      "step": 93270
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.43495063901431e-05,
+      "loss": 1.2279,
+      "step": 93280
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.434835846024336e-05,
+      "loss": 1.2099,
+      "step": 93290
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.43472105303436e-05,
+      "loss": 1.422,
+      "step": 93300
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.434606260044387e-05,
+      "loss": 1.2364,
+      "step": 93310
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.43449146705441e-05,
+      "loss": 1.3553,
+      "step": 93320
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.434376674064437e-05,
+      "loss": 1.5051,
+      "step": 93330
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.434261881074462e-05,
+      "loss": 1.3169,
+      "step": 93340
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.434147088084487e-05,
+      "loss": 1.1666,
+      "step": 93350
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.434032295094512e-05,
+      "loss": 1.3784,
+      "step": 93360
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.433917502104538e-05,
+      "loss": 1.3288,
+      "step": 93370
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.433802709114563e-05,
+      "loss": 1.3976,
+      "step": 93380
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 6.433687916124588e-05,
+      "loss": 1.1261,
+      "step": 93390
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.433573123134613e-05,
+      "loss": 1.239,
+      "step": 93400
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.433458330144638e-05,
+      "loss": 1.2716,
+      "step": 93410
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.433343537154664e-05,
+      "loss": 1.2532,
+      "step": 93420
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.43322874416469e-05,
+      "loss": 1.2413,
+      "step": 93430
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.433113951174714e-05,
+      "loss": 1.3008,
+      "step": 93440
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.43299915818474e-05,
+      "loss": 1.2759,
+      "step": 93450
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.432884365194766e-05,
+      "loss": 1.2413,
+      "step": 93460
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.432769572204789e-05,
+      "loss": 1.4076,
+      "step": 93470
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.432654779214816e-05,
+      "loss": 1.3874,
+      "step": 93480
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.43253998622484e-05,
+      "loss": 1.4963,
+      "step": 93490
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.432425193234866e-05,
+      "loss": 1.4014,
+      "step": 93500
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.43231040024489e-05,
+      "loss": 1.2639,
+      "step": 93510
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.432195607254917e-05,
+      "loss": 1.382,
+      "step": 93520
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.432080814264942e-05,
+      "loss": 1.214,
+      "step": 93530
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.431966021274967e-05,
+      "loss": 1.3119,
+      "step": 93540
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.431851228284992e-05,
+      "loss": 1.2189,
+      "step": 93550
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.431736435295017e-05,
+      "loss": 1.4173,
+      "step": 93560
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.431621642305043e-05,
+      "loss": 1.3115,
+      "step": 93570
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.431506849315068e-05,
+      "loss": 1.3345,
+      "step": 93580
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.431392056325093e-05,
+      "loss": 1.3282,
+      "step": 93590
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.431277263335118e-05,
+      "loss": 1.292,
+      "step": 93600
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 6.431162470345145e-05,
+      "loss": 1.2506,
+      "step": 93610
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.431047677355168e-05,
+      "loss": 1.2506,
+      "step": 93620
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.430932884365195e-05,
+      "loss": 1.3574,
+      "step": 93630
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.43081809137522e-05,
+      "loss": 1.308,
+      "step": 93640
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.430703298385244e-05,
+      "loss": 1.398,
+      "step": 93650
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.43058850539527e-05,
+      "loss": 1.3249,
+      "step": 93660
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.430473712405296e-05,
+      "loss": 1.5764,
+      "step": 93670
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.430358919415321e-05,
+      "loss": 1.3597,
+      "step": 93680
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.430244126425346e-05,
+      "loss": 1.3786,
+      "step": 93690
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.430129333435371e-05,
+      "loss": 1.2908,
+      "step": 93700
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.430014540445396e-05,
+      "loss": 1.2381,
+      "step": 93710
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.429899747455422e-05,
+      "loss": 1.2864,
+      "step": 93720
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.429784954465447e-05,
+      "loss": 1.3634,
+      "step": 93730
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.429670161475472e-05,
+      "loss": 1.2643,
+      "step": 93740
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.429555368485497e-05,
+      "loss": 1.4407,
+      "step": 93750
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.429440575495523e-05,
+      "loss": 1.2964,
+      "step": 93760
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.429325782505547e-05,
+      "loss": 1.3315,
+      "step": 93770
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.429210989515573e-05,
+      "loss": 1.1468,
+      "step": 93780
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.429096196525598e-05,
+      "loss": 1.2677,
+      "step": 93790
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.428981403535623e-05,
+      "loss": 1.3048,
+      "step": 93800
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.428866610545648e-05,
+      "loss": 1.3311,
+      "step": 93810
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 6.428751817555675e-05,
+      "loss": 1.2455,
+      "step": 93820
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.4286370245657e-05,
+      "loss": 1.307,
+      "step": 93830
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.428522231575725e-05,
+      "loss": 1.4773,
+      "step": 93840
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.42840743858575e-05,
+      "loss": 1.3088,
+      "step": 93850
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.428292645595775e-05,
+      "loss": 1.2626,
+      "step": 93860
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.428177852605801e-05,
+      "loss": 1.3686,
+      "step": 93870
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.428063059615826e-05,
+      "loss": 1.1626,
+      "step": 93880
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.427948266625851e-05,
+      "loss": 1.37,
+      "step": 93890
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.427833473635876e-05,
+      "loss": 1.5476,
+      "step": 93900
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.427718680645902e-05,
+      "loss": 1.1405,
+      "step": 93910
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.427603887655926e-05,
+      "loss": 1.2685,
+      "step": 93920
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.427489094665952e-05,
+      "loss": 1.4179,
+      "step": 93930
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.427374301675977e-05,
+      "loss": 1.2611,
+      "step": 93940
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.427259508686002e-05,
+      "loss": 1.476,
+      "step": 93950
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.427144715696027e-05,
+      "loss": 1.2627,
+      "step": 93960
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.427029922706054e-05,
+      "loss": 1.135,
+      "step": 93970
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.426915129716078e-05,
+      "loss": 1.4502,
+      "step": 93980
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.426800336726103e-05,
+      "loss": 1.2938,
+      "step": 93990
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.426685543736128e-05,
+      "loss": 1.3441,
+      "step": 94000
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.426570750746153e-05,
+      "loss": 1.2239,
+      "step": 94010
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.42645595775618e-05,
+      "loss": 1.3233,
+      "step": 94020
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.426341164766205e-05,
+      "loss": 1.3539,
+      "step": 94030
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 6.42622637177623e-05,
+      "loss": 1.2385,
+      "step": 94040
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.426111578786255e-05,
+      "loss": 1.3045,
+      "step": 94050
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.425996785796281e-05,
+      "loss": 1.2705,
+      "step": 94060
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.425881992806305e-05,
+      "loss": 1.335,
+      "step": 94070
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.425767199816331e-05,
+      "loss": 1.408,
+      "step": 94080
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.425652406826356e-05,
+      "loss": 1.1979,
+      "step": 94090
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.425537613836381e-05,
+      "loss": 1.3013,
+      "step": 94100
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.425422820846406e-05,
+      "loss": 1.4879,
+      "step": 94110
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.425308027856432e-05,
+      "loss": 1.2983,
+      "step": 94120
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.425193234866457e-05,
+      "loss": 1.4629,
+      "step": 94130
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.425078441876482e-05,
+      "loss": 1.4255,
+      "step": 94140
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.424963648886507e-05,
+      "loss": 1.3887,
+      "step": 94150
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.424848855896532e-05,
+      "loss": 1.3771,
+      "step": 94160
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.424734062906559e-05,
+      "loss": 1.2391,
+      "step": 94170
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.424619269916584e-05,
+      "loss": 1.3652,
+      "step": 94180
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.424504476926609e-05,
+      "loss": 1.2743,
+      "step": 94190
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.424389683936634e-05,
+      "loss": 1.1962,
+      "step": 94200
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.42427489094666e-05,
+      "loss": 1.2707,
+      "step": 94210
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.424160097956684e-05,
+      "loss": 1.167,
+      "step": 94220
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.42404530496671e-05,
+      "loss": 1.2806,
+      "step": 94230
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.423930511976735e-05,
+      "loss": 1.3931,
+      "step": 94240
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.42381571898676e-05,
+      "loss": 1.1611,
+      "step": 94250
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 6.423700925996785e-05,
+      "loss": 1.3273,
+      "step": 94260
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.423586133006811e-05,
+      "loss": 1.212,
+      "step": 94270
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.423471340016836e-05,
+      "loss": 1.4021,
+      "step": 94280
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.423356547026861e-05,
+      "loss": 1.374,
+      "step": 94290
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.423241754036886e-05,
+      "loss": 1.3136,
+      "step": 94300
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.423126961046911e-05,
+      "loss": 1.4121,
+      "step": 94310
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.423012168056937e-05,
+      "loss": 1.2848,
+      "step": 94320
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.422897375066962e-05,
+      "loss": 1.4457,
+      "step": 94330
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.422782582076987e-05,
+      "loss": 1.2752,
+      "step": 94340
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.422667789087012e-05,
+      "loss": 1.2634,
+      "step": 94350
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.422552996097039e-05,
+      "loss": 1.2881,
+      "step": 94360
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.422438203107062e-05,
+      "loss": 1.2819,
+      "step": 94370
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.422323410117089e-05,
+      "loss": 1.2992,
+      "step": 94380
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.422208617127114e-05,
+      "loss": 1.2927,
+      "step": 94390
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.422093824137139e-05,
+      "loss": 1.3093,
+      "step": 94400
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.421979031147164e-05,
+      "loss": 1.3586,
+      "step": 94410
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.42186423815719e-05,
+      "loss": 1.2663,
+      "step": 94420
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.421749445167215e-05,
+      "loss": 1.2789,
+      "step": 94430
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.42163465217724e-05,
+      "loss": 1.2335,
+      "step": 94440
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.421519859187265e-05,
+      "loss": 1.3508,
+      "step": 94450
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.42140506619729e-05,
+      "loss": 1.2549,
+      "step": 94460
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.421290273207316e-05,
+      "loss": 1.4042,
+      "step": 94470
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 6.421175480217341e-05,
+      "loss": 1.3193,
+      "step": 94480
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.421060687227366e-05,
+      "loss": 1.3983,
+      "step": 94490
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.420945894237391e-05,
+      "loss": 1.4183,
+      "step": 94500
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.420831101247418e-05,
+      "loss": 1.3204,
+      "step": 94510
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.420716308257441e-05,
+      "loss": 1.4436,
+      "step": 94520
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.420601515267468e-05,
+      "loss": 1.3259,
+      "step": 94530
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.420486722277493e-05,
+      "loss": 1.2691,
+      "step": 94540
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.420371929287518e-05,
+      "loss": 1.3011,
+      "step": 94550
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.420257136297543e-05,
+      "loss": 1.4689,
+      "step": 94560
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.420142343307568e-05,
+      "loss": 1.298,
+      "step": 94570
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.420027550317594e-05,
+      "loss": 1.3776,
+      "step": 94580
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.419912757327619e-05,
+      "loss": 1.3013,
+      "step": 94590
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.419797964337644e-05,
+      "loss": 1.4043,
+      "step": 94600
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.419683171347669e-05,
+      "loss": 1.3356,
+      "step": 94610
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.419568378357695e-05,
+      "loss": 1.4255,
+      "step": 94620
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.419453585367719e-05,
+      "loss": 1.2485,
+      "step": 94630
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.419338792377745e-05,
+      "loss": 1.4593,
+      "step": 94640
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.41922399938777e-05,
+      "loss": 1.4704,
+      "step": 94650
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.419109206397795e-05,
+      "loss": 1.3063,
+      "step": 94660
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.41899441340782e-05,
+      "loss": 1.4193,
+      "step": 94670
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.418879620417846e-05,
+      "loss": 1.2848,
+      "step": 94680
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 6.418764827427871e-05,
+      "loss": 1.2803,
+      "step": 94690
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.418650034437896e-05,
+      "loss": 1.2526,
+      "step": 94700
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.418535241447921e-05,
+      "loss": 1.3107,
+      "step": 94710
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.418420448457946e-05,
+      "loss": 1.4144,
+      "step": 94720
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.418305655467973e-05,
+      "loss": 1.282,
+      "step": 94730
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.418190862477998e-05,
+      "loss": 1.39,
+      "step": 94740
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.418076069488023e-05,
+      "loss": 1.4625,
+      "step": 94750
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.417961276498048e-05,
+      "loss": 1.258,
+      "step": 94760
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.417846483508074e-05,
+      "loss": 1.2041,
+      "step": 94770
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.417731690518098e-05,
+      "loss": 1.3552,
+      "step": 94780
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.417616897528124e-05,
+      "loss": 1.2969,
+      "step": 94790
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.417502104538149e-05,
+      "loss": 1.2537,
+      "step": 94800
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.417387311548174e-05,
+      "loss": 1.3055,
+      "step": 94810
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.417272518558199e-05,
+      "loss": 1.3271,
+      "step": 94820
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.417157725568225e-05,
+      "loss": 1.3587,
+      "step": 94830
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.41704293257825e-05,
+      "loss": 1.2509,
+      "step": 94840
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.416928139588275e-05,
+      "loss": 1.2415,
+      "step": 94850
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.4168133465983e-05,
+      "loss": 1.2958,
+      "step": 94860
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.416698553608325e-05,
+      "loss": 1.4123,
+      "step": 94870
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.416583760618352e-05,
+      "loss": 1.3233,
+      "step": 94880
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.416468967628377e-05,
+      "loss": 1.4158,
+      "step": 94890
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.416354174638402e-05,
+      "loss": 1.2212,
+      "step": 94900
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 6.416239381648427e-05,
+      "loss": 1.36,
+      "step": 94910
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.416124588658453e-05,
+      "loss": 1.4612,
+      "step": 94920
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.416009795668476e-05,
+      "loss": 1.3287,
+      "step": 94930
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.415895002678503e-05,
+      "loss": 1.2645,
+      "step": 94940
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.415780209688528e-05,
+      "loss": 1.268,
+      "step": 94950
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.415665416698553e-05,
+      "loss": 1.3606,
+      "step": 94960
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.415550623708578e-05,
+      "loss": 1.1879,
+      "step": 94970
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.415435830718604e-05,
+      "loss": 1.4989,
+      "step": 94980
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.415321037728629e-05,
+      "loss": 1.2689,
+      "step": 94990
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.415206244738654e-05,
+      "loss": 1.234,
+      "step": 95000
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.415091451748679e-05,
+      "loss": 1.4111,
+      "step": 95010
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.414976658758704e-05,
+      "loss": 1.2794,
+      "step": 95020
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.41486186576873e-05,
+      "loss": 1.3338,
+      "step": 95030
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.414747072778755e-05,
+      "loss": 1.2067,
+      "step": 95040
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.41463227978878e-05,
+      "loss": 1.291,
+      "step": 95050
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.414517486798805e-05,
+      "loss": 1.367,
+      "step": 95060
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.414402693808832e-05,
+      "loss": 1.3038,
+      "step": 95070
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.414287900818855e-05,
+      "loss": 1.4444,
+      "step": 95080
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.414173107828882e-05,
+      "loss": 1.4273,
+      "step": 95090
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.414058314838907e-05,
+      "loss": 1.2804,
+      "step": 95100
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.413943521848932e-05,
+      "loss": 1.3446,
+      "step": 95110
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.413828728858957e-05,
+      "loss": 1.22,
+      "step": 95120
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 6.413713935868983e-05,
+      "loss": 1.2149,
+      "step": 95130
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.413599142879008e-05,
+      "loss": 1.3785,
+      "step": 95140
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.413484349889033e-05,
+      "loss": 1.3267,
+      "step": 95150
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.413369556899058e-05,
+      "loss": 1.4088,
+      "step": 95160
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.413254763909083e-05,
+      "loss": 1.2044,
+      "step": 95170
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.413139970919109e-05,
+      "loss": 1.3218,
+      "step": 95180
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.413025177929134e-05,
+      "loss": 1.2743,
+      "step": 95190
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.412910384939159e-05,
+      "loss": 1.3605,
+      "step": 95200
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.412795591949184e-05,
+      "loss": 1.3385,
+      "step": 95210
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.41268079895921e-05,
+      "loss": 1.2952,
+      "step": 95220
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.412566005969234e-05,
+      "loss": 1.2431,
+      "step": 95230
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.41245121297926e-05,
+      "loss": 1.3244,
+      "step": 95240
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.412336419989286e-05,
+      "loss": 1.4388,
+      "step": 95250
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.41222162699931e-05,
+      "loss": 1.2701,
+      "step": 95260
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.412106834009335e-05,
+      "loss": 1.2746,
+      "step": 95270
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.411992041019362e-05,
+      "loss": 1.4199,
+      "step": 95280
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.411877248029387e-05,
+      "loss": 1.3936,
+      "step": 95290
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.411762455039412e-05,
+      "loss": 1.3711,
+      "step": 95300
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.411647662049437e-05,
+      "loss": 1.4034,
+      "step": 95310
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.411532869059462e-05,
+      "loss": 1.3033,
+      "step": 95320
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.411418076069488e-05,
+      "loss": 1.2575,
+      "step": 95330
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.411303283079513e-05,
+      "loss": 1.2271,
+      "step": 95340
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 6.411188490089538e-05,
+      "loss": 1.2542,
+      "step": 95350
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.411073697099563e-05,
+      "loss": 1.3584,
+      "step": 95360
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.41095890410959e-05,
+      "loss": 1.392,
+      "step": 95370
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.410844111119613e-05,
+      "loss": 1.3627,
+      "step": 95380
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.41072931812964e-05,
+      "loss": 1.1577,
+      "step": 95390
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.410614525139664e-05,
+      "loss": 1.4485,
+      "step": 95400
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.41049973214969e-05,
+      "loss": 1.2454,
+      "step": 95410
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.410384939159714e-05,
+      "loss": 1.2761,
+      "step": 95420
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.41027014616974e-05,
+      "loss": 1.274,
+      "step": 95430
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.410155353179766e-05,
+      "loss": 1.2228,
+      "step": 95440
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.41004056018979e-05,
+      "loss": 1.1797,
+      "step": 95450
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.409925767199816e-05,
+      "loss": 1.2089,
+      "step": 95460
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.40981097420984e-05,
+      "loss": 1.2698,
+      "step": 95470
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.409696181219867e-05,
+      "loss": 1.3403,
+      "step": 95480
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.409581388229892e-05,
+      "loss": 1.3715,
+      "step": 95490
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.409466595239917e-05,
+      "loss": 1.3386,
+      "step": 95500
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.409351802249942e-05,
+      "loss": 1.3765,
+      "step": 95510
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.409237009259968e-05,
+      "loss": 1.4188,
+      "step": 95520
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.409122216269992e-05,
+      "loss": 1.3024,
+      "step": 95530
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.409007423280018e-05,
+      "loss": 1.2932,
+      "step": 95540
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.408892630290043e-05,
+      "loss": 1.3932,
+      "step": 95550
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.408777837300068e-05,
+      "loss": 1.386,
+      "step": 95560
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 6.408663044310093e-05,
+      "loss": 1.1919,
+      "step": 95570
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.40854825132012e-05,
+      "loss": 1.3823,
+      "step": 95580
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.408433458330145e-05,
+      "loss": 1.2108,
+      "step": 95590
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.40831866534017e-05,
+      "loss": 1.5211,
+      "step": 95600
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.408203872350194e-05,
+      "loss": 1.2588,
+      "step": 95610
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.40808907936022e-05,
+      "loss": 1.3344,
+      "step": 95620
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.407974286370246e-05,
+      "loss": 1.2181,
+      "step": 95630
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.407859493380271e-05,
+      "loss": 1.448,
+      "step": 95640
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.407744700390296e-05,
+      "loss": 1.3837,
+      "step": 95650
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.407629907400321e-05,
+      "loss": 1.4078,
+      "step": 95660
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.407515114410347e-05,
+      "loss": 1.2318,
+      "step": 95670
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.407400321420371e-05,
+      "loss": 1.3043,
+      "step": 95680
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.407285528430397e-05,
+      "loss": 1.3754,
+      "step": 95690
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.407170735440422e-05,
+      "loss": 1.451,
+      "step": 95700
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.407055942450447e-05,
+      "loss": 1.2645,
+      "step": 95710
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.406941149460472e-05,
+      "loss": 1.3922,
+      "step": 95720
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.406826356470498e-05,
+      "loss": 1.3492,
+      "step": 95730
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.406711563480523e-05,
+      "loss": 1.2883,
+      "step": 95740
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.406596770490548e-05,
+      "loss": 1.3746,
+      "step": 95750
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.406481977500573e-05,
+      "loss": 1.3392,
+      "step": 95760
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.406367184510598e-05,
+      "loss": 1.3531,
+      "step": 95770
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 6.406252391520625e-05,
+      "loss": 1.2622,
+      "step": 95780
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.40613759853065e-05,
+      "loss": 1.441,
+      "step": 95790
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.406022805540675e-05,
+      "loss": 1.3145,
+      "step": 95800
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.4059080125507e-05,
+      "loss": 1.2272,
+      "step": 95810
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.405793219560726e-05,
+      "loss": 1.1941,
+      "step": 95820
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.40567842657075e-05,
+      "loss": 1.3836,
+      "step": 95830
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.405563633580776e-05,
+      "loss": 1.3651,
+      "step": 95840
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.405448840590801e-05,
+      "loss": 1.2622,
+      "step": 95850
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.405334047600826e-05,
+      "loss": 1.2533,
+      "step": 95860
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.405219254610851e-05,
+      "loss": 1.3761,
+      "step": 95870
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.405104461620877e-05,
+      "loss": 1.3161,
+      "step": 95880
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.404989668630902e-05,
+      "loss": 1.3488,
+      "step": 95890
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.404874875640927e-05,
+      "loss": 1.3449,
+      "step": 95900
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.404760082650952e-05,
+      "loss": 1.2871,
+      "step": 95910
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.404645289660977e-05,
+      "loss": 1.3076,
+      "step": 95920
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.404530496671003e-05,
+      "loss": 1.2871,
+      "step": 95930
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.404415703681027e-05,
+      "loss": 1.4054,
+      "step": 95940
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.404300910691053e-05,
+      "loss": 1.3412,
+      "step": 95950
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.404186117701078e-05,
+      "loss": 1.3083,
+      "step": 95960
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.404071324711103e-05,
+      "loss": 1.348,
+      "step": 95970
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.403956531721128e-05,
+      "loss": 1.3247,
+      "step": 95980
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.403841738731155e-05,
+      "loss": 1.4006,
+      "step": 95990
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 6.40372694574118e-05,
+      "loss": 1.3268,
+      "step": 96000
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.403612152751205e-05,
+      "loss": 1.2695,
+      "step": 96010
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.40349735976123e-05,
+      "loss": 1.247,
+      "step": 96020
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.403382566771255e-05,
+      "loss": 1.1527,
+      "step": 96030
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.403267773781281e-05,
+      "loss": 1.3456,
+      "step": 96040
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.403152980791306e-05,
+      "loss": 1.4191,
+      "step": 96050
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.403038187801331e-05,
+      "loss": 1.3665,
+      "step": 96060
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.402923394811356e-05,
+      "loss": 1.3719,
+      "step": 96070
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.402808601821382e-05,
+      "loss": 1.2502,
+      "step": 96080
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.402693808831406e-05,
+      "loss": 1.298,
+      "step": 96090
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.402579015841432e-05,
+      "loss": 1.2845,
+      "step": 96100
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.402464222851457e-05,
+      "loss": 1.5786,
+      "step": 96110
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.402349429861482e-05,
+      "loss": 1.4199,
+      "step": 96120
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.402234636871507e-05,
+      "loss": 1.1592,
+      "step": 96130
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.402119843881534e-05,
+      "loss": 1.2681,
+      "step": 96140
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.402005050891559e-05,
+      "loss": 1.2904,
+      "step": 96150
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.401890257901584e-05,
+      "loss": 1.2228,
+      "step": 96160
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.401775464911609e-05,
+      "loss": 1.2652,
+      "step": 96170
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.401660671921634e-05,
+      "loss": 1.2694,
+      "step": 96180
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.40154587893166e-05,
+      "loss": 1.2569,
+      "step": 96190
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.401431085941685e-05,
+      "loss": 1.1423,
+      "step": 96200
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.40131629295171e-05,
+      "loss": 1.3092,
+      "step": 96210
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 6.401201499961735e-05,
+      "loss": 1.3717,
+      "step": 96220
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.401086706971761e-05,
+      "loss": 1.4051,
+      "step": 96230
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.400971913981785e-05,
+      "loss": 1.3169,
+      "step": 96240
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.400857120991811e-05,
+      "loss": 1.3081,
+      "step": 96250
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.400742328001836e-05,
+      "loss": 1.4371,
+      "step": 96260
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.400627535011861e-05,
+      "loss": 1.4571,
+      "step": 96270
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.400512742021886e-05,
+      "loss": 1.3621,
+      "step": 96280
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.400397949031912e-05,
+      "loss": 1.2999,
+      "step": 96290
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.400283156041937e-05,
+      "loss": 1.2345,
+      "step": 96300
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.400168363051962e-05,
+      "loss": 1.3023,
+      "step": 96310
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.400053570061987e-05,
+      "loss": 1.3632,
+      "step": 96320
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.399938777072012e-05,
+      "loss": 1.3546,
+      "step": 96330
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.399823984082039e-05,
+      "loss": 1.2613,
+      "step": 96340
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.399709191092064e-05,
+      "loss": 1.4399,
+      "step": 96350
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.399594398102089e-05,
+      "loss": 1.4021,
+      "step": 96360
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.399479605112114e-05,
+      "loss": 1.2637,
+      "step": 96370
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.39936481212214e-05,
+      "loss": 1.2213,
+      "step": 96380
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.399250019132164e-05,
+      "loss": 1.5746,
+      "step": 96390
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.39913522614219e-05,
+      "loss": 1.5673,
+      "step": 96400
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.399020433152215e-05,
+      "loss": 1.3759,
+      "step": 96410
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.39890564016224e-05,
+      "loss": 1.2475,
+      "step": 96420
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.398790847172265e-05,
+      "loss": 1.257,
+      "step": 96430
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.398676054182291e-05,
+      "loss": 1.3029,
+      "step": 96440
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.398561261192316e-05,
+      "loss": 1.1569,
+      "step": 96450
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.398446468202341e-05,
+      "loss": 1.3535,
+      "step": 96460
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.398331675212366e-05,
+      "loss": 1.2424,
+      "step": 96470
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.398216882222391e-05,
+      "loss": 1.1813,
+      "step": 96480
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.398102089232418e-05,
+      "loss": 1.314,
+      "step": 96490
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.397987296242443e-05,
+      "loss": 1.2956,
+      "step": 96500
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.397872503252468e-05,
+      "loss": 1.3172,
+      "step": 96510
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.397757710262493e-05,
+      "loss": 1.3021,
+      "step": 96520
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.397642917272519e-05,
+      "loss": 1.1128,
+      "step": 96530
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.397528124282543e-05,
+      "loss": 1.2848,
+      "step": 96540
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.397413331292569e-05,
+      "loss": 1.1761,
+      "step": 96550
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.397298538302594e-05,
+      "loss": 1.2513,
+      "step": 96560
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.397183745312619e-05,
+      "loss": 1.2472,
+      "step": 96570
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.397068952322644e-05,
+      "loss": 1.2041,
+      "step": 96580
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.39695415933267e-05,
+      "loss": 1.4692,
+      "step": 96590
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.396839366342695e-05,
+      "loss": 1.4263,
+      "step": 96600
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.39672457335272e-05,
+      "loss": 1.2666,
+      "step": 96610
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.396609780362745e-05,
+      "loss": 1.3459,
+      "step": 96620
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.39649498737277e-05,
+      "loss": 1.3995,
+      "step": 96630
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.396380194382796e-05,
+      "loss": 1.3244,
+      "step": 96640
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.396265401392821e-05,
+      "loss": 1.3593,
+      "step": 96650
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.396150608402846e-05,
+      "loss": 1.2017,
+      "step": 96660
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.396035815412871e-05,
+      "loss": 1.4495,
+      "step": 96670
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.395921022422898e-05,
+      "loss": 1.33,
+      "step": 96680
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.395806229432921e-05,
+      "loss": 1.5375,
+      "step": 96690
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.395691436442948e-05,
+      "loss": 1.3068,
+      "step": 96700
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.395576643452973e-05,
+      "loss": 1.6526,
+      "step": 96710
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.395461850462998e-05,
+      "loss": 1.3259,
+      "step": 96720
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.395347057473023e-05,
+      "loss": 1.2881,
+      "step": 96730
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.395232264483049e-05,
+      "loss": 1.38,
+      "step": 96740
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.395117471493074e-05,
+      "loss": 1.1923,
+      "step": 96750
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.395002678503099e-05,
+      "loss": 1.1943,
+      "step": 96760
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.394887885513124e-05,
+      "loss": 1.3264,
+      "step": 96770
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.394773092523149e-05,
+      "loss": 1.2413,
+      "step": 96780
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.394658299533175e-05,
+      "loss": 1.1379,
+      "step": 96790
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.3945435065432e-05,
+      "loss": 1.2789,
+      "step": 96800
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.394428713553225e-05,
+      "loss": 1.3233,
+      "step": 96810
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.39431392056325e-05,
+      "loss": 1.3034,
+      "step": 96820
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.394199127573277e-05,
+      "loss": 1.3037,
+      "step": 96830
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.3940843345833e-05,
+      "loss": 1.3535,
+      "step": 96840
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.393969541593327e-05,
+      "loss": 1.2803,
+      "step": 96850
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.393854748603352e-05,
+      "loss": 1.3469,
+      "step": 96860
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.393739955613377e-05,
+      "loss": 1.1984,
+      "step": 96870
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.393625162623402e-05,
+      "loss": 1.2871,
+      "step": 96880
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.393510369633428e-05,
+      "loss": 1.2256,
+      "step": 96890
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.393395576643453e-05,
+      "loss": 1.2809,
+      "step": 96900
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.393280783653478e-05,
+      "loss": 1.2369,
+      "step": 96910
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.393165990663503e-05,
+      "loss": 1.2799,
+      "step": 96920
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.393051197673528e-05,
+      "loss": 1.368,
+      "step": 96930
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.392936404683554e-05,
+      "loss": 1.3497,
+      "step": 96940
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.392821611693579e-05,
+      "loss": 1.4152,
+      "step": 96950
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.392706818703604e-05,
+      "loss": 1.3035,
+      "step": 96960
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.392592025713629e-05,
+      "loss": 1.4377,
+      "step": 96970
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.392477232723655e-05,
+      "loss": 1.2292,
+      "step": 96980
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.392362439733679e-05,
+      "loss": 1.2635,
+      "step": 96990
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.392247646743705e-05,
+      "loss": 1.3159,
+      "step": 97000
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.39213285375373e-05,
+      "loss": 1.3203,
+      "step": 97010
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.392018060763755e-05,
+      "loss": 1.2779,
+      "step": 97020
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.39190326777378e-05,
+      "loss": 1.2291,
+      "step": 97030
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.391788474783807e-05,
+      "loss": 1.3108,
+      "step": 97040
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.391673681793832e-05,
+      "loss": 1.3539,
+      "step": 97050
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.391558888803857e-05,
+      "loss": 1.2014,
+      "step": 97060
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.391444095813882e-05,
+      "loss": 1.3091,
+      "step": 97070
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.391329302823907e-05,
+      "loss": 1.5279,
+      "step": 97080
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.391214509833933e-05,
+      "loss": 1.5235,
+      "step": 97090
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.391099716843958e-05,
+      "loss": 1.291,
+      "step": 97100
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.390984923853983e-05,
+      "loss": 1.2637,
+      "step": 97110
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.390870130864008e-05,
+      "loss": 1.2709,
+      "step": 97120
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.390755337874034e-05,
+      "loss": 1.2905,
+      "step": 97130
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.390640544884058e-05,
+      "loss": 1.4063,
+      "step": 97140
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.390525751894084e-05,
+      "loss": 1.3594,
+      "step": 97150
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.390410958904109e-05,
+      "loss": 1.3185,
+      "step": 97160
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.390296165914134e-05,
+      "loss": 1.2279,
+      "step": 97170
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.390181372924159e-05,
+      "loss": 1.2181,
+      "step": 97180
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.390066579934186e-05,
+      "loss": 1.3477,
+      "step": 97190
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.38995178694421e-05,
+      "loss": 1.1987,
+      "step": 97200
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.389836993954236e-05,
+      "loss": 1.3461,
+      "step": 97210
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.38972220096426e-05,
+      "loss": 1.2787,
+      "step": 97220
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.389607407974285e-05,
+      "loss": 1.3417,
+      "step": 97230
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.389492614984312e-05,
+      "loss": 1.1621,
+      "step": 97240
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.389377821994337e-05,
+      "loss": 1.2998,
+      "step": 97250
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.389263029004362e-05,
+      "loss": 1.3139,
+      "step": 97260
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.389148236014387e-05,
+      "loss": 1.4012,
+      "step": 97270
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.389033443024412e-05,
+      "loss": 1.4367,
+      "step": 97280
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.388918650034437e-05,
+      "loss": 1.4464,
+      "step": 97290
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.388803857044463e-05,
+      "loss": 1.3379,
+      "step": 97300
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.388689064054488e-05,
+      "loss": 1.2056,
+      "step": 97310
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 6.388574271064513e-05,
+      "loss": 1.345,
+      "step": 97320
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 6.388459478074538e-05,
+      "loss": 1.376,
+      "step": 97330
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 6.388344685084563e-05,
+      "loss": 1.2122,
+      "step": 97340
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 6.38822989209459e-05,
+      "loss": 1.478,
+      "step": 97350
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 6.388115099104614e-05,
+      "loss": 1.3144,
+      "step": 97360
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 6.38800030611464e-05,
+      "loss": 1.307,
+      "step": 97370
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 6.387885513124664e-05,
+      "loss": 1.362,
+      "step": 97380
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 6.38777072013469e-05,
+      "loss": 1.311,
+      "step": 97390
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 6.387655927144714e-05,
+      "loss": 1.2466,
+      "step": 97400
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 6.38754113415474e-05,
+      "loss": 1.2595,
+      "step": 97410
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 6.387426341164766e-05,
+      "loss": 1.3348,
+      "step": 97420
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 6.38731154817479e-05,
+      "loss": 1.2388,
+      "step": 97430
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 6.387196755184816e-05,
+      "loss": 1.2228,
+      "step": 97440
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 6.387081962194842e-05,
+      "loss": 1.3693,
+      "step": 97450
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 6.386967169204867e-05,
+      "loss": 1.3033,
+      "step": 97460
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 6.386852376214892e-05,
+      "loss": 1.4659,
+      "step": 97470
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 6.386737583224917e-05,
+      "loss": 1.3848,
+      "step": 97480
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 6.386622790234942e-05,
+      "loss": 1.4148,
+      "step": 97490
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 6.386507997244968e-05,
+      "loss": 1.3942,
+      "step": 97500
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 6.386393204254993e-05,
+      "loss": 1.3852,
+      "step": 97510
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 6.386278411265018e-05,
+      "loss": 1.3444,
+      "step": 97520
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 6.386163618275043e-05,
+      "loss": 1.2146,
+      "step": 97530
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 6.38604882528507e-05,
+      "loss": 1.1993,
+      "step": 97540
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 6.385934032295093e-05,
+      "loss": 1.4108,
+      "step": 97550
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 6.38581923930512e-05,
+      "loss": 1.4184,
+      "step": 97560
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 6.385704446315144e-05,
+      "loss": 1.37,
+      "step": 97570
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 6.38558965332517e-05,
+      "loss": 1.4049,
+      "step": 97580
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 6.385474860335194e-05,
+      "loss": 1.3109,
+      "step": 97590
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 6.385360067345221e-05,
+      "loss": 1.145,
+      "step": 97600
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 6.385245274355246e-05,
+      "loss": 1.2696,
+      "step": 97610
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 6.385130481365271e-05,
+      "loss": 1.2895,
+      "step": 97620
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 6.385015688375296e-05,
+      "loss": 1.2487,
+      "step": 97630
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 6.384900895385321e-05,
+      "loss": 1.3701,
+      "step": 97640
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 6.384786102395347e-05,
+      "loss": 1.3083,
+      "step": 97650
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 6.384671309405372e-05,
+      "loss": 1.4427,
+      "step": 97660
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 6.384556516415397e-05,
+      "loss": 1.5579,
+      "step": 97670
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 6.384441723425422e-05,
+      "loss": 1.4361,
+      "step": 97680
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 6.384326930435448e-05,
+      "loss": 1.3223,
+      "step": 97690
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 6.384212137445472e-05,
+      "loss": 1.2991,
+      "step": 97700
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 6.384097344455498e-05,
+      "loss": 1.2672,
+      "step": 97710
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 6.383982551465523e-05,
+      "loss": 1.2768,
+      "step": 97720
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 6.383867758475548e-05,
+      "loss": 1.4,
+      "step": 97730
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 6.383752965485573e-05,
+      "loss": 1.3999,
+      "step": 97740
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 6.3836381724956e-05,
+      "loss": 1.2711,
+      "step": 97750
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 6.383523379505625e-05,
+      "loss": 1.4563,
+      "step": 97760
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 6.38340858651565e-05,
+      "loss": 1.3316,
+      "step": 97770
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 6.383293793525675e-05,
+      "loss": 1.2808,
+      "step": 97780
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 6.3831790005357e-05,
+      "loss": 1.3335,
+      "step": 97790
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 6.383064207545726e-05,
+      "loss": 1.3896,
+      "step": 97800
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 6.382949414555751e-05,
+      "loss": 1.2832,
+      "step": 97810
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 6.382834621565776e-05,
+      "loss": 1.3203,
+      "step": 97820
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 6.382719828575801e-05,
+      "loss": 1.3634,
+      "step": 97830
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 6.382605035585827e-05,
+      "loss": 1.3228,
+      "step": 97840
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 6.382490242595851e-05,
+      "loss": 1.2153,
+      "step": 97850
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 6.382375449605877e-05,
+      "loss": 1.2905,
+      "step": 97860
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 6.382260656615902e-05,
+      "loss": 1.4213,
+      "step": 97870
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 6.382145863625927e-05,
+      "loss": 1.2601,
+      "step": 97880
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 6.382031070635952e-05,
+      "loss": 1.3095,
+      "step": 97890
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 6.381916277645978e-05,
+      "loss": 1.3972,
+      "step": 97900
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 6.381801484656003e-05,
+      "loss": 1.2224,
+      "step": 97910
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 6.381686691666028e-05,
+      "loss": 1.2683,
+      "step": 97920
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 6.381571898676053e-05,
+      "loss": 1.3584,
+      "step": 97930
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 6.381457105686078e-05,
+      "loss": 1.4505,
+      "step": 97940
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 6.381342312696105e-05,
+      "loss": 1.4558,
+      "step": 97950
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 6.38122751970613e-05,
+      "loss": 1.4414,
+      "step": 97960
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 6.381112726716155e-05,
+      "loss": 1.2127,
+      "step": 97970
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 6.38099793372618e-05,
+      "loss": 1.2117,
+      "step": 97980
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 6.380883140736206e-05,
+      "loss": 1.3621,
+      "step": 97990
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 6.38076834774623e-05,
+      "loss": 1.3903,
+      "step": 98000
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 6.380653554756256e-05,
+      "loss": 1.453,
+      "step": 98010
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 6.380538761766281e-05,
+      "loss": 1.2098,
+      "step": 98020
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 6.380423968776306e-05,
+      "loss": 1.4257,
+      "step": 98030
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 6.380309175786331e-05,
+      "loss": 1.1059,
+      "step": 98040
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 6.380194382796357e-05,
+      "loss": 1.2642,
+      "step": 98050
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 6.380079589806382e-05,
+      "loss": 1.2783,
+      "step": 98060
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 6.379964796816407e-05,
+      "loss": 1.3907,
+      "step": 98070
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 6.379850003826432e-05,
+      "loss": 1.3225,
+      "step": 98080
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 6.379735210836457e-05,
+      "loss": 1.4436,
+      "step": 98090
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 6.379620417846484e-05,
+      "loss": 1.2149,
+      "step": 98100
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 6.379505624856509e-05,
+      "loss": 1.4273,
+      "step": 98110
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 6.379390831866534e-05,
+      "loss": 1.1914,
+      "step": 98120
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 6.379276038876559e-05,
+      "loss": 1.2266,
+      "step": 98130
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 6.379161245886585e-05,
+      "loss": 1.3856,
+      "step": 98140
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 6.379046452896609e-05,
+      "loss": 1.4474,
+      "step": 98150
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 6.378931659906635e-05,
+      "loss": 1.3261,
+      "step": 98160
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 6.37881686691666e-05,
+      "loss": 1.4,
+      "step": 98170
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 6.378702073926685e-05,
+      "loss": 1.2931,
+      "step": 98180
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.37858728093671e-05,
+      "loss": 1.3609,
+      "step": 98190
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.378472487946736e-05,
+      "loss": 1.3044,
+      "step": 98200
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.378357694956761e-05,
+      "loss": 1.2577,
+      "step": 98210
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.378242901966786e-05,
+      "loss": 1.2157,
+      "step": 98220
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.378128108976811e-05,
+      "loss": 1.3919,
+      "step": 98230
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.378013315986836e-05,
+      "loss": 1.365,
+      "step": 98240
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.377898522996862e-05,
+      "loss": 1.3063,
+      "step": 98250
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.377783730006887e-05,
+      "loss": 1.3448,
+      "step": 98260
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.377668937016912e-05,
+      "loss": 1.344,
+      "step": 98270
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.377554144026937e-05,
+      "loss": 1.386,
+      "step": 98280
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.377439351036964e-05,
+      "loss": 1.2214,
+      "step": 98290
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.377324558046987e-05,
+      "loss": 1.3012,
+      "step": 98300
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.377209765057014e-05,
+      "loss": 1.3297,
+      "step": 98310
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.377094972067039e-05,
+      "loss": 1.2984,
+      "step": 98320
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.376980179077064e-05,
+      "loss": 1.2903,
+      "step": 98330
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.376865386087089e-05,
+      "loss": 1.2098,
+      "step": 98340
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.376750593097115e-05,
+      "loss": 1.2944,
+      "step": 98350
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.37663580010714e-05,
+      "loss": 1.1104,
+      "step": 98360
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.376521007117165e-05,
+      "loss": 1.4719,
+      "step": 98370
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.37640621412719e-05,
+      "loss": 1.2509,
+      "step": 98380
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.376291421137215e-05,
+      "loss": 1.3883,
+      "step": 98390
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 6.376176628147241e-05,
+      "loss": 1.3636,
+      "step": 98400
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.376061835157266e-05,
+      "loss": 1.3801,
+      "step": 98410
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.375947042167291e-05,
+      "loss": 1.3436,
+      "step": 98420
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.375832249177316e-05,
+      "loss": 1.3069,
+      "step": 98430
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.375717456187343e-05,
+      "loss": 1.3163,
+      "step": 98440
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.375602663197366e-05,
+      "loss": 1.1002,
+      "step": 98450
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.375487870207393e-05,
+      "loss": 1.3734,
+      "step": 98460
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.375373077217418e-05,
+      "loss": 1.3082,
+      "step": 98470
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.375258284227443e-05,
+      "loss": 1.2656,
+      "step": 98480
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.375143491237468e-05,
+      "loss": 1.182,
+      "step": 98490
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.375028698247494e-05,
+      "loss": 1.5219,
+      "step": 98500
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.374913905257519e-05,
+      "loss": 1.1923,
+      "step": 98510
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.374799112267544e-05,
+      "loss": 1.4234,
+      "step": 98520
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.374684319277569e-05,
+      "loss": 1.4779,
+      "step": 98530
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.374569526287594e-05,
+      "loss": 1.3917,
+      "step": 98540
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.37445473329762e-05,
+      "loss": 1.2915,
+      "step": 98550
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.374339940307645e-05,
+      "loss": 1.4283,
+      "step": 98560
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.37422514731767e-05,
+      "loss": 1.3253,
+      "step": 98570
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.374110354327695e-05,
+      "loss": 1.3816,
+      "step": 98580
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.373995561337721e-05,
+      "loss": 1.3218,
+      "step": 98590
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.373880768347745e-05,
+      "loss": 1.2994,
+      "step": 98600
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.373765975357771e-05,
+      "loss": 1.3496,
+      "step": 98610
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.373651182367796e-05,
+      "loss": 1.3124,
+      "step": 98620
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.373536389377821e-05,
+      "loss": 1.3323,
+      "step": 98630
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.373421596387846e-05,
+      "loss": 1.4783,
+      "step": 98640
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.373306803397871e-05,
+      "loss": 1.3903,
+      "step": 98650
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.373192010407898e-05,
+      "loss": 1.3725,
+      "step": 98660
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.373077217417923e-05,
+      "loss": 1.3753,
+      "step": 98670
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.372962424427948e-05,
+      "loss": 1.3875,
+      "step": 98680
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.372847631437973e-05,
+      "loss": 1.3514,
+      "step": 98690
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.372732838447999e-05,
+      "loss": 1.3185,
+      "step": 98700
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.372618045458023e-05,
+      "loss": 1.3082,
+      "step": 98710
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.372503252468049e-05,
+      "loss": 1.2072,
+      "step": 98720
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.372388459478074e-05,
+      "loss": 1.4675,
+      "step": 98730
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.372273666488099e-05,
+      "loss": 1.2468,
+      "step": 98740
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.372158873498124e-05,
+      "loss": 1.2105,
+      "step": 98750
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.37204408050815e-05,
+      "loss": 1.3907,
+      "step": 98760
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.371929287518175e-05,
+      "loss": 1.3112,
+      "step": 98770
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.3718144945282e-05,
+      "loss": 1.2761,
+      "step": 98780
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.371699701538225e-05,
+      "loss": 1.3272,
+      "step": 98790
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.37158490854825e-05,
+      "loss": 1.3361,
+      "step": 98800
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.371470115558277e-05,
+      "loss": 1.257,
+      "step": 98810
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.371355322568302e-05,
+      "loss": 1.4276,
+      "step": 98820
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.371240529578327e-05,
+      "loss": 1.3706,
+      "step": 98830
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.371125736588352e-05,
+      "loss": 1.2005,
+      "step": 98840
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.371010943598378e-05,
+      "loss": 1.2725,
+      "step": 98850
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.370896150608401e-05,
+      "loss": 1.3719,
+      "step": 98860
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.370781357618428e-05,
+      "loss": 1.2845,
+      "step": 98870
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.370666564628453e-05,
+      "loss": 1.4034,
+      "step": 98880
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.370551771638478e-05,
+      "loss": 1.2779,
+      "step": 98890
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.370436978648503e-05,
+      "loss": 1.3755,
+      "step": 98900
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.370322185658529e-05,
+      "loss": 1.4227,
+      "step": 98910
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.370207392668554e-05,
+      "loss": 1.2227,
+      "step": 98920
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.370092599678579e-05,
+      "loss": 1.3466,
+      "step": 98930
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.369977806688604e-05,
+      "loss": 1.4333,
+      "step": 98940
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.369863013698629e-05,
+      "loss": 1.2855,
+      "step": 98950
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.369748220708655e-05,
+      "loss": 1.3099,
+      "step": 98960
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.36963342771868e-05,
+      "loss": 1.2793,
+      "step": 98970
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.369518634728705e-05,
+      "loss": 1.3152,
+      "step": 98980
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.36940384173873e-05,
+      "loss": 1.3002,
+      "step": 98990
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.369289048748757e-05,
+      "loss": 1.2423,
+      "step": 99000
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.36917425575878e-05,
+      "loss": 1.2717,
+      "step": 99010
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.369059462768807e-05,
+      "loss": 1.323,
+      "step": 99020
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.368944669778832e-05,
+      "loss": 1.2237,
+      "step": 99030
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.368829876788857e-05,
+      "loss": 1.2819,
+      "step": 99040
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.368715083798882e-05,
+      "loss": 1.1535,
+      "step": 99050
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.368600290808908e-05,
+      "loss": 1.2699,
+      "step": 99060
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.368485497818933e-05,
+      "loss": 1.2749,
+      "step": 99070
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.368370704828958e-05,
+      "loss": 1.4022,
+      "step": 99080
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.368255911838983e-05,
+      "loss": 1.3155,
+      "step": 99090
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.368141118849008e-05,
+      "loss": 1.4511,
+      "step": 99100
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.368026325859034e-05,
+      "loss": 1.2393,
+      "step": 99110
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.367911532869059e-05,
+      "loss": 1.2599,
+      "step": 99120
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.367796739879084e-05,
+      "loss": 1.1838,
+      "step": 99130
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.367681946889109e-05,
+      "loss": 1.3537,
+      "step": 99140
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.367567153899136e-05,
+      "loss": 1.2485,
+      "step": 99150
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.367452360909159e-05,
+      "loss": 1.1353,
+      "step": 99160
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.367337567919186e-05,
+      "loss": 1.3295,
+      "step": 99170
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.36722277492921e-05,
+      "loss": 1.5718,
+      "step": 99180
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.367107981939235e-05,
+      "loss": 1.2734,
+      "step": 99190
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.36699318894926e-05,
+      "loss": 1.2468,
+      "step": 99200
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.366878395959287e-05,
+      "loss": 1.3688,
+      "step": 99210
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.366763602969312e-05,
+      "loss": 1.3508,
+      "step": 99220
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.366648809979337e-05,
+      "loss": 1.3801,
+      "step": 99230
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.366534016989362e-05,
+      "loss": 1.2255,
+      "step": 99240
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.366419223999387e-05,
+      "loss": 1.3815,
+      "step": 99250
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.366304431009413e-05,
+      "loss": 1.3365,
+      "step": 99260
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.366189638019438e-05,
+      "loss": 1.2289,
+      "step": 99270
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.366074845029463e-05,
+      "loss": 1.3326,
+      "step": 99280
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.365960052039488e-05,
+      "loss": 1.3718,
+      "step": 99290
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.365845259049514e-05,
+      "loss": 1.3566,
+      "step": 99300
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.365730466059538e-05,
+      "loss": 1.3131,
+      "step": 99310
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.365615673069564e-05,
+      "loss": 1.3604,
+      "step": 99320
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.36550088007959e-05,
+      "loss": 1.2166,
+      "step": 99330
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.365386087089614e-05,
+      "loss": 1.2763,
+      "step": 99340
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.365271294099639e-05,
+      "loss": 1.3661,
+      "step": 99350
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.365156501109666e-05,
+      "loss": 1.1833,
+      "step": 99360
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.36504170811969e-05,
+      "loss": 1.3116,
+      "step": 99370
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.364926915129716e-05,
+      "loss": 1.213,
+      "step": 99380
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.36481212213974e-05,
+      "loss": 1.4127,
+      "step": 99390
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.364697329149766e-05,
+      "loss": 1.2644,
+      "step": 99400
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.364582536159792e-05,
+      "loss": 1.4492,
+      "step": 99410
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.364467743169817e-05,
+      "loss": 1.3017,
+      "step": 99420
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.364352950179842e-05,
+      "loss": 1.3372,
+      "step": 99430
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.364238157189867e-05,
+      "loss": 1.336,
+      "step": 99440
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.364123364199893e-05,
+      "loss": 1.4386,
+      "step": 99450
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.364008571209917e-05,
+      "loss": 1.3051,
+      "step": 99460
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.363893778219943e-05,
+      "loss": 1.3122,
+      "step": 99470
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.363778985229968e-05,
+      "loss": 1.0705,
+      "step": 99480
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 6.363664192239993e-05,
+      "loss": 1.3739,
+      "step": 99490
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.363549399250018e-05,
+      "loss": 1.5142,
+      "step": 99500
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.363434606260045e-05,
+      "loss": 1.2641,
+      "step": 99510
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.36331981327007e-05,
+      "loss": 1.3282,
+      "step": 99520
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.363205020280094e-05,
+      "loss": 1.2818,
+      "step": 99530
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.36309022729012e-05,
+      "loss": 1.3216,
+      "step": 99540
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.362975434300144e-05,
+      "loss": 1.3834,
+      "step": 99550
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.362860641310171e-05,
+      "loss": 1.4215,
+      "step": 99560
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.362745848320196e-05,
+      "loss": 1.3322,
+      "step": 99570
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.362631055330221e-05,
+      "loss": 1.2897,
+      "step": 99580
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.362516262340246e-05,
+      "loss": 1.2731,
+      "step": 99590
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.362401469350272e-05,
+      "loss": 1.3082,
+      "step": 99600
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.362286676360296e-05,
+      "loss": 1.3602,
+      "step": 99610
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.362171883370322e-05,
+      "loss": 1.4056,
+      "step": 99620
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.362057090380347e-05,
+      "loss": 1.2535,
+      "step": 99630
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.361942297390372e-05,
+      "loss": 1.4055,
+      "step": 99640
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.361827504400397e-05,
+      "loss": 1.2922,
+      "step": 99650
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.361712711410423e-05,
+      "loss": 1.2671,
+      "step": 99660
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.361597918420448e-05,
+      "loss": 1.4016,
+      "step": 99670
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.361483125430473e-05,
+      "loss": 1.2107,
+      "step": 99680
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.361368332440498e-05,
+      "loss": 1.1573,
+      "step": 99690
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.361253539450523e-05,
+      "loss": 1.2821,
+      "step": 99700
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 6.36113874646055e-05,
+      "loss": 1.2887,
+      "step": 99710
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.361023953470575e-05,
+      "loss": 1.2734,
+      "step": 99720
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.3609091604806e-05,
+      "loss": 1.3605,
+      "step": 99730
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.360794367490625e-05,
+      "loss": 1.2854,
+      "step": 99740
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.360679574500651e-05,
+      "loss": 1.3528,
+      "step": 99750
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.360564781510675e-05,
+      "loss": 1.2287,
+      "step": 99760
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.360449988520701e-05,
+      "loss": 1.2672,
+      "step": 99770
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.360335195530726e-05,
+      "loss": 1.2388,
+      "step": 99780
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.360220402540751e-05,
+      "loss": 1.3277,
+      "step": 99790
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.360105609550776e-05,
+      "loss": 1.313,
+      "step": 99800
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.359990816560802e-05,
+      "loss": 1.2995,
+      "step": 99810
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.359876023570827e-05,
+      "loss": 1.3034,
+      "step": 99820
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.359761230580852e-05,
+      "loss": 1.4501,
+      "step": 99830
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.359646437590877e-05,
+      "loss": 1.2474,
+      "step": 99840
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.359531644600902e-05,
+      "loss": 1.2569,
+      "step": 99850
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.359416851610928e-05,
+      "loss": 1.3094,
+      "step": 99860
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.359302058620953e-05,
+      "loss": 1.3536,
+      "step": 99870
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.359187265630978e-05,
+      "loss": 1.2991,
+      "step": 99880
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.359072472641003e-05,
+      "loss": 1.2928,
+      "step": 99890
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.35895767965103e-05,
+      "loss": 1.1717,
+      "step": 99900
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.358842886661053e-05,
+      "loss": 1.2674,
+      "step": 99910
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.35872809367108e-05,
+      "loss": 1.3909,
+      "step": 99920
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 6.358613300681105e-05,
+      "loss": 1.2806,
+      "step": 99930
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.35849850769113e-05,
+      "loss": 1.3183,
+      "step": 99940
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.358383714701155e-05,
+      "loss": 1.3834,
+      "step": 99950
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.358268921711181e-05,
+      "loss": 1.4592,
+      "step": 99960
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.358154128721206e-05,
+      "loss": 1.3867,
+      "step": 99970
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.358039335731231e-05,
+      "loss": 1.2121,
+      "step": 99980
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.357924542741256e-05,
+      "loss": 1.368,
+      "step": 99990
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.357809749751281e-05,
+      "loss": 1.2296,
+      "step": 100000
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.357694956761307e-05,
+      "loss": 1.2499,
+      "step": 100010
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.357580163771331e-05,
+      "loss": 1.1717,
+      "step": 100020
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.357465370781357e-05,
+      "loss": 1.4083,
+      "step": 100030
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.357350577791382e-05,
+      "loss": 1.4559,
+      "step": 100040
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.357235784801407e-05,
+      "loss": 1.282,
+      "step": 100050
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.357120991811432e-05,
+      "loss": 1.3168,
+      "step": 100060
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.357006198821459e-05,
+      "loss": 1.1919,
+      "step": 100070
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.356891405831484e-05,
+      "loss": 1.3922,
+      "step": 100080
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.356776612841509e-05,
+      "loss": 1.26,
+      "step": 100090
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.356661819851534e-05,
+      "loss": 1.4954,
+      "step": 100100
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.356547026861559e-05,
+      "loss": 1.2193,
+      "step": 100110
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.356432233871585e-05,
+      "loss": 1.2594,
+      "step": 100120
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.35631744088161e-05,
+      "loss": 1.2231,
+      "step": 100130
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 6.356202647891635e-05,
+      "loss": 1.2128,
+      "step": 100140
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.35608785490166e-05,
+      "loss": 1.401,
+      "step": 100150
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.355973061911686e-05,
+      "loss": 1.1716,
+      "step": 100160
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.35585826892171e-05,
+      "loss": 1.2584,
+      "step": 100170
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.355743475931736e-05,
+      "loss": 1.3279,
+      "step": 100180
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.355628682941761e-05,
+      "loss": 1.3532,
+      "step": 100190
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.355513889951786e-05,
+      "loss": 1.4343,
+      "step": 100200
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.355399096961811e-05,
+      "loss": 1.3772,
+      "step": 100210
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.355284303971837e-05,
+      "loss": 1.1452,
+      "step": 100220
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.355169510981862e-05,
+      "loss": 1.2814,
+      "step": 100230
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.355054717991887e-05,
+      "loss": 1.217,
+      "step": 100240
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.354939925001912e-05,
+      "loss": 1.452,
+      "step": 100250
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.354825132011937e-05,
+      "loss": 1.424,
+      "step": 100260
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.354710339021964e-05,
+      "loss": 1.1726,
+      "step": 100270
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.354595546031989e-05,
+      "loss": 1.3985,
+      "step": 100280
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.354480753042014e-05,
+      "loss": 1.1664,
+      "step": 100290
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.354365960052039e-05,
+      "loss": 1.3401,
+      "step": 100300
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.354251167062065e-05,
+      "loss": 1.3925,
+      "step": 100310
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.354136374072089e-05,
+      "loss": 1.1904,
+      "step": 100320
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.354021581082115e-05,
+      "loss": 1.2537,
+      "step": 100330
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.35390678809214e-05,
+      "loss": 1.3697,
+      "step": 100340
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.353791995102165e-05,
+      "loss": 1.2801,
+      "step": 100350
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 6.35367720211219e-05,
+      "loss": 1.3095,
+      "step": 100360
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.353562409122216e-05,
+      "loss": 1.3034,
+      "step": 100370
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.353447616132241e-05,
+      "loss": 1.2391,
+      "step": 100380
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.353332823142266e-05,
+      "loss": 1.2934,
+      "step": 100390
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.353218030152291e-05,
+      "loss": 1.2624,
+      "step": 100400
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.353103237162316e-05,
+      "loss": 1.2797,
+      "step": 100410
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.352988444172343e-05,
+      "loss": 1.3518,
+      "step": 100420
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.352873651182368e-05,
+      "loss": 1.2628,
+      "step": 100430
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.352758858192393e-05,
+      "loss": 1.2357,
+      "step": 100440
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.352644065202418e-05,
+      "loss": 1.2508,
+      "step": 100450
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.352529272212444e-05,
+      "loss": 1.3097,
+      "step": 100460
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.352414479222468e-05,
+      "loss": 1.2989,
+      "step": 100470
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.352299686232494e-05,
+      "loss": 1.3963,
+      "step": 100480
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.352184893242519e-05,
+      "loss": 1.3267,
+      "step": 100490
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.352070100252544e-05,
+      "loss": 1.3105,
+      "step": 100500
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.351955307262569e-05,
+      "loss": 1.3417,
+      "step": 100510
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.351840514272595e-05,
+      "loss": 1.2245,
+      "step": 100520
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.35172572128262e-05,
+      "loss": 1.3123,
+      "step": 100530
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.351610928292645e-05,
+      "loss": 1.2204,
+      "step": 100540
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.35149613530267e-05,
+      "loss": 1.238,
+      "step": 100550
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.351381342312695e-05,
+      "loss": 1.2221,
+      "step": 100560
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.351266549322721e-05,
+      "loss": 1.2433,
+      "step": 100570
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 6.351151756332746e-05,
+      "loss": 1.3962,
+      "step": 100580
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.351036963342771e-05,
+      "loss": 1.3347,
+      "step": 100590
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.350922170352796e-05,
+      "loss": 1.4681,
+      "step": 100600
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.350807377362823e-05,
+      "loss": 1.4259,
+      "step": 100610
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.350692584372846e-05,
+      "loss": 1.3484,
+      "step": 100620
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.350577791382873e-05,
+      "loss": 1.1895,
+      "step": 100630
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.350462998392898e-05,
+      "loss": 1.2431,
+      "step": 100640
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.350348205402923e-05,
+      "loss": 1.4081,
+      "step": 100650
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.350233412412948e-05,
+      "loss": 1.3122,
+      "step": 100660
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.350118619422974e-05,
+      "loss": 1.327,
+      "step": 100670
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.350003826432999e-05,
+      "loss": 1.1764,
+      "step": 100680
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.349889033443024e-05,
+      "loss": 1.1798,
+      "step": 100690
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.349774240453049e-05,
+      "loss": 1.2928,
+      "step": 100700
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.349659447463074e-05,
+      "loss": 1.381,
+      "step": 100710
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.3495446544731e-05,
+      "loss": 1.3136,
+      "step": 100720
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.349429861483125e-05,
+      "loss": 1.3042,
+      "step": 100730
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.34931506849315e-05,
+      "loss": 1.3095,
+      "step": 100740
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.349200275503175e-05,
+      "loss": 1.1826,
+      "step": 100750
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.349085482513202e-05,
+      "loss": 1.2133,
+      "step": 100760
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.348970689523225e-05,
+      "loss": 1.3947,
+      "step": 100770
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.348855896533252e-05,
+      "loss": 1.2309,
+      "step": 100780
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.348741103543277e-05,
+      "loss": 1.3239,
+      "step": 100790
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 6.348626310553302e-05,
+      "loss": 1.3071,
+      "step": 100800
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.348511517563326e-05,
+      "loss": 1.2799,
+      "step": 100810
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.348396724573353e-05,
+      "loss": 1.3197,
+      "step": 100820
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.348281931583378e-05,
+      "loss": 1.4869,
+      "step": 100830
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.348167138593403e-05,
+      "loss": 1.3757,
+      "step": 100840
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.348052345603428e-05,
+      "loss": 1.3111,
+      "step": 100850
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.347937552613453e-05,
+      "loss": 1.3409,
+      "step": 100860
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.347822759623479e-05,
+      "loss": 1.3423,
+      "step": 100870
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.347707966633504e-05,
+      "loss": 1.3476,
+      "step": 100880
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.347593173643529e-05,
+      "loss": 1.3394,
+      "step": 100890
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.347478380653554e-05,
+      "loss": 1.2776,
+      "step": 100900
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.34736358766358e-05,
+      "loss": 1.271,
+      "step": 100910
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.347248794673604e-05,
+      "loss": 1.4771,
+      "step": 100920
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.34713400168363e-05,
+      "loss": 1.2799,
+      "step": 100930
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.347019208693655e-05,
+      "loss": 1.3596,
+      "step": 100940
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.34690441570368e-05,
+      "loss": 1.5777,
+      "step": 100950
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.346789622713705e-05,
+      "loss": 1.3179,
+      "step": 100960
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.346674829723732e-05,
+      "loss": 1.3464,
+      "step": 100970
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.346560036733757e-05,
+      "loss": 1.4711,
+      "step": 100980
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.346445243743782e-05,
+      "loss": 1.425,
+      "step": 100990
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.346330450753807e-05,
+      "loss": 1.3276,
+      "step": 101000
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.346215657763832e-05,
+      "loss": 1.3209,
+      "step": 101010
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 6.346100864773858e-05,
+      "loss": 1.2149,
+      "step": 101020
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.345986071783883e-05,
+      "loss": 1.2776,
+      "step": 101030
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.345871278793908e-05,
+      "loss": 1.3246,
+      "step": 101040
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.345756485803933e-05,
+      "loss": 1.3107,
+      "step": 101050
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.345641692813959e-05,
+      "loss": 1.3116,
+      "step": 101060
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.345526899823983e-05,
+      "loss": 1.3727,
+      "step": 101070
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.345412106834009e-05,
+      "loss": 1.1365,
+      "step": 101080
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.345297313844034e-05,
+      "loss": 1.2765,
+      "step": 101090
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.345182520854059e-05,
+      "loss": 1.3521,
+      "step": 101100
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.345067727864084e-05,
+      "loss": 1.4431,
+      "step": 101110
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.34495293487411e-05,
+      "loss": 1.351,
+      "step": 101120
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.344838141884136e-05,
+      "loss": 1.5228,
+      "step": 101130
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.34472334889416e-05,
+      "loss": 1.1689,
+      "step": 101140
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.344608555904185e-05,
+      "loss": 1.2469,
+      "step": 101150
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.34449376291421e-05,
+      "loss": 1.3458,
+      "step": 101160
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.344378969924237e-05,
+      "loss": 1.2977,
+      "step": 101170
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.344264176934262e-05,
+      "loss": 1.3492,
+      "step": 101180
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.344149383944287e-05,
+      "loss": 1.2245,
+      "step": 101190
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.344034590954312e-05,
+      "loss": 1.2878,
+      "step": 101200
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.343919797964338e-05,
+      "loss": 1.265,
+      "step": 101210
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.343805004974362e-05,
+      "loss": 1.3657,
+      "step": 101220
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.343690211984388e-05,
+      "loss": 1.3057,
+      "step": 101230
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.343575418994413e-05,
+      "loss": 1.3516,
+      "step": 101240
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.343460626004438e-05,
+      "loss": 1.231,
+      "step": 101250
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.343345833014463e-05,
+      "loss": 1.2635,
+      "step": 101260
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.34323104002449e-05,
+      "loss": 1.5044,
+      "step": 101270
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.343116247034514e-05,
+      "loss": 1.4087,
+      "step": 101280
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.34300145404454e-05,
+      "loss": 1.3721,
+      "step": 101290
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.342886661054564e-05,
+      "loss": 1.234,
+      "step": 101300
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.342771868064589e-05,
+      "loss": 1.4145,
+      "step": 101310
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.342657075074616e-05,
+      "loss": 1.3059,
+      "step": 101320
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.34254228208464e-05,
+      "loss": 1.2715,
+      "step": 101330
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.342427489094666e-05,
+      "loss": 1.3695,
+      "step": 101340
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.34231269610469e-05,
+      "loss": 1.4027,
+      "step": 101350
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.342197903114717e-05,
+      "loss": 1.5043,
+      "step": 101360
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.34208311012474e-05,
+      "loss": 1.2269,
+      "step": 101370
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.341968317134767e-05,
+      "loss": 1.3207,
+      "step": 101380
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.341853524144792e-05,
+      "loss": 1.2903,
+      "step": 101390
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.341738731154817e-05,
+      "loss": 1.2564,
+      "step": 101400
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.341623938164842e-05,
+      "loss": 1.2448,
+      "step": 101410
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.341509145174867e-05,
+      "loss": 1.2799,
+      "step": 101420
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.341394352184893e-05,
+      "loss": 1.192,
+      "step": 101430
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.341279559194918e-05,
+      "loss": 1.3678,
+      "step": 101440
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.341164766204943e-05,
+      "loss": 1.2047,
+      "step": 101450
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.341049973214968e-05,
+      "loss": 1.3196,
+      "step": 101460
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.340935180224995e-05,
+      "loss": 1.3408,
+      "step": 101470
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.340820387235018e-05,
+      "loss": 1.2876,
+      "step": 101480
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.340705594245044e-05,
+      "loss": 1.2751,
+      "step": 101490
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.34059080125507e-05,
+      "loss": 1.3719,
+      "step": 101500
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.340476008265094e-05,
+      "loss": 1.2936,
+      "step": 101510
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.34036121527512e-05,
+      "loss": 1.2783,
+      "step": 101520
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.340246422285146e-05,
+      "loss": 1.2685,
+      "step": 101530
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.340131629295171e-05,
+      "loss": 1.3793,
+      "step": 101540
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.340016836305196e-05,
+      "loss": 1.2265,
+      "step": 101550
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.339902043315221e-05,
+      "loss": 1.3286,
+      "step": 101560
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.339787250325246e-05,
+      "loss": 1.3097,
+      "step": 101570
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.339672457335272e-05,
+      "loss": 1.2332,
+      "step": 101580
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.339557664345297e-05,
+      "loss": 1.3336,
+      "step": 101590
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.339442871355322e-05,
+      "loss": 1.2099,
+      "step": 101600
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.339328078365347e-05,
+      "loss": 1.2714,
+      "step": 101610
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.339213285375373e-05,
+      "loss": 1.3516,
+      "step": 101620
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.339098492385397e-05,
+      "loss": 1.3371,
+      "step": 101630
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.338983699395423e-05,
+      "loss": 1.3259,
+      "step": 101640
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.338868906405448e-05,
+      "loss": 1.3291,
+      "step": 101650
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.338754113415473e-05,
+      "loss": 1.2894,
+      "step": 101660
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.338639320425498e-05,
+      "loss": 1.3006,
+      "step": 101670
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.338524527435525e-05,
+      "loss": 1.3344,
+      "step": 101680
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.33840973444555e-05,
+      "loss": 1.4464,
+      "step": 101690
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.338294941455575e-05,
+      "loss": 1.4423,
+      "step": 101700
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.3381801484656e-05,
+      "loss": 1.2945,
+      "step": 101710
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.338065355475625e-05,
+      "loss": 1.3222,
+      "step": 101720
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.337950562485651e-05,
+      "loss": 1.1411,
+      "step": 101730
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.337835769495676e-05,
+      "loss": 1.2692,
+      "step": 101740
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.337720976505701e-05,
+      "loss": 1.1903,
+      "step": 101750
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.337606183515726e-05,
+      "loss": 1.5031,
+      "step": 101760
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.337491390525752e-05,
+      "loss": 1.3579,
+      "step": 101770
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.337376597535776e-05,
+      "loss": 1.5994,
+      "step": 101780
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.337261804545802e-05,
+      "loss": 1.5555,
+      "step": 101790
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.337147011555827e-05,
+      "loss": 1.266,
+      "step": 101800
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.337032218565852e-05,
+      "loss": 1.3841,
+      "step": 101810
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.336917425575877e-05,
+      "loss": 1.4257,
+      "step": 101820
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.336802632585903e-05,
+      "loss": 1.3851,
+      "step": 101830
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.336687839595928e-05,
+      "loss": 1.3113,
+      "step": 101840
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.336573046605953e-05,
+      "loss": 1.2437,
+      "step": 101850
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.336458253615978e-05,
+      "loss": 1.2966,
+      "step": 101860
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.336343460626003e-05,
+      "loss": 1.3438,
+      "step": 101870
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.33622866763603e-05,
+      "loss": 1.286,
+      "step": 101880
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.336113874646055e-05,
+      "loss": 1.2001,
+      "step": 101890
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.33599908165608e-05,
+      "loss": 1.2734,
+      "step": 101900
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.335884288666105e-05,
+      "loss": 1.2354,
+      "step": 101910
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.335769495676131e-05,
+      "loss": 1.4084,
+      "step": 101920
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.335654702686155e-05,
+      "loss": 1.3744,
+      "step": 101930
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.335539909696181e-05,
+      "loss": 1.2804,
+      "step": 101940
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.335425116706206e-05,
+      "loss": 1.404,
+      "step": 101950
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.335310323716231e-05,
+      "loss": 1.3409,
+      "step": 101960
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.335195530726256e-05,
+      "loss": 1.377,
+      "step": 101970
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.335080737736282e-05,
+      "loss": 1.37,
+      "step": 101980
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.334965944746307e-05,
+      "loss": 1.0892,
+      "step": 101990
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.334851151756332e-05,
+      "loss": 1.209,
+      "step": 102000
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.334736358766357e-05,
+      "loss": 1.3372,
+      "step": 102010
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.334621565776382e-05,
+      "loss": 1.353,
+      "step": 102020
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.334506772786409e-05,
+      "loss": 1.395,
+      "step": 102030
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.334391979796434e-05,
+      "loss": 1.2321,
+      "step": 102040
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.334277186806459e-05,
+      "loss": 1.2726,
+      "step": 102050
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.334162393816484e-05,
+      "loss": 1.3501,
+      "step": 102060
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.33404760082651e-05,
+      "loss": 1.3577,
+      "step": 102070
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.333932807836534e-05,
+      "loss": 1.2498,
+      "step": 102080
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.33381801484656e-05,
+      "loss": 1.2487,
+      "step": 102090
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.333703221856585e-05,
+      "loss": 1.399,
+      "step": 102100
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.33358842886661e-05,
+      "loss": 1.2809,
+      "step": 102110
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.333473635876635e-05,
+      "loss": 1.3148,
+      "step": 102120
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.333358842886661e-05,
+      "loss": 1.5041,
+      "step": 102130
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.333244049896686e-05,
+      "loss": 1.3434,
+      "step": 102140
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.333129256906711e-05,
+      "loss": 1.2107,
+      "step": 102150
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.333014463916736e-05,
+      "loss": 1.3946,
+      "step": 102160
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.332899670926761e-05,
+      "loss": 1.1489,
+      "step": 102170
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.332784877936787e-05,
+      "loss": 1.3338,
+      "step": 102180
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.332670084946812e-05,
+      "loss": 1.5032,
+      "step": 102190
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.332555291956837e-05,
+      "loss": 1.3042,
+      "step": 102200
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.332440498966862e-05,
+      "loss": 1.4674,
+      "step": 102210
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.332325705976889e-05,
+      "loss": 1.1885,
+      "step": 102220
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.332210912986912e-05,
+      "loss": 1.4354,
+      "step": 102230
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.332096119996939e-05,
+      "loss": 1.3069,
+      "step": 102240
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.331981327006964e-05,
+      "loss": 1.2177,
+      "step": 102250
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.331866534016989e-05,
+      "loss": 1.264,
+      "step": 102260
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.331751741027014e-05,
+      "loss": 1.2599,
+      "step": 102270
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.33163694803704e-05,
+      "loss": 1.2535,
+      "step": 102280
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.331522155047065e-05,
+      "loss": 1.3178,
+      "step": 102290
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.33140736205709e-05,
+      "loss": 1.2909,
+      "step": 102300
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.331292569067115e-05,
+      "loss": 1.2206,
+      "step": 102310
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.33117777607714e-05,
+      "loss": 1.391,
+      "step": 102320
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.331062983087166e-05,
+      "loss": 1.3374,
+      "step": 102330
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.330948190097191e-05,
+      "loss": 1.2455,
+      "step": 102340
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.330833397107216e-05,
+      "loss": 1.1919,
+      "step": 102350
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.330718604117241e-05,
+      "loss": 1.2622,
+      "step": 102360
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.330603811127268e-05,
+      "loss": 1.2377,
+      "step": 102370
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.330489018137291e-05,
+      "loss": 1.3119,
+      "step": 102380
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.330374225147318e-05,
+      "loss": 1.2734,
+      "step": 102390
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.330259432157343e-05,
+      "loss": 1.301,
+      "step": 102400
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.330144639167368e-05,
+      "loss": 1.2865,
+      "step": 102410
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.330029846177393e-05,
+      "loss": 1.3577,
+      "step": 102420
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.329915053187419e-05,
+      "loss": 1.4535,
+      "step": 102430
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.329800260197444e-05,
+      "loss": 1.3427,
+      "step": 102440
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.329685467207469e-05,
+      "loss": 1.4076,
+      "step": 102450
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.329570674217494e-05,
+      "loss": 1.2275,
+      "step": 102460
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.329455881227519e-05,
+      "loss": 1.43,
+      "step": 102470
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.329341088237545e-05,
+      "loss": 1.3618,
+      "step": 102480
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.32922629524757e-05,
+      "loss": 1.4502,
+      "step": 102490
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.329111502257595e-05,
+      "loss": 1.3867,
+      "step": 102500
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.32899670926762e-05,
+      "loss": 1.2703,
+      "step": 102510
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.328881916277646e-05,
+      "loss": 1.4441,
+      "step": 102520
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.32876712328767e-05,
+      "loss": 1.3558,
+      "step": 102530
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.328652330297696e-05,
+      "loss": 1.3042,
+      "step": 102540
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.328537537307721e-05,
+      "loss": 1.2352,
+      "step": 102550
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.328422744317746e-05,
+      "loss": 1.3543,
+      "step": 102560
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.328307951327771e-05,
+      "loss": 1.239,
+      "step": 102570
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.328193158337798e-05,
+      "loss": 1.3729,
+      "step": 102580
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.328078365347823e-05,
+      "loss": 1.4432,
+      "step": 102590
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.327963572357848e-05,
+      "loss": 1.4485,
+      "step": 102600
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.327848779367873e-05,
+      "loss": 1.1902,
+      "step": 102610
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.327733986377898e-05,
+      "loss": 1.5292,
+      "step": 102620
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.327619193387924e-05,
+      "loss": 1.3053,
+      "step": 102630
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.327504400397949e-05,
+      "loss": 1.2606,
+      "step": 102640
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.327389607407974e-05,
+      "loss": 1.3063,
+      "step": 102650
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.327274814417999e-05,
+      "loss": 1.3509,
+      "step": 102660
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.327160021428025e-05,
+      "loss": 1.2142,
+      "step": 102670
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.327045228438049e-05,
+      "loss": 1.2515,
+      "step": 102680
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.326930435448075e-05,
+      "loss": 1.1595,
+      "step": 102690
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.3268156424581e-05,
+      "loss": 1.4176,
+      "step": 102700
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.326700849468125e-05,
+      "loss": 1.2463,
+      "step": 102710
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.32658605647815e-05,
+      "loss": 1.264,
+      "step": 102720
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.326471263488177e-05,
+      "loss": 1.2372,
+      "step": 102730
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.326356470498202e-05,
+      "loss": 1.2573,
+      "step": 102740
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.326241677508227e-05,
+      "loss": 1.2404,
+      "step": 102750
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 6.326126884518252e-05,
+      "loss": 1.2038,
+      "step": 102760
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.326012091528276e-05,
+      "loss": 1.3029,
+      "step": 102770
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.325897298538303e-05,
+      "loss": 1.3457,
+      "step": 102780
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.325782505548326e-05,
+      "loss": 1.2875,
+      "step": 102790
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.325667712558353e-05,
+      "loss": 1.2928,
+      "step": 102800
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.325552919568378e-05,
+      "loss": 1.5916,
+      "step": 102810
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.325438126578403e-05,
+      "loss": 1.2272,
+      "step": 102820
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.325323333588428e-05,
+      "loss": 1.1753,
+      "step": 102830
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.325208540598454e-05,
+      "loss": 1.3391,
+      "step": 102840
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.325093747608479e-05,
+      "loss": 1.3958,
+      "step": 102850
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.324978954618504e-05,
+      "loss": 1.3057,
+      "step": 102860
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.324864161628529e-05,
+      "loss": 1.4584,
+      "step": 102870
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.324749368638554e-05,
+      "loss": 1.4536,
+      "step": 102880
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.32463457564858e-05,
+      "loss": 1.2825,
+      "step": 102890
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.324519782658605e-05,
+      "loss": 1.2303,
+      "step": 102900
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.32440498966863e-05,
+      "loss": 1.3478,
+      "step": 102910
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.324290196678655e-05,
+      "loss": 1.3474,
+      "step": 102920
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.324175403688682e-05,
+      "loss": 1.5546,
+      "step": 102930
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.324060610698705e-05,
+      "loss": 1.2302,
+      "step": 102940
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.323945817708732e-05,
+      "loss": 1.2865,
+      "step": 102950
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.323831024718757e-05,
+      "loss": 1.3576,
+      "step": 102960
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.323716231728782e-05,
+      "loss": 1.3512,
+      "step": 102970
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 6.323601438738807e-05,
+      "loss": 1.3352,
+      "step": 102980
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.323486645748833e-05,
+      "loss": 1.3077,
+      "step": 102990
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.323371852758858e-05,
+      "loss": 1.108,
+      "step": 103000
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.323257059768883e-05,
+      "loss": 1.2989,
+      "step": 103010
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.323142266778908e-05,
+      "loss": 1.3766,
+      "step": 103020
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.323027473788933e-05,
+      "loss": 1.3054,
+      "step": 103030
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.322912680798959e-05,
+      "loss": 1.3347,
+      "step": 103040
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.322797887808984e-05,
+      "loss": 1.2164,
+      "step": 103050
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.322683094819009e-05,
+      "loss": 1.4026,
+      "step": 103060
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.322568301829034e-05,
+      "loss": 1.2903,
+      "step": 103070
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.32245350883906e-05,
+      "loss": 1.3844,
+      "step": 103080
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.322338715849084e-05,
+      "loss": 1.3025,
+      "step": 103090
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.32222392285911e-05,
+      "loss": 1.2289,
+      "step": 103100
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.322109129869135e-05,
+      "loss": 1.2586,
+      "step": 103110
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.32199433687916e-05,
+      "loss": 1.3253,
+      "step": 103120
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.321879543889185e-05,
+      "loss": 1.2889,
+      "step": 103130
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.321764750899212e-05,
+      "loss": 1.3127,
+      "step": 103140
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.321649957909237e-05,
+      "loss": 1.2468,
+      "step": 103150
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.321535164919262e-05,
+      "loss": 1.1902,
+      "step": 103160
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.321420371929287e-05,
+      "loss": 1.3613,
+      "step": 103170
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.321305578939312e-05,
+      "loss": 1.2222,
+      "step": 103180
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.321190785949338e-05,
+      "loss": 1.3531,
+      "step": 103190
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 6.321075992959363e-05,
+      "loss": 1.2828,
+      "step": 103200
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.320961199969388e-05,
+      "loss": 1.3238,
+      "step": 103210
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.320846406979413e-05,
+      "loss": 1.2939,
+      "step": 103220
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.32073161398944e-05,
+      "loss": 1.4538,
+      "step": 103230
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.320616820999463e-05,
+      "loss": 1.5036,
+      "step": 103240
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.32050202800949e-05,
+      "loss": 1.2606,
+      "step": 103250
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.320387235019514e-05,
+      "loss": 1.2623,
+      "step": 103260
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.320272442029539e-05,
+      "loss": 1.4198,
+      "step": 103270
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.320157649039564e-05,
+      "loss": 1.2018,
+      "step": 103280
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.32004285604959e-05,
+      "loss": 1.2782,
+      "step": 103290
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.319928063059616e-05,
+      "loss": 1.2626,
+      "step": 103300
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.31981327006964e-05,
+      "loss": 1.2111,
+      "step": 103310
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.319698477079666e-05,
+      "loss": 1.275,
+      "step": 103320
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.31958368408969e-05,
+      "loss": 1.3351,
+      "step": 103330
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.319468891099717e-05,
+      "loss": 1.3645,
+      "step": 103340
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.319354098109742e-05,
+      "loss": 1.323,
+      "step": 103350
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.319239305119767e-05,
+      "loss": 1.3321,
+      "step": 103360
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.319124512129792e-05,
+      "loss": 1.343,
+      "step": 103370
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.319009719139818e-05,
+      "loss": 1.479,
+      "step": 103380
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.318894926149842e-05,
+      "loss": 1.1609,
+      "step": 103390
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.318780133159868e-05,
+      "loss": 1.3993,
+      "step": 103400
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.318665340169893e-05,
+      "loss": 1.2623,
+      "step": 103410
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.318550547179918e-05,
+      "loss": 1.3493,
+      "step": 103420
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.318435754189943e-05,
+      "loss": 1.243,
+      "step": 103430
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.31832096119997e-05,
+      "loss": 1.2894,
+      "step": 103440
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.318206168209994e-05,
+      "loss": 1.3415,
+      "step": 103450
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.31809137522002e-05,
+      "loss": 1.2385,
+      "step": 103460
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.317976582230044e-05,
+      "loss": 1.4547,
+      "step": 103470
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.31786178924007e-05,
+      "loss": 1.2454,
+      "step": 103480
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.317746996250096e-05,
+      "loss": 1.2689,
+      "step": 103490
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.317632203260121e-05,
+      "loss": 1.2139,
+      "step": 103500
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.317517410270146e-05,
+      "loss": 1.3716,
+      "step": 103510
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.317402617280171e-05,
+      "loss": 1.287,
+      "step": 103520
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.317287824290197e-05,
+      "loss": 1.4698,
+      "step": 103530
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.317173031300221e-05,
+      "loss": 1.2305,
+      "step": 103540
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.317058238310247e-05,
+      "loss": 1.2859,
+      "step": 103550
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.316943445320272e-05,
+      "loss": 1.2596,
+      "step": 103560
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.316828652330297e-05,
+      "loss": 1.3045,
+      "step": 103570
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.316713859340322e-05,
+      "loss": 1.2363,
+      "step": 103580
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.316599066350348e-05,
+      "loss": 1.2673,
+      "step": 103590
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.316484273360373e-05,
+      "loss": 1.3019,
+      "step": 103600
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.316369480370398e-05,
+      "loss": 1.5029,
+      "step": 103610
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.316254687380423e-05,
+      "loss": 1.2524,
+      "step": 103620
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.316139894390448e-05,
+      "loss": 1.2428,
+      "step": 103630
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.316025101400475e-05,
+      "loss": 1.1994,
+      "step": 103640
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.3159103084105e-05,
+      "loss": 1.3534,
+      "step": 103650
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.315795515420525e-05,
+      "loss": 1.4654,
+      "step": 103660
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.31568072243055e-05,
+      "loss": 1.2628,
+      "step": 103670
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.315565929440576e-05,
+      "loss": 1.2256,
+      "step": 103680
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.3154511364506e-05,
+      "loss": 1.3526,
+      "step": 103690
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.315336343460626e-05,
+      "loss": 1.2441,
+      "step": 103700
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.315221550470651e-05,
+      "loss": 1.4132,
+      "step": 103710
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.315106757480676e-05,
+      "loss": 1.32,
+      "step": 103720
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.314991964490701e-05,
+      "loss": 1.2547,
+      "step": 103730
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.314877171500727e-05,
+      "loss": 1.3157,
+      "step": 103740
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.314762378510752e-05,
+      "loss": 1.2048,
+      "step": 103750
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.314647585520777e-05,
+      "loss": 1.2201,
+      "step": 103760
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.314532792530802e-05,
+      "loss": 1.2959,
+      "step": 103770
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.314417999540827e-05,
+      "loss": 1.293,
+      "step": 103780
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.314303206550853e-05,
+      "loss": 1.3069,
+      "step": 103790
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.314188413560878e-05,
+      "loss": 1.283,
+      "step": 103800
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.314073620570903e-05,
+      "loss": 1.3178,
+      "step": 103810
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.313958827580928e-05,
+      "loss": 1.3544,
+      "step": 103820
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.313844034590955e-05,
+      "loss": 1.3138,
+      "step": 103830
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.313729241600978e-05,
+      "loss": 1.2984,
+      "step": 103840
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.313614448611005e-05,
+      "loss": 1.2675,
+      "step": 103850
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.31349965562103e-05,
+      "loss": 1.2197,
+      "step": 103860
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.313384862631055e-05,
+      "loss": 1.5626,
+      "step": 103870
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.31327006964108e-05,
+      "loss": 1.3102,
+      "step": 103880
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.313155276651106e-05,
+      "loss": 1.3912,
+      "step": 103890
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.313040483661131e-05,
+      "loss": 1.1865,
+      "step": 103900
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.312925690671156e-05,
+      "loss": 1.2635,
+      "step": 103910
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.312810897681181e-05,
+      "loss": 1.432,
+      "step": 103920
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.312696104691206e-05,
+      "loss": 1.2473,
+      "step": 103930
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.312581311701232e-05,
+      "loss": 1.4166,
+      "step": 103940
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.312466518711257e-05,
+      "loss": 1.2776,
+      "step": 103950
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.312351725721282e-05,
+      "loss": 1.2682,
+      "step": 103960
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.312236932731307e-05,
+      "loss": 1.4952,
+      "step": 103970
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.312122139741334e-05,
+      "loss": 1.3857,
+      "step": 103980
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.312007346751357e-05,
+      "loss": 1.357,
+      "step": 103990
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.311892553761384e-05,
+      "loss": 1.3878,
+      "step": 104000
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.311777760771409e-05,
+      "loss": 1.3977,
+      "step": 104010
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.311662967781434e-05,
+      "loss": 1.2581,
+      "step": 104020
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.311548174791459e-05,
+      "loss": 1.3606,
+      "step": 104030
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.311433381801485e-05,
+      "loss": 1.2116,
+      "step": 104040
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.31131858881151e-05,
+      "loss": 1.3073,
+      "step": 104050
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.311203795821535e-05,
+      "loss": 1.3266,
+      "step": 104060
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 6.31108900283156e-05,
+      "loss": 1.2643,
+      "step": 104070
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.310974209841585e-05,
+      "loss": 1.4333,
+      "step": 104080
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.310859416851611e-05,
+      "loss": 1.4412,
+      "step": 104090
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.310744623861636e-05,
+      "loss": 1.5666,
+      "step": 104100
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.310629830871661e-05,
+      "loss": 1.401,
+      "step": 104110
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.310515037881686e-05,
+      "loss": 1.2547,
+      "step": 104120
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.310400244891712e-05,
+      "loss": 1.4311,
+      "step": 104130
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.310285451901736e-05,
+      "loss": 1.445,
+      "step": 104140
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.310170658911762e-05,
+      "loss": 1.2876,
+      "step": 104150
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.310055865921787e-05,
+      "loss": 1.3695,
+      "step": 104160
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.309941072931812e-05,
+      "loss": 1.1995,
+      "step": 104170
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.309826279941837e-05,
+      "loss": 1.2072,
+      "step": 104180
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.309711486951862e-05,
+      "loss": 1.3791,
+      "step": 104190
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.309596693961889e-05,
+      "loss": 1.2084,
+      "step": 104200
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.309481900971914e-05,
+      "loss": 1.408,
+      "step": 104210
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.309367107981939e-05,
+      "loss": 1.2339,
+      "step": 104220
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.309252314991964e-05,
+      "loss": 1.2844,
+      "step": 104230
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.30913752200199e-05,
+      "loss": 1.3683,
+      "step": 104240
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.309022729012014e-05,
+      "loss": 1.2534,
+      "step": 104250
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.30890793602204e-05,
+      "loss": 1.3994,
+      "step": 104260
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.308793143032065e-05,
+      "loss": 1.2429,
+      "step": 104270
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 6.30867835004209e-05,
+      "loss": 1.5305,
+      "step": 104280
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.308563557052115e-05,
+      "loss": 1.2816,
+      "step": 104290
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.308448764062141e-05,
+      "loss": 1.3281,
+      "step": 104300
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.308333971072166e-05,
+      "loss": 1.2888,
+      "step": 104310
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.308219178082191e-05,
+      "loss": 1.2595,
+      "step": 104320
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.308104385092216e-05,
+      "loss": 1.2853,
+      "step": 104330
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.307989592102241e-05,
+      "loss": 1.2593,
+      "step": 104340
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.307874799112268e-05,
+      "loss": 1.2997,
+      "step": 104350
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.307760006122293e-05,
+      "loss": 1.3445,
+      "step": 104360
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.307645213132318e-05,
+      "loss": 1.2851,
+      "step": 104370
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.307530420142343e-05,
+      "loss": 1.3131,
+      "step": 104380
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.307415627152369e-05,
+      "loss": 1.2788,
+      "step": 104390
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.307300834162392e-05,
+      "loss": 1.4489,
+      "step": 104400
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.307186041172419e-05,
+      "loss": 1.1796,
+      "step": 104410
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.307071248182444e-05,
+      "loss": 1.2527,
+      "step": 104420
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.306956455192469e-05,
+      "loss": 1.388,
+      "step": 104430
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.306841662202494e-05,
+      "loss": 1.3189,
+      "step": 104440
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.30672686921252e-05,
+      "loss": 1.4013,
+      "step": 104450
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.306612076222545e-05,
+      "loss": 1.373,
+      "step": 104460
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.30649728323257e-05,
+      "loss": 1.2172,
+      "step": 104470
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.306382490242595e-05,
+      "loss": 1.4063,
+      "step": 104480
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.30626769725262e-05,
+      "loss": 1.3949,
+      "step": 104490
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 6.306152904262646e-05,
+      "loss": 1.2298,
+      "step": 104500
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.306038111272671e-05,
+      "loss": 1.3715,
+      "step": 104510
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.305923318282696e-05,
+      "loss": 1.4342,
+      "step": 104520
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.305808525292721e-05,
+      "loss": 1.3457,
+      "step": 104530
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.305693732302748e-05,
+      "loss": 1.2938,
+      "step": 104540
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.305578939312771e-05,
+      "loss": 1.3129,
+      "step": 104550
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.305464146322798e-05,
+      "loss": 1.2225,
+      "step": 104560
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.305349353332823e-05,
+      "loss": 1.297,
+      "step": 104570
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.305234560342848e-05,
+      "loss": 1.2353,
+      "step": 104580
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.305119767352873e-05,
+      "loss": 1.2102,
+      "step": 104590
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.305004974362899e-05,
+      "loss": 1.223,
+      "step": 104600
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.304890181372924e-05,
+      "loss": 1.3267,
+      "step": 104610
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.304775388382949e-05,
+      "loss": 1.3838,
+      "step": 104620
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.304660595392974e-05,
+      "loss": 1.087,
+      "step": 104630
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.304545802402999e-05,
+      "loss": 1.3165,
+      "step": 104640
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.304431009413025e-05,
+      "loss": 1.4231,
+      "step": 104650
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.30431621642305e-05,
+      "loss": 1.4528,
+      "step": 104660
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.304201423433075e-05,
+      "loss": 1.4126,
+      "step": 104670
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.3040866304431e-05,
+      "loss": 1.4649,
+      "step": 104680
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.303971837453127e-05,
+      "loss": 1.3366,
+      "step": 104690
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.30385704446315e-05,
+      "loss": 1.2663,
+      "step": 104700
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.303742251473177e-05,
+      "loss": 1.529,
+      "step": 104710
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 6.303627458483202e-05,
+      "loss": 1.3746,
+      "step": 104720
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.303512665493226e-05,
+      "loss": 1.4843,
+      "step": 104730
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.303397872503251e-05,
+      "loss": 1.336,
+      "step": 104740
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.303283079513278e-05,
+      "loss": 1.296,
+      "step": 104750
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.303168286523303e-05,
+      "loss": 1.5152,
+      "step": 104760
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.303053493533328e-05,
+      "loss": 1.4285,
+      "step": 104770
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.302938700543353e-05,
+      "loss": 1.2973,
+      "step": 104780
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.302823907553378e-05,
+      "loss": 1.4402,
+      "step": 104790
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.302709114563404e-05,
+      "loss": 1.2081,
+      "step": 104800
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.302594321573429e-05,
+      "loss": 1.2664,
+      "step": 104810
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.302479528583454e-05,
+      "loss": 1.3396,
+      "step": 104820
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.302364735593479e-05,
+      "loss": 1.3642,
+      "step": 104830
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.302249942603505e-05,
+      "loss": 1.4042,
+      "step": 104840
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.302135149613529e-05,
+      "loss": 1.3167,
+      "step": 104850
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.302020356623555e-05,
+      "loss": 1.3041,
+      "step": 104860
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.30190556363358e-05,
+      "loss": 1.4282,
+      "step": 104870
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.301790770643605e-05,
+      "loss": 1.2733,
+      "step": 104880
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.30167597765363e-05,
+      "loss": 1.3052,
+      "step": 104890
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.301561184663657e-05,
+      "loss": 1.2261,
+      "step": 104900
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.301446391673682e-05,
+      "loss": 1.3972,
+      "step": 104910
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.301331598683707e-05,
+      "loss": 1.3448,
+      "step": 104920
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.301216805693732e-05,
+      "loss": 1.297,
+      "step": 104930
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 6.301102012703757e-05,
+      "loss": 1.2919,
+      "step": 104940
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.300987219713783e-05,
+      "loss": 1.2959,
+      "step": 104950
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.300872426723808e-05,
+      "loss": 1.407,
+      "step": 104960
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.300757633733833e-05,
+      "loss": 1.2958,
+      "step": 104970
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.300642840743858e-05,
+      "loss": 1.3811,
+      "step": 104980
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.300528047753884e-05,
+      "loss": 1.4045,
+      "step": 104990
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.300413254763908e-05,
+      "loss": 1.3817,
+      "step": 105000
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.300298461773934e-05,
+      "loss": 1.2869,
+      "step": 105010
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.300183668783959e-05,
+      "loss": 1.3048,
+      "step": 105020
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.300068875793984e-05,
+      "loss": 1.1332,
+      "step": 105030
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.299954082804009e-05,
+      "loss": 1.1618,
+      "step": 105040
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.299839289814036e-05,
+      "loss": 1.3076,
+      "step": 105050
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.29972449682406e-05,
+      "loss": 1.408,
+      "step": 105060
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.299609703834085e-05,
+      "loss": 1.2279,
+      "step": 105070
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.29949491084411e-05,
+      "loss": 1.1766,
+      "step": 105080
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.299380117854135e-05,
+      "loss": 1.3709,
+      "step": 105090
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.299265324864162e-05,
+      "loss": 1.2007,
+      "step": 105100
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.299150531874187e-05,
+      "loss": 1.3111,
+      "step": 105110
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.299035738884212e-05,
+      "loss": 1.3016,
+      "step": 105120
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.298920945894237e-05,
+      "loss": 1.3155,
+      "step": 105130
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.298806152904263e-05,
+      "loss": 1.2885,
+      "step": 105140
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.298691359914287e-05,
+      "loss": 1.2694,
+      "step": 105150
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.298576566924313e-05,
+      "loss": 1.3565,
+      "step": 105160
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.298461773934338e-05,
+      "loss": 1.3434,
+      "step": 105170
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.298346980944363e-05,
+      "loss": 1.3957,
+      "step": 105180
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.298232187954388e-05,
+      "loss": 1.1998,
+      "step": 105190
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.298117394964414e-05,
+      "loss": 1.3664,
+      "step": 105200
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.29800260197444e-05,
+      "loss": 1.2548,
+      "step": 105210
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.297887808984464e-05,
+      "loss": 1.287,
+      "step": 105220
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.297773015994489e-05,
+      "loss": 1.2482,
+      "step": 105230
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.297658223004514e-05,
+      "loss": 1.2871,
+      "step": 105240
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.29754343001454e-05,
+      "loss": 1.4108,
+      "step": 105250
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.297428637024566e-05,
+      "loss": 1.2936,
+      "step": 105260
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.29731384403459e-05,
+      "loss": 1.3006,
+      "step": 105270
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.297199051044616e-05,
+      "loss": 1.3576,
+      "step": 105280
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.297084258054642e-05,
+      "loss": 1.2149,
+      "step": 105290
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.296969465064666e-05,
+      "loss": 1.3569,
+      "step": 105300
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.296854672074692e-05,
+      "loss": 1.2206,
+      "step": 105310
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.296739879084717e-05,
+      "loss": 1.3637,
+      "step": 105320
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.296625086094742e-05,
+      "loss": 1.2112,
+      "step": 105330
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.296510293104767e-05,
+      "loss": 1.4269,
+      "step": 105340
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.296395500114793e-05,
+      "loss": 1.2679,
+      "step": 105350
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.296280707124818e-05,
+      "loss": 1.4089,
+      "step": 105360
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.296165914134843e-05,
+      "loss": 1.3412,
+      "step": 105370
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.296051121144868e-05,
+      "loss": 1.1692,
+      "step": 105380
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.295936328154893e-05,
+      "loss": 1.306,
+      "step": 105390
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.29582153516492e-05,
+      "loss": 1.2652,
+      "step": 105400
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.295706742174944e-05,
+      "loss": 1.3353,
+      "step": 105410
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.29559194918497e-05,
+      "loss": 1.3185,
+      "step": 105420
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.295477156194994e-05,
+      "loss": 1.2511,
+      "step": 105430
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.295362363205021e-05,
+      "loss": 1.4105,
+      "step": 105440
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.295247570215044e-05,
+      "loss": 1.4458,
+      "step": 105450
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.295132777225071e-05,
+      "loss": 1.2627,
+      "step": 105460
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.295017984235096e-05,
+      "loss": 1.4437,
+      "step": 105470
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.294903191245121e-05,
+      "loss": 1.3724,
+      "step": 105480
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.294788398255146e-05,
+      "loss": 1.275,
+      "step": 105490
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.294673605265172e-05,
+      "loss": 1.1633,
+      "step": 105500
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.294558812275197e-05,
+      "loss": 1.2735,
+      "step": 105510
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.294444019285222e-05,
+      "loss": 1.1838,
+      "step": 105520
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.294329226295247e-05,
+      "loss": 1.4898,
+      "step": 105530
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.294214433305272e-05,
+      "loss": 1.3828,
+      "step": 105540
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.294099640315298e-05,
+      "loss": 1.2872,
+      "step": 105550
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.293984847325322e-05,
+      "loss": 1.2428,
+      "step": 105560
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.293870054335348e-05,
+      "loss": 1.1991,
+      "step": 105570
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.293755261345373e-05,
+      "loss": 1.3898,
+      "step": 105580
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.293640468355398e-05,
+      "loss": 1.3959,
+      "step": 105590
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.293525675365423e-05,
+      "loss": 1.2558,
+      "step": 105600
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.29341088237545e-05,
+      "loss": 1.3013,
+      "step": 105610
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.293296089385475e-05,
+      "loss": 1.4075,
+      "step": 105620
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.2931812963955e-05,
+      "loss": 1.3716,
+      "step": 105630
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.293066503405525e-05,
+      "loss": 1.3718,
+      "step": 105640
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.29295171041555e-05,
+      "loss": 1.3266,
+      "step": 105650
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.292836917425576e-05,
+      "loss": 1.2753,
+      "step": 105660
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.292722124435601e-05,
+      "loss": 1.3199,
+      "step": 105670
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.292607331445626e-05,
+      "loss": 1.3379,
+      "step": 105680
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.292492538455651e-05,
+      "loss": 1.3051,
+      "step": 105690
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.292377745465677e-05,
+      "loss": 1.372,
+      "step": 105700
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.292262952475701e-05,
+      "loss": 1.4006,
+      "step": 105710
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.292148159485727e-05,
+      "loss": 1.2312,
+      "step": 105720
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.292033366495752e-05,
+      "loss": 1.3092,
+      "step": 105730
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.291918573505777e-05,
+      "loss": 1.3661,
+      "step": 105740
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.291803780515802e-05,
+      "loss": 1.2398,
+      "step": 105750
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.291688987525828e-05,
+      "loss": 1.433,
+      "step": 105760
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.291574194535853e-05,
+      "loss": 1.2474,
+      "step": 105770
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.291459401545878e-05,
+      "loss": 1.3699,
+      "step": 105780
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.291344608555903e-05,
+      "loss": 1.3619,
+      "step": 105790
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.291229815565928e-05,
+      "loss": 1.3903,
+      "step": 105800
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.291115022575955e-05,
+      "loss": 1.2747,
+      "step": 105810
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.29100022958598e-05,
+      "loss": 1.1906,
+      "step": 105820
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.290885436596005e-05,
+      "loss": 1.3881,
+      "step": 105830
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.29077064360603e-05,
+      "loss": 1.3584,
+      "step": 105840
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.290655850616056e-05,
+      "loss": 1.3393,
+      "step": 105850
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.29054105762608e-05,
+      "loss": 1.605,
+      "step": 105860
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.290426264636106e-05,
+      "loss": 1.3515,
+      "step": 105870
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.290311471646131e-05,
+      "loss": 1.1725,
+      "step": 105880
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.290196678656156e-05,
+      "loss": 1.2974,
+      "step": 105890
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.290081885666181e-05,
+      "loss": 1.3805,
+      "step": 105900
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.289967092676207e-05,
+      "loss": 1.3357,
+      "step": 105910
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.289852299686232e-05,
+      "loss": 1.3796,
+      "step": 105920
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.289737506696257e-05,
+      "loss": 1.3311,
+      "step": 105930
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.289622713706282e-05,
+      "loss": 1.4098,
+      "step": 105940
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.289507920716307e-05,
+      "loss": 1.2527,
+      "step": 105950
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.289393127726334e-05,
+      "loss": 1.1708,
+      "step": 105960
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.289278334736359e-05,
+      "loss": 1.3561,
+      "step": 105970
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.289163541746384e-05,
+      "loss": 1.2162,
+      "step": 105980
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.289048748756409e-05,
+      "loss": 1.2169,
+      "step": 105990
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.288933955766435e-05,
+      "loss": 1.344,
+      "step": 106000
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.288819162776459e-05,
+      "loss": 1.268,
+      "step": 106010
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.288704369786485e-05,
+      "loss": 1.4362,
+      "step": 106020
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.28858957679651e-05,
+      "loss": 1.3164,
+      "step": 106030
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.288474783806535e-05,
+      "loss": 1.2664,
+      "step": 106040
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.28835999081656e-05,
+      "loss": 1.2362,
+      "step": 106050
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.288245197826586e-05,
+      "loss": 1.459,
+      "step": 106060
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.288130404836611e-05,
+      "loss": 1.3617,
+      "step": 106070
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.288015611846636e-05,
+      "loss": 1.3548,
+      "step": 106080
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.287900818856661e-05,
+      "loss": 1.4469,
+      "step": 106090
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.287786025866686e-05,
+      "loss": 1.4318,
+      "step": 106100
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.287671232876712e-05,
+      "loss": 1.3245,
+      "step": 106110
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.287556439886737e-05,
+      "loss": 1.2182,
+      "step": 106120
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.287441646896762e-05,
+      "loss": 1.099,
+      "step": 106130
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.287326853906787e-05,
+      "loss": 1.2876,
+      "step": 106140
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.287212060916814e-05,
+      "loss": 1.3623,
+      "step": 106150
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.287097267926837e-05,
+      "loss": 1.2771,
+      "step": 106160
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.286982474936864e-05,
+      "loss": 1.3138,
+      "step": 106170
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.286867681946889e-05,
+      "loss": 1.261,
+      "step": 106180
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.286752888956914e-05,
+      "loss": 1.3352,
+      "step": 106190
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.286638095966939e-05,
+      "loss": 1.4083,
+      "step": 106200
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.286523302976965e-05,
+      "loss": 1.4165,
+      "step": 106210
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.28640850998699e-05,
+      "loss": 1.2224,
+      "step": 106220
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.286293716997015e-05,
+      "loss": 1.2067,
+      "step": 106230
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.28617892400704e-05,
+      "loss": 1.2802,
+      "step": 106240
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 6.286064131017065e-05,
+      "loss": 1.3178,
+      "step": 106250
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.285949338027091e-05,
+      "loss": 1.2788,
+      "step": 106260
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.285834545037116e-05,
+      "loss": 1.2424,
+      "step": 106270
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.285719752047141e-05,
+      "loss": 1.4491,
+      "step": 106280
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.285604959057166e-05,
+      "loss": 1.2854,
+      "step": 106290
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.285490166067193e-05,
+      "loss": 1.2703,
+      "step": 106300
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.285375373077216e-05,
+      "loss": 1.276,
+      "step": 106310
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.285260580087243e-05,
+      "loss": 1.605,
+      "step": 106320
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.285145787097268e-05,
+      "loss": 1.2356,
+      "step": 106330
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.285030994107293e-05,
+      "loss": 1.2768,
+      "step": 106340
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.284916201117317e-05,
+      "loss": 1.2948,
+      "step": 106350
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.284801408127344e-05,
+      "loss": 1.2,
+      "step": 106360
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.284686615137369e-05,
+      "loss": 1.4107,
+      "step": 106370
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.284571822147394e-05,
+      "loss": 1.292,
+      "step": 106380
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.284457029157419e-05,
+      "loss": 1.3133,
+      "step": 106390
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.284342236167444e-05,
+      "loss": 1.2767,
+      "step": 106400
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.28422744317747e-05,
+      "loss": 1.3285,
+      "step": 106410
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.284112650187495e-05,
+      "loss": 1.2654,
+      "step": 106420
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.28399785719752e-05,
+      "loss": 1.2889,
+      "step": 106430
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.283883064207545e-05,
+      "loss": 1.251,
+      "step": 106440
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.283768271217571e-05,
+      "loss": 1.3114,
+      "step": 106450
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 6.283653478227595e-05,
+      "loss": 1.4269,
+      "step": 106460
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.283538685237621e-05,
+      "loss": 1.4746,
+      "step": 106470
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.283423892247646e-05,
+      "loss": 1.3167,
+      "step": 106480
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.283309099257671e-05,
+      "loss": 1.3633,
+      "step": 106490
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.283194306267696e-05,
+      "loss": 1.3965,
+      "step": 106500
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.283079513277723e-05,
+      "loss": 1.2919,
+      "step": 106510
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.282964720287748e-05,
+      "loss": 1.3481,
+      "step": 106520
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.282849927297773e-05,
+      "loss": 1.486,
+      "step": 106530
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.282735134307798e-05,
+      "loss": 1.2427,
+      "step": 106540
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.282620341317823e-05,
+      "loss": 1.5005,
+      "step": 106550
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.282505548327849e-05,
+      "loss": 1.3719,
+      "step": 106560
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.282390755337874e-05,
+      "loss": 1.3386,
+      "step": 106570
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.282275962347899e-05,
+      "loss": 1.1905,
+      "step": 106580
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.282161169357924e-05,
+      "loss": 1.4288,
+      "step": 106590
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.28204637636795e-05,
+      "loss": 1.4878,
+      "step": 106600
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.281931583377974e-05,
+      "loss": 1.3871,
+      "step": 106610
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.281816790388e-05,
+      "loss": 1.3982,
+      "step": 106620
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.281701997398025e-05,
+      "loss": 1.2762,
+      "step": 106630
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.28158720440805e-05,
+      "loss": 1.336,
+      "step": 106640
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.281472411418075e-05,
+      "loss": 1.2279,
+      "step": 106650
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.281357618428102e-05,
+      "loss": 1.3725,
+      "step": 106660
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.281242825438127e-05,
+      "loss": 1.2461,
+      "step": 106670
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.281128032448152e-05,
+      "loss": 1.2649,
+      "step": 106680
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.281013239458176e-05,
+      "loss": 1.3247,
+      "step": 106690
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.280898446468201e-05,
+      "loss": 1.4011,
+      "step": 106700
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.280783653478228e-05,
+      "loss": 1.2332,
+      "step": 106710
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.280668860488253e-05,
+      "loss": 1.4741,
+      "step": 106720
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.280554067498278e-05,
+      "loss": 1.4298,
+      "step": 106730
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.280439274508303e-05,
+      "loss": 1.28,
+      "step": 106740
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.280324481518329e-05,
+      "loss": 1.1823,
+      "step": 106750
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.280209688528353e-05,
+      "loss": 1.1885,
+      "step": 106760
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.280094895538379e-05,
+      "loss": 1.3674,
+      "step": 106770
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.279980102548404e-05,
+      "loss": 1.3197,
+      "step": 106780
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.279865309558429e-05,
+      "loss": 1.3674,
+      "step": 106790
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.279750516568454e-05,
+      "loss": 1.3986,
+      "step": 106800
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.27963572357848e-05,
+      "loss": 1.3851,
+      "step": 106810
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.279520930588505e-05,
+      "loss": 1.2551,
+      "step": 106820
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.27940613759853e-05,
+      "loss": 1.2732,
+      "step": 106830
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.279291344608555e-05,
+      "loss": 1.2664,
+      "step": 106840
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.27917655161858e-05,
+      "loss": 1.1448,
+      "step": 106850
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.279061758628607e-05,
+      "loss": 1.3196,
+      "step": 106860
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.278946965638632e-05,
+      "loss": 1.2166,
+      "step": 106870
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.278832172648657e-05,
+      "loss": 1.2781,
+      "step": 106880
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.278717379658682e-05,
+      "loss": 1.3466,
+      "step": 106890
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 6.278602586668707e-05,
+      "loss": 1.4071,
+      "step": 106900
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.278487793678732e-05,
+      "loss": 1.2981,
+      "step": 106910
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.278373000688758e-05,
+      "loss": 1.2756,
+      "step": 106920
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.278258207698782e-05,
+      "loss": 1.3122,
+      "step": 106930
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.278143414708808e-05,
+      "loss": 1.2267,
+      "step": 106940
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.278028621718833e-05,
+      "loss": 1.3729,
+      "step": 106950
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.277913828728858e-05,
+      "loss": 1.2059,
+      "step": 106960
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.277799035738884e-05,
+      "loss": 1.2871,
+      "step": 106970
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.277684242748909e-05,
+      "loss": 1.3206,
+      "step": 106980
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.277569449758934e-05,
+      "loss": 1.3146,
+      "step": 106990
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.277454656768959e-05,
+      "loss": 1.2919,
+      "step": 107000
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.277339863778986e-05,
+      "loss": 1.0999,
+      "step": 107010
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.277225070789009e-05,
+      "loss": 1.3269,
+      "step": 107020
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.277110277799035e-05,
+      "loss": 1.1897,
+      "step": 107030
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.27699548480906e-05,
+      "loss": 1.1465,
+      "step": 107040
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.276880691819085e-05,
+      "loss": 1.3198,
+      "step": 107050
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.27676589882911e-05,
+      "loss": 1.3842,
+      "step": 107060
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.276651105839137e-05,
+      "loss": 1.3458,
+      "step": 107070
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.27653631284916e-05,
+      "loss": 1.3189,
+      "step": 107080
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.276421519859187e-05,
+      "loss": 1.4678,
+      "step": 107090
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.276306726869212e-05,
+      "loss": 1.4477,
+      "step": 107100
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.276191933879237e-05,
+      "loss": 1.2207,
+      "step": 107110
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.276077140889263e-05,
+      "loss": 1.3572,
+      "step": 107120
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.275962347899288e-05,
+      "loss": 1.2475,
+      "step": 107130
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.275847554909313e-05,
+      "loss": 1.254,
+      "step": 107140
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.275732761919338e-05,
+      "loss": 1.4442,
+      "step": 107150
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.275617968929364e-05,
+      "loss": 1.2943,
+      "step": 107160
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.275503175939388e-05,
+      "loss": 1.435,
+      "step": 107170
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.275388382949414e-05,
+      "loss": 1.2478,
+      "step": 107180
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.275273589959439e-05,
+      "loss": 1.3812,
+      "step": 107190
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.275158796969464e-05,
+      "loss": 1.1624,
+      "step": 107200
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.275044003979489e-05,
+      "loss": 1.373,
+      "step": 107210
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.274929210989516e-05,
+      "loss": 1.2899,
+      "step": 107220
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.274814417999539e-05,
+      "loss": 1.2787,
+      "step": 107230
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.274699625009566e-05,
+      "loss": 1.4591,
+      "step": 107240
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.27458483201959e-05,
+      "loss": 1.2036,
+      "step": 107250
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.274470039029616e-05,
+      "loss": 1.3172,
+      "step": 107260
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.274355246039642e-05,
+      "loss": 1.2214,
+      "step": 107270
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.274240453049667e-05,
+      "loss": 1.3667,
+      "step": 107280
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.274125660059692e-05,
+      "loss": 1.4264,
+      "step": 107290
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.274010867069717e-05,
+      "loss": 1.3147,
+      "step": 107300
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.273896074079743e-05,
+      "loss": 1.3687,
+      "step": 107310
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.273781281089767e-05,
+      "loss": 1.5812,
+      "step": 107320
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.273666488099793e-05,
+      "loss": 1.2913,
+      "step": 107330
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.273551695109818e-05,
+      "loss": 1.3798,
+      "step": 107340
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.273436902119843e-05,
+      "loss": 1.2707,
+      "step": 107350
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.273322109129868e-05,
+      "loss": 1.209,
+      "step": 107360
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.273207316139894e-05,
+      "loss": 1.2886,
+      "step": 107370
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.27309252314992e-05,
+      "loss": 1.4018,
+      "step": 107380
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.272977730159944e-05,
+      "loss": 1.4581,
+      "step": 107390
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.27286293716997e-05,
+      "loss": 1.3851,
+      "step": 107400
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.272748144179994e-05,
+      "loss": 1.2863,
+      "step": 107410
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.272633351190021e-05,
+      "loss": 1.2208,
+      "step": 107420
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.272518558200046e-05,
+      "loss": 1.293,
+      "step": 107430
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.272403765210071e-05,
+      "loss": 1.2676,
+      "step": 107440
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.272288972220096e-05,
+      "loss": 1.3856,
+      "step": 107450
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.272174179230122e-05,
+      "loss": 1.3697,
+      "step": 107460
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.272059386240146e-05,
+      "loss": 1.3362,
+      "step": 107470
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.271944593250172e-05,
+      "loss": 1.3417,
+      "step": 107480
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.271829800260197e-05,
+      "loss": 1.1829,
+      "step": 107490
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.271715007270222e-05,
+      "loss": 1.3464,
+      "step": 107500
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.271600214280247e-05,
+      "loss": 1.2738,
+      "step": 107510
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.271485421290273e-05,
+      "loss": 1.3666,
+      "step": 107520
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.271370628300298e-05,
+      "loss": 1.2389,
+      "step": 107530
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.271255835310323e-05,
+      "loss": 1.1679,
+      "step": 107540
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.271141042320348e-05,
+      "loss": 1.1579,
+      "step": 107550
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.271026249330373e-05,
+      "loss": 1.2382,
+      "step": 107560
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.2709114563404e-05,
+      "loss": 1.3098,
+      "step": 107570
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.270796663350425e-05,
+      "loss": 1.3263,
+      "step": 107580
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.27068187036045e-05,
+      "loss": 1.2511,
+      "step": 107590
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.270567077370475e-05,
+      "loss": 1.2328,
+      "step": 107600
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.270452284380501e-05,
+      "loss": 1.3186,
+      "step": 107610
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.270337491390525e-05,
+      "loss": 1.4163,
+      "step": 107620
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.270222698400551e-05,
+      "loss": 1.3588,
+      "step": 107630
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.270107905410576e-05,
+      "loss": 1.3959,
+      "step": 107640
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.269993112420601e-05,
+      "loss": 1.2775,
+      "step": 107650
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.269878319430626e-05,
+      "loss": 1.2202,
+      "step": 107660
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.269763526440652e-05,
+      "loss": 1.2607,
+      "step": 107670
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.269648733450677e-05,
+      "loss": 1.255,
+      "step": 107680
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.269533940460702e-05,
+      "loss": 1.3081,
+      "step": 107690
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.269419147470727e-05,
+      "loss": 1.2581,
+      "step": 107700
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.269304354480752e-05,
+      "loss": 1.1808,
+      "step": 107710
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.269189561490778e-05,
+      "loss": 1.2914,
+      "step": 107720
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.269074768500803e-05,
+      "loss": 1.3613,
+      "step": 107730
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.268959975510828e-05,
+      "loss": 1.2326,
+      "step": 107740
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.268845182520853e-05,
+      "loss": 1.3952,
+      "step": 107750
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.26873038953088e-05,
+      "loss": 1.3121,
+      "step": 107760
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.268615596540903e-05,
+      "loss": 1.3623,
+      "step": 107770
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.26850080355093e-05,
+      "loss": 1.2893,
+      "step": 107780
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.268386010560955e-05,
+      "loss": 1.1575,
+      "step": 107790
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.26827121757098e-05,
+      "loss": 1.2641,
+      "step": 107800
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.268156424581005e-05,
+      "loss": 1.3587,
+      "step": 107810
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.268041631591031e-05,
+      "loss": 1.2972,
+      "step": 107820
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.267926838601056e-05,
+      "loss": 1.3421,
+      "step": 107830
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.267812045611081e-05,
+      "loss": 1.2666,
+      "step": 107840
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.267697252621106e-05,
+      "loss": 1.472,
+      "step": 107850
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.267582459631131e-05,
+      "loss": 1.3129,
+      "step": 107860
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.267467666641157e-05,
+      "loss": 1.3474,
+      "step": 107870
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.267352873651182e-05,
+      "loss": 1.0948,
+      "step": 107880
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.267238080661207e-05,
+      "loss": 1.35,
+      "step": 107890
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.267123287671232e-05,
+      "loss": 1.3912,
+      "step": 107900
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.267008494681259e-05,
+      "loss": 1.2653,
+      "step": 107910
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.266893701691282e-05,
+      "loss": 1.2622,
+      "step": 107920
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.266778908701309e-05,
+      "loss": 1.3263,
+      "step": 107930
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.266664115711334e-05,
+      "loss": 1.2976,
+      "step": 107940
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.266549322721359e-05,
+      "loss": 1.3789,
+      "step": 107950
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.266434529731384e-05,
+      "loss": 1.3252,
+      "step": 107960
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.26631973674141e-05,
+      "loss": 1.3405,
+      "step": 107970
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.266204943751435e-05,
+      "loss": 1.2725,
+      "step": 107980
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.26609015076146e-05,
+      "loss": 1.4776,
+      "step": 107990
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.265975357771485e-05,
+      "loss": 1.3098,
+      "step": 108000
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.26586056478151e-05,
+      "loss": 1.2081,
+      "step": 108010
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.265745771791536e-05,
+      "loss": 1.3335,
+      "step": 108020
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.265630978801561e-05,
+      "loss": 1.2175,
+      "step": 108030
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.265516185811586e-05,
+      "loss": 1.3574,
+      "step": 108040
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.265401392821611e-05,
+      "loss": 1.3609,
+      "step": 108050
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.265286599831637e-05,
+      "loss": 1.3949,
+      "step": 108060
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.265171806841661e-05,
+      "loss": 1.3851,
+      "step": 108070
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.265057013851687e-05,
+      "loss": 1.3009,
+      "step": 108080
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.264942220861712e-05,
+      "loss": 1.2682,
+      "step": 108090
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.264827427871737e-05,
+      "loss": 1.2257,
+      "step": 108100
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.264712634881762e-05,
+      "loss": 1.2805,
+      "step": 108110
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.264597841891789e-05,
+      "loss": 1.3047,
+      "step": 108120
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.264483048901814e-05,
+      "loss": 1.2617,
+      "step": 108130
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.264368255911839e-05,
+      "loss": 1.1382,
+      "step": 108140
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.264253462921864e-05,
+      "loss": 1.3262,
+      "step": 108150
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.264138669931889e-05,
+      "loss": 1.4506,
+      "step": 108160
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.264023876941915e-05,
+      "loss": 1.3898,
+      "step": 108170
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.26390908395194e-05,
+      "loss": 1.3264,
+      "step": 108180
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.263794290961965e-05,
+      "loss": 1.3425,
+      "step": 108190
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.26367949797199e-05,
+      "loss": 1.2172,
+      "step": 108200
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.263564704982016e-05,
+      "loss": 1.3316,
+      "step": 108210
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.26344991199204e-05,
+      "loss": 1.2588,
+      "step": 108220
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.263335119002066e-05,
+      "loss": 1.2713,
+      "step": 108230
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.263220326012091e-05,
+      "loss": 1.2131,
+      "step": 108240
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.263105533022116e-05,
+      "loss": 1.2475,
+      "step": 108250
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.262990740032141e-05,
+      "loss": 1.3695,
+      "step": 108260
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.262875947042166e-05,
+      "loss": 1.1573,
+      "step": 108270
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.262761154052193e-05,
+      "loss": 1.3246,
+      "step": 108280
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.262646361062218e-05,
+      "loss": 1.4369,
+      "step": 108290
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.262531568072243e-05,
+      "loss": 1.3706,
+      "step": 108300
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.262416775082267e-05,
+      "loss": 1.3818,
+      "step": 108310
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.262301982092294e-05,
+      "loss": 1.2572,
+      "step": 108320
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.262187189102317e-05,
+      "loss": 1.3978,
+      "step": 108330
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.262072396112344e-05,
+      "loss": 1.2429,
+      "step": 108340
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.261957603122369e-05,
+      "loss": 1.1749,
+      "step": 108350
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.261842810132394e-05,
+      "loss": 1.2754,
+      "step": 108360
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.261728017142419e-05,
+      "loss": 1.2932,
+      "step": 108370
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.261613224152445e-05,
+      "loss": 1.3588,
+      "step": 108380
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.261498431162469e-05,
+      "loss": 1.329,
+      "step": 108390
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.261383638172495e-05,
+      "loss": 1.3454,
+      "step": 108400
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.26126884518252e-05,
+      "loss": 1.4835,
+      "step": 108410
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.261154052192545e-05,
+      "loss": 1.3031,
+      "step": 108420
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.261039259202571e-05,
+      "loss": 1.3286,
+      "step": 108430
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.260924466212596e-05,
+      "loss": 1.2771,
+      "step": 108440
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.260809673222621e-05,
+      "loss": 1.207,
+      "step": 108450
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.260694880232646e-05,
+      "loss": 1.4173,
+      "step": 108460
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.260580087242673e-05,
+      "loss": 1.3033,
+      "step": 108470
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.260465294252696e-05,
+      "loss": 1.2234,
+      "step": 108480
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.260350501262723e-05,
+      "loss": 1.2793,
+      "step": 108490
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.260235708272748e-05,
+      "loss": 1.2488,
+      "step": 108500
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.260120915282773e-05,
+      "loss": 1.2531,
+      "step": 108510
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.260006122292798e-05,
+      "loss": 1.264,
+      "step": 108520
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.259891329302824e-05,
+      "loss": 1.3345,
+      "step": 108530
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.259776536312848e-05,
+      "loss": 1.2776,
+      "step": 108540
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.259661743322874e-05,
+      "loss": 1.301,
+      "step": 108550
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.259546950332899e-05,
+      "loss": 1.3473,
+      "step": 108560
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.259432157342924e-05,
+      "loss": 1.4023,
+      "step": 108570
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.25931736435295e-05,
+      "loss": 1.3465,
+      "step": 108580
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.259202571362975e-05,
+      "loss": 1.2574,
+      "step": 108590
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.259087778373e-05,
+      "loss": 1.269,
+      "step": 108600
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.258972985383025e-05,
+      "loss": 1.2186,
+      "step": 108610
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.258858192393052e-05,
+      "loss": 1.3129,
+      "step": 108620
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.258743399403075e-05,
+      "loss": 1.215,
+      "step": 108630
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.258628606413102e-05,
+      "loss": 1.2983,
+      "step": 108640
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.258513813423126e-05,
+      "loss": 1.2752,
+      "step": 108650
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.258399020433151e-05,
+      "loss": 1.3641,
+      "step": 108660
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.258284227443176e-05,
+      "loss": 1.2465,
+      "step": 108670
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.258169434453203e-05,
+      "loss": 1.3151,
+      "step": 108680
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.258054641463226e-05,
+      "loss": 1.143,
+      "step": 108690
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.257939848473253e-05,
+      "loss": 1.2313,
+      "step": 108700
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.257825055483278e-05,
+      "loss": 1.5211,
+      "step": 108710
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.257710262493303e-05,
+      "loss": 1.3225,
+      "step": 108720
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.257595469503329e-05,
+      "loss": 1.1988,
+      "step": 108730
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.257480676513354e-05,
+      "loss": 1.1501,
+      "step": 108740
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.257365883523379e-05,
+      "loss": 1.2619,
+      "step": 108750
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.257251090533404e-05,
+      "loss": 1.3401,
+      "step": 108760
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.25713629754343e-05,
+      "loss": 1.3727,
+      "step": 108770
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.257021504553454e-05,
+      "loss": 1.3794,
+      "step": 108780
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.25690671156348e-05,
+      "loss": 1.2678,
+      "step": 108790
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.256791918573505e-05,
+      "loss": 1.3645,
+      "step": 108800
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.25667712558353e-05,
+      "loss": 1.2013,
+      "step": 108810
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.256562332593555e-05,
+      "loss": 1.3755,
+      "step": 108820
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.256447539603582e-05,
+      "loss": 1.2586,
+      "step": 108830
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.256332746613605e-05,
+      "loss": 1.4838,
+      "step": 108840
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.256217953623632e-05,
+      "loss": 1.3241,
+      "step": 108850
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.256103160633657e-05,
+      "loss": 1.3024,
+      "step": 108860
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.255988367643682e-05,
+      "loss": 1.3974,
+      "step": 108870
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.255873574653708e-05,
+      "loss": 1.3554,
+      "step": 108880
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.255758781663733e-05,
+      "loss": 1.4884,
+      "step": 108890
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.255643988673758e-05,
+      "loss": 1.3529,
+      "step": 108900
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.255529195683783e-05,
+      "loss": 1.1515,
+      "step": 108910
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.255414402693809e-05,
+      "loss": 1.3416,
+      "step": 108920
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.255299609703833e-05,
+      "loss": 1.2678,
+      "step": 108930
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.255184816713859e-05,
+      "loss": 1.2592,
+      "step": 108940
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.255070023723884e-05,
+      "loss": 1.3672,
+      "step": 108950
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.254955230733909e-05,
+      "loss": 1.2715,
+      "step": 108960
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.254840437743934e-05,
+      "loss": 1.1987,
+      "step": 108970
+    },
+    {
+      "epoch": 5.0,
+      "eval_cer": 0.6701074597949316,
+      "eval_em": 0.007357212003872217,
+      "eval_f1": 0.007357212003872217,
+      "eval_loss": 1.226467251777649,
+      "eval_runtime": 2267.8815,
+      "eval_samples_per_second": 4.555,
+      "eval_steps_per_second": 0.57,
+      "eval_wer": 0.9926427879961278,
+      "step": 108976
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.25472564475396e-05,
+      "loss": 1.2755,
+      "step": 108980
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.254610851763984e-05,
+      "loss": 1.1448,
+      "step": 108990
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.25449605877401e-05,
+      "loss": 1.3636,
+      "step": 109000
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.254381265784035e-05,
+      "loss": 1.2663,
+      "step": 109010
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.25426647279406e-05,
+      "loss": 1.1419,
+      "step": 109020
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.254151679804087e-05,
+      "loss": 1.0951,
+      "step": 109030
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.254036886814112e-05,
+      "loss": 1.2082,
+      "step": 109040
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.253922093824137e-05,
+      "loss": 1.3267,
+      "step": 109050
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.253807300834162e-05,
+      "loss": 1.1967,
+      "step": 109060
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.253692507844188e-05,
+      "loss": 1.3903,
+      "step": 109070
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.253577714854212e-05,
+      "loss": 1.3975,
+      "step": 109080
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.253462921864238e-05,
+      "loss": 1.2424,
+      "step": 109090
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.253348128874263e-05,
+      "loss": 1.6031,
+      "step": 109100
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.253233335884288e-05,
+      "loss": 1.2211,
+      "step": 109110
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.253118542894313e-05,
+      "loss": 1.1989,
+      "step": 109120
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.25300374990434e-05,
+      "loss": 1.2433,
+      "step": 109130
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.252888956914363e-05,
+      "loss": 1.3057,
+      "step": 109140
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.252774163924389e-05,
+      "loss": 1.2153,
+      "step": 109150
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.252659370934414e-05,
+      "loss": 1.1373,
+      "step": 109160
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.252544577944439e-05,
+      "loss": 1.2116,
+      "step": 109170
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.252429784954466e-05,
+      "loss": 1.2886,
+      "step": 109180
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.25231499196449e-05,
+      "loss": 1.1668,
+      "step": 109190
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.252200198974516e-05,
+      "loss": 1.1792,
+      "step": 109200
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.25208540598454e-05,
+      "loss": 1.3433,
+      "step": 109210
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.251970612994567e-05,
+      "loss": 1.3795,
+      "step": 109220
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.25185582000459e-05,
+      "loss": 1.1784,
+      "step": 109230
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.251741027014617e-05,
+      "loss": 1.2835,
+      "step": 109240
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.251626234024642e-05,
+      "loss": 1.2531,
+      "step": 109250
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.251511441034667e-05,
+      "loss": 1.1446,
+      "step": 109260
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.251396648044692e-05,
+      "loss": 1.1654,
+      "step": 109270
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.251281855054718e-05,
+      "loss": 1.2325,
+      "step": 109280
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.251167062064742e-05,
+      "loss": 1.2586,
+      "step": 109290
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 6.251052269074768e-05,
+      "loss": 1.1584,
+      "step": 109300
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.250937476084793e-05,
+      "loss": 1.1793,
+      "step": 109310
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.250822683094818e-05,
+      "loss": 1.2307,
+      "step": 109320
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.250707890104844e-05,
+      "loss": 1.254,
+      "step": 109330
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.25059309711487e-05,
+      "loss": 1.2903,
+      "step": 109340
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.250478304124894e-05,
+      "loss": 1.3069,
+      "step": 109350
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.25036351113492e-05,
+      "loss": 1.3139,
+      "step": 109360
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.250248718144946e-05,
+      "loss": 1.3097,
+      "step": 109370
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.25013392515497e-05,
+      "loss": 1.3123,
+      "step": 109380
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.250019132164996e-05,
+      "loss": 1.3362,
+      "step": 109390
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.249904339175021e-05,
+      "loss": 1.4057,
+      "step": 109400
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.249789546185046e-05,
+      "loss": 1.2769,
+      "step": 109410
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.249674753195071e-05,
+      "loss": 1.2493,
+      "step": 109420
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.249559960205097e-05,
+      "loss": 1.295,
+      "step": 109430
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.24944516721512e-05,
+      "loss": 1.1702,
+      "step": 109440
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.249330374225147e-05,
+      "loss": 1.2639,
+      "step": 109450
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.249215581235172e-05,
+      "loss": 1.2557,
+      "step": 109460
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.249100788245197e-05,
+      "loss": 1.2574,
+      "step": 109470
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.248985995255223e-05,
+      "loss": 1.3116,
+      "step": 109480
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.248871202265248e-05,
+      "loss": 1.3783,
+      "step": 109490
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.248756409275273e-05,
+      "loss": 1.2776,
+      "step": 109500
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.248641616285298e-05,
+      "loss": 1.3304,
+      "step": 109510
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 6.248526823295325e-05,
+      "loss": 1.3105,
+      "step": 109520
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.248412030305348e-05,
+      "loss": 1.2194,
+      "step": 109530
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.248297237315375e-05,
+      "loss": 1.3335,
+      "step": 109540
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.2481824443254e-05,
+      "loss": 1.244,
+      "step": 109550
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.248067651335425e-05,
+      "loss": 1.3454,
+      "step": 109560
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.24795285834545e-05,
+      "loss": 1.2098,
+      "step": 109570
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.247838065355476e-05,
+      "loss": 1.3274,
+      "step": 109580
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.2477232723655e-05,
+      "loss": 1.0526,
+      "step": 109590
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.247608479375526e-05,
+      "loss": 1.2441,
+      "step": 109600
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.247493686385551e-05,
+      "loss": 1.2608,
+      "step": 109610
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.247378893395576e-05,
+      "loss": 1.3031,
+      "step": 109620
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.247264100405602e-05,
+      "loss": 1.1908,
+      "step": 109630
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.247149307415626e-05,
+      "loss": 1.2704,
+      "step": 109640
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.247034514425652e-05,
+      "loss": 1.0785,
+      "step": 109650
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.246919721435677e-05,
+      "loss": 1.3532,
+      "step": 109660
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.246804928445702e-05,
+      "loss": 1.2296,
+      "step": 109670
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.246690135455727e-05,
+      "loss": 1.3025,
+      "step": 109680
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.246575342465753e-05,
+      "loss": 1.2752,
+      "step": 109690
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.246460549475777e-05,
+      "loss": 1.3365,
+      "step": 109700
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.246345756485803e-05,
+      "loss": 1.1584,
+      "step": 109710
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.246230963495828e-05,
+      "loss": 1.369,
+      "step": 109720
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 6.246116170505853e-05,
+      "loss": 1.2991,
+      "step": 109730
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.246001377515878e-05,
+      "loss": 1.3296,
+      "step": 109740
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.245886584525905e-05,
+      "loss": 1.1849,
+      "step": 109750
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.24577179153593e-05,
+      "loss": 1.3261,
+      "step": 109760
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.245656998545955e-05,
+      "loss": 1.3723,
+      "step": 109770
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.245542205555981e-05,
+      "loss": 1.2706,
+      "step": 109780
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.245427412566005e-05,
+      "loss": 1.1005,
+      "step": 109790
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.245312619576031e-05,
+      "loss": 1.2639,
+      "step": 109800
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.245197826586056e-05,
+      "loss": 1.4461,
+      "step": 109810
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.245083033596081e-05,
+      "loss": 1.2572,
+      "step": 109820
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.244968240606106e-05,
+      "loss": 1.3772,
+      "step": 109830
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.244853447616132e-05,
+      "loss": 1.2386,
+      "step": 109840
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.244738654626156e-05,
+      "loss": 1.2125,
+      "step": 109850
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.244623861636182e-05,
+      "loss": 1.1528,
+      "step": 109860
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.244509068646207e-05,
+      "loss": 1.2576,
+      "step": 109870
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.244394275656232e-05,
+      "loss": 1.3314,
+      "step": 109880
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.244279482666257e-05,
+      "loss": 1.1292,
+      "step": 109890
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.244164689676284e-05,
+      "loss": 1.5068,
+      "step": 109900
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.244049896686309e-05,
+      "loss": 1.1751,
+      "step": 109910
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.243935103696334e-05,
+      "loss": 1.1685,
+      "step": 109920
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.24382031070636e-05,
+      "loss": 1.2504,
+      "step": 109930
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.243705517716383e-05,
+      "loss": 1.2359,
+      "step": 109940
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 6.24359072472641e-05,
+      "loss": 1.2002,
+      "step": 109950
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.243475931736435e-05,
+      "loss": 1.3594,
+      "step": 109960
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.24336113874646e-05,
+      "loss": 1.1829,
+      "step": 109970
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.243246345756485e-05,
+      "loss": 1.3924,
+      "step": 109980
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.243131552766511e-05,
+      "loss": 1.3255,
+      "step": 109990
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.243016759776535e-05,
+      "loss": 1.2869,
+      "step": 110000
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.242901966786561e-05,
+      "loss": 1.3145,
+      "step": 110010
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.242787173796586e-05,
+      "loss": 1.2704,
+      "step": 110020
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.242672380806611e-05,
+      "loss": 1.2102,
+      "step": 110030
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.242557587816636e-05,
+      "loss": 1.243,
+      "step": 110040
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.242442794826662e-05,
+      "loss": 1.2879,
+      "step": 110050
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.242328001836687e-05,
+      "loss": 1.4002,
+      "step": 110060
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.242213208846712e-05,
+      "loss": 1.2867,
+      "step": 110070
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.242098415856739e-05,
+      "loss": 1.2612,
+      "step": 110080
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.241983622866762e-05,
+      "loss": 1.1833,
+      "step": 110090
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.241868829876789e-05,
+      "loss": 1.2798,
+      "step": 110100
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.241754036886814e-05,
+      "loss": 1.221,
+      "step": 110110
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.241639243896839e-05,
+      "loss": 1.1625,
+      "step": 110120
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.241524450906864e-05,
+      "loss": 1.1825,
+      "step": 110130
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.24140965791689e-05,
+      "loss": 1.1754,
+      "step": 110140
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.241294864926914e-05,
+      "loss": 1.1613,
+      "step": 110150
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.24118007193694e-05,
+      "loss": 1.3552,
+      "step": 110160
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 6.241065278946965e-05,
+      "loss": 1.251,
+      "step": 110170
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.24095048595699e-05,
+      "loss": 1.2298,
+      "step": 110180
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.240835692967015e-05,
+      "loss": 1.1935,
+      "step": 110190
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.240720899977041e-05,
+      "loss": 1.426,
+      "step": 110200
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.240606106987066e-05,
+      "loss": 1.3004,
+      "step": 110210
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.240491313997091e-05,
+      "loss": 1.2942,
+      "step": 110220
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.240376521007118e-05,
+      "loss": 1.2251,
+      "step": 110230
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.240261728017141e-05,
+      "loss": 1.2654,
+      "step": 110240
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.240146935027168e-05,
+      "loss": 1.1249,
+      "step": 110250
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.240032142037193e-05,
+      "loss": 1.36,
+      "step": 110260
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.239917349047217e-05,
+      "loss": 1.1996,
+      "step": 110270
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.239802556057242e-05,
+      "loss": 1.2027,
+      "step": 110280
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.239687763067269e-05,
+      "loss": 1.3242,
+      "step": 110290
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.239572970077292e-05,
+      "loss": 1.2673,
+      "step": 110300
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.239458177087319e-05,
+      "loss": 1.3504,
+      "step": 110310
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.239343384097344e-05,
+      "loss": 1.2246,
+      "step": 110320
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.239228591107369e-05,
+      "loss": 1.0792,
+      "step": 110330
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.239113798117394e-05,
+      "loss": 1.1724,
+      "step": 110340
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.23899900512742e-05,
+      "loss": 1.2537,
+      "step": 110350
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.238884212137445e-05,
+      "loss": 1.3254,
+      "step": 110360
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.23876941914747e-05,
+      "loss": 1.3907,
+      "step": 110370
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.238654626157496e-05,
+      "loss": 1.1788,
+      "step": 110380
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 6.23853983316752e-05,
+      "loss": 1.2431,
+      "step": 110390
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.238425040177546e-05,
+      "loss": 1.345,
+      "step": 110400
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.238310247187571e-05,
+      "loss": 1.3315,
+      "step": 110410
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.238195454197596e-05,
+      "loss": 1.2621,
+      "step": 110420
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.238080661207621e-05,
+      "loss": 1.3347,
+      "step": 110430
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.237965868217648e-05,
+      "loss": 1.4177,
+      "step": 110440
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.237851075227671e-05,
+      "loss": 1.2835,
+      "step": 110450
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.237736282237698e-05,
+      "loss": 1.2696,
+      "step": 110460
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.237621489247723e-05,
+      "loss": 1.1881,
+      "step": 110470
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.237506696257748e-05,
+      "loss": 1.2119,
+      "step": 110480
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.237391903267773e-05,
+      "loss": 1.1961,
+      "step": 110490
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.237277110277799e-05,
+      "loss": 1.2653,
+      "step": 110500
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.237162317287824e-05,
+      "loss": 1.4203,
+      "step": 110510
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.237047524297849e-05,
+      "loss": 1.1576,
+      "step": 110520
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.236932731307875e-05,
+      "loss": 1.2402,
+      "step": 110530
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.236817938317899e-05,
+      "loss": 1.3514,
+      "step": 110540
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.236703145327925e-05,
+      "loss": 1.2254,
+      "step": 110550
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.23658835233795e-05,
+      "loss": 1.1886,
+      "step": 110560
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.236473559347975e-05,
+      "loss": 1.2088,
+      "step": 110570
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.236358766358e-05,
+      "loss": 1.259,
+      "step": 110580
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.236243973368027e-05,
+      "loss": 1.1989,
+      "step": 110590
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.23612918037805e-05,
+      "loss": 1.2551,
+      "step": 110600
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 6.236014387388076e-05,
+      "loss": 1.2976,
+      "step": 110610
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.235899594398101e-05,
+      "loss": 1.1818,
+      "step": 110620
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.235784801408126e-05,
+      "loss": 1.158,
+      "step": 110630
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.235670008418151e-05,
+      "loss": 1.2972,
+      "step": 110640
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.235555215428178e-05,
+      "loss": 1.2581,
+      "step": 110650
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.235440422438203e-05,
+      "loss": 1.1816,
+      "step": 110660
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.235325629448228e-05,
+      "loss": 1.1715,
+      "step": 110670
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.235210836458254e-05,
+      "loss": 1.2018,
+      "step": 110680
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.235096043468278e-05,
+      "loss": 1.2521,
+      "step": 110690
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.234981250478304e-05,
+      "loss": 1.1971,
+      "step": 110700
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.234866457488329e-05,
+      "loss": 1.3356,
+      "step": 110710
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.234751664498354e-05,
+      "loss": 1.3303,
+      "step": 110720
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.234636871508379e-05,
+      "loss": 1.2178,
+      "step": 110730
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.234522078518405e-05,
+      "loss": 1.1583,
+      "step": 110740
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.234407285528429e-05,
+      "loss": 1.2019,
+      "step": 110750
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.234292492538455e-05,
+      "loss": 1.1288,
+      "step": 110760
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.23417769954848e-05,
+      "loss": 1.4161,
+      "step": 110770
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.234062906558505e-05,
+      "loss": 1.2597,
+      "step": 110780
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.23394811356853e-05,
+      "loss": 1.2703,
+      "step": 110790
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.233833320578557e-05,
+      "loss": 1.2425,
+      "step": 110800
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.233718527588582e-05,
+      "loss": 1.3034,
+      "step": 110810
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 6.233603734598607e-05,
+      "loss": 1.2225,
+      "step": 110820
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.233488941608633e-05,
+      "loss": 1.276,
+      "step": 110830
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.233374148618657e-05,
+      "loss": 1.4799,
+      "step": 110840
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.233259355628683e-05,
+      "loss": 1.2735,
+      "step": 110850
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.233144562638708e-05,
+      "loss": 1.311,
+      "step": 110860
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.233029769648733e-05,
+      "loss": 1.3277,
+      "step": 110870
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.232914976658758e-05,
+      "loss": 1.3647,
+      "step": 110880
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.232800183668784e-05,
+      "loss": 1.1732,
+      "step": 110890
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.232685390678808e-05,
+      "loss": 1.2119,
+      "step": 110900
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.232570597688834e-05,
+      "loss": 1.2253,
+      "step": 110910
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.232455804698859e-05,
+      "loss": 1.2086,
+      "step": 110920
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.232341011708884e-05,
+      "loss": 1.2027,
+      "step": 110930
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.232226218718909e-05,
+      "loss": 1.315,
+      "step": 110940
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.232111425728935e-05,
+      "loss": 1.1814,
+      "step": 110950
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.23199663273896e-05,
+      "loss": 1.2208,
+      "step": 110960
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.231881839748985e-05,
+      "loss": 1.2239,
+      "step": 110970
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.231767046759012e-05,
+      "loss": 1.2037,
+      "step": 110980
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.231652253769035e-05,
+      "loss": 1.2152,
+      "step": 110990
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.231537460779062e-05,
+      "loss": 1.3341,
+      "step": 111000
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.231422667789085e-05,
+      "loss": 1.3555,
+      "step": 111010
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.231307874799112e-05,
+      "loss": 1.2547,
+      "step": 111020
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.231193081809137e-05,
+      "loss": 1.3935,
+      "step": 111030
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 6.231078288819162e-05,
+      "loss": 1.2458,
+      "step": 111040
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.230963495829187e-05,
+      "loss": 1.3472,
+      "step": 111050
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.230848702839213e-05,
+      "loss": 1.2913,
+      "step": 111060
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.230733909849238e-05,
+      "loss": 1.335,
+      "step": 111070
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.230619116859263e-05,
+      "loss": 1.2438,
+      "step": 111080
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.230504323869288e-05,
+      "loss": 1.2471,
+      "step": 111090
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.230389530879313e-05,
+      "loss": 1.2787,
+      "step": 111100
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.230274737889339e-05,
+      "loss": 1.2897,
+      "step": 111110
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.230159944899364e-05,
+      "loss": 1.4621,
+      "step": 111120
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.230045151909389e-05,
+      "loss": 1.2323,
+      "step": 111130
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.229930358919414e-05,
+      "loss": 1.2391,
+      "step": 111140
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.22981556592944e-05,
+      "loss": 1.2784,
+      "step": 111150
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.229700772939464e-05,
+      "loss": 1.2798,
+      "step": 111160
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.22958597994949e-05,
+      "loss": 1.2728,
+      "step": 111170
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.229471186959516e-05,
+      "loss": 1.1982,
+      "step": 111180
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.22935639396954e-05,
+      "loss": 1.1881,
+      "step": 111190
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.229241600979566e-05,
+      "loss": 1.2284,
+      "step": 111200
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.229126807989592e-05,
+      "loss": 1.2858,
+      "step": 111210
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.229012014999617e-05,
+      "loss": 1.3332,
+      "step": 111220
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.228897222009642e-05,
+      "loss": 1.1867,
+      "step": 111230
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.228782429019667e-05,
+      "loss": 1.3506,
+      "step": 111240
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.228667636029692e-05,
+      "loss": 1.3865,
+      "step": 111250
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 6.228552843039718e-05,
+      "loss": 1.2474,
+      "step": 111260
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.228438050049743e-05,
+      "loss": 1.1616,
+      "step": 111270
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.228323257059768e-05,
+      "loss": 1.4417,
+      "step": 111280
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.228208464069793e-05,
+      "loss": 1.2896,
+      "step": 111290
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.22809367107982e-05,
+      "loss": 1.0669,
+      "step": 111300
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.227978878089843e-05,
+      "loss": 1.3279,
+      "step": 111310
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.22786408509987e-05,
+      "loss": 1.2742,
+      "step": 111320
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.227749292109894e-05,
+      "loss": 1.1968,
+      "step": 111330
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.22763449911992e-05,
+      "loss": 1.2012,
+      "step": 111340
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.227519706129944e-05,
+      "loss": 1.3709,
+      "step": 111350
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.227404913139971e-05,
+      "loss": 1.2826,
+      "step": 111360
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.227290120149996e-05,
+      "loss": 1.3132,
+      "step": 111370
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.227175327160021e-05,
+      "loss": 1.3428,
+      "step": 111380
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.227060534170046e-05,
+      "loss": 1.3702,
+      "step": 111390
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.22694574118007e-05,
+      "loss": 1.283,
+      "step": 111400
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.226830948190097e-05,
+      "loss": 1.2045,
+      "step": 111410
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.226716155200122e-05,
+      "loss": 1.2249,
+      "step": 111420
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.226601362210147e-05,
+      "loss": 1.2117,
+      "step": 111430
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.226486569220172e-05,
+      "loss": 1.2532,
+      "step": 111440
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.226371776230198e-05,
+      "loss": 1.242,
+      "step": 111450
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.226256983240222e-05,
+      "loss": 1.327,
+      "step": 111460
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.226142190250248e-05,
+      "loss": 1.3057,
+      "step": 111470
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 6.226027397260273e-05,
+      "loss": 1.2414,
+      "step": 111480
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.225912604270298e-05,
+      "loss": 1.2631,
+      "step": 111490
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.225797811280323e-05,
+      "loss": 1.1345,
+      "step": 111500
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.22568301829035e-05,
+      "loss": 1.2949,
+      "step": 111510
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.225568225300375e-05,
+      "loss": 1.2471,
+      "step": 111520
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.2254534323104e-05,
+      "loss": 1.3686,
+      "step": 111530
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.225338639320425e-05,
+      "loss": 1.2128,
+      "step": 111540
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.22522384633045e-05,
+      "loss": 1.2788,
+      "step": 111550
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.225109053340476e-05,
+      "loss": 1.1615,
+      "step": 111560
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.224994260350501e-05,
+      "loss": 1.3913,
+      "step": 111570
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.224879467360526e-05,
+      "loss": 1.1452,
+      "step": 111580
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.224764674370551e-05,
+      "loss": 1.2105,
+      "step": 111590
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.224649881380577e-05,
+      "loss": 1.2266,
+      "step": 111600
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.224535088390601e-05,
+      "loss": 1.0598,
+      "step": 111610
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.224420295400627e-05,
+      "loss": 1.2881,
+      "step": 111620
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.224305502410652e-05,
+      "loss": 1.1471,
+      "step": 111630
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.224190709420677e-05,
+      "loss": 1.1688,
+      "step": 111640
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.224075916430702e-05,
+      "loss": 1.2948,
+      "step": 111650
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.223961123440728e-05,
+      "loss": 1.0524,
+      "step": 111660
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.223846330450753e-05,
+      "loss": 1.2609,
+      "step": 111670
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.223731537460778e-05,
+      "loss": 1.3965,
+      "step": 111680
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.223616744470803e-05,
+      "loss": 1.2582,
+      "step": 111690
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 6.223501951480828e-05,
+      "loss": 1.3036,
+      "step": 111700
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.223387158490855e-05,
+      "loss": 1.3149,
+      "step": 111710
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.22327236550088e-05,
+      "loss": 1.1641,
+      "step": 111720
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.223157572510905e-05,
+      "loss": 1.308,
+      "step": 111730
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.22304277952093e-05,
+      "loss": 1.281,
+      "step": 111740
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.222927986530956e-05,
+      "loss": 1.2508,
+      "step": 111750
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.22281319354098e-05,
+      "loss": 1.4217,
+      "step": 111760
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.222698400551006e-05,
+      "loss": 1.2726,
+      "step": 111770
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.222583607561031e-05,
+      "loss": 1.3041,
+      "step": 111780
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.222468814571056e-05,
+      "loss": 1.3199,
+      "step": 111790
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.222354021581081e-05,
+      "loss": 1.4252,
+      "step": 111800
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.222239228591107e-05,
+      "loss": 1.2139,
+      "step": 111810
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.222124435601132e-05,
+      "loss": 1.2986,
+      "step": 111820
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.222009642611157e-05,
+      "loss": 1.1987,
+      "step": 111830
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.221894849621182e-05,
+      "loss": 1.4063,
+      "step": 111840
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.221780056631207e-05,
+      "loss": 1.3587,
+      "step": 111850
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.221665263641234e-05,
+      "loss": 1.2728,
+      "step": 111860
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.221550470651259e-05,
+      "loss": 1.2512,
+      "step": 111870
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.221435677661284e-05,
+      "loss": 1.2158,
+      "step": 111880
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.221320884671309e-05,
+      "loss": 1.3802,
+      "step": 111890
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.221206091681335e-05,
+      "loss": 1.206,
+      "step": 111900
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 6.221091298691358e-05,
+      "loss": 1.2771,
+      "step": 111910
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.220976505701385e-05,
+      "loss": 1.2919,
+      "step": 111920
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.22086171271141e-05,
+      "loss": 1.1324,
+      "step": 111930
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.220746919721435e-05,
+      "loss": 1.2213,
+      "step": 111940
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.22063212673146e-05,
+      "loss": 1.4598,
+      "step": 111950
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.220517333741486e-05,
+      "loss": 1.3362,
+      "step": 111960
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.220402540751511e-05,
+      "loss": 1.2533,
+      "step": 111970
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.220287747761536e-05,
+      "loss": 1.3632,
+      "step": 111980
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.220172954771561e-05,
+      "loss": 1.2078,
+      "step": 111990
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.220058161781586e-05,
+      "loss": 1.2787,
+      "step": 112000
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.219943368791612e-05,
+      "loss": 1.343,
+      "step": 112010
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.219828575801637e-05,
+      "loss": 1.2246,
+      "step": 112020
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.219713782811662e-05,
+      "loss": 1.3269,
+      "step": 112030
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.219598989821687e-05,
+      "loss": 1.3246,
+      "step": 112040
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.219484196831714e-05,
+      "loss": 1.1609,
+      "step": 112050
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.219369403841737e-05,
+      "loss": 1.2141,
+      "step": 112060
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.219254610851764e-05,
+      "loss": 1.2628,
+      "step": 112070
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.219139817861789e-05,
+      "loss": 1.2757,
+      "step": 112080
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.219025024871814e-05,
+      "loss": 1.2852,
+      "step": 112090
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.218910231881839e-05,
+      "loss": 1.2161,
+      "step": 112100
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.218795438891865e-05,
+      "loss": 1.2376,
+      "step": 112110
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.21868064590189e-05,
+      "loss": 1.2103,
+      "step": 112120
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 6.218565852911915e-05,
+      "loss": 1.2567,
+      "step": 112130
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.21845105992194e-05,
+      "loss": 1.3039,
+      "step": 112140
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.218336266931965e-05,
+      "loss": 1.189,
+      "step": 112150
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.218221473941991e-05,
+      "loss": 1.2018,
+      "step": 112160
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.218106680952016e-05,
+      "loss": 1.2837,
+      "step": 112170
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.217991887962041e-05,
+      "loss": 1.2881,
+      "step": 112180
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.217877094972066e-05,
+      "loss": 1.262,
+      "step": 112190
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.217762301982093e-05,
+      "loss": 1.342,
+      "step": 112200
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.217647508992116e-05,
+      "loss": 1.2723,
+      "step": 112210
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.217532716002143e-05,
+      "loss": 1.4389,
+      "step": 112220
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.217417923012167e-05,
+      "loss": 1.1081,
+      "step": 112230
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.217303130022192e-05,
+      "loss": 1.2617,
+      "step": 112240
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.217188337032217e-05,
+      "loss": 1.2393,
+      "step": 112250
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.217073544042244e-05,
+      "loss": 1.2945,
+      "step": 112260
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.216958751052269e-05,
+      "loss": 1.2096,
+      "step": 112270
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.216843958062294e-05,
+      "loss": 1.2167,
+      "step": 112280
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.216729165072319e-05,
+      "loss": 1.2935,
+      "step": 112290
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.216614372082344e-05,
+      "loss": 1.3016,
+      "step": 112300
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.21649957909237e-05,
+      "loss": 1.1894,
+      "step": 112310
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.216384786102395e-05,
+      "loss": 1.4241,
+      "step": 112320
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.21626999311242e-05,
+      "loss": 1.1529,
+      "step": 112330
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.216155200122445e-05,
+      "loss": 1.3001,
+      "step": 112340
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 6.216040407132471e-05,
+      "loss": 1.1446,
+      "step": 112350
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.215925614142495e-05,
+      "loss": 1.3228,
+      "step": 112360
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.215810821152521e-05,
+      "loss": 1.1461,
+      "step": 112370
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.215696028162546e-05,
+      "loss": 1.3785,
+      "step": 112380
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.215581235172571e-05,
+      "loss": 1.1403,
+      "step": 112390
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.215466442182596e-05,
+      "loss": 1.3446,
+      "step": 112400
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.215351649192621e-05,
+      "loss": 1.1544,
+      "step": 112410
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.215236856202648e-05,
+      "loss": 1.3469,
+      "step": 112420
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.215122063212673e-05,
+      "loss": 1.181,
+      "step": 112430
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.215007270222698e-05,
+      "loss": 1.2804,
+      "step": 112440
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.214892477232723e-05,
+      "loss": 1.2227,
+      "step": 112450
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.214777684242749e-05,
+      "loss": 1.1599,
+      "step": 112460
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.214662891252773e-05,
+      "loss": 1.1913,
+      "step": 112470
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.214548098262799e-05,
+      "loss": 1.3901,
+      "step": 112480
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.214433305272824e-05,
+      "loss": 1.2393,
+      "step": 112490
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.214318512282849e-05,
+      "loss": 1.1253,
+      "step": 112500
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.214203719292874e-05,
+      "loss": 1.208,
+      "step": 112510
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.2140889263029e-05,
+      "loss": 1.3032,
+      "step": 112520
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.213974133312925e-05,
+      "loss": 1.3219,
+      "step": 112530
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.21385934032295e-05,
+      "loss": 1.2166,
+      "step": 112540
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.213744547332975e-05,
+      "loss": 1.2769,
+      "step": 112550
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.213629754343e-05,
+      "loss": 1.2674,
+      "step": 112560
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.213514961353026e-05,
+      "loss": 1.3616,
+      "step": 112570
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.213400168363051e-05,
+      "loss": 1.2845,
+      "step": 112580
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.213285375373076e-05,
+      "loss": 1.2156,
+      "step": 112590
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.213170582383101e-05,
+      "loss": 1.2098,
+      "step": 112600
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.213055789393128e-05,
+      "loss": 1.27,
+      "step": 112610
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.212940996403151e-05,
+      "loss": 1.1999,
+      "step": 112620
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.212826203413178e-05,
+      "loss": 1.218,
+      "step": 112630
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.212711410423203e-05,
+      "loss": 1.2523,
+      "step": 112640
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.212596617433228e-05,
+      "loss": 1.3686,
+      "step": 112650
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.212481824443253e-05,
+      "loss": 1.2836,
+      "step": 112660
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.212367031453279e-05,
+      "loss": 1.2234,
+      "step": 112670
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.212252238463304e-05,
+      "loss": 1.1272,
+      "step": 112680
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.212137445473329e-05,
+      "loss": 1.2022,
+      "step": 112690
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.212022652483354e-05,
+      "loss": 1.3595,
+      "step": 112700
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.211907859493379e-05,
+      "loss": 1.1144,
+      "step": 112710
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.211793066503405e-05,
+      "loss": 1.4919,
+      "step": 112720
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.21167827351343e-05,
+      "loss": 1.384,
+      "step": 112730
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.211563480523455e-05,
+      "loss": 1.2682,
+      "step": 112740
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.21144868753348e-05,
+      "loss": 1.2221,
+      "step": 112750
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.211333894543507e-05,
+      "loss": 1.3235,
+      "step": 112760
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.21121910155353e-05,
+      "loss": 1.204,
+      "step": 112770
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.211104308563557e-05,
+      "loss": 1.3252,
+      "step": 112780
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.210989515573582e-05,
+      "loss": 1.368,
+      "step": 112790
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.210874722583607e-05,
+      "loss": 1.3737,
+      "step": 112800
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.210759929593632e-05,
+      "loss": 1.2668,
+      "step": 112810
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.210645136603658e-05,
+      "loss": 1.2385,
+      "step": 112820
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.210530343613683e-05,
+      "loss": 1.2753,
+      "step": 112830
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.210415550623708e-05,
+      "loss": 1.1977,
+      "step": 112840
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.210300757633733e-05,
+      "loss": 1.209,
+      "step": 112850
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.210185964643758e-05,
+      "loss": 1.3948,
+      "step": 112860
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.210071171653784e-05,
+      "loss": 1.3416,
+      "step": 112870
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.209956378663809e-05,
+      "loss": 1.2651,
+      "step": 112880
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.209841585673834e-05,
+      "loss": 1.1745,
+      "step": 112890
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.209726792683859e-05,
+      "loss": 1.2095,
+      "step": 112900
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.209611999693885e-05,
+      "loss": 1.1347,
+      "step": 112910
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.209497206703909e-05,
+      "loss": 1.3235,
+      "step": 112920
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.209382413713935e-05,
+      "loss": 1.3803,
+      "step": 112930
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.20926762072396e-05,
+      "loss": 1.2081,
+      "step": 112940
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.209152827733985e-05,
+      "loss": 1.146,
+      "step": 112950
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.20903803474401e-05,
+      "loss": 1.3203,
+      "step": 112960
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.208923241754037e-05,
+      "loss": 1.3089,
+      "step": 112970
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.208808448764062e-05,
+      "loss": 1.2595,
+      "step": 112980
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.208693655774087e-05,
+      "loss": 1.3401,
+      "step": 112990
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.208578862784112e-05,
+      "loss": 1.1963,
+      "step": 113000
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.208464069794137e-05,
+      "loss": 1.2092,
+      "step": 113010
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.208349276804163e-05,
+      "loss": 1.2767,
+      "step": 113020
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.208234483814188e-05,
+      "loss": 1.1451,
+      "step": 113030
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.208119690824213e-05,
+      "loss": 1.1546,
+      "step": 113040
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.208004897834238e-05,
+      "loss": 1.3443,
+      "step": 113050
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.207890104844264e-05,
+      "loss": 1.2893,
+      "step": 113060
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.207775311854288e-05,
+      "loss": 1.2001,
+      "step": 113070
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.207660518864314e-05,
+      "loss": 1.1739,
+      "step": 113080
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.207545725874339e-05,
+      "loss": 1.3335,
+      "step": 113090
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.207430932884364e-05,
+      "loss": 1.229,
+      "step": 113100
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.207316139894389e-05,
+      "loss": 1.1885,
+      "step": 113110
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.207201346904416e-05,
+      "loss": 1.2456,
+      "step": 113120
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.20708655391444e-05,
+      "loss": 1.3677,
+      "step": 113130
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.206971760924466e-05,
+      "loss": 1.2428,
+      "step": 113140
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.20685696793449e-05,
+      "loss": 1.4088,
+      "step": 113150
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.206742174944516e-05,
+      "loss": 1.301,
+      "step": 113160
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.206627381954542e-05,
+      "loss": 1.3691,
+      "step": 113170
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.206512588964567e-05,
+      "loss": 1.2517,
+      "step": 113180
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.206397795974592e-05,
+      "loss": 1.2194,
+      "step": 113190
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.206283002984617e-05,
+      "loss": 1.1859,
+      "step": 113200
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.206168209994643e-05,
+      "loss": 1.2068,
+      "step": 113210
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.206053417004667e-05,
+      "loss": 1.132,
+      "step": 113220
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.205938624014693e-05,
+      "loss": 1.2159,
+      "step": 113230
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.205823831024718e-05,
+      "loss": 1.2722,
+      "step": 113240
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.205709038034743e-05,
+      "loss": 1.2996,
+      "step": 113250
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.205594245044768e-05,
+      "loss": 1.2676,
+      "step": 113260
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.205479452054794e-05,
+      "loss": 1.0219,
+      "step": 113270
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.20536465906482e-05,
+      "loss": 1.1308,
+      "step": 113280
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.205249866074844e-05,
+      "loss": 1.2227,
+      "step": 113290
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.20513507308487e-05,
+      "loss": 1.1434,
+      "step": 113300
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.205020280094894e-05,
+      "loss": 1.3572,
+      "step": 113310
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.204905487104921e-05,
+      "loss": 1.1202,
+      "step": 113320
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.204790694114946e-05,
+      "loss": 1.2759,
+      "step": 113330
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.204675901124971e-05,
+      "loss": 1.277,
+      "step": 113340
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.204561108134996e-05,
+      "loss": 1.2669,
+      "step": 113350
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.204446315145022e-05,
+      "loss": 1.2931,
+      "step": 113360
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.204331522155046e-05,
+      "loss": 1.3266,
+      "step": 113370
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.204216729165072e-05,
+      "loss": 1.3323,
+      "step": 113380
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.204101936175097e-05,
+      "loss": 1.2445,
+      "step": 113390
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.203987143185122e-05,
+      "loss": 1.2513,
+      "step": 113400
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.203872350195147e-05,
+      "loss": 1.4208,
+      "step": 113410
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.203757557205173e-05,
+      "loss": 1.3773,
+      "step": 113420
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.203642764215198e-05,
+      "loss": 1.1632,
+      "step": 113430
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.203527971225223e-05,
+      "loss": 1.3211,
+      "step": 113440
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.203413178235248e-05,
+      "loss": 1.2305,
+      "step": 113450
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.203298385245273e-05,
+      "loss": 1.2776,
+      "step": 113460
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.2031835922553e-05,
+      "loss": 1.1763,
+      "step": 113470
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.203068799265325e-05,
+      "loss": 1.1835,
+      "step": 113480
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.20295400627535e-05,
+      "loss": 1.3705,
+      "step": 113490
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.202839213285375e-05,
+      "loss": 1.3542,
+      "step": 113500
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.202724420295401e-05,
+      "loss": 1.2785,
+      "step": 113510
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.202609627305424e-05,
+      "loss": 1.2134,
+      "step": 113520
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.202494834315451e-05,
+      "loss": 1.4034,
+      "step": 113530
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.202380041325476e-05,
+      "loss": 1.2355,
+      "step": 113540
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.202265248335501e-05,
+      "loss": 1.2753,
+      "step": 113550
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.202150455345526e-05,
+      "loss": 1.4464,
+      "step": 113560
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.202035662355552e-05,
+      "loss": 1.1526,
+      "step": 113570
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.201920869365577e-05,
+      "loss": 1.1715,
+      "step": 113580
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.201806076375602e-05,
+      "loss": 1.1844,
+      "step": 113590
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.201691283385627e-05,
+      "loss": 1.2312,
+      "step": 113600
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.201576490395652e-05,
+      "loss": 1.1257,
+      "step": 113610
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.201461697405678e-05,
+      "loss": 1.1887,
+      "step": 113620
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.201346904415703e-05,
+      "loss": 1.3753,
+      "step": 113630
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.201232111425728e-05,
+      "loss": 1.3382,
+      "step": 113640
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.201117318435753e-05,
+      "loss": 1.3252,
+      "step": 113650
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.20100252544578e-05,
+      "loss": 1.4186,
+      "step": 113660
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.200887732455803e-05,
+      "loss": 1.2809,
+      "step": 113670
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.20077293946583e-05,
+      "loss": 1.3678,
+      "step": 113680
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.200658146475855e-05,
+      "loss": 1.2507,
+      "step": 113690
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.20054335348588e-05,
+      "loss": 1.287,
+      "step": 113700
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.200428560495905e-05,
+      "loss": 1.2798,
+      "step": 113710
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.200313767505931e-05,
+      "loss": 1.2147,
+      "step": 113720
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.200198974515956e-05,
+      "loss": 1.3192,
+      "step": 113730
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.200084181525981e-05,
+      "loss": 1.1976,
+      "step": 113740
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.199969388536006e-05,
+      "loss": 1.2962,
+      "step": 113750
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.199854595546031e-05,
+      "loss": 1.2593,
+      "step": 113760
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.199739802556057e-05,
+      "loss": 1.2749,
+      "step": 113770
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.199625009566081e-05,
+      "loss": 1.2296,
+      "step": 113780
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.199510216576107e-05,
+      "loss": 1.2744,
+      "step": 113790
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.199395423586132e-05,
+      "loss": 1.273,
+      "step": 113800
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.199280630596157e-05,
+      "loss": 1.1841,
+      "step": 113810
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.199165837606182e-05,
+      "loss": 1.2759,
+      "step": 113820
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.199051044616209e-05,
+      "loss": 1.3643,
+      "step": 113830
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.198936251626234e-05,
+      "loss": 1.2701,
+      "step": 113840
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.198821458636259e-05,
+      "loss": 1.2896,
+      "step": 113850
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.198706665646283e-05,
+      "loss": 1.2609,
+      "step": 113860
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.198591872656308e-05,
+      "loss": 1.2319,
+      "step": 113870
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.198477079666335e-05,
+      "loss": 1.1907,
+      "step": 113880
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.19836228667636e-05,
+      "loss": 1.2385,
+      "step": 113890
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.198247493686385e-05,
+      "loss": 1.2983,
+      "step": 113900
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.19813270069641e-05,
+      "loss": 1.4374,
+      "step": 113910
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.198017907706436e-05,
+      "loss": 1.1166,
+      "step": 113920
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.19790311471646e-05,
+      "loss": 1.1211,
+      "step": 113930
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.197788321726486e-05,
+      "loss": 1.3619,
+      "step": 113940
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.197673528736511e-05,
+      "loss": 1.259,
+      "step": 113950
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.197558735746536e-05,
+      "loss": 1.1873,
+      "step": 113960
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.197443942756561e-05,
+      "loss": 1.2418,
+      "step": 113970
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.197329149766587e-05,
+      "loss": 1.3639,
+      "step": 113980
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.197214356776612e-05,
+      "loss": 1.2676,
+      "step": 113990
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.197099563786637e-05,
+      "loss": 1.1731,
+      "step": 114000
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.196984770796662e-05,
+      "loss": 1.1501,
+      "step": 114010
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.196869977806687e-05,
+      "loss": 1.3694,
+      "step": 114020
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.196755184816714e-05,
+      "loss": 1.2866,
+      "step": 114030
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.196640391826739e-05,
+      "loss": 1.3244,
+      "step": 114040
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.196525598836764e-05,
+      "loss": 1.3276,
+      "step": 114050
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.196410805846789e-05,
+      "loss": 1.1687,
+      "step": 114060
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.196296012856815e-05,
+      "loss": 1.4842,
+      "step": 114070
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.196181219866839e-05,
+      "loss": 1.2101,
+      "step": 114080
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.196066426876865e-05,
+      "loss": 1.2187,
+      "step": 114090
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.19595163388689e-05,
+      "loss": 1.3697,
+      "step": 114100
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.195836840896915e-05,
+      "loss": 1.2857,
+      "step": 114110
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.19572204790694e-05,
+      "loss": 1.2638,
+      "step": 114120
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.195607254916966e-05,
+      "loss": 1.4066,
+      "step": 114130
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.195492461926991e-05,
+      "loss": 1.2591,
+      "step": 114140
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.195377668937016e-05,
+      "loss": 1.2929,
+      "step": 114150
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.195262875947041e-05,
+      "loss": 1.1692,
+      "step": 114160
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.195148082957066e-05,
+      "loss": 1.2827,
+      "step": 114170
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.195033289967093e-05,
+      "loss": 1.2722,
+      "step": 114180
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.194918496977117e-05,
+      "loss": 1.2613,
+      "step": 114190
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.194803703987142e-05,
+      "loss": 1.2523,
+      "step": 114200
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.194688910997167e-05,
+      "loss": 1.3277,
+      "step": 114210
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.194574118007194e-05,
+      "loss": 1.2385,
+      "step": 114220
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.194459325017217e-05,
+      "loss": 1.0859,
+      "step": 114230
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.194344532027244e-05,
+      "loss": 1.341,
+      "step": 114240
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.194229739037269e-05,
+      "loss": 1.0768,
+      "step": 114250
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.194114946047294e-05,
+      "loss": 1.3348,
+      "step": 114260
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.194000153057319e-05,
+      "loss": 1.2188,
+      "step": 114270
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.193885360067345e-05,
+      "loss": 1.3749,
+      "step": 114280
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.19377056707737e-05,
+      "loss": 1.277,
+      "step": 114290
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.193655774087395e-05,
+      "loss": 1.4269,
+      "step": 114300
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.19354098109742e-05,
+      "loss": 1.2292,
+      "step": 114310
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.193426188107445e-05,
+      "loss": 1.2878,
+      "step": 114320
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.193311395117471e-05,
+      "loss": 1.2531,
+      "step": 114330
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.193196602127496e-05,
+      "loss": 1.2083,
+      "step": 114340
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.193081809137521e-05,
+      "loss": 1.3204,
+      "step": 114350
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.192967016147546e-05,
+      "loss": 1.1391,
+      "step": 114360
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.192852223157573e-05,
+      "loss": 1.3118,
+      "step": 114370
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.192737430167596e-05,
+      "loss": 1.3048,
+      "step": 114380
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.192622637177623e-05,
+      "loss": 1.2072,
+      "step": 114390
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.192507844187648e-05,
+      "loss": 1.2437,
+      "step": 114400
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.192393051197673e-05,
+      "loss": 1.1464,
+      "step": 114410
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.192278258207698e-05,
+      "loss": 1.3337,
+      "step": 114420
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.192163465217724e-05,
+      "loss": 1.2921,
+      "step": 114430
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.192048672227749e-05,
+      "loss": 1.1214,
+      "step": 114440
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.191933879237774e-05,
+      "loss": 1.1737,
+      "step": 114450
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.191819086247799e-05,
+      "loss": 1.3641,
+      "step": 114460
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.191704293257824e-05,
+      "loss": 1.3119,
+      "step": 114470
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.19158950026785e-05,
+      "loss": 1.1649,
+      "step": 114480
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.191474707277875e-05,
+      "loss": 1.2393,
+      "step": 114490
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.1913599142879e-05,
+      "loss": 1.4615,
+      "step": 114500
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.191245121297925e-05,
+      "loss": 1.388,
+      "step": 114510
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.191130328307952e-05,
+      "loss": 1.2606,
+      "step": 114520
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.191015535317975e-05,
+      "loss": 1.187,
+      "step": 114530
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.190900742328001e-05,
+      "loss": 1.2654,
+      "step": 114540
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.190785949338026e-05,
+      "loss": 1.1817,
+      "step": 114550
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.190671156348051e-05,
+      "loss": 1.1231,
+      "step": 114560
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.190556363358076e-05,
+      "loss": 1.1459,
+      "step": 114570
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.190441570368103e-05,
+      "loss": 1.2561,
+      "step": 114580
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.190326777378128e-05,
+      "loss": 1.2512,
+      "step": 114590
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.190211984388153e-05,
+      "loss": 1.2014,
+      "step": 114600
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.190097191398178e-05,
+      "loss": 1.298,
+      "step": 114610
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.189982398408203e-05,
+      "loss": 1.4453,
+      "step": 114620
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.189867605418229e-05,
+      "loss": 1.2462,
+      "step": 114630
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.189752812428254e-05,
+      "loss": 1.2321,
+      "step": 114640
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.189638019438279e-05,
+      "loss": 1.1786,
+      "step": 114650
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.189523226448304e-05,
+      "loss": 1.1967,
+      "step": 114660
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.18940843345833e-05,
+      "loss": 1.404,
+      "step": 114670
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.189293640468354e-05,
+      "loss": 1.1811,
+      "step": 114680
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.18917884747838e-05,
+      "loss": 1.2755,
+      "step": 114690
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.189064054488405e-05,
+      "loss": 1.2165,
+      "step": 114700
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.18894926149843e-05,
+      "loss": 1.2294,
+      "step": 114710
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.188834468508455e-05,
+      "loss": 1.2657,
+      "step": 114720
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.188719675518482e-05,
+      "loss": 1.1787,
+      "step": 114730
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.188604882528507e-05,
+      "loss": 1.3203,
+      "step": 114740
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.188490089538532e-05,
+      "loss": 1.3713,
+      "step": 114750
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.188375296548557e-05,
+      "loss": 1.3211,
+      "step": 114760
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.188260503558582e-05,
+      "loss": 1.336,
+      "step": 114770
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.188145710568608e-05,
+      "loss": 1.3715,
+      "step": 114780
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.188030917578633e-05,
+      "loss": 1.2649,
+      "step": 114790
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.187916124588658e-05,
+      "loss": 1.2168,
+      "step": 114800
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.187801331598683e-05,
+      "loss": 1.2308,
+      "step": 114810
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.187686538608709e-05,
+      "loss": 1.3203,
+      "step": 114820
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.187571745618733e-05,
+      "loss": 1.2627,
+      "step": 114830
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.187456952628759e-05,
+      "loss": 1.1939,
+      "step": 114840
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.187342159638784e-05,
+      "loss": 1.2952,
+      "step": 114850
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.187227366648809e-05,
+      "loss": 1.2656,
+      "step": 114860
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.187112573658834e-05,
+      "loss": 1.2372,
+      "step": 114870
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.18699778066886e-05,
+      "loss": 1.217,
+      "step": 114880
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.186882987678885e-05,
+      "loss": 1.1805,
+      "step": 114890
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.18676819468891e-05,
+      "loss": 1.4087,
+      "step": 114900
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.186653401698935e-05,
+      "loss": 1.3909,
+      "step": 114910
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.18653860870896e-05,
+      "loss": 1.2271,
+      "step": 114920
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.186423815718987e-05,
+      "loss": 1.2877,
+      "step": 114930
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.186309022729012e-05,
+      "loss": 1.2205,
+      "step": 114940
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.186194229739037e-05,
+      "loss": 1.1567,
+      "step": 114950
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.186079436749062e-05,
+      "loss": 1.2569,
+      "step": 114960
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.185964643759088e-05,
+      "loss": 1.3083,
+      "step": 114970
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.185849850769112e-05,
+      "loss": 1.2901,
+      "step": 114980
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.185735057779138e-05,
+      "loss": 1.1879,
+      "step": 114990
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.185620264789163e-05,
+      "loss": 1.3623,
+      "step": 115000
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.185505471799188e-05,
+      "loss": 1.1385,
+      "step": 115010
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.185390678809213e-05,
+      "loss": 1.3449,
+      "step": 115020
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.185275885819239e-05,
+      "loss": 1.2428,
+      "step": 115030
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.185161092829264e-05,
+      "loss": 1.2063,
+      "step": 115040
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.185046299839289e-05,
+      "loss": 1.1764,
+      "step": 115050
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.184931506849314e-05,
+      "loss": 1.2469,
+      "step": 115060
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.184816713859339e-05,
+      "loss": 1.2807,
+      "step": 115070
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.184701920869366e-05,
+      "loss": 1.1991,
+      "step": 115080
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.18458712787939e-05,
+      "loss": 1.3007,
+      "step": 115090
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.184472334889416e-05,
+      "loss": 1.2083,
+      "step": 115100
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.18435754189944e-05,
+      "loss": 1.3379,
+      "step": 115110
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.184242748909466e-05,
+      "loss": 1.2511,
+      "step": 115120
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.18412795591949e-05,
+      "loss": 1.1898,
+      "step": 115130
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.184013162929517e-05,
+      "loss": 1.2291,
+      "step": 115140
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.183898369939542e-05,
+      "loss": 1.2056,
+      "step": 115150
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.183783576949567e-05,
+      "loss": 1.2419,
+      "step": 115160
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.183668783959592e-05,
+      "loss": 1.238,
+      "step": 115170
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.183553990969617e-05,
+      "loss": 1.2728,
+      "step": 115180
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.183439197979643e-05,
+      "loss": 1.2114,
+      "step": 115190
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.183324404989668e-05,
+      "loss": 1.3677,
+      "step": 115200
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.183209611999693e-05,
+      "loss": 1.2302,
+      "step": 115210
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.183094819009718e-05,
+      "loss": 1.2414,
+      "step": 115220
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.182980026019744e-05,
+      "loss": 1.308,
+      "step": 115230
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.182865233029768e-05,
+      "loss": 1.2079,
+      "step": 115240
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.182750440039794e-05,
+      "loss": 1.3018,
+      "step": 115250
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.18263564704982e-05,
+      "loss": 1.1658,
+      "step": 115260
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.182520854059844e-05,
+      "loss": 1.3805,
+      "step": 115270
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.18240606106987e-05,
+      "loss": 1.2685,
+      "step": 115280
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.182291268079896e-05,
+      "loss": 1.3493,
+      "step": 115290
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.182176475089921e-05,
+      "loss": 1.3064,
+      "step": 115300
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.182061682099946e-05,
+      "loss": 1.3492,
+      "step": 115310
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.18194688910997e-05,
+      "loss": 1.2879,
+      "step": 115320
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.181832096119996e-05,
+      "loss": 1.2413,
+      "step": 115330
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.181717303130022e-05,
+      "loss": 1.1951,
+      "step": 115340
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.181602510140047e-05,
+      "loss": 1.2098,
+      "step": 115350
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.181487717150072e-05,
+      "loss": 1.1093,
+      "step": 115360
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.181372924160097e-05,
+      "loss": 1.1382,
+      "step": 115370
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.181258131170123e-05,
+      "loss": 1.3389,
+      "step": 115380
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.181143338180147e-05,
+      "loss": 1.217,
+      "step": 115390
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 6.181028545190173e-05,
+      "loss": 1.073,
+      "step": 115400
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.180913752200198e-05,
+      "loss": 1.456,
+      "step": 115410
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.180798959210223e-05,
+      "loss": 1.3063,
+      "step": 115420
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.180684166220248e-05,
+      "loss": 1.3017,
+      "step": 115430
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.180569373230275e-05,
+      "loss": 1.1948,
+      "step": 115440
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.1804545802403e-05,
+      "loss": 1.4008,
+      "step": 115450
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.180339787250325e-05,
+      "loss": 1.2919,
+      "step": 115460
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.18022499426035e-05,
+      "loss": 1.2463,
+      "step": 115470
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.180110201270374e-05,
+      "loss": 1.2105,
+      "step": 115480
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.179995408280401e-05,
+      "loss": 1.3081,
+      "step": 115490
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.179880615290426e-05,
+      "loss": 1.2788,
+      "step": 115500
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.179765822300451e-05,
+      "loss": 1.1826,
+      "step": 115510
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.179651029310476e-05,
+      "loss": 1.1559,
+      "step": 115520
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.179536236320502e-05,
+      "loss": 1.1556,
+      "step": 115530
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.179421443330526e-05,
+      "loss": 1.2461,
+      "step": 115540
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.179306650340552e-05,
+      "loss": 1.1919,
+      "step": 115550
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.179191857350577e-05,
+      "loss": 1.295,
+      "step": 115560
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.179077064360602e-05,
+      "loss": 1.2145,
+      "step": 115570
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.178962271370627e-05,
+      "loss": 1.2591,
+      "step": 115580
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.178847478380653e-05,
+      "loss": 1.2868,
+      "step": 115590
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.178732685390678e-05,
+      "loss": 1.2669,
+      "step": 115600
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.178617892400703e-05,
+      "loss": 1.4935,
+      "step": 115610
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.178503099410728e-05,
+      "loss": 1.1791,
+      "step": 115620
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.178388306420753e-05,
+      "loss": 1.2526,
+      "step": 115630
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.17827351343078e-05,
+      "loss": 1.2138,
+      "step": 115640
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.178158720440805e-05,
+      "loss": 1.2673,
+      "step": 115650
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.17804392745083e-05,
+      "loss": 1.2293,
+      "step": 115660
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.177929134460855e-05,
+      "loss": 1.1922,
+      "step": 115670
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.177814341470881e-05,
+      "loss": 1.3972,
+      "step": 115680
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.177699548480905e-05,
+      "loss": 1.266,
+      "step": 115690
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.177584755490931e-05,
+      "loss": 1.2909,
+      "step": 115700
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.177469962500956e-05,
+      "loss": 1.0967,
+      "step": 115710
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.177355169510981e-05,
+      "loss": 1.2498,
+      "step": 115720
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.177240376521006e-05,
+      "loss": 1.3134,
+      "step": 115730
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.177125583531032e-05,
+      "loss": 1.2154,
+      "step": 115740
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.177010790541057e-05,
+      "loss": 1.2496,
+      "step": 115750
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.176895997551082e-05,
+      "loss": 1.1104,
+      "step": 115760
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.176781204561107e-05,
+      "loss": 1.3369,
+      "step": 115770
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.176666411571132e-05,
+      "loss": 1.2043,
+      "step": 115780
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.176551618581159e-05,
+      "loss": 1.3261,
+      "step": 115790
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.176436825591184e-05,
+      "loss": 1.2939,
+      "step": 115800
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.176322032601209e-05,
+      "loss": 1.1972,
+      "step": 115810
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.176207239611233e-05,
+      "loss": 1.3024,
+      "step": 115820
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.17609244662126e-05,
+      "loss": 1.2548,
+      "step": 115830
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.175977653631283e-05,
+      "loss": 1.3514,
+      "step": 115840
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.17586286064131e-05,
+      "loss": 1.2139,
+      "step": 115850
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.175748067651335e-05,
+      "loss": 1.1876,
+      "step": 115860
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.17563327466136e-05,
+      "loss": 1.284,
+      "step": 115870
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.175518481671385e-05,
+      "loss": 1.2242,
+      "step": 115880
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.175403688681411e-05,
+      "loss": 1.2175,
+      "step": 115890
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.175288895691436e-05,
+      "loss": 1.3095,
+      "step": 115900
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.175174102701461e-05,
+      "loss": 1.2619,
+      "step": 115910
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.175059309711486e-05,
+      "loss": 1.1708,
+      "step": 115920
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.174944516721511e-05,
+      "loss": 1.3124,
+      "step": 115930
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.174829723731537e-05,
+      "loss": 1.4597,
+      "step": 115940
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.174714930741562e-05,
+      "loss": 1.2442,
+      "step": 115950
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.174600137751587e-05,
+      "loss": 1.2373,
+      "step": 115960
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.174485344761612e-05,
+      "loss": 1.3006,
+      "step": 115970
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.174370551771639e-05,
+      "loss": 1.337,
+      "step": 115980
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.174255758781662e-05,
+      "loss": 1.2612,
+      "step": 115990
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.174140965791689e-05,
+      "loss": 1.4215,
+      "step": 116000
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.174026172801714e-05,
+      "loss": 1.4734,
+      "step": 116010
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.173911379811739e-05,
+      "loss": 1.2424,
+      "step": 116020
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.173796586821764e-05,
+      "loss": 1.2483,
+      "step": 116030
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.17368179383179e-05,
+      "loss": 1.3123,
+      "step": 116040
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.173567000841815e-05,
+      "loss": 1.1982,
+      "step": 116050
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.17345220785184e-05,
+      "loss": 1.0465,
+      "step": 116060
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.173337414861865e-05,
+      "loss": 1.3115,
+      "step": 116070
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.17322262187189e-05,
+      "loss": 1.3636,
+      "step": 116080
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.173107828881916e-05,
+      "loss": 1.3134,
+      "step": 116090
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.172993035891941e-05,
+      "loss": 1.3712,
+      "step": 116100
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.172878242901966e-05,
+      "loss": 1.165,
+      "step": 116110
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.172763449911991e-05,
+      "loss": 1.2211,
+      "step": 116120
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.172648656922018e-05,
+      "loss": 1.2434,
+      "step": 116130
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.172533863932041e-05,
+      "loss": 1.165,
+      "step": 116140
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.172419070942067e-05,
+      "loss": 1.1866,
+      "step": 116150
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.172304277952092e-05,
+      "loss": 1.3596,
+      "step": 116160
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.172189484962117e-05,
+      "loss": 1.2136,
+      "step": 116170
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.172074691972142e-05,
+      "loss": 1.3371,
+      "step": 116180
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.171959898982169e-05,
+      "loss": 1.289,
+      "step": 116190
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.171845105992194e-05,
+      "loss": 1.222,
+      "step": 116200
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.171730313002219e-05,
+      "loss": 1.2665,
+      "step": 116210
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.171615520012244e-05,
+      "loss": 1.2197,
+      "step": 116220
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.171500727022269e-05,
+      "loss": 1.2489,
+      "step": 116230
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.171385934032295e-05,
+      "loss": 1.0756,
+      "step": 116240
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.17127114104232e-05,
+      "loss": 1.3827,
+      "step": 116250
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.171156348052345e-05,
+      "loss": 1.1499,
+      "step": 116260
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.17104155506237e-05,
+      "loss": 1.3074,
+      "step": 116270
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.170926762072396e-05,
+      "loss": 1.424,
+      "step": 116280
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.17081196908242e-05,
+      "loss": 1.286,
+      "step": 116290
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.170697176092446e-05,
+      "loss": 1.2973,
+      "step": 116300
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.170582383102471e-05,
+      "loss": 1.3496,
+      "step": 116310
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.170467590112496e-05,
+      "loss": 1.1481,
+      "step": 116320
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.170352797122521e-05,
+      "loss": 1.3145,
+      "step": 116330
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.170238004132548e-05,
+      "loss": 1.3328,
+      "step": 116340
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.170123211142573e-05,
+      "loss": 1.2058,
+      "step": 116350
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.170008418152598e-05,
+      "loss": 1.1186,
+      "step": 116360
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.169893625162623e-05,
+      "loss": 1.2074,
+      "step": 116370
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.169778832172648e-05,
+      "loss": 1.37,
+      "step": 116380
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.169664039182674e-05,
+      "loss": 1.079,
+      "step": 116390
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.169549246192699e-05,
+      "loss": 1.3965,
+      "step": 116400
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.169434453202724e-05,
+      "loss": 1.174,
+      "step": 116410
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.169319660212749e-05,
+      "loss": 1.3632,
+      "step": 116420
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.169204867222775e-05,
+      "loss": 1.1745,
+      "step": 116430
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.169090074232799e-05,
+      "loss": 1.2878,
+      "step": 116440
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.168975281242825e-05,
+      "loss": 1.3158,
+      "step": 116450
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.16886048825285e-05,
+      "loss": 1.315,
+      "step": 116460
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.168745695262875e-05,
+      "loss": 1.2712,
+      "step": 116470
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.1686309022729e-05,
+      "loss": 1.3607,
+      "step": 116480
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.168516109282925e-05,
+      "loss": 1.3238,
+      "step": 116490
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.168401316292951e-05,
+      "loss": 1.3861,
+      "step": 116500
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.168286523302976e-05,
+      "loss": 1.2806,
+      "step": 116510
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.168171730313001e-05,
+      "loss": 1.1885,
+      "step": 116520
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.168056937323026e-05,
+      "loss": 1.2616,
+      "step": 116530
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.167942144333053e-05,
+      "loss": 1.2619,
+      "step": 116540
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.167827351343076e-05,
+      "loss": 1.3384,
+      "step": 116550
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.167712558353103e-05,
+      "loss": 1.2965,
+      "step": 116560
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.167597765363128e-05,
+      "loss": 1.1004,
+      "step": 116570
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.167482972373153e-05,
+      "loss": 1.3594,
+      "step": 116580
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.167368179383178e-05,
+      "loss": 1.2553,
+      "step": 116590
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.167253386393204e-05,
+      "loss": 1.2792,
+      "step": 116600
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.167138593403229e-05,
+      "loss": 1.2945,
+      "step": 116610
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.167023800413254e-05,
+      "loss": 1.2539,
+      "step": 116620
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.166909007423279e-05,
+      "loss": 1.2842,
+      "step": 116630
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.166794214433304e-05,
+      "loss": 1.3162,
+      "step": 116640
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.16667942144333e-05,
+      "loss": 1.2541,
+      "step": 116650
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.166564628453355e-05,
+      "loss": 1.3444,
+      "step": 116660
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.16644983546338e-05,
+      "loss": 1.2183,
+      "step": 116670
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.166335042473405e-05,
+      "loss": 1.3382,
+      "step": 116680
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.166220249483432e-05,
+      "loss": 1.2302,
+      "step": 116690
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.166105456493455e-05,
+      "loss": 1.1763,
+      "step": 116700
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.165990663503482e-05,
+      "loss": 1.3395,
+      "step": 116710
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.165875870513507e-05,
+      "loss": 1.2566,
+      "step": 116720
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.165761077523532e-05,
+      "loss": 1.2724,
+      "step": 116730
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.165646284533557e-05,
+      "loss": 1.2995,
+      "step": 116740
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.165531491543583e-05,
+      "loss": 1.346,
+      "step": 116750
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.165416698553608e-05,
+      "loss": 1.3173,
+      "step": 116760
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.165301905563633e-05,
+      "loss": 1.1616,
+      "step": 116770
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.165187112573658e-05,
+      "loss": 1.3085,
+      "step": 116780
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.165072319583683e-05,
+      "loss": 1.1963,
+      "step": 116790
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.164957526593709e-05,
+      "loss": 1.2317,
+      "step": 116800
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.164842733603734e-05,
+      "loss": 1.2758,
+      "step": 116810
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.164727940613759e-05,
+      "loss": 1.2242,
+      "step": 116820
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.164613147623784e-05,
+      "loss": 1.2249,
+      "step": 116830
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.16449835463381e-05,
+      "loss": 1.2237,
+      "step": 116840
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.164383561643834e-05,
+      "loss": 1.2132,
+      "step": 116850
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.16426876865386e-05,
+      "loss": 1.1579,
+      "step": 116860
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.164153975663885e-05,
+      "loss": 1.2117,
+      "step": 116870
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.16403918267391e-05,
+      "loss": 1.2717,
+      "step": 116880
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.163924389683935e-05,
+      "loss": 1.2287,
+      "step": 116890
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.163809596693962e-05,
+      "loss": 1.3028,
+      "step": 116900
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.163694803703987e-05,
+      "loss": 1.3306,
+      "step": 116910
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.163580010714012e-05,
+      "loss": 1.2873,
+      "step": 116920
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 6.163465217724037e-05,
+      "loss": 1.4111,
+      "step": 116930
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 6.163350424734062e-05,
+      "loss": 1.2798,
+      "step": 116940
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 6.163235631744088e-05,
+      "loss": 1.4264,
+      "step": 116950
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 6.163120838754113e-05,
+      "loss": 1.4334,
+      "step": 116960
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 6.163006045764138e-05,
+      "loss": 1.221,
+      "step": 116970
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 6.162891252774163e-05,
+      "loss": 1.3134,
+      "step": 116980
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 6.162776459784189e-05,
+      "loss": 1.2519,
+      "step": 116990
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 6.162661666794213e-05,
+      "loss": 1.3388,
+      "step": 117000
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 6.162546873804239e-05,
+      "loss": 1.2944,
+      "step": 117010
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 6.162432080814264e-05,
+      "loss": 1.206,
+      "step": 117020
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 6.162317287824289e-05,
+      "loss": 1.3424,
+      "step": 117030
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 6.162202494834314e-05,
+      "loss": 1.3106,
+      "step": 117040
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 6.16208770184434e-05,
+      "loss": 1.1695,
+      "step": 117050
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 6.161972908854366e-05,
+      "loss": 1.243,
+      "step": 117060
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 6.16185811586439e-05,
+      "loss": 1.3088,
+      "step": 117070
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 6.161743322874416e-05,
+      "loss": 1.2207,
+      "step": 117080
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 6.16162852988444e-05,
+      "loss": 1.2762,
+      "step": 117090
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 6.161513736894467e-05,
+      "loss": 1.2091,
+      "step": 117100
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 6.161398943904492e-05,
+      "loss": 1.2106,
+      "step": 117110
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 6.161284150914517e-05,
+      "loss": 1.3898,
+      "step": 117120
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 6.161169357924542e-05,
+      "loss": 1.3324,
+      "step": 117130
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 6.161054564934568e-05,
+      "loss": 1.228,
+      "step": 117140
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.160939771944592e-05,
+      "loss": 1.3368,
+      "step": 117150
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.160824978954618e-05,
+      "loss": 1.4224,
+      "step": 117160
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.160710185964643e-05,
+      "loss": 1.3813,
+      "step": 117170
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.160595392974668e-05,
+      "loss": 1.233,
+      "step": 117180
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.160480599984693e-05,
+      "loss": 1.2697,
+      "step": 117190
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.16036580699472e-05,
+      "loss": 1.2428,
+      "step": 117200
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.160251014004744e-05,
+      "loss": 1.233,
+      "step": 117210
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.16013622101477e-05,
+      "loss": 1.2795,
+      "step": 117220
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.160021428024794e-05,
+      "loss": 1.1645,
+      "step": 117230
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.15990663503482e-05,
+      "loss": 1.267,
+      "step": 117240
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.159791842044846e-05,
+      "loss": 1.2655,
+      "step": 117250
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.159677049054871e-05,
+      "loss": 1.2865,
+      "step": 117260
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.159562256064896e-05,
+      "loss": 1.27,
+      "step": 117270
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.15944746307492e-05,
+      "loss": 1.1226,
+      "step": 117280
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.159332670084947e-05,
+      "loss": 1.2799,
+      "step": 117290
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.15921787709497e-05,
+      "loss": 1.285,
+      "step": 117300
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.159103084104997e-05,
+      "loss": 1.3857,
+      "step": 117310
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.158988291115022e-05,
+      "loss": 1.3911,
+      "step": 117320
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.158873498125047e-05,
+      "loss": 1.3409,
+      "step": 117330
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.158758705135072e-05,
+      "loss": 1.3097,
+      "step": 117340
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.158643912145098e-05,
+      "loss": 1.2142,
+      "step": 117350
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 6.158529119155123e-05,
+      "loss": 1.3683,
+      "step": 117360
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.158414326165148e-05,
+      "loss": 1.2078,
+      "step": 117370
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.158299533175173e-05,
+      "loss": 1.2696,
+      "step": 117380
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.158184740185198e-05,
+      "loss": 1.3048,
+      "step": 117390
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.158069947195225e-05,
+      "loss": 1.1292,
+      "step": 117400
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.15795515420525e-05,
+      "loss": 1.3668,
+      "step": 117410
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.157840361215275e-05,
+      "loss": 1.255,
+      "step": 117420
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.1577255682253e-05,
+      "loss": 1.2965,
+      "step": 117430
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.157610775235326e-05,
+      "loss": 1.2567,
+      "step": 117440
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.15749598224535e-05,
+      "loss": 1.2677,
+      "step": 117450
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.157381189255376e-05,
+      "loss": 1.4348,
+      "step": 117460
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.157266396265401e-05,
+      "loss": 1.2372,
+      "step": 117470
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.157151603275426e-05,
+      "loss": 1.2856,
+      "step": 117480
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.157036810285451e-05,
+      "loss": 1.3984,
+      "step": 117490
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.156922017295477e-05,
+      "loss": 1.1969,
+      "step": 117500
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.156807224305502e-05,
+      "loss": 1.4149,
+      "step": 117510
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.156692431315527e-05,
+      "loss": 1.2227,
+      "step": 117520
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.156577638325552e-05,
+      "loss": 1.2415,
+      "step": 117530
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.156462845335577e-05,
+      "loss": 1.3641,
+      "step": 117540
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.156348052345603e-05,
+      "loss": 1.3253,
+      "step": 117550
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.156233259355628e-05,
+      "loss": 1.3899,
+      "step": 117560
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.156118466365653e-05,
+      "loss": 1.3621,
+      "step": 117570
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 6.156003673375678e-05,
+      "loss": 1.3177,
+      "step": 117580
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.155888880385705e-05,
+      "loss": 1.2896,
+      "step": 117590
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.155774087395728e-05,
+      "loss": 1.3033,
+      "step": 117600
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.155659294405755e-05,
+      "loss": 1.308,
+      "step": 117610
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.15554450141578e-05,
+      "loss": 1.2488,
+      "step": 117620
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.155429708425805e-05,
+      "loss": 1.2026,
+      "step": 117630
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.15531491543583e-05,
+      "loss": 1.1981,
+      "step": 117640
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.155200122445856e-05,
+      "loss": 1.1667,
+      "step": 117650
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.155085329455881e-05,
+      "loss": 1.2571,
+      "step": 117660
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.154970536465906e-05,
+      "loss": 1.2778,
+      "step": 117670
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.154855743475931e-05,
+      "loss": 1.2953,
+      "step": 117680
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.154740950485956e-05,
+      "loss": 1.3572,
+      "step": 117690
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.154626157495982e-05,
+      "loss": 1.3446,
+      "step": 117700
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.154511364506007e-05,
+      "loss": 1.3437,
+      "step": 117710
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.154396571516032e-05,
+      "loss": 1.1891,
+      "step": 117720
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.154281778526057e-05,
+      "loss": 1.2643,
+      "step": 117730
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.154166985536084e-05,
+      "loss": 1.2257,
+      "step": 117740
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.154052192546107e-05,
+      "loss": 1.25,
+      "step": 117750
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.153937399556134e-05,
+      "loss": 1.2008,
+      "step": 117760
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.153822606566159e-05,
+      "loss": 1.2754,
+      "step": 117770
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.153707813576183e-05,
+      "loss": 1.2069,
+      "step": 117780
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.153593020586208e-05,
+      "loss": 1.243,
+      "step": 117790
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 6.153478227596235e-05,
+      "loss": 1.3651,
+      "step": 117800
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.15336343460626e-05,
+      "loss": 1.3058,
+      "step": 117810
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.153248641616285e-05,
+      "loss": 1.2871,
+      "step": 117820
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.15313384862631e-05,
+      "loss": 1.3207,
+      "step": 117830
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.153019055636335e-05,
+      "loss": 1.2046,
+      "step": 117840
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.152904262646361e-05,
+      "loss": 1.1819,
+      "step": 117850
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.152789469656385e-05,
+      "loss": 1.2309,
+      "step": 117860
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.152674676666411e-05,
+      "loss": 1.1706,
+      "step": 117870
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.152559883676436e-05,
+      "loss": 1.2424,
+      "step": 117880
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.152445090686461e-05,
+      "loss": 1.3022,
+      "step": 117890
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.152330297696486e-05,
+      "loss": 1.302,
+      "step": 117900
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.152215504706512e-05,
+      "loss": 1.1845,
+      "step": 117910
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.152100711716537e-05,
+      "loss": 1.2711,
+      "step": 117920
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.151985918726562e-05,
+      "loss": 1.2819,
+      "step": 117930
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.151871125736587e-05,
+      "loss": 1.1803,
+      "step": 117940
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.151756332746612e-05,
+      "loss": 1.3418,
+      "step": 117950
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.151641539756639e-05,
+      "loss": 1.1505,
+      "step": 117960
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.151526746766664e-05,
+      "loss": 1.1663,
+      "step": 117970
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.151411953776689e-05,
+      "loss": 1.2024,
+      "step": 117980
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.151297160786714e-05,
+      "loss": 1.4707,
+      "step": 117990
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.15118236779674e-05,
+      "loss": 1.3092,
+      "step": 118000
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.151067574806764e-05,
+      "loss": 1.2208,
+      "step": 118010
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 6.15095278181679e-05,
+      "loss": 1.2126,
+      "step": 118020
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.150837988826815e-05,
+      "loss": 1.2913,
+      "step": 118030
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.15072319583684e-05,
+      "loss": 1.1835,
+      "step": 118040
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.150608402846865e-05,
+      "loss": 1.259,
+      "step": 118050
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.150493609856891e-05,
+      "loss": 1.2337,
+      "step": 118060
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.150378816866916e-05,
+      "loss": 1.2469,
+      "step": 118070
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.150264023876941e-05,
+      "loss": 1.1616,
+      "step": 118080
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.150149230886966e-05,
+      "loss": 1.2148,
+      "step": 118090
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.150034437896991e-05,
+      "loss": 1.261,
+      "step": 118100
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.149919644907017e-05,
+      "loss": 1.3555,
+      "step": 118110
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.149804851917042e-05,
+      "loss": 1.1749,
+      "step": 118120
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.149690058927067e-05,
+      "loss": 1.2852,
+      "step": 118130
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.149575265937092e-05,
+      "loss": 1.3398,
+      "step": 118140
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.149460472947119e-05,
+      "loss": 1.3022,
+      "step": 118150
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.149345679957142e-05,
+      "loss": 1.2907,
+      "step": 118160
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.149230886967169e-05,
+      "loss": 1.2745,
+      "step": 118170
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.149116093977194e-05,
+      "loss": 1.0838,
+      "step": 118180
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.149001300987219e-05,
+      "loss": 1.2535,
+      "step": 118190
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.148886507997244e-05,
+      "loss": 1.2331,
+      "step": 118200
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.14877171500727e-05,
+      "loss": 1.2514,
+      "step": 118210
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.148656922017295e-05,
+      "loss": 1.203,
+      "step": 118220
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 6.14854212902732e-05,
+      "loss": 1.2194,
+      "step": 118230
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.148427336037345e-05,
+      "loss": 1.3384,
+      "step": 118240
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.14831254304737e-05,
+      "loss": 1.2749,
+      "step": 118250
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.148197750057396e-05,
+      "loss": 1.3119,
+      "step": 118260
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.148082957067421e-05,
+      "loss": 1.3255,
+      "step": 118270
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.147968164077446e-05,
+      "loss": 1.2149,
+      "step": 118280
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.147853371087471e-05,
+      "loss": 1.2312,
+      "step": 118290
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.147738578097498e-05,
+      "loss": 1.1324,
+      "step": 118300
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.147623785107521e-05,
+      "loss": 1.2702,
+      "step": 118310
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.147508992117548e-05,
+      "loss": 1.2005,
+      "step": 118320
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.147394199127573e-05,
+      "loss": 1.248,
+      "step": 118330
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.147279406137598e-05,
+      "loss": 1.2511,
+      "step": 118340
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.147164613147623e-05,
+      "loss": 1.2446,
+      "step": 118350
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.147049820157649e-05,
+      "loss": 1.2872,
+      "step": 118360
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.146935027167674e-05,
+      "loss": 1.3358,
+      "step": 118370
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.146820234177699e-05,
+      "loss": 1.1648,
+      "step": 118380
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.146705441187724e-05,
+      "loss": 1.1721,
+      "step": 118390
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.146590648197749e-05,
+      "loss": 1.2268,
+      "step": 118400
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.146475855207775e-05,
+      "loss": 1.2051,
+      "step": 118410
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.1463610622178e-05,
+      "loss": 1.2385,
+      "step": 118420
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.146246269227825e-05,
+      "loss": 1.2746,
+      "step": 118430
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.14613147623785e-05,
+      "loss": 1.1465,
+      "step": 118440
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 6.146016683247876e-05,
+      "loss": 1.3729,
+      "step": 118450
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.1459018902579e-05,
+      "loss": 1.3002,
+      "step": 118460
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.145787097267926e-05,
+      "loss": 1.1391,
+      "step": 118470
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.145672304277951e-05,
+      "loss": 1.2221,
+      "step": 118480
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.145557511287976e-05,
+      "loss": 1.3149,
+      "step": 118490
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.145442718298001e-05,
+      "loss": 1.1875,
+      "step": 118500
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.145327925308028e-05,
+      "loss": 1.1605,
+      "step": 118510
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.145213132318053e-05,
+      "loss": 1.3102,
+      "step": 118520
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.145098339328078e-05,
+      "loss": 1.2245,
+      "step": 118530
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.144983546338103e-05,
+      "loss": 1.228,
+      "step": 118540
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.144868753348128e-05,
+      "loss": 1.3225,
+      "step": 118550
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.144753960358154e-05,
+      "loss": 1.2649,
+      "step": 118560
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.144639167368179e-05,
+      "loss": 1.3682,
+      "step": 118570
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.144524374378204e-05,
+      "loss": 1.1689,
+      "step": 118580
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.144409581388229e-05,
+      "loss": 1.2276,
+      "step": 118590
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.144294788398255e-05,
+      "loss": 1.1389,
+      "step": 118600
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.144179995408279e-05,
+      "loss": 1.2473,
+      "step": 118610
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.144065202418305e-05,
+      "loss": 1.2557,
+      "step": 118620
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.14395040942833e-05,
+      "loss": 1.1365,
+      "step": 118630
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.143835616438355e-05,
+      "loss": 1.3279,
+      "step": 118640
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.14372082344838e-05,
+      "loss": 1.1567,
+      "step": 118650
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.143606030458407e-05,
+      "loss": 1.1663,
+      "step": 118660
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 6.143491237468432e-05,
+      "loss": 1.2304,
+      "step": 118670
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.143376444478457e-05,
+      "loss": 1.3074,
+      "step": 118680
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.143261651488482e-05,
+      "loss": 1.2122,
+      "step": 118690
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.143146858498507e-05,
+      "loss": 1.338,
+      "step": 118700
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.143032065508533e-05,
+      "loss": 1.3964,
+      "step": 118710
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.142917272518558e-05,
+      "loss": 1.2978,
+      "step": 118720
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.142802479528583e-05,
+      "loss": 1.3013,
+      "step": 118730
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.142687686538608e-05,
+      "loss": 1.3363,
+      "step": 118740
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.142572893548634e-05,
+      "loss": 1.2595,
+      "step": 118750
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.142458100558658e-05,
+      "loss": 1.1897,
+      "step": 118760
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.142343307568684e-05,
+      "loss": 1.2578,
+      "step": 118770
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.142228514578709e-05,
+      "loss": 1.266,
+      "step": 118780
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.142113721588734e-05,
+      "loss": 1.3831,
+      "step": 118790
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.141998928598759e-05,
+      "loss": 1.3656,
+      "step": 118800
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.141884135608785e-05,
+      "loss": 1.2753,
+      "step": 118810
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.14176934261881e-05,
+      "loss": 1.2059,
+      "step": 118820
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.141654549628835e-05,
+      "loss": 1.238,
+      "step": 118830
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.14153975663886e-05,
+      "loss": 1.2936,
+      "step": 118840
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.141424963648885e-05,
+      "loss": 1.225,
+      "step": 118850
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.141310170658912e-05,
+      "loss": 1.2021,
+      "step": 118860
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.141195377668937e-05,
+      "loss": 1.196,
+      "step": 118870
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.141080584678962e-05,
+      "loss": 1.2293,
+      "step": 118880
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 6.140965791688987e-05,
+      "loss": 1.2125,
+      "step": 118890
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.140850998699013e-05,
+      "loss": 1.2707,
+      "step": 118900
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.140736205709037e-05,
+      "loss": 1.2799,
+      "step": 118910
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.140621412719063e-05,
+      "loss": 1.3738,
+      "step": 118920
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.140506619729088e-05,
+      "loss": 1.2945,
+      "step": 118930
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.140391826739113e-05,
+      "loss": 1.169,
+      "step": 118940
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.140277033749138e-05,
+      "loss": 1.2272,
+      "step": 118950
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.140162240759164e-05,
+      "loss": 1.1675,
+      "step": 118960
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.140047447769189e-05,
+      "loss": 1.2471,
+      "step": 118970
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.139932654779214e-05,
+      "loss": 1.2634,
+      "step": 118980
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.139817861789239e-05,
+      "loss": 1.2334,
+      "step": 118990
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.139703068799264e-05,
+      "loss": 1.2587,
+      "step": 119000
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.13958827580929e-05,
+      "loss": 1.2186,
+      "step": 119010
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.139473482819316e-05,
+      "loss": 1.114,
+      "step": 119020
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.13935868982934e-05,
+      "loss": 1.1488,
+      "step": 119030
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.139243896839366e-05,
+      "loss": 1.5065,
+      "step": 119040
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.139129103849392e-05,
+      "loss": 1.2729,
+      "step": 119050
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.139014310859416e-05,
+      "loss": 1.3975,
+      "step": 119060
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.138899517869442e-05,
+      "loss": 1.3421,
+      "step": 119070
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.138784724879467e-05,
+      "loss": 1.2199,
+      "step": 119080
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.138669931889492e-05,
+      "loss": 1.2132,
+      "step": 119090
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.138555138899517e-05,
+      "loss": 1.2777,
+      "step": 119100
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 6.138440345909543e-05,
+      "loss": 1.3445,
+      "step": 119110
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.138325552919568e-05,
+      "loss": 1.2447,
+      "step": 119120
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.138210759929593e-05,
+      "loss": 1.2375,
+      "step": 119130
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.138095966939618e-05,
+      "loss": 1.2263,
+      "step": 119140
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.137981173949643e-05,
+      "loss": 1.1842,
+      "step": 119150
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.13786638095967e-05,
+      "loss": 1.2196,
+      "step": 119160
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.137751587969694e-05,
+      "loss": 1.2606,
+      "step": 119170
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.13763679497972e-05,
+      "loss": 1.2369,
+      "step": 119180
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.137522001989744e-05,
+      "loss": 1.2242,
+      "step": 119190
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.137407208999771e-05,
+      "loss": 1.1796,
+      "step": 119200
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.137292416009794e-05,
+      "loss": 1.4021,
+      "step": 119210
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.137177623019821e-05,
+      "loss": 1.4366,
+      "step": 119220
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.137062830029846e-05,
+      "loss": 1.2724,
+      "step": 119230
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.13694803703987e-05,
+      "loss": 1.1595,
+      "step": 119240
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.136833244049896e-05,
+      "loss": 1.4065,
+      "step": 119250
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.13671845105992e-05,
+      "loss": 1.3377,
+      "step": 119260
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.136603658069947e-05,
+      "loss": 1.0984,
+      "step": 119270
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.136488865079972e-05,
+      "loss": 1.1928,
+      "step": 119280
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.136374072089997e-05,
+      "loss": 1.3269,
+      "step": 119290
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.136259279100022e-05,
+      "loss": 1.2544,
+      "step": 119300
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.136144486110048e-05,
+      "loss": 1.2373,
+      "step": 119310
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 6.136029693120072e-05,
+      "loss": 1.2864,
+      "step": 119320
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.135914900130098e-05,
+      "loss": 1.125,
+      "step": 119330
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.135800107140123e-05,
+      "loss": 1.4838,
+      "step": 119340
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.135685314150148e-05,
+      "loss": 1.0905,
+      "step": 119350
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.135570521160173e-05,
+      "loss": 1.1288,
+      "step": 119360
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.1354557281702e-05,
+      "loss": 1.1967,
+      "step": 119370
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.135340935180225e-05,
+      "loss": 1.249,
+      "step": 119380
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.13522614219025e-05,
+      "loss": 1.1899,
+      "step": 119390
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.135111349200274e-05,
+      "loss": 1.1964,
+      "step": 119400
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.1349965562103e-05,
+      "loss": 1.341,
+      "step": 119410
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.134881763220326e-05,
+      "loss": 1.2468,
+      "step": 119420
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.134766970230351e-05,
+      "loss": 1.237,
+      "step": 119430
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.134652177240376e-05,
+      "loss": 1.115,
+      "step": 119440
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.134537384250401e-05,
+      "loss": 1.2818,
+      "step": 119450
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.134422591260427e-05,
+      "loss": 1.3234,
+      "step": 119460
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.134307798270451e-05,
+      "loss": 1.3665,
+      "step": 119470
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.134193005280477e-05,
+      "loss": 1.3386,
+      "step": 119480
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.134078212290502e-05,
+      "loss": 1.1802,
+      "step": 119490
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.133963419300527e-05,
+      "loss": 1.2359,
+      "step": 119500
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.133848626310552e-05,
+      "loss": 1.2002,
+      "step": 119510
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.133733833320578e-05,
+      "loss": 1.2962,
+      "step": 119520
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.133619040330603e-05,
+      "loss": 1.2603,
+      "step": 119530
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 6.133504247340628e-05,
+      "loss": 1.3382,
+      "step": 119540
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.133389454350653e-05,
+      "loss": 1.1374,
+      "step": 119550
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.133274661360678e-05,
+      "loss": 1.1614,
+      "step": 119560
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.133159868370705e-05,
+      "loss": 1.2585,
+      "step": 119570
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.13304507538073e-05,
+      "loss": 1.3117,
+      "step": 119580
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.132930282390755e-05,
+      "loss": 1.2376,
+      "step": 119590
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.13281548940078e-05,
+      "loss": 1.2153,
+      "step": 119600
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.132700696410806e-05,
+      "loss": 1.1731,
+      "step": 119610
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.13258590342083e-05,
+      "loss": 1.1228,
+      "step": 119620
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.132471110430856e-05,
+      "loss": 1.4252,
+      "step": 119630
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.132356317440881e-05,
+      "loss": 1.2631,
+      "step": 119640
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.132241524450906e-05,
+      "loss": 1.0678,
+      "step": 119650
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.132126731460931e-05,
+      "loss": 1.286,
+      "step": 119660
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.132011938470957e-05,
+      "loss": 1.3343,
+      "step": 119670
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.131897145480982e-05,
+      "loss": 1.201,
+      "step": 119680
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.131782352491007e-05,
+      "loss": 1.2867,
+      "step": 119690
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.131667559501032e-05,
+      "loss": 1.1635,
+      "step": 119700
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.131552766511057e-05,
+      "loss": 1.1691,
+      "step": 119710
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.131437973521084e-05,
+      "loss": 1.2961,
+      "step": 119720
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.131323180531109e-05,
+      "loss": 1.1847,
+      "step": 119730
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.131208387541133e-05,
+      "loss": 1.3762,
+      "step": 119740
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.131093594551158e-05,
+      "loss": 1.2429,
+      "step": 119750
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 6.130978801561185e-05,
+      "loss": 1.1827,
+      "step": 119760
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.130864008571208e-05,
+      "loss": 1.3232,
+      "step": 119770
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.130749215581235e-05,
+      "loss": 1.3344,
+      "step": 119780
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.13063442259126e-05,
+      "loss": 1.353,
+      "step": 119790
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.130519629601285e-05,
+      "loss": 1.1588,
+      "step": 119800
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.13040483661131e-05,
+      "loss": 1.2161,
+      "step": 119810
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.130290043621336e-05,
+      "loss": 1.3559,
+      "step": 119820
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.130175250631361e-05,
+      "loss": 1.1035,
+      "step": 119830
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.130060457641386e-05,
+      "loss": 1.1932,
+      "step": 119840
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.129945664651411e-05,
+      "loss": 1.1949,
+      "step": 119850
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.129830871661436e-05,
+      "loss": 1.3427,
+      "step": 119860
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.129716078671462e-05,
+      "loss": 1.2723,
+      "step": 119870
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.129601285681487e-05,
+      "loss": 1.2471,
+      "step": 119880
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.129486492691512e-05,
+      "loss": 1.307,
+      "step": 119890
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.129371699701537e-05,
+      "loss": 1.344,
+      "step": 119900
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.129256906711564e-05,
+      "loss": 1.2935,
+      "step": 119910
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.129142113721587e-05,
+      "loss": 1.244,
+      "step": 119920
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.129027320731614e-05,
+      "loss": 1.2087,
+      "step": 119930
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.128912527741639e-05,
+      "loss": 1.3164,
+      "step": 119940
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.128797734751664e-05,
+      "loss": 1.3004,
+      "step": 119950
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.128682941761689e-05,
+      "loss": 1.2122,
+      "step": 119960
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.128568148771715e-05,
+      "loss": 1.238,
+      "step": 119970
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 6.12845335578174e-05,
+      "loss": 1.2611,
+      "step": 119980
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.128338562791765e-05,
+      "loss": 1.1411,
+      "step": 119990
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.12822376980179e-05,
+      "loss": 1.2625,
+      "step": 120000
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.128108976811815e-05,
+      "loss": 1.3652,
+      "step": 120010
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.127994183821841e-05,
+      "loss": 1.254,
+      "step": 120020
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.127879390831866e-05,
+      "loss": 1.1782,
+      "step": 120030
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.127764597841891e-05,
+      "loss": 1.2313,
+      "step": 120040
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.127649804851916e-05,
+      "loss": 1.2863,
+      "step": 120050
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.127535011861943e-05,
+      "loss": 1.3956,
+      "step": 120060
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.127420218871966e-05,
+      "loss": 1.3394,
+      "step": 120070
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.127305425881992e-05,
+      "loss": 1.274,
+      "step": 120080
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.127190632892017e-05,
+      "loss": 1.2413,
+      "step": 120090
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.127075839902042e-05,
+      "loss": 1.236,
+      "step": 120100
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.126961046912067e-05,
+      "loss": 1.2972,
+      "step": 120110
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.126846253922094e-05,
+      "loss": 1.2137,
+      "step": 120120
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.126731460932119e-05,
+      "loss": 1.29,
+      "step": 120130
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.126616667942144e-05,
+      "loss": 1.372,
+      "step": 120140
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.126501874952169e-05,
+      "loss": 1.3069,
+      "step": 120150
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.126387081962194e-05,
+      "loss": 1.3261,
+      "step": 120160
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.12627228897222e-05,
+      "loss": 1.276,
+      "step": 120170
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.126157495982245e-05,
+      "loss": 1.2393,
+      "step": 120180
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.12604270299227e-05,
+      "loss": 1.3076,
+      "step": 120190
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 6.125927910002295e-05,
+      "loss": 1.2149,
+      "step": 120200
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 6.125813117012321e-05,
+      "loss": 1.2371,
+      "step": 120210
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 6.125698324022345e-05,
+      "loss": 1.1809,
+      "step": 120220
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 6.125583531032371e-05,
+      "loss": 1.1842,
+      "step": 120230
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 6.125468738042396e-05,
+      "loss": 1.3114,
+      "step": 120240
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 6.125353945052421e-05,
+      "loss": 1.1463,
+      "step": 120250
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 6.125239152062446e-05,
+      "loss": 1.2514,
+      "step": 120260
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 6.125124359072473e-05,
+      "loss": 1.2964,
+      "step": 120270
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 6.125009566082498e-05,
+      "loss": 1.2388,
+      "step": 120280
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 6.124894773092523e-05,
+      "loss": 1.2969,
+      "step": 120290
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 6.124779980102548e-05,
+      "loss": 1.345,
+      "step": 120300
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 6.124665187112573e-05,
+      "loss": 1.2415,
+      "step": 120310
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 6.124550394122599e-05,
+      "loss": 1.3169,
+      "step": 120320
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 6.124435601132624e-05,
+      "loss": 1.1418,
+      "step": 120330
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 6.124320808142649e-05,
+      "loss": 1.3195,
+      "step": 120340
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 6.124206015152674e-05,
+      "loss": 1.1591,
+      "step": 120350
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 6.1240912221627e-05,
+      "loss": 1.1581,
+      "step": 120360
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 6.123976429172724e-05,
+      "loss": 1.4207,
+      "step": 120370
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 6.12386163618275e-05,
+      "loss": 1.2361,
+      "step": 120380
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 6.123746843192775e-05,
+      "loss": 1.317,
+      "step": 120390
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 6.1236320502028e-05,
+      "loss": 1.368,
+      "step": 120400
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 6.123517257212825e-05,
+      "loss": 1.303,
+      "step": 120410
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 6.123402464222851e-05,
+      "loss": 1.3137,
+      "step": 120420
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 6.123287671232876e-05,
+      "loss": 1.2523,
+      "step": 120430
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 6.123172878242901e-05,
+      "loss": 1.099,
+      "step": 120440
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 6.123058085252926e-05,
+      "loss": 1.2836,
+      "step": 120450
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 6.122943292262951e-05,
+      "loss": 1.6075,
+      "step": 120460
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 6.122828499272978e-05,
+      "loss": 1.3481,
+      "step": 120470
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 6.122713706283003e-05,
+      "loss": 1.2207,
+      "step": 120480
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 6.122598913293028e-05,
+      "loss": 1.3611,
+      "step": 120490
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 6.122484120303053e-05,
+      "loss": 1.1927,
+      "step": 120500
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 6.122369327313079e-05,
+      "loss": 1.2243,
+      "step": 120510
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 6.122254534323103e-05,
+      "loss": 1.2574,
+      "step": 120520
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 6.122139741333129e-05,
+      "loss": 1.3317,
+      "step": 120530
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 6.122024948343154e-05,
+      "loss": 1.18,
+      "step": 120540
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 6.121910155353179e-05,
+      "loss": 1.2479,
+      "step": 120550
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 6.121795362363204e-05,
+      "loss": 1.1472,
+      "step": 120560
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 6.12168056937323e-05,
+      "loss": 1.2002,
+      "step": 120570
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 6.121565776383255e-05,
+      "loss": 1.2759,
+      "step": 120580
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 6.12145098339328e-05,
+      "loss": 1.1354,
+      "step": 120590
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 6.121336190403305e-05,
+      "loss": 1.0822,
+      "step": 120600
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 6.12122139741333e-05,
+      "loss": 1.142,
+      "step": 120610
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 6.121106604423357e-05,
+      "loss": 1.2728,
+      "step": 120620
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 6.12099181143338e-05,
+      "loss": 1.1919,
+      "step": 120630
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.120877018443407e-05,
+      "loss": 1.3743,
+      "step": 120640
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.120762225453432e-05,
+      "loss": 1.3016,
+      "step": 120650
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.120647432463457e-05,
+      "loss": 1.1414,
+      "step": 120660
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.120532639473482e-05,
+      "loss": 1.1804,
+      "step": 120670
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.120417846483508e-05,
+      "loss": 1.1394,
+      "step": 120680
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.120303053493533e-05,
+      "loss": 1.2168,
+      "step": 120690
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.120188260503558e-05,
+      "loss": 1.2989,
+      "step": 120700
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.120073467513583e-05,
+      "loss": 1.199,
+      "step": 120710
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.119958674523608e-05,
+      "loss": 1.2219,
+      "step": 120720
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.119843881533634e-05,
+      "loss": 1.2955,
+      "step": 120730
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.119729088543659e-05,
+      "loss": 1.2249,
+      "step": 120740
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.119614295553684e-05,
+      "loss": 1.3916,
+      "step": 120750
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.119499502563709e-05,
+      "loss": 1.2276,
+      "step": 120760
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.119384709573735e-05,
+      "loss": 1.2641,
+      "step": 120770
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.119269916583759e-05,
+      "loss": 1.1774,
+      "step": 120780
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.119155123593785e-05,
+      "loss": 1.203,
+      "step": 120790
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.11904033060381e-05,
+      "loss": 1.3348,
+      "step": 120800
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.118925537613835e-05,
+      "loss": 1.1606,
+      "step": 120810
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.11881074462386e-05,
+      "loss": 1.2345,
+      "step": 120820
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.118695951633887e-05,
+      "loss": 1.3027,
+      "step": 120830
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.118581158643912e-05,
+      "loss": 1.291,
+      "step": 120840
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 6.118466365653937e-05,
+      "loss": 1.2948,
+      "step": 120850
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.118351572663962e-05,
+      "loss": 1.1669,
+      "step": 120860
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.118236779673987e-05,
+      "loss": 1.2561,
+      "step": 120870
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.118121986684013e-05,
+      "loss": 1.1823,
+      "step": 120880
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.118007193694038e-05,
+      "loss": 1.1231,
+      "step": 120890
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.117892400704063e-05,
+      "loss": 1.3863,
+      "step": 120900
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.117777607714088e-05,
+      "loss": 1.2471,
+      "step": 120910
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.117662814724114e-05,
+      "loss": 1.3041,
+      "step": 120920
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.117548021734138e-05,
+      "loss": 1.2465,
+      "step": 120930
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.117433228744164e-05,
+      "loss": 1.4307,
+      "step": 120940
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.117318435754189e-05,
+      "loss": 1.264,
+      "step": 120950
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.117203642764214e-05,
+      "loss": 1.284,
+      "step": 120960
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.117088849774239e-05,
+      "loss": 1.1803,
+      "step": 120970
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.116974056784266e-05,
+      "loss": 1.2734,
+      "step": 120980
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.11685926379429e-05,
+      "loss": 1.1563,
+      "step": 120990
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.116744470804316e-05,
+      "loss": 1.2508,
+      "step": 121000
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.11662967781434e-05,
+      "loss": 1.2695,
+      "step": 121010
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.116514884824366e-05,
+      "loss": 1.3154,
+      "step": 121020
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.116400091834392e-05,
+      "loss": 1.4869,
+      "step": 121030
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.116285298844417e-05,
+      "loss": 1.3615,
+      "step": 121040
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.116170505854442e-05,
+      "loss": 1.2126,
+      "step": 121050
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.116055712864467e-05,
+      "loss": 1.1972,
+      "step": 121060
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 6.115940919874493e-05,
+      "loss": 1.1974,
+      "step": 121070
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.115826126884517e-05,
+      "loss": 1.3864,
+      "step": 121080
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.115711333894543e-05,
+      "loss": 1.2939,
+      "step": 121090
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.115596540904568e-05,
+      "loss": 1.3539,
+      "step": 121100
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.115481747914593e-05,
+      "loss": 1.1374,
+      "step": 121110
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.115366954924618e-05,
+      "loss": 1.1898,
+      "step": 121120
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.115252161934644e-05,
+      "loss": 1.4164,
+      "step": 121130
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.11513736894467e-05,
+      "loss": 1.3847,
+      "step": 121140
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.115022575954694e-05,
+      "loss": 1.1513,
+      "step": 121150
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.11490778296472e-05,
+      "loss": 1.2045,
+      "step": 121160
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.114792989974744e-05,
+      "loss": 1.3608,
+      "step": 121170
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.114678196984771e-05,
+      "loss": 1.2199,
+      "step": 121180
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.114563403994796e-05,
+      "loss": 1.2735,
+      "step": 121190
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.11444861100482e-05,
+      "loss": 1.2078,
+      "step": 121200
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.114333818014846e-05,
+      "loss": 1.2933,
+      "step": 121210
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.114219025024872e-05,
+      "loss": 1.2498,
+      "step": 121220
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.114104232034896e-05,
+      "loss": 1.2942,
+      "step": 121230
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.113989439044922e-05,
+      "loss": 1.3075,
+      "step": 121240
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.113874646054947e-05,
+      "loss": 1.2451,
+      "step": 121250
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.113759853064972e-05,
+      "loss": 1.2487,
+      "step": 121260
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.113645060074997e-05,
+      "loss": 1.4184,
+      "step": 121270
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.113530267085023e-05,
+      "loss": 1.24,
+      "step": 121280
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 6.113415474095048e-05,
+      "loss": 1.2444,
+      "step": 121290
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.113300681105073e-05,
+      "loss": 1.1498,
+      "step": 121300
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.113185888115098e-05,
+      "loss": 1.2039,
+      "step": 121310
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.113071095125123e-05,
+      "loss": 1.2333,
+      "step": 121320
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.11295630213515e-05,
+      "loss": 1.3481,
+      "step": 121330
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.112841509145175e-05,
+      "loss": 1.2443,
+      "step": 121340
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.1127267161552e-05,
+      "loss": 1.2243,
+      "step": 121350
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.112611923165224e-05,
+      "loss": 1.286,
+      "step": 121360
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.112497130175251e-05,
+      "loss": 1.3241,
+      "step": 121370
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.112382337185274e-05,
+      "loss": 1.1434,
+      "step": 121380
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.112267544195301e-05,
+      "loss": 1.2393,
+      "step": 121390
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.112152751205326e-05,
+      "loss": 1.2987,
+      "step": 121400
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.112037958215351e-05,
+      "loss": 1.2962,
+      "step": 121410
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.111923165225376e-05,
+      "loss": 1.3036,
+      "step": 121420
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.111808372235402e-05,
+      "loss": 1.2502,
+      "step": 121430
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.111693579245427e-05,
+      "loss": 1.337,
+      "step": 121440
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.111578786255452e-05,
+      "loss": 1.33,
+      "step": 121450
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.111463993265477e-05,
+      "loss": 1.2844,
+      "step": 121460
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.111349200275502e-05,
+      "loss": 1.2142,
+      "step": 121470
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.111234407285528e-05,
+      "loss": 1.2056,
+      "step": 121480
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.111119614295553e-05,
+      "loss": 1.2237,
+      "step": 121490
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 6.111004821305578e-05,
+      "loss": 1.2753,
+      "step": 121500
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 6.110890028315603e-05,
+      "loss": 1.5313,
+      "step": 121510
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 6.11077523532563e-05,
+      "loss": 1.2892,
+      "step": 121520
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 6.110660442335653e-05,
+      "loss": 1.3324,
+      "step": 121530
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 6.11054564934568e-05,
+      "loss": 1.2424,
+      "step": 121540
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 6.110430856355705e-05,
+      "loss": 1.2683,
+      "step": 121550
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 6.11031606336573e-05,
+      "loss": 1.4517,
+      "step": 121560
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 6.110201270375755e-05,
+      "loss": 1.3118,
+      "step": 121570
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 6.110086477385781e-05,
+      "loss": 1.2935,
+      "step": 121580
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 6.109971684395806e-05,
+      "loss": 1.3842,
+      "step": 121590
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 6.109856891405831e-05,
+      "loss": 1.2796,
+      "step": 121600
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 6.109742098415856e-05,
+      "loss": 1.2709,
+      "step": 121610
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 6.109627305425881e-05,
+      "loss": 1.1711,
+      "step": 121620
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 6.109512512435907e-05,
+      "loss": 1.2949,
+      "step": 121630
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 6.109397719445932e-05,
+      "loss": 1.3616,
+      "step": 121640
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 6.109282926455957e-05,
+      "loss": 1.2596,
+      "step": 121650
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 6.109168133465982e-05,
+      "loss": 1.4412,
+      "step": 121660
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 6.109053340476009e-05,
+      "loss": 1.2541,
+      "step": 121670
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 6.108938547486032e-05,
+      "loss": 1.3744,
+      "step": 121680
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 6.108823754496058e-05,
+      "loss": 1.2604,
+      "step": 121690
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 6.108708961506083e-05,
+      "loss": 1.2211,
+      "step": 121700
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 6.108594168516108e-05,
+      "loss": 1.1703,
+      "step": 121710
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 6.108479375526133e-05,
+      "loss": 1.3641,
+      "step": 121720
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.10836458253616e-05,
+      "loss": 1.3748,
+      "step": 121730
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.108249789546185e-05,
+      "loss": 1.4157,
+      "step": 121740
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.10813499655621e-05,
+      "loss": 1.1825,
+      "step": 121750
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.108020203566235e-05,
+      "loss": 1.3754,
+      "step": 121760
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.10790541057626e-05,
+      "loss": 1.2829,
+      "step": 121770
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.107790617586286e-05,
+      "loss": 1.1932,
+      "step": 121780
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.107675824596311e-05,
+      "loss": 1.2147,
+      "step": 121790
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.107561031606336e-05,
+      "loss": 1.1584,
+      "step": 121800
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.107446238616361e-05,
+      "loss": 1.3599,
+      "step": 121810
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.107331445626387e-05,
+      "loss": 1.2725,
+      "step": 121820
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.107216652636411e-05,
+      "loss": 1.2876,
+      "step": 121830
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.107101859646437e-05,
+      "loss": 1.2339,
+      "step": 121840
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.106987066656462e-05,
+      "loss": 1.1715,
+      "step": 121850
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.106872273666487e-05,
+      "loss": 1.3673,
+      "step": 121860
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.106757480676512e-05,
+      "loss": 1.1634,
+      "step": 121870
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.106642687686539e-05,
+      "loss": 1.255,
+      "step": 121880
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.106527894696564e-05,
+      "loss": 1.2454,
+      "step": 121890
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.106413101706589e-05,
+      "loss": 1.2189,
+      "step": 121900
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.106298308716614e-05,
+      "loss": 1.2693,
+      "step": 121910
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.106183515726639e-05,
+      "loss": 1.161,
+      "step": 121920
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.106068722736665e-05,
+      "loss": 1.3349,
+      "step": 121930
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 6.10595392974669e-05,
+      "loss": 1.1862,
+      "step": 121940
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 6.105839136756715e-05,
+      "loss": 1.2752,
+      "step": 121950
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 6.10572434376674e-05,
+      "loss": 1.2229,
+      "step": 121960
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 6.105609550776766e-05,
+      "loss": 1.2312,
+      "step": 121970
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 6.10549475778679e-05,
+      "loss": 1.3384,
+      "step": 121980
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 6.105379964796816e-05,
+      "loss": 1.18,
+      "step": 121990
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 6.105265171806841e-05,
+      "loss": 1.3945,
+      "step": 122000
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 6.105150378816866e-05,
+      "loss": 1.3428,
+      "step": 122010
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 6.105035585826891e-05,
+      "loss": 1.1612,
+      "step": 122020
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 6.104920792836916e-05,
+      "loss": 1.3077,
+      "step": 122030
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 6.104805999846942e-05,
+      "loss": 1.1706,
+      "step": 122040
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 6.104691206856967e-05,
+      "loss": 1.1608,
+      "step": 122050
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 6.104576413866992e-05,
+      "loss": 1.432,
+      "step": 122060
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 6.104461620877017e-05,
+      "loss": 1.2932,
+      "step": 122070
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 6.104346827887044e-05,
+      "loss": 1.199,
+      "step": 122080
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 6.104232034897067e-05,
+      "loss": 1.2401,
+      "step": 122090
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 6.104117241907094e-05,
+      "loss": 1.1643,
+      "step": 122100
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 6.104002448917119e-05,
+      "loss": 1.2607,
+      "step": 122110
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 6.103887655927144e-05,
+      "loss": 1.2859,
+      "step": 122120
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 6.103772862937169e-05,
+      "loss": 1.2794,
+      "step": 122130
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 6.103658069947195e-05,
+      "loss": 1.1964,
+      "step": 122140
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 6.10354327695722e-05,
+      "loss": 1.1996,
+      "step": 122150
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 6.103428483967245e-05,
+      "loss": 1.3295,
+      "step": 122160
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 6.103313690977271e-05,
+      "loss": 1.2314,
+      "step": 122170
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 6.1031988979872957e-05,
+      "loss": 1.2951,
+      "step": 122180
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 6.1030841049973206e-05,
+      "loss": 1.2152,
+      "step": 122190
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 6.102969312007346e-05,
+      "loss": 1.2097,
+      "step": 122200
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 6.102854519017371e-05,
+      "loss": 1.3108,
+      "step": 122210
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 6.102739726027397e-05,
+      "loss": 1.1798,
+      "step": 122220
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 6.102624933037422e-05,
+      "loss": 1.3042,
+      "step": 122230
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 6.102510140047447e-05,
+      "loss": 1.2427,
+      "step": 122240
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 6.1023953470574726e-05,
+      "loss": 1.3267,
+      "step": 122250
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 6.102280554067498e-05,
+      "loss": 1.2978,
+      "step": 122260
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 6.1021657610775226e-05,
+      "loss": 1.3607,
+      "step": 122270
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 6.102050968087548e-05,
+      "loss": 1.3429,
+      "step": 122280
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 6.101936175097574e-05,
+      "loss": 1.2835,
+      "step": 122290
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 6.101821382107598e-05,
+      "loss": 1.4389,
+      "step": 122300
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 6.101706589117624e-05,
+      "loss": 1.2999,
+      "step": 122310
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 6.1015917961276495e-05,
+      "loss": 1.2863,
+      "step": 122320
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 6.1014770031376745e-05,
+      "loss": 1.2682,
+      "step": 122330
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 6.1013622101476995e-05,
+      "loss": 1.306,
+      "step": 122340
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 6.101247417157725e-05,
+      "loss": 1.2714,
+      "step": 122350
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 6.10113262416775e-05,
+      "loss": 1.187,
+      "step": 122360
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 6.101017831177776e-05,
+      "loss": 1.278,
+      "step": 122370
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 6.100903038187801e-05,
+      "loss": 1.3543,
+      "step": 122380
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 6.100788245197826e-05,
+      "loss": 1.3998,
+      "step": 122390
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 6.1006734522078514e-05,
+      "loss": 1.0913,
+      "step": 122400
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 6.100558659217877e-05,
+      "loss": 1.2689,
+      "step": 122410
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 6.1004438662279014e-05,
+      "loss": 1.1489,
+      "step": 122420
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 6.100329073237927e-05,
+      "loss": 1.1694,
+      "step": 122430
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 6.100214280247953e-05,
+      "loss": 1.2407,
+      "step": 122440
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 6.100099487257977e-05,
+      "loss": 1.2626,
+      "step": 122450
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 6.099984694268003e-05,
+      "loss": 1.2679,
+      "step": 122460
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 6.0998699012780284e-05,
+      "loss": 1.2893,
+      "step": 122470
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 6.0997551082880534e-05,
+      "loss": 1.3746,
+      "step": 122480
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 6.0996403152980783e-05,
+      "loss": 1.2681,
+      "step": 122490
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 6.099525522308104e-05,
+      "loss": 1.3389,
+      "step": 122500
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 6.099410729318129e-05,
+      "loss": 1.2357,
+      "step": 122510
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 6.0992959363281547e-05,
+      "loss": 1.282,
+      "step": 122520
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 6.0991811433381796e-05,
+      "loss": 1.1886,
+      "step": 122530
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 6.0990663503482046e-05,
+      "loss": 1.2905,
+      "step": 122540
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 6.09895155735823e-05,
+      "loss": 1.1167,
+      "step": 122550
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 6.098836764368256e-05,
+      "loss": 1.3597,
+      "step": 122560
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 6.09872197137828e-05,
+      "loss": 1.1241,
+      "step": 122570
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 6.098607178388306e-05,
+      "loss": 1.3626,
+      "step": 122580
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 6.0984923853983316e-05,
+      "loss": 1.1771,
+      "step": 122590
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 6.098377592408356e-05,
+      "loss": 1.2095,
+      "step": 122600
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 6.0982627994183816e-05,
+      "loss": 1.1485,
+      "step": 122610
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 6.098148006428407e-05,
+      "loss": 1.2802,
+      "step": 122620
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 6.098033213438432e-05,
+      "loss": 1.299,
+      "step": 122630
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 6.097918420448457e-05,
+      "loss": 1.2461,
+      "step": 122640
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 6.097803627458483e-05,
+      "loss": 1.1635,
+      "step": 122650
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 6.097688834468508e-05,
+      "loss": 1.1205,
+      "step": 122660
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 6.0975740414785335e-05,
+      "loss": 1.2961,
+      "step": 122670
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 6.0974592484885585e-05,
+      "loss": 1.3011,
+      "step": 122680
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 6.0973444554985835e-05,
+      "loss": 1.2565,
+      "step": 122690
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 6.097229662508609e-05,
+      "loss": 1.342,
+      "step": 122700
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 6.097114869518635e-05,
+      "loss": 1.2617,
+      "step": 122710
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 6.097000076528659e-05,
+      "loss": 1.3054,
+      "step": 122720
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 6.096885283538685e-05,
+      "loss": 1.2565,
+      "step": 122730
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 6.0967704905487104e-05,
+      "loss": 1.3283,
+      "step": 122740
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 6.096655697558735e-05,
+      "loss": 1.1479,
+      "step": 122750
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 6.0965409045687604e-05,
+      "loss": 1.2052,
+      "step": 122760
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 6.096426111578786e-05,
+      "loss": 1.1931,
+      "step": 122770
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 6.096311318588811e-05,
+      "loss": 1.0733,
+      "step": 122780
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 6.096196525598836e-05,
+      "loss": 1.2104,
+      "step": 122790
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 6.096081732608862e-05,
+      "loss": 1.3189,
+      "step": 122800
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 6.095966939618887e-05,
+      "loss": 1.1728,
+      "step": 122810
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.0958521466289123e-05,
+      "loss": 1.3136,
+      "step": 122820
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.095737353638937e-05,
+      "loss": 1.3089,
+      "step": 122830
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.095622560648962e-05,
+      "loss": 1.1763,
+      "step": 122840
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.095507767658988e-05,
+      "loss": 1.1538,
+      "step": 122850
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.0953929746690136e-05,
+      "loss": 1.2365,
+      "step": 122860
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.095278181679038e-05,
+      "loss": 1.1407,
+      "step": 122870
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.0951633886890636e-05,
+      "loss": 1.3565,
+      "step": 122880
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.095048595699089e-05,
+      "loss": 1.2669,
+      "step": 122890
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.0949338027091136e-05,
+      "loss": 1.2809,
+      "step": 122900
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.094819009719139e-05,
+      "loss": 1.2102,
+      "step": 122910
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.094704216729165e-05,
+      "loss": 1.0798,
+      "step": 122920
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.09458942373919e-05,
+      "loss": 1.4949,
+      "step": 122930
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.094474630749215e-05,
+      "loss": 1.2423,
+      "step": 122940
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.0943598377592405e-05,
+      "loss": 1.245,
+      "step": 122950
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.0942450447692655e-05,
+      "loss": 1.3488,
+      "step": 122960
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.094130251779291e-05,
+      "loss": 1.2419,
+      "step": 122970
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.094015458789316e-05,
+      "loss": 1.3526,
+      "step": 122980
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.093900665799341e-05,
+      "loss": 1.3089,
+      "step": 122990
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.093785872809367e-05,
+      "loss": 1.1898,
+      "step": 123000
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.0936710798193925e-05,
+      "loss": 1.3552,
+      "step": 123010
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.093556286829417e-05,
+      "loss": 1.1006,
+      "step": 123020
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 6.0934414938394425e-05,
+      "loss": 1.3578,
+      "step": 123030
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 6.093326700849468e-05,
+      "loss": 1.2596,
+      "step": 123040
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 6.0932119078594924e-05,
+      "loss": 1.3946,
+      "step": 123050
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 6.093097114869518e-05,
+      "loss": 1.2032,
+      "step": 123060
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 6.092982321879544e-05,
+      "loss": 1.2299,
+      "step": 123070
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 6.092867528889569e-05,
+      "loss": 1.2213,
+      "step": 123080
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 6.092752735899594e-05,
+      "loss": 1.4384,
+      "step": 123090
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 6.0926379429096194e-05,
+      "loss": 1.2842,
+      "step": 123100
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 6.0925231499196444e-05,
+      "loss": 1.3168,
+      "step": 123110
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 6.09240835692967e-05,
+      "loss": 1.2653,
+      "step": 123120
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 6.092293563939695e-05,
+      "loss": 1.1618,
+      "step": 123130
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 6.09217877094972e-05,
+      "loss": 1.2695,
+      "step": 123140
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 6.092063977959746e-05,
+      "loss": 1.1255,
+      "step": 123150
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 6.091949184969771e-05,
+      "loss": 1.1938,
+      "step": 123160
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 6.0918343919797956e-05,
+      "loss": 1.2751,
+      "step": 123170
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 6.091719598989821e-05,
+      "loss": 1.2598,
+      "step": 123180
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 6.091604805999847e-05,
+      "loss": 1.3105,
+      "step": 123190
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 6.091490013009871e-05,
+      "loss": 1.2608,
+      "step": 123200
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 6.091375220019897e-05,
+      "loss": 1.064,
+      "step": 123210
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 6.0912604270299226e-05,
+      "loss": 1.2127,
+      "step": 123220
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 6.0911456340399476e-05,
+      "loss": 1.3175,
+      "step": 123230
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 6.0910308410499726e-05,
+      "loss": 1.2005,
+      "step": 123240
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 6.090916048059998e-05,
+      "loss": 1.2143,
+      "step": 123250
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 6.090801255070023e-05,
+      "loss": 1.2263,
+      "step": 123260
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 6.090686462080049e-05,
+      "loss": 1.261,
+      "step": 123270
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 6.090571669090074e-05,
+      "loss": 1.4334,
+      "step": 123280
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 6.090456876100099e-05,
+      "loss": 1.3148,
+      "step": 123290
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 6.0903420831101245e-05,
+      "loss": 1.1807,
+      "step": 123300
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 6.09022729012015e-05,
+      "loss": 1.2165,
+      "step": 123310
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 6.0901124971301745e-05,
+      "loss": 1.434,
+      "step": 123320
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 6.0899977041402e-05,
+      "loss": 1.2359,
+      "step": 123330
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 6.089882911150226e-05,
+      "loss": 1.3323,
+      "step": 123340
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 6.08976811816025e-05,
+      "loss": 1.1552,
+      "step": 123350
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 6.089653325170276e-05,
+      "loss": 1.266,
+      "step": 123360
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 6.089538532180301e-05,
+      "loss": 1.3585,
+      "step": 123370
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 6.0894237391903264e-05,
+      "loss": 1.2306,
+      "step": 123380
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 6.0893089462003514e-05,
+      "loss": 1.3339,
+      "step": 123390
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 6.0891941532103764e-05,
+      "loss": 1.2164,
+      "step": 123400
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 6.089079360220402e-05,
+      "loss": 1.133,
+      "step": 123410
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 6.088964567230428e-05,
+      "loss": 1.2237,
+      "step": 123420
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 6.088849774240452e-05,
+      "loss": 1.2962,
+      "step": 123430
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 6.088734981250478e-05,
+      "loss": 1.2765,
+      "step": 123440
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 6.0886201882605034e-05,
+      "loss": 1.1619,
+      "step": 123450
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 6.088505395270528e-05,
+      "loss": 1.27,
+      "step": 123460
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 6.0883906022805533e-05,
+      "loss": 1.176,
+      "step": 123470
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 6.088275809290579e-05,
+      "loss": 1.1546,
+      "step": 123480
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 6.088161016300604e-05,
+      "loss": 1.1491,
+      "step": 123490
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 6.088046223310629e-05,
+      "loss": 1.2295,
+      "step": 123500
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 6.0879314303206546e-05,
+      "loss": 1.0994,
+      "step": 123510
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 6.0878166373306796e-05,
+      "loss": 1.1182,
+      "step": 123520
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 6.087701844340705e-05,
+      "loss": 1.3862,
+      "step": 123530
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 6.08758705135073e-05,
+      "loss": 1.4835,
+      "step": 123540
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 6.087472258360755e-05,
+      "loss": 1.2241,
+      "step": 123550
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 6.087357465370781e-05,
+      "loss": 1.3477,
+      "step": 123560
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 6.0872426723808066e-05,
+      "loss": 1.1989,
+      "step": 123570
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 6.087127879390831e-05,
+      "loss": 1.4209,
+      "step": 123580
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 6.0870130864008566e-05,
+      "loss": 1.3071,
+      "step": 123590
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 6.086898293410882e-05,
+      "loss": 1.2735,
+      "step": 123600
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 6.0867835004209065e-05,
+      "loss": 1.245,
+      "step": 123610
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 6.086668707430932e-05,
+      "loss": 1.2661,
+      "step": 123620
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 6.086553914440958e-05,
+      "loss": 1.2486,
+      "step": 123630
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 6.086439121450983e-05,
+      "loss": 1.2912,
+      "step": 123640
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 6.086324328461008e-05,
+      "loss": 1.2861,
+      "step": 123650
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 6.0862095354710335e-05,
+      "loss": 1.329,
+      "step": 123660
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 6.0860947424810585e-05,
+      "loss": 1.3267,
+      "step": 123670
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 6.085979949491084e-05,
+      "loss": 1.3792,
+      "step": 123680
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 6.085865156501109e-05,
+      "loss": 1.3507,
+      "step": 123690
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 6.085750363511134e-05,
+      "loss": 1.2962,
+      "step": 123700
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 6.08563557052116e-05,
+      "loss": 1.2272,
+      "step": 123710
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 6.0855207775311854e-05,
+      "loss": 1.2893,
+      "step": 123720
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 6.08540598454121e-05,
+      "loss": 1.1606,
+      "step": 123730
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 6.0852911915512354e-05,
+      "loss": 1.2458,
+      "step": 123740
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 6.085176398561261e-05,
+      "loss": 1.1455,
+      "step": 123750
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 6.0850616055712854e-05,
+      "loss": 1.3809,
+      "step": 123760
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 6.084946812581311e-05,
+      "loss": 1.2443,
+      "step": 123770
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 6.084832019591337e-05,
+      "loss": 1.0752,
+      "step": 123780
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 6.084717226601362e-05,
+      "loss": 1.3464,
+      "step": 123790
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 6.084602433611387e-05,
+      "loss": 1.438,
+      "step": 123800
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 6.084487640621412e-05,
+      "loss": 1.2662,
+      "step": 123810
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 6.084372847631437e-05,
+      "loss": 1.1847,
+      "step": 123820
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 6.084258054641463e-05,
+      "loss": 1.1325,
+      "step": 123830
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 6.084143261651488e-05,
+      "loss": 1.1895,
+      "step": 123840
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 6.084028468661513e-05,
+      "loss": 1.1063,
+      "step": 123850
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 6.0839136756715386e-05,
+      "loss": 1.2978,
+      "step": 123860
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 6.083798882681564e-05,
+      "loss": 1.2513,
+      "step": 123870
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 6.0836840896915886e-05,
+      "loss": 1.2213,
+      "step": 123880
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 6.083569296701614e-05,
+      "loss": 1.3211,
+      "step": 123890
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 6.08345450371164e-05,
+      "loss": 1.2605,
+      "step": 123900
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 6.083339710721664e-05,
+      "loss": 1.2876,
+      "step": 123910
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 6.08322491773169e-05,
+      "loss": 1.2993,
+      "step": 123920
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 6.0831101247417155e-05,
+      "loss": 1.3013,
+      "step": 123930
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 6.0829953317517405e-05,
+      "loss": 1.2183,
+      "step": 123940
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 6.0828805387617655e-05,
+      "loss": 1.361,
+      "step": 123950
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 6.082765745771791e-05,
+      "loss": 1.2067,
+      "step": 123960
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 6.082650952781816e-05,
+      "loss": 1.2665,
+      "step": 123970
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 6.082536159791842e-05,
+      "loss": 1.2078,
+      "step": 123980
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 6.082421366801867e-05,
+      "loss": 1.1848,
+      "step": 123990
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 6.082306573811892e-05,
+      "loss": 1.1927,
+      "step": 124000
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 6.0821917808219175e-05,
+      "loss": 1.2036,
+      "step": 124010
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 6.082076987831943e-05,
+      "loss": 1.4959,
+      "step": 124020
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 6.0819621948419674e-05,
+      "loss": 1.0663,
+      "step": 124030
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 6.081847401851993e-05,
+      "loss": 1.2934,
+      "step": 124040
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 6.081732608862019e-05,
+      "loss": 1.3154,
+      "step": 124050
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 6.081617815872043e-05,
+      "loss": 1.2259,
+      "step": 124060
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 6.081503022882069e-05,
+      "loss": 1.3303,
+      "step": 124070
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 6.0813882298920944e-05,
+      "loss": 1.1615,
+      "step": 124080
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 6.0812734369021194e-05,
+      "loss": 1.2175,
+      "step": 124090
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 6.0811586439121444e-05,
+      "loss": 1.2737,
+      "step": 124100
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 6.08104385092217e-05,
+      "loss": 1.3348,
+      "step": 124110
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 6.080929057932195e-05,
+      "loss": 1.1037,
+      "step": 124120
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.080814264942221e-05,
+      "loss": 1.1687,
+      "step": 124130
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.0806994719522457e-05,
+      "loss": 1.1995,
+      "step": 124140
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.0805846789622706e-05,
+      "loss": 1.2842,
+      "step": 124150
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.080469885972296e-05,
+      "loss": 1.3324,
+      "step": 124160
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.080355092982322e-05,
+      "loss": 1.2017,
+      "step": 124170
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.080240299992346e-05,
+      "loss": 1.1053,
+      "step": 124180
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.080125507002372e-05,
+      "loss": 1.3997,
+      "step": 124190
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.0800107140123976e-05,
+      "loss": 1.2811,
+      "step": 124200
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.079895921022422e-05,
+      "loss": 1.3174,
+      "step": 124210
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.0797811280324476e-05,
+      "loss": 1.0849,
+      "step": 124220
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.079666335042473e-05,
+      "loss": 1.4456,
+      "step": 124230
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.079551542052498e-05,
+      "loss": 1.3967,
+      "step": 124240
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.079436749062523e-05,
+      "loss": 1.3683,
+      "step": 124250
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.079321956072549e-05,
+      "loss": 1.2532,
+      "step": 124260
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.079207163082574e-05,
+      "loss": 1.2303,
+      "step": 124270
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.0790923700925995e-05,
+      "loss": 1.2404,
+      "step": 124280
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.0789775771026245e-05,
+      "loss": 1.3221,
+      "step": 124290
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.0788627841126495e-05,
+      "loss": 1.2726,
+      "step": 124300
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.078747991122675e-05,
+      "loss": 1.2058,
+      "step": 124310
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.078633198132701e-05,
+      "loss": 1.1437,
+      "step": 124320
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.078518405142725e-05,
+      "loss": 1.2269,
+      "step": 124330
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 6.078403612152751e-05,
+      "loss": 1.1606,
+      "step": 124340
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 6.0782888191627765e-05,
+      "loss": 1.2593,
+      "step": 124350
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 6.078174026172801e-05,
+      "loss": 1.3569,
+      "step": 124360
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 6.0780592331828264e-05,
+      "loss": 1.2963,
+      "step": 124370
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 6.077944440192852e-05,
+      "loss": 1.2405,
+      "step": 124380
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 6.077829647202877e-05,
+      "loss": 1.2427,
+      "step": 124390
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 6.077714854212902e-05,
+      "loss": 1.31,
+      "step": 124400
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 6.077600061222928e-05,
+      "loss": 1.1967,
+      "step": 124410
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 6.077485268232953e-05,
+      "loss": 1.2973,
+      "step": 124420
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 6.0773704752429784e-05,
+      "loss": 1.2558,
+      "step": 124430
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 6.0772556822530034e-05,
+      "loss": 1.1821,
+      "step": 124440
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 6.0771408892630283e-05,
+      "loss": 1.1244,
+      "step": 124450
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 6.077026096273054e-05,
+      "loss": 1.3207,
+      "step": 124460
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 6.07691130328308e-05,
+      "loss": 1.3766,
+      "step": 124470
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 6.076796510293104e-05,
+      "loss": 1.3204,
+      "step": 124480
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 6.0766817173031296e-05,
+      "loss": 1.219,
+      "step": 124490
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 6.076566924313155e-05,
+      "loss": 1.2011,
+      "step": 124500
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 6.0764521313231796e-05,
+      "loss": 1.3313,
+      "step": 124510
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 6.076337338333205e-05,
+      "loss": 1.2884,
+      "step": 124520
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 6.076222545343231e-05,
+      "loss": 1.1919,
+      "step": 124530
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 6.076107752353256e-05,
+      "loss": 1.3158,
+      "step": 124540
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 6.075992959363281e-05,
+      "loss": 1.1754,
+      "step": 124550
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.0758781663733066e-05,
+      "loss": 1.1321,
+      "step": 124560
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.0757633733833316e-05,
+      "loss": 1.3229,
+      "step": 124570
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.075648580393357e-05,
+      "loss": 1.388,
+      "step": 124580
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.075533787403382e-05,
+      "loss": 1.218,
+      "step": 124590
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.075418994413407e-05,
+      "loss": 1.2969,
+      "step": 124600
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.075304201423433e-05,
+      "loss": 1.2507,
+      "step": 124610
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.0751894084334585e-05,
+      "loss": 1.1614,
+      "step": 124620
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.075074615443483e-05,
+      "loss": 1.2784,
+      "step": 124630
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.0749598224535085e-05,
+      "loss": 1.3031,
+      "step": 124640
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.074845029463534e-05,
+      "loss": 1.2023,
+      "step": 124650
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.0747302364735585e-05,
+      "loss": 1.3276,
+      "step": 124660
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.074615443483584e-05,
+      "loss": 1.2369,
+      "step": 124670
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.07450065049361e-05,
+      "loss": 1.2369,
+      "step": 124680
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.074385857503635e-05,
+      "loss": 1.232,
+      "step": 124690
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.07427106451366e-05,
+      "loss": 1.2899,
+      "step": 124700
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.0741562715236854e-05,
+      "loss": 1.2598,
+      "step": 124710
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.0740414785337104e-05,
+      "loss": 1.2859,
+      "step": 124720
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.073926685543736e-05,
+      "loss": 1.4341,
+      "step": 124730
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.0738118925537604e-05,
+      "loss": 1.292,
+      "step": 124740
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.073697099563786e-05,
+      "loss": 1.3636,
+      "step": 124750
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.073582306573812e-05,
+      "loss": 1.3045,
+      "step": 124760
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 6.073467513583836e-05,
+      "loss": 1.2563,
+      "step": 124770
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 6.073352720593862e-05,
+      "loss": 1.3673,
+      "step": 124780
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 6.073237927603887e-05,
+      "loss": 1.2435,
+      "step": 124790
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 6.073123134613912e-05,
+      "loss": 1.2527,
+      "step": 124800
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 6.073008341623937e-05,
+      "loss": 1.1243,
+      "step": 124810
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 6.072893548633963e-05,
+      "loss": 1.1897,
+      "step": 124820
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 6.072778755643988e-05,
+      "loss": 1.1558,
+      "step": 124830
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 6.0726639626540136e-05,
+      "loss": 1.2619,
+      "step": 124840
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 6.0725491696640386e-05,
+      "loss": 1.2555,
+      "step": 124850
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 6.0724343766740636e-05,
+      "loss": 1.2575,
+      "step": 124860
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 6.072319583684089e-05,
+      "loss": 1.2851,
+      "step": 124870
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 6.072204790694115e-05,
+      "loss": 1.2694,
+      "step": 124880
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 6.072089997704139e-05,
+      "loss": 1.381,
+      "step": 124890
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 6.071975204714165e-05,
+      "loss": 1.1469,
+      "step": 124900
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 6.0718604117241905e-05,
+      "loss": 1.5318,
+      "step": 124910
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 6.071745618734215e-05,
+      "loss": 1.3231,
+      "step": 124920
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 6.0716308257442405e-05,
+      "loss": 1.3079,
+      "step": 124930
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 6.071516032754266e-05,
+      "loss": 1.2858,
+      "step": 124940
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 6.071401239764291e-05,
+      "loss": 1.3799,
+      "step": 124950
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 6.071286446774316e-05,
+      "loss": 1.1932,
+      "step": 124960
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 6.071171653784342e-05,
+      "loss": 1.2837,
+      "step": 124970
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 6.071056860794367e-05,
+      "loss": 1.3141,
+      "step": 124980
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 6.0709420678043925e-05,
+      "loss": 1.171,
+      "step": 124990
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 6.0708272748144174e-05,
+      "loss": 1.2662,
+      "step": 125000
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 6.0707124818244424e-05,
+      "loss": 1.2103,
+      "step": 125010
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 6.070597688834468e-05,
+      "loss": 1.229,
+      "step": 125020
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 6.070482895844494e-05,
+      "loss": 1.2927,
+      "step": 125030
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 6.070368102854518e-05,
+      "loss": 1.2935,
+      "step": 125040
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 6.070253309864544e-05,
+      "loss": 1.3249,
+      "step": 125050
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 6.0701385168745694e-05,
+      "loss": 1.3148,
+      "step": 125060
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 6.070023723884594e-05,
+      "loss": 1.1468,
+      "step": 125070
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 6.0699089308946194e-05,
+      "loss": 1.3538,
+      "step": 125080
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 6.069794137904645e-05,
+      "loss": 1.4518,
+      "step": 125090
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 6.06967934491467e-05,
+      "loss": 1.4905,
+      "step": 125100
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 6.069564551924695e-05,
+      "loss": 1.2013,
+      "step": 125110
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 6.0694497589347207e-05,
+      "loss": 1.2762,
+      "step": 125120
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 6.0693349659447456e-05,
+      "loss": 1.1939,
+      "step": 125130
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 6.069220172954771e-05,
+      "loss": 1.3309,
+      "step": 125140
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 6.069105379964796e-05,
+      "loss": 1.2215,
+      "step": 125150
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 6.068990586974821e-05,
+      "loss": 1.1673,
+      "step": 125160
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 6.068875793984847e-05,
+      "loss": 1.2586,
+      "step": 125170
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 6.0687610009948726e-05,
+      "loss": 1.1806,
+      "step": 125180
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 6.068646208004897e-05,
+      "loss": 1.1051,
+      "step": 125190
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 6.0685314150149226e-05,
+      "loss": 1.3152,
+      "step": 125200
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 6.068416622024948e-05,
+      "loss": 1.2457,
+      "step": 125210
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.0683018290349725e-05,
+      "loss": 1.2241,
+      "step": 125220
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.068187036044998e-05,
+      "loss": 1.2749,
+      "step": 125230
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.068072243055024e-05,
+      "loss": 1.1948,
+      "step": 125240
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.067957450065049e-05,
+      "loss": 1.2712,
+      "step": 125250
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.067842657075074e-05,
+      "loss": 1.2387,
+      "step": 125260
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.0677278640850995e-05,
+      "loss": 1.0782,
+      "step": 125270
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.0676130710951245e-05,
+      "loss": 1.4914,
+      "step": 125280
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.06749827810515e-05,
+      "loss": 1.2276,
+      "step": 125290
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.067383485115175e-05,
+      "loss": 1.2932,
+      "step": 125300
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.0672686921252e-05,
+      "loss": 1.3369,
+      "step": 125310
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.067153899135226e-05,
+      "loss": 1.1584,
+      "step": 125320
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.0670391061452515e-05,
+      "loss": 1.3719,
+      "step": 125330
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.066924313155276e-05,
+      "loss": 1.3129,
+      "step": 125340
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.0668095201653014e-05,
+      "loss": 1.2336,
+      "step": 125350
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.066694727175327e-05,
+      "loss": 1.1723,
+      "step": 125360
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.0665799341853514e-05,
+      "loss": 1.2991,
+      "step": 125370
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.066465141195377e-05,
+      "loss": 1.3175,
+      "step": 125380
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.066350348205403e-05,
+      "loss": 1.2803,
+      "step": 125390
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.066235555215428e-05,
+      "loss": 1.2962,
+      "step": 125400
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.066120762225453e-05,
+      "loss": 1.2048,
+      "step": 125410
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.0660059692354784e-05,
+      "loss": 1.1975,
+      "step": 125420
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 6.0658911762455033e-05,
+      "loss": 1.327,
+      "step": 125430
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 6.065776383255529e-05,
+      "loss": 1.4083,
+      "step": 125440
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 6.065661590265554e-05,
+      "loss": 1.295,
+      "step": 125450
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 6.065546797275579e-05,
+      "loss": 1.2147,
+      "step": 125460
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 6.0654320042856046e-05,
+      "loss": 1.2115,
+      "step": 125470
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 6.06531721129563e-05,
+      "loss": 1.312,
+      "step": 125480
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 6.0652024183056546e-05,
+      "loss": 1.3141,
+      "step": 125490
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 6.06508762531568e-05,
+      "loss": 1.2697,
+      "step": 125500
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 6.064972832325706e-05,
+      "loss": 1.2196,
+      "step": 125510
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 6.06485803933573e-05,
+      "loss": 1.3302,
+      "step": 125520
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 6.064743246345756e-05,
+      "loss": 1.2724,
+      "step": 125530
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 6.0646284533557816e-05,
+      "loss": 1.3554,
+      "step": 125540
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 6.0645136603658066e-05,
+      "loss": 1.2362,
+      "step": 125550
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 6.0643988673758315e-05,
+      "loss": 1.3168,
+      "step": 125560
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 6.064284074385857e-05,
+      "loss": 1.2631,
+      "step": 125570
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 6.064169281395882e-05,
+      "loss": 1.3195,
+      "step": 125580
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 6.064054488405908e-05,
+      "loss": 1.3049,
+      "step": 125590
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 6.063939695415933e-05,
+      "loss": 1.3105,
+      "step": 125600
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 6.063824902425958e-05,
+      "loss": 1.1795,
+      "step": 125610
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 6.0637101094359835e-05,
+      "loss": 1.2562,
+      "step": 125620
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 6.063595316446009e-05,
+      "loss": 1.1457,
+      "step": 125630
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 6.0634805234560335e-05,
+      "loss": 1.3033,
+      "step": 125640
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.063365730466059e-05,
+      "loss": 1.1939,
+      "step": 125650
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.063250937476085e-05,
+      "loss": 1.2021,
+      "step": 125660
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.063136144486109e-05,
+      "loss": 1.3166,
+      "step": 125670
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.063021351496135e-05,
+      "loss": 1.2526,
+      "step": 125680
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.0629065585061604e-05,
+      "loss": 1.3527,
+      "step": 125690
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.0627917655161854e-05,
+      "loss": 1.4234,
+      "step": 125700
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.0626769725262104e-05,
+      "loss": 1.2285,
+      "step": 125710
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.062562179536236e-05,
+      "loss": 1.3857,
+      "step": 125720
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.062447386546261e-05,
+      "loss": 1.1963,
+      "step": 125730
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.062332593556287e-05,
+      "loss": 1.2954,
+      "step": 125740
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.062217800566312e-05,
+      "loss": 1.2584,
+      "step": 125750
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.062103007576337e-05,
+      "loss": 1.1337,
+      "step": 125760
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.061988214586362e-05,
+      "loss": 1.1558,
+      "step": 125770
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.061873421596388e-05,
+      "loss": 1.0891,
+      "step": 125780
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.061758628606412e-05,
+      "loss": 1.1125,
+      "step": 125790
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.061643835616438e-05,
+      "loss": 1.3031,
+      "step": 125800
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.0615290426264636e-05,
+      "loss": 1.1158,
+      "step": 125810
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.061414249636488e-05,
+      "loss": 1.1823,
+      "step": 125820
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.0612994566465136e-05,
+      "loss": 1.2125,
+      "step": 125830
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.061184663656539e-05,
+      "loss": 1.2346,
+      "step": 125840
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.061069870666564e-05,
+      "loss": 1.2105,
+      "step": 125850
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 6.060955077676589e-05,
+      "loss": 1.3503,
+      "step": 125860
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.060840284686615e-05,
+      "loss": 1.2499,
+      "step": 125870
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.06072549169664e-05,
+      "loss": 1.4569,
+      "step": 125880
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.0606106987066655e-05,
+      "loss": 1.2162,
+      "step": 125890
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.0604959057166905e-05,
+      "loss": 1.2347,
+      "step": 125900
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.0603811127267155e-05,
+      "loss": 1.215,
+      "step": 125910
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.060266319736741e-05,
+      "loss": 1.1411,
+      "step": 125920
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.060151526746767e-05,
+      "loss": 1.1775,
+      "step": 125930
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.060036733756791e-05,
+      "loss": 1.1858,
+      "step": 125940
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.059921940766817e-05,
+      "loss": 1.3008,
+      "step": 125950
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.0598071477768425e-05,
+      "loss": 1.0971,
+      "step": 125960
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.059692354786867e-05,
+      "loss": 1.2912,
+      "step": 125970
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.0595775617968924e-05,
+      "loss": 1.2374,
+      "step": 125980
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.059462768806918e-05,
+      "loss": 1.1439,
+      "step": 125990
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.059347975816943e-05,
+      "loss": 1.3546,
+      "step": 126000
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.059233182826968e-05,
+      "loss": 1.343,
+      "step": 126010
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.059118389836994e-05,
+      "loss": 1.309,
+      "step": 126020
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.059003596847019e-05,
+      "loss": 1.1915,
+      "step": 126030
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.0588888038570444e-05,
+      "loss": 1.1541,
+      "step": 126040
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.0587740108670694e-05,
+      "loss": 1.3285,
+      "step": 126050
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.0586592178770944e-05,
+      "loss": 1.2642,
+      "step": 126060
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.05854442488712e-05,
+      "loss": 1.2212,
+      "step": 126070
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.058429631897146e-05,
+      "loss": 1.1787,
+      "step": 126080
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.05831483890717e-05,
+      "loss": 1.2798,
+      "step": 126090
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.0582000459171957e-05,
+      "loss": 1.2145,
+      "step": 126100
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.05808525292722e-05,
+      "loss": 1.4415,
+      "step": 126110
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.0579704599372456e-05,
+      "loss": 1.2508,
+      "step": 126120
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.057855666947271e-05,
+      "loss": 1.2739,
+      "step": 126130
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.057740873957296e-05,
+      "loss": 1.1614,
+      "step": 126140
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.057626080967322e-05,
+      "loss": 1.1171,
+      "step": 126150
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.057511287977347e-05,
+      "loss": 1.1985,
+      "step": 126160
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.057396494987372e-05,
+      "loss": 1.2765,
+      "step": 126170
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.0572817019973976e-05,
+      "loss": 1.2229,
+      "step": 126180
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.057166909007423e-05,
+      "loss": 1.1881,
+      "step": 126190
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.0570521160174475e-05,
+      "loss": 1.3335,
+      "step": 126200
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.056937323027473e-05,
+      "loss": 1.1741,
+      "step": 126210
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.056822530037499e-05,
+      "loss": 1.2068,
+      "step": 126220
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.056707737047523e-05,
+      "loss": 1.3798,
+      "step": 126230
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.056592944057549e-05,
+      "loss": 1.2356,
+      "step": 126240
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.0564781510675745e-05,
+      "loss": 1.1794,
+      "step": 126250
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.056363358077599e-05,
+      "loss": 1.2366,
+      "step": 126260
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.0562485650876245e-05,
+      "loss": 1.178,
+      "step": 126270
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.05613377209765e-05,
+      "loss": 1.3517,
+      "step": 126280
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.056018979107675e-05,
+      "loss": 1.1864,
+      "step": 126290
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.055904186117701e-05,
+      "loss": 1.2478,
+      "step": 126300
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.055789393127726e-05,
+      "loss": 1.0777,
+      "step": 126310
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.055674600137751e-05,
+      "loss": 1.2816,
+      "step": 126320
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.0555598071477764e-05,
+      "loss": 1.2692,
+      "step": 126330
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.055445014157802e-05,
+      "loss": 1.2097,
+      "step": 126340
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.0553302211678264e-05,
+      "loss": 1.3104,
+      "step": 126350
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.055215428177852e-05,
+      "loss": 1.3407,
+      "step": 126360
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.055100635187878e-05,
+      "loss": 1.3834,
+      "step": 126370
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.054985842197902e-05,
+      "loss": 1.1048,
+      "step": 126380
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.054871049207928e-05,
+      "loss": 1.2486,
+      "step": 126390
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.0547562562179534e-05,
+      "loss": 1.2684,
+      "step": 126400
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.0546414632279777e-05,
+      "loss": 1.2637,
+      "step": 126410
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.054526670238003e-05,
+      "loss": 1.4546,
+      "step": 126420
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.054411877248029e-05,
+      "loss": 1.2301,
+      "step": 126430
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.054297084258054e-05,
+      "loss": 1.2562,
+      "step": 126440
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.0541822912680796e-05,
+      "loss": 1.1528,
+      "step": 126450
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.0540674982781046e-05,
+      "loss": 1.2175,
+      "step": 126460
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.0539527052881296e-05,
+      "loss": 1.1847,
+      "step": 126470
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.053837912298155e-05,
+      "loss": 1.2874,
+      "step": 126480
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.053723119308181e-05,
+      "loss": 1.3995,
+      "step": 126490
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.053608326318205e-05,
+      "loss": 1.13,
+      "step": 126500
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.053493533328231e-05,
+      "loss": 1.3048,
+      "step": 126510
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.0533787403382566e-05,
+      "loss": 1.3524,
+      "step": 126520
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 6.053263947348281e-05,
+      "loss": 1.2281,
+      "step": 126530
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 6.0531491543583065e-05,
+      "loss": 1.2798,
+      "step": 126540
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 6.053034361368332e-05,
+      "loss": 1.2268,
+      "step": 126550
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 6.0529195683783565e-05,
+      "loss": 1.3156,
+      "step": 126560
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 6.052804775388382e-05,
+      "loss": 1.2783,
+      "step": 126570
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 6.052689982398408e-05,
+      "loss": 1.3281,
+      "step": 126580
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 6.052575189408433e-05,
+      "loss": 1.3211,
+      "step": 126590
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 6.0524603964184585e-05,
+      "loss": 1.1722,
+      "step": 126600
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 6.0523456034284835e-05,
+      "loss": 1.2181,
+      "step": 126610
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 6.0522308104385085e-05,
+      "loss": 1.2948,
+      "step": 126620
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 6.052116017448534e-05,
+      "loss": 1.261,
+      "step": 126630
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 6.05200122445856e-05,
+      "loss": 1.2164,
+      "step": 126640
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 6.051886431468584e-05,
+      "loss": 1.2821,
+      "step": 126650
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 6.05177163847861e-05,
+      "loss": 1.2397,
+      "step": 126660
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 6.0516568454886354e-05,
+      "loss": 1.1902,
+      "step": 126670
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 6.05154205249866e-05,
+      "loss": 1.4084,
+      "step": 126680
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 6.0514272595086854e-05,
+      "loss": 1.276,
+      "step": 126690
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 6.051312466518711e-05,
+      "loss": 1.3673,
+      "step": 126700
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 6.0511976735287354e-05,
+      "loss": 1.3179,
+      "step": 126710
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 6.051082880538761e-05,
+      "loss": 1.2116,
+      "step": 126720
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 6.050968087548787e-05,
+      "loss": 1.0923,
+      "step": 126730
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 6.050853294558812e-05,
+      "loss": 1.3163,
+      "step": 126740
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 6.050738501568837e-05,
+      "loss": 1.2587,
+      "step": 126750
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 6.050623708578862e-05,
+      "loss": 1.2983,
+      "step": 126760
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 6.050508915588887e-05,
+      "loss": 1.4159,
+      "step": 126770
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 6.050394122598913e-05,
+      "loss": 1.1713,
+      "step": 126780
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 6.0502793296089386e-05,
+      "loss": 1.1601,
+      "step": 126790
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 6.050164536618963e-05,
+      "loss": 1.3271,
+      "step": 126800
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 6.0500497436289886e-05,
+      "loss": 1.322,
+      "step": 126810
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 6.049934950639014e-05,
+      "loss": 1.3593,
+      "step": 126820
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 6.0498201576490386e-05,
+      "loss": 1.3414,
+      "step": 126830
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 6.049705364659064e-05,
+      "loss": 1.2515,
+      "step": 126840
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 6.04959057166909e-05,
+      "loss": 1.1923,
+      "step": 126850
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 6.049475778679114e-05,
+      "loss": 1.3016,
+      "step": 126860
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 6.04936098568914e-05,
+      "loss": 1.3645,
+      "step": 126870
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 6.0492461926991655e-05,
+      "loss": 1.3123,
+      "step": 126880
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 6.0491313997091905e-05,
+      "loss": 1.3282,
+      "step": 126890
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 6.049016606719216e-05,
+      "loss": 1.2521,
+      "step": 126900
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 6.048901813729241e-05,
+      "loss": 1.2829,
+      "step": 126910
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 6.048787020739266e-05,
+      "loss": 1.1914,
+      "step": 126920
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 6.048672227749292e-05,
+      "loss": 1.2421,
+      "step": 126930
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 6.0485574347593175e-05,
+      "loss": 1.3534,
+      "step": 126940
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 6.048442641769342e-05,
+      "loss": 1.2337,
+      "step": 126950
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.0483278487793674e-05,
+      "loss": 1.2083,
+      "step": 126960
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.048213055789393e-05,
+      "loss": 1.4097,
+      "step": 126970
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.0480982627994174e-05,
+      "loss": 1.3665,
+      "step": 126980
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.047983469809443e-05,
+      "loss": 1.282,
+      "step": 126990
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.047868676819469e-05,
+      "loss": 1.2883,
+      "step": 127000
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.047753883829493e-05,
+      "loss": 1.3212,
+      "step": 127010
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.047639090839519e-05,
+      "loss": 1.1315,
+      "step": 127020
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.0475242978495444e-05,
+      "loss": 1.2977,
+      "step": 127030
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.0474095048595694e-05,
+      "loss": 1.3148,
+      "step": 127040
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.047294711869595e-05,
+      "loss": 1.3474,
+      "step": 127050
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.04717991887962e-05,
+      "loss": 1.1923,
+      "step": 127060
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.047065125889645e-05,
+      "loss": 1.2268,
+      "step": 127070
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.0469503328996707e-05,
+      "loss": 1.3032,
+      "step": 127080
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.046835539909696e-05,
+      "loss": 1.232,
+      "step": 127090
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.0467207469197206e-05,
+      "loss": 1.2061,
+      "step": 127100
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.046605953929746e-05,
+      "loss": 1.2294,
+      "step": 127110
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.046491160939772e-05,
+      "loss": 1.2642,
+      "step": 127120
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.046376367949796e-05,
+      "loss": 1.4445,
+      "step": 127130
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.046261574959822e-05,
+      "loss": 1.4281,
+      "step": 127140
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.0461467819698476e-05,
+      "loss": 1.311,
+      "step": 127150
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.046031988979872e-05,
+      "loss": 1.1617,
+      "step": 127160
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 6.0459171959898976e-05,
+      "loss": 1.3289,
+      "step": 127170
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 6.045802402999923e-05,
+      "loss": 1.2384,
+      "step": 127180
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 6.045687610009948e-05,
+      "loss": 1.2229,
+      "step": 127190
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 6.045572817019974e-05,
+      "loss": 1.18,
+      "step": 127200
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 6.045458024029999e-05,
+      "loss": 1.1963,
+      "step": 127210
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 6.045343231040024e-05,
+      "loss": 1.0428,
+      "step": 127220
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 6.0452284380500495e-05,
+      "loss": 1.2098,
+      "step": 127230
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 6.045113645060075e-05,
+      "loss": 1.2844,
+      "step": 127240
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 6.0449988520700995e-05,
+      "loss": 1.3302,
+      "step": 127250
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 6.044884059080125e-05,
+      "loss": 1.1561,
+      "step": 127260
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 6.044769266090151e-05,
+      "loss": 1.259,
+      "step": 127270
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 6.044654473100175e-05,
+      "loss": 1.2197,
+      "step": 127280
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 6.044539680110201e-05,
+      "loss": 1.2541,
+      "step": 127290
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 6.0444248871202264e-05,
+      "loss": 1.2381,
+      "step": 127300
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 6.044310094130251e-05,
+      "loss": 1.2646,
+      "step": 127310
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 6.0441953011402764e-05,
+      "loss": 1.1117,
+      "step": 127320
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 6.044080508150302e-05,
+      "loss": 1.1854,
+      "step": 127330
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 6.043965715160327e-05,
+      "loss": 1.289,
+      "step": 127340
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 6.043850922170353e-05,
+      "loss": 1.2251,
+      "step": 127350
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 6.043736129180378e-05,
+      "loss": 1.3596,
+      "step": 127360
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 6.043621336190403e-05,
+      "loss": 1.2808,
+      "step": 127370
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 6.0435065432004284e-05,
+      "loss": 1.1927,
+      "step": 127380
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 6.043391750210454e-05,
+      "loss": 1.1697,
+      "step": 127390
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 6.043276957220478e-05,
+      "loss": 0.9921,
+      "step": 127400
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 6.043162164230504e-05,
+      "loss": 1.1728,
+      "step": 127410
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 6.0430473712405296e-05,
+      "loss": 1.327,
+      "step": 127420
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 6.042932578250554e-05,
+      "loss": 1.2662,
+      "step": 127430
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 6.0428177852605796e-05,
+      "loss": 1.3971,
+      "step": 127440
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 6.042702992270605e-05,
+      "loss": 1.1779,
+      "step": 127450
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 6.0425881992806296e-05,
+      "loss": 1.3866,
+      "step": 127460
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 6.042473406290655e-05,
+      "loss": 1.2889,
+      "step": 127470
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 6.04235861330068e-05,
+      "loss": 1.2024,
+      "step": 127480
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 6.042243820310706e-05,
+      "loss": 1.1764,
+      "step": 127490
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 6.0421290273207316e-05,
+      "loss": 1.2505,
+      "step": 127500
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 6.042014234330756e-05,
+      "loss": 1.2621,
+      "step": 127510
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 6.0418994413407815e-05,
+      "loss": 1.3687,
+      "step": 127520
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 6.041784648350807e-05,
+      "loss": 1.1875,
+      "step": 127530
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 6.0416698553608315e-05,
+      "loss": 1.327,
+      "step": 127540
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 6.041555062370857e-05,
+      "loss": 1.2823,
+      "step": 127550
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 6.041440269380883e-05,
+      "loss": 1.4594,
+      "step": 127560
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 6.041325476390907e-05,
+      "loss": 1.3265,
+      "step": 127570
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 6.041210683400933e-05,
+      "loss": 1.2636,
+      "step": 127580
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 6.0410958904109585e-05,
+      "loss": 1.3029,
+      "step": 127590
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 6.0409810974209835e-05,
+      "loss": 1.2815,
+      "step": 127600
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 6.0408663044310084e-05,
+      "loss": 1.303,
+      "step": 127610
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.040751511441034e-05,
+      "loss": 1.264,
+      "step": 127620
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.040636718451059e-05,
+      "loss": 1.2763,
+      "step": 127630
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.040521925461085e-05,
+      "loss": 1.3612,
+      "step": 127640
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.0404071324711104e-05,
+      "loss": 1.1623,
+      "step": 127650
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.040292339481135e-05,
+      "loss": 1.322,
+      "step": 127660
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.0401775464911604e-05,
+      "loss": 1.2319,
+      "step": 127670
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.040062753501186e-05,
+      "loss": 1.314,
+      "step": 127680
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.0399479605112104e-05,
+      "loss": 1.308,
+      "step": 127690
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.039833167521236e-05,
+      "loss": 1.1884,
+      "step": 127700
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.039718374531262e-05,
+      "loss": 1.3095,
+      "step": 127710
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.039603581541286e-05,
+      "loss": 1.2201,
+      "step": 127720
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.0394887885513117e-05,
+      "loss": 1.2747,
+      "step": 127730
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.039373995561337e-05,
+      "loss": 1.1642,
+      "step": 127740
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.039259202571362e-05,
+      "loss": 1.2396,
+      "step": 127750
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.039144409581387e-05,
+      "loss": 1.2797,
+      "step": 127760
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.039029616591413e-05,
+      "loss": 1.0684,
+      "step": 127770
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.038914823601438e-05,
+      "loss": 1.3616,
+      "step": 127780
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.0388000306114636e-05,
+      "loss": 1.2877,
+      "step": 127790
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.038685237621489e-05,
+      "loss": 1.2058,
+      "step": 127800
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.0385704446315136e-05,
+      "loss": 1.2665,
+      "step": 127810
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 6.038455651641539e-05,
+      "loss": 1.4808,
+      "step": 127820
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.038340858651565e-05,
+      "loss": 1.2624,
+      "step": 127830
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.038226065661589e-05,
+      "loss": 1.2782,
+      "step": 127840
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.038111272671615e-05,
+      "loss": 1.2579,
+      "step": 127850
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.0379964796816405e-05,
+      "loss": 1.2643,
+      "step": 127860
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.037881686691665e-05,
+      "loss": 1.513,
+      "step": 127870
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.0377668937016905e-05,
+      "loss": 1.3643,
+      "step": 127880
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.037652100711716e-05,
+      "loss": 1.4085,
+      "step": 127890
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.037537307721741e-05,
+      "loss": 1.222,
+      "step": 127900
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.037422514731766e-05,
+      "loss": 1.2603,
+      "step": 127910
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.037307721741792e-05,
+      "loss": 1.1784,
+      "step": 127920
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.037192928751817e-05,
+      "loss": 1.1988,
+      "step": 127930
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.0370781357618424e-05,
+      "loss": 1.0537,
+      "step": 127940
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.036963342771868e-05,
+      "loss": 1.257,
+      "step": 127950
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.0368485497818924e-05,
+      "loss": 1.4868,
+      "step": 127960
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.036733756791918e-05,
+      "loss": 1.2309,
+      "step": 127970
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.036618963801944e-05,
+      "loss": 1.4895,
+      "step": 127980
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.036504170811968e-05,
+      "loss": 1.2462,
+      "step": 127990
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.036389377821994e-05,
+      "loss": 1.1126,
+      "step": 128000
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.0362745848320194e-05,
+      "loss": 1.1597,
+      "step": 128010
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.036159791842044e-05,
+      "loss": 1.2131,
+      "step": 128020
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.0360449988520693e-05,
+      "loss": 1.1714,
+      "step": 128030
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 6.035930205862095e-05,
+      "loss": 1.2008,
+      "step": 128040
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.03581541287212e-05,
+      "loss": 1.3723,
+      "step": 128050
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.035700619882145e-05,
+      "loss": 1.1066,
+      "step": 128060
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.0355858268921706e-05,
+      "loss": 1.1814,
+      "step": 128070
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.0354710339021956e-05,
+      "loss": 1.2234,
+      "step": 128080
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.035356240912221e-05,
+      "loss": 1.3407,
+      "step": 128090
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.035241447922247e-05,
+      "loss": 1.3945,
+      "step": 128100
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.035126654932271e-05,
+      "loss": 1.3017,
+      "step": 128110
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.035011861942297e-05,
+      "loss": 1.1616,
+      "step": 128120
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.0348970689523226e-05,
+      "loss": 1.23,
+      "step": 128130
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.034782275962347e-05,
+      "loss": 1.2234,
+      "step": 128140
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.0346674829723726e-05,
+      "loss": 1.339,
+      "step": 128150
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.034552689982398e-05,
+      "loss": 1.2239,
+      "step": 128160
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.0344378969924225e-05,
+      "loss": 1.2752,
+      "step": 128170
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.034323104002448e-05,
+      "loss": 1.2442,
+      "step": 128180
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.034208311012474e-05,
+      "loss": 1.2418,
+      "step": 128190
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.034093518022499e-05,
+      "loss": 1.2635,
+      "step": 128200
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.033978725032524e-05,
+      "loss": 1.24,
+      "step": 128210
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.0338639320425495e-05,
+      "loss": 1.3754,
+      "step": 128220
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.0337491390525745e-05,
+      "loss": 1.5877,
+      "step": 128230
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.0336343460626e-05,
+      "loss": 1.1491,
+      "step": 128240
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.033519553072626e-05,
+      "loss": 1.2719,
+      "step": 128250
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.03340476008265e-05,
+      "loss": 1.271,
+      "step": 128260
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 6.033289967092676e-05,
+      "loss": 1.2501,
+      "step": 128270
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 6.0331751741027014e-05,
+      "loss": 1.2985,
+      "step": 128280
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 6.033060381112726e-05,
+      "loss": 1.3285,
+      "step": 128290
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 6.0329455881227514e-05,
+      "loss": 1.3334,
+      "step": 128300
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 6.032830795132777e-05,
+      "loss": 1.1548,
+      "step": 128310
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 6.0327160021428014e-05,
+      "loss": 1.222,
+      "step": 128320
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 6.032601209152827e-05,
+      "loss": 1.373,
+      "step": 128330
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 6.032486416162853e-05,
+      "loss": 1.2933,
+      "step": 128340
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 6.032371623172878e-05,
+      "loss": 1.2046,
+      "step": 128350
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 6.032256830182903e-05,
+      "loss": 1.3132,
+      "step": 128360
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 6.032142037192928e-05,
+      "loss": 1.2575,
+      "step": 128370
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 6.032027244202953e-05,
+      "loss": 1.1368,
+      "step": 128380
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 6.031912451212979e-05,
+      "loss": 1.1207,
+      "step": 128390
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 6.0317976582230046e-05,
+      "loss": 1.2373,
+      "step": 128400
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 6.031682865233029e-05,
+      "loss": 1.0856,
+      "step": 128410
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 6.0315680722430546e-05,
+      "loss": 1.2898,
+      "step": 128420
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 6.03145327925308e-05,
+      "loss": 1.2764,
+      "step": 128430
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 6.0313384862631046e-05,
+      "loss": 1.1798,
+      "step": 128440
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 6.03122369327313e-05,
+      "loss": 1.2434,
+      "step": 128450
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 6.031108900283156e-05,
+      "loss": 1.2984,
+      "step": 128460
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 6.03099410729318e-05,
+      "loss": 1.1983,
+      "step": 128470
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 6.030879314303206e-05,
+      "loss": 1.2123,
+      "step": 128480
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 6.0307645213132316e-05,
+      "loss": 1.0694,
+      "step": 128490
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 6.0306497283232565e-05,
+      "loss": 1.302,
+      "step": 128500
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 6.0305349353332815e-05,
+      "loss": 1.2048,
+      "step": 128510
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 6.030420142343307e-05,
+      "loss": 1.2898,
+      "step": 128520
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 6.030305349353332e-05,
+      "loss": 1.2983,
+      "step": 128530
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 6.030190556363358e-05,
+      "loss": 1.2339,
+      "step": 128540
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 6.0300757633733835e-05,
+      "loss": 1.2404,
+      "step": 128550
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 6.029960970383408e-05,
+      "loss": 1.3,
+      "step": 128560
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 6.0298461773934335e-05,
+      "loss": 1.2495,
+      "step": 128570
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 6.029731384403459e-05,
+      "loss": 1.2507,
+      "step": 128580
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 6.0296165914134834e-05,
+      "loss": 1.287,
+      "step": 128590
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 6.029501798423509e-05,
+      "loss": 1.247,
+      "step": 128600
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 6.029387005433535e-05,
+      "loss": 1.2009,
+      "step": 128610
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 6.029272212443559e-05,
+      "loss": 1.2426,
+      "step": 128620
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 6.029157419453585e-05,
+      "loss": 1.3729,
+      "step": 128630
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 6.0290426264636104e-05,
+      "loss": 1.2733,
+      "step": 128640
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 6.0289278334736354e-05,
+      "loss": 1.2505,
+      "step": 128650
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 6.0288130404836604e-05,
+      "loss": 1.2902,
+      "step": 128660
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 6.028698247493686e-05,
+      "loss": 1.354,
+      "step": 128670
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 6.028583454503711e-05,
+      "loss": 1.4119,
+      "step": 128680
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 6.028468661513737e-05,
+      "loss": 1.1663,
+      "step": 128690
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 6.0283538685237623e-05,
+      "loss": 1.1465,
+      "step": 128700
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 6.0282390755337867e-05,
+      "loss": 1.3579,
+      "step": 128710
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 6.028124282543812e-05,
+      "loss": 1.2414,
+      "step": 128720
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 6.028009489553838e-05,
+      "loss": 1.2856,
+      "step": 128730
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 6.027894696563862e-05,
+      "loss": 1.4213,
+      "step": 128740
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 6.027779903573888e-05,
+      "loss": 1.3237,
+      "step": 128750
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 6.0276651105839136e-05,
+      "loss": 1.2845,
+      "step": 128760
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 6.027550317593938e-05,
+      "loss": 1.2722,
+      "step": 128770
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 6.0274355246039636e-05,
+      "loss": 1.2667,
+      "step": 128780
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 6.027320731613989e-05,
+      "loss": 1.3458,
+      "step": 128790
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 6.027205938624014e-05,
+      "loss": 1.1922,
+      "step": 128800
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 6.027091145634039e-05,
+      "loss": 1.3425,
+      "step": 128810
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 6.026976352644065e-05,
+      "loss": 1.1006,
+      "step": 128820
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 6.02686155965409e-05,
+      "loss": 1.2939,
+      "step": 128830
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 6.0267467666641155e-05,
+      "loss": 1.1207,
+      "step": 128840
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 6.02663197367414e-05,
+      "loss": 1.2779,
+      "step": 128850
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 6.0265171806841655e-05,
+      "loss": 1.1772,
+      "step": 128860
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 6.026402387694191e-05,
+      "loss": 1.1642,
+      "step": 128870
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 6.0262875947042155e-05,
+      "loss": 1.341,
+      "step": 128880
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 6.026172801714241e-05,
+      "loss": 1.2421,
+      "step": 128890
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 6.026058008724267e-05,
+      "loss": 1.321,
+      "step": 128900
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 6.025943215734292e-05,
+      "loss": 1.3145,
+      "step": 128910
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.025828422744317e-05,
+      "loss": 1.3482,
+      "step": 128920
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.0257136297543424e-05,
+      "loss": 1.1465,
+      "step": 128930
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.0255988367643674e-05,
+      "loss": 1.3112,
+      "step": 128940
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.025484043774393e-05,
+      "loss": 1.2785,
+      "step": 128950
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.025369250784418e-05,
+      "loss": 1.2065,
+      "step": 128960
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.025254457794443e-05,
+      "loss": 1.159,
+      "step": 128970
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.025139664804469e-05,
+      "loss": 1.2732,
+      "step": 128980
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.0250248718144944e-05,
+      "loss": 1.1829,
+      "step": 128990
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.024910078824519e-05,
+      "loss": 1.2443,
+      "step": 129000
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.0247952858345443e-05,
+      "loss": 1.232,
+      "step": 129010
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.02468049284457e-05,
+      "loss": 1.3439,
+      "step": 129020
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.024565699854594e-05,
+      "loss": 1.1604,
+      "step": 129030
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.02445090686462e-05,
+      "loss": 1.2852,
+      "step": 129040
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.0243361138746456e-05,
+      "loss": 1.1983,
+      "step": 129050
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.0242213208846706e-05,
+      "loss": 1.4271,
+      "step": 129060
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.0241065278946956e-05,
+      "loss": 1.3689,
+      "step": 129070
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.023991734904721e-05,
+      "loss": 1.337,
+      "step": 129080
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.023876941914746e-05,
+      "loss": 1.2364,
+      "step": 129090
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.023762148924772e-05,
+      "loss": 1.2805,
+      "step": 129100
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.023647355934797e-05,
+      "loss": 1.3822,
+      "step": 129110
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.023532562944822e-05,
+      "loss": 1.2935,
+      "step": 129120
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.0234177699548476e-05,
+      "loss": 1.3721,
+      "step": 129130
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.023302976964873e-05,
+      "loss": 1.3958,
+      "step": 129140
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.0231881839748975e-05,
+      "loss": 1.3036,
+      "step": 129150
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.023073390984923e-05,
+      "loss": 1.3467,
+      "step": 129160
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.022958597994949e-05,
+      "loss": 1.187,
+      "step": 129170
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.022843805004973e-05,
+      "loss": 1.3804,
+      "step": 129180
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.022729012014999e-05,
+      "loss": 1.2837,
+      "step": 129190
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.0226142190250245e-05,
+      "loss": 1.2949,
+      "step": 129200
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.0224994260350495e-05,
+      "loss": 1.3553,
+      "step": 129210
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.0223846330450745e-05,
+      "loss": 1.3633,
+      "step": 129220
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.0222698400551e-05,
+      "loss": 1.2231,
+      "step": 129230
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.022155047065125e-05,
+      "loss": 1.3915,
+      "step": 129240
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.022040254075151e-05,
+      "loss": 1.3058,
+      "step": 129250
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.021925461085176e-05,
+      "loss": 1.3406,
+      "step": 129260
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.021810668095201e-05,
+      "loss": 1.2271,
+      "step": 129270
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.0216958751052264e-05,
+      "loss": 1.2608,
+      "step": 129280
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.021581082115252e-05,
+      "loss": 1.2919,
+      "step": 129290
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.0214662891252764e-05,
+      "loss": 1.4033,
+      "step": 129300
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.021351496135302e-05,
+      "loss": 1.2585,
+      "step": 129310
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.021236703145328e-05,
+      "loss": 1.1682,
+      "step": 129320
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.021121910155352e-05,
+      "loss": 1.2615,
+      "step": 129330
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.021007117165378e-05,
+      "loss": 1.3397,
+      "step": 129340
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.020892324175403e-05,
+      "loss": 1.335,
+      "step": 129350
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.020777531185428e-05,
+      "loss": 1.2807,
+      "step": 129360
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.020662738195453e-05,
+      "loss": 1.2782,
+      "step": 129370
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.020547945205479e-05,
+      "loss": 1.3226,
+      "step": 129380
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.020433152215504e-05,
+      "loss": 1.1833,
+      "step": 129390
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.0203183592255296e-05,
+      "loss": 1.2231,
+      "step": 129400
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.0202035662355546e-05,
+      "loss": 1.3251,
+      "step": 129410
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.0200887732455796e-05,
+      "loss": 1.3561,
+      "step": 129420
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.019973980255605e-05,
+      "loss": 1.3014,
+      "step": 129430
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.019859187265631e-05,
+      "loss": 1.2157,
+      "step": 129440
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.019744394275655e-05,
+      "loss": 1.1922,
+      "step": 129450
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.019629601285681e-05,
+      "loss": 1.2248,
+      "step": 129460
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.0195148082957066e-05,
+      "loss": 1.4354,
+      "step": 129470
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.019400015305731e-05,
+      "loss": 1.2655,
+      "step": 129480
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.0192852223157565e-05,
+      "loss": 1.2438,
+      "step": 129490
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.019170429325782e-05,
+      "loss": 1.3627,
+      "step": 129500
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.019055636335807e-05,
+      "loss": 1.2081,
+      "step": 129510
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.018940843345832e-05,
+      "loss": 1.248,
+      "step": 129520
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.018826050355858e-05,
+      "loss": 1.3124,
+      "step": 129530
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.018711257365883e-05,
+      "loss": 1.2268,
+      "step": 129540
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.0185964643759085e-05,
+      "loss": 1.2742,
+      "step": 129550
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.0184816713859335e-05,
+      "loss": 1.172,
+      "step": 129560
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.0183668783959584e-05,
+      "loss": 1.15,
+      "step": 129570
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.018252085405984e-05,
+      "loss": 1.2079,
+      "step": 129580
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.01813729241601e-05,
+      "loss": 1.2267,
+      "step": 129590
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.018022499426034e-05,
+      "loss": 1.4855,
+      "step": 129600
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.01790770643606e-05,
+      "loss": 1.2729,
+      "step": 129610
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.0177929134460854e-05,
+      "loss": 1.2233,
+      "step": 129620
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.01767812045611e-05,
+      "loss": 1.3708,
+      "step": 129630
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.0175633274661354e-05,
+      "loss": 1.1994,
+      "step": 129640
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.017448534476161e-05,
+      "loss": 1.3532,
+      "step": 129650
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.017333741486186e-05,
+      "loss": 1.0971,
+      "step": 129660
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.017218948496211e-05,
+      "loss": 1.1571,
+      "step": 129670
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.017104155506237e-05,
+      "loss": 1.4601,
+      "step": 129680
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.0169893625162617e-05,
+      "loss": 1.2367,
+      "step": 129690
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.016874569526287e-05,
+      "loss": 1.1865,
+      "step": 129700
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.016759776536312e-05,
+      "loss": 1.2673,
+      "step": 129710
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.016644983546337e-05,
+      "loss": 1.2111,
+      "step": 129720
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.016530190556363e-05,
+      "loss": 1.2966,
+      "step": 129730
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.0164153975663886e-05,
+      "loss": 1.2949,
+      "step": 129740
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.016300604576413e-05,
+      "loss": 1.3447,
+      "step": 129750
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.0161858115864386e-05,
+      "loss": 1.3392,
+      "step": 129760
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.016071018596464e-05,
+      "loss": 1.4041,
+      "step": 129770
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.0159562256064886e-05,
+      "loss": 1.3363,
+      "step": 129780
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 6.015841432616514e-05,
+      "loss": 1.1787,
+      "step": 129790
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 6.01572663962654e-05,
+      "loss": 1.2388,
+      "step": 129800
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 6.015611846636565e-05,
+      "loss": 1.1916,
+      "step": 129810
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 6.01549705364659e-05,
+      "loss": 1.268,
+      "step": 129820
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 6.0153822606566155e-05,
+      "loss": 1.4321,
+      "step": 129830
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 6.0152674676666405e-05,
+      "loss": 1.305,
+      "step": 129840
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 6.015152674676666e-05,
+      "loss": 1.2871,
+      "step": 129850
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 6.015037881686691e-05,
+      "loss": 1.3135,
+      "step": 129860
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 6.014923088696716e-05,
+      "loss": 1.2618,
+      "step": 129870
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 6.014808295706742e-05,
+      "loss": 1.3033,
+      "step": 129880
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 6.0146935027167675e-05,
+      "loss": 1.2489,
+      "step": 129890
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 6.014578709726792e-05,
+      "loss": 1.2214,
+      "step": 129900
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 6.0144639167368174e-05,
+      "loss": 1.2552,
+      "step": 129910
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 6.014349123746843e-05,
+      "loss": 1.1994,
+      "step": 129920
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 6.0142343307568674e-05,
+      "loss": 1.2559,
+      "step": 129930
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 6.014119537766893e-05,
+      "loss": 1.1303,
+      "step": 129940
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 6.014004744776919e-05,
+      "loss": 1.2487,
+      "step": 129950
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 6.013889951786944e-05,
+      "loss": 1.2144,
+      "step": 129960
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 6.013775158796969e-05,
+      "loss": 1.4027,
+      "step": 129970
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 6.0136603658069944e-05,
+      "loss": 1.2277,
+      "step": 129980
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 6.0135455728170193e-05,
+      "loss": 1.185,
+      "step": 129990
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 6.013430779827045e-05,
+      "loss": 1.2197,
+      "step": 130000
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.01331598683707e-05,
+      "loss": 1.2778,
+      "step": 130010
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.013201193847095e-05,
+      "loss": 1.3158,
+      "step": 130020
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.0130864008571206e-05,
+      "loss": 1.2937,
+      "step": 130030
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.012971607867146e-05,
+      "loss": 1.2363,
+      "step": 130040
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.0128568148771706e-05,
+      "loss": 1.1566,
+      "step": 130050
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.012742021887196e-05,
+      "loss": 1.2755,
+      "step": 130060
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.012627228897222e-05,
+      "loss": 1.3541,
+      "step": 130070
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.012512435907246e-05,
+      "loss": 1.2774,
+      "step": 130080
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.012397642917272e-05,
+      "loss": 1.1348,
+      "step": 130090
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.0122828499272976e-05,
+      "loss": 1.1677,
+      "step": 130100
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.0121680569373226e-05,
+      "loss": 1.2098,
+      "step": 130110
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.0120532639473475e-05,
+      "loss": 1.3483,
+      "step": 130120
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.011938470957373e-05,
+      "loss": 1.2437,
+      "step": 130130
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.011823677967398e-05,
+      "loss": 1.1873,
+      "step": 130140
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.011708884977424e-05,
+      "loss": 1.3315,
+      "step": 130150
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.011594091987449e-05,
+      "loss": 1.4983,
+      "step": 130160
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.011479298997474e-05,
+      "loss": 1.2135,
+      "step": 130170
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.0113645060074995e-05,
+      "loss": 1.1308,
+      "step": 130180
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.011249713017525e-05,
+      "loss": 1.2474,
+      "step": 130190
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.0111349200275495e-05,
+      "loss": 1.2421,
+      "step": 130200
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.011020127037575e-05,
+      "loss": 1.3005,
+      "step": 130210
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 6.0109053340476e-05,
+      "loss": 1.4422,
+      "step": 130220
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.010790541057625e-05,
+      "loss": 1.3508,
+      "step": 130230
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.010675748067651e-05,
+      "loss": 1.329,
+      "step": 130240
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.010560955077676e-05,
+      "loss": 1.2623,
+      "step": 130250
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.0104461620877014e-05,
+      "loss": 1.1628,
+      "step": 130260
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.0103313690977264e-05,
+      "loss": 1.334,
+      "step": 130270
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.0102165761077514e-05,
+      "loss": 1.1433,
+      "step": 130280
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.010101783117777e-05,
+      "loss": 1.4066,
+      "step": 130290
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.009986990127803e-05,
+      "loss": 1.2422,
+      "step": 130300
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.009872197137827e-05,
+      "loss": 1.2759,
+      "step": 130310
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.009757404147853e-05,
+      "loss": 1.3637,
+      "step": 130320
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.009642611157878e-05,
+      "loss": 1.2737,
+      "step": 130330
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.0095278181679026e-05,
+      "loss": 1.3902,
+      "step": 130340
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.009413025177928e-05,
+      "loss": 1.2159,
+      "step": 130350
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.009298232187954e-05,
+      "loss": 1.2287,
+      "step": 130360
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.009183439197979e-05,
+      "loss": 1.2669,
+      "step": 130370
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.009068646208004e-05,
+      "loss": 1.2441,
+      "step": 130380
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.0089538532180296e-05,
+      "loss": 1.238,
+      "step": 130390
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.0088390602280546e-05,
+      "loss": 1.2024,
+      "step": 130400
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.00872426723808e-05,
+      "loss": 1.2389,
+      "step": 130410
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.008609474248105e-05,
+      "loss": 1.29,
+      "step": 130420
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.00849468125813e-05,
+      "loss": 1.2013,
+      "step": 130430
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.008379888268156e-05,
+      "loss": 1.2407,
+      "step": 130440
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.0082650952781815e-05,
+      "loss": 1.1856,
+      "step": 130450
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.008150302288206e-05,
+      "loss": 1.4886,
+      "step": 130460
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.0080355092982315e-05,
+      "loss": 1.3114,
+      "step": 130470
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.007920716308257e-05,
+      "loss": 1.3026,
+      "step": 130480
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.0078059233182815e-05,
+      "loss": 1.2494,
+      "step": 130490
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.007691130328307e-05,
+      "loss": 1.2134,
+      "step": 130500
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.007576337338333e-05,
+      "loss": 1.3931,
+      "step": 130510
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.007461544348358e-05,
+      "loss": 1.2774,
+      "step": 130520
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.007346751358383e-05,
+      "loss": 1.3904,
+      "step": 130530
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.0072319583684085e-05,
+      "loss": 1.3509,
+      "step": 130540
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.0071171653784334e-05,
+      "loss": 1.3885,
+      "step": 130550
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.007002372388459e-05,
+      "loss": 1.3333,
+      "step": 130560
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.006887579398484e-05,
+      "loss": 1.1595,
+      "step": 130570
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.006772786408509e-05,
+      "loss": 1.2818,
+      "step": 130580
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.006657993418535e-05,
+      "loss": 1.3905,
+      "step": 130590
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.0065432004285604e-05,
+      "loss": 1.184,
+      "step": 130600
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.006428407438585e-05,
+      "loss": 1.1095,
+      "step": 130610
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.0063136144486104e-05,
+      "loss": 1.2096,
+      "step": 130620
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.006198821458636e-05,
+      "loss": 1.2678,
+      "step": 130630
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.0060840284686603e-05,
+      "loss": 1.3032,
+      "step": 130640
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.005969235478686e-05,
+      "loss": 1.3019,
+      "step": 130650
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.005854442488712e-05,
+      "loss": 1.2333,
+      "step": 130660
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.0057396494987367e-05,
+      "loss": 1.3113,
+      "step": 130670
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.0056248565087616e-05,
+      "loss": 1.1853,
+      "step": 130680
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.005510063518787e-05,
+      "loss": 1.3078,
+      "step": 130690
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.005395270528812e-05,
+      "loss": 1.3006,
+      "step": 130700
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.005280477538838e-05,
+      "loss": 1.3132,
+      "step": 130710
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.005165684548863e-05,
+      "loss": 1.1167,
+      "step": 130720
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.005050891558888e-05,
+      "loss": 1.2627,
+      "step": 130730
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.0049360985689136e-05,
+      "loss": 1.1471,
+      "step": 130740
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.004821305578939e-05,
+      "loss": 1.2766,
+      "step": 130750
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.0047065125889636e-05,
+      "loss": 1.249,
+      "step": 130760
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.004591719598989e-05,
+      "loss": 1.3754,
+      "step": 130770
+    },
+    {
+      "epoch": 6.0,
+      "eval_cer": 0.6698123939387187,
+      "eval_em": 0.007357212003872217,
+      "eval_f1": 0.007357212003872217,
+      "eval_loss": 1.2156100273132324,
+      "eval_runtime": 1714.3444,
+      "eval_samples_per_second": 6.026,
+      "eval_steps_per_second": 0.754,
+      "eval_wer": 0.9926427879961278,
+      "step": 130771
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.004476926609015e-05,
+      "loss": 1.3006,
+      "step": 130780
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.004362133619039e-05,
+      "loss": 1.2517,
+      "step": 130790
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.004247340629065e-05,
+      "loss": 1.3235,
+      "step": 130800
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.0041325476390905e-05,
+      "loss": 1.191,
+      "step": 130810
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.0040177546491155e-05,
+      "loss": 1.2505,
+      "step": 130820
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.0039029616591405e-05,
+      "loss": 1.2141,
+      "step": 130830
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.003788168669166e-05,
+      "loss": 1.2687,
+      "step": 130840
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.003673375679191e-05,
+      "loss": 1.3275,
+      "step": 130850
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.003558582689217e-05,
+      "loss": 1.1088,
+      "step": 130860
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.003443789699242e-05,
+      "loss": 1.1989,
+      "step": 130870
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.003328996709267e-05,
+      "loss": 1.143,
+      "step": 130880
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 6.0032142037192924e-05,
+      "loss": 1.1099,
+      "step": 130890
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 6.003099410729318e-05,
+      "loss": 1.3726,
+      "step": 130900
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 6.0029846177393424e-05,
+      "loss": 1.2167,
+      "step": 130910
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 6.002869824749368e-05,
+      "loss": 1.1205,
+      "step": 130920
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 6.002755031759394e-05,
+      "loss": 1.2237,
+      "step": 130930
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 6.002640238769418e-05,
+      "loss": 1.166,
+      "step": 130940
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 6.002525445779444e-05,
+      "loss": 1.1325,
+      "step": 130950
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 6.0024106527894694e-05,
+      "loss": 1.2262,
+      "step": 130960
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 6.0022958597994943e-05,
+      "loss": 1.0841,
+      "step": 130970
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 6.002181066809519e-05,
+      "loss": 1.2709,
+      "step": 130980
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 6.002066273819545e-05,
+      "loss": 1.1979,
+      "step": 130990
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 6.00195148082957e-05,
+      "loss": 1.1153,
+      "step": 131000
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 6.0018366878395956e-05,
+      "loss": 1.1822,
+      "step": 131010
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 6.0017218948496206e-05,
+      "loss": 1.3262,
+      "step": 131020
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 6.0016071018596456e-05,
+      "loss": 1.2548,
+      "step": 131030
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 6.001492308869671e-05,
+      "loss": 1.2089,
+      "step": 131040
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 6.001377515879697e-05,
+      "loss": 1.1354,
+      "step": 131050
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 6.001262722889721e-05,
+      "loss": 1.2644,
+      "step": 131060
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 6.001147929899747e-05,
+      "loss": 1.2436,
+      "step": 131070
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 6.0010331369097726e-05,
+      "loss": 1.3389,
+      "step": 131080
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 6.000918343919797e-05,
+      "loss": 1.1547,
+      "step": 131090
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 6.0008035509298225e-05,
+      "loss": 1.2382,
+      "step": 131100
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 6.000688757939848e-05,
+      "loss": 1.154,
+      "step": 131110
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 6.000573964949873e-05,
+      "loss": 1.1748,
+      "step": 131120
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 6.000459171959898e-05,
+      "loss": 1.2237,
+      "step": 131130
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 6.000344378969924e-05,
+      "loss": 1.1382,
+      "step": 131140
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 6.000229585979949e-05,
+      "loss": 1.0915,
+      "step": 131150
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 6.0001147929899745e-05,
+      "loss": 1.0924,
+      "step": 131160
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 1.25,
+      "step": 131170
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 5.9998852070100245e-05,
+      "loss": 1.3355,
+      "step": 131180
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 5.99977041402005e-05,
+      "loss": 1.256,
+      "step": 131190
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 5.999655621030076e-05,
+      "loss": 1.321,
+      "step": 131200
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 5.9995408280401e-05,
+      "loss": 1.1096,
+      "step": 131210
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 5.999426035050126e-05,
+      "loss": 1.2195,
+      "step": 131220
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 5.9993112420601514e-05,
+      "loss": 1.2089,
+      "step": 131230
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 5.999196449070176e-05,
+      "loss": 1.304,
+      "step": 131240
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 5.9990816560802014e-05,
+      "loss": 1.1898,
+      "step": 131250
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 5.998966863090227e-05,
+      "loss": 1.2049,
+      "step": 131260
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 5.998852070100252e-05,
+      "loss": 1.2588,
+      "step": 131270
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 5.998737277110277e-05,
+      "loss": 1.2228,
+      "step": 131280
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 5.998622484120303e-05,
+      "loss": 1.2374,
+      "step": 131290
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 5.998507691130328e-05,
+      "loss": 1.2225,
+      "step": 131300
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 5.998392898140353e-05,
+      "loss": 1.2533,
+      "step": 131310
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.998278105150378e-05,
+      "loss": 1.3457,
+      "step": 131320
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.998163312160403e-05,
+      "loss": 1.1429,
+      "step": 131330
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.998048519170429e-05,
+      "loss": 1.2429,
+      "step": 131340
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.9979337261804546e-05,
+      "loss": 1.2377,
+      "step": 131350
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.997818933190479e-05,
+      "loss": 1.1917,
+      "step": 131360
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.9977041402005046e-05,
+      "loss": 1.1325,
+      "step": 131370
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.99758934721053e-05,
+      "loss": 1.1933,
+      "step": 131380
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.9974745542205546e-05,
+      "loss": 1.1348,
+      "step": 131390
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.99735976123058e-05,
+      "loss": 1.3099,
+      "step": 131400
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.997244968240606e-05,
+      "loss": 1.2401,
+      "step": 131410
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.997130175250631e-05,
+      "loss": 1.2352,
+      "step": 131420
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.997015382260656e-05,
+      "loss": 1.1531,
+      "step": 131430
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.9969005892706815e-05,
+      "loss": 1.262,
+      "step": 131440
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.9967857962807065e-05,
+      "loss": 1.235,
+      "step": 131450
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.996671003290732e-05,
+      "loss": 1.2395,
+      "step": 131460
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.996556210300757e-05,
+      "loss": 1.2576,
+      "step": 131470
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.996441417310782e-05,
+      "loss": 1.1838,
+      "step": 131480
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.996326624320808e-05,
+      "loss": 1.1327,
+      "step": 131490
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.9962118313308335e-05,
+      "loss": 1.2113,
+      "step": 131500
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.996097038340858e-05,
+      "loss": 1.1586,
+      "step": 131510
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.9959822453508835e-05,
+      "loss": 1.2011,
+      "step": 131520
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.995867452360909e-05,
+      "loss": 1.218,
+      "step": 131530
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.9957526593709334e-05,
+      "loss": 1.3915,
+      "step": 131540
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.995637866380959e-05,
+      "loss": 1.088,
+      "step": 131550
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.995523073390985e-05,
+      "loss": 1.3769,
+      "step": 131560
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.99540828040101e-05,
+      "loss": 1.2545,
+      "step": 131570
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.995293487411035e-05,
+      "loss": 1.2377,
+      "step": 131580
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.99517869442106e-05,
+      "loss": 1.2523,
+      "step": 131590
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.9950639014310854e-05,
+      "loss": 1.1516,
+      "step": 131600
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.994949108441111e-05,
+      "loss": 1.1788,
+      "step": 131610
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.994834315451135e-05,
+      "loss": 1.1979,
+      "step": 131620
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.994719522461161e-05,
+      "loss": 1.1838,
+      "step": 131630
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.994604729471187e-05,
+      "loss": 1.1049,
+      "step": 131640
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.994489936481211e-05,
+      "loss": 1.1558,
+      "step": 131650
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.9943751434912366e-05,
+      "loss": 1.12,
+      "step": 131660
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.994260350501262e-05,
+      "loss": 1.1816,
+      "step": 131670
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.994145557511287e-05,
+      "loss": 1.2809,
+      "step": 131680
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.994030764521312e-05,
+      "loss": 1.175,
+      "step": 131690
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.993915971531338e-05,
+      "loss": 1.258,
+      "step": 131700
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.993801178541363e-05,
+      "loss": 1.2639,
+      "step": 131710
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.9936863855513886e-05,
+      "loss": 1.1218,
+      "step": 131720
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.9935715925614136e-05,
+      "loss": 1.2699,
+      "step": 131730
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.9934567995714386e-05,
+      "loss": 1.1331,
+      "step": 131740
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.993342006581464e-05,
+      "loss": 1.3144,
+      "step": 131750
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.99322721359149e-05,
+      "loss": 1.298,
+      "step": 131760
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.993112420601514e-05,
+      "loss": 1.1905,
+      "step": 131770
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.99299762761154e-05,
+      "loss": 1.0831,
+      "step": 131780
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.9928828346215655e-05,
+      "loss": 1.0902,
+      "step": 131790
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.99276804163159e-05,
+      "loss": 1.0654,
+      "step": 131800
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.9926532486416155e-05,
+      "loss": 1.1086,
+      "step": 131810
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.992538455651641e-05,
+      "loss": 1.2308,
+      "step": 131820
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.992423662661666e-05,
+      "loss": 1.3301,
+      "step": 131830
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.992308869671691e-05,
+      "loss": 1.0937,
+      "step": 131840
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.992194076681717e-05,
+      "loss": 1.208,
+      "step": 131850
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.992079283691742e-05,
+      "loss": 1.1911,
+      "step": 131860
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.9919644907017674e-05,
+      "loss": 1.0372,
+      "step": 131870
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.9918496977117924e-05,
+      "loss": 1.2585,
+      "step": 131880
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.9917349047218174e-05,
+      "loss": 1.3056,
+      "step": 131890
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.991620111731843e-05,
+      "loss": 1.2154,
+      "step": 131900
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.991505318741869e-05,
+      "loss": 1.2341,
+      "step": 131910
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.991390525751893e-05,
+      "loss": 1.09,
+      "step": 131920
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.991275732761919e-05,
+      "loss": 1.257,
+      "step": 131930
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.9911609397719444e-05,
+      "loss": 1.2293,
+      "step": 131940
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.991046146781969e-05,
+      "loss": 1.3578,
+      "step": 131950
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.990931353791994e-05,
+      "loss": 1.1898,
+      "step": 131960
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.99081656080202e-05,
+      "loss": 1.2095,
+      "step": 131970
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.990701767812045e-05,
+      "loss": 1.1171,
+      "step": 131980
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.99058697482207e-05,
+      "loss": 1.2938,
+      "step": 131990
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.9904721818320956e-05,
+      "loss": 1.2102,
+      "step": 132000
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.9903573888421206e-05,
+      "loss": 1.1395,
+      "step": 132010
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.990242595852146e-05,
+      "loss": 1.2052,
+      "step": 132020
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.990127802862171e-05,
+      "loss": 1.3304,
+      "step": 132030
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.990013009872196e-05,
+      "loss": 1.2548,
+      "step": 132040
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.989898216882222e-05,
+      "loss": 1.1455,
+      "step": 132050
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.9897834238922476e-05,
+      "loss": 1.2929,
+      "step": 132060
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.989668630902272e-05,
+      "loss": 1.1665,
+      "step": 132070
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.9895538379122975e-05,
+      "loss": 1.2027,
+      "step": 132080
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.989439044922323e-05,
+      "loss": 1.2696,
+      "step": 132090
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.9893242519323475e-05,
+      "loss": 1.2854,
+      "step": 132100
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.989209458942373e-05,
+      "loss": 1.1012,
+      "step": 132110
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.989094665952399e-05,
+      "loss": 1.1932,
+      "step": 132120
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.988979872962424e-05,
+      "loss": 1.18,
+      "step": 132130
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.988865079972449e-05,
+      "loss": 1.18,
+      "step": 132140
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.9887502869824745e-05,
+      "loss": 1.1865,
+      "step": 132150
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.9886354939924995e-05,
+      "loss": 1.2232,
+      "step": 132160
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.988520701002525e-05,
+      "loss": 1.0631,
+      "step": 132170
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.98840590801255e-05,
+      "loss": 1.1752,
+      "step": 132180
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.988291115022575e-05,
+      "loss": 1.3909,
+      "step": 132190
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.988176322032601e-05,
+      "loss": 1.1527,
+      "step": 132200
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.9880615290426264e-05,
+      "loss": 1.201,
+      "step": 132210
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.987946736052651e-05,
+      "loss": 1.3527,
+      "step": 132220
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.9878319430626764e-05,
+      "loss": 1.3101,
+      "step": 132230
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.987717150072702e-05,
+      "loss": 1.1546,
+      "step": 132240
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.9876023570827264e-05,
+      "loss": 1.0901,
+      "step": 132250
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.987487564092752e-05,
+      "loss": 1.1397,
+      "step": 132260
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.987372771102778e-05,
+      "loss": 1.2679,
+      "step": 132270
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.987257978112803e-05,
+      "loss": 1.1959,
+      "step": 132280
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.9871431851228277e-05,
+      "loss": 1.317,
+      "step": 132290
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.987028392132853e-05,
+      "loss": 1.1035,
+      "step": 132300
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.986913599142878e-05,
+      "loss": 1.1472,
+      "step": 132310
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.986798806152904e-05,
+      "loss": 1.0938,
+      "step": 132320
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.986684013162929e-05,
+      "loss": 1.1917,
+      "step": 132330
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.986569220172954e-05,
+      "loss": 1.2165,
+      "step": 132340
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.9864544271829796e-05,
+      "loss": 1.3457,
+      "step": 132350
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.986339634193005e-05,
+      "loss": 1.4151,
+      "step": 132360
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.9862248412030296e-05,
+      "loss": 1.1197,
+      "step": 132370
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.986110048213055e-05,
+      "loss": 1.183,
+      "step": 132380
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.985995255223081e-05,
+      "loss": 1.0994,
+      "step": 132390
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.985880462233105e-05,
+      "loss": 1.1439,
+      "step": 132400
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.985765669243131e-05,
+      "loss": 1.2909,
+      "step": 132410
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.9856508762531565e-05,
+      "loss": 1.2737,
+      "step": 132420
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.9855360832631815e-05,
+      "loss": 1.1639,
+      "step": 132430
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.9854212902732065e-05,
+      "loss": 1.2564,
+      "step": 132440
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.985306497283232e-05,
+      "loss": 1.1913,
+      "step": 132450
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.985191704293257e-05,
+      "loss": 1.0607,
+      "step": 132460
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.985076911303283e-05,
+      "loss": 1.2583,
+      "step": 132470
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.984962118313308e-05,
+      "loss": 1.3908,
+      "step": 132480
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.984847325323333e-05,
+      "loss": 1.2417,
+      "step": 132490
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.9847325323333585e-05,
+      "loss": 1.2135,
+      "step": 132500
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.984617739343384e-05,
+      "loss": 1.2314,
+      "step": 132510
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.9845029463534084e-05,
+      "loss": 1.3052,
+      "step": 132520
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.984388153363434e-05,
+      "loss": 1.2068,
+      "step": 132530
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.98427336037346e-05,
+      "loss": 1.2873,
+      "step": 132540
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.984158567383484e-05,
+      "loss": 1.2897,
+      "step": 132550
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.98404377439351e-05,
+      "loss": 1.2166,
+      "step": 132560
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.9839289814035354e-05,
+      "loss": 1.4277,
+      "step": 132570
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.9838141884135604e-05,
+      "loss": 1.2162,
+      "step": 132580
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.9836993954235854e-05,
+      "loss": 1.2755,
+      "step": 132590
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.983584602433611e-05,
+      "loss": 1.3169,
+      "step": 132600
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.983469809443636e-05,
+      "loss": 1.2627,
+      "step": 132610
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.983355016453662e-05,
+      "loss": 1.3002,
+      "step": 132620
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.9832402234636866e-05,
+      "loss": 1.2868,
+      "step": 132630
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.9831254304737116e-05,
+      "loss": 1.209,
+      "step": 132640
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.983010637483737e-05,
+      "loss": 1.3129,
+      "step": 132650
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.982895844493763e-05,
+      "loss": 1.0976,
+      "step": 132660
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.982781051503787e-05,
+      "loss": 1.1702,
+      "step": 132670
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.982666258513813e-05,
+      "loss": 1.2212,
+      "step": 132680
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.9825514655238386e-05,
+      "loss": 1.1666,
+      "step": 132690
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.982436672533863e-05,
+      "loss": 1.3476,
+      "step": 132700
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.9823218795438886e-05,
+      "loss": 1.1689,
+      "step": 132710
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.982207086553914e-05,
+      "loss": 1.2458,
+      "step": 132720
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.982092293563939e-05,
+      "loss": 1.1549,
+      "step": 132730
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.981977500573964e-05,
+      "loss": 1.2438,
+      "step": 132740
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.98186270758399e-05,
+      "loss": 1.1363,
+      "step": 132750
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.981747914594015e-05,
+      "loss": 1.2265,
+      "step": 132760
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.9816331216040405e-05,
+      "loss": 1.2413,
+      "step": 132770
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.9815183286140655e-05,
+      "loss": 1.1821,
+      "step": 132780
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.9814035356240905e-05,
+      "loss": 1.25,
+      "step": 132790
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.981288742634116e-05,
+      "loss": 1.1931,
+      "step": 132800
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.981173949644142e-05,
+      "loss": 1.2685,
+      "step": 132810
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.981059156654166e-05,
+      "loss": 1.2143,
+      "step": 132820
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.980944363664192e-05,
+      "loss": 1.2018,
+      "step": 132830
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.9808295706742174e-05,
+      "loss": 1.231,
+      "step": 132840
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.980714777684242e-05,
+      "loss": 1.0943,
+      "step": 132850
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.9805999846942674e-05,
+      "loss": 1.1174,
+      "step": 132860
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.980485191704293e-05,
+      "loss": 1.1319,
+      "step": 132870
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.980370398714318e-05,
+      "loss": 1.1909,
+      "step": 132880
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.980255605724343e-05,
+      "loss": 1.2753,
+      "step": 132890
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.980140812734369e-05,
+      "loss": 1.2706,
+      "step": 132900
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.980026019744394e-05,
+      "loss": 1.2217,
+      "step": 132910
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.9799112267544194e-05,
+      "loss": 1.2059,
+      "step": 132920
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.9797964337644443e-05,
+      "loss": 1.1575,
+      "step": 132930
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.979681640774469e-05,
+      "loss": 1.1959,
+      "step": 132940
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.979566847784495e-05,
+      "loss": 1.2569,
+      "step": 132950
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.979452054794519e-05,
+      "loss": 1.1138,
+      "step": 132960
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.979337261804545e-05,
+      "loss": 1.2711,
+      "step": 132970
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.9792224688145706e-05,
+      "loss": 1.1161,
+      "step": 132980
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.9791076758245956e-05,
+      "loss": 1.2348,
+      "step": 132990
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.9789928828346206e-05,
+      "loss": 1.2493,
+      "step": 133000
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.978878089844646e-05,
+      "loss": 1.1314,
+      "step": 133010
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.978763296854671e-05,
+      "loss": 1.2011,
+      "step": 133020
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.978648503864697e-05,
+      "loss": 1.2416,
+      "step": 133030
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.978533710874722e-05,
+      "loss": 1.2551,
+      "step": 133040
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.978418917884747e-05,
+      "loss": 1.2271,
+      "step": 133050
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.9783041248947725e-05,
+      "loss": 1.2373,
+      "step": 133060
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.978189331904798e-05,
+      "loss": 1.2215,
+      "step": 133070
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.9780745389148225e-05,
+      "loss": 1.1643,
+      "step": 133080
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.977959745924848e-05,
+      "loss": 1.2715,
+      "step": 133090
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.977844952934874e-05,
+      "loss": 1.1287,
+      "step": 133100
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.977730159944898e-05,
+      "loss": 1.1673,
+      "step": 133110
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.977615366954924e-05,
+      "loss": 1.22,
+      "step": 133120
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.9775005739649495e-05,
+      "loss": 1.1476,
+      "step": 133130
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.9773857809749745e-05,
+      "loss": 1.1904,
+      "step": 133140
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.9772709879849994e-05,
+      "loss": 1.0739,
+      "step": 133150
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.977156194995025e-05,
+      "loss": 1.3689,
+      "step": 133160
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.97704140200505e-05,
+      "loss": 1.2656,
+      "step": 133170
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.976926609015076e-05,
+      "loss": 1.2072,
+      "step": 133180
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.976811816025101e-05,
+      "loss": 1.4371,
+      "step": 133190
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.976697023035126e-05,
+      "loss": 1.1693,
+      "step": 133200
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.9765822300451514e-05,
+      "loss": 1.1067,
+      "step": 133210
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.976467437055177e-05,
+      "loss": 1.1599,
+      "step": 133220
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.9763526440652014e-05,
+      "loss": 1.2326,
+      "step": 133230
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.976237851075227e-05,
+      "loss": 1.3011,
+      "step": 133240
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.976123058085253e-05,
+      "loss": 1.2741,
+      "step": 133250
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.976008265095277e-05,
+      "loss": 1.1164,
+      "step": 133260
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.9758934721053027e-05,
+      "loss": 1.3429,
+      "step": 133270
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.975778679115328e-05,
+      "loss": 1.1739,
+      "step": 133280
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.975663886125353e-05,
+      "loss": 1.192,
+      "step": 133290
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.975549093135378e-05,
+      "loss": 1.2131,
+      "step": 133300
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.975434300145404e-05,
+      "loss": 1.2803,
+      "step": 133310
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.975319507155429e-05,
+      "loss": 1.2266,
+      "step": 133320
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.9752047141654546e-05,
+      "loss": 1.2503,
+      "step": 133330
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.9750899211754796e-05,
+      "loss": 1.2376,
+      "step": 133340
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.9749751281855046e-05,
+      "loss": 1.1379,
+      "step": 133350
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.97486033519553e-05,
+      "loss": 1.1337,
+      "step": 133360
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.974745542205556e-05,
+      "loss": 1.4034,
+      "step": 133370
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.97463074921558e-05,
+      "loss": 1.2513,
+      "step": 133380
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.974515956225606e-05,
+      "loss": 1.1902,
+      "step": 133390
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.9744011632356315e-05,
+      "loss": 1.1197,
+      "step": 133400
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.974286370245656e-05,
+      "loss": 1.143,
+      "step": 133410
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.9741715772556815e-05,
+      "loss": 1.2332,
+      "step": 133420
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.974056784265707e-05,
+      "loss": 1.1973,
+      "step": 133430
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.973941991275732e-05,
+      "loss": 1.2163,
+      "step": 133440
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.973827198285757e-05,
+      "loss": 1.1645,
+      "step": 133450
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.973712405295783e-05,
+      "loss": 1.2096,
+      "step": 133460
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.973597612305808e-05,
+      "loss": 1.2255,
+      "step": 133470
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.9734828193158335e-05,
+      "loss": 1.1566,
+      "step": 133480
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.9733680263258584e-05,
+      "loss": 1.3568,
+      "step": 133490
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.9732532333358834e-05,
+      "loss": 1.1286,
+      "step": 133500
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.973138440345909e-05,
+      "loss": 1.1962,
+      "step": 133510
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.973023647355935e-05,
+      "loss": 1.1828,
+      "step": 133520
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.972908854365959e-05,
+      "loss": 1.2584,
+      "step": 133530
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.972794061375985e-05,
+      "loss": 1.2299,
+      "step": 133540
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.9726792683860104e-05,
+      "loss": 1.2126,
+      "step": 133550
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.972564475396035e-05,
+      "loss": 1.1767,
+      "step": 133560
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.9724496824060604e-05,
+      "loss": 1.1191,
+      "step": 133570
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.972334889416086e-05,
+      "loss": 1.1672,
+      "step": 133580
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.972220096426111e-05,
+      "loss": 1.3347,
+      "step": 133590
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.972105303436136e-05,
+      "loss": 1.1577,
+      "step": 133600
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.9719905104461616e-05,
+      "loss": 1.2186,
+      "step": 133610
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.9718757174561866e-05,
+      "loss": 1.1805,
+      "step": 133620
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.971760924466212e-05,
+      "loss": 1.2374,
+      "step": 133630
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.971646131476237e-05,
+      "loss": 1.3542,
+      "step": 133640
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.971531338486262e-05,
+      "loss": 1.2042,
+      "step": 133650
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.971416545496288e-05,
+      "loss": 1.1636,
+      "step": 133660
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.9713017525063136e-05,
+      "loss": 1.1656,
+      "step": 133670
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.971186959516338e-05,
+      "loss": 1.218,
+      "step": 133680
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.9710721665263636e-05,
+      "loss": 1.1814,
+      "step": 133690
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.970957373536389e-05,
+      "loss": 1.134,
+      "step": 133700
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.9708425805464135e-05,
+      "loss": 1.1584,
+      "step": 133710
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.970727787556439e-05,
+      "loss": 1.3088,
+      "step": 133720
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.970612994566465e-05,
+      "loss": 1.1488,
+      "step": 133730
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.97049820157649e-05,
+      "loss": 1.2637,
+      "step": 133740
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.970383408586515e-05,
+      "loss": 1.2042,
+      "step": 133750
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.9702686155965405e-05,
+      "loss": 1.2312,
+      "step": 133760
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.9701538226065655e-05,
+      "loss": 1.2149,
+      "step": 133770
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.970039029616591e-05,
+      "loss": 1.239,
+      "step": 133780
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.969924236626616e-05,
+      "loss": 1.2758,
+      "step": 133790
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.969809443636641e-05,
+      "loss": 1.2401,
+      "step": 133800
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.969694650646667e-05,
+      "loss": 1.2241,
+      "step": 133810
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.9695798576566924e-05,
+      "loss": 1.1729,
+      "step": 133820
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.969465064666717e-05,
+      "loss": 1.3164,
+      "step": 133830
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.9693502716767424e-05,
+      "loss": 1.2084,
+      "step": 133840
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.969235478686768e-05,
+      "loss": 1.1332,
+      "step": 133850
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.9691206856967924e-05,
+      "loss": 1.1049,
+      "step": 133860
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.969005892706818e-05,
+      "loss": 1.2808,
+      "step": 133870
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.968891099716844e-05,
+      "loss": 1.1962,
+      "step": 133880
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.968776306726869e-05,
+      "loss": 1.1812,
+      "step": 133890
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.968661513736894e-05,
+      "loss": 1.1656,
+      "step": 133900
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.9685467207469193e-05,
+      "loss": 1.2007,
+      "step": 133910
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.968431927756944e-05,
+      "loss": 1.1701,
+      "step": 133920
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.96831713476697e-05,
+      "loss": 1.1804,
+      "step": 133930
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.968202341776995e-05,
+      "loss": 1.2577,
+      "step": 133940
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.96808754878702e-05,
+      "loss": 1.1698,
+      "step": 133950
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.9679727557970456e-05,
+      "loss": 1.0875,
+      "step": 133960
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.967857962807071e-05,
+      "loss": 1.1182,
+      "step": 133970
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.9677431698170956e-05,
+      "loss": 1.2326,
+      "step": 133980
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.967628376827121e-05,
+      "loss": 1.3909,
+      "step": 133990
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.967513583837147e-05,
+      "loss": 1.1617,
+      "step": 134000
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.967398790847171e-05,
+      "loss": 1.0511,
+      "step": 134010
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.967283997857197e-05,
+      "loss": 1.2612,
+      "step": 134020
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.9671692048672226e-05,
+      "loss": 1.2373,
+      "step": 134030
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.9670544118772475e-05,
+      "loss": 1.1692,
+      "step": 134040
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.9669396188872725e-05,
+      "loss": 1.1591,
+      "step": 134050
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.966824825897298e-05,
+      "loss": 1.1533,
+      "step": 134060
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.966710032907323e-05,
+      "loss": 1.2975,
+      "step": 134070
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.966595239917349e-05,
+      "loss": 1.2113,
+      "step": 134080
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.966480446927374e-05,
+      "loss": 1.1435,
+      "step": 134090
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.966365653937399e-05,
+      "loss": 1.2458,
+      "step": 134100
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.9662508609474245e-05,
+      "loss": 1.0976,
+      "step": 134110
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.96613606795745e-05,
+      "loss": 1.1831,
+      "step": 134120
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.9660212749674744e-05,
+      "loss": 1.2418,
+      "step": 134130
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.9659064819775e-05,
+      "loss": 1.1678,
+      "step": 134140
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.965791688987526e-05,
+      "loss": 1.1794,
+      "step": 134150
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.96567689599755e-05,
+      "loss": 1.2689,
+      "step": 134160
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.965562103007576e-05,
+      "loss": 1.224,
+      "step": 134170
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.9654473100176014e-05,
+      "loss": 1.3154,
+      "step": 134180
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.9653325170276264e-05,
+      "loss": 1.2302,
+      "step": 134190
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.9652177240376514e-05,
+      "loss": 1.2228,
+      "step": 134200
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.965102931047677e-05,
+      "loss": 1.2513,
+      "step": 134210
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.964988138057702e-05,
+      "loss": 1.1411,
+      "step": 134220
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.964873345067728e-05,
+      "loss": 1.2141,
+      "step": 134230
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.964758552077753e-05,
+      "loss": 1.2012,
+      "step": 134240
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.9646437590877777e-05,
+      "loss": 1.1751,
+      "step": 134250
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.964528966097803e-05,
+      "loss": 1.2359,
+      "step": 134260
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.964414173107829e-05,
+      "loss": 1.2475,
+      "step": 134270
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.964299380117853e-05,
+      "loss": 1.1613,
+      "step": 134280
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.964184587127879e-05,
+      "loss": 1.2351,
+      "step": 134290
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.9640697941379046e-05,
+      "loss": 1.3427,
+      "step": 134300
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.963955001147929e-05,
+      "loss": 1.2213,
+      "step": 134310
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.9638402081579546e-05,
+      "loss": 1.3029,
+      "step": 134320
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.9637254151679796e-05,
+      "loss": 1.1539,
+      "step": 134330
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.963610622178005e-05,
+      "loss": 1.2892,
+      "step": 134340
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.96349582918803e-05,
+      "loss": 1.2501,
+      "step": 134350
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.963381036198055e-05,
+      "loss": 1.1704,
+      "step": 134360
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.963266243208081e-05,
+      "loss": 1.1399,
+      "step": 134370
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.9631514502181065e-05,
+      "loss": 1.2072,
+      "step": 134380
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.963036657228131e-05,
+      "loss": 1.0554,
+      "step": 134390
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.9629218642381565e-05,
+      "loss": 1.2545,
+      "step": 134400
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.962807071248182e-05,
+      "loss": 1.0287,
+      "step": 134410
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.9626922782582065e-05,
+      "loss": 1.2703,
+      "step": 134420
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.962577485268232e-05,
+      "loss": 1.17,
+      "step": 134430
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.962462692278258e-05,
+      "loss": 1.3225,
+      "step": 134440
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.962347899288283e-05,
+      "loss": 1.1629,
+      "step": 134450
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.962233106298308e-05,
+      "loss": 1.1733,
+      "step": 134460
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.9621183133083334e-05,
+      "loss": 1.1925,
+      "step": 134470
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.9620035203183584e-05,
+      "loss": 1.2692,
+      "step": 134480
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.961888727328384e-05,
+      "loss": 1.1652,
+      "step": 134490
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.961773934338409e-05,
+      "loss": 1.1778,
+      "step": 134500
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.961659141348434e-05,
+      "loss": 1.317,
+      "step": 134510
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.96154434835846e-05,
+      "loss": 1.1584,
+      "step": 134520
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.9614295553684854e-05,
+      "loss": 1.1475,
+      "step": 134530
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.96131476237851e-05,
+      "loss": 1.1753,
+      "step": 134540
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.9611999693885354e-05,
+      "loss": 1.1522,
+      "step": 134550
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.961085176398561e-05,
+      "loss": 1.142,
+      "step": 134560
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.960970383408585e-05,
+      "loss": 1.1235,
+      "step": 134570
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.960855590418611e-05,
+      "loss": 1.2755,
+      "step": 134580
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.9607407974286366e-05,
+      "loss": 1.1691,
+      "step": 134590
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.9606260044386616e-05,
+      "loss": 1.2269,
+      "step": 134600
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.9605112114486866e-05,
+      "loss": 1.1704,
+      "step": 134610
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.960396418458712e-05,
+      "loss": 1.316,
+      "step": 134620
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.960281625468737e-05,
+      "loss": 1.1297,
+      "step": 134630
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.960166832478763e-05,
+      "loss": 1.1609,
+      "step": 134640
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.960052039488788e-05,
+      "loss": 1.1608,
+      "step": 134650
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.959937246498813e-05,
+      "loss": 1.2112,
+      "step": 134660
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.9598224535088386e-05,
+      "loss": 1.2191,
+      "step": 134670
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.959707660518864e-05,
+      "loss": 1.2618,
+      "step": 134680
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.9595928675288885e-05,
+      "loss": 1.0907,
+      "step": 134690
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.959478074538914e-05,
+      "loss": 1.2979,
+      "step": 134700
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.95936328154894e-05,
+      "loss": 1.2531,
+      "step": 134710
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.959248488558964e-05,
+      "loss": 1.2601,
+      "step": 134720
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.95913369556899e-05,
+      "loss": 1.0861,
+      "step": 134730
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.9590189025790155e-05,
+      "loss": 1.2384,
+      "step": 134740
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.9589041095890405e-05,
+      "loss": 1.0944,
+      "step": 134750
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.9587893165990655e-05,
+      "loss": 1.2319,
+      "step": 134760
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.958674523609091e-05,
+      "loss": 1.1448,
+      "step": 134770
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.958559730619116e-05,
+      "loss": 1.2599,
+      "step": 134780
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.958444937629142e-05,
+      "loss": 1.1642,
+      "step": 134790
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.958330144639167e-05,
+      "loss": 1.2419,
+      "step": 134800
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.958215351649192e-05,
+      "loss": 1.2369,
+      "step": 134810
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.9581005586592174e-05,
+      "loss": 1.2691,
+      "step": 134820
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.957985765669243e-05,
+      "loss": 1.195,
+      "step": 134830
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.9578709726792674e-05,
+      "loss": 1.0639,
+      "step": 134840
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.957756179689293e-05,
+      "loss": 1.1755,
+      "step": 134850
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.957641386699319e-05,
+      "loss": 1.2056,
+      "step": 134860
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.957526593709343e-05,
+      "loss": 1.1581,
+      "step": 134870
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.957411800719369e-05,
+      "loss": 1.2061,
+      "step": 134880
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.9572970077293943e-05,
+      "loss": 1.17,
+      "step": 134890
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.957182214739419e-05,
+      "loss": 1.2102,
+      "step": 134900
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.957067421749444e-05,
+      "loss": 1.2325,
+      "step": 134910
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.95695262875947e-05,
+      "loss": 1.1624,
+      "step": 134920
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.956837835769495e-05,
+      "loss": 1.2266,
+      "step": 134930
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.9567230427795206e-05,
+      "loss": 1.2689,
+      "step": 134940
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.9566082497895456e-05,
+      "loss": 1.1836,
+      "step": 134950
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.9564934567995706e-05,
+      "loss": 1.1689,
+      "step": 134960
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.956378663809596e-05,
+      "loss": 1.2592,
+      "step": 134970
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.956263870819622e-05,
+      "loss": 1.2543,
+      "step": 134980
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.956149077829646e-05,
+      "loss": 1.1294,
+      "step": 134990
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.956034284839672e-05,
+      "loss": 1.1277,
+      "step": 135000
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.9559194918496976e-05,
+      "loss": 1.2441,
+      "step": 135010
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.955804698859722e-05,
+      "loss": 1.2946,
+      "step": 135020
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.9556899058697475e-05,
+      "loss": 1.1457,
+      "step": 135030
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.955575112879773e-05,
+      "loss": 1.3528,
+      "step": 135040
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.955460319889798e-05,
+      "loss": 1.0513,
+      "step": 135050
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.955345526899823e-05,
+      "loss": 1.2457,
+      "step": 135060
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.955230733909849e-05,
+      "loss": 1.2655,
+      "step": 135070
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.955115940919874e-05,
+      "loss": 1.2477,
+      "step": 135080
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.9550011479298995e-05,
+      "loss": 1.2769,
+      "step": 135090
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.9548863549399245e-05,
+      "loss": 1.2,
+      "step": 135100
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.9547715619499494e-05,
+      "loss": 1.2317,
+      "step": 135110
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.954656768959975e-05,
+      "loss": 1.1769,
+      "step": 135120
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.954541975970001e-05,
+      "loss": 1.1751,
+      "step": 135130
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.954427182980025e-05,
+      "loss": 1.2382,
+      "step": 135140
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.954312389990051e-05,
+      "loss": 1.192,
+      "step": 135150
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.9541975970000764e-05,
+      "loss": 1.1922,
+      "step": 135160
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.954082804010101e-05,
+      "loss": 1.1876,
+      "step": 135170
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.9539680110201264e-05,
+      "loss": 1.2318,
+      "step": 135180
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.953853218030152e-05,
+      "loss": 1.4356,
+      "step": 135190
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.953738425040177e-05,
+      "loss": 1.2032,
+      "step": 135200
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.953623632050202e-05,
+      "loss": 1.2034,
+      "step": 135210
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.953508839060228e-05,
+      "loss": 1.2153,
+      "step": 135220
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.9533940460702527e-05,
+      "loss": 1.1511,
+      "step": 135230
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.953279253080278e-05,
+      "loss": 1.1658,
+      "step": 135240
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.953164460090303e-05,
+      "loss": 1.2157,
+      "step": 135250
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.953049667100328e-05,
+      "loss": 1.3182,
+      "step": 135260
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.952934874110354e-05,
+      "loss": 1.2638,
+      "step": 135270
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.9528200811203796e-05,
+      "loss": 1.2717,
+      "step": 135280
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.952705288130404e-05,
+      "loss": 1.2485,
+      "step": 135290
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.9525904951404296e-05,
+      "loss": 1.4227,
+      "step": 135300
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.952475702150455e-05,
+      "loss": 1.1543,
+      "step": 135310
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.9523609091604796e-05,
+      "loss": 1.2686,
+      "step": 135320
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.952246116170505e-05,
+      "loss": 1.1143,
+      "step": 135330
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.952131323180531e-05,
+      "loss": 1.2163,
+      "step": 135340
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.952016530190556e-05,
+      "loss": 1.1568,
+      "step": 135350
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.951901737200581e-05,
+      "loss": 1.2231,
+      "step": 135360
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.9517869442106065e-05,
+      "loss": 1.0997,
+      "step": 135370
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.9516721512206315e-05,
+      "loss": 1.2678,
+      "step": 135380
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.951557358230657e-05,
+      "loss": 1.1431,
+      "step": 135390
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.951442565240682e-05,
+      "loss": 1.2674,
+      "step": 135400
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.951327772250707e-05,
+      "loss": 1.195,
+      "step": 135410
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.951212979260733e-05,
+      "loss": 1.1983,
+      "step": 135420
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.9510981862707585e-05,
+      "loss": 1.2319,
+      "step": 135430
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.950983393280783e-05,
+      "loss": 1.0789,
+      "step": 135440
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.9508686002908084e-05,
+      "loss": 1.2518,
+      "step": 135450
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.950753807300834e-05,
+      "loss": 1.1996,
+      "step": 135460
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.9506390143108584e-05,
+      "loss": 1.1848,
+      "step": 135470
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.950524221320884e-05,
+      "loss": 1.2234,
+      "step": 135480
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.95040942833091e-05,
+      "loss": 1.3191,
+      "step": 135490
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.950294635340935e-05,
+      "loss": 1.2576,
+      "step": 135500
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.95017984235096e-05,
+      "loss": 1.1892,
+      "step": 135510
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.9500650493609854e-05,
+      "loss": 1.2169,
+      "step": 135520
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.9499502563710104e-05,
+      "loss": 1.2379,
+      "step": 135530
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.949835463381036e-05,
+      "loss": 1.2535,
+      "step": 135540
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.949720670391061e-05,
+      "loss": 1.1779,
+      "step": 135550
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.949605877401086e-05,
+      "loss": 1.3096,
+      "step": 135560
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.9494910844111116e-05,
+      "loss": 1.1742,
+      "step": 135570
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.949376291421137e-05,
+      "loss": 1.0503,
+      "step": 135580
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.9492614984311616e-05,
+      "loss": 1.1584,
+      "step": 135590
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.949146705441187e-05,
+      "loss": 1.1283,
+      "step": 135600
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.949031912451213e-05,
+      "loss": 1.1674,
+      "step": 135610
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.948917119461237e-05,
+      "loss": 1.2151,
+      "step": 135620
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.948802326471263e-05,
+      "loss": 1.1953,
+      "step": 135630
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.9486875334812886e-05,
+      "loss": 1.1317,
+      "step": 135640
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.9485727404913136e-05,
+      "loss": 1.0711,
+      "step": 135650
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.9484579475013385e-05,
+      "loss": 1.1744,
+      "step": 135660
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.948343154511364e-05,
+      "loss": 1.2668,
+      "step": 135670
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.948228361521389e-05,
+      "loss": 1.2856,
+      "step": 135680
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.948113568531415e-05,
+      "loss": 1.1091,
+      "step": 135690
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.947998775541439e-05,
+      "loss": 1.2455,
+      "step": 135700
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.947883982551465e-05,
+      "loss": 1.2248,
+      "step": 135710
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.9477691895614905e-05,
+      "loss": 1.2772,
+      "step": 135720
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.947654396571515e-05,
+      "loss": 1.3114,
+      "step": 135730
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.9475396035815405e-05,
+      "loss": 1.2502,
+      "step": 135740
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.947424810591566e-05,
+      "loss": 1.1095,
+      "step": 135750
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.947310017601591e-05,
+      "loss": 1.1306,
+      "step": 135760
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.947195224611616e-05,
+      "loss": 1.2513,
+      "step": 135770
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.947080431621642e-05,
+      "loss": 1.1785,
+      "step": 135780
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.946965638631667e-05,
+      "loss": 1.2917,
+      "step": 135790
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.9468508456416924e-05,
+      "loss": 1.3595,
+      "step": 135800
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.9467360526517174e-05,
+      "loss": 1.1708,
+      "step": 135810
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.9466212596617424e-05,
+      "loss": 1.1338,
+      "step": 135820
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.946506466671768e-05,
+      "loss": 1.116,
+      "step": 135830
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.946391673681794e-05,
+      "loss": 1.3317,
+      "step": 135840
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.946276880691818e-05,
+      "loss": 1.1513,
+      "step": 135850
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.946162087701844e-05,
+      "loss": 1.2507,
+      "step": 135860
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.9460472947118693e-05,
+      "loss": 1.2656,
+      "step": 135870
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.9459325017218937e-05,
+      "loss": 1.2247,
+      "step": 135880
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.945817708731919e-05,
+      "loss": 1.1225,
+      "step": 135890
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.945702915741945e-05,
+      "loss": 1.2888,
+      "step": 135900
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.94558812275197e-05,
+      "loss": 1.363,
+      "step": 135910
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.945473329761995e-05,
+      "loss": 1.2294,
+      "step": 135920
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.9453585367720206e-05,
+      "loss": 1.3502,
+      "step": 135930
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.9452437437820456e-05,
+      "loss": 1.1917,
+      "step": 135940
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.945128950792071e-05,
+      "loss": 1.267,
+      "step": 135950
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.945014157802096e-05,
+      "loss": 1.1826,
+      "step": 135960
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.944899364812121e-05,
+      "loss": 1.1805,
+      "step": 135970
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.944784571822147e-05,
+      "loss": 1.2225,
+      "step": 135980
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.9446697788321726e-05,
+      "loss": 1.1876,
+      "step": 135990
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.944554985842197e-05,
+      "loss": 1.1558,
+      "step": 136000
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.9444401928522225e-05,
+      "loss": 1.0867,
+      "step": 136010
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.944325399862248e-05,
+      "loss": 1.3428,
+      "step": 136020
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.9442106068722725e-05,
+      "loss": 1.1597,
+      "step": 136030
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.944095813882298e-05,
+      "loss": 1.2137,
+      "step": 136040
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.943981020892324e-05,
+      "loss": 1.1756,
+      "step": 136050
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.943866227902349e-05,
+      "loss": 1.2516,
+      "step": 136060
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.943751434912374e-05,
+      "loss": 1.1641,
+      "step": 136070
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.9436366419223995e-05,
+      "loss": 1.1878,
+      "step": 136080
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.9435218489324244e-05,
+      "loss": 1.3162,
+      "step": 136090
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.94340705594245e-05,
+      "loss": 1.0878,
+      "step": 136100
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.943292262952475e-05,
+      "loss": 1.1904,
+      "step": 136110
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.9431774699625e-05,
+      "loss": 1.0413,
+      "step": 136120
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.943062676972526e-05,
+      "loss": 1.3343,
+      "step": 136130
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.9429478839825514e-05,
+      "loss": 1.1429,
+      "step": 136140
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.942833090992576e-05,
+      "loss": 1.2543,
+      "step": 136150
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.9427182980026014e-05,
+      "loss": 1.1936,
+      "step": 136160
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.942603505012627e-05,
+      "loss": 1.3081,
+      "step": 136170
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.9424887120226513e-05,
+      "loss": 1.2779,
+      "step": 136180
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.942373919032677e-05,
+      "loss": 1.2799,
+      "step": 136190
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.942259126042703e-05,
+      "loss": 1.2134,
+      "step": 136200
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.9421443330527277e-05,
+      "loss": 1.2419,
+      "step": 136210
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.9420295400627526e-05,
+      "loss": 1.1732,
+      "step": 136220
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.941914747072778e-05,
+      "loss": 1.1279,
+      "step": 136230
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.941799954082803e-05,
+      "loss": 1.2825,
+      "step": 136240
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.941685161092829e-05,
+      "loss": 1.1664,
+      "step": 136250
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.941570368102854e-05,
+      "loss": 1.1312,
+      "step": 136260
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.941455575112879e-05,
+      "loss": 1.331,
+      "step": 136270
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.9413407821229046e-05,
+      "loss": 1.264,
+      "step": 136280
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.94122598913293e-05,
+      "loss": 1.3138,
+      "step": 136290
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.9411111961429546e-05,
+      "loss": 1.3367,
+      "step": 136300
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.94099640315298e-05,
+      "loss": 1.2344,
+      "step": 136310
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.940881610163006e-05,
+      "loss": 1.1908,
+      "step": 136320
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.94076681717303e-05,
+      "loss": 1.1493,
+      "step": 136330
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.940652024183056e-05,
+      "loss": 1.2816,
+      "step": 136340
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.9405372311930815e-05,
+      "loss": 1.1571,
+      "step": 136350
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.9404224382031065e-05,
+      "loss": 1.2141,
+      "step": 136360
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.9403076452131315e-05,
+      "loss": 1.1788,
+      "step": 136370
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.940192852223157e-05,
+      "loss": 1.224,
+      "step": 136380
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.940078059233182e-05,
+      "loss": 1.2359,
+      "step": 136390
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.939963266243208e-05,
+      "loss": 1.2227,
+      "step": 136400
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.939848473253233e-05,
+      "loss": 1.0981,
+      "step": 136410
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.939733680263258e-05,
+      "loss": 1.1883,
+      "step": 136420
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.9396188872732834e-05,
+      "loss": 1.3529,
+      "step": 136430
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.939504094283309e-05,
+      "loss": 1.2377,
+      "step": 136440
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.9393893012933334e-05,
+      "loss": 1.2413,
+      "step": 136450
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.939274508303359e-05,
+      "loss": 1.3753,
+      "step": 136460
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.939159715313385e-05,
+      "loss": 1.1834,
+      "step": 136470
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.939044922323409e-05,
+      "loss": 1.1771,
+      "step": 136480
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.938930129333435e-05,
+      "loss": 1.2164,
+      "step": 136490
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.9388153363434604e-05,
+      "loss": 1.1138,
+      "step": 136500
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.9387005433534854e-05,
+      "loss": 1.4431,
+      "step": 136510
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.93858575036351e-05,
+      "loss": 1.1243,
+      "step": 136520
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.938470957373536e-05,
+      "loss": 1.1878,
+      "step": 136530
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.938356164383561e-05,
+      "loss": 1.2222,
+      "step": 136540
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.9382413713935866e-05,
+      "loss": 1.0571,
+      "step": 136550
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.9381265784036116e-05,
+      "loss": 1.1521,
+      "step": 136560
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.9380117854136366e-05,
+      "loss": 1.1895,
+      "step": 136570
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.937896992423662e-05,
+      "loss": 1.4027,
+      "step": 136580
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.937782199433688e-05,
+      "loss": 1.2395,
+      "step": 136590
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.937667406443712e-05,
+      "loss": 1.251,
+      "step": 136600
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.937552613453738e-05,
+      "loss": 1.1705,
+      "step": 136610
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.9374378204637636e-05,
+      "loss": 1.2904,
+      "step": 136620
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.937323027473788e-05,
+      "loss": 1.1379,
+      "step": 136630
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.9372082344838135e-05,
+      "loss": 1.2431,
+      "step": 136640
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.937093441493839e-05,
+      "loss": 1.0988,
+      "step": 136650
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.936978648503864e-05,
+      "loss": 1.2089,
+      "step": 136660
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.936863855513889e-05,
+      "loss": 1.2785,
+      "step": 136670
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.936749062523915e-05,
+      "loss": 1.0961,
+      "step": 136680
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.93663426953394e-05,
+      "loss": 1.3022,
+      "step": 136690
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.9365194765439655e-05,
+      "loss": 1.3167,
+      "step": 136700
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.9364046835539905e-05,
+      "loss": 1.2459,
+      "step": 136710
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.9362898905640155e-05,
+      "loss": 1.057,
+      "step": 136720
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.936175097574041e-05,
+      "loss": 1.205,
+      "step": 136730
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.936060304584067e-05,
+      "loss": 1.2384,
+      "step": 136740
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.935945511594091e-05,
+      "loss": 1.318,
+      "step": 136750
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.935830718604117e-05,
+      "loss": 1.2719,
+      "step": 136760
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.9357159256141424e-05,
+      "loss": 1.1158,
+      "step": 136770
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.935601132624167e-05,
+      "loss": 1.2791,
+      "step": 136780
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.9354863396341924e-05,
+      "loss": 1.368,
+      "step": 136790
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.935371546644218e-05,
+      "loss": 1.1078,
+      "step": 136800
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.935256753654243e-05,
+      "loss": 1.2463,
+      "step": 136810
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.935141960664268e-05,
+      "loss": 1.1889,
+      "step": 136820
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.935027167674294e-05,
+      "loss": 1.2049,
+      "step": 136830
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.934912374684319e-05,
+      "loss": 1.2186,
+      "step": 136840
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.9347975816943443e-05,
+      "loss": 1.1158,
+      "step": 136850
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.934682788704369e-05,
+      "loss": 1.0719,
+      "step": 136860
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.934567995714394e-05,
+      "loss": 1.2305,
+      "step": 136870
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.93445320272442e-05,
+      "loss": 1.1797,
+      "step": 136880
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.9343384097344456e-05,
+      "loss": 1.1708,
+      "step": 136890
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.93422361674447e-05,
+      "loss": 1.1936,
+      "step": 136900
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.9341088237544956e-05,
+      "loss": 1.1353,
+      "step": 136910
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.933994030764521e-05,
+      "loss": 1.1658,
+      "step": 136920
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.9338792377745456e-05,
+      "loss": 1.2002,
+      "step": 136930
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.933764444784571e-05,
+      "loss": 1.2467,
+      "step": 136940
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.933649651794597e-05,
+      "loss": 1.1679,
+      "step": 136950
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.933534858804622e-05,
+      "loss": 1.1852,
+      "step": 136960
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.933420065814647e-05,
+      "loss": 1.2923,
+      "step": 136970
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.9333052728246725e-05,
+      "loss": 1.1424,
+      "step": 136980
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.9331904798346975e-05,
+      "loss": 1.3409,
+      "step": 136990
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.933075686844723e-05,
+      "loss": 1.2153,
+      "step": 137000
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.932960893854748e-05,
+      "loss": 1.2589,
+      "step": 137010
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.932846100864773e-05,
+      "loss": 1.2954,
+      "step": 137020
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.932731307874799e-05,
+      "loss": 1.2092,
+      "step": 137030
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.9326165148848245e-05,
+      "loss": 1.3631,
+      "step": 137040
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.932501721894849e-05,
+      "loss": 1.338,
+      "step": 137050
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.9323869289048745e-05,
+      "loss": 1.2326,
+      "step": 137060
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.9322721359149e-05,
+      "loss": 1.2304,
+      "step": 137070
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.9321573429249244e-05,
+      "loss": 1.1739,
+      "step": 137080
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.93204254993495e-05,
+      "loss": 1.3188,
+      "step": 137090
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.931927756944975e-05,
+      "loss": 1.0329,
+      "step": 137100
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.931812963955001e-05,
+      "loss": 1.2262,
+      "step": 137110
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.931698170965026e-05,
+      "loss": 1.111,
+      "step": 137120
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.931583377975051e-05,
+      "loss": 1.2202,
+      "step": 137130
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.9314685849850764e-05,
+      "loss": 1.1638,
+      "step": 137140
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.931353791995102e-05,
+      "loss": 1.2011,
+      "step": 137150
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.9312389990051263e-05,
+      "loss": 1.3761,
+      "step": 137160
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.931124206015152e-05,
+      "loss": 1.4595,
+      "step": 137170
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.931009413025178e-05,
+      "loss": 1.2033,
+      "step": 137180
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.930894620035202e-05,
+      "loss": 1.2709,
+      "step": 137190
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.9307798270452276e-05,
+      "loss": 1.2923,
+      "step": 137200
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.930665034055253e-05,
+      "loss": 1.2775,
+      "step": 137210
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.930550241065278e-05,
+      "loss": 1.2028,
+      "step": 137220
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.930435448075303e-05,
+      "loss": 1.1983,
+      "step": 137230
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.930320655085329e-05,
+      "loss": 1.3459,
+      "step": 137240
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.930205862095354e-05,
+      "loss": 1.2075,
+      "step": 137250
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.9300910691053796e-05,
+      "loss": 1.1963,
+      "step": 137260
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.9299762761154046e-05,
+      "loss": 1.1293,
+      "step": 137270
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.9298614831254296e-05,
+      "loss": 1.1276,
+      "step": 137280
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.929746690135455e-05,
+      "loss": 1.1774,
+      "step": 137290
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.929631897145481e-05,
+      "loss": 1.3289,
+      "step": 137300
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.929517104155505e-05,
+      "loss": 1.2767,
+      "step": 137310
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.929402311165531e-05,
+      "loss": 1.2047,
+      "step": 137320
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.9292875181755565e-05,
+      "loss": 1.1091,
+      "step": 137330
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.929172725185581e-05,
+      "loss": 1.1504,
+      "step": 137340
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.9290579321956065e-05,
+      "loss": 1.1547,
+      "step": 137350
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.928943139205632e-05,
+      "loss": 1.3374,
+      "step": 137360
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.928828346215657e-05,
+      "loss": 1.1433,
+      "step": 137370
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.928713553225682e-05,
+      "loss": 1.2063,
+      "step": 137380
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.928598760235708e-05,
+      "loss": 1.2598,
+      "step": 137390
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.928483967245733e-05,
+      "loss": 1.2117,
+      "step": 137400
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.9283691742557584e-05,
+      "loss": 1.1588,
+      "step": 137410
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.9282543812657834e-05,
+      "loss": 1.1872,
+      "step": 137420
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.9281395882758084e-05,
+      "loss": 1.1635,
+      "step": 137430
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.928024795285834e-05,
+      "loss": 1.1899,
+      "step": 137440
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.92791000229586e-05,
+      "loss": 1.1851,
+      "step": 137450
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.927795209305884e-05,
+      "loss": 1.2285,
+      "step": 137460
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.92768041631591e-05,
+      "loss": 1.1964,
+      "step": 137470
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.9275656233259354e-05,
+      "loss": 1.0602,
+      "step": 137480
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.92745083033596e-05,
+      "loss": 1.3173,
+      "step": 137490
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.927336037345985e-05,
+      "loss": 1.169,
+      "step": 137500
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.927221244356011e-05,
+      "loss": 1.4079,
+      "step": 137510
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.927106451366036e-05,
+      "loss": 1.0504,
+      "step": 137520
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.926991658376061e-05,
+      "loss": 1.2561,
+      "step": 137530
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.9268768653860866e-05,
+      "loss": 1.3676,
+      "step": 137540
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.9267620723961116e-05,
+      "loss": 1.3146,
+      "step": 137550
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.926647279406137e-05,
+      "loss": 1.3263,
+      "step": 137560
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.926532486416162e-05,
+      "loss": 1.2225,
+      "step": 137570
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.926417693426187e-05,
+      "loss": 1.2303,
+      "step": 137580
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.926302900436213e-05,
+      "loss": 1.0327,
+      "step": 137590
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.9261881074462386e-05,
+      "loss": 1.1416,
+      "step": 137600
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.926073314456263e-05,
+      "loss": 1.244,
+      "step": 137610
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.9259585214662885e-05,
+      "loss": 1.2078,
+      "step": 137620
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.925843728476314e-05,
+      "loss": 1.1396,
+      "step": 137630
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.9257289354863385e-05,
+      "loss": 1.2084,
+      "step": 137640
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.925614142496364e-05,
+      "loss": 1.2686,
+      "step": 137650
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.92549934950639e-05,
+      "loss": 1.1361,
+      "step": 137660
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.925384556516415e-05,
+      "loss": 1.2033,
+      "step": 137670
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.92526976352644e-05,
+      "loss": 1.2662,
+      "step": 137680
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.9251549705364655e-05,
+      "loss": 1.2474,
+      "step": 137690
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.9250401775464905e-05,
+      "loss": 1.1287,
+      "step": 137700
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.924925384556516e-05,
+      "loss": 1.1801,
+      "step": 137710
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.924810591566541e-05,
+      "loss": 1.2676,
+      "step": 137720
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.924695798576566e-05,
+      "loss": 1.2021,
+      "step": 137730
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.924581005586592e-05,
+      "loss": 1.1948,
+      "step": 137740
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.9244662125966174e-05,
+      "loss": 1.3666,
+      "step": 137750
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.924351419606642e-05,
+      "loss": 1.1599,
+      "step": 137760
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.9242366266166674e-05,
+      "loss": 1.2267,
+      "step": 137770
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.924121833626693e-05,
+      "loss": 1.2883,
+      "step": 137780
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.9240070406367174e-05,
+      "loss": 1.186,
+      "step": 137790
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.923892247646743e-05,
+      "loss": 1.1446,
+      "step": 137800
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.923777454656769e-05,
+      "loss": 1.1294,
+      "step": 137810
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.923662661666794e-05,
+      "loss": 1.2668,
+      "step": 137820
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.923547868676819e-05,
+      "loss": 1.2413,
+      "step": 137830
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.923433075686844e-05,
+      "loss": 1.2875,
+      "step": 137840
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.923318282696869e-05,
+      "loss": 1.2023,
+      "step": 137850
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.923203489706895e-05,
+      "loss": 1.1662,
+      "step": 137860
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.92308869671692e-05,
+      "loss": 1.1038,
+      "step": 137870
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.922973903726945e-05,
+      "loss": 1.2231,
+      "step": 137880
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.9228591107369706e-05,
+      "loss": 1.1862,
+      "step": 137890
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.922744317746996e-05,
+      "loss": 1.1325,
+      "step": 137900
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.9226295247570206e-05,
+      "loss": 1.114,
+      "step": 137910
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.922514731767046e-05,
+      "loss": 1.3021,
+      "step": 137920
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.922399938777072e-05,
+      "loss": 1.2398,
+      "step": 137930
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.922285145787096e-05,
+      "loss": 1.33,
+      "step": 137940
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.922170352797122e-05,
+      "loss": 1.2151,
+      "step": 137950
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.9220555598071475e-05,
+      "loss": 1.2495,
+      "step": 137960
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.9219407668171725e-05,
+      "loss": 1.2741,
+      "step": 137970
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.9218259738271975e-05,
+      "loss": 1.2053,
+      "step": 137980
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.921711180837223e-05,
+      "loss": 1.2054,
+      "step": 137990
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.921596387847248e-05,
+      "loss": 1.2729,
+      "step": 138000
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.921481594857274e-05,
+      "loss": 1.356,
+      "step": 138010
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.921366801867299e-05,
+      "loss": 1.2009,
+      "step": 138020
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.921252008877324e-05,
+      "loss": 1.1909,
+      "step": 138030
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.9211372158873495e-05,
+      "loss": 1.1451,
+      "step": 138040
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.921022422897375e-05,
+      "loss": 1.283,
+      "step": 138050
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.9209076299073994e-05,
+      "loss": 1.1731,
+      "step": 138060
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.920792836917425e-05,
+      "loss": 1.2276,
+      "step": 138070
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.920678043927451e-05,
+      "loss": 1.2783,
+      "step": 138080
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.920563250937475e-05,
+      "loss": 1.2148,
+      "step": 138090
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.920448457947501e-05,
+      "loss": 1.1852,
+      "step": 138100
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.9203336649575264e-05,
+      "loss": 1.2708,
+      "step": 138110
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.9202188719675514e-05,
+      "loss": 1.2044,
+      "step": 138120
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.9201040789775764e-05,
+      "loss": 1.2182,
+      "step": 138130
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.919989285987602e-05,
+      "loss": 1.1711,
+      "step": 138140
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.919874492997627e-05,
+      "loss": 1.062,
+      "step": 138150
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.919759700007653e-05,
+      "loss": 1.1955,
+      "step": 138160
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.9196449070176777e-05,
+      "loss": 1.1606,
+      "step": 138170
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.9195301140277026e-05,
+      "loss": 1.3438,
+      "step": 138180
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.919415321037728e-05,
+      "loss": 1.1652,
+      "step": 138190
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.919300528047754e-05,
+      "loss": 1.2863,
+      "step": 138200
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.919185735057778e-05,
+      "loss": 1.2935,
+      "step": 138210
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.919070942067804e-05,
+      "loss": 1.3023,
+      "step": 138220
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.9189561490778296e-05,
+      "loss": 1.3091,
+      "step": 138230
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.918841356087854e-05,
+      "loss": 1.0693,
+      "step": 138240
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.9187265630978796e-05,
+      "loss": 1.2232,
+      "step": 138250
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.918611770107905e-05,
+      "loss": 1.1984,
+      "step": 138260
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.91849697711793e-05,
+      "loss": 1.237,
+      "step": 138270
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.918382184127955e-05,
+      "loss": 1.2688,
+      "step": 138280
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.918267391137981e-05,
+      "loss": 1.0993,
+      "step": 138290
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.918152598148006e-05,
+      "loss": 1.2015,
+      "step": 138300
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.9180378051580315e-05,
+      "loss": 1.2075,
+      "step": 138310
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.9179230121680565e-05,
+      "loss": 1.227,
+      "step": 138320
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.9178082191780815e-05,
+      "loss": 1.0811,
+      "step": 138330
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.917693426188107e-05,
+      "loss": 1.2312,
+      "step": 138340
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.917578633198133e-05,
+      "loss": 1.3221,
+      "step": 138350
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.917463840208157e-05,
+      "loss": 1.2546,
+      "step": 138360
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.917349047218183e-05,
+      "loss": 1.1522,
+      "step": 138370
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.9172342542282084e-05,
+      "loss": 1.2797,
+      "step": 138380
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.917119461238233e-05,
+      "loss": 1.163,
+      "step": 138390
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.9170046682482584e-05,
+      "loss": 1.274,
+      "step": 138400
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.916889875258284e-05,
+      "loss": 1.1708,
+      "step": 138410
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.916775082268309e-05,
+      "loss": 1.2343,
+      "step": 138420
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.916660289278334e-05,
+      "loss": 1.2795,
+      "step": 138430
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.91654549628836e-05,
+      "loss": 1.2404,
+      "step": 138440
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.916430703298385e-05,
+      "loss": 1.3118,
+      "step": 138450
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.9163159103084104e-05,
+      "loss": 1.2666,
+      "step": 138460
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.916201117318435e-05,
+      "loss": 1.3055,
+      "step": 138470
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.91608632432846e-05,
+      "loss": 1.3151,
+      "step": 138480
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.915971531338486e-05,
+      "loss": 1.1833,
+      "step": 138490
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.91585673834851e-05,
+      "loss": 1.1822,
+      "step": 138500
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.915741945358536e-05,
+      "loss": 1.3645,
+      "step": 138510
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.9156271523685616e-05,
+      "loss": 1.157,
+      "step": 138520
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.9155123593785866e-05,
+      "loss": 1.1966,
+      "step": 138530
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.9153975663886116e-05,
+      "loss": 1.2922,
+      "step": 138540
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.915282773398637e-05,
+      "loss": 1.2803,
+      "step": 138550
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.915167980408662e-05,
+      "loss": 1.1783,
+      "step": 138560
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.915053187418688e-05,
+      "loss": 1.3565,
+      "step": 138570
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.914938394428713e-05,
+      "loss": 1.3876,
+      "step": 138580
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.914823601438738e-05,
+      "loss": 1.3168,
+      "step": 138590
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.9147088084487635e-05,
+      "loss": 1.2305,
+      "step": 138600
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.914594015458789e-05,
+      "loss": 1.2039,
+      "step": 138610
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.9144792224688135e-05,
+      "loss": 1.1725,
+      "step": 138620
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.914364429478839e-05,
+      "loss": 1.2948,
+      "step": 138630
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.914249636488865e-05,
+      "loss": 1.3364,
+      "step": 138640
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.914134843498889e-05,
+      "loss": 1.2942,
+      "step": 138650
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.914020050508915e-05,
+      "loss": 1.2562,
+      "step": 138660
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.9139052575189405e-05,
+      "loss": 1.1343,
+      "step": 138670
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.9137904645289655e-05,
+      "loss": 1.0913,
+      "step": 138680
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.9136756715389904e-05,
+      "loss": 1.1205,
+      "step": 138690
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.913560878549016e-05,
+      "loss": 1.1139,
+      "step": 138700
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.913446085559041e-05,
+      "loss": 1.251,
+      "step": 138710
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.913331292569067e-05,
+      "loss": 1.1462,
+      "step": 138720
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.913216499579092e-05,
+      "loss": 1.274,
+      "step": 138730
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.913101706589117e-05,
+      "loss": 1.2129,
+      "step": 138740
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.9129869135991424e-05,
+      "loss": 1.2392,
+      "step": 138750
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.912872120609168e-05,
+      "loss": 1.4059,
+      "step": 138760
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.9127573276191924e-05,
+      "loss": 1.2428,
+      "step": 138770
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.912642534629218e-05,
+      "loss": 1.0866,
+      "step": 138780
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.912527741639244e-05,
+      "loss": 1.2177,
+      "step": 138790
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.912412948649268e-05,
+      "loss": 1.315,
+      "step": 138800
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.912298155659294e-05,
+      "loss": 1.2319,
+      "step": 138810
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.912183362669319e-05,
+      "loss": 1.198,
+      "step": 138820
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.912068569679344e-05,
+      "loss": 1.3386,
+      "step": 138830
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.911953776689369e-05,
+      "loss": 1.1308,
+      "step": 138840
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.911838983699395e-05,
+      "loss": 1.2253,
+      "step": 138850
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.91172419070942e-05,
+      "loss": 1.2084,
+      "step": 138860
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.9116093977194456e-05,
+      "loss": 1.2469,
+      "step": 138870
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.9114946047294706e-05,
+      "loss": 1.2849,
+      "step": 138880
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.9113798117394956e-05,
+      "loss": 1.3478,
+      "step": 138890
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.911265018749521e-05,
+      "loss": 1.2015,
+      "step": 138900
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.911150225759547e-05,
+      "loss": 1.2194,
+      "step": 138910
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.911035432769571e-05,
+      "loss": 1.2607,
+      "step": 138920
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.910920639779597e-05,
+      "loss": 1.1974,
+      "step": 138930
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.9108058467896225e-05,
+      "loss": 1.3047,
+      "step": 138940
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.910691053799647e-05,
+      "loss": 1.2549,
+      "step": 138950
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.9105762608096725e-05,
+      "loss": 1.2083,
+      "step": 138960
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.910461467819698e-05,
+      "loss": 1.1494,
+      "step": 138970
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.910346674829723e-05,
+      "loss": 1.242,
+      "step": 138980
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.910231881839748e-05,
+      "loss": 1.199,
+      "step": 138990
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.910117088849774e-05,
+      "loss": 1.1402,
+      "step": 139000
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.910002295859799e-05,
+      "loss": 1.3164,
+      "step": 139010
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.9098875028698245e-05,
+      "loss": 1.2596,
+      "step": 139020
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.9097727098798494e-05,
+      "loss": 1.182,
+      "step": 139030
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.9096579168898744e-05,
+      "loss": 1.1452,
+      "step": 139040
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.9095431238999e-05,
+      "loss": 1.237,
+      "step": 139050
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.909428330909926e-05,
+      "loss": 1.225,
+      "step": 139060
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.90931353791995e-05,
+      "loss": 1.2742,
+      "step": 139070
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.909198744929976e-05,
+      "loss": 1.2031,
+      "step": 139080
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.9090839519400014e-05,
+      "loss": 1.2502,
+      "step": 139090
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.908969158950026e-05,
+      "loss": 1.1731,
+      "step": 139100
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.9088543659600514e-05,
+      "loss": 1.3144,
+      "step": 139110
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.908739572970077e-05,
+      "loss": 1.2106,
+      "step": 139120
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.908624779980102e-05,
+      "loss": 1.2385,
+      "step": 139130
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.908509986990127e-05,
+      "loss": 1.1729,
+      "step": 139140
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.9083951940001527e-05,
+      "loss": 1.109,
+      "step": 139150
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.9082804010101776e-05,
+      "loss": 1.2719,
+      "step": 139160
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.908165608020203e-05,
+      "loss": 1.2562,
+      "step": 139170
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.908050815030228e-05,
+      "loss": 1.12,
+      "step": 139180
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.907936022040253e-05,
+      "loss": 1.1691,
+      "step": 139190
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.907821229050279e-05,
+      "loss": 1.1228,
+      "step": 139200
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.9077064360603046e-05,
+      "loss": 1.19,
+      "step": 139210
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.907591643070329e-05,
+      "loss": 1.3229,
+      "step": 139220
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.9074768500803546e-05,
+      "loss": 1.2374,
+      "step": 139230
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.90736205709038e-05,
+      "loss": 1.166,
+      "step": 139240
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.9072472641004045e-05,
+      "loss": 1.291,
+      "step": 139250
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.90713247111043e-05,
+      "loss": 1.1978,
+      "step": 139260
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.907017678120456e-05,
+      "loss": 1.3073,
+      "step": 139270
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.906902885130481e-05,
+      "loss": 1.1315,
+      "step": 139280
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.906788092140506e-05,
+      "loss": 1.3106,
+      "step": 139290
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.9066732991505315e-05,
+      "loss": 1.2675,
+      "step": 139300
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.9065585061605565e-05,
+      "loss": 1.2486,
+      "step": 139310
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.906443713170582e-05,
+      "loss": 1.2108,
+      "step": 139320
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.906328920180607e-05,
+      "loss": 1.3444,
+      "step": 139330
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.906214127190632e-05,
+      "loss": 1.2565,
+      "step": 139340
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.906099334200658e-05,
+      "loss": 1.2999,
+      "step": 139350
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.9059845412106834e-05,
+      "loss": 1.2344,
+      "step": 139360
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.905869748220708e-05,
+      "loss": 1.1374,
+      "step": 139370
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.9057549552307334e-05,
+      "loss": 1.1853,
+      "step": 139380
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.905640162240759e-05,
+      "loss": 1.3404,
+      "step": 139390
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.9055253692507834e-05,
+      "loss": 1.1503,
+      "step": 139400
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.905410576260809e-05,
+      "loss": 1.2069,
+      "step": 139410
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.905295783270835e-05,
+      "loss": 1.2491,
+      "step": 139420
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.90518099028086e-05,
+      "loss": 1.1768,
+      "step": 139430
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.905066197290885e-05,
+      "loss": 1.276,
+      "step": 139440
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.9049514043009103e-05,
+      "loss": 1.1377,
+      "step": 139450
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.904836611310935e-05,
+      "loss": 1.1556,
+      "step": 139460
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.904721818320961e-05,
+      "loss": 1.228,
+      "step": 139470
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.904607025330986e-05,
+      "loss": 1.2202,
+      "step": 139480
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.904492232341011e-05,
+      "loss": 1.2113,
+      "step": 139490
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.9043774393510366e-05,
+      "loss": 1.2547,
+      "step": 139500
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.904262646361062e-05,
+      "loss": 1.1327,
+      "step": 139510
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.9041478533710866e-05,
+      "loss": 1.169,
+      "step": 139520
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.904033060381112e-05,
+      "loss": 1.2046,
+      "step": 139530
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.903918267391138e-05,
+      "loss": 1.0974,
+      "step": 139540
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.903803474401162e-05,
+      "loss": 1.3085,
+      "step": 139550
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.903688681411188e-05,
+      "loss": 1.2064,
+      "step": 139560
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.9035738884212136e-05,
+      "loss": 1.2447,
+      "step": 139570
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.9034590954312385e-05,
+      "loss": 1.1772,
+      "step": 139580
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.9033443024412635e-05,
+      "loss": 1.2201,
+      "step": 139590
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.903229509451289e-05,
+      "loss": 1.1911,
+      "step": 139600
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.903114716461314e-05,
+      "loss": 1.1418,
+      "step": 139610
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.90299992347134e-05,
+      "loss": 1.0162,
+      "step": 139620
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.902885130481365e-05,
+      "loss": 1.1317,
+      "step": 139630
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.90277033749139e-05,
+      "loss": 1.2046,
+      "step": 139640
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.9026555445014155e-05,
+      "loss": 1.3526,
+      "step": 139650
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.902540751511441e-05,
+      "loss": 1.0954,
+      "step": 139660
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.9024259585214654e-05,
+      "loss": 1.1863,
+      "step": 139670
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.902311165531491e-05,
+      "loss": 1.2661,
+      "step": 139680
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.902196372541517e-05,
+      "loss": 1.2094,
+      "step": 139690
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.902081579551541e-05,
+      "loss": 1.3148,
+      "step": 139700
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.901966786561567e-05,
+      "loss": 1.1098,
+      "step": 139710
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.9018519935715924e-05,
+      "loss": 1.2071,
+      "step": 139720
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.9017372005816174e-05,
+      "loss": 1.3051,
+      "step": 139730
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.9016224075916424e-05,
+      "loss": 1.4491,
+      "step": 139740
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.901507614601668e-05,
+      "loss": 1.0735,
+      "step": 139750
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.901392821611693e-05,
+      "loss": 1.2151,
+      "step": 139760
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.901278028621719e-05,
+      "loss": 1.2254,
+      "step": 139770
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.901163235631744e-05,
+      "loss": 1.2415,
+      "step": 139780
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.901048442641769e-05,
+      "loss": 1.2082,
+      "step": 139790
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.900933649651794e-05,
+      "loss": 1.2067,
+      "step": 139800
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.90081885666182e-05,
+      "loss": 1.2903,
+      "step": 139810
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.900704063671844e-05,
+      "loss": 1.2034,
+      "step": 139820
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.90058927068187e-05,
+      "loss": 1.2371,
+      "step": 139830
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.900474477691895e-05,
+      "loss": 1.2725,
+      "step": 139840
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.90035968470192e-05,
+      "loss": 1.338,
+      "step": 139850
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.9002448917119456e-05,
+      "loss": 1.1553,
+      "step": 139860
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.9001300987219706e-05,
+      "loss": 1.2701,
+      "step": 139870
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.900015305731996e-05,
+      "loss": 1.1871,
+      "step": 139880
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.899900512742021e-05,
+      "loss": 1.1614,
+      "step": 139890
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.899785719752046e-05,
+      "loss": 1.2419,
+      "step": 139900
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.899670926762072e-05,
+      "loss": 1.2011,
+      "step": 139910
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.8995561337720975e-05,
+      "loss": 1.3273,
+      "step": 139920
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.899441340782122e-05,
+      "loss": 1.2839,
+      "step": 139930
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.8993265477921475e-05,
+      "loss": 1.3654,
+      "step": 139940
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.899211754802173e-05,
+      "loss": 1.2386,
+      "step": 139950
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.8990969618121975e-05,
+      "loss": 1.1011,
+      "step": 139960
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.898982168822223e-05,
+      "loss": 1.1817,
+      "step": 139970
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.898867375832249e-05,
+      "loss": 1.2377,
+      "step": 139980
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.898752582842274e-05,
+      "loss": 1.2036,
+      "step": 139990
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.898637789852299e-05,
+      "loss": 1.3331,
+      "step": 140000
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.8985229968623244e-05,
+      "loss": 1.4233,
+      "step": 140010
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.8984082038723494e-05,
+      "loss": 1.3521,
+      "step": 140020
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.898293410882375e-05,
+      "loss": 1.2425,
+      "step": 140030
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.8981786178924e-05,
+      "loss": 1.1073,
+      "step": 140040
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.898063824902425e-05,
+      "loss": 1.2016,
+      "step": 140050
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.897949031912451e-05,
+      "loss": 1.2811,
+      "step": 140060
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.8978342389224764e-05,
+      "loss": 1.2352,
+      "step": 140070
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.897719445932501e-05,
+      "loss": 1.344,
+      "step": 140080
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.8976046529425264e-05,
+      "loss": 1.2672,
+      "step": 140090
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.897489859952552e-05,
+      "loss": 1.2775,
+      "step": 140100
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.897375066962576e-05,
+      "loss": 1.2105,
+      "step": 140110
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.897260273972602e-05,
+      "loss": 1.1915,
+      "step": 140120
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.8971454809826277e-05,
+      "loss": 1.0433,
+      "step": 140130
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.8970306879926526e-05,
+      "loss": 1.3035,
+      "step": 140140
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.8969158950026776e-05,
+      "loss": 1.2377,
+      "step": 140150
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.896801102012703e-05,
+      "loss": 1.1583,
+      "step": 140160
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.896686309022728e-05,
+      "loss": 1.1866,
+      "step": 140170
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.896571516032754e-05,
+      "loss": 1.2427,
+      "step": 140180
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.896456723042779e-05,
+      "loss": 1.1846,
+      "step": 140190
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.896341930052804e-05,
+      "loss": 1.261,
+      "step": 140200
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.8962271370628296e-05,
+      "loss": 1.1689,
+      "step": 140210
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.896112344072855e-05,
+      "loss": 1.3248,
+      "step": 140220
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.8959975510828795e-05,
+      "loss": 1.2513,
+      "step": 140230
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.895882758092905e-05,
+      "loss": 1.2335,
+      "step": 140240
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.895767965102931e-05,
+      "loss": 1.1249,
+      "step": 140250
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.895653172112955e-05,
+      "loss": 1.1473,
+      "step": 140260
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.895538379122981e-05,
+      "loss": 1.1607,
+      "step": 140270
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.8954235861330065e-05,
+      "loss": 1.2656,
+      "step": 140280
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.8953087931430315e-05,
+      "loss": 1.3065,
+      "step": 140290
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.8951940001530565e-05,
+      "loss": 1.3108,
+      "step": 140300
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.895079207163082e-05,
+      "loss": 1.2204,
+      "step": 140310
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.894964414173107e-05,
+      "loss": 1.2466,
+      "step": 140320
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.894849621183133e-05,
+      "loss": 1.2976,
+      "step": 140330
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.894734828193158e-05,
+      "loss": 1.1922,
+      "step": 140340
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.894620035203183e-05,
+      "loss": 1.3566,
+      "step": 140350
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.8945052422132084e-05,
+      "loss": 1.2507,
+      "step": 140360
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.894390449223234e-05,
+      "loss": 1.2636,
+      "step": 140370
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.8942756562332584e-05,
+      "loss": 1.3234,
+      "step": 140380
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.894160863243284e-05,
+      "loss": 1.2196,
+      "step": 140390
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.89404607025331e-05,
+      "loss": 1.3144,
+      "step": 140400
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.893931277263334e-05,
+      "loss": 1.1457,
+      "step": 140410
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.89381648427336e-05,
+      "loss": 1.2507,
+      "step": 140420
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.8937016912833853e-05,
+      "loss": 1.238,
+      "step": 140430
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.89358689829341e-05,
+      "loss": 1.2049,
+      "step": 140440
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.893472105303435e-05,
+      "loss": 1.1299,
+      "step": 140450
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.893357312313461e-05,
+      "loss": 1.1509,
+      "step": 140460
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.893242519323486e-05,
+      "loss": 1.2452,
+      "step": 140470
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.8931277263335116e-05,
+      "loss": 1.1538,
+      "step": 140480
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.8930129333435366e-05,
+      "loss": 1.2094,
+      "step": 140490
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.8928981403535616e-05,
+      "loss": 1.1767,
+      "step": 140500
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.892783347363587e-05,
+      "loss": 1.2104,
+      "step": 140510
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.892668554373613e-05,
+      "loss": 1.2124,
+      "step": 140520
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.892553761383637e-05,
+      "loss": 1.1619,
+      "step": 140530
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.892438968393663e-05,
+      "loss": 1.2129,
+      "step": 140540
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.8923241754036886e-05,
+      "loss": 1.2086,
+      "step": 140550
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.892209382413713e-05,
+      "loss": 1.221,
+      "step": 140560
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.8920945894237385e-05,
+      "loss": 1.1788,
+      "step": 140570
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.891979796433764e-05,
+      "loss": 1.2099,
+      "step": 140580
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.891865003443789e-05,
+      "loss": 1.2812,
+      "step": 140590
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.891750210453814e-05,
+      "loss": 1.2541,
+      "step": 140600
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.89163541746384e-05,
+      "loss": 1.348,
+      "step": 140610
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.891520624473865e-05,
+      "loss": 1.2285,
+      "step": 140620
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.8914058314838905e-05,
+      "loss": 1.3128,
+      "step": 140630
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.8912910384939155e-05,
+      "loss": 1.1837,
+      "step": 140640
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.8911762455039404e-05,
+      "loss": 1.1658,
+      "step": 140650
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.891061452513966e-05,
+      "loss": 1.3466,
+      "step": 140660
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.890946659523992e-05,
+      "loss": 1.1754,
+      "step": 140670
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.890831866534016e-05,
+      "loss": 1.1916,
+      "step": 140680
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.890717073544042e-05,
+      "loss": 1.0963,
+      "step": 140690
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.8906022805540674e-05,
+      "loss": 1.2061,
+      "step": 140700
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.890487487564092e-05,
+      "loss": 1.1156,
+      "step": 140710
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.8903726945741174e-05,
+      "loss": 1.1826,
+      "step": 140720
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.890257901584143e-05,
+      "loss": 1.1663,
+      "step": 140730
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.890143108594168e-05,
+      "loss": 1.2295,
+      "step": 140740
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.890028315604193e-05,
+      "loss": 1.1271,
+      "step": 140750
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.889913522614219e-05,
+      "loss": 1.2557,
+      "step": 140760
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.889798729624244e-05,
+      "loss": 1.3438,
+      "step": 140770
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.889683936634269e-05,
+      "loss": 1.1941,
+      "step": 140780
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.889569143644294e-05,
+      "loss": 1.2864,
+      "step": 140790
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.889454350654319e-05,
+      "loss": 1.229,
+      "step": 140800
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.889339557664345e-05,
+      "loss": 1.308,
+      "step": 140810
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.8892247646743706e-05,
+      "loss": 1.1623,
+      "step": 140820
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.889109971684395e-05,
+      "loss": 1.1529,
+      "step": 140830
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.8889951786944206e-05,
+      "loss": 1.1813,
+      "step": 140840
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.888880385704446e-05,
+      "loss": 1.2733,
+      "step": 140850
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.8887655927144706e-05,
+      "loss": 1.2057,
+      "step": 140860
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.888650799724496e-05,
+      "loss": 1.4033,
+      "step": 140870
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.888536006734522e-05,
+      "loss": 1.3578,
+      "step": 140880
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.888421213744547e-05,
+      "loss": 1.1789,
+      "step": 140890
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.888306420754572e-05,
+      "loss": 1.111,
+      "step": 140900
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.8881916277645975e-05,
+      "loss": 1.3918,
+      "step": 140910
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.8880768347746225e-05,
+      "loss": 1.3731,
+      "step": 140920
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.887962041784648e-05,
+      "loss": 1.3587,
+      "step": 140930
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.887847248794673e-05,
+      "loss": 1.1463,
+      "step": 140940
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.887732455804698e-05,
+      "loss": 1.1785,
+      "step": 140950
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.887617662814724e-05,
+      "loss": 1.1246,
+      "step": 140960
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.8875028698247495e-05,
+      "loss": 1.1739,
+      "step": 140970
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.887388076834774e-05,
+      "loss": 1.1985,
+      "step": 140980
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.8872732838447994e-05,
+      "loss": 1.2476,
+      "step": 140990
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.887158490854825e-05,
+      "loss": 1.141,
+      "step": 141000
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.8870436978648494e-05,
+      "loss": 1.2068,
+      "step": 141010
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.886928904874875e-05,
+      "loss": 1.1493,
+      "step": 141020
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.886814111884901e-05,
+      "loss": 1.1578,
+      "step": 141030
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.886699318894926e-05,
+      "loss": 1.2702,
+      "step": 141040
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.886584525904951e-05,
+      "loss": 1.1319,
+      "step": 141050
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.8864697329149764e-05,
+      "loss": 1.3093,
+      "step": 141060
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.8863549399250014e-05,
+      "loss": 1.0723,
+      "step": 141070
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.886240146935027e-05,
+      "loss": 1.2076,
+      "step": 141080
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.886125353945052e-05,
+      "loss": 1.2464,
+      "step": 141090
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.886010560955077e-05,
+      "loss": 1.2252,
+      "step": 141100
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.8858957679651027e-05,
+      "loss": 1.1184,
+      "step": 141110
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.885780974975128e-05,
+      "loss": 1.2921,
+      "step": 141120
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.8856661819851526e-05,
+      "loss": 1.2434,
+      "step": 141130
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.885551388995178e-05,
+      "loss": 1.2611,
+      "step": 141140
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.885436596005204e-05,
+      "loss": 1.3172,
+      "step": 141150
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.885321803015228e-05,
+      "loss": 1.1572,
+      "step": 141160
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.885207010025254e-05,
+      "loss": 1.2549,
+      "step": 141170
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.8850922170352796e-05,
+      "loss": 1.2842,
+      "step": 141180
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.8849774240453046e-05,
+      "loss": 1.1478,
+      "step": 141190
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.8848626310553296e-05,
+      "loss": 1.0046,
+      "step": 141200
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.8847478380653545e-05,
+      "loss": 1.1816,
+      "step": 141210
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.88463304507538e-05,
+      "loss": 1.1647,
+      "step": 141220
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.884518252085406e-05,
+      "loss": 1.2243,
+      "step": 141230
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.88440345909543e-05,
+      "loss": 1.2062,
+      "step": 141240
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.884288666105456e-05,
+      "loss": 1.0166,
+      "step": 141250
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.8841738731154815e-05,
+      "loss": 1.214,
+      "step": 141260
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.884059080125506e-05,
+      "loss": 1.1477,
+      "step": 141270
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.8839442871355315e-05,
+      "loss": 1.2626,
+      "step": 141280
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.883829494145557e-05,
+      "loss": 1.2114,
+      "step": 141290
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.883714701155582e-05,
+      "loss": 1.1116,
+      "step": 141300
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.883599908165607e-05,
+      "loss": 1.2429,
+      "step": 141310
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.883485115175633e-05,
+      "loss": 1.094,
+      "step": 141320
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.883370322185658e-05,
+      "loss": 1.2216,
+      "step": 141330
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.8832555291956834e-05,
+      "loss": 1.1507,
+      "step": 141340
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.8831407362057084e-05,
+      "loss": 1.3056,
+      "step": 141350
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.8830259432157334e-05,
+      "loss": 1.109,
+      "step": 141360
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.882911150225759e-05,
+      "loss": 1.1564,
+      "step": 141370
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.882796357235785e-05,
+      "loss": 1.2241,
+      "step": 141380
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.882681564245809e-05,
+      "loss": 1.289,
+      "step": 141390
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.882566771255835e-05,
+      "loss": 1.2697,
+      "step": 141400
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.8824519782658603e-05,
+      "loss": 1.2746,
+      "step": 141410
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.8823371852758847e-05,
+      "loss": 1.3748,
+      "step": 141420
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.88222239228591e-05,
+      "loss": 1.1104,
+      "step": 141430
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.882107599295936e-05,
+      "loss": 1.0978,
+      "step": 141440
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.881992806305961e-05,
+      "loss": 1.1245,
+      "step": 141450
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.881878013315986e-05,
+      "loss": 1.2168,
+      "step": 141460
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.8817632203260116e-05,
+      "loss": 1.1714,
+      "step": 141470
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.8816484273360366e-05,
+      "loss": 1.256,
+      "step": 141480
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.881533634346062e-05,
+      "loss": 1.2053,
+      "step": 141490
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.881418841356087e-05,
+      "loss": 1.207,
+      "step": 141500
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.881304048366112e-05,
+      "loss": 1.1546,
+      "step": 141510
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.881189255376138e-05,
+      "loss": 1.0955,
+      "step": 141520
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.8810744623861636e-05,
+      "loss": 1.1508,
+      "step": 141530
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.880959669396188e-05,
+      "loss": 1.1238,
+      "step": 141540
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.8808448764062135e-05,
+      "loss": 1.1301,
+      "step": 141550
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.880730083416239e-05,
+      "loss": 1.2696,
+      "step": 141560
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.8806152904262635e-05,
+      "loss": 1.1686,
+      "step": 141570
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.880500497436289e-05,
+      "loss": 1.0726,
+      "step": 141580
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.880385704446315e-05,
+      "loss": 1.1541,
+      "step": 141590
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.88027091145634e-05,
+      "loss": 1.3362,
+      "step": 141600
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.880156118466365e-05,
+      "loss": 1.2437,
+      "step": 141610
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.8800413254763905e-05,
+      "loss": 1.2366,
+      "step": 141620
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.8799265324864154e-05,
+      "loss": 1.1134,
+      "step": 141630
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.879811739496441e-05,
+      "loss": 1.2122,
+      "step": 141640
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.879696946506466e-05,
+      "loss": 1.2502,
+      "step": 141650
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.879582153516491e-05,
+      "loss": 1.2891,
+      "step": 141660
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.879467360526517e-05,
+      "loss": 1.2241,
+      "step": 141670
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.8793525675365424e-05,
+      "loss": 1.2625,
+      "step": 141680
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.879237774546567e-05,
+      "loss": 1.234,
+      "step": 141690
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.8791229815565924e-05,
+      "loss": 1.2169,
+      "step": 141700
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.879008188566618e-05,
+      "loss": 1.2558,
+      "step": 141710
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.8788933955766424e-05,
+      "loss": 1.5301,
+      "step": 141720
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.878778602586668e-05,
+      "loss": 1.2287,
+      "step": 141730
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.878663809596694e-05,
+      "loss": 1.1728,
+      "step": 141740
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.878549016606719e-05,
+      "loss": 1.193,
+      "step": 141750
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.8784342236167436e-05,
+      "loss": 1.1872,
+      "step": 141760
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.878319430626769e-05,
+      "loss": 1.2635,
+      "step": 141770
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.878204637636794e-05,
+      "loss": 1.2074,
+      "step": 141780
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.87808984464682e-05,
+      "loss": 1.1521,
+      "step": 141790
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.877975051656845e-05,
+      "loss": 1.095,
+      "step": 141800
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.87786025866687e-05,
+      "loss": 1.2867,
+      "step": 141810
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.8777454656768956e-05,
+      "loss": 1.1802,
+      "step": 141820
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.877630672686921e-05,
+      "loss": 1.2059,
+      "step": 141830
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.8775158796969456e-05,
+      "loss": 1.1265,
+      "step": 141840
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.877401086706971e-05,
+      "loss": 1.2503,
+      "step": 141850
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.877286293716997e-05,
+      "loss": 1.0865,
+      "step": 141860
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.877171500727021e-05,
+      "loss": 1.0954,
+      "step": 141870
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.877056707737047e-05,
+      "loss": 1.0783,
+      "step": 141880
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.8769419147470725e-05,
+      "loss": 1.2351,
+      "step": 141890
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.8768271217570975e-05,
+      "loss": 1.2655,
+      "step": 141900
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.8767123287671225e-05,
+      "loss": 1.1414,
+      "step": 141910
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.876597535777148e-05,
+      "loss": 1.311,
+      "step": 141920
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.876482742787173e-05,
+      "loss": 1.2114,
+      "step": 141930
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.876367949797199e-05,
+      "loss": 1.2229,
+      "step": 141940
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.876253156807224e-05,
+      "loss": 1.1783,
+      "step": 141950
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.876138363817249e-05,
+      "loss": 1.1076,
+      "step": 141960
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.8760235708272744e-05,
+      "loss": 1.2116,
+      "step": 141970
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.8759087778373e-05,
+      "loss": 1.2658,
+      "step": 141980
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.8757939848473244e-05,
+      "loss": 1.339,
+      "step": 141990
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.87567919185735e-05,
+      "loss": 1.2607,
+      "step": 142000
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.875564398867376e-05,
+      "loss": 1.1394,
+      "step": 142010
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.8754496058774e-05,
+      "loss": 1.1614,
+      "step": 142020
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.875334812887426e-05,
+      "loss": 1.1914,
+      "step": 142030
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.8752200198974514e-05,
+      "loss": 1.1646,
+      "step": 142040
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.8751052269074764e-05,
+      "loss": 1.1752,
+      "step": 142050
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.8749904339175013e-05,
+      "loss": 1.1527,
+      "step": 142060
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.874875640927527e-05,
+      "loss": 1.2652,
+      "step": 142070
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.874760847937552e-05,
+      "loss": 1.1142,
+      "step": 142080
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.8746460549475777e-05,
+      "loss": 1.2291,
+      "step": 142090
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.8745312619576026e-05,
+      "loss": 1.1983,
+      "step": 142100
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.8744164689676276e-05,
+      "loss": 1.2156,
+      "step": 142110
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.874301675977653e-05,
+      "loss": 1.1565,
+      "step": 142120
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.874186882987679e-05,
+      "loss": 1.2366,
+      "step": 142130
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.874072089997703e-05,
+      "loss": 1.2614,
+      "step": 142140
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.873957297007729e-05,
+      "loss": 1.2618,
+      "step": 142150
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.8738425040177546e-05,
+      "loss": 1.2003,
+      "step": 142160
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.873727711027779e-05,
+      "loss": 1.2312,
+      "step": 142170
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.8736129180378046e-05,
+      "loss": 1.3458,
+      "step": 142180
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.87349812504783e-05,
+      "loss": 1.2944,
+      "step": 142190
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.873383332057855e-05,
+      "loss": 1.223,
+      "step": 142200
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.87326853906788e-05,
+      "loss": 1.255,
+      "step": 142210
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.873153746077906e-05,
+      "loss": 1.2802,
+      "step": 142220
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.873038953087931e-05,
+      "loss": 1.392,
+      "step": 142230
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.8729241600979565e-05,
+      "loss": 1.1585,
+      "step": 142240
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.8728093671079815e-05,
+      "loss": 1.2449,
+      "step": 142250
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.8726945741180065e-05,
+      "loss": 1.2277,
+      "step": 142260
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.872579781128032e-05,
+      "loss": 1.3187,
+      "step": 142270
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.872464988138058e-05,
+      "loss": 1.2784,
+      "step": 142280
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.872350195148082e-05,
+      "loss": 1.1974,
+      "step": 142290
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.872235402158108e-05,
+      "loss": 1.2761,
+      "step": 142300
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.8721206091681334e-05,
+      "loss": 1.2292,
+      "step": 142310
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.872005816178158e-05,
+      "loss": 1.2686,
+      "step": 142320
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.8718910231881834e-05,
+      "loss": 1.1328,
+      "step": 142330
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.871776230198209e-05,
+      "loss": 1.1952,
+      "step": 142340
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.871661437208234e-05,
+      "loss": 1.2035,
+      "step": 142350
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.871546644218259e-05,
+      "loss": 1.2179,
+      "step": 142360
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.871431851228285e-05,
+      "loss": 1.1743,
+      "step": 142370
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.87131705823831e-05,
+      "loss": 1.1825,
+      "step": 142380
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.8712022652483353e-05,
+      "loss": 1.2356,
+      "step": 142390
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.87108747225836e-05,
+      "loss": 1.3361,
+      "step": 142400
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.870972679268385e-05,
+      "loss": 1.2367,
+      "step": 142410
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.870857886278411e-05,
+      "loss": 1.2553,
+      "step": 142420
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.8707430932884366e-05,
+      "loss": 1.2703,
+      "step": 142430
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.870628300298461e-05,
+      "loss": 1.0036,
+      "step": 142440
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.8705135073084866e-05,
+      "loss": 1.2702,
+      "step": 142450
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.870398714318512e-05,
+      "loss": 1.2762,
+      "step": 142460
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.8702839213285366e-05,
+      "loss": 1.1537,
+      "step": 142470
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.870169128338562e-05,
+      "loss": 1.1764,
+      "step": 142480
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.870054335348588e-05,
+      "loss": 1.212,
+      "step": 142490
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.869939542358613e-05,
+      "loss": 1.1433,
+      "step": 142500
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.869824749368638e-05,
+      "loss": 1.4271,
+      "step": 142510
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.8697099563786635e-05,
+      "loss": 1.274,
+      "step": 142520
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.8695951633886885e-05,
+      "loss": 1.1907,
+      "step": 142530
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.869480370398714e-05,
+      "loss": 1.0179,
+      "step": 142540
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.869365577408739e-05,
+      "loss": 1.2009,
+      "step": 142550
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.869250784418764e-05,
+      "loss": 1.1441,
+      "step": 142560
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.86913599142879e-05,
+      "loss": 1.2606,
+      "step": 142570
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.869021198438814e-05,
+      "loss": 1.3037,
+      "step": 142580
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.86890640544884e-05,
+      "loss": 1.168,
+      "step": 142590
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.8687916124588655e-05,
+      "loss": 1.355,
+      "step": 142600
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.8686768194688904e-05,
+      "loss": 1.153,
+      "step": 142610
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.8685620264789154e-05,
+      "loss": 1.1601,
+      "step": 142620
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.868447233488941e-05,
+      "loss": 1.1181,
+      "step": 142630
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.868332440498966e-05,
+      "loss": 1.2814,
+      "step": 142640
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.868217647508992e-05,
+      "loss": 1.231,
+      "step": 142650
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.868102854519017e-05,
+      "loss": 1.205,
+      "step": 142660
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.867988061529042e-05,
+      "loss": 1.1861,
+      "step": 142670
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.8678732685390674e-05,
+      "loss": 1.2177,
+      "step": 142680
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.867758475549093e-05,
+      "loss": 1.4232,
+      "step": 142690
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.8676436825591173e-05,
+      "loss": 1.149,
+      "step": 142700
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.867528889569143e-05,
+      "loss": 1.1701,
+      "step": 142710
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.867414096579169e-05,
+      "loss": 1.1889,
+      "step": 142720
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.867299303589193e-05,
+      "loss": 1.1367,
+      "step": 142730
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.8671845105992186e-05,
+      "loss": 1.2224,
+      "step": 142740
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.867069717609244e-05,
+      "loss": 1.1195,
+      "step": 142750
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.866954924619269e-05,
+      "loss": 1.2855,
+      "step": 142760
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.866840131629294e-05,
+      "loss": 1.2855,
+      "step": 142770
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.86672533863932e-05,
+      "loss": 1.1741,
+      "step": 142780
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.866610545649345e-05,
+      "loss": 1.1849,
+      "step": 142790
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.8664957526593706e-05,
+      "loss": 1.306,
+      "step": 142800
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.8663809596693956e-05,
+      "loss": 1.1635,
+      "step": 142810
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.8662661666794206e-05,
+      "loss": 1.2111,
+      "step": 142820
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.866151373689446e-05,
+      "loss": 1.3781,
+      "step": 142830
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.866036580699472e-05,
+      "loss": 1.3468,
+      "step": 142840
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.865921787709496e-05,
+      "loss": 1.1855,
+      "step": 142850
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.865806994719522e-05,
+      "loss": 1.3064,
+      "step": 142860
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.8656922017295475e-05,
+      "loss": 1.2651,
+      "step": 142870
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.865577408739572e-05,
+      "loss": 1.1105,
+      "step": 142880
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.8654626157495975e-05,
+      "loss": 1.2382,
+      "step": 142890
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.865347822759623e-05,
+      "loss": 1.3583,
+      "step": 142900
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.865233029769648e-05,
+      "loss": 1.1964,
+      "step": 142910
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.865118236779673e-05,
+      "loss": 1.3045,
+      "step": 142920
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.865003443789699e-05,
+      "loss": 1.164,
+      "step": 142930
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.864888650799724e-05,
+      "loss": 1.2391,
+      "step": 142940
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.8647738578097494e-05,
+      "loss": 1.3212,
+      "step": 142950
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.8646590648197744e-05,
+      "loss": 1.351,
+      "step": 142960
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.8645442718297994e-05,
+      "loss": 1.2021,
+      "step": 142970
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.864429478839825e-05,
+      "loss": 1.2072,
+      "step": 142980
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.864314685849851e-05,
+      "loss": 1.2461,
+      "step": 142990
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.864199892859875e-05,
+      "loss": 1.2221,
+      "step": 143000
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.864085099869901e-05,
+      "loss": 1.2466,
+      "step": 143010
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.8639703068799264e-05,
+      "loss": 1.2412,
+      "step": 143020
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.863855513889951e-05,
+      "loss": 1.0135,
+      "step": 143030
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.8637407208999763e-05,
+      "loss": 1.3027,
+      "step": 143040
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.863625927910002e-05,
+      "loss": 1.1327,
+      "step": 143050
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.863511134920027e-05,
+      "loss": 1.4479,
+      "step": 143060
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.863396341930052e-05,
+      "loss": 1.1348,
+      "step": 143070
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.8632815489400776e-05,
+      "loss": 1.2502,
+      "step": 143080
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.8631667559501026e-05,
+      "loss": 1.171,
+      "step": 143090
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.863051962960128e-05,
+      "loss": 1.1369,
+      "step": 143100
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.862937169970153e-05,
+      "loss": 1.1663,
+      "step": 143110
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.862822376980178e-05,
+      "loss": 1.1706,
+      "step": 143120
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.862707583990204e-05,
+      "loss": 1.2009,
+      "step": 143130
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.8625927910002296e-05,
+      "loss": 1.2592,
+      "step": 143140
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.862477998010254e-05,
+      "loss": 1.2402,
+      "step": 143150
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.8623632050202796e-05,
+      "loss": 1.3086,
+      "step": 143160
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.862248412030305e-05,
+      "loss": 1.2502,
+      "step": 143170
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.8621336190403295e-05,
+      "loss": 1.3706,
+      "step": 143180
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.862018826050355e-05,
+      "loss": 1.174,
+      "step": 143190
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.861904033060381e-05,
+      "loss": 1.1749,
+      "step": 143200
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.861789240070406e-05,
+      "loss": 1.3227,
+      "step": 143210
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.861674447080431e-05,
+      "loss": 1.1262,
+      "step": 143220
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.8615596540904565e-05,
+      "loss": 1.2154,
+      "step": 143230
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.8614448611004815e-05,
+      "loss": 1.1533,
+      "step": 143240
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.861330068110507e-05,
+      "loss": 1.3105,
+      "step": 143250
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.861215275120532e-05,
+      "loss": 1.2349,
+      "step": 143260
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.861100482130557e-05,
+      "loss": 1.1712,
+      "step": 143270
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.860985689140583e-05,
+      "loss": 1.1733,
+      "step": 143280
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.8608708961506084e-05,
+      "loss": 1.2049,
+      "step": 143290
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.860756103160633e-05,
+      "loss": 1.3001,
+      "step": 143300
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.8606413101706584e-05,
+      "loss": 1.1985,
+      "step": 143310
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.860526517180684e-05,
+      "loss": 1.2602,
+      "step": 143320
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.8604117241907084e-05,
+      "loss": 1.1344,
+      "step": 143330
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.860296931200734e-05,
+      "loss": 1.1773,
+      "step": 143340
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.86018213821076e-05,
+      "loss": 1.1132,
+      "step": 143350
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.860067345220785e-05,
+      "loss": 1.2256,
+      "step": 143360
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.85995255223081e-05,
+      "loss": 1.22,
+      "step": 143370
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.859837759240835e-05,
+      "loss": 1.2108,
+      "step": 143380
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.85972296625086e-05,
+      "loss": 1.2544,
+      "step": 143390
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.859608173260886e-05,
+      "loss": 1.2019,
+      "step": 143400
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.859493380270911e-05,
+      "loss": 1.341,
+      "step": 143410
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.859378587280936e-05,
+      "loss": 1.1518,
+      "step": 143420
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.8592637942909616e-05,
+      "loss": 1.2153,
+      "step": 143430
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.859149001300987e-05,
+      "loss": 1.2325,
+      "step": 143440
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.8590342083110116e-05,
+      "loss": 1.1598,
+      "step": 143450
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.858919415321037e-05,
+      "loss": 1.2582,
+      "step": 143460
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.858804622331063e-05,
+      "loss": 1.2635,
+      "step": 143470
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.858689829341087e-05,
+      "loss": 1.1914,
+      "step": 143480
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.858575036351113e-05,
+      "loss": 1.2718,
+      "step": 143490
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.8584602433611385e-05,
+      "loss": 1.137,
+      "step": 143500
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.8583454503711635e-05,
+      "loss": 1.2536,
+      "step": 143510
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.8582306573811885e-05,
+      "loss": 1.2458,
+      "step": 143520
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.858115864391214e-05,
+      "loss": 1.1792,
+      "step": 143530
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.858001071401239e-05,
+      "loss": 1.1744,
+      "step": 143540
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.857886278411265e-05,
+      "loss": 1.1895,
+      "step": 143550
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.85777148542129e-05,
+      "loss": 1.1432,
+      "step": 143560
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.857656692431315e-05,
+      "loss": 1.1675,
+      "step": 143570
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.8575418994413405e-05,
+      "loss": 1.1554,
+      "step": 143580
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.857427106451366e-05,
+      "loss": 1.2377,
+      "step": 143590
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.8573123134613904e-05,
+      "loss": 1.1417,
+      "step": 143600
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.857197520471416e-05,
+      "loss": 1.25,
+      "step": 143610
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.857082727481442e-05,
+      "loss": 1.1102,
+      "step": 143620
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.856967934491466e-05,
+      "loss": 1.3712,
+      "step": 143630
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.856853141501492e-05,
+      "loss": 1.348,
+      "step": 143640
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.8567383485115174e-05,
+      "loss": 1.1657,
+      "step": 143650
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.8566235555215424e-05,
+      "loss": 1.201,
+      "step": 143660
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.8565087625315674e-05,
+      "loss": 1.0918,
+      "step": 143670
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.856393969541593e-05,
+      "loss": 1.2318,
+      "step": 143680
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.856279176551618e-05,
+      "loss": 1.1988,
+      "step": 143690
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.856164383561644e-05,
+      "loss": 1.2355,
+      "step": 143700
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.8560495905716687e-05,
+      "loss": 1.2044,
+      "step": 143710
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.8559347975816936e-05,
+      "loss": 1.3375,
+      "step": 143720
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.855820004591719e-05,
+      "loss": 1.2017,
+      "step": 143730
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.855705211601745e-05,
+      "loss": 1.2305,
+      "step": 143740
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.855590418611769e-05,
+      "loss": 1.2388,
+      "step": 143750
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.855475625621795e-05,
+      "loss": 1.2177,
+      "step": 143760
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.8553608326318206e-05,
+      "loss": 1.2911,
+      "step": 143770
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.855246039641845e-05,
+      "loss": 1.2102,
+      "step": 143780
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.8551312466518706e-05,
+      "loss": 1.2399,
+      "step": 143790
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.855016453661896e-05,
+      "loss": 1.1732,
+      "step": 143800
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.854901660671921e-05,
+      "loss": 1.247,
+      "step": 143810
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.854786867681946e-05,
+      "loss": 1.3443,
+      "step": 143820
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.854672074691972e-05,
+      "loss": 1.2085,
+      "step": 143830
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.854557281701997e-05,
+      "loss": 1.1827,
+      "step": 143840
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.8544424887120225e-05,
+      "loss": 1.4397,
+      "step": 143850
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.8543276957220475e-05,
+      "loss": 1.2706,
+      "step": 143860
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.8542129027320725e-05,
+      "loss": 1.1193,
+      "step": 143870
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.854098109742098e-05,
+      "loss": 1.1435,
+      "step": 143880
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.853983316752124e-05,
+      "loss": 1.2237,
+      "step": 143890
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.853868523762148e-05,
+      "loss": 1.1375,
+      "step": 143900
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.853753730772174e-05,
+      "loss": 1.1593,
+      "step": 143910
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.8536389377821995e-05,
+      "loss": 1.1932,
+      "step": 143920
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.853524144792224e-05,
+      "loss": 1.1059,
+      "step": 143930
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.8534093518022494e-05,
+      "loss": 1.4135,
+      "step": 143940
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.8532945588122744e-05,
+      "loss": 1.1468,
+      "step": 143950
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.8531797658223e-05,
+      "loss": 1.2059,
+      "step": 143960
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.853064972832325e-05,
+      "loss": 1.3684,
+      "step": 143970
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.85295017984235e-05,
+      "loss": 1.279,
+      "step": 143980
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.852835386852376e-05,
+      "loss": 1.1427,
+      "step": 143990
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.8527205938624014e-05,
+      "loss": 1.2056,
+      "step": 144000
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.852605800872426e-05,
+      "loss": 1.1683,
+      "step": 144010
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.8524910078824513e-05,
+      "loss": 1.172,
+      "step": 144020
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.852376214892477e-05,
+      "loss": 1.1913,
+      "step": 144030
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.852261421902501e-05,
+      "loss": 1.2239,
+      "step": 144040
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.852146628912527e-05,
+      "loss": 1.1659,
+      "step": 144050
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.8520318359225526e-05,
+      "loss": 1.3482,
+      "step": 144060
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.8519170429325776e-05,
+      "loss": 1.2052,
+      "step": 144070
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.8518022499426026e-05,
+      "loss": 1.2878,
+      "step": 144080
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.851687456952628e-05,
+      "loss": 1.0545,
+      "step": 144090
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.851572663962653e-05,
+      "loss": 1.1784,
+      "step": 144100
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.851457870972679e-05,
+      "loss": 1.2655,
+      "step": 144110
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.851343077982704e-05,
+      "loss": 1.2931,
+      "step": 144120
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.851228284992729e-05,
+      "loss": 1.1734,
+      "step": 144130
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.8511134920027546e-05,
+      "loss": 1.2226,
+      "step": 144140
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.85099869901278e-05,
+      "loss": 1.324,
+      "step": 144150
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.8508839060228045e-05,
+      "loss": 1.304,
+      "step": 144160
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.85076911303283e-05,
+      "loss": 1.2188,
+      "step": 144170
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.850654320042856e-05,
+      "loss": 1.2196,
+      "step": 144180
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.85053952705288e-05,
+      "loss": 1.1547,
+      "step": 144190
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.850424734062906e-05,
+      "loss": 1.098,
+      "step": 144200
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.8503099410729315e-05,
+      "loss": 1.2057,
+      "step": 144210
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.8501951480829565e-05,
+      "loss": 1.0277,
+      "step": 144220
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.8500803550929815e-05,
+      "loss": 1.256,
+      "step": 144230
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.849965562103007e-05,
+      "loss": 1.2484,
+      "step": 144240
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.849850769113032e-05,
+      "loss": 1.2921,
+      "step": 144250
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.849735976123058e-05,
+      "loss": 1.2375,
+      "step": 144260
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.849621183133083e-05,
+      "loss": 1.2462,
+      "step": 144270
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.849506390143108e-05,
+      "loss": 1.0799,
+      "step": 144280
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.8493915971531334e-05,
+      "loss": 1.2993,
+      "step": 144290
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.849276804163159e-05,
+      "loss": 1.3114,
+      "step": 144300
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.8491620111731834e-05,
+      "loss": 1.1807,
+      "step": 144310
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.849047218183209e-05,
+      "loss": 1.1151,
+      "step": 144320
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.848932425193235e-05,
+      "loss": 1.3235,
+      "step": 144330
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.848817632203259e-05,
+      "loss": 1.2629,
+      "step": 144340
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.848702839213285e-05,
+      "loss": 1.1454,
+      "step": 144350
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.84858804622331e-05,
+      "loss": 1.2586,
+      "step": 144360
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.848473253233335e-05,
+      "loss": 1.1364,
+      "step": 144370
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.84835846024336e-05,
+      "loss": 1.1508,
+      "step": 144380
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.848243667253386e-05,
+      "loss": 1.2587,
+      "step": 144390
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.848128874263411e-05,
+      "loss": 1.1034,
+      "step": 144400
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.8480140812734366e-05,
+      "loss": 1.2457,
+      "step": 144410
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.8478992882834616e-05,
+      "loss": 1.1676,
+      "step": 144420
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.8477844952934866e-05,
+      "loss": 1.2678,
+      "step": 144430
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.847669702303512e-05,
+      "loss": 1.3444,
+      "step": 144440
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.847554909313538e-05,
+      "loss": 1.2461,
+      "step": 144450
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.847440116323562e-05,
+      "loss": 1.2005,
+      "step": 144460
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.847325323333588e-05,
+      "loss": 1.1883,
+      "step": 144470
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.8472105303436135e-05,
+      "loss": 1.1155,
+      "step": 144480
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.847095737353638e-05,
+      "loss": 1.2691,
+      "step": 144490
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.8469809443636635e-05,
+      "loss": 1.193,
+      "step": 144500
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.846866151373689e-05,
+      "loss": 1.265,
+      "step": 144510
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.846751358383714e-05,
+      "loss": 1.1883,
+      "step": 144520
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.846636565393739e-05,
+      "loss": 1.2082,
+      "step": 144530
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.846521772403765e-05,
+      "loss": 1.273,
+      "step": 144540
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.84640697941379e-05,
+      "loss": 1.0202,
+      "step": 144550
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.8462921864238155e-05,
+      "loss": 1.1321,
+      "step": 144560
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.8461773934338404e-05,
+      "loss": 1.2094,
+      "step": 144570
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.8460626004438654e-05,
+      "loss": 1.2606,
+      "step": 144580
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.845947807453891e-05,
+      "loss": 1.3093,
+      "step": 144590
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.845833014463917e-05,
+      "loss": 1.1875,
+      "step": 144600
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.845718221473941e-05,
+      "loss": 1.2939,
+      "step": 144610
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.845603428483967e-05,
+      "loss": 1.2478,
+      "step": 144620
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.8454886354939924e-05,
+      "loss": 1.3019,
+      "step": 144630
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.845373842504017e-05,
+      "loss": 1.2604,
+      "step": 144640
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.8452590495140424e-05,
+      "loss": 1.041,
+      "step": 144650
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.845144256524068e-05,
+      "loss": 1.2661,
+      "step": 144660
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.845029463534093e-05,
+      "loss": 1.2281,
+      "step": 144670
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.844914670544118e-05,
+      "loss": 1.2161,
+      "step": 144680
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.8447998775541437e-05,
+      "loss": 1.2437,
+      "step": 144690
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.8446850845641686e-05,
+      "loss": 1.1784,
+      "step": 144700
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.844570291574194e-05,
+      "loss": 1.198,
+      "step": 144710
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.844455498584219e-05,
+      "loss": 1.1448,
+      "step": 144720
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.844340705594244e-05,
+      "loss": 1.1596,
+      "step": 144730
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.84422591260427e-05,
+      "loss": 1.1626,
+      "step": 144740
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.8441111196142956e-05,
+      "loss": 1.3156,
+      "step": 144750
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.84399632662432e-05,
+      "loss": 1.2447,
+      "step": 144760
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.8438815336343456e-05,
+      "loss": 1.1458,
+      "step": 144770
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.843766740644371e-05,
+      "loss": 1.1384,
+      "step": 144780
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.8436519476543955e-05,
+      "loss": 1.3439,
+      "step": 144790
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.843537154664421e-05,
+      "loss": 1.3019,
+      "step": 144800
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.843422361674447e-05,
+      "loss": 1.2785,
+      "step": 144810
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.843307568684472e-05,
+      "loss": 1.305,
+      "step": 144820
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.843192775694497e-05,
+      "loss": 1.239,
+      "step": 144830
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.8430779827045225e-05,
+      "loss": 1.2348,
+      "step": 144840
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.8429631897145475e-05,
+      "loss": 1.1316,
+      "step": 144850
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.842848396724573e-05,
+      "loss": 1.3163,
+      "step": 144860
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.842733603734598e-05,
+      "loss": 1.2452,
+      "step": 144870
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.842618810744623e-05,
+      "loss": 1.3371,
+      "step": 144880
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.842504017754649e-05,
+      "loss": 1.2448,
+      "step": 144890
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.8423892247646745e-05,
+      "loss": 1.1266,
+      "step": 144900
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.842274431774699e-05,
+      "loss": 1.1842,
+      "step": 144910
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.8421596387847244e-05,
+      "loss": 1.107,
+      "step": 144920
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.84204484579475e-05,
+      "loss": 1.3275,
+      "step": 144930
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.8419300528047744e-05,
+      "loss": 1.2624,
+      "step": 144940
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.8418152598148e-05,
+      "loss": 1.1829,
+      "step": 144950
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.841700466824826e-05,
+      "loss": 1.3632,
+      "step": 144960
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.841585673834851e-05,
+      "loss": 1.1552,
+      "step": 144970
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.841470880844876e-05,
+      "loss": 1.2143,
+      "step": 144980
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.8413560878549014e-05,
+      "loss": 1.2273,
+      "step": 144990
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.8412412948649263e-05,
+      "loss": 1.2009,
+      "step": 145000
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.841126501874952e-05,
+      "loss": 1.0886,
+      "step": 145010
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.841011708884977e-05,
+      "loss": 1.2735,
+      "step": 145020
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.840896915895002e-05,
+      "loss": 1.2339,
+      "step": 145030
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.8407821229050276e-05,
+      "loss": 1.2383,
+      "step": 145040
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.840667329915053e-05,
+      "loss": 1.2073,
+      "step": 145050
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.8405525369250776e-05,
+      "loss": 1.305,
+      "step": 145060
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.840437743935103e-05,
+      "loss": 1.1422,
+      "step": 145070
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.840322950945129e-05,
+      "loss": 1.2174,
+      "step": 145080
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.840208157955153e-05,
+      "loss": 1.1375,
+      "step": 145090
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.840093364965179e-05,
+      "loss": 1.2422,
+      "step": 145100
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.8399785719752046e-05,
+      "loss": 1.308,
+      "step": 145110
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.8398637789852296e-05,
+      "loss": 1.361,
+      "step": 145120
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.8397489859952545e-05,
+      "loss": 1.2538,
+      "step": 145130
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.83963419300528e-05,
+      "loss": 1.1646,
+      "step": 145140
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.839519400015305e-05,
+      "loss": 1.1906,
+      "step": 145150
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.839404607025331e-05,
+      "loss": 1.2538,
+      "step": 145160
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.839289814035356e-05,
+      "loss": 1.1113,
+      "step": 145170
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.839175021045381e-05,
+      "loss": 1.1485,
+      "step": 145180
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.8390602280554065e-05,
+      "loss": 1.2394,
+      "step": 145190
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.838945435065432e-05,
+      "loss": 1.2877,
+      "step": 145200
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.8388306420754565e-05,
+      "loss": 1.3073,
+      "step": 145210
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.838715849085482e-05,
+      "loss": 1.2419,
+      "step": 145220
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.838601056095508e-05,
+      "loss": 1.381,
+      "step": 145230
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.838486263105532e-05,
+      "loss": 1.1914,
+      "step": 145240
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.838371470115558e-05,
+      "loss": 1.2355,
+      "step": 145250
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.8382566771255834e-05,
+      "loss": 1.1852,
+      "step": 145260
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.8381418841356084e-05,
+      "loss": 1.3046,
+      "step": 145270
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.8380270911456334e-05,
+      "loss": 1.1915,
+      "step": 145280
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.837912298155659e-05,
+      "loss": 1.1865,
+      "step": 145290
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.837797505165684e-05,
+      "loss": 1.1539,
+      "step": 145300
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.83768271217571e-05,
+      "loss": 1.172,
+      "step": 145310
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.837567919185734e-05,
+      "loss": 1.1519,
+      "step": 145320
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.83745312619576e-05,
+      "loss": 1.1514,
+      "step": 145330
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.837338333205785e-05,
+      "loss": 1.2965,
+      "step": 145340
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.8372235402158096e-05,
+      "loss": 1.2352,
+      "step": 145350
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.837108747225835e-05,
+      "loss": 1.3279,
+      "step": 145360
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.836993954235861e-05,
+      "loss": 1.3556,
+      "step": 145370
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.836879161245886e-05,
+      "loss": 1.159,
+      "step": 145380
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.836764368255911e-05,
+      "loss": 1.2138,
+      "step": 145390
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.8366495752659366e-05,
+      "loss": 1.2052,
+      "step": 145400
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.8365347822759616e-05,
+      "loss": 1.1991,
+      "step": 145410
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.836419989285987e-05,
+      "loss": 1.3018,
+      "step": 145420
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.836305196296012e-05,
+      "loss": 1.1607,
+      "step": 145430
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.836190403306037e-05,
+      "loss": 1.2701,
+      "step": 145440
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.836075610316063e-05,
+      "loss": 1.0505,
+      "step": 145450
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.8359608173260885e-05,
+      "loss": 1.1685,
+      "step": 145460
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.835846024336113e-05,
+      "loss": 1.1289,
+      "step": 145470
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.8357312313461385e-05,
+      "loss": 1.0702,
+      "step": 145480
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.835616438356164e-05,
+      "loss": 1.2453,
+      "step": 145490
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.8355016453661885e-05,
+      "loss": 1.1247,
+      "step": 145500
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.835386852376214e-05,
+      "loss": 1.4349,
+      "step": 145510
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.83527205938624e-05,
+      "loss": 1.1715,
+      "step": 145520
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.835157266396265e-05,
+      "loss": 1.1327,
+      "step": 145530
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.83504247340629e-05,
+      "loss": 1.14,
+      "step": 145540
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.8349276804163154e-05,
+      "loss": 1.1491,
+      "step": 145550
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.8348128874263404e-05,
+      "loss": 1.1757,
+      "step": 145560
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.834698094436366e-05,
+      "loss": 1.3399,
+      "step": 145570
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.834583301446391e-05,
+      "loss": 1.1828,
+      "step": 145580
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.834468508456416e-05,
+      "loss": 1.1555,
+      "step": 145590
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.834353715466442e-05,
+      "loss": 1.2854,
+      "step": 145600
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.8342389224764674e-05,
+      "loss": 1.214,
+      "step": 145610
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.834124129486492e-05,
+      "loss": 1.148,
+      "step": 145620
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.8340093364965174e-05,
+      "loss": 1.2397,
+      "step": 145630
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.833894543506543e-05,
+      "loss": 1.2461,
+      "step": 145640
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.833779750516567e-05,
+      "loss": 1.3309,
+      "step": 145650
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.833664957526593e-05,
+      "loss": 1.123,
+      "step": 145660
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.8335501645366187e-05,
+      "loss": 1.0678,
+      "step": 145670
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.8334353715466436e-05,
+      "loss": 1.2286,
+      "step": 145680
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.8333205785566686e-05,
+      "loss": 1.1157,
+      "step": 145690
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.833205785566694e-05,
+      "loss": 1.1462,
+      "step": 145700
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.833090992576719e-05,
+      "loss": 1.1547,
+      "step": 145710
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.832976199586745e-05,
+      "loss": 1.2626,
+      "step": 145720
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.83286140659677e-05,
+      "loss": 1.1306,
+      "step": 145730
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.832746613606795e-05,
+      "loss": 1.1543,
+      "step": 145740
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.8326318206168206e-05,
+      "loss": 1.0985,
+      "step": 145750
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.832517027626846e-05,
+      "loss": 1.2552,
+      "step": 145760
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.8324022346368705e-05,
+      "loss": 1.1807,
+      "step": 145770
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.832287441646896e-05,
+      "loss": 1.2754,
+      "step": 145780
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.832172648656922e-05,
+      "loss": 1.1001,
+      "step": 145790
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.832057855666946e-05,
+      "loss": 1.1778,
+      "step": 145800
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.831943062676972e-05,
+      "loss": 1.372,
+      "step": 145810
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.8318282696869975e-05,
+      "loss": 1.092,
+      "step": 145820
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.8317134766970225e-05,
+      "loss": 1.3605,
+      "step": 145830
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.8315986837070475e-05,
+      "loss": 1.2636,
+      "step": 145840
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.831483890717073e-05,
+      "loss": 1.2697,
+      "step": 145850
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.831369097727098e-05,
+      "loss": 1.0542,
+      "step": 145860
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.831254304737124e-05,
+      "loss": 1.1504,
+      "step": 145870
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.831139511747149e-05,
+      "loss": 1.1777,
+      "step": 145880
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.831024718757174e-05,
+      "loss": 1.2454,
+      "step": 145890
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.8309099257671994e-05,
+      "loss": 1.1971,
+      "step": 145900
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.830795132777225e-05,
+      "loss": 1.2469,
+      "step": 145910
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.8306803397872494e-05,
+      "loss": 1.1519,
+      "step": 145920
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.830565546797275e-05,
+      "loss": 1.1302,
+      "step": 145930
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.830450753807301e-05,
+      "loss": 1.1217,
+      "step": 145940
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.830335960817325e-05,
+      "loss": 1.0786,
+      "step": 145950
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.830221167827351e-05,
+      "loss": 1.1963,
+      "step": 145960
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.8301063748373764e-05,
+      "loss": 1.3222,
+      "step": 145970
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.8299915818474013e-05,
+      "loss": 1.2129,
+      "step": 145980
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.829876788857426e-05,
+      "loss": 1.2199,
+      "step": 145990
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.829761995867452e-05,
+      "loss": 1.1944,
+      "step": 146000
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.829647202877477e-05,
+      "loss": 1.3053,
+      "step": 146010
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.8295324098875026e-05,
+      "loss": 1.2762,
+      "step": 146020
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.8294176168975276e-05,
+      "loss": 1.2367,
+      "step": 146030
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.8293028239075526e-05,
+      "loss": 1.3629,
+      "step": 146040
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.829188030917578e-05,
+      "loss": 1.2336,
+      "step": 146050
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.829073237927604e-05,
+      "loss": 1.2518,
+      "step": 146060
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.828958444937628e-05,
+      "loss": 1.0837,
+      "step": 146070
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.828843651947654e-05,
+      "loss": 1.244,
+      "step": 146080
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.8287288589576796e-05,
+      "loss": 1.3589,
+      "step": 146090
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.828614065967704e-05,
+      "loss": 1.2268,
+      "step": 146100
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.8284992729777295e-05,
+      "loss": 1.252,
+      "step": 146110
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.828384479987755e-05,
+      "loss": 1.2579,
+      "step": 146120
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.82826968699778e-05,
+      "loss": 1.3241,
+      "step": 146130
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.828154894007805e-05,
+      "loss": 1.1024,
+      "step": 146140
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.828040101017831e-05,
+      "loss": 1.3608,
+      "step": 146150
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.827925308027856e-05,
+      "loss": 1.3353,
+      "step": 146160
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.8278105150378815e-05,
+      "loss": 1.1948,
+      "step": 146170
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.8276957220479065e-05,
+      "loss": 1.2074,
+      "step": 146180
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.8275809290579315e-05,
+      "loss": 1.292,
+      "step": 146190
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.827466136067957e-05,
+      "loss": 1.2505,
+      "step": 146200
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.827351343077983e-05,
+      "loss": 1.2066,
+      "step": 146210
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.827236550088007e-05,
+      "loss": 1.355,
+      "step": 146220
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.827121757098033e-05,
+      "loss": 1.2211,
+      "step": 146230
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.8270069641080584e-05,
+      "loss": 1.1765,
+      "step": 146240
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.826892171118083e-05,
+      "loss": 1.1675,
+      "step": 146250
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.8267773781281084e-05,
+      "loss": 1.2054,
+      "step": 146260
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.826662585138134e-05,
+      "loss": 1.2848,
+      "step": 146270
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.826547792148159e-05,
+      "loss": 1.2307,
+      "step": 146280
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.826432999158184e-05,
+      "loss": 1.278,
+      "step": 146290
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.82631820616821e-05,
+      "loss": 1.2742,
+      "step": 146300
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.826203413178235e-05,
+      "loss": 1.2347,
+      "step": 146310
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.82608862018826e-05,
+      "loss": 1.2073,
+      "step": 146320
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.825973827198285e-05,
+      "loss": 1.2756,
+      "step": 146330
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.82585903420831e-05,
+      "loss": 1.2055,
+      "step": 146340
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.825744241218336e-05,
+      "loss": 1.3026,
+      "step": 146350
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.8256294482283616e-05,
+      "loss": 1.2622,
+      "step": 146360
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.825514655238386e-05,
+      "loss": 1.1302,
+      "step": 146370
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.8253998622484116e-05,
+      "loss": 1.3009,
+      "step": 146380
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.825285069258437e-05,
+      "loss": 1.2252,
+      "step": 146390
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.8251702762684616e-05,
+      "loss": 1.2544,
+      "step": 146400
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.825055483278487e-05,
+      "loss": 1.1539,
+      "step": 146410
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.824940690288513e-05,
+      "loss": 1.1845,
+      "step": 146420
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.824825897298538e-05,
+      "loss": 1.2104,
+      "step": 146430
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.824711104308563e-05,
+      "loss": 1.2485,
+      "step": 146440
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.8245963113185885e-05,
+      "loss": 1.304,
+      "step": 146450
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.8244815183286135e-05,
+      "loss": 1.2046,
+      "step": 146460
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.824366725338639e-05,
+      "loss": 1.261,
+      "step": 146470
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.824251932348664e-05,
+      "loss": 1.1271,
+      "step": 146480
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.824137139358689e-05,
+      "loss": 1.1788,
+      "step": 146490
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.824022346368715e-05,
+      "loss": 1.1992,
+      "step": 146500
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.8239075533787405e-05,
+      "loss": 1.1718,
+      "step": 146510
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.823792760388765e-05,
+      "loss": 1.1571,
+      "step": 146520
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.8236779673987904e-05,
+      "loss": 1.2028,
+      "step": 146530
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.823563174408816e-05,
+      "loss": 1.1826,
+      "step": 146540
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.8234483814188404e-05,
+      "loss": 1.2033,
+      "step": 146550
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.823333588428866e-05,
+      "loss": 1.2332,
+      "step": 146560
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.823218795438892e-05,
+      "loss": 1.2318,
+      "step": 146570
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.823104002448917e-05,
+      "loss": 1.3368,
+      "step": 146580
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.822989209458942e-05,
+      "loss": 1.134,
+      "step": 146590
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.8228744164689674e-05,
+      "loss": 1.105,
+      "step": 146600
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.8227596234789924e-05,
+      "loss": 1.2565,
+      "step": 146610
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.822644830489018e-05,
+      "loss": 1.2493,
+      "step": 146620
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.822530037499043e-05,
+      "loss": 1.2589,
+      "step": 146630
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.822415244509068e-05,
+      "loss": 1.2714,
+      "step": 146640
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.8223004515190937e-05,
+      "loss": 1.2049,
+      "step": 146650
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.822185658529119e-05,
+      "loss": 1.2022,
+      "step": 146660
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.8220708655391436e-05,
+      "loss": 1.3685,
+      "step": 146670
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.821956072549169e-05,
+      "loss": 1.2816,
+      "step": 146680
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.821841279559194e-05,
+      "loss": 1.2814,
+      "step": 146690
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.821726486569219e-05,
+      "loss": 1.2592,
+      "step": 146700
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.821611693579245e-05,
+      "loss": 1.1059,
+      "step": 146710
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.82149690058927e-05,
+      "loss": 1.106,
+      "step": 146720
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.8213821075992956e-05,
+      "loss": 1.2496,
+      "step": 146730
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.8212673146093206e-05,
+      "loss": 1.3111,
+      "step": 146740
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.8211525216193455e-05,
+      "loss": 1.0925,
+      "step": 146750
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.821037728629371e-05,
+      "loss": 1.3072,
+      "step": 146760
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.820922935639397e-05,
+      "loss": 1.3638,
+      "step": 146770
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.820808142649421e-05,
+      "loss": 1.2097,
+      "step": 146780
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.820693349659447e-05,
+      "loss": 1.1486,
+      "step": 146790
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.8205785566694725e-05,
+      "loss": 1.1912,
+      "step": 146800
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.820463763679497e-05,
+      "loss": 1.1714,
+      "step": 146810
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.8203489706895225e-05,
+      "loss": 1.1451,
+      "step": 146820
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.820234177699548e-05,
+      "loss": 1.1666,
+      "step": 146830
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.820119384709573e-05,
+      "loss": 1.118,
+      "step": 146840
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.820004591719598e-05,
+      "loss": 1.0782,
+      "step": 146850
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.819889798729624e-05,
+      "loss": 1.2202,
+      "step": 146860
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.819775005739649e-05,
+      "loss": 1.3435,
+      "step": 146870
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.8196602127496744e-05,
+      "loss": 1.2783,
+      "step": 146880
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.8195454197596994e-05,
+      "loss": 1.2029,
+      "step": 146890
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.8194306267697244e-05,
+      "loss": 1.0759,
+      "step": 146900
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.81931583377975e-05,
+      "loss": 1.2253,
+      "step": 146910
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.819201040789776e-05,
+      "loss": 1.0294,
+      "step": 146920
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.8190862477998e-05,
+      "loss": 1.2121,
+      "step": 146930
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.818971454809826e-05,
+      "loss": 1.293,
+      "step": 146940
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.8188566618198514e-05,
+      "loss": 1.0791,
+      "step": 146950
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.818741868829876e-05,
+      "loss": 1.337,
+      "step": 146960
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.818627075839901e-05,
+      "loss": 1.3277,
+      "step": 146970
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.818512282849927e-05,
+      "loss": 1.1504,
+      "step": 146980
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.818397489859952e-05,
+      "loss": 1.1762,
+      "step": 146990
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.818282696869977e-05,
+      "loss": 1.1474,
+      "step": 147000
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.8181679038800026e-05,
+      "loss": 1.2084,
+      "step": 147010
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.8180531108900276e-05,
+      "loss": 1.2265,
+      "step": 147020
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.817938317900053e-05,
+      "loss": 1.1322,
+      "step": 147030
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.817823524910078e-05,
+      "loss": 1.1734,
+      "step": 147040
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.817708731920103e-05,
+      "loss": 1.3196,
+      "step": 147050
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.817593938930129e-05,
+      "loss": 1.1712,
+      "step": 147060
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.8174791459401546e-05,
+      "loss": 1.1237,
+      "step": 147070
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.817364352950179e-05,
+      "loss": 1.3735,
+      "step": 147080
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.8172495599602045e-05,
+      "loss": 1.2876,
+      "step": 147090
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.81713476697023e-05,
+      "loss": 1.1618,
+      "step": 147100
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.8170199739802545e-05,
+      "loss": 1.1884,
+      "step": 147110
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.81690518099028e-05,
+      "loss": 1.3297,
+      "step": 147120
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.816790388000306e-05,
+      "loss": 1.1287,
+      "step": 147130
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.816675595010331e-05,
+      "loss": 1.2127,
+      "step": 147140
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.816560802020356e-05,
+      "loss": 1.2527,
+      "step": 147150
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.8164460090303815e-05,
+      "loss": 1.0696,
+      "step": 147160
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.8163312160404065e-05,
+      "loss": 1.4118,
+      "step": 147170
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.816216423050432e-05,
+      "loss": 1.2037,
+      "step": 147180
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.816101630060457e-05,
+      "loss": 1.0467,
+      "step": 147190
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.815986837070482e-05,
+      "loss": 1.3428,
+      "step": 147200
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.815872044080508e-05,
+      "loss": 1.06,
+      "step": 147210
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.8157572510905334e-05,
+      "loss": 1.1531,
+      "step": 147220
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.815642458100558e-05,
+      "loss": 1.239,
+      "step": 147230
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.8155276651105834e-05,
+      "loss": 1.2047,
+      "step": 147240
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.815412872120609e-05,
+      "loss": 1.2073,
+      "step": 147250
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.8152980791306334e-05,
+      "loss": 1.3658,
+      "step": 147260
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.815183286140659e-05,
+      "loss": 1.25,
+      "step": 147270
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.815068493150685e-05,
+      "loss": 1.1937,
+      "step": 147280
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.81495370016071e-05,
+      "loss": 1.2015,
+      "step": 147290
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.8148389071707347e-05,
+      "loss": 1.2572,
+      "step": 147300
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.81472411418076e-05,
+      "loss": 1.1358,
+      "step": 147310
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.814609321190785e-05,
+      "loss": 1.267,
+      "step": 147320
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.814494528200811e-05,
+      "loss": 1.2523,
+      "step": 147330
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.814379735210836e-05,
+      "loss": 1.2811,
+      "step": 147340
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.814264942220861e-05,
+      "loss": 1.1695,
+      "step": 147350
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.8141501492308866e-05,
+      "loss": 1.1332,
+      "step": 147360
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.814035356240912e-05,
+      "loss": 1.2746,
+      "step": 147370
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.8139205632509366e-05,
+      "loss": 1.1769,
+      "step": 147380
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.813805770260962e-05,
+      "loss": 1.2711,
+      "step": 147390
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.813690977270988e-05,
+      "loss": 1.1446,
+      "step": 147400
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.813576184281012e-05,
+      "loss": 1.2395,
+      "step": 147410
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.813461391291038e-05,
+      "loss": 1.151,
+      "step": 147420
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.8133465983010635e-05,
+      "loss": 1.2911,
+      "step": 147430
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.8132318053110885e-05,
+      "loss": 1.1837,
+      "step": 147440
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.8131170123211135e-05,
+      "loss": 1.1607,
+      "step": 147450
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.813002219331139e-05,
+      "loss": 1.2325,
+      "step": 147460
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.812887426341164e-05,
+      "loss": 1.3158,
+      "step": 147470
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.81277263335119e-05,
+      "loss": 1.3466,
+      "step": 147480
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.812657840361215e-05,
+      "loss": 1.3679,
+      "step": 147490
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.81254304737124e-05,
+      "loss": 1.2242,
+      "step": 147500
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.8124282543812654e-05,
+      "loss": 1.1679,
+      "step": 147510
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.812313461391291e-05,
+      "loss": 1.1722,
+      "step": 147520
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.8121986684013154e-05,
+      "loss": 1.2328,
+      "step": 147530
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.812083875411341e-05,
+      "loss": 1.055,
+      "step": 147540
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.811969082421367e-05,
+      "loss": 1.1715,
+      "step": 147550
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.811854289431391e-05,
+      "loss": 1.1965,
+      "step": 147560
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.811739496441417e-05,
+      "loss": 1.2728,
+      "step": 147570
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.8116247034514424e-05,
+      "loss": 1.2003,
+      "step": 147580
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.8115099104614674e-05,
+      "loss": 1.269,
+      "step": 147590
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.8113951174714923e-05,
+      "loss": 1.2866,
+      "step": 147600
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.811280324481518e-05,
+      "loss": 1.1554,
+      "step": 147610
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.811165531491543e-05,
+      "loss": 1.1941,
+      "step": 147620
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.8110507385015687e-05,
+      "loss": 1.1486,
+      "step": 147630
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.8109359455115936e-05,
+      "loss": 1.2537,
+      "step": 147640
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.8108211525216186e-05,
+      "loss": 1.2648,
+      "step": 147650
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.810706359531644e-05,
+      "loss": 1.2076,
+      "step": 147660
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.81059156654167e-05,
+      "loss": 1.207,
+      "step": 147670
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.810476773551694e-05,
+      "loss": 1.3622,
+      "step": 147680
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.81036198056172e-05,
+      "loss": 1.2226,
+      "step": 147690
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.8102471875717456e-05,
+      "loss": 1.2993,
+      "step": 147700
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.81013239458177e-05,
+      "loss": 1.0752,
+      "step": 147710
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.8100176015917956e-05,
+      "loss": 1.3492,
+      "step": 147720
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.809902808601821e-05,
+      "loss": 1.0583,
+      "step": 147730
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.809788015611846e-05,
+      "loss": 1.1919,
+      "step": 147740
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.809673222621871e-05,
+      "loss": 1.2452,
+      "step": 147750
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.809558429631897e-05,
+      "loss": 1.3307,
+      "step": 147760
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.809443636641922e-05,
+      "loss": 1.2884,
+      "step": 147770
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.8093288436519475e-05,
+      "loss": 1.2024,
+      "step": 147780
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.8092140506619725e-05,
+      "loss": 1.3767,
+      "step": 147790
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.8090992576719975e-05,
+      "loss": 1.2097,
+      "step": 147800
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.808984464682023e-05,
+      "loss": 1.1732,
+      "step": 147810
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.808869671692049e-05,
+      "loss": 1.1348,
+      "step": 147820
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.808754878702073e-05,
+      "loss": 1.0905,
+      "step": 147830
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.808640085712099e-05,
+      "loss": 1.2065,
+      "step": 147840
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.8085252927221244e-05,
+      "loss": 1.2178,
+      "step": 147850
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.808410499732149e-05,
+      "loss": 1.1913,
+      "step": 147860
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.8082957067421744e-05,
+      "loss": 1.1832,
+      "step": 147870
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.8081809137522e-05,
+      "loss": 1.3154,
+      "step": 147880
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.808066120762225e-05,
+      "loss": 1.26,
+      "step": 147890
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.80795132777225e-05,
+      "loss": 1.1252,
+      "step": 147900
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.807836534782276e-05,
+      "loss": 1.3062,
+      "step": 147910
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.807721741792301e-05,
+      "loss": 1.1559,
+      "step": 147920
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.8076069488023264e-05,
+      "loss": 1.1291,
+      "step": 147930
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.807492155812351e-05,
+      "loss": 1.1064,
+      "step": 147940
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.807377362822376e-05,
+      "loss": 1.2213,
+      "step": 147950
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.807262569832402e-05,
+      "loss": 1.1695,
+      "step": 147960
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.8071477768424277e-05,
+      "loss": 1.1846,
+      "step": 147970
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.807032983852452e-05,
+      "loss": 1.2853,
+      "step": 147980
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.8069181908624776e-05,
+      "loss": 1.1561,
+      "step": 147990
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.806803397872503e-05,
+      "loss": 1.0991,
+      "step": 148000
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.8066886048825276e-05,
+      "loss": 1.2922,
+      "step": 148010
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.806573811892553e-05,
+      "loss": 1.2223,
+      "step": 148020
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.806459018902579e-05,
+      "loss": 1.0591,
+      "step": 148030
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.806344225912604e-05,
+      "loss": 1.2643,
+      "step": 148040
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.806229432922629e-05,
+      "loss": 1.2964,
+      "step": 148050
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.806114639932654e-05,
+      "loss": 1.1982,
+      "step": 148060
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.8059998469426795e-05,
+      "loss": 1.2438,
+      "step": 148070
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.805885053952705e-05,
+      "loss": 1.252,
+      "step": 148080
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.8057702609627295e-05,
+      "loss": 1.1285,
+      "step": 148090
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.805655467972755e-05,
+      "loss": 1.1542,
+      "step": 148100
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.805540674982781e-05,
+      "loss": 1.2356,
+      "step": 148110
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.805425881992805e-05,
+      "loss": 1.1573,
+      "step": 148120
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.805311089002831e-05,
+      "loss": 1.2283,
+      "step": 148130
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.8051962960128565e-05,
+      "loss": 1.1833,
+      "step": 148140
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.8050815030228815e-05,
+      "loss": 1.0767,
+      "step": 148150
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.8049667100329064e-05,
+      "loss": 1.3059,
+      "step": 148160
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.804851917042932e-05,
+      "loss": 1.0597,
+      "step": 148170
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.804737124052957e-05,
+      "loss": 1.2132,
+      "step": 148180
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.804622331062983e-05,
+      "loss": 1.2372,
+      "step": 148190
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.804507538073008e-05,
+      "loss": 1.32,
+      "step": 148200
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.804392745083033e-05,
+      "loss": 1.1867,
+      "step": 148210
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.8042779520930584e-05,
+      "loss": 1.3125,
+      "step": 148220
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.804163159103084e-05,
+      "loss": 1.2972,
+      "step": 148230
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.8040483661131084e-05,
+      "loss": 1.2031,
+      "step": 148240
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.803933573123134e-05,
+      "loss": 1.1479,
+      "step": 148250
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.80381878013316e-05,
+      "loss": 1.172,
+      "step": 148260
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.803703987143184e-05,
+      "loss": 1.2212,
+      "step": 148270
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.8035891941532097e-05,
+      "loss": 1.1453,
+      "step": 148280
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.803474401163235e-05,
+      "loss": 1.1141,
+      "step": 148290
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.80335960817326e-05,
+      "loss": 1.1782,
+      "step": 148300
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.803244815183285e-05,
+      "loss": 1.1596,
+      "step": 148310
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.803130022193311e-05,
+      "loss": 1.1863,
+      "step": 148320
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.803015229203336e-05,
+      "loss": 0.9671,
+      "step": 148330
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.8029004362133616e-05,
+      "loss": 1.1435,
+      "step": 148340
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.8027856432233866e-05,
+      "loss": 1.1383,
+      "step": 148350
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.8026708502334116e-05,
+      "loss": 1.3606,
+      "step": 148360
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.802556057243437e-05,
+      "loss": 1.3057,
+      "step": 148370
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.802441264253463e-05,
+      "loss": 1.2038,
+      "step": 148380
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.802326471263487e-05,
+      "loss": 1.177,
+      "step": 148390
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.802211678273513e-05,
+      "loss": 1.18,
+      "step": 148400
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.8020968852835385e-05,
+      "loss": 1.2748,
+      "step": 148410
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.801982092293563e-05,
+      "loss": 1.1971,
+      "step": 148420
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.8018672993035885e-05,
+      "loss": 1.0827,
+      "step": 148430
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.801752506313614e-05,
+      "loss": 1.1715,
+      "step": 148440
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.801637713323639e-05,
+      "loss": 1.2237,
+      "step": 148450
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.801522920333664e-05,
+      "loss": 1.2257,
+      "step": 148460
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.80140812734369e-05,
+      "loss": 1.1037,
+      "step": 148470
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.801293334353715e-05,
+      "loss": 1.3396,
+      "step": 148480
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.8011785413637404e-05,
+      "loss": 1.1518,
+      "step": 148490
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.8010637483737654e-05,
+      "loss": 1.2347,
+      "step": 148500
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.8009489553837904e-05,
+      "loss": 1.3092,
+      "step": 148510
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.800834162393816e-05,
+      "loss": 1.1061,
+      "step": 148520
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.800719369403842e-05,
+      "loss": 1.2776,
+      "step": 148530
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.800604576413866e-05,
+      "loss": 1.2572,
+      "step": 148540
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.800489783423892e-05,
+      "loss": 1.2407,
+      "step": 148550
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.8003749904339174e-05,
+      "loss": 1.1842,
+      "step": 148560
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.800260197443942e-05,
+      "loss": 1.2748,
+      "step": 148570
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.8001454044539673e-05,
+      "loss": 1.2474,
+      "step": 148580
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.800030611463993e-05,
+      "loss": 1.2768,
+      "step": 148590
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.799915818474018e-05,
+      "loss": 1.2114,
+      "step": 148600
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.799801025484043e-05,
+      "loss": 1.3279,
+      "step": 148610
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.7996862324940686e-05,
+      "loss": 1.2524,
+      "step": 148620
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.7995714395040936e-05,
+      "loss": 1.2001,
+      "step": 148630
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.799456646514119e-05,
+      "loss": 1.2783,
+      "step": 148640
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.799341853524144e-05,
+      "loss": 1.2852,
+      "step": 148650
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.799227060534169e-05,
+      "loss": 1.3486,
+      "step": 148660
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.799112267544195e-05,
+      "loss": 1.204,
+      "step": 148670
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.7989974745542206e-05,
+      "loss": 1.2181,
+      "step": 148680
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.798882681564245e-05,
+      "loss": 1.2465,
+      "step": 148690
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.7987678885742706e-05,
+      "loss": 1.3114,
+      "step": 148700
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.798653095584296e-05,
+      "loss": 1.3346,
+      "step": 148710
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.7985383025943205e-05,
+      "loss": 1.1389,
+      "step": 148720
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.798423509604346e-05,
+      "loss": 1.2178,
+      "step": 148730
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.798308716614372e-05,
+      "loss": 1.1976,
+      "step": 148740
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.798193923624397e-05,
+      "loss": 1.1267,
+      "step": 148750
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.798079130634422e-05,
+      "loss": 1.3227,
+      "step": 148760
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.7979643376444475e-05,
+      "loss": 1.159,
+      "step": 148770
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.7978495446544725e-05,
+      "loss": 1.262,
+      "step": 148780
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.797734751664498e-05,
+      "loss": 1.2233,
+      "step": 148790
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.797619958674523e-05,
+      "loss": 1.2894,
+      "step": 148800
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.797505165684548e-05,
+      "loss": 1.1609,
+      "step": 148810
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.797390372694574e-05,
+      "loss": 1.2928,
+      "step": 148820
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.7972755797045994e-05,
+      "loss": 1.1556,
+      "step": 148830
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.797160786714624e-05,
+      "loss": 1.3698,
+      "step": 148840
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.7970459937246494e-05,
+      "loss": 1.2221,
+      "step": 148850
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.796931200734675e-05,
+      "loss": 1.2523,
+      "step": 148860
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.7968164077446994e-05,
+      "loss": 1.2801,
+      "step": 148870
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.796701614754725e-05,
+      "loss": 1.1927,
+      "step": 148880
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.796586821764751e-05,
+      "loss": 1.1977,
+      "step": 148890
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.796472028774776e-05,
+      "loss": 1.2462,
+      "step": 148900
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.796357235784801e-05,
+      "loss": 1.2322,
+      "step": 148910
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.796242442794826e-05,
+      "loss": 1.3311,
+      "step": 148920
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.796127649804851e-05,
+      "loss": 1.1895,
+      "step": 148930
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.796012856814877e-05,
+      "loss": 1.126,
+      "step": 148940
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.795898063824902e-05,
+      "loss": 1.3024,
+      "step": 148950
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.795783270834927e-05,
+      "loss": 1.0892,
+      "step": 148960
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.7956684778449526e-05,
+      "loss": 1.2868,
+      "step": 148970
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.795553684854978e-05,
+      "loss": 1.2552,
+      "step": 148980
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.7954388918650026e-05,
+      "loss": 1.2379,
+      "step": 148990
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.795324098875028e-05,
+      "loss": 1.2334,
+      "step": 149000
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.795209305885054e-05,
+      "loss": 1.1863,
+      "step": 149010
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.795094512895078e-05,
+      "loss": 1.2015,
+      "step": 149020
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.794979719905104e-05,
+      "loss": 1.1134,
+      "step": 149030
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.7948649269151296e-05,
+      "loss": 1.1607,
+      "step": 149040
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.7947501339251545e-05,
+      "loss": 1.1373,
+      "step": 149050
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.7946353409351795e-05,
+      "loss": 1.1868,
+      "step": 149060
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.794520547945205e-05,
+      "loss": 1.2627,
+      "step": 149070
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.79440575495523e-05,
+      "loss": 1.1839,
+      "step": 149080
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.794290961965256e-05,
+      "loss": 1.3532,
+      "step": 149090
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.794176168975281e-05,
+      "loss": 1.0912,
+      "step": 149100
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.794061375985306e-05,
+      "loss": 1.2645,
+      "step": 149110
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.7939465829953315e-05,
+      "loss": 1.2639,
+      "step": 149120
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.793831790005357e-05,
+      "loss": 1.0907,
+      "step": 149130
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.7937169970153814e-05,
+      "loss": 1.1554,
+      "step": 149140
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.793602204025407e-05,
+      "loss": 1.3028,
+      "step": 149150
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.793487411035433e-05,
+      "loss": 1.1408,
+      "step": 149160
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.793372618045457e-05,
+      "loss": 1.1835,
+      "step": 149170
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.793257825055483e-05,
+      "loss": 1.29,
+      "step": 149180
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.7931430320655084e-05,
+      "loss": 1.3122,
+      "step": 149190
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.7930282390755334e-05,
+      "loss": 1.3495,
+      "step": 149200
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.7929134460855584e-05,
+      "loss": 1.1911,
+      "step": 149210
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.792798653095584e-05,
+      "loss": 1.1948,
+      "step": 149220
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.792683860105609e-05,
+      "loss": 1.3507,
+      "step": 149230
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.792569067115635e-05,
+      "loss": 1.1289,
+      "step": 149240
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.79245427412566e-05,
+      "loss": 1.4318,
+      "step": 149250
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.7923394811356847e-05,
+      "loss": 1.3914,
+      "step": 149260
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.79222468814571e-05,
+      "loss": 1.3692,
+      "step": 149270
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.792109895155736e-05,
+      "loss": 1.2855,
+      "step": 149280
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.79199510216576e-05,
+      "loss": 1.0915,
+      "step": 149290
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.791880309175786e-05,
+      "loss": 1.2233,
+      "step": 149300
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.7917655161858116e-05,
+      "loss": 1.3021,
+      "step": 149310
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.791650723195836e-05,
+      "loss": 1.1731,
+      "step": 149320
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.7915359302058616e-05,
+      "loss": 1.0874,
+      "step": 149330
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.791421137215887e-05,
+      "loss": 1.1875,
+      "step": 149340
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.791306344225912e-05,
+      "loss": 1.2129,
+      "step": 149350
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.791191551235937e-05,
+      "loss": 1.2172,
+      "step": 149360
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.791076758245963e-05,
+      "loss": 1.3421,
+      "step": 149370
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.790961965255988e-05,
+      "loss": 1.0947,
+      "step": 149380
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.7908471722660135e-05,
+      "loss": 1.2223,
+      "step": 149390
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.7907323792760385e-05,
+      "loss": 1.0609,
+      "step": 149400
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.7906175862860635e-05,
+      "loss": 1.1906,
+      "step": 149410
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.790502793296089e-05,
+      "loss": 1.1894,
+      "step": 149420
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.7903880003061135e-05,
+      "loss": 1.2926,
+      "step": 149430
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.790273207316139e-05,
+      "loss": 1.1491,
+      "step": 149440
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.790158414326165e-05,
+      "loss": 1.2672,
+      "step": 149450
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.79004362133619e-05,
+      "loss": 1.1514,
+      "step": 149460
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.789928828346215e-05,
+      "loss": 1.1466,
+      "step": 149470
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.7898140353562404e-05,
+      "loss": 1.2138,
+      "step": 149480
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.7896992423662654e-05,
+      "loss": 1.2354,
+      "step": 149490
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.789584449376291e-05,
+      "loss": 1.1269,
+      "step": 149500
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.789469656386316e-05,
+      "loss": 1.1044,
+      "step": 149510
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.789354863396341e-05,
+      "loss": 1.303,
+      "step": 149520
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.789240070406367e-05,
+      "loss": 1.1935,
+      "step": 149530
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.7891252774163924e-05,
+      "loss": 1.1824,
+      "step": 149540
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.789010484426417e-05,
+      "loss": 1.3891,
+      "step": 149550
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.7888956914364423e-05,
+      "loss": 1.2738,
+      "step": 149560
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.788780898446468e-05,
+      "loss": 1.1027,
+      "step": 149570
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.788666105456492e-05,
+      "loss": 1.3411,
+      "step": 149580
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.788551312466518e-05,
+      "loss": 1.151,
+      "step": 149590
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.7884365194765436e-05,
+      "loss": 1.1969,
+      "step": 149600
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.7883217264865686e-05,
+      "loss": 1.2416,
+      "step": 149610
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.7882069334965936e-05,
+      "loss": 1.2189,
+      "step": 149620
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.788092140506619e-05,
+      "loss": 1.2537,
+      "step": 149630
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.787977347516644e-05,
+      "loss": 1.2313,
+      "step": 149640
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.78786255452667e-05,
+      "loss": 1.2085,
+      "step": 149650
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.787747761536695e-05,
+      "loss": 1.0781,
+      "step": 149660
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.78763296854672e-05,
+      "loss": 1.1629,
+      "step": 149670
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.7875181755567456e-05,
+      "loss": 1.137,
+      "step": 149680
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.787403382566771e-05,
+      "loss": 1.214,
+      "step": 149690
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.7872885895767955e-05,
+      "loss": 1.1633,
+      "step": 149700
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.787173796586821e-05,
+      "loss": 1.2953,
+      "step": 149710
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.787059003596847e-05,
+      "loss": 1.2919,
+      "step": 149720
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.786944210606871e-05,
+      "loss": 1.3222,
+      "step": 149730
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.786829417616897e-05,
+      "loss": 1.4117,
+      "step": 149740
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.7867146246269225e-05,
+      "loss": 1.2514,
+      "step": 149750
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.7865998316369475e-05,
+      "loss": 1.191,
+      "step": 149760
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.7864850386469725e-05,
+      "loss": 1.1631,
+      "step": 149770
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.786370245656998e-05,
+      "loss": 1.3282,
+      "step": 149780
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.786255452667023e-05,
+      "loss": 1.0986,
+      "step": 149790
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.786140659677049e-05,
+      "loss": 1.1169,
+      "step": 149800
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.786025866687074e-05,
+      "loss": 1.123,
+      "step": 149810
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.785911073697099e-05,
+      "loss": 1.1812,
+      "step": 149820
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.7857962807071244e-05,
+      "loss": 1.313,
+      "step": 149830
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.78568148771715e-05,
+      "loss": 1.3592,
+      "step": 149840
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.7855666947271744e-05,
+      "loss": 1.2033,
+      "step": 149850
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.7854519017372e-05,
+      "loss": 1.2479,
+      "step": 149860
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.785337108747226e-05,
+      "loss": 1.1718,
+      "step": 149870
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.78522231575725e-05,
+      "loss": 1.2622,
+      "step": 149880
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.785107522767276e-05,
+      "loss": 1.2248,
+      "step": 149890
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.784992729777301e-05,
+      "loss": 1.2103,
+      "step": 149900
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.784877936787326e-05,
+      "loss": 1.1902,
+      "step": 149910
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.784763143797351e-05,
+      "loss": 1.3325,
+      "step": 149920
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.784648350807377e-05,
+      "loss": 1.2778,
+      "step": 149930
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.784533557817402e-05,
+      "loss": 1.2728,
+      "step": 149940
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.7844187648274276e-05,
+      "loss": 1.1313,
+      "step": 149950
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.7843039718374526e-05,
+      "loss": 1.1711,
+      "step": 149960
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.7841891788474776e-05,
+      "loss": 1.2195,
+      "step": 149970
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.784074385857503e-05,
+      "loss": 1.2511,
+      "step": 149980
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.783959592867529e-05,
+      "loss": 1.1647,
+      "step": 149990
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.783844799877553e-05,
+      "loss": 1.2313,
+      "step": 150000
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.783730006887579e-05,
+      "loss": 1.2708,
+      "step": 150010
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.7836152138976046e-05,
+      "loss": 1.171,
+      "step": 150020
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.783500420907629e-05,
+      "loss": 1.2146,
+      "step": 150030
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.7833856279176545e-05,
+      "loss": 1.0629,
+      "step": 150040
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.78327083492768e-05,
+      "loss": 1.3088,
+      "step": 150050
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.783156041937705e-05,
+      "loss": 1.2128,
+      "step": 150060
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.78304124894773e-05,
+      "loss": 1.2351,
+      "step": 150070
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.782926455957756e-05,
+      "loss": 1.2738,
+      "step": 150080
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.782811662967781e-05,
+      "loss": 1.1451,
+      "step": 150090
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.7826968699778065e-05,
+      "loss": 1.3306,
+      "step": 150100
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.7825820769878315e-05,
+      "loss": 1.2169,
+      "step": 150110
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.7824672839978564e-05,
+      "loss": 1.0849,
+      "step": 150120
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.782352491007882e-05,
+      "loss": 1.2691,
+      "step": 150130
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.782237698017908e-05,
+      "loss": 1.1717,
+      "step": 150140
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.782122905027932e-05,
+      "loss": 1.2417,
+      "step": 150150
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.782008112037958e-05,
+      "loss": 1.4322,
+      "step": 150160
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.7818933190479834e-05,
+      "loss": 1.2828,
+      "step": 150170
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.781778526058008e-05,
+      "loss": 1.1261,
+      "step": 150180
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.7816637330680334e-05,
+      "loss": 1.1887,
+      "step": 150190
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.781548940078059e-05,
+      "loss": 1.1961,
+      "step": 150200
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.781434147088084e-05,
+      "loss": 0.9976,
+      "step": 150210
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.781319354098109e-05,
+      "loss": 1.3033,
+      "step": 150220
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.781204561108135e-05,
+      "loss": 1.146,
+      "step": 150230
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.7810897681181597e-05,
+      "loss": 1.2377,
+      "step": 150240
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.780974975128185e-05,
+      "loss": 1.251,
+      "step": 150250
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.78086018213821e-05,
+      "loss": 1.2074,
+      "step": 150260
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.780745389148235e-05,
+      "loss": 1.252,
+      "step": 150270
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.780630596158261e-05,
+      "loss": 1.3221,
+      "step": 150280
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.7805158031682866e-05,
+      "loss": 1.4239,
+      "step": 150290
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.780401010178311e-05,
+      "loss": 1.1361,
+      "step": 150300
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.7802862171883366e-05,
+      "loss": 1.145,
+      "step": 150310
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.780171424198362e-05,
+      "loss": 1.1274,
+      "step": 150320
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.7800566312083866e-05,
+      "loss": 1.319,
+      "step": 150330
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.779941838218412e-05,
+      "loss": 1.117,
+      "step": 150340
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.779827045228438e-05,
+      "loss": 1.1504,
+      "step": 150350
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.779712252238463e-05,
+      "loss": 1.18,
+      "step": 150360
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.779597459248488e-05,
+      "loss": 1.1899,
+      "step": 150370
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.7794826662585135e-05,
+      "loss": 1.2873,
+      "step": 150380
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.7793678732685385e-05,
+      "loss": 1.1025,
+      "step": 150390
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.779253080278564e-05,
+      "loss": 1.2931,
+      "step": 150400
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.779138287288589e-05,
+      "loss": 1.3346,
+      "step": 150410
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.779023494298614e-05,
+      "loss": 1.1945,
+      "step": 150420
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.77890870130864e-05,
+      "loss": 1.1475,
+      "step": 150430
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.7787939083186655e-05,
+      "loss": 1.1847,
+      "step": 150440
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.77867911532869e-05,
+      "loss": 1.304,
+      "step": 150450
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.7785643223387154e-05,
+      "loss": 1.2586,
+      "step": 150460
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.778449529348741e-05,
+      "loss": 1.1288,
+      "step": 150470
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.7783347363587654e-05,
+      "loss": 1.2566,
+      "step": 150480
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.778219943368791e-05,
+      "loss": 1.0508,
+      "step": 150490
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.778105150378817e-05,
+      "loss": 1.2758,
+      "step": 150500
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.777990357388842e-05,
+      "loss": 1.2181,
+      "step": 150510
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.777875564398867e-05,
+      "loss": 1.1582,
+      "step": 150520
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.7777607714088924e-05,
+      "loss": 1.1868,
+      "step": 150530
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.7776459784189173e-05,
+      "loss": 1.1764,
+      "step": 150540
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.777531185428943e-05,
+      "loss": 1.2205,
+      "step": 150550
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.777416392438968e-05,
+      "loss": 1.2227,
+      "step": 150560
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.777301599448993e-05,
+      "loss": 1.2307,
+      "step": 150570
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.7771868064590186e-05,
+      "loss": 1.1466,
+      "step": 150580
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.777072013469044e-05,
+      "loss": 1.2544,
+      "step": 150590
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.7769572204790686e-05,
+      "loss": 1.2562,
+      "step": 150600
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.776842427489094e-05,
+      "loss": 1.345,
+      "step": 150610
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.77672763449912e-05,
+      "loss": 1.3211,
+      "step": 150620
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.776612841509144e-05,
+      "loss": 1.2385,
+      "step": 150630
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.77649804851917e-05,
+      "loss": 1.2199,
+      "step": 150640
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.7763832555291956e-05,
+      "loss": 1.3354,
+      "step": 150650
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.7762684625392206e-05,
+      "loss": 1.2055,
+      "step": 150660
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.7761536695492455e-05,
+      "loss": 1.2404,
+      "step": 150670
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.776038876559271e-05,
+      "loss": 1.3094,
+      "step": 150680
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.775924083569296e-05,
+      "loss": 1.1338,
+      "step": 150690
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.775809290579322e-05,
+      "loss": 1.2942,
+      "step": 150700
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.775694497589347e-05,
+      "loss": 1.3918,
+      "step": 150710
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.775579704599372e-05,
+      "loss": 1.2259,
+      "step": 150720
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.7754649116093975e-05,
+      "loss": 1.1338,
+      "step": 150730
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.775350118619423e-05,
+      "loss": 1.212,
+      "step": 150740
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.7752353256294475e-05,
+      "loss": 1.1448,
+      "step": 150750
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.775120532639473e-05,
+      "loss": 1.2357,
+      "step": 150760
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.775005739649499e-05,
+      "loss": 1.306,
+      "step": 150770
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.774890946659523e-05,
+      "loss": 1.2083,
+      "step": 150780
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.774776153669549e-05,
+      "loss": 1.272,
+      "step": 150790
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.774661360679574e-05,
+      "loss": 1.2637,
+      "step": 150800
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.7745465676895994e-05,
+      "loss": 1.2971,
+      "step": 150810
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.7744317746996244e-05,
+      "loss": 1.2176,
+      "step": 150820
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.7743169817096494e-05,
+      "loss": 1.2136,
+      "step": 150830
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.774202188719675e-05,
+      "loss": 1.1252,
+      "step": 150840
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.774087395729701e-05,
+      "loss": 1.1652,
+      "step": 150850
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.773972602739725e-05,
+      "loss": 1.1473,
+      "step": 150860
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.773857809749751e-05,
+      "loss": 1.1025,
+      "step": 150870
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.773743016759776e-05,
+      "loss": 1.1801,
+      "step": 150880
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.7736282237698006e-05,
+      "loss": 1.3379,
+      "step": 150890
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.773513430779826e-05,
+      "loss": 1.4789,
+      "step": 150900
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.773398637789852e-05,
+      "loss": 1.2094,
+      "step": 150910
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.773283844799877e-05,
+      "loss": 1.1147,
+      "step": 150920
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.773169051809902e-05,
+      "loss": 1.2229,
+      "step": 150930
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.7730542588199276e-05,
+      "loss": 1.1049,
+      "step": 150940
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.7729394658299526e-05,
+      "loss": 1.1922,
+      "step": 150950
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.772824672839978e-05,
+      "loss": 1.193,
+      "step": 150960
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.772709879850003e-05,
+      "loss": 1.2726,
+      "step": 150970
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.772595086860028e-05,
+      "loss": 1.2548,
+      "step": 150980
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.772480293870054e-05,
+      "loss": 1.3231,
+      "step": 150990
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.7723655008800796e-05,
+      "loss": 1.2718,
+      "step": 151000
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.772250707890104e-05,
+      "loss": 1.1817,
+      "step": 151010
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.7721359149001295e-05,
+      "loss": 1.2982,
+      "step": 151020
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.772021121910155e-05,
+      "loss": 1.1693,
+      "step": 151030
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.7719063289201795e-05,
+      "loss": 1.2242,
+      "step": 151040
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.771791535930205e-05,
+      "loss": 1.204,
+      "step": 151050
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.771676742940231e-05,
+      "loss": 1.1668,
+      "step": 151060
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.771561949950256e-05,
+      "loss": 1.1812,
+      "step": 151070
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.771447156960281e-05,
+      "loss": 1.3116,
+      "step": 151080
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.7713323639703065e-05,
+      "loss": 1.1382,
+      "step": 151090
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.7712175709803314e-05,
+      "loss": 1.3471,
+      "step": 151100
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.771102777990357e-05,
+      "loss": 1.2456,
+      "step": 151110
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.770987985000382e-05,
+      "loss": 1.2767,
+      "step": 151120
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.770873192010407e-05,
+      "loss": 1.2383,
+      "step": 151130
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.770758399020433e-05,
+      "loss": 1.2713,
+      "step": 151140
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.7706436060304584e-05,
+      "loss": 1.2424,
+      "step": 151150
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.770528813040483e-05,
+      "loss": 1.2414,
+      "step": 151160
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.7704140200505084e-05,
+      "loss": 1.2199,
+      "step": 151170
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.770299227060534e-05,
+      "loss": 1.233,
+      "step": 151180
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.7701844340705583e-05,
+      "loss": 1.249,
+      "step": 151190
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.770069641080584e-05,
+      "loss": 1.2716,
+      "step": 151200
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.76995484809061e-05,
+      "loss": 1.2618,
+      "step": 151210
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.7698400551006347e-05,
+      "loss": 1.1883,
+      "step": 151220
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.7697252621106596e-05,
+      "loss": 1.1114,
+      "step": 151230
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.769610469120685e-05,
+      "loss": 1.2623,
+      "step": 151240
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.76949567613071e-05,
+      "loss": 1.2569,
+      "step": 151250
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.769380883140736e-05,
+      "loss": 1.216,
+      "step": 151260
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.769266090150761e-05,
+      "loss": 1.0508,
+      "step": 151270
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.769151297160786e-05,
+      "loss": 1.3272,
+      "step": 151280
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.7690365041708116e-05,
+      "loss": 1.2303,
+      "step": 151290
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.768921711180837e-05,
+      "loss": 1.2195,
+      "step": 151300
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.7688069181908616e-05,
+      "loss": 1.2094,
+      "step": 151310
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.768692125200887e-05,
+      "loss": 1.2469,
+      "step": 151320
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.768577332210913e-05,
+      "loss": 1.2758,
+      "step": 151330
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.768462539220937e-05,
+      "loss": 1.1674,
+      "step": 151340
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.768347746230963e-05,
+      "loss": 1.2847,
+      "step": 151350
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.7682329532409885e-05,
+      "loss": 1.2103,
+      "step": 151360
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.7681181602510135e-05,
+      "loss": 1.2802,
+      "step": 151370
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.7680033672610385e-05,
+      "loss": 1.2596,
+      "step": 151380
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.767888574271064e-05,
+      "loss": 1.1246,
+      "step": 151390
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.767773781281089e-05,
+      "loss": 1.101,
+      "step": 151400
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.767658988291115e-05,
+      "loss": 1.2991,
+      "step": 151410
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.76754419530114e-05,
+      "loss": 1.0798,
+      "step": 151420
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.767429402311165e-05,
+      "loss": 1.2236,
+      "step": 151430
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.7673146093211904e-05,
+      "loss": 1.1787,
+      "step": 151440
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.767199816331216e-05,
+      "loss": 1.2297,
+      "step": 151450
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.7670850233412404e-05,
+      "loss": 1.2743,
+      "step": 151460
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.766970230351266e-05,
+      "loss": 1.2167,
+      "step": 151470
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.766855437361292e-05,
+      "loss": 1.1598,
+      "step": 151480
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.766740644371316e-05,
+      "loss": 1.3078,
+      "step": 151490
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.766625851381342e-05,
+      "loss": 1.2848,
+      "step": 151500
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.7665110583913674e-05,
+      "loss": 1.1504,
+      "step": 151510
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.7663962654013923e-05,
+      "loss": 1.1993,
+      "step": 151520
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.766281472411417e-05,
+      "loss": 1.2175,
+      "step": 151530
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.766166679421443e-05,
+      "loss": 1.1298,
+      "step": 151540
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.766051886431468e-05,
+      "loss": 1.3546,
+      "step": 151550
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.7659370934414936e-05,
+      "loss": 1.1339,
+      "step": 151560
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.7658223004515186e-05,
+      "loss": 1.1426,
+      "step": 151570
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.7657075074615436e-05,
+      "loss": 1.2081,
+      "step": 151580
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.765592714471569e-05,
+      "loss": 1.269,
+      "step": 151590
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.765477921481595e-05,
+      "loss": 1.1794,
+      "step": 151600
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.765363128491619e-05,
+      "loss": 1.2106,
+      "step": 151610
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.765248335501645e-05,
+      "loss": 1.198,
+      "step": 151620
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.7651335425116706e-05,
+      "loss": 1.1992,
+      "step": 151630
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.765018749521695e-05,
+      "loss": 1.2492,
+      "step": 151640
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.7649039565317205e-05,
+      "loss": 1.3592,
+      "step": 151650
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.764789163541746e-05,
+      "loss": 1.3471,
+      "step": 151660
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.764674370551771e-05,
+      "loss": 1.1674,
+      "step": 151670
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.764559577561796e-05,
+      "loss": 1.1844,
+      "step": 151680
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.764444784571822e-05,
+      "loss": 1.1857,
+      "step": 151690
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.764329991581847e-05,
+      "loss": 1.3173,
+      "step": 151700
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.7642151985918725e-05,
+      "loss": 1.2452,
+      "step": 151710
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.7641004056018975e-05,
+      "loss": 1.4783,
+      "step": 151720
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.7639856126119225e-05,
+      "loss": 1.3385,
+      "step": 151730
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.763870819621948e-05,
+      "loss": 1.2867,
+      "step": 151740
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.763756026631974e-05,
+      "loss": 1.1983,
+      "step": 151750
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.763641233641998e-05,
+      "loss": 1.266,
+      "step": 151760
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.763526440652024e-05,
+      "loss": 1.333,
+      "step": 151770
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.7634116476620494e-05,
+      "loss": 1.2715,
+      "step": 151780
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.763296854672074e-05,
+      "loss": 1.2287,
+      "step": 151790
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.7631820616820994e-05,
+      "loss": 1.1677,
+      "step": 151800
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.763067268692125e-05,
+      "loss": 1.1752,
+      "step": 151810
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.76295247570215e-05,
+      "loss": 1.3649,
+      "step": 151820
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.762837682712175e-05,
+      "loss": 1.1973,
+      "step": 151830
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.762722889722201e-05,
+      "loss": 1.3813,
+      "step": 151840
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.762608096732226e-05,
+      "loss": 1.1816,
+      "step": 151850
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.762493303742251e-05,
+      "loss": 1.1861,
+      "step": 151860
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.762378510752276e-05,
+      "loss": 1.1516,
+      "step": 151870
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.762263717762301e-05,
+      "loss": 1.2645,
+      "step": 151880
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.762148924772327e-05,
+      "loss": 1.2256,
+      "step": 151890
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.7620341317823526e-05,
+      "loss": 1.2588,
+      "step": 151900
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.761919338792377e-05,
+      "loss": 1.3783,
+      "step": 151910
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.7618045458024026e-05,
+      "loss": 1.1377,
+      "step": 151920
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.761689752812428e-05,
+      "loss": 1.3236,
+      "step": 151930
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.7615749598224526e-05,
+      "loss": 1.2304,
+      "step": 151940
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.761460166832478e-05,
+      "loss": 1.2585,
+      "step": 151950
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.761345373842504e-05,
+      "loss": 1.241,
+      "step": 151960
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.761230580852529e-05,
+      "loss": 1.1575,
+      "step": 151970
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.761115787862554e-05,
+      "loss": 1.2538,
+      "step": 151980
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.7610009948725795e-05,
+      "loss": 1.154,
+      "step": 151990
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.7608862018826045e-05,
+      "loss": 1.1161,
+      "step": 152000
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.76077140889263e-05,
+      "loss": 1.3408,
+      "step": 152010
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.760656615902655e-05,
+      "loss": 1.2641,
+      "step": 152020
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.76054182291268e-05,
+      "loss": 1.3551,
+      "step": 152030
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.760427029922706e-05,
+      "loss": 1.253,
+      "step": 152040
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.7603122369327315e-05,
+      "loss": 1.2893,
+      "step": 152050
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.760197443942756e-05,
+      "loss": 1.0924,
+      "step": 152060
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.7600826509527815e-05,
+      "loss": 1.1547,
+      "step": 152070
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.759967857962807e-05,
+      "loss": 1.1974,
+      "step": 152080
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.7598530649728314e-05,
+      "loss": 1.2018,
+      "step": 152090
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.759738271982857e-05,
+      "loss": 1.0664,
+      "step": 152100
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.759623478992883e-05,
+      "loss": 1.3379,
+      "step": 152110
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.759508686002908e-05,
+      "loss": 1.2797,
+      "step": 152120
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.759393893012933e-05,
+      "loss": 1.184,
+      "step": 152130
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.7592791000229584e-05,
+      "loss": 1.1691,
+      "step": 152140
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.7591643070329834e-05,
+      "loss": 1.2046,
+      "step": 152150
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.759049514043009e-05,
+      "loss": 1.2976,
+      "step": 152160
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.758934721053034e-05,
+      "loss": 1.2719,
+      "step": 152170
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.758819928063059e-05,
+      "loss": 1.1843,
+      "step": 152180
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.758705135073085e-05,
+      "loss": 1.0073,
+      "step": 152190
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.758590342083109e-05,
+      "loss": 1.1244,
+      "step": 152200
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.7584755490931346e-05,
+      "loss": 1.1997,
+      "step": 152210
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.75836075610316e-05,
+      "loss": 1.224,
+      "step": 152220
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.758245963113185e-05,
+      "loss": 1.1637,
+      "step": 152230
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.75813117012321e-05,
+      "loss": 1.3188,
+      "step": 152240
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.758016377133236e-05,
+      "loss": 1.1178,
+      "step": 152250
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.757901584143261e-05,
+      "loss": 1.1616,
+      "step": 152260
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.7577867911532866e-05,
+      "loss": 1.2081,
+      "step": 152270
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.7576719981633116e-05,
+      "loss": 1.1968,
+      "step": 152280
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.7575572051733366e-05,
+      "loss": 1.3167,
+      "step": 152290
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.757442412183362e-05,
+      "loss": 1.236,
+      "step": 152300
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.757327619193388e-05,
+      "loss": 1.2544,
+      "step": 152310
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.757212826203412e-05,
+      "loss": 1.2544,
+      "step": 152320
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.757098033213438e-05,
+      "loss": 1.2676,
+      "step": 152330
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.7569832402234635e-05,
+      "loss": 1.2778,
+      "step": 152340
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.756868447233488e-05,
+      "loss": 1.1211,
+      "step": 152350
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.7567536542435135e-05,
+      "loss": 1.4489,
+      "step": 152360
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.756638861253539e-05,
+      "loss": 1.1626,
+      "step": 152370
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.756524068263564e-05,
+      "loss": 1.317,
+      "step": 152380
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.756409275273589e-05,
+      "loss": 1.1884,
+      "step": 152390
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.756294482283615e-05,
+      "loss": 1.2945,
+      "step": 152400
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.75617968929364e-05,
+      "loss": 1.2511,
+      "step": 152410
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.7560648963036654e-05,
+      "loss": 1.2407,
+      "step": 152420
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.7559501033136904e-05,
+      "loss": 1.1322,
+      "step": 152430
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.7558353103237154e-05,
+      "loss": 1.2209,
+      "step": 152440
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 5.755720517333741e-05,
+      "loss": 1.1676,
+      "step": 152450
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.755605724343767e-05,
+      "loss": 1.2333,
+      "step": 152460
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.755490931353791e-05,
+      "loss": 1.1933,
+      "step": 152470
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.755376138363817e-05,
+      "loss": 1.2847,
+      "step": 152480
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.7552613453738424e-05,
+      "loss": 1.17,
+      "step": 152490
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.755146552383867e-05,
+      "loss": 1.1392,
+      "step": 152500
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.755031759393892e-05,
+      "loss": 1.3324,
+      "step": 152510
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.754916966403918e-05,
+      "loss": 1.2984,
+      "step": 152520
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.754802173413943e-05,
+      "loss": 1.1913,
+      "step": 152530
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.754687380423968e-05,
+      "loss": 1.2846,
+      "step": 152540
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.7545725874339936e-05,
+      "loss": 1.1627,
+      "step": 152550
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.7544577944440186e-05,
+      "loss": 1.1833,
+      "step": 152560
+    },
+    {
+      "epoch": 7.0,
+      "eval_cer": 0.6691760555699544,
+      "eval_em": 0.007357212003872217,
+      "eval_f1": 0.007357212003872217,
+      "eval_loss": 1.204297423362732,
+      "eval_runtime": 1710.6265,
+      "eval_samples_per_second": 6.039,
+      "eval_steps_per_second": 0.755,
+      "eval_wer": 0.9926427879961278,
+      "step": 152567
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.754343001454044e-05,
+      "loss": 1.225,
+      "step": 152570
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.754228208464069e-05,
+      "loss": 1.1719,
+      "step": 152580
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.754113415474094e-05,
+      "loss": 1.136,
+      "step": 152590
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.75399862248412e-05,
+      "loss": 1.0386,
+      "step": 152600
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.7538838294941456e-05,
+      "loss": 1.1958,
+      "step": 152610
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.75376903650417e-05,
+      "loss": 1.1272,
+      "step": 152620
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.7536542435141955e-05,
+      "loss": 1.0637,
+      "step": 152630
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.753539450524221e-05,
+      "loss": 1.1666,
+      "step": 152640
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.7534246575342455e-05,
+      "loss": 1.2472,
+      "step": 152650
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.753309864544271e-05,
+      "loss": 1.1689,
+      "step": 152660
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 5.753195071554297e-05,
+      "loss": 1.1068,
+      "step": 152670
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.753080278564322e-05,
+      "loss": 1.1142,
+      "step": 152680
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.752965485574347e-05,
+      "loss": 1.2151,
+      "step": 152690
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.7528506925843725e-05,
+      "loss": 1.1252,
+      "step": 152700
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.7527358995943975e-05,
+      "loss": 1.2088,
+      "step": 152710
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.752621106604423e-05,
+      "loss": 1.1664,
+      "step": 152720
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.752506313614448e-05,
+      "loss": 1.0511,
+      "step": 152730
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.752391520624473e-05,
+      "loss": 1.108,
+      "step": 152740
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.752276727634499e-05,
+      "loss": 1.1614,
+      "step": 152750
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.7521619346445244e-05,
+      "loss": 1.0686,
+      "step": 152760
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.752047141654549e-05,
+      "loss": 1.1755,
+      "step": 152770
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.7519323486645744e-05,
+      "loss": 1.1264,
+      "step": 152780
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.7518175556746e-05,
+      "loss": 1.1962,
+      "step": 152790
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.7517027626846244e-05,
+      "loss": 1.1171,
+      "step": 152800
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.75158796969465e-05,
+      "loss": 1.3427,
+      "step": 152810
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.751473176704676e-05,
+      "loss": 1.217,
+      "step": 152820
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.751358383714701e-05,
+      "loss": 1.1213,
+      "step": 152830
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.7512435907247257e-05,
+      "loss": 1.1054,
+      "step": 152840
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.751128797734751e-05,
+      "loss": 1.1487,
+      "step": 152850
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.751014004744776e-05,
+      "loss": 1.2838,
+      "step": 152860
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.750899211754802e-05,
+      "loss": 1.0958,
+      "step": 152870
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.750784418764827e-05,
+      "loss": 1.0084,
+      "step": 152880
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 5.750669625774852e-05,
+      "loss": 1.2237,
+      "step": 152890
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.7505548327848776e-05,
+      "loss": 1.2703,
+      "step": 152900
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.750440039794903e-05,
+      "loss": 1.1673,
+      "step": 152910
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.7503252468049276e-05,
+      "loss": 1.1878,
+      "step": 152920
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.750210453814953e-05,
+      "loss": 1.116,
+      "step": 152930
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.750095660824979e-05,
+      "loss": 1.044,
+      "step": 152940
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.749980867835003e-05,
+      "loss": 1.1158,
+      "step": 152950
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.749866074845029e-05,
+      "loss": 1.2061,
+      "step": 152960
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.7497512818550545e-05,
+      "loss": 1.2339,
+      "step": 152970
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.7496364888650795e-05,
+      "loss": 1.1697,
+      "step": 152980
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.7495216958751045e-05,
+      "loss": 1.2248,
+      "step": 152990
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.74940690288513e-05,
+      "loss": 1.2206,
+      "step": 153000
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.749292109895155e-05,
+      "loss": 1.1433,
+      "step": 153010
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.749177316905181e-05,
+      "loss": 1.0138,
+      "step": 153020
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.749062523915206e-05,
+      "loss": 1.1445,
+      "step": 153030
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.748947730925231e-05,
+      "loss": 1.1827,
+      "step": 153040
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.7488329379352565e-05,
+      "loss": 1.2193,
+      "step": 153050
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.748718144945282e-05,
+      "loss": 1.3281,
+      "step": 153060
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.7486033519553064e-05,
+      "loss": 1.1164,
+      "step": 153070
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.748488558965332e-05,
+      "loss": 1.2815,
+      "step": 153080
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.748373765975358e-05,
+      "loss": 1.0955,
+      "step": 153090
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.748258972985382e-05,
+      "loss": 1.2118,
+      "step": 153100
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 5.748144179995408e-05,
+      "loss": 1.0924,
+      "step": 153110
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.7480293870054334e-05,
+      "loss": 1.0757,
+      "step": 153120
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.7479145940154584e-05,
+      "loss": 1.2297,
+      "step": 153130
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.7477998010254834e-05,
+      "loss": 1.1371,
+      "step": 153140
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.747685008035509e-05,
+      "loss": 1.0983,
+      "step": 153150
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.747570215045534e-05,
+      "loss": 1.1146,
+      "step": 153160
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.74745542205556e-05,
+      "loss": 1.2589,
+      "step": 153170
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.7473406290655846e-05,
+      "loss": 1.1645,
+      "step": 153180
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.7472258360756096e-05,
+      "loss": 1.0964,
+      "step": 153190
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.747111043085635e-05,
+      "loss": 1.0953,
+      "step": 153200
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.746996250095661e-05,
+      "loss": 1.2364,
+      "step": 153210
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.746881457105685e-05,
+      "loss": 1.0906,
+      "step": 153220
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.746766664115711e-05,
+      "loss": 1.1918,
+      "step": 153230
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.7466518711257366e-05,
+      "loss": 1.0323,
+      "step": 153240
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.746537078135761e-05,
+      "loss": 1.1773,
+      "step": 153250
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.7464222851457866e-05,
+      "loss": 1.0063,
+      "step": 153260
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.746307492155812e-05,
+      "loss": 1.283,
+      "step": 153270
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.746192699165837e-05,
+      "loss": 1.2543,
+      "step": 153280
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.746077906175862e-05,
+      "loss": 1.1064,
+      "step": 153290
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.745963113185888e-05,
+      "loss": 1.1977,
+      "step": 153300
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.745848320195913e-05,
+      "loss": 1.284,
+      "step": 153310
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 5.7457335272059385e-05,
+      "loss": 1.2186,
+      "step": 153320
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.7456187342159635e-05,
+      "loss": 1.132,
+      "step": 153330
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.7455039412259885e-05,
+      "loss": 1.1642,
+      "step": 153340
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.745389148236014e-05,
+      "loss": 1.0354,
+      "step": 153350
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.74527435524604e-05,
+      "loss": 1.0804,
+      "step": 153360
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.745159562256064e-05,
+      "loss": 1.2036,
+      "step": 153370
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.74504476926609e-05,
+      "loss": 1.1483,
+      "step": 153380
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.7449299762761154e-05,
+      "loss": 1.097,
+      "step": 153390
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.74481518328614e-05,
+      "loss": 1.0615,
+      "step": 153400
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.7447003902961654e-05,
+      "loss": 1.1688,
+      "step": 153410
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.744585597306191e-05,
+      "loss": 1.3854,
+      "step": 153420
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.744470804316216e-05,
+      "loss": 1.087,
+      "step": 153430
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.744356011326241e-05,
+      "loss": 1.0566,
+      "step": 153440
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.744241218336267e-05,
+      "loss": 1.2456,
+      "step": 153450
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.744126425346292e-05,
+      "loss": 1.1317,
+      "step": 153460
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.7440116323563174e-05,
+      "loss": 1.1098,
+      "step": 153470
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.7438968393663423e-05,
+      "loss": 0.9965,
+      "step": 153480
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.743782046376367e-05,
+      "loss": 1.339,
+      "step": 153490
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.743667253386393e-05,
+      "loss": 1.1785,
+      "step": 153500
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.7435524603964187e-05,
+      "loss": 1.2095,
+      "step": 153510
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.743437667406443e-05,
+      "loss": 1.2142,
+      "step": 153520
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.7433228744164686e-05,
+      "loss": 1.1886,
+      "step": 153530
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 5.743208081426494e-05,
+      "loss": 1.0697,
+      "step": 153540
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.7430932884365186e-05,
+      "loss": 1.0941,
+      "step": 153550
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.742978495446544e-05,
+      "loss": 1.1505,
+      "step": 153560
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.742863702456569e-05,
+      "loss": 1.1261,
+      "step": 153570
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.742748909466595e-05,
+      "loss": 1.2606,
+      "step": 153580
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.74263411647662e-05,
+      "loss": 1.1903,
+      "step": 153590
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.742519323486645e-05,
+      "loss": 1.0683,
+      "step": 153600
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.7424045304966705e-05,
+      "loss": 1.1958,
+      "step": 153610
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.742289737506696e-05,
+      "loss": 1.194,
+      "step": 153620
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.7421749445167205e-05,
+      "loss": 1.1453,
+      "step": 153630
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.742060151526746e-05,
+      "loss": 1.2401,
+      "step": 153640
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.741945358536772e-05,
+      "loss": 1.1226,
+      "step": 153650
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.741830565546796e-05,
+      "loss": 1.1434,
+      "step": 153660
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.741715772556822e-05,
+      "loss": 1.2507,
+      "step": 153670
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.7416009795668475e-05,
+      "loss": 1.0675,
+      "step": 153680
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.7414861865768725e-05,
+      "loss": 1.1669,
+      "step": 153690
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.7413713935868974e-05,
+      "loss": 1.1422,
+      "step": 153700
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.741256600596923e-05,
+      "loss": 1.0666,
+      "step": 153710
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.741141807606948e-05,
+      "loss": 1.1454,
+      "step": 153720
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.741027014616974e-05,
+      "loss": 1.2211,
+      "step": 153730
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.740912221626999e-05,
+      "loss": 1.2336,
+      "step": 153740
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.740797428637024e-05,
+      "loss": 1.1544,
+      "step": 153750
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 5.7406826356470494e-05,
+      "loss": 1.3027,
+      "step": 153760
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.740567842657075e-05,
+      "loss": 1.0123,
+      "step": 153770
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.7404530496670994e-05,
+      "loss": 1.3349,
+      "step": 153780
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.740338256677125e-05,
+      "loss": 1.1735,
+      "step": 153790
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.740223463687151e-05,
+      "loss": 1.1092,
+      "step": 153800
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.740108670697175e-05,
+      "loss": 1.0599,
+      "step": 153810
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.7399938777072007e-05,
+      "loss": 1.0873,
+      "step": 153820
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.739879084717226e-05,
+      "loss": 1.2468,
+      "step": 153830
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.739764291727251e-05,
+      "loss": 1.1008,
+      "step": 153840
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.739649498737276e-05,
+      "loss": 1.194,
+      "step": 153850
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.739534705747302e-05,
+      "loss": 1.2236,
+      "step": 153860
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.739419912757327e-05,
+      "loss": 1.2716,
+      "step": 153870
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.7393051197673526e-05,
+      "loss": 1.1638,
+      "step": 153880
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.7391903267773776e-05,
+      "loss": 1.072,
+      "step": 153890
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.7390755337874026e-05,
+      "loss": 1.1814,
+      "step": 153900
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.738960740797428e-05,
+      "loss": 1.0305,
+      "step": 153910
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.738845947807454e-05,
+      "loss": 1.3393,
+      "step": 153920
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.738731154817478e-05,
+      "loss": 1.2695,
+      "step": 153930
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.738616361827504e-05,
+      "loss": 1.1448,
+      "step": 153940
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.7385015688375295e-05,
+      "loss": 1.3267,
+      "step": 153950
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.738386775847554e-05,
+      "loss": 1.0967,
+      "step": 153960
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.7382719828575795e-05,
+      "loss": 1.0893,
+      "step": 153970
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 5.738157189867605e-05,
+      "loss": 1.1665,
+      "step": 153980
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.73804239687763e-05,
+      "loss": 1.1165,
+      "step": 153990
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.737927603887655e-05,
+      "loss": 1.1836,
+      "step": 154000
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.737812810897681e-05,
+      "loss": 1.1318,
+      "step": 154010
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.737698017907706e-05,
+      "loss": 1.1002,
+      "step": 154020
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.7375832249177315e-05,
+      "loss": 1.1416,
+      "step": 154030
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.7374684319277564e-05,
+      "loss": 1.0389,
+      "step": 154040
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.7373536389377814e-05,
+      "loss": 1.2377,
+      "step": 154050
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.737238845947807e-05,
+      "loss": 1.1476,
+      "step": 154060
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.737124052957833e-05,
+      "loss": 1.1186,
+      "step": 154070
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.737009259967857e-05,
+      "loss": 1.1797,
+      "step": 154080
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.736894466977883e-05,
+      "loss": 1.2198,
+      "step": 154090
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.7367796739879084e-05,
+      "loss": 1.2474,
+      "step": 154100
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.736664880997933e-05,
+      "loss": 1.1287,
+      "step": 154110
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.7365500880079584e-05,
+      "loss": 1.2187,
+      "step": 154120
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.736435295017984e-05,
+      "loss": 1.1531,
+      "step": 154130
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.736320502028009e-05,
+      "loss": 0.9894,
+      "step": 154140
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.736205709038034e-05,
+      "loss": 1.0942,
+      "step": 154150
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.7360909160480596e-05,
+      "loss": 1.1032,
+      "step": 154160
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.7359761230580846e-05,
+      "loss": 1.1853,
+      "step": 154170
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.73586133006811e-05,
+      "loss": 1.1694,
+      "step": 154180
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.735746537078135e-05,
+      "loss": 1.1521,
+      "step": 154190
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 5.73563174408816e-05,
+      "loss": 1.1309,
+      "step": 154200
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.735516951098186e-05,
+      "loss": 1.1427,
+      "step": 154210
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.7354021581082116e-05,
+      "loss": 1.1686,
+      "step": 154220
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.735287365118236e-05,
+      "loss": 1.0691,
+      "step": 154230
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.7351725721282616e-05,
+      "loss": 1.2346,
+      "step": 154240
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.735057779138287e-05,
+      "loss": 1.133,
+      "step": 154250
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.7349429861483115e-05,
+      "loss": 1.2223,
+      "step": 154260
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.734828193158337e-05,
+      "loss": 1.1405,
+      "step": 154270
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.734713400168363e-05,
+      "loss": 1.2054,
+      "step": 154280
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.734598607178388e-05,
+      "loss": 1.1545,
+      "step": 154290
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.734483814188413e-05,
+      "loss": 1.2567,
+      "step": 154300
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.7343690211984385e-05,
+      "loss": 1.1714,
+      "step": 154310
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.7342542282084635e-05,
+      "loss": 1.1049,
+      "step": 154320
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.734139435218489e-05,
+      "loss": 1.1836,
+      "step": 154330
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.734024642228514e-05,
+      "loss": 1.3192,
+      "step": 154340
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.733909849238539e-05,
+      "loss": 1.3149,
+      "step": 154350
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.733795056248565e-05,
+      "loss": 1.2363,
+      "step": 154360
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.7336802632585904e-05,
+      "loss": 1.1917,
+      "step": 154370
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.733565470268615e-05,
+      "loss": 1.1908,
+      "step": 154380
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.7334506772786404e-05,
+      "loss": 1.2282,
+      "step": 154390
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.733335884288666e-05,
+      "loss": 1.1686,
+      "step": 154400
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 5.7332210912986904e-05,
+      "loss": 1.1895,
+      "step": 154410
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.733106298308716e-05,
+      "loss": 1.2993,
+      "step": 154420
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.732991505318742e-05,
+      "loss": 1.1921,
+      "step": 154430
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.732876712328767e-05,
+      "loss": 1.072,
+      "step": 154440
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.732761919338792e-05,
+      "loss": 1.1212,
+      "step": 154450
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.7326471263488173e-05,
+      "loss": 1.1777,
+      "step": 154460
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.732532333358842e-05,
+      "loss": 1.1229,
+      "step": 154470
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.732417540368868e-05,
+      "loss": 1.0777,
+      "step": 154480
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.732302747378893e-05,
+      "loss": 1.1867,
+      "step": 154490
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.732187954388918e-05,
+      "loss": 1.0919,
+      "step": 154500
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.7320731613989436e-05,
+      "loss": 1.2355,
+      "step": 154510
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.731958368408969e-05,
+      "loss": 1.141,
+      "step": 154520
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.7318435754189936e-05,
+      "loss": 1.1038,
+      "step": 154530
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.731728782429019e-05,
+      "loss": 1.2717,
+      "step": 154540
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.731613989439045e-05,
+      "loss": 1.0808,
+      "step": 154550
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.731499196449069e-05,
+      "loss": 1.1771,
+      "step": 154560
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.731384403459095e-05,
+      "loss": 1.0702,
+      "step": 154570
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.7312696104691206e-05,
+      "loss": 1.0589,
+      "step": 154580
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.7311548174791455e-05,
+      "loss": 1.1692,
+      "step": 154590
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.7310400244891705e-05,
+      "loss": 1.3209,
+      "step": 154600
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.730925231499196e-05,
+      "loss": 1.274,
+      "step": 154610
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.730810438509221e-05,
+      "loss": 1.2427,
+      "step": 154620
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 5.730695645519247e-05,
+      "loss": 1.132,
+      "step": 154630
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.730580852529272e-05,
+      "loss": 1.0749,
+      "step": 154640
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.730466059539297e-05,
+      "loss": 1.1873,
+      "step": 154650
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.7303512665493225e-05,
+      "loss": 1.2184,
+      "step": 154660
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.730236473559348e-05,
+      "loss": 1.1561,
+      "step": 154670
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.7301216805693724e-05,
+      "loss": 1.2883,
+      "step": 154680
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.730006887579398e-05,
+      "loss": 1.2549,
+      "step": 154690
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.729892094589424e-05,
+      "loss": 1.0574,
+      "step": 154700
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.729777301599448e-05,
+      "loss": 1.0695,
+      "step": 154710
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.729662508609474e-05,
+      "loss": 1.1228,
+      "step": 154720
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.7295477156194994e-05,
+      "loss": 1.272,
+      "step": 154730
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.7294329226295244e-05,
+      "loss": 1.2385,
+      "step": 154740
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.7293181296395494e-05,
+      "loss": 1.0822,
+      "step": 154750
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.729203336649575e-05,
+      "loss": 1.2271,
+      "step": 154760
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.7290885436596e-05,
+      "loss": 1.3043,
+      "step": 154770
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.728973750669626e-05,
+      "loss": 1.2862,
+      "step": 154780
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.728858957679651e-05,
+      "loss": 1.1533,
+      "step": 154790
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.7287441646896757e-05,
+      "loss": 1.2131,
+      "step": 154800
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.728629371699701e-05,
+      "loss": 1.2104,
+      "step": 154810
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.728514578709727e-05,
+      "loss": 1.1877,
+      "step": 154820
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.728399785719751e-05,
+      "loss": 1.2489,
+      "step": 154830
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.728284992729777e-05,
+      "loss": 1.2468,
+      "step": 154840
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 5.7281701997398026e-05,
+      "loss": 1.2642,
+      "step": 154850
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.728055406749827e-05,
+      "loss": 1.1631,
+      "step": 154860
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.7279406137598526e-05,
+      "loss": 1.16,
+      "step": 154870
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.727825820769878e-05,
+      "loss": 1.1235,
+      "step": 154880
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.727711027779903e-05,
+      "loss": 1.1903,
+      "step": 154890
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.727596234789928e-05,
+      "loss": 1.2467,
+      "step": 154900
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.727481441799954e-05,
+      "loss": 1.1951,
+      "step": 154910
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.727366648809979e-05,
+      "loss": 1.261,
+      "step": 154920
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.7272518558200045e-05,
+      "loss": 1.1687,
+      "step": 154930
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.727137062830029e-05,
+      "loss": 1.0988,
+      "step": 154940
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.7270222698400545e-05,
+      "loss": 1.2364,
+      "step": 154950
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.72690747685008e-05,
+      "loss": 1.1182,
+      "step": 154960
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.7267926838601045e-05,
+      "loss": 1.1008,
+      "step": 154970
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.72667789087013e-05,
+      "loss": 1.1778,
+      "step": 154980
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.726563097880156e-05,
+      "loss": 1.0055,
+      "step": 154990
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.726448304890181e-05,
+      "loss": 1.1147,
+      "step": 155000
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.726333511900206e-05,
+      "loss": 1.05,
+      "step": 155010
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.7262187189102314e-05,
+      "loss": 1.1793,
+      "step": 155020
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.7261039259202564e-05,
+      "loss": 1.2729,
+      "step": 155030
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.725989132930282e-05,
+      "loss": 1.1661,
+      "step": 155040
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.725874339940307e-05,
+      "loss": 1.2289,
+      "step": 155050
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.725759546950332e-05,
+      "loss": 1.0143,
+      "step": 155060
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 5.725644753960358e-05,
+      "loss": 1.1183,
+      "step": 155070
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.7255299609703834e-05,
+      "loss": 1.1895,
+      "step": 155080
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.725415167980408e-05,
+      "loss": 1.1856,
+      "step": 155090
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.7253003749904334e-05,
+      "loss": 1.1471,
+      "step": 155100
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.725185582000459e-05,
+      "loss": 1.136,
+      "step": 155110
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.725070789010483e-05,
+      "loss": 1.1649,
+      "step": 155120
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.724955996020509e-05,
+      "loss": 1.2562,
+      "step": 155130
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.7248412030305346e-05,
+      "loss": 1.2147,
+      "step": 155140
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.7247264100405596e-05,
+      "loss": 1.1008,
+      "step": 155150
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.7246116170505846e-05,
+      "loss": 1.2219,
+      "step": 155160
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.72449682406061e-05,
+      "loss": 1.084,
+      "step": 155170
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.724382031070635e-05,
+      "loss": 1.1775,
+      "step": 155180
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.724267238080661e-05,
+      "loss": 1.2259,
+      "step": 155190
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.724152445090686e-05,
+      "loss": 1.3156,
+      "step": 155200
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.724037652100711e-05,
+      "loss": 1.1366,
+      "step": 155210
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.7239228591107366e-05,
+      "loss": 1.0927,
+      "step": 155220
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.723808066120762e-05,
+      "loss": 1.1526,
+      "step": 155230
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.7236932731307865e-05,
+      "loss": 1.2322,
+      "step": 155240
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.723578480140812e-05,
+      "loss": 1.1153,
+      "step": 155250
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.723463687150838e-05,
+      "loss": 1.2752,
+      "step": 155260
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.723348894160862e-05,
+      "loss": 1.1983,
+      "step": 155270
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.723234101170888e-05,
+      "loss": 1.2407,
+      "step": 155280
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 5.7231193081809135e-05,
+      "loss": 1.1277,
+      "step": 155290
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.7230045151909385e-05,
+      "loss": 1.1662,
+      "step": 155300
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.7228897222009635e-05,
+      "loss": 1.2104,
+      "step": 155310
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.722774929210989e-05,
+      "loss": 1.2413,
+      "step": 155320
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.722660136221014e-05,
+      "loss": 1.0828,
+      "step": 155330
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.72254534323104e-05,
+      "loss": 1.2215,
+      "step": 155340
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.722430550241065e-05,
+      "loss": 1.12,
+      "step": 155350
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.72231575725109e-05,
+      "loss": 1.2413,
+      "step": 155360
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.7222009642611154e-05,
+      "loss": 1.3082,
+      "step": 155370
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.722086171271141e-05,
+      "loss": 1.1846,
+      "step": 155380
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.7219713782811654e-05,
+      "loss": 1.1991,
+      "step": 155390
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.721856585291191e-05,
+      "loss": 1.0624,
+      "step": 155400
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.721741792301217e-05,
+      "loss": 1.2656,
+      "step": 155410
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.721626999311241e-05,
+      "loss": 1.0675,
+      "step": 155420
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.721512206321267e-05,
+      "loss": 1.1745,
+      "step": 155430
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.7213974133312923e-05,
+      "loss": 1.1726,
+      "step": 155440
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.721282620341317e-05,
+      "loss": 1.0714,
+      "step": 155450
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.721167827351342e-05,
+      "loss": 1.1913,
+      "step": 155460
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.721053034361368e-05,
+      "loss": 1.1059,
+      "step": 155470
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.720938241371393e-05,
+      "loss": 1.1679,
+      "step": 155480
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.7208234483814186e-05,
+      "loss": 1.1064,
+      "step": 155490
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 5.7207086553914436e-05,
+      "loss": 1.1899,
+      "step": 155500
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.7205938624014686e-05,
+      "loss": 1.1024,
+      "step": 155510
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.720479069411494e-05,
+      "loss": 1.1474,
+      "step": 155520
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.72036427642152e-05,
+      "loss": 1.1158,
+      "step": 155530
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.720249483431544e-05,
+      "loss": 1.1247,
+      "step": 155540
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.72013469044157e-05,
+      "loss": 1.1392,
+      "step": 155550
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.7200198974515956e-05,
+      "loss": 1.201,
+      "step": 155560
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.71990510446162e-05,
+      "loss": 1.1368,
+      "step": 155570
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.7197903114716455e-05,
+      "loss": 1.3053,
+      "step": 155580
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.719675518481671e-05,
+      "loss": 1.1376,
+      "step": 155590
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.719560725491696e-05,
+      "loss": 1.1993,
+      "step": 155600
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.719445932501721e-05,
+      "loss": 1.1452,
+      "step": 155610
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.719331139511747e-05,
+      "loss": 1.1404,
+      "step": 155620
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.719216346521772e-05,
+      "loss": 1.1676,
+      "step": 155630
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.7191015535317975e-05,
+      "loss": 1.2223,
+      "step": 155640
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.7189867605418225e-05,
+      "loss": 1.0762,
+      "step": 155650
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.7188719675518474e-05,
+      "loss": 1.0628,
+      "step": 155660
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.718757174561873e-05,
+      "loss": 1.1896,
+      "step": 155670
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.718642381571899e-05,
+      "loss": 1.1392,
+      "step": 155680
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.718527588581923e-05,
+      "loss": 1.2008,
+      "step": 155690
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.718412795591949e-05,
+      "loss": 1.1218,
+      "step": 155700
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.7182980026019744e-05,
+      "loss": 1.1663,
+      "step": 155710
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 5.718183209611999e-05,
+      "loss": 1.1887,
+      "step": 155720
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.7180684166220244e-05,
+      "loss": 1.0358,
+      "step": 155730
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.71795362363205e-05,
+      "loss": 1.2066,
+      "step": 155740
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.717838830642075e-05,
+      "loss": 1.2438,
+      "step": 155750
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.7177240376521e-05,
+      "loss": 1.035,
+      "step": 155760
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.717609244662126e-05,
+      "loss": 1.0969,
+      "step": 155770
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.7174944516721507e-05,
+      "loss": 1.277,
+      "step": 155780
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.717379658682176e-05,
+      "loss": 1.0381,
+      "step": 155790
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.717264865692201e-05,
+      "loss": 1.2683,
+      "step": 155800
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.717150072702226e-05,
+      "loss": 1.2224,
+      "step": 155810
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.717035279712252e-05,
+      "loss": 1.3004,
+      "step": 155820
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.7169204867222776e-05,
+      "loss": 1.1501,
+      "step": 155830
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.716805693732302e-05,
+      "loss": 1.1748,
+      "step": 155840
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.7166909007423276e-05,
+      "loss": 1.2497,
+      "step": 155850
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.716576107752353e-05,
+      "loss": 1.2382,
+      "step": 155860
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.7164613147623776e-05,
+      "loss": 1.1206,
+      "step": 155870
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.716346521772403e-05,
+      "loss": 1.221,
+      "step": 155880
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.716231728782429e-05,
+      "loss": 1.3756,
+      "step": 155890
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.716116935792454e-05,
+      "loss": 1.1975,
+      "step": 155900
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.716002142802479e-05,
+      "loss": 1.2379,
+      "step": 155910
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.7158873498125045e-05,
+      "loss": 1.195,
+      "step": 155920
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.7157725568225295e-05,
+      "loss": 1.186,
+      "step": 155930
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 5.715657763832555e-05,
+      "loss": 1.1342,
+      "step": 155940
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.71554297084258e-05,
+      "loss": 1.0172,
+      "step": 155950
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.715428177852605e-05,
+      "loss": 1.0371,
+      "step": 155960
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.715313384862631e-05,
+      "loss": 1.2378,
+      "step": 155970
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.7151985918726565e-05,
+      "loss": 1.1455,
+      "step": 155980
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.715083798882681e-05,
+      "loss": 1.1342,
+      "step": 155990
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.7149690058927064e-05,
+      "loss": 1.1729,
+      "step": 156000
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.714854212902732e-05,
+      "loss": 1.2515,
+      "step": 156010
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.7147394199127564e-05,
+      "loss": 1.1963,
+      "step": 156020
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.714624626922782e-05,
+      "loss": 1.1207,
+      "step": 156030
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.714509833932808e-05,
+      "loss": 1.1432,
+      "step": 156040
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.714395040942833e-05,
+      "loss": 1.1669,
+      "step": 156050
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.714280247952858e-05,
+      "loss": 1.153,
+      "step": 156060
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.7141654549628834e-05,
+      "loss": 1.3008,
+      "step": 156070
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.7140506619729084e-05,
+      "loss": 1.2205,
+      "step": 156080
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.713935868982934e-05,
+      "loss": 1.2151,
+      "step": 156090
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.713821075992959e-05,
+      "loss": 1.1429,
+      "step": 156100
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.713706283002984e-05,
+      "loss": 1.2936,
+      "step": 156110
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.7135914900130096e-05,
+      "loss": 1.1479,
+      "step": 156120
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.713476697023035e-05,
+      "loss": 1.1264,
+      "step": 156130
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.7133619040330596e-05,
+      "loss": 1.1465,
+      "step": 156140
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.713247111043085e-05,
+      "loss": 1.2072,
+      "step": 156150
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 5.713132318053111e-05,
+      "loss": 1.2543,
+      "step": 156160
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.713017525063135e-05,
+      "loss": 1.1222,
+      "step": 156170
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.712902732073161e-05,
+      "loss": 1.1786,
+      "step": 156180
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.7127879390831866e-05,
+      "loss": 1.207,
+      "step": 156190
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.7126731460932116e-05,
+      "loss": 1.1226,
+      "step": 156200
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.7125583531032366e-05,
+      "loss": 1.1713,
+      "step": 156210
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.712443560113262e-05,
+      "loss": 1.2581,
+      "step": 156220
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.712328767123287e-05,
+      "loss": 1.173,
+      "step": 156230
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.712213974133313e-05,
+      "loss": 1.1323,
+      "step": 156240
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.712099181143338e-05,
+      "loss": 1.2226,
+      "step": 156250
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.711984388153363e-05,
+      "loss": 1.095,
+      "step": 156260
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.7118695951633885e-05,
+      "loss": 1.1076,
+      "step": 156270
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.711754802173414e-05,
+      "loss": 1.3327,
+      "step": 156280
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.7116400091834385e-05,
+      "loss": 1.1308,
+      "step": 156290
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.711525216193464e-05,
+      "loss": 1.0812,
+      "step": 156300
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.711410423203489e-05,
+      "loss": 1.2247,
+      "step": 156310
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.711295630213514e-05,
+      "loss": 1.2981,
+      "step": 156320
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.71118083722354e-05,
+      "loss": 1.153,
+      "step": 156330
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.711066044233565e-05,
+      "loss": 1.1282,
+      "step": 156340
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.7109512512435904e-05,
+      "loss": 1.2487,
+      "step": 156350
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.7108364582536154e-05,
+      "loss": 1.1537,
+      "step": 156360
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.7107216652636404e-05,
+      "loss": 1.3111,
+      "step": 156370
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 5.710606872273666e-05,
+      "loss": 1.1578,
+      "step": 156380
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.710492079283692e-05,
+      "loss": 1.3046,
+      "step": 156390
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.710377286293716e-05,
+      "loss": 1.2455,
+      "step": 156400
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.710262493303742e-05,
+      "loss": 1.1366,
+      "step": 156410
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.7101477003137673e-05,
+      "loss": 1.2147,
+      "step": 156420
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.7100329073237917e-05,
+      "loss": 1.1205,
+      "step": 156430
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.709918114333817e-05,
+      "loss": 1.1391,
+      "step": 156440
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.709803321343843e-05,
+      "loss": 1.154,
+      "step": 156450
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.709688528353868e-05,
+      "loss": 1.079,
+      "step": 156460
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.709573735363893e-05,
+      "loss": 1.1884,
+      "step": 156470
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.7094589423739186e-05,
+      "loss": 1.2147,
+      "step": 156480
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.7093441493839436e-05,
+      "loss": 1.1689,
+      "step": 156490
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.709229356393969e-05,
+      "loss": 1.2556,
+      "step": 156500
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.709114563403994e-05,
+      "loss": 1.2077,
+      "step": 156510
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.708999770414019e-05,
+      "loss": 1.0889,
+      "step": 156520
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.708884977424045e-05,
+      "loss": 1.0873,
+      "step": 156530
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.7087701844340706e-05,
+      "loss": 1.2429,
+      "step": 156540
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.708655391444095e-05,
+      "loss": 1.08,
+      "step": 156550
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.7085405984541205e-05,
+      "loss": 1.1324,
+      "step": 156560
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.708425805464146e-05,
+      "loss": 1.1294,
+      "step": 156570
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.7083110124741705e-05,
+      "loss": 1.2294,
+      "step": 156580
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 5.708196219484196e-05,
+      "loss": 1.0755,
+      "step": 156590
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.708081426494222e-05,
+      "loss": 1.1719,
+      "step": 156600
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.707966633504247e-05,
+      "loss": 1.1235,
+      "step": 156610
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.707851840514272e-05,
+      "loss": 1.1437,
+      "step": 156620
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.7077370475242975e-05,
+      "loss": 1.2018,
+      "step": 156630
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.7076222545343224e-05,
+      "loss": 1.1594,
+      "step": 156640
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.707507461544348e-05,
+      "loss": 1.2005,
+      "step": 156650
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.707392668554373e-05,
+      "loss": 1.404,
+      "step": 156660
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.707277875564398e-05,
+      "loss": 1.0672,
+      "step": 156670
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.707163082574424e-05,
+      "loss": 1.1874,
+      "step": 156680
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.7070482895844494e-05,
+      "loss": 1.1075,
+      "step": 156690
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.706933496594474e-05,
+      "loss": 1.2641,
+      "step": 156700
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.7068187036044994e-05,
+      "loss": 1.1253,
+      "step": 156710
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.706703910614525e-05,
+      "loss": 1.2713,
+      "step": 156720
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.7065891176245493e-05,
+      "loss": 1.2223,
+      "step": 156730
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.706474324634575e-05,
+      "loss": 1.1133,
+      "step": 156740
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.706359531644601e-05,
+      "loss": 1.0607,
+      "step": 156750
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.7062447386546257e-05,
+      "loss": 1.1104,
+      "step": 156760
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.7061299456646506e-05,
+      "loss": 1.307,
+      "step": 156770
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.706015152674676e-05,
+      "loss": 1.1547,
+      "step": 156780
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.705900359684701e-05,
+      "loss": 1.189,
+      "step": 156790
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.705785566694727e-05,
+      "loss": 1.1982,
+      "step": 156800
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 5.705670773704752e-05,
+      "loss": 1.2015,
+      "step": 156810
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.705555980714777e-05,
+      "loss": 1.1077,
+      "step": 156820
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.7054411877248026e-05,
+      "loss": 1.2232,
+      "step": 156830
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.705326394734828e-05,
+      "loss": 1.2446,
+      "step": 156840
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.7052116017448526e-05,
+      "loss": 1.132,
+      "step": 156850
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.705096808754878e-05,
+      "loss": 1.251,
+      "step": 156860
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.704982015764904e-05,
+      "loss": 1.1885,
+      "step": 156870
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.704867222774928e-05,
+      "loss": 1.2126,
+      "step": 156880
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.704752429784954e-05,
+      "loss": 1.1699,
+      "step": 156890
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.7046376367949795e-05,
+      "loss": 1.2852,
+      "step": 156900
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.7045228438050045e-05,
+      "loss": 1.0911,
+      "step": 156910
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.7044080508150295e-05,
+      "loss": 1.117,
+      "step": 156920
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.704293257825055e-05,
+      "loss": 1.1757,
+      "step": 156930
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.70417846483508e-05,
+      "loss": 1.2046,
+      "step": 156940
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.704063671845106e-05,
+      "loss": 1.2119,
+      "step": 156950
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.703948878855131e-05,
+      "loss": 1.1181,
+      "step": 156960
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.703834085865156e-05,
+      "loss": 1.117,
+      "step": 156970
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.7037192928751814e-05,
+      "loss": 1.269,
+      "step": 156980
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.703604499885207e-05,
+      "loss": 1.2263,
+      "step": 156990
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.7034897068952314e-05,
+      "loss": 1.1367,
+      "step": 157000
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.703374913905257e-05,
+      "loss": 1.2353,
+      "step": 157010
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.703260120915283e-05,
+      "loss": 1.1382,
+      "step": 157020
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 5.703145327925307e-05,
+      "loss": 1.141,
+      "step": 157030
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.703030534935333e-05,
+      "loss": 1.2313,
+      "step": 157040
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.7029157419453584e-05,
+      "loss": 1.1274,
+      "step": 157050
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.7028009489553834e-05,
+      "loss": 1.1107,
+      "step": 157060
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.702686155965408e-05,
+      "loss": 1.1786,
+      "step": 157070
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.702571362975434e-05,
+      "loss": 1.1304,
+      "step": 157080
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.702456569985459e-05,
+      "loss": 1.2402,
+      "step": 157090
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.7023417769954846e-05,
+      "loss": 1.29,
+      "step": 157100
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.7022269840055096e-05,
+      "loss": 1.1561,
+      "step": 157110
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.7021121910155346e-05,
+      "loss": 1.0554,
+      "step": 157120
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.70199739802556e-05,
+      "loss": 1.1033,
+      "step": 157130
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.701882605035586e-05,
+      "loss": 1.1641,
+      "step": 157140
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.70176781204561e-05,
+      "loss": 0.9628,
+      "step": 157150
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.701653019055636e-05,
+      "loss": 1.1188,
+      "step": 157160
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.7015382260656616e-05,
+      "loss": 1.118,
+      "step": 157170
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.701423433075686e-05,
+      "loss": 1.2392,
+      "step": 157180
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.7013086400857115e-05,
+      "loss": 1.1,
+      "step": 157190
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.701193847095737e-05,
+      "loss": 1.2647,
+      "step": 157200
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.701079054105762e-05,
+      "loss": 1.0295,
+      "step": 157210
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.700964261115787e-05,
+      "loss": 1.1027,
+      "step": 157220
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.700849468125813e-05,
+      "loss": 1.1161,
+      "step": 157230
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.700734675135838e-05,
+      "loss": 1.192,
+      "step": 157240
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 5.7006198821458635e-05,
+      "loss": 1.1255,
+      "step": 157250
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.7005050891558885e-05,
+      "loss": 1.1226,
+      "step": 157260
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.7003902961659135e-05,
+      "loss": 1.1787,
+      "step": 157270
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.700275503175939e-05,
+      "loss": 1.2067,
+      "step": 157280
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.700160710185965e-05,
+      "loss": 1.1933,
+      "step": 157290
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.700045917195989e-05,
+      "loss": 1.0454,
+      "step": 157300
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.699931124206015e-05,
+      "loss": 1.0952,
+      "step": 157310
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.6998163312160404e-05,
+      "loss": 1.1794,
+      "step": 157320
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.699701538226065e-05,
+      "loss": 1.2317,
+      "step": 157330
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.6995867452360904e-05,
+      "loss": 1.1787,
+      "step": 157340
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.699471952246116e-05,
+      "loss": 1.0245,
+      "step": 157350
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.699357159256141e-05,
+      "loss": 1.1538,
+      "step": 157360
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.699242366266166e-05,
+      "loss": 1.2091,
+      "step": 157370
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.699127573276192e-05,
+      "loss": 1.2705,
+      "step": 157380
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.699012780286217e-05,
+      "loss": 1.3072,
+      "step": 157390
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.6988979872962423e-05,
+      "loss": 1.1991,
+      "step": 157400
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.698783194306267e-05,
+      "loss": 1.1094,
+      "step": 157410
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.698668401316292e-05,
+      "loss": 1.2027,
+      "step": 157420
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.698553608326318e-05,
+      "loss": 1.1797,
+      "step": 157430
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.6984388153363436e-05,
+      "loss": 1.2145,
+      "step": 157440
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.698324022346368e-05,
+      "loss": 1.055,
+      "step": 157450
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.6982092293563936e-05,
+      "loss": 1.1489,
+      "step": 157460
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 5.698094436366419e-05,
+      "loss": 1.0931,
+      "step": 157470
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.6979796433764436e-05,
+      "loss": 1.2369,
+      "step": 157480
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.697864850386469e-05,
+      "loss": 1.1352,
+      "step": 157490
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.697750057396495e-05,
+      "loss": 1.1982,
+      "step": 157500
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.69763526440652e-05,
+      "loss": 1.1917,
+      "step": 157510
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.697520471416545e-05,
+      "loss": 1.1547,
+      "step": 157520
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.6974056784265705e-05,
+      "loss": 1.1907,
+      "step": 157530
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.6972908854365955e-05,
+      "loss": 1.074,
+      "step": 157540
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.697176092446621e-05,
+      "loss": 1.2059,
+      "step": 157550
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.697061299456646e-05,
+      "loss": 1.2468,
+      "step": 157560
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.696946506466671e-05,
+      "loss": 1.2977,
+      "step": 157570
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.696831713476697e-05,
+      "loss": 1.1735,
+      "step": 157580
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.6967169204867225e-05,
+      "loss": 1.0467,
+      "step": 157590
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.696602127496747e-05,
+      "loss": 1.2072,
+      "step": 157600
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.6964873345067725e-05,
+      "loss": 1.1217,
+      "step": 157610
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.696372541516798e-05,
+      "loss": 1.1751,
+      "step": 157620
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.6962577485268224e-05,
+      "loss": 1.1454,
+      "step": 157630
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.696142955536848e-05,
+      "loss": 1.1905,
+      "step": 157640
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.696028162546874e-05,
+      "loss": 1.1314,
+      "step": 157650
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.695913369556899e-05,
+      "loss": 1.1973,
+      "step": 157660
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.695798576566924e-05,
+      "loss": 1.1869,
+      "step": 157670
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 5.695683783576949e-05,
+      "loss": 1.0708,
+      "step": 157680
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.6955689905869744e-05,
+      "loss": 1.2818,
+      "step": 157690
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.695454197597e-05,
+      "loss": 1.184,
+      "step": 157700
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.6953394046070243e-05,
+      "loss": 1.4397,
+      "step": 157710
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.69522461161705e-05,
+      "loss": 1.3299,
+      "step": 157720
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.695109818627076e-05,
+      "loss": 1.1541,
+      "step": 157730
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.6949950256371e-05,
+      "loss": 1.1289,
+      "step": 157740
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.6948802326471256e-05,
+      "loss": 1.2118,
+      "step": 157750
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.694765439657151e-05,
+      "loss": 1.094,
+      "step": 157760
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.694650646667176e-05,
+      "loss": 1.1627,
+      "step": 157770
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.694535853677201e-05,
+      "loss": 1.2683,
+      "step": 157780
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.694421060687227e-05,
+      "loss": 1.1355,
+      "step": 157790
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.694306267697252e-05,
+      "loss": 1.0894,
+      "step": 157800
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.6941914747072776e-05,
+      "loss": 1.0825,
+      "step": 157810
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.6940766817173026e-05,
+      "loss": 1.0675,
+      "step": 157820
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.6939618887273276e-05,
+      "loss": 1.2184,
+      "step": 157830
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.693847095737353e-05,
+      "loss": 1.0918,
+      "step": 157840
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.693732302747379e-05,
+      "loss": 1.1304,
+      "step": 157850
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.693617509757403e-05,
+      "loss": 1.3013,
+      "step": 157860
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.693502716767429e-05,
+      "loss": 1.1818,
+      "step": 157870
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.6933879237774545e-05,
+      "loss": 1.2191,
+      "step": 157880
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.693273130787479e-05,
+      "loss": 1.1957,
+      "step": 157890
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 5.6931583377975045e-05,
+      "loss": 1.1822,
+      "step": 157900
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.69304354480753e-05,
+      "loss": 1.1378,
+      "step": 157910
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.692928751817555e-05,
+      "loss": 1.0844,
+      "step": 157920
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.69281395882758e-05,
+      "loss": 1.116,
+      "step": 157930
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.692699165837606e-05,
+      "loss": 1.1193,
+      "step": 157940
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.692584372847631e-05,
+      "loss": 1.1314,
+      "step": 157950
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.6924695798576564e-05,
+      "loss": 1.272,
+      "step": 157960
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.6923547868676814e-05,
+      "loss": 1.141,
+      "step": 157970
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.6922399938777064e-05,
+      "loss": 1.2007,
+      "step": 157980
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.692125200887732e-05,
+      "loss": 1.0963,
+      "step": 157990
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.692010407897758e-05,
+      "loss": 1.2733,
+      "step": 158000
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.691895614907782e-05,
+      "loss": 1.0034,
+      "step": 158010
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.691780821917808e-05,
+      "loss": 1.2601,
+      "step": 158020
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.6916660289278334e-05,
+      "loss": 1.2133,
+      "step": 158030
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.691551235937858e-05,
+      "loss": 1.2154,
+      "step": 158040
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.691436442947883e-05,
+      "loss": 1.107,
+      "step": 158050
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.691321649957909e-05,
+      "loss": 1.2047,
+      "step": 158060
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.691206856967934e-05,
+      "loss": 1.3269,
+      "step": 158070
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.691092063977959e-05,
+      "loss": 1.041,
+      "step": 158080
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.6909772709879846e-05,
+      "loss": 1.2262,
+      "step": 158090
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.6908624779980096e-05,
+      "loss": 1.1247,
+      "step": 158100
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.690747685008035e-05,
+      "loss": 1.3052,
+      "step": 158110
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 5.69063289201806e-05,
+      "loss": 1.2403,
+      "step": 158120
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.690518099028085e-05,
+      "loss": 1.2478,
+      "step": 158130
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.690403306038111e-05,
+      "loss": 1.1435,
+      "step": 158140
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.6902885130481366e-05,
+      "loss": 1.1093,
+      "step": 158150
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.690173720058161e-05,
+      "loss": 1.2384,
+      "step": 158160
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.6900589270681865e-05,
+      "loss": 1.2899,
+      "step": 158170
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.689944134078212e-05,
+      "loss": 1.272,
+      "step": 158180
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.6898293410882365e-05,
+      "loss": 1.2955,
+      "step": 158190
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.689714548098262e-05,
+      "loss": 1.141,
+      "step": 158200
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.689599755108288e-05,
+      "loss": 1.1905,
+      "step": 158210
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.689484962118313e-05,
+      "loss": 1.2088,
+      "step": 158220
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.689370169128338e-05,
+      "loss": 1.2028,
+      "step": 158230
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.6892553761383635e-05,
+      "loss": 1.1377,
+      "step": 158240
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.6891405831483885e-05,
+      "loss": 1.2562,
+      "step": 158250
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.689025790158414e-05,
+      "loss": 1.2086,
+      "step": 158260
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.688910997168439e-05,
+      "loss": 1.1221,
+      "step": 158270
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.688796204178464e-05,
+      "loss": 1.1891,
+      "step": 158280
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.68868141118849e-05,
+      "loss": 1.0769,
+      "step": 158290
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.6885666181985154e-05,
+      "loss": 1.1478,
+      "step": 158300
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.68845182520854e-05,
+      "loss": 1.1125,
+      "step": 158310
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.6883370322185654e-05,
+      "loss": 1.2264,
+      "step": 158320
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.688222239228591e-05,
+      "loss": 1.1651,
+      "step": 158330
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 5.6881074462386154e-05,
+      "loss": 1.3553,
+      "step": 158340
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.687992653248641e-05,
+      "loss": 1.1444,
+      "step": 158350
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.687877860258667e-05,
+      "loss": 1.2509,
+      "step": 158360
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.687763067268692e-05,
+      "loss": 0.9169,
+      "step": 158370
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.687648274278717e-05,
+      "loss": 1.1298,
+      "step": 158380
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.687533481288742e-05,
+      "loss": 1.1968,
+      "step": 158390
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.687418688298767e-05,
+      "loss": 1.0661,
+      "step": 158400
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.687303895308793e-05,
+      "loss": 1.3063,
+      "step": 158410
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.687189102318818e-05,
+      "loss": 1.2464,
+      "step": 158420
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.687074309328843e-05,
+      "loss": 1.2849,
+      "step": 158430
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.6869595163388686e-05,
+      "loss": 1.2752,
+      "step": 158440
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.686844723348894e-05,
+      "loss": 1.1996,
+      "step": 158450
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.6867299303589186e-05,
+      "loss": 1.1944,
+      "step": 158460
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.686615137368944e-05,
+      "loss": 1.2009,
+      "step": 158470
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.68650034437897e-05,
+      "loss": 1.1491,
+      "step": 158480
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.686385551388994e-05,
+      "loss": 1.043,
+      "step": 158490
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.68627075839902e-05,
+      "loss": 1.2173,
+      "step": 158500
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.6861559654090455e-05,
+      "loss": 1.2757,
+      "step": 158510
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.6860411724190705e-05,
+      "loss": 1.2503,
+      "step": 158520
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.6859263794290955e-05,
+      "loss": 1.1035,
+      "step": 158530
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.685811586439121e-05,
+      "loss": 1.0713,
+      "step": 158540
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.685696793449146e-05,
+      "loss": 1.0933,
+      "step": 158550
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 5.685582000459172e-05,
+      "loss": 1.1435,
+      "step": 158560
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.685467207469197e-05,
+      "loss": 1.1467,
+      "step": 158570
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.685352414479222e-05,
+      "loss": 1.1333,
+      "step": 158580
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.6852376214892475e-05,
+      "loss": 1.1741,
+      "step": 158590
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.685122828499273e-05,
+      "loss": 1.1136,
+      "step": 158600
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.6850080355092974e-05,
+      "loss": 1.0886,
+      "step": 158610
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.684893242519323e-05,
+      "loss": 1.2159,
+      "step": 158620
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.684778449529349e-05,
+      "loss": 1.1598,
+      "step": 158630
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.684663656539373e-05,
+      "loss": 1.1481,
+      "step": 158640
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.684548863549399e-05,
+      "loss": 1.1368,
+      "step": 158650
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.6844340705594244e-05,
+      "loss": 1.2002,
+      "step": 158660
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.6843192775694494e-05,
+      "loss": 1.2652,
+      "step": 158670
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.6842044845794744e-05,
+      "loss": 1.1655,
+      "step": 158680
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.6840896915895e-05,
+      "loss": 1.1919,
+      "step": 158690
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.683974898599525e-05,
+      "loss": 1.2378,
+      "step": 158700
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.683860105609551e-05,
+      "loss": 1.0609,
+      "step": 158710
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.6837453126195757e-05,
+      "loss": 1.1637,
+      "step": 158720
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.6836305196296006e-05,
+      "loss": 1.067,
+      "step": 158730
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.683515726639626e-05,
+      "loss": 1.2903,
+      "step": 158740
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.683400933649652e-05,
+      "loss": 1.1989,
+      "step": 158750
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.683286140659676e-05,
+      "loss": 1.1101,
+      "step": 158760
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 5.683171347669702e-05,
+      "loss": 1.188,
+      "step": 158770
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.6830565546797276e-05,
+      "loss": 1.1877,
+      "step": 158780
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.682941761689752e-05,
+      "loss": 1.1845,
+      "step": 158790
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.6828269686997776e-05,
+      "loss": 1.2907,
+      "step": 158800
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.682712175709803e-05,
+      "loss": 1.2142,
+      "step": 158810
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.682597382719828e-05,
+      "loss": 1.1626,
+      "step": 158820
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.682482589729853e-05,
+      "loss": 1.1519,
+      "step": 158830
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.682367796739879e-05,
+      "loss": 1.1699,
+      "step": 158840
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.682253003749904e-05,
+      "loss": 1.1679,
+      "step": 158850
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.6821382107599295e-05,
+      "loss": 1.0788,
+      "step": 158860
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.6820234177699545e-05,
+      "loss": 1.1727,
+      "step": 158870
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.6819086247799795e-05,
+      "loss": 1.1602,
+      "step": 158880
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.681793831790005e-05,
+      "loss": 1.1439,
+      "step": 158890
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.681679038800031e-05,
+      "loss": 1.2228,
+      "step": 158900
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.681564245810055e-05,
+      "loss": 1.1159,
+      "step": 158910
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.681449452820081e-05,
+      "loss": 1.1844,
+      "step": 158920
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.6813346598301064e-05,
+      "loss": 1.287,
+      "step": 158930
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.681219866840131e-05,
+      "loss": 1.2913,
+      "step": 158940
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.6811050738501564e-05,
+      "loss": 1.111,
+      "step": 158950
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.680990280860182e-05,
+      "loss": 1.2312,
+      "step": 158960
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.680875487870207e-05,
+      "loss": 1.2146,
+      "step": 158970
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.680760694880232e-05,
+      "loss": 1.1532,
+      "step": 158980
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 5.680645901890258e-05,
+      "loss": 1.0603,
+      "step": 158990
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.680531108900283e-05,
+      "loss": 1.1772,
+      "step": 159000
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.6804163159103084e-05,
+      "loss": 1.1162,
+      "step": 159010
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.6803015229203333e-05,
+      "loss": 1.2722,
+      "step": 159020
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.680186729930358e-05,
+      "loss": 1.0205,
+      "step": 159030
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.680071936940384e-05,
+      "loss": 1.215,
+      "step": 159040
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.679957143950408e-05,
+      "loss": 1.1781,
+      "step": 159050
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.679842350960434e-05,
+      "loss": 1.2885,
+      "step": 159060
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.6797275579704596e-05,
+      "loss": 1.2126,
+      "step": 159070
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.6796127649804846e-05,
+      "loss": 1.2187,
+      "step": 159080
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.6794979719905096e-05,
+      "loss": 1.2579,
+      "step": 159090
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.679383179000535e-05,
+      "loss": 1.034,
+      "step": 159100
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.67926838601056e-05,
+      "loss": 1.1638,
+      "step": 159110
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.679153593020586e-05,
+      "loss": 1.2012,
+      "step": 159120
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.679038800030611e-05,
+      "loss": 1.252,
+      "step": 159130
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.678924007040636e-05,
+      "loss": 1.1968,
+      "step": 159140
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.6788092140506615e-05,
+      "loss": 1.3323,
+      "step": 159150
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.678694421060687e-05,
+      "loss": 1.1401,
+      "step": 159160
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.6785796280707115e-05,
+      "loss": 1.4159,
+      "step": 159170
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.678464835080737e-05,
+      "loss": 1.2741,
+      "step": 159180
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.678350042090763e-05,
+      "loss": 1.1799,
+      "step": 159190
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.678235249100787e-05,
+      "loss": 1.2596,
+      "step": 159200
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 5.678120456110813e-05,
+      "loss": 1.218,
+      "step": 159210
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.6780056631208385e-05,
+      "loss": 1.1452,
+      "step": 159220
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.6778908701308635e-05,
+      "loss": 1.216,
+      "step": 159230
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.6777760771408885e-05,
+      "loss": 1.1178,
+      "step": 159240
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.677661284150914e-05,
+      "loss": 1.1824,
+      "step": 159250
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.677546491160939e-05,
+      "loss": 1.2089,
+      "step": 159260
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.677431698170965e-05,
+      "loss": 1.1611,
+      "step": 159270
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.67731690518099e-05,
+      "loss": 1.2335,
+      "step": 159280
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.677202112191015e-05,
+      "loss": 1.1829,
+      "step": 159290
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.6770873192010404e-05,
+      "loss": 1.2204,
+      "step": 159300
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.676972526211066e-05,
+      "loss": 1.2623,
+      "step": 159310
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.6768577332210904e-05,
+      "loss": 1.2471,
+      "step": 159320
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.676742940231116e-05,
+      "loss": 1.1642,
+      "step": 159330
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.676628147241142e-05,
+      "loss": 1.1249,
+      "step": 159340
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.676513354251166e-05,
+      "loss": 1.1593,
+      "step": 159350
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.676398561261192e-05,
+      "loss": 1.3518,
+      "step": 159360
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.676283768271217e-05,
+      "loss": 1.1133,
+      "step": 159370
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.676168975281242e-05,
+      "loss": 1.3279,
+      "step": 159380
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.676054182291267e-05,
+      "loss": 1.1402,
+      "step": 159390
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.675939389301293e-05,
+      "loss": 1.0028,
+      "step": 159400
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.675824596311318e-05,
+      "loss": 1.0746,
+      "step": 159410
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.6757098033213436e-05,
+      "loss": 1.1702,
+      "step": 159420
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 5.6755950103313686e-05,
+      "loss": 1.0958,
+      "step": 159430
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.6754802173413936e-05,
+      "loss": 1.1172,
+      "step": 159440
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.675365424351419e-05,
+      "loss": 1.213,
+      "step": 159450
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.675250631361445e-05,
+      "loss": 1.2205,
+      "step": 159460
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.675135838371469e-05,
+      "loss": 1.144,
+      "step": 159470
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.675021045381495e-05,
+      "loss": 1.1487,
+      "step": 159480
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.6749062523915205e-05,
+      "loss": 1.1246,
+      "step": 159490
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.674791459401545e-05,
+      "loss": 0.9883,
+      "step": 159500
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.6746766664115705e-05,
+      "loss": 1.2499,
+      "step": 159510
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.674561873421596e-05,
+      "loss": 1.1831,
+      "step": 159520
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.674447080431621e-05,
+      "loss": 1.127,
+      "step": 159530
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.674332287441646e-05,
+      "loss": 1.0468,
+      "step": 159540
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.674217494451672e-05,
+      "loss": 1.2689,
+      "step": 159550
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.674102701461697e-05,
+      "loss": 1.1658,
+      "step": 159560
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.6739879084717225e-05,
+      "loss": 1.2671,
+      "step": 159570
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.6738731154817474e-05,
+      "loss": 1.2149,
+      "step": 159580
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.6737583224917724e-05,
+      "loss": 1.1149,
+      "step": 159590
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.673643529501798e-05,
+      "loss": 1.1637,
+      "step": 159600
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.673528736511824e-05,
+      "loss": 1.121,
+      "step": 159610
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.673413943521848e-05,
+      "loss": 1.1135,
+      "step": 159620
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.673299150531874e-05,
+      "loss": 1.2946,
+      "step": 159630
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.6731843575418994e-05,
+      "loss": 1.1599,
+      "step": 159640
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 5.673069564551924e-05,
+      "loss": 1.2602,
+      "step": 159650
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.6729547715619494e-05,
+      "loss": 1.1977,
+      "step": 159660
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.672839978571975e-05,
+      "loss": 1.1859,
+      "step": 159670
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.672725185582e-05,
+      "loss": 1.2068,
+      "step": 159680
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.672610392592025e-05,
+      "loss": 1.0197,
+      "step": 159690
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.6724955996020507e-05,
+      "loss": 1.257,
+      "step": 159700
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.6723808066120756e-05,
+      "loss": 1.2211,
+      "step": 159710
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.672266013622101e-05,
+      "loss": 1.1786,
+      "step": 159720
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.672151220632126e-05,
+      "loss": 1.1751,
+      "step": 159730
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.672036427642151e-05,
+      "loss": 1.1347,
+      "step": 159740
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.671921634652177e-05,
+      "loss": 1.3109,
+      "step": 159750
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.6718068416622026e-05,
+      "loss": 1.2637,
+      "step": 159760
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.671692048672227e-05,
+      "loss": 1.2231,
+      "step": 159770
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.6715772556822526e-05,
+      "loss": 1.2174,
+      "step": 159780
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.671462462692278e-05,
+      "loss": 1.187,
+      "step": 159790
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.6713476697023025e-05,
+      "loss": 1.1004,
+      "step": 159800
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.671232876712328e-05,
+      "loss": 1.1493,
+      "step": 159810
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.671118083722354e-05,
+      "loss": 1.1294,
+      "step": 159820
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.671003290732379e-05,
+      "loss": 1.1021,
+      "step": 159830
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.670888497742404e-05,
+      "loss": 1.1342,
+      "step": 159840
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.6707737047524295e-05,
+      "loss": 1.1404,
+      "step": 159850
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 5.6706589117624545e-05,
+      "loss": 1.3232,
+      "step": 159860
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.67054411877248e-05,
+      "loss": 1.1433,
+      "step": 159870
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.670429325782505e-05,
+      "loss": 1.1039,
+      "step": 159880
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.67031453279253e-05,
+      "loss": 1.0969,
+      "step": 159890
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.670199739802556e-05,
+      "loss": 1.0925,
+      "step": 159900
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.6700849468125814e-05,
+      "loss": 1.2194,
+      "step": 159910
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.669970153822606e-05,
+      "loss": 1.2997,
+      "step": 159920
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.6698553608326314e-05,
+      "loss": 1.1755,
+      "step": 159930
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.669740567842657e-05,
+      "loss": 1.1195,
+      "step": 159940
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.6696257748526814e-05,
+      "loss": 1.0603,
+      "step": 159950
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.669510981862707e-05,
+      "loss": 1.3014,
+      "step": 159960
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.669396188872733e-05,
+      "loss": 0.9695,
+      "step": 159970
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.669281395882758e-05,
+      "loss": 1.2041,
+      "step": 159980
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.669166602892783e-05,
+      "loss": 1.1621,
+      "step": 159990
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.6690518099028083e-05,
+      "loss": 1.1267,
+      "step": 160000
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.668937016912833e-05,
+      "loss": 1.0791,
+      "step": 160010
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.668822223922859e-05,
+      "loss": 1.1378,
+      "step": 160020
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.668707430932884e-05,
+      "loss": 1.1483,
+      "step": 160030
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.668592637942909e-05,
+      "loss": 1.1612,
+      "step": 160040
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.6684778449529346e-05,
+      "loss": 1.2791,
+      "step": 160050
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.66836305196296e-05,
+      "loss": 1.3505,
+      "step": 160060
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.6682482589729846e-05,
+      "loss": 1.2104,
+      "step": 160070
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 5.66813346598301e-05,
+      "loss": 1.1761,
+      "step": 160080
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.668018672993036e-05,
+      "loss": 1.1254,
+      "step": 160090
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.66790388000306e-05,
+      "loss": 1.0182,
+      "step": 160100
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.667789087013086e-05,
+      "loss": 1.1369,
+      "step": 160110
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.6676742940231116e-05,
+      "loss": 1.0818,
+      "step": 160120
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.6675595010331365e-05,
+      "loss": 1.2803,
+      "step": 160130
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.6674447080431615e-05,
+      "loss": 1.126,
+      "step": 160140
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.667329915053187e-05,
+      "loss": 1.1284,
+      "step": 160150
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.667215122063212e-05,
+      "loss": 1.2549,
+      "step": 160160
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.667100329073238e-05,
+      "loss": 1.1736,
+      "step": 160170
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.666985536083263e-05,
+      "loss": 1.2036,
+      "step": 160180
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.666870743093288e-05,
+      "loss": 1.1885,
+      "step": 160190
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.6667559501033135e-05,
+      "loss": 1.1574,
+      "step": 160200
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.666641157113339e-05,
+      "loss": 1.3201,
+      "step": 160210
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.6665263641233634e-05,
+      "loss": 1.1332,
+      "step": 160220
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.666411571133389e-05,
+      "loss": 1.22,
+      "step": 160230
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.666296778143415e-05,
+      "loss": 1.1055,
+      "step": 160240
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.666181985153439e-05,
+      "loss": 1.1219,
+      "step": 160250
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.666067192163465e-05,
+      "loss": 1.0942,
+      "step": 160260
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.6659523991734904e-05,
+      "loss": 1.0532,
+      "step": 160270
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.6658376061835154e-05,
+      "loss": 1.0928,
+      "step": 160280
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.6657228131935404e-05,
+      "loss": 1.1725,
+      "step": 160290
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 5.665608020203566e-05,
+      "loss": 1.2626,
+      "step": 160300
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.665493227213591e-05,
+      "loss": 1.1981,
+      "step": 160310
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.665378434223617e-05,
+      "loss": 1.2918,
+      "step": 160320
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.665263641233642e-05,
+      "loss": 1.045,
+      "step": 160330
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.665148848243667e-05,
+      "loss": 1.1286,
+      "step": 160340
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.665034055253692e-05,
+      "loss": 1.111,
+      "step": 160350
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.664919262263718e-05,
+      "loss": 1.1336,
+      "step": 160360
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.664804469273742e-05,
+      "loss": 1.1658,
+      "step": 160370
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.664689676283768e-05,
+      "loss": 1.1477,
+      "step": 160380
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.6645748832937936e-05,
+      "loss": 1.1929,
+      "step": 160390
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.664460090303818e-05,
+      "loss": 1.0597,
+      "step": 160400
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.6643452973138436e-05,
+      "loss": 1.3323,
+      "step": 160410
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.6642305043238686e-05,
+      "loss": 1.0927,
+      "step": 160420
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.664115711333894e-05,
+      "loss": 1.0221,
+      "step": 160430
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.664000918343919e-05,
+      "loss": 1.3118,
+      "step": 160440
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.663886125353944e-05,
+      "loss": 1.1681,
+      "step": 160450
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.66377133236397e-05,
+      "loss": 1.0322,
+      "step": 160460
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.6636565393739955e-05,
+      "loss": 1.1151,
+      "step": 160470
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.66354174638402e-05,
+      "loss": 1.2509,
+      "step": 160480
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.6634269533940455e-05,
+      "loss": 1.1557,
+      "step": 160490
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.663312160404071e-05,
+      "loss": 1.1209,
+      "step": 160500
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.6631973674140955e-05,
+      "loss": 1.0983,
+      "step": 160510
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 5.663082574424121e-05,
+      "loss": 1.1611,
+      "step": 160520
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.662967781434147e-05,
+      "loss": 1.1668,
+      "step": 160530
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.662852988444172e-05,
+      "loss": 1.156,
+      "step": 160540
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.662738195454197e-05,
+      "loss": 1.1935,
+      "step": 160550
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.6626234024642224e-05,
+      "loss": 1.143,
+      "step": 160560
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.6625086094742474e-05,
+      "loss": 1.2369,
+      "step": 160570
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.662393816484273e-05,
+      "loss": 1.082,
+      "step": 160580
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.662279023494298e-05,
+      "loss": 1.2218,
+      "step": 160590
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.662164230504323e-05,
+      "loss": 1.3883,
+      "step": 160600
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.662049437514349e-05,
+      "loss": 1.1219,
+      "step": 160610
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.6619346445243744e-05,
+      "loss": 1.1729,
+      "step": 160620
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.661819851534399e-05,
+      "loss": 1.242,
+      "step": 160630
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.6617050585444244e-05,
+      "loss": 1.1557,
+      "step": 160640
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.66159026555445e-05,
+      "loss": 1.1216,
+      "step": 160650
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.661475472564474e-05,
+      "loss": 1.2744,
+      "step": 160660
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.6613606795745e-05,
+      "loss": 1.1773,
+      "step": 160670
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.6612458865845257e-05,
+      "loss": 1.1776,
+      "step": 160680
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.6611310935945506e-05,
+      "loss": 1.1111,
+      "step": 160690
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.6610163006045756e-05,
+      "loss": 1.1982,
+      "step": 160700
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.660901507614601e-05,
+      "loss": 1.1451,
+      "step": 160710
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.660786714624626e-05,
+      "loss": 1.1655,
+      "step": 160720
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.660671921634652e-05,
+      "loss": 1.1993,
+      "step": 160730
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 5.660557128644677e-05,
+      "loss": 1.1597,
+      "step": 160740
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.660442335654702e-05,
+      "loss": 1.0316,
+      "step": 160750
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.6603275426647276e-05,
+      "loss": 1.1793,
+      "step": 160760
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.660212749674753e-05,
+      "loss": 1.1907,
+      "step": 160770
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.6600979566847775e-05,
+      "loss": 1.202,
+      "step": 160780
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.659983163694803e-05,
+      "loss": 1.0585,
+      "step": 160790
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.659868370704829e-05,
+      "loss": 1.2153,
+      "step": 160800
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.659753577714853e-05,
+      "loss": 1.1881,
+      "step": 160810
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.659638784724879e-05,
+      "loss": 1.1753,
+      "step": 160820
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.6595239917349045e-05,
+      "loss": 1.1052,
+      "step": 160830
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.6594091987449295e-05,
+      "loss": 1.1909,
+      "step": 160840
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.6592944057549545e-05,
+      "loss": 1.1176,
+      "step": 160850
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.65917961276498e-05,
+      "loss": 1.1847,
+      "step": 160860
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.659064819775005e-05,
+      "loss": 1.102,
+      "step": 160870
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.658950026785031e-05,
+      "loss": 1.0661,
+      "step": 160880
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.658835233795056e-05,
+      "loss": 1.0474,
+      "step": 160890
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.658720440805081e-05,
+      "loss": 1.1929,
+      "step": 160900
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.6586056478151064e-05,
+      "loss": 1.2846,
+      "step": 160910
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.658490854825132e-05,
+      "loss": 1.225,
+      "step": 160920
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.6583760618351564e-05,
+      "loss": 1.1773,
+      "step": 160930
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.658261268845182e-05,
+      "loss": 1.1222,
+      "step": 160940
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 5.658146475855208e-05,
+      "loss": 1.0428,
+      "step": 160950
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.658031682865232e-05,
+      "loss": 1.0715,
+      "step": 160960
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.657916889875258e-05,
+      "loss": 1.1463,
+      "step": 160970
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.6578020968852833e-05,
+      "loss": 1.1722,
+      "step": 160980
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.657687303895308e-05,
+      "loss": 1.146,
+      "step": 160990
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.657572510905333e-05,
+      "loss": 1.1586,
+      "step": 161000
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.657457717915359e-05,
+      "loss": 1.3311,
+      "step": 161010
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.657342924925384e-05,
+      "loss": 1.2625,
+      "step": 161020
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.6572281319354096e-05,
+      "loss": 1.171,
+      "step": 161030
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.6571133389454346e-05,
+      "loss": 1.098,
+      "step": 161040
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.6569985459554596e-05,
+      "loss": 1.1888,
+      "step": 161050
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.656883752965485e-05,
+      "loss": 1.1922,
+      "step": 161060
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.656768959975511e-05,
+      "loss": 1.0959,
+      "step": 161070
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.656654166985535e-05,
+      "loss": 1.1758,
+      "step": 161080
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.656539373995561e-05,
+      "loss": 1.06,
+      "step": 161090
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.6564245810055866e-05,
+      "loss": 1.1741,
+      "step": 161100
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.656309788015611e-05,
+      "loss": 1.1863,
+      "step": 161110
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.6561949950256365e-05,
+      "loss": 1.233,
+      "step": 161120
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.656080202035662e-05,
+      "loss": 1.2075,
+      "step": 161130
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.655965409045687e-05,
+      "loss": 1.1971,
+      "step": 161140
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.655850616055712e-05,
+      "loss": 1.1577,
+      "step": 161150
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.655735823065738e-05,
+      "loss": 1.3332,
+      "step": 161160
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 5.655621030075763e-05,
+      "loss": 1.1753,
+      "step": 161170
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.6555062370857885e-05,
+      "loss": 1.1798,
+      "step": 161180
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.6553914440958135e-05,
+      "loss": 1.1502,
+      "step": 161190
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.6552766511058384e-05,
+      "loss": 1.2067,
+      "step": 161200
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.655161858115864e-05,
+      "loss": 1.2294,
+      "step": 161210
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.65504706512589e-05,
+      "loss": 1.2027,
+      "step": 161220
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.654932272135914e-05,
+      "loss": 1.2187,
+      "step": 161230
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.65481747914594e-05,
+      "loss": 1.0979,
+      "step": 161240
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.6547026861559654e-05,
+      "loss": 1.1898,
+      "step": 161250
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.65458789316599e-05,
+      "loss": 1.0934,
+      "step": 161260
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.6544731001760154e-05,
+      "loss": 1.1259,
+      "step": 161270
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.654358307186041e-05,
+      "loss": 1.1685,
+      "step": 161280
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.654243514196066e-05,
+      "loss": 1.2727,
+      "step": 161290
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.654128721206091e-05,
+      "loss": 1.2395,
+      "step": 161300
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.654013928216117e-05,
+      "loss": 1.1683,
+      "step": 161310
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.653899135226142e-05,
+      "loss": 1.132,
+      "step": 161320
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.653784342236167e-05,
+      "loss": 1.04,
+      "step": 161330
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.653669549246192e-05,
+      "loss": 1.055,
+      "step": 161340
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.653554756256217e-05,
+      "loss": 1.1264,
+      "step": 161350
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.653439963266243e-05,
+      "loss": 1.3999,
+      "step": 161360
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.6533251702762686e-05,
+      "loss": 1.2255,
+      "step": 161370
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.653210377286293e-05,
+      "loss": 1.1376,
+      "step": 161380
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 5.6530955842963186e-05,
+      "loss": 1.1677,
+      "step": 161390
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.652980791306344e-05,
+      "loss": 1.1914,
+      "step": 161400
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.6528659983163686e-05,
+      "loss": 1.2141,
+      "step": 161410
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.652751205326394e-05,
+      "loss": 1.1263,
+      "step": 161420
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.65263641233642e-05,
+      "loss": 1.3232,
+      "step": 161430
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.652521619346445e-05,
+      "loss": 1.2192,
+      "step": 161440
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.65240682635647e-05,
+      "loss": 1.1377,
+      "step": 161450
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.6522920333664955e-05,
+      "loss": 1.0869,
+      "step": 161460
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.6521772403765205e-05,
+      "loss": 1.1105,
+      "step": 161470
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.652062447386546e-05,
+      "loss": 1.1607,
+      "step": 161480
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.651947654396571e-05,
+      "loss": 1.2231,
+      "step": 161490
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.651832861406596e-05,
+      "loss": 1.0771,
+      "step": 161500
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.651718068416622e-05,
+      "loss": 1.2736,
+      "step": 161510
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.6516032754266475e-05,
+      "loss": 1.1124,
+      "step": 161520
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.651488482436672e-05,
+      "loss": 1.0874,
+      "step": 161530
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.6513736894466974e-05,
+      "loss": 1.1593,
+      "step": 161540
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.651258896456723e-05,
+      "loss": 1.1946,
+      "step": 161550
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.6511441034667474e-05,
+      "loss": 1.161,
+      "step": 161560
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.651029310476773e-05,
+      "loss": 1.2483,
+      "step": 161570
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.650914517486799e-05,
+      "loss": 1.1695,
+      "step": 161580
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.650799724496824e-05,
+      "loss": 1.1848,
+      "step": 161590
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.650684931506849e-05,
+      "loss": 1.129,
+      "step": 161600
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 5.6505701385168744e-05,
+      "loss": 1.1665,
+      "step": 161610
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.6504553455268994e-05,
+      "loss": 1.165,
+      "step": 161620
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.650340552536925e-05,
+      "loss": 1.072,
+      "step": 161630
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.65022575954695e-05,
+      "loss": 1.2726,
+      "step": 161640
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.650110966556975e-05,
+      "loss": 1.1424,
+      "step": 161650
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.6499961735670007e-05,
+      "loss": 1.1647,
+      "step": 161660
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.649881380577026e-05,
+      "loss": 1.1444,
+      "step": 161670
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.6497665875870506e-05,
+      "loss": 1.1429,
+      "step": 161680
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.649651794597076e-05,
+      "loss": 1.2841,
+      "step": 161690
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.649537001607102e-05,
+      "loss": 1.2506,
+      "step": 161700
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.649422208617126e-05,
+      "loss": 1.2352,
+      "step": 161710
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.649307415627152e-05,
+      "loss": 1.1321,
+      "step": 161720
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.6491926226371776e-05,
+      "loss": 1.3486,
+      "step": 161730
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.6490778296472026e-05,
+      "loss": 1.2139,
+      "step": 161740
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.6489630366572276e-05,
+      "loss": 1.2567,
+      "step": 161750
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.648848243667253e-05,
+      "loss": 1.1591,
+      "step": 161760
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.648733450677278e-05,
+      "loss": 1.1198,
+      "step": 161770
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.648618657687304e-05,
+      "loss": 1.2793,
+      "step": 161780
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.648503864697328e-05,
+      "loss": 1.1623,
+      "step": 161790
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.648389071707354e-05,
+      "loss": 1.0922,
+      "step": 161800
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.6482742787173795e-05,
+      "loss": 1.0614,
+      "step": 161810
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.648159485727404e-05,
+      "loss": 1.1548,
+      "step": 161820
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 5.6480446927374295e-05,
+      "loss": 1.2967,
+      "step": 161830
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.647929899747455e-05,
+      "loss": 1.2796,
+      "step": 161840
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.6478151067574794e-05,
+      "loss": 1.1689,
+      "step": 161850
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.647700313767505e-05,
+      "loss": 1.2515,
+      "step": 161860
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.647585520777531e-05,
+      "loss": 1.0203,
+      "step": 161870
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.647470727787556e-05,
+      "loss": 1.1198,
+      "step": 161880
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.6473559347975814e-05,
+      "loss": 1.2336,
+      "step": 161890
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.6472411418076064e-05,
+      "loss": 1.1203,
+      "step": 161900
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.6471263488176314e-05,
+      "loss": 1.2496,
+      "step": 161910
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.647011555827657e-05,
+      "loss": 1.3018,
+      "step": 161920
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.646896762837683e-05,
+      "loss": 1.1491,
+      "step": 161930
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.646781969847707e-05,
+      "loss": 1.2742,
+      "step": 161940
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.646667176857733e-05,
+      "loss": 1.1099,
+      "step": 161950
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.6465523838677583e-05,
+      "loss": 1.1645,
+      "step": 161960
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.6464375908777827e-05,
+      "loss": 1.1605,
+      "step": 161970
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.646322797887808e-05,
+      "loss": 1.104,
+      "step": 161980
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.646208004897834e-05,
+      "loss": 1.1338,
+      "step": 161990
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.646093211907858e-05,
+      "loss": 1.1736,
+      "step": 162000
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.645978418917884e-05,
+      "loss": 1.1831,
+      "step": 162010
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.6458636259279096e-05,
+      "loss": 1.1427,
+      "step": 162020
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.6457488329379346e-05,
+      "loss": 1.2639,
+      "step": 162030
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 5.64563403994796e-05,
+      "loss": 1.1557,
+      "step": 162040
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.645519246957985e-05,
+      "loss": 1.3315,
+      "step": 162050
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.64540445396801e-05,
+      "loss": 1.2472,
+      "step": 162060
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.645289660978036e-05,
+      "loss": 1.0327,
+      "step": 162070
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.6451748679880616e-05,
+      "loss": 1.1841,
+      "step": 162080
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.645060074998086e-05,
+      "loss": 1.2582,
+      "step": 162090
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.6449452820081115e-05,
+      "loss": 1.1482,
+      "step": 162100
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.644830489018137e-05,
+      "loss": 1.2633,
+      "step": 162110
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.6447156960281615e-05,
+      "loss": 1.1036,
+      "step": 162120
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.644600903038187e-05,
+      "loss": 1.2067,
+      "step": 162130
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.644486110048213e-05,
+      "loss": 1.1668,
+      "step": 162140
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.644371317058237e-05,
+      "loss": 1.1048,
+      "step": 162150
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.644256524068263e-05,
+      "loss": 1.1928,
+      "step": 162160
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.6441417310782885e-05,
+      "loss": 1.1728,
+      "step": 162170
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.6440269380883134e-05,
+      "loss": 1.1931,
+      "step": 162180
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.643912145098339e-05,
+      "loss": 1.1997,
+      "step": 162190
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.643797352108364e-05,
+      "loss": 1.211,
+      "step": 162200
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.643682559118389e-05,
+      "loss": 1.2542,
+      "step": 162210
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.643567766128415e-05,
+      "loss": 1.2377,
+      "step": 162220
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.6434529731384404e-05,
+      "loss": 1.0828,
+      "step": 162230
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.643338180148465e-05,
+      "loss": 1.1566,
+      "step": 162240
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.6432233871584904e-05,
+      "loss": 1.1389,
+      "step": 162250
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 5.643108594168516e-05,
+      "loss": 1.1728,
+      "step": 162260
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.6429938011785404e-05,
+      "loss": 1.0705,
+      "step": 162270
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.642879008188566e-05,
+      "loss": 1.0919,
+      "step": 162280
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.642764215198592e-05,
+      "loss": 1.1435,
+      "step": 162290
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.642649422208616e-05,
+      "loss": 1.095,
+      "step": 162300
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.6425346292186416e-05,
+      "loss": 1.1393,
+      "step": 162310
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.642419836228667e-05,
+      "loss": 1.109,
+      "step": 162320
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.642305043238692e-05,
+      "loss": 1.3679,
+      "step": 162330
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.642190250248718e-05,
+      "loss": 1.0855,
+      "step": 162340
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.642075457258743e-05,
+      "loss": 1.208,
+      "step": 162350
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.641960664268768e-05,
+      "loss": 1.1942,
+      "step": 162360
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.6418458712787936e-05,
+      "loss": 1.1713,
+      "step": 162370
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.641731078288819e-05,
+      "loss": 1.0899,
+      "step": 162380
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.6416162852988436e-05,
+      "loss": 1.2271,
+      "step": 162390
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.641501492308869e-05,
+      "loss": 1.2056,
+      "step": 162400
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.641386699318895e-05,
+      "loss": 1.1405,
+      "step": 162410
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.641271906328919e-05,
+      "loss": 1.3005,
+      "step": 162420
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.641157113338945e-05,
+      "loss": 1.2307,
+      "step": 162430
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.6410423203489705e-05,
+      "loss": 1.14,
+      "step": 162440
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.640927527358995e-05,
+      "loss": 1.1793,
+      "step": 162450
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.6408127343690205e-05,
+      "loss": 1.2044,
+      "step": 162460
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.640697941379046e-05,
+      "loss": 1.0638,
+      "step": 162470
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 5.640583148389071e-05,
+      "loss": 1.1244,
+      "step": 162480
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.640468355399097e-05,
+      "loss": 1.2278,
+      "step": 162490
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.640353562409122e-05,
+      "loss": 1.1467,
+      "step": 162500
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.640238769419147e-05,
+      "loss": 1.1391,
+      "step": 162510
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.6401239764291724e-05,
+      "loss": 1.209,
+      "step": 162520
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.640009183439198e-05,
+      "loss": 1.1778,
+      "step": 162530
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.6398943904492224e-05,
+      "loss": 1.1476,
+      "step": 162540
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.639779597459248e-05,
+      "loss": 1.1954,
+      "step": 162550
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.639664804469274e-05,
+      "loss": 1.1313,
+      "step": 162560
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.639550011479298e-05,
+      "loss": 1.1421,
+      "step": 162570
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.639435218489324e-05,
+      "loss": 1.2588,
+      "step": 162580
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.6393204254993494e-05,
+      "loss": 1.2559,
+      "step": 162590
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.639205632509374e-05,
+      "loss": 1.0828,
+      "step": 162600
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.6390908395193993e-05,
+      "loss": 1.3671,
+      "step": 162610
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.638976046529425e-05,
+      "loss": 1.2394,
+      "step": 162620
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.63886125353945e-05,
+      "loss": 1.1396,
+      "step": 162630
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.6387464605494757e-05,
+      "loss": 1.2449,
+      "step": 162640
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.6386316675595006e-05,
+      "loss": 1.1409,
+      "step": 162650
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.6385168745695256e-05,
+      "loss": 1.1994,
+      "step": 162660
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.638402081579551e-05,
+      "loss": 1.2297,
+      "step": 162670
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.638287288589577e-05,
+      "loss": 1.0345,
+      "step": 162680
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.638172495599601e-05,
+      "loss": 1.047,
+      "step": 162690
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.638057702609627e-05,
+      "loss": 1.2159,
+      "step": 162700
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.6379429096196526e-05,
+      "loss": 1.2396,
+      "step": 162710
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.637828116629677e-05,
+      "loss": 1.12,
+      "step": 162720
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.6377133236397026e-05,
+      "loss": 1.2337,
+      "step": 162730
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.637598530649728e-05,
+      "loss": 1.2229,
+      "step": 162740
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.6374837376597525e-05,
+      "loss": 1.1754,
+      "step": 162750
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.637368944669778e-05,
+      "loss": 1.1174,
+      "step": 162760
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.637254151679804e-05,
+      "loss": 1.2309,
+      "step": 162770
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.637139358689829e-05,
+      "loss": 1.0935,
+      "step": 162780
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.6370245656998545e-05,
+      "loss": 1.2026,
+      "step": 162790
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.6369097727098795e-05,
+      "loss": 1.3185,
+      "step": 162800
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.6367949797199045e-05,
+      "loss": 1.1828,
+      "step": 162810
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.63668018672993e-05,
+      "loss": 1.1361,
+      "step": 162820
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.636565393739956e-05,
+      "loss": 1.1014,
+      "step": 162830
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.63645060074998e-05,
+      "loss": 1.1098,
+      "step": 162840
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.636335807760006e-05,
+      "loss": 1.1233,
+      "step": 162850
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.6362210147700314e-05,
+      "loss": 1.2046,
+      "step": 162860
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.636106221780056e-05,
+      "loss": 1.1506,
+      "step": 162870
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.6359914287900814e-05,
+      "loss": 1.0854,
+      "step": 162880
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.635876635800107e-05,
+      "loss": 1.0376,
+      "step": 162890
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.6357618428101314e-05,
+      "loss": 1.209,
+      "step": 162900
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 5.635647049820157e-05,
+      "loss": 1.1583,
+      "step": 162910
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.635532256830183e-05,
+      "loss": 1.1246,
+      "step": 162920
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.635417463840208e-05,
+      "loss": 1.2138,
+      "step": 162930
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.6353026708502333e-05,
+      "loss": 1.1953,
+      "step": 162940
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.635187877860258e-05,
+      "loss": 1.1665,
+      "step": 162950
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.635073084870283e-05,
+      "loss": 1.1079,
+      "step": 162960
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.634958291880309e-05,
+      "loss": 1.2004,
+      "step": 162970
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.6348434988903346e-05,
+      "loss": 1.2681,
+      "step": 162980
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.634728705900359e-05,
+      "loss": 1.2031,
+      "step": 162990
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.6346139129103846e-05,
+      "loss": 1.1386,
+      "step": 163000
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.63449911992041e-05,
+      "loss": 1.1437,
+      "step": 163010
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.6343843269304346e-05,
+      "loss": 1.1636,
+      "step": 163020
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.63426953394046e-05,
+      "loss": 1.2214,
+      "step": 163030
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.634154740950486e-05,
+      "loss": 1.235,
+      "step": 163040
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.63403994796051e-05,
+      "loss": 1.067,
+      "step": 163050
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.633925154970536e-05,
+      "loss": 1.2491,
+      "step": 163060
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.6338103619805615e-05,
+      "loss": 1.2458,
+      "step": 163070
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.6336955689905865e-05,
+      "loss": 1.1539,
+      "step": 163080
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.633580776000612e-05,
+      "loss": 1.224,
+      "step": 163090
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.633465983010637e-05,
+      "loss": 1.1989,
+      "step": 163100
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.633351190020662e-05,
+      "loss": 1.1585,
+      "step": 163110
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.633236397030688e-05,
+      "loss": 1.2192,
+      "step": 163120
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 5.6331216040407135e-05,
+      "loss": 1.119,
+      "step": 163130
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.633006811050738e-05,
+      "loss": 1.1656,
+      "step": 163140
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.6328920180607635e-05,
+      "loss": 1.1988,
+      "step": 163150
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.632777225070788e-05,
+      "loss": 1.1142,
+      "step": 163160
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.6326624320808134e-05,
+      "loss": 1.0829,
+      "step": 163170
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.632547639090839e-05,
+      "loss": 1.1803,
+      "step": 163180
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.632432846100864e-05,
+      "loss": 1.2765,
+      "step": 163190
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.632318053110889e-05,
+      "loss": 1.1758,
+      "step": 163200
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.632203260120915e-05,
+      "loss": 1.1036,
+      "step": 163210
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.63208846713094e-05,
+      "loss": 1.287,
+      "step": 163220
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.6319736741409654e-05,
+      "loss": 1.0767,
+      "step": 163230
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.631858881150991e-05,
+      "loss": 1.1745,
+      "step": 163240
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.6317440881610154e-05,
+      "loss": 1.2137,
+      "step": 163250
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.631629295171041e-05,
+      "loss": 1.088,
+      "step": 163260
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.631514502181067e-05,
+      "loss": 1.1511,
+      "step": 163270
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.631399709191091e-05,
+      "loss": 1.1362,
+      "step": 163280
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.6312849162011166e-05,
+      "loss": 1.0666,
+      "step": 163290
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.631170123211142e-05,
+      "loss": 1.1711,
+      "step": 163300
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.6310553302211666e-05,
+      "loss": 1.1723,
+      "step": 163310
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.630940537231192e-05,
+      "loss": 1.2115,
+      "step": 163320
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.630825744241218e-05,
+      "loss": 1.2692,
+      "step": 163330
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.630710951251243e-05,
+      "loss": 1.3061,
+      "step": 163340
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 5.630596158261268e-05,
+      "loss": 1.1595,
+      "step": 163350
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.6304813652712936e-05,
+      "loss": 1.1249,
+      "step": 163360
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.6303665722813186e-05,
+      "loss": 1.1601,
+      "step": 163370
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.630251779291344e-05,
+      "loss": 1.1192,
+      "step": 163380
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.63013698630137e-05,
+      "loss": 1.1556,
+      "step": 163390
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.630022193311394e-05,
+      "loss": 1.2234,
+      "step": 163400
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.62990740032142e-05,
+      "loss": 1.1258,
+      "step": 163410
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.6297926073314455e-05,
+      "loss": 1.1916,
+      "step": 163420
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.62967781434147e-05,
+      "loss": 1.2503,
+      "step": 163430
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.6295630213514955e-05,
+      "loss": 1.1673,
+      "step": 163440
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.629448228361521e-05,
+      "loss": 1.1655,
+      "step": 163450
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.6293334353715455e-05,
+      "loss": 1.1301,
+      "step": 163460
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.629218642381571e-05,
+      "loss": 1.2285,
+      "step": 163470
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.629103849391597e-05,
+      "loss": 1.2212,
+      "step": 163480
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.628989056401622e-05,
+      "loss": 1.1858,
+      "step": 163490
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.628874263411647e-05,
+      "loss": 1.1548,
+      "step": 163500
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.6287594704216724e-05,
+      "loss": 1.0972,
+      "step": 163510
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.6286446774316974e-05,
+      "loss": 1.1202,
+      "step": 163520
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.628529884441723e-05,
+      "loss": 1.1809,
+      "step": 163530
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.628415091451749e-05,
+      "loss": 1.2993,
+      "step": 163540
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.628300298461773e-05,
+      "loss": 1.1812,
+      "step": 163550
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.628185505471799e-05,
+      "loss": 1.0883,
+      "step": 163560
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 5.6280707124818244e-05,
+      "loss": 1.1366,
+      "step": 163570
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 5.627955919491849e-05,
+      "loss": 1.3412,
+      "step": 163580
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 5.6278411265018743e-05,
+      "loss": 1.2345,
+      "step": 163590
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 5.6277263335119e-05,
+      "loss": 1.2285,
+      "step": 163600
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 5.627611540521924e-05,
+      "loss": 1.2794,
+      "step": 163610
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 5.62749674753195e-05,
+      "loss": 1.1522,
+      "step": 163620
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 5.6273819545419756e-05,
+      "loss": 1.1622,
+      "step": 163630
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 5.6272671615520006e-05,
+      "loss": 1.2202,
+      "step": 163640
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 5.6271523685620256e-05,
+      "loss": 1.1718,
+      "step": 163650
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 5.627037575572051e-05,
+      "loss": 1.0853,
+      "step": 163660
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 5.626922782582076e-05,
+      "loss": 0.9936,
+      "step": 163670
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 5.626807989592102e-05,
+      "loss": 1.1679,
+      "step": 163680
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 5.6266931966021276e-05,
+      "loss": 1.3627,
+      "step": 163690
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 5.626578403612152e-05,
+      "loss": 1.1948,
+      "step": 163700
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 5.6264636106221776e-05,
+      "loss": 1.2174,
+      "step": 163710
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 5.626348817632203e-05,
+      "loss": 1.1665,
+      "step": 163720
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 5.6262340246422275e-05,
+      "loss": 1.2283,
+      "step": 163730
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 5.626119231652253e-05,
+      "loss": 1.1988,
+      "step": 163740
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 5.626004438662279e-05,
+      "loss": 1.1722,
+      "step": 163750
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 5.625889645672303e-05,
+      "loss": 1.1319,
+      "step": 163760
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 5.625774852682329e-05,
+      "loss": 1.0864,
+      "step": 163770
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 5.6256600596923545e-05,
+      "loss": 1.1638,
+      "step": 163780
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 5.6255452667023795e-05,
+      "loss": 1.0513,
+      "step": 163790
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 5.6254304737124045e-05,
+      "loss": 1.1784,
+      "step": 163800
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 5.62531568072243e-05,
+      "loss": 1.3696,
+      "step": 163810
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 5.625200887732455e-05,
+      "loss": 1.1164,
+      "step": 163820
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 5.625086094742481e-05,
+      "loss": 1.1551,
+      "step": 163830
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 5.6249713017525064e-05,
+      "loss": 1.1281,
+      "step": 163840
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 5.624856508762531e-05,
+      "loss": 1.113,
+      "step": 163850
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 5.6247417157725564e-05,
+      "loss": 1.2381,
+      "step": 163860
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 5.624626922782582e-05,
+      "loss": 1.1547,
+      "step": 163870
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 5.6245121297926064e-05,
+      "loss": 1.175,
+      "step": 163880
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 5.624397336802632e-05,
+      "loss": 1.1576,
+      "step": 163890
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 5.624282543812658e-05,
+      "loss": 1.1525,
+      "step": 163900
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 5.624167750822682e-05,
+      "loss": 1.1026,
+      "step": 163910
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 5.624052957832708e-05,
+      "loss": 1.2069,
+      "step": 163920
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 5.623938164842733e-05,
+      "loss": 1.3041,
+      "step": 163930
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 5.623823371852758e-05,
+      "loss": 1.1315,
+      "step": 163940
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 5.623708578862784e-05,
+      "loss": 1.2852,
+      "step": 163950
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 5.623593785872809e-05,
+      "loss": 1.0271,
+      "step": 163960
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 5.623478992882834e-05,
+      "loss": 1.1641,
+      "step": 163970
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 5.6233641998928596e-05,
+      "loss": 1.246,
+      "step": 163980
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 5.623249406902885e-05,
+      "loss": 1.1225,
+      "step": 163990
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 5.6231346139129096e-05,
+      "loss": 1.2966,
+      "step": 164000
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.623019820922935e-05,
+      "loss": 1.2852,
+      "step": 164010
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.622905027932961e-05,
+      "loss": 1.1472,
+      "step": 164020
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.622790234942985e-05,
+      "loss": 1.159,
+      "step": 164030
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.622675441953011e-05,
+      "loss": 1.1996,
+      "step": 164040
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.6225606489630365e-05,
+      "loss": 1.1415,
+      "step": 164050
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.622445855973061e-05,
+      "loss": 1.0944,
+      "step": 164060
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.6223310629830865e-05,
+      "loss": 1.3379,
+      "step": 164070
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.622216269993112e-05,
+      "loss": 1.2316,
+      "step": 164080
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.622101477003137e-05,
+      "loss": 1.1266,
+      "step": 164090
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.621986684013163e-05,
+      "loss": 1.0799,
+      "step": 164100
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.621871891023188e-05,
+      "loss": 1.2449,
+      "step": 164110
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.621757098033213e-05,
+      "loss": 1.0869,
+      "step": 164120
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.6216423050432385e-05,
+      "loss": 1.1838,
+      "step": 164130
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.621527512053264e-05,
+      "loss": 1.1349,
+      "step": 164140
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.6214127190632884e-05,
+      "loss": 1.2103,
+      "step": 164150
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.621297926073314e-05,
+      "loss": 1.1725,
+      "step": 164160
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.62118313308334e-05,
+      "loss": 1.3009,
+      "step": 164170
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.621068340093364e-05,
+      "loss": 1.1449,
+      "step": 164180
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.62095354710339e-05,
+      "loss": 1.194,
+      "step": 164190
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.6208387541134154e-05,
+      "loss": 1.2236,
+      "step": 164200
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.62072396112344e-05,
+      "loss": 1.2719,
+      "step": 164210
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.6206091681334654e-05,
+      "loss": 1.1555,
+      "step": 164220
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.620494375143491e-05,
+      "loss": 1.0615,
+      "step": 164230
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.620379582153516e-05,
+      "loss": 1.1683,
+      "step": 164240
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.620264789163542e-05,
+      "loss": 1.1254,
+      "step": 164250
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.620149996173567e-05,
+      "loss": 1.2624,
+      "step": 164260
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.6200352031835916e-05,
+      "loss": 1.2714,
+      "step": 164270
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.619920410193617e-05,
+      "loss": 1.0454,
+      "step": 164280
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.619805617203643e-05,
+      "loss": 1.1869,
+      "step": 164290
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.619690824213667e-05,
+      "loss": 1.2576,
+      "step": 164300
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.619576031223693e-05,
+      "loss": 1.2264,
+      "step": 164310
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.6194612382337186e-05,
+      "loss": 1.1339,
+      "step": 164320
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.619346445243743e-05,
+      "loss": 1.206,
+      "step": 164330
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.6192316522537686e-05,
+      "loss": 1.1716,
+      "step": 164340
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.619116859263794e-05,
+      "loss": 1.1554,
+      "step": 164350
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.6190020662738185e-05,
+      "loss": 1.1303,
+      "step": 164360
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.618887273283844e-05,
+      "loss": 1.2265,
+      "step": 164370
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.61877248029387e-05,
+      "loss": 1.0684,
+      "step": 164380
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.618657687303895e-05,
+      "loss": 1.0536,
+      "step": 164390
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.6185428943139205e-05,
+      "loss": 1.2672,
+      "step": 164400
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.6184281013239455e-05,
+      "loss": 1.1163,
+      "step": 164410
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.6183133083339705e-05,
+      "loss": 1.2015,
+      "step": 164420
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.618198515343996e-05,
+      "loss": 1.2272,
+      "step": 164430
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.618083722354022e-05,
+      "loss": 1.1447,
+      "step": 164440
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.617968929364046e-05,
+      "loss": 1.0225,
+      "step": 164450
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.617854136374072e-05,
+      "loss": 1.171,
+      "step": 164460
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.6177393433840975e-05,
+      "loss": 1.2058,
+      "step": 164470
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.617624550394122e-05,
+      "loss": 1.1338,
+      "step": 164480
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.6175097574041474e-05,
+      "loss": 0.9637,
+      "step": 164490
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.617394964414173e-05,
+      "loss": 1.1212,
+      "step": 164500
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.6172801714241974e-05,
+      "loss": 1.2047,
+      "step": 164510
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.617165378434223e-05,
+      "loss": 1.1863,
+      "step": 164520
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.617050585444248e-05,
+      "loss": 1.1269,
+      "step": 164530
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.616935792454274e-05,
+      "loss": 1.1232,
+      "step": 164540
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.6168209994642994e-05,
+      "loss": 1.107,
+      "step": 164550
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.616706206474324e-05,
+      "loss": 1.1383,
+      "step": 164560
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.6165914134843493e-05,
+      "loss": 1.1834,
+      "step": 164570
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.616476620494375e-05,
+      "loss": 1.1815,
+      "step": 164580
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.616361827504399e-05,
+      "loss": 1.2153,
+      "step": 164590
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.616247034514425e-05,
+      "loss": 1.2268,
+      "step": 164600
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.6161322415244506e-05,
+      "loss": 1.0497,
+      "step": 164610
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.616017448534475e-05,
+      "loss": 1.2127,
+      "step": 164620
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.6159026555445006e-05,
+      "loss": 1.1083,
+      "step": 164630
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.615787862554526e-05,
+      "loss": 1.078,
+      "step": 164640
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.615673069564551e-05,
+      "loss": 1.0775,
+      "step": 164650
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.615558276574576e-05,
+      "loss": 1.0564,
+      "step": 164660
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.615443483584602e-05,
+      "loss": 1.1857,
+      "step": 164670
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.615328690594627e-05,
+      "loss": 1.0549,
+      "step": 164680
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.6152138976046526e-05,
+      "loss": 1.1652,
+      "step": 164690
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.615099104614678e-05,
+      "loss": 1.0828,
+      "step": 164700
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.6149843116247025e-05,
+      "loss": 1.1463,
+      "step": 164710
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.614869518634728e-05,
+      "loss": 1.3028,
+      "step": 164720
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.614754725644754e-05,
+      "loss": 1.1996,
+      "step": 164730
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.614639932654778e-05,
+      "loss": 1.0686,
+      "step": 164740
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.614525139664804e-05,
+      "loss": 1.2416,
+      "step": 164750
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.6144103466748295e-05,
+      "loss": 1.2436,
+      "step": 164760
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.614295553684854e-05,
+      "loss": 1.211,
+      "step": 164770
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.6141807606948795e-05,
+      "loss": 1.1768,
+      "step": 164780
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.614065967704905e-05,
+      "loss": 1.1612,
+      "step": 164790
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.61395117471493e-05,
+      "loss": 1.1177,
+      "step": 164800
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.613836381724955e-05,
+      "loss": 1.1642,
+      "step": 164810
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.613721588734981e-05,
+      "loss": 1.1139,
+      "step": 164820
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.613606795745006e-05,
+      "loss": 1.1037,
+      "step": 164830
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.6134920027550314e-05,
+      "loss": 1.0519,
+      "step": 164840
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.613377209765057e-05,
+      "loss": 1.2081,
+      "step": 164850
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.6132624167750814e-05,
+      "loss": 1.072,
+      "step": 164860
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.613147623785107e-05,
+      "loss": 1.12,
+      "step": 164870
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.613032830795133e-05,
+      "loss": 1.1476,
+      "step": 164880
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.612918037805157e-05,
+      "loss": 1.2695,
+      "step": 164890
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.612803244815183e-05,
+      "loss": 1.2597,
+      "step": 164900
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.612688451825208e-05,
+      "loss": 1.1064,
+      "step": 164910
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.6125736588352326e-05,
+      "loss": 1.1906,
+      "step": 164920
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.612458865845258e-05,
+      "loss": 1.1472,
+      "step": 164930
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.612344072855284e-05,
+      "loss": 1.3103,
+      "step": 164940
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.612229279865309e-05,
+      "loss": 1.2321,
+      "step": 164950
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.612114486875334e-05,
+      "loss": 1.3162,
+      "step": 164960
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.6119996938853596e-05,
+      "loss": 1.1725,
+      "step": 164970
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.6118849008953846e-05,
+      "loss": 1.2843,
+      "step": 164980
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.61177010790541e-05,
+      "loss": 1.1059,
+      "step": 164990
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.611655314915436e-05,
+      "loss": 1.1566,
+      "step": 165000
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.61154052192546e-05,
+      "loss": 1.169,
+      "step": 165010
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.611425728935486e-05,
+      "loss": 1.1733,
+      "step": 165020
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.6113109359455115e-05,
+      "loss": 1.2255,
+      "step": 165030
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.611196142955536e-05,
+      "loss": 1.1767,
+      "step": 165040
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.6110813499655615e-05,
+      "loss": 1.2534,
+      "step": 165050
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.610966556975587e-05,
+      "loss": 1.2745,
+      "step": 165060
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.6108517639856115e-05,
+      "loss": 1.0859,
+      "step": 165070
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.610736970995637e-05,
+      "loss": 1.0756,
+      "step": 165080
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.610622178005663e-05,
+      "loss": 1.1501,
+      "step": 165090
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.610507385015688e-05,
+      "loss": 1.2141,
+      "step": 165100
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.610392592025713e-05,
+      "loss": 1.2478,
+      "step": 165110
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.6102777990357384e-05,
+      "loss": 1.181,
+      "step": 165120
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.6101630060457634e-05,
+      "loss": 1.1558,
+      "step": 165130
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.610048213055789e-05,
+      "loss": 1.1243,
+      "step": 165140
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.609933420065815e-05,
+      "loss": 1.2693,
+      "step": 165150
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.609818627075839e-05,
+      "loss": 1.1782,
+      "step": 165160
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.609703834085865e-05,
+      "loss": 1.1008,
+      "step": 165170
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.6095890410958904e-05,
+      "loss": 1.1865,
+      "step": 165180
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.609474248105915e-05,
+      "loss": 1.2568,
+      "step": 165190
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.6093594551159404e-05,
+      "loss": 1.09,
+      "step": 165200
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.609244662125966e-05,
+      "loss": 1.2632,
+      "step": 165210
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.60912986913599e-05,
+      "loss": 1.1645,
+      "step": 165220
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.609015076146016e-05,
+      "loss": 1.1671,
+      "step": 165230
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.6089002831560417e-05,
+      "loss": 1.1051,
+      "step": 165240
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.6087854901660666e-05,
+      "loss": 1.2062,
+      "step": 165250
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.6086706971760916e-05,
+      "loss": 1.1927,
+      "step": 165260
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.608555904186117e-05,
+      "loss": 1.2198,
+      "step": 165270
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.608441111196142e-05,
+      "loss": 1.1306,
+      "step": 165280
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.608326318206168e-05,
+      "loss": 1.2606,
+      "step": 165290
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.6082115252161936e-05,
+      "loss": 1.2786,
+      "step": 165300
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.608096732226218e-05,
+      "loss": 1.2224,
+      "step": 165310
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.6079819392362436e-05,
+      "loss": 1.1157,
+      "step": 165320
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.607867146246269e-05,
+      "loss": 1.2054,
+      "step": 165330
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.6077523532562935e-05,
+      "loss": 1.1914,
+      "step": 165340
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.607637560266319e-05,
+      "loss": 1.1973,
+      "step": 165350
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.607522767276345e-05,
+      "loss": 1.1758,
+      "step": 165360
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.607407974286369e-05,
+      "loss": 1.4131,
+      "step": 165370
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.607293181296395e-05,
+      "loss": 1.2639,
+      "step": 165380
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.6071783883064205e-05,
+      "loss": 1.1207,
+      "step": 165390
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.6070635953164455e-05,
+      "loss": 1.1873,
+      "step": 165400
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.6069488023264705e-05,
+      "loss": 1.3134,
+      "step": 165410
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.606834009336496e-05,
+      "loss": 1.2844,
+      "step": 165420
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.606719216346521e-05,
+      "loss": 1.1952,
+      "step": 165430
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.606604423356547e-05,
+      "loss": 1.0882,
+      "step": 165440
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.6064896303665725e-05,
+      "loss": 1.2003,
+      "step": 165450
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.606374837376597e-05,
+      "loss": 1.0767,
+      "step": 165460
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.6062600443866224e-05,
+      "loss": 1.1907,
+      "step": 165470
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.606145251396648e-05,
+      "loss": 1.0228,
+      "step": 165480
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.6060304584066724e-05,
+      "loss": 1.2328,
+      "step": 165490
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.605915665416698e-05,
+      "loss": 1.2133,
+      "step": 165500
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.605800872426724e-05,
+      "loss": 1.2015,
+      "step": 165510
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.605686079436748e-05,
+      "loss": 1.1519,
+      "step": 165520
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.605571286446774e-05,
+      "loss": 1.2811,
+      "step": 165530
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.6054564934567994e-05,
+      "loss": 1.1476,
+      "step": 165540
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.6053417004668243e-05,
+      "loss": 1.3261,
+      "step": 165550
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.605226907476849e-05,
+      "loss": 1.1947,
+      "step": 165560
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.605112114486875e-05,
+      "loss": 1.2839,
+      "step": 165570
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.6049973214969e-05,
+      "loss": 1.0665,
+      "step": 165580
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.6048825285069256e-05,
+      "loss": 1.0964,
+      "step": 165590
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.604767735516951e-05,
+      "loss": 1.2123,
+      "step": 165600
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.6046529425269756e-05,
+      "loss": 1.137,
+      "step": 165610
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.604538149537001e-05,
+      "loss": 1.1624,
+      "step": 165620
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.604423356547027e-05,
+      "loss": 1.2542,
+      "step": 165630
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.604308563557051e-05,
+      "loss": 1.1872,
+      "step": 165640
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.604193770567077e-05,
+      "loss": 1.2034,
+      "step": 165650
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.6040789775771026e-05,
+      "loss": 1.1391,
+      "step": 165660
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.603964184587127e-05,
+      "loss": 1.2711,
+      "step": 165670
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.6038493915971525e-05,
+      "loss": 1.251,
+      "step": 165680
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.603734598607178e-05,
+      "loss": 1.1548,
+      "step": 165690
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.603619805617203e-05,
+      "loss": 1.2864,
+      "step": 165700
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.603505012627228e-05,
+      "loss": 1.1505,
+      "step": 165710
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.603390219637254e-05,
+      "loss": 1.1623,
+      "step": 165720
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.603275426647279e-05,
+      "loss": 1.1875,
+      "step": 165730
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.6031606336573045e-05,
+      "loss": 1.2054,
+      "step": 165740
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.60304584066733e-05,
+      "loss": 1.1654,
+      "step": 165750
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 5.6029310476773545e-05,
+      "loss": 1.3066,
+      "step": 165760
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 5.60281625468738e-05,
+      "loss": 1.1494,
+      "step": 165770
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 5.602701461697406e-05,
+      "loss": 1.1384,
+      "step": 165780
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 5.60258666870743e-05,
+      "loss": 1.1867,
+      "step": 165790
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 5.602471875717456e-05,
+      "loss": 1.1394,
+      "step": 165800
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 5.6023570827274814e-05,
+      "loss": 1.2374,
+      "step": 165810
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 5.602242289737506e-05,
+      "loss": 1.1563,
+      "step": 165820
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 5.6021274967475314e-05,
+      "loss": 1.1437,
+      "step": 165830
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 5.602012703757557e-05,
+      "loss": 1.1812,
+      "step": 165840
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 5.601897910767582e-05,
+      "loss": 1.3056,
+      "step": 165850
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 5.601783117777607e-05,
+      "loss": 1.1447,
+      "step": 165860
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 5.601668324787633e-05,
+      "loss": 1.1235,
+      "step": 165870
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 5.601553531797658e-05,
+      "loss": 1.1863,
+      "step": 165880
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 5.601438738807683e-05,
+      "loss": 1.1605,
+      "step": 165890
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 5.6013239458177076e-05,
+      "loss": 1.2628,
+      "step": 165900
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 5.601209152827733e-05,
+      "loss": 1.2129,
+      "step": 165910
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 5.601094359837759e-05,
+      "loss": 1.2685,
+      "step": 165920
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 5.600979566847783e-05,
+      "loss": 1.1422,
+      "step": 165930
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 5.600864773857809e-05,
+      "loss": 1.1402,
+      "step": 165940
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 5.6007499808678346e-05,
+      "loss": 1.1338,
+      "step": 165950
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 5.6006351878778596e-05,
+      "loss": 1.3708,
+      "step": 165960
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 5.6005203948878846e-05,
+      "loss": 1.0397,
+      "step": 165970
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 5.60040560189791e-05,
+      "loss": 1.1335,
+      "step": 165980
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 5.600290808907935e-05,
+      "loss": 1.1011,
+      "step": 165990
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 5.600176015917961e-05,
+      "loss": 1.244,
+      "step": 166000
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 5.600061222927986e-05,
+      "loss": 1.2031,
+      "step": 166010
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 5.599946429938011e-05,
+      "loss": 1.1106,
+      "step": 166020
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 5.5998316369480365e-05,
+      "loss": 1.2073,
+      "step": 166030
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 5.599716843958062e-05,
+      "loss": 1.2777,
+      "step": 166040
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 5.5996020509680865e-05,
+      "loss": 1.0991,
+      "step": 166050
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 5.599487257978112e-05,
+      "loss": 1.2431,
+      "step": 166060
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 5.599372464988138e-05,
+      "loss": 1.1021,
+      "step": 166070
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 5.599257671998162e-05,
+      "loss": 1.2396,
+      "step": 166080
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 5.599142879008188e-05,
+      "loss": 1.1312,
+      "step": 166090
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 5.5990280860182134e-05,
+      "loss": 1.4264,
+      "step": 166100
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 5.5989132930282384e-05,
+      "loss": 1.2642,
+      "step": 166110
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 5.5987985000382634e-05,
+      "loss": 1.0547,
+      "step": 166120
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 5.598683707048289e-05,
+      "loss": 1.1091,
+      "step": 166130
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 5.598568914058314e-05,
+      "loss": 1.1481,
+      "step": 166140
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 5.59845412106834e-05,
+      "loss": 1.0646,
+      "step": 166150
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 5.598339328078365e-05,
+      "loss": 1.2793,
+      "step": 166160
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 5.59822453508839e-05,
+      "loss": 1.1906,
+      "step": 166170
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 5.5981097420984154e-05,
+      "loss": 1.3665,
+      "step": 166180
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 5.597994949108441e-05,
+      "loss": 1.0905,
+      "step": 166190
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 5.597880156118465e-05,
+      "loss": 1.0887,
+      "step": 166200
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 5.597765363128491e-05,
+      "loss": 1.0573,
+      "step": 166210
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 5.5976505701385167e-05,
+      "loss": 1.1331,
+      "step": 166220
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 5.597535777148541e-05,
+      "loss": 1.1119,
+      "step": 166230
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 5.5974209841585666e-05,
+      "loss": 1.1977,
+      "step": 166240
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 5.597306191168592e-05,
+      "loss": 1.0839,
+      "step": 166250
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 5.597191398178617e-05,
+      "loss": 1.2186,
+      "step": 166260
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 5.597076605188642e-05,
+      "loss": 1.1983,
+      "step": 166270
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 5.596961812198668e-05,
+      "loss": 1.1928,
+      "step": 166280
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 5.596847019208693e-05,
+      "loss": 1.1804,
+      "step": 166290
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 5.5967322262187186e-05,
+      "loss": 1.0908,
+      "step": 166300
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 5.5966174332287436e-05,
+      "loss": 1.3712,
+      "step": 166310
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 5.5965026402387685e-05,
+      "loss": 1.2182,
+      "step": 166320
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 5.596387847248794e-05,
+      "loss": 1.1533,
+      "step": 166330
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 5.59627305425882e-05,
+      "loss": 1.1962,
+      "step": 166340
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 5.596158261268844e-05,
+      "loss": 1.0456,
+      "step": 166350
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 5.59604346827887e-05,
+      "loss": 1.2488,
+      "step": 166360
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 5.5959286752888955e-05,
+      "loss": 1.1286,
+      "step": 166370
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 5.59581388229892e-05,
+      "loss": 1.0505,
+      "step": 166380
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 5.5956990893089455e-05,
+      "loss": 1.2086,
+      "step": 166390
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 5.595584296318971e-05,
+      "loss": 1.1619,
+      "step": 166400
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 5.595469503328996e-05,
+      "loss": 1.2181,
+      "step": 166410
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 5.595354710339021e-05,
+      "loss": 1.2324,
+      "step": 166420
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 5.595239917349047e-05,
+      "loss": 1.2838,
+      "step": 166430
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 5.595125124359072e-05,
+      "loss": 1.2406,
+      "step": 166440
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 5.5950103313690974e-05,
+      "loss": 1.2372,
+      "step": 166450
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 5.5948955383791224e-05,
+      "loss": 1.1987,
+      "step": 166460
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 5.5947807453891474e-05,
+      "loss": 1.264,
+      "step": 166470
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 5.594665952399173e-05,
+      "loss": 1.1866,
+      "step": 166480
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 5.594551159409199e-05,
+      "loss": 1.172,
+      "step": 166490
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 5.594436366419223e-05,
+      "loss": 1.1339,
+      "step": 166500
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 5.594321573429249e-05,
+      "loss": 1.132,
+      "step": 166510
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 5.5942067804392744e-05,
+      "loss": 1.2005,
+      "step": 166520
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 5.594091987449299e-05,
+      "loss": 1.2804,
+      "step": 166530
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 5.593977194459324e-05,
+      "loss": 1.2233,
+      "step": 166540
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 5.59386240146935e-05,
+      "loss": 1.1705,
+      "step": 166550
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 5.593747608479375e-05,
+      "loss": 1.1314,
+      "step": 166560
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 5.5936328154894e-05,
+      "loss": 1.1004,
+      "step": 166570
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 5.5935180224994256e-05,
+      "loss": 1.2138,
+      "step": 166580
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 5.5934032295094506e-05,
+      "loss": 1.2916,
+      "step": 166590
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 5.593288436519476e-05,
+      "loss": 1.0926,
+      "step": 166600
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 5.593173643529501e-05,
+      "loss": 1.1941,
+      "step": 166610
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 5.593058850539526e-05,
+      "loss": 1.128,
+      "step": 166620
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 5.592944057549552e-05,
+      "loss": 1.2882,
+      "step": 166630
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 5.5928292645595776e-05,
+      "loss": 1.1402,
+      "step": 166640
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 5.592714471569602e-05,
+      "loss": 1.2187,
+      "step": 166650
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 5.5925996785796275e-05,
+      "loss": 1.1684,
+      "step": 166660
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 5.592484885589653e-05,
+      "loss": 1.0569,
+      "step": 166670
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 5.5923700925996775e-05,
+      "loss": 1.2655,
+      "step": 166680
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 5.592255299609703e-05,
+      "loss": 1.1196,
+      "step": 166690
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 5.592140506619729e-05,
+      "loss": 1.2069,
+      "step": 166700
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 5.592025713629754e-05,
+      "loss": 1.1906,
+      "step": 166710
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 5.591910920639779e-05,
+      "loss": 1.1857,
+      "step": 166720
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 5.5917961276498045e-05,
+      "loss": 1.2661,
+      "step": 166730
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 5.5916813346598295e-05,
+      "loss": 1.1182,
+      "step": 166740
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 5.591566541669855e-05,
+      "loss": 1.1136,
+      "step": 166750
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 5.59145174867988e-05,
+      "loss": 1.1663,
+      "step": 166760
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 5.591336955689905e-05,
+      "loss": 1.1103,
+      "step": 166770
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 5.591222162699931e-05,
+      "loss": 1.1047,
+      "step": 166780
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 5.5911073697099564e-05,
+      "loss": 1.1165,
+      "step": 166790
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 5.590992576719981e-05,
+      "loss": 1.1898,
+      "step": 166800
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 5.5908777837300064e-05,
+      "loss": 1.1749,
+      "step": 166810
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 5.590762990740032e-05,
+      "loss": 1.1203,
+      "step": 166820
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 5.5906481977500564e-05,
+      "loss": 1.109,
+      "step": 166830
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 5.590533404760082e-05,
+      "loss": 1.0977,
+      "step": 166840
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 5.590418611770108e-05,
+      "loss": 1.1826,
+      "step": 166850
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 5.590303818780133e-05,
+      "loss": 1.1476,
+      "step": 166860
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 5.5901890257901577e-05,
+      "loss": 1.1775,
+      "step": 166870
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 5.590074232800183e-05,
+      "loss": 1.2112,
+      "step": 166880
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 5.589959439810208e-05,
+      "loss": 1.1549,
+      "step": 166890
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 5.589844646820234e-05,
+      "loss": 1.2128,
+      "step": 166900
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 5.589729853830259e-05,
+      "loss": 1.1483,
+      "step": 166910
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 5.589615060840284e-05,
+      "loss": 1.1795,
+      "step": 166920
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 5.5895002678503096e-05,
+      "loss": 1.1995,
+      "step": 166930
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 5.589385474860335e-05,
+      "loss": 1.196,
+      "step": 166940
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 5.5892706818703596e-05,
+      "loss": 1.1349,
+      "step": 166950
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 5.589155888880385e-05,
+      "loss": 1.1976,
+      "step": 166960
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 5.589041095890411e-05,
+      "loss": 1.2562,
+      "step": 166970
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 5.588926302900435e-05,
+      "loss": 1.2735,
+      "step": 166980
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 5.588811509910461e-05,
+      "loss": 1.2235,
+      "step": 166990
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 5.5886967169204865e-05,
+      "loss": 1.1405,
+      "step": 167000
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 5.5885819239305115e-05,
+      "loss": 1.1817,
+      "step": 167010
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 5.5884671309405365e-05,
+      "loss": 1.2156,
+      "step": 167020
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 5.588352337950562e-05,
+      "loss": 1.1186,
+      "step": 167030
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 5.588237544960587e-05,
+      "loss": 1.1811,
+      "step": 167040
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 5.588122751970613e-05,
+      "loss": 1.1839,
+      "step": 167050
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 5.588007958980638e-05,
+      "loss": 1.1472,
+      "step": 167060
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 5.587893165990663e-05,
+      "loss": 1.2202,
+      "step": 167070
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 5.5877783730006884e-05,
+      "loss": 1.1532,
+      "step": 167080
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 5.587663580010714e-05,
+      "loss": 1.1105,
+      "step": 167090
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 5.5875487870207384e-05,
+      "loss": 1.2223,
+      "step": 167100
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 5.587433994030764e-05,
+      "loss": 1.1637,
+      "step": 167110
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 5.58731920104079e-05,
+      "loss": 1.1016,
+      "step": 167120
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 5.587204408050814e-05,
+      "loss": 1.1918,
+      "step": 167130
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 5.58708961506084e-05,
+      "loss": 1.1473,
+      "step": 167140
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 5.5869748220708654e-05,
+      "loss": 1.1465,
+      "step": 167150
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 5.5868600290808904e-05,
+      "loss": 1.1243,
+      "step": 167160
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 5.5867452360909153e-05,
+      "loss": 1.2398,
+      "step": 167170
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 5.586630443100941e-05,
+      "loss": 1.211,
+      "step": 167180
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 5.586515650110966e-05,
+      "loss": 1.2858,
+      "step": 167190
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 5.5864008571209917e-05,
+      "loss": 1.1587,
+      "step": 167200
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 5.5862860641310166e-05,
+      "loss": 1.3532,
+      "step": 167210
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 5.5861712711410416e-05,
+      "loss": 1.2767,
+      "step": 167220
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 5.586056478151067e-05,
+      "loss": 1.2294,
+      "step": 167230
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 5.585941685161093e-05,
+      "loss": 1.1792,
+      "step": 167240
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 5.585826892171117e-05,
+      "loss": 1.189,
+      "step": 167250
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 5.585712099181143e-05,
+      "loss": 1.1527,
+      "step": 167260
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 5.585597306191168e-05,
+      "loss": 1.227,
+      "step": 167270
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 5.585482513201193e-05,
+      "loss": 1.2648,
+      "step": 167280
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 5.5853677202112186e-05,
+      "loss": 1.1953,
+      "step": 167290
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 5.5852529272212435e-05,
+      "loss": 1.2251,
+      "step": 167300
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 5.585138134231269e-05,
+      "loss": 1.1322,
+      "step": 167310
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 5.585023341241294e-05,
+      "loss": 1.1853,
+      "step": 167320
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 5.584908548251319e-05,
+      "loss": 1.1168,
+      "step": 167330
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 5.584793755261345e-05,
+      "loss": 1.0177,
+      "step": 167340
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 5.5846789622713705e-05,
+      "loss": 1.2724,
+      "step": 167350
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 5.584564169281395e-05,
+      "loss": 1.188,
+      "step": 167360
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 5.5844493762914205e-05,
+      "loss": 1.1984,
+      "step": 167370
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 5.584334583301446e-05,
+      "loss": 1.0795,
+      "step": 167380
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 5.5842197903114704e-05,
+      "loss": 1.2152,
+      "step": 167390
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 5.584104997321496e-05,
+      "loss": 1.1566,
+      "step": 167400
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 5.583990204331522e-05,
+      "loss": 1.1233,
+      "step": 167410
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 5.583875411341547e-05,
+      "loss": 1.2063,
+      "step": 167420
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 5.583760618351572e-05,
+      "loss": 1.1607,
+      "step": 167430
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 5.5836458253615974e-05,
+      "loss": 1.1525,
+      "step": 167440
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 5.5835310323716224e-05,
+      "loss": 1.178,
+      "step": 167450
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 5.583416239381648e-05,
+      "loss": 1.2207,
+      "step": 167460
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 5.583301446391673e-05,
+      "loss": 1.1424,
+      "step": 167470
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 5.583186653401698e-05,
+      "loss": 1.066,
+      "step": 167480
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 5.583071860411724e-05,
+      "loss": 1.1014,
+      "step": 167490
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 5.5829570674217494e-05,
+      "loss": 1.2017,
+      "step": 167500
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 5.582842274431774e-05,
+      "loss": 1.2422,
+      "step": 167510
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 5.582727481441799e-05,
+      "loss": 1.1217,
+      "step": 167520
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 5.582612688451825e-05,
+      "loss": 1.3131,
+      "step": 167530
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 5.582497895461849e-05,
+      "loss": 1.2858,
+      "step": 167540
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 5.582383102471875e-05,
+      "loss": 1.2306,
+      "step": 167550
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 5.5822683094819006e-05,
+      "loss": 1.1672,
+      "step": 167560
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 5.5821535164919256e-05,
+      "loss": 1.1542,
+      "step": 167570
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 5.5820387235019506e-05,
+      "loss": 1.2808,
+      "step": 167580
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 5.581923930511976e-05,
+      "loss": 1.2815,
+      "step": 167590
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 5.581809137522001e-05,
+      "loss": 1.1163,
+      "step": 167600
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 5.581694344532027e-05,
+      "loss": 1.0656,
+      "step": 167610
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 5.581579551542052e-05,
+      "loss": 1.2264,
+      "step": 167620
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 5.581464758552077e-05,
+      "loss": 1.1449,
+      "step": 167630
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 5.5813499655621025e-05,
+      "loss": 1.1746,
+      "step": 167640
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 5.581235172572128e-05,
+      "loss": 1.255,
+      "step": 167650
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 5.5811203795821525e-05,
+      "loss": 1.2854,
+      "step": 167660
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 5.581005586592178e-05,
+      "loss": 1.156,
+      "step": 167670
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 5.580890793602204e-05,
+      "loss": 1.1822,
+      "step": 167680
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 5.580776000612228e-05,
+      "loss": 1.2485,
+      "step": 167690
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 5.580661207622254e-05,
+      "loss": 1.2803,
+      "step": 167700
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 5.5805464146322795e-05,
+      "loss": 1.1701,
+      "step": 167710
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 5.5804316216423045e-05,
+      "loss": 1.2026,
+      "step": 167720
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 5.5803168286523294e-05,
+      "loss": 1.2119,
+      "step": 167730
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 5.580202035662355e-05,
+      "loss": 1.1557,
+      "step": 167740
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 5.58008724267238e-05,
+      "loss": 1.1859,
+      "step": 167750
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 5.579972449682406e-05,
+      "loss": 1.181,
+      "step": 167760
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 5.579857656692431e-05,
+      "loss": 1.1813,
+      "step": 167770
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 5.579742863702456e-05,
+      "loss": 1.2268,
+      "step": 167780
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 5.5796280707124814e-05,
+      "loss": 1.1269,
+      "step": 167790
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 5.579513277722507e-05,
+      "loss": 1.1123,
+      "step": 167800
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 5.5793984847325314e-05,
+      "loss": 1.1528,
+      "step": 167810
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 5.579283691742557e-05,
+      "loss": 1.153,
+      "step": 167820
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 5.579168898752583e-05,
+      "loss": 1.2019,
+      "step": 167830
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 5.579054105762607e-05,
+      "loss": 1.1942,
+      "step": 167840
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 5.5789393127726327e-05,
+      "loss": 1.0946,
+      "step": 167850
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 5.578824519782658e-05,
+      "loss": 1.16,
+      "step": 167860
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 5.578709726792683e-05,
+      "loss": 1.128,
+      "step": 167870
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 5.578594933802708e-05,
+      "loss": 1.3274,
+      "step": 167880
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 5.578480140812734e-05,
+      "loss": 1.333,
+      "step": 167890
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 5.578365347822759e-05,
+      "loss": 1.1533,
+      "step": 167900
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 5.5782505548327846e-05,
+      "loss": 1.24,
+      "step": 167910
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 5.5781357618428096e-05,
+      "loss": 1.3853,
+      "step": 167920
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 5.5780209688528346e-05,
+      "loss": 1.1266,
+      "step": 167930
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 5.57790617586286e-05,
+      "loss": 1.159,
+      "step": 167940
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 5.577791382872886e-05,
+      "loss": 1.2527,
+      "step": 167950
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 5.57767658988291e-05,
+      "loss": 1.1509,
+      "step": 167960
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 5.577561796892936e-05,
+      "loss": 1.1438,
+      "step": 167970
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 5.5774470039029615e-05,
+      "loss": 1.2769,
+      "step": 167980
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 5.577332210912986e-05,
+      "loss": 1.2239,
+      "step": 167990
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 5.5772174179230115e-05,
+      "loss": 1.2086,
+      "step": 168000
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 5.577102624933037e-05,
+      "loss": 1.226,
+      "step": 168010
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 5.576987831943062e-05,
+      "loss": 1.3343,
+      "step": 168020
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 5.576873038953087e-05,
+      "loss": 1.2329,
+      "step": 168030
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 5.576758245963113e-05,
+      "loss": 1.1201,
+      "step": 168040
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 5.576643452973138e-05,
+      "loss": 1.2002,
+      "step": 168050
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 5.5765286599831634e-05,
+      "loss": 1.2424,
+      "step": 168060
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 5.5764138669931884e-05,
+      "loss": 1.3188,
+      "step": 168070
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 5.5762990740032134e-05,
+      "loss": 1.0981,
+      "step": 168080
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 5.576184281013239e-05,
+      "loss": 1.077,
+      "step": 168090
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 5.576069488023265e-05,
+      "loss": 1.0902,
+      "step": 168100
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 5.575954695033289e-05,
+      "loss": 1.2252,
+      "step": 168110
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 5.575839902043315e-05,
+      "loss": 1.1875,
+      "step": 168120
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 5.5757251090533404e-05,
+      "loss": 1.1225,
+      "step": 168130
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 5.575610316063365e-05,
+      "loss": 1.0961,
+      "step": 168140
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 5.5754955230733903e-05,
+      "loss": 1.1995,
+      "step": 168150
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 5.575380730083416e-05,
+      "loss": 1.2256,
+      "step": 168160
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 5.575265937093441e-05,
+      "loss": 1.1725,
+      "step": 168170
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 5.575151144103466e-05,
+      "loss": 1.0798,
+      "step": 168180
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 5.5750363511134916e-05,
+      "loss": 1.1171,
+      "step": 168190
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 5.5749215581235166e-05,
+      "loss": 1.2718,
+      "step": 168200
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 5.574806765133542e-05,
+      "loss": 1.0767,
+      "step": 168210
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 5.574691972143567e-05,
+      "loss": 1.1817,
+      "step": 168220
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 5.574577179153592e-05,
+      "loss": 1.1967,
+      "step": 168230
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 5.574462386163618e-05,
+      "loss": 1.1255,
+      "step": 168240
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 5.5743475931736436e-05,
+      "loss": 1.1918,
+      "step": 168250
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 5.574232800183668e-05,
+      "loss": 1.34,
+      "step": 168260
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 5.5741180071936936e-05,
+      "loss": 1.1772,
+      "step": 168270
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 5.574003214203719e-05,
+      "loss": 1.2572,
+      "step": 168280
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 5.5738884212137435e-05,
+      "loss": 1.2138,
+      "step": 168290
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 5.573773628223769e-05,
+      "loss": 1.1064,
+      "step": 168300
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 5.573658835233795e-05,
+      "loss": 1.1837,
+      "step": 168310
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 5.57354404224382e-05,
+      "loss": 1.1991,
+      "step": 168320
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 5.573429249253845e-05,
+      "loss": 1.174,
+      "step": 168330
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 5.5733144562638705e-05,
+      "loss": 1.234,
+      "step": 168340
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 5.5731996632738955e-05,
+      "loss": 1.2014,
+      "step": 168350
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 5.573084870283921e-05,
+      "loss": 1.1737,
+      "step": 168360
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 5.572970077293946e-05,
+      "loss": 1.1201,
+      "step": 168370
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 5.572855284303971e-05,
+      "loss": 1.2356,
+      "step": 168380
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 5.572740491313997e-05,
+      "loss": 1.2444,
+      "step": 168390
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 5.5726256983240224e-05,
+      "loss": 1.0993,
+      "step": 168400
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 5.572510905334047e-05,
+      "loss": 1.2134,
+      "step": 168410
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 5.5723961123440724e-05,
+      "loss": 1.243,
+      "step": 168420
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 5.572281319354098e-05,
+      "loss": 0.9862,
+      "step": 168430
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 5.5721665263641224e-05,
+      "loss": 1.0386,
+      "step": 168440
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 5.572051733374148e-05,
+      "loss": 1.1415,
+      "step": 168450
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 5.571936940384174e-05,
+      "loss": 1.343,
+      "step": 168460
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 5.571822147394199e-05,
+      "loss": 1.2624,
+      "step": 168470
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 5.571707354404224e-05,
+      "loss": 1.1569,
+      "step": 168480
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 5.5715925614142493e-05,
+      "loss": 1.2496,
+      "step": 168490
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 5.571477768424274e-05,
+      "loss": 1.2279,
+      "step": 168500
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 5.5713629754343e-05,
+      "loss": 1.2245,
+      "step": 168510
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 5.571248182444325e-05,
+      "loss": 1.2402,
+      "step": 168520
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 5.57113338945435e-05,
+      "loss": 1.1481,
+      "step": 168530
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 5.5710185964643756e-05,
+      "loss": 1.2871,
+      "step": 168540
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 5.570903803474401e-05,
+      "loss": 1.2691,
+      "step": 168550
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 5.5707890104844256e-05,
+      "loss": 1.1792,
+      "step": 168560
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 5.570674217494451e-05,
+      "loss": 1.2443,
+      "step": 168570
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 5.570559424504477e-05,
+      "loss": 1.1742,
+      "step": 168580
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 5.570444631514501e-05,
+      "loss": 1.3612,
+      "step": 168590
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 5.570329838524527e-05,
+      "loss": 1.1437,
+      "step": 168600
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 5.5702150455345526e-05,
+      "loss": 1.1608,
+      "step": 168610
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 5.5701002525445775e-05,
+      "loss": 1.2274,
+      "step": 168620
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 5.5699854595546025e-05,
+      "loss": 1.1382,
+      "step": 168630
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 5.569870666564628e-05,
+      "loss": 1.1347,
+      "step": 168640
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 5.569755873574653e-05,
+      "loss": 1.1737,
+      "step": 168650
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 5.569641080584679e-05,
+      "loss": 1.0906,
+      "step": 168660
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 5.569526287594703e-05,
+      "loss": 1.108,
+      "step": 168670
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 5.569411494604729e-05,
+      "loss": 1.2749,
+      "step": 168680
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 5.5692967016147545e-05,
+      "loss": 1.2574,
+      "step": 168690
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 5.569181908624779e-05,
+      "loss": 1.2936,
+      "step": 168700
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 5.5690671156348044e-05,
+      "loss": 1.0815,
+      "step": 168710
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 5.56895232264483e-05,
+      "loss": 1.2202,
+      "step": 168720
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 5.568837529654855e-05,
+      "loss": 1.0555,
+      "step": 168730
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 5.56872273666488e-05,
+      "loss": 1.1331,
+      "step": 168740
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 5.568607943674906e-05,
+      "loss": 1.1558,
+      "step": 168750
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 5.568493150684931e-05,
+      "loss": 1.146,
+      "step": 168760
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 5.5683783576949564e-05,
+      "loss": 1.2311,
+      "step": 168770
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 5.5682635647049814e-05,
+      "loss": 1.1525,
+      "step": 168780
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 5.5681487717150064e-05,
+      "loss": 1.1216,
+      "step": 168790
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 5.568033978725032e-05,
+      "loss": 1.2772,
+      "step": 168800
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 5.567919185735058e-05,
+      "loss": 1.1867,
+      "step": 168810
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 5.567804392745082e-05,
+      "loss": 1.1345,
+      "step": 168820
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 5.5676895997551077e-05,
+      "loss": 1.2313,
+      "step": 168830
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 5.567574806765133e-05,
+      "loss": 1.2184,
+      "step": 168840
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 5.5674600137751576e-05,
+      "loss": 1.2433,
+      "step": 168850
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 5.567345220785183e-05,
+      "loss": 1.1719,
+      "step": 168860
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 5.567230427795209e-05,
+      "loss": 1.2566,
+      "step": 168870
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 5.567115634805234e-05,
+      "loss": 1.106,
+      "step": 168880
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 5.567000841815259e-05,
+      "loss": 1.1453,
+      "step": 168890
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 5.5668860488252846e-05,
+      "loss": 1.154,
+      "step": 168900
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 5.5667712558353096e-05,
+      "loss": 1.1635,
+      "step": 168910
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 5.566656462845335e-05,
+      "loss": 1.209,
+      "step": 168920
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 5.56654166985536e-05,
+      "loss": 1.2084,
+      "step": 168930
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 5.566426876865385e-05,
+      "loss": 1.0094,
+      "step": 168940
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 5.566312083875411e-05,
+      "loss": 1.2484,
+      "step": 168950
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 5.5661972908854365e-05,
+      "loss": 1.1353,
+      "step": 168960
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 5.566082497895461e-05,
+      "loss": 1.1556,
+      "step": 168970
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 5.5659677049054865e-05,
+      "loss": 0.9634,
+      "step": 168980
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 5.565852911915512e-05,
+      "loss": 1.2798,
+      "step": 168990
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 5.5657381189255365e-05,
+      "loss": 1.0835,
+      "step": 169000
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 5.565623325935562e-05,
+      "loss": 1.0773,
+      "step": 169010
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 5.565508532945588e-05,
+      "loss": 1.2407,
+      "step": 169020
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 5.565393739955613e-05,
+      "loss": 1.135,
+      "step": 169030
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 5.565278946965638e-05,
+      "loss": 1.2188,
+      "step": 169040
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 5.5651641539756634e-05,
+      "loss": 1.1696,
+      "step": 169050
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 5.5650493609856884e-05,
+      "loss": 1.1391,
+      "step": 169060
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 5.564934567995714e-05,
+      "loss": 1.2522,
+      "step": 169070
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 5.564819775005739e-05,
+      "loss": 1.2827,
+      "step": 169080
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 5.564704982015764e-05,
+      "loss": 1.2276,
+      "step": 169090
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 5.56459018902579e-05,
+      "loss": 1.2839,
+      "step": 169100
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 5.5644753960358154e-05,
+      "loss": 1.1054,
+      "step": 169110
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 5.56436060304584e-05,
+      "loss": 1.0694,
+      "step": 169120
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 5.5642458100558653e-05,
+      "loss": 1.3271,
+      "step": 169130
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 5.564131017065891e-05,
+      "loss": 1.0985,
+      "step": 169140
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 5.564016224075915e-05,
+      "loss": 1.2441,
+      "step": 169150
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 5.563901431085941e-05,
+      "loss": 1.302,
+      "step": 169160
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 5.5637866380959666e-05,
+      "loss": 1.189,
+      "step": 169170
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 5.5636718451059916e-05,
+      "loss": 1.1223,
+      "step": 169180
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 5.5635570521160166e-05,
+      "loss": 1.3166,
+      "step": 169190
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 5.563442259126042e-05,
+      "loss": 1.0847,
+      "step": 169200
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 5.563327466136067e-05,
+      "loss": 1.181,
+      "step": 169210
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 5.563212673146093e-05,
+      "loss": 1.4033,
+      "step": 169220
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 5.563097880156118e-05,
+      "loss": 1.2351,
+      "step": 169230
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 5.562983087166143e-05,
+      "loss": 1.2074,
+      "step": 169240
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 5.5628682941761686e-05,
+      "loss": 1.2444,
+      "step": 169250
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 5.562753501186194e-05,
+      "loss": 1.1305,
+      "step": 169260
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 5.5626387081962185e-05,
+      "loss": 1.3168,
+      "step": 169270
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 5.562523915206244e-05,
+      "loss": 1.239,
+      "step": 169280
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 5.56240912221627e-05,
+      "loss": 1.1198,
+      "step": 169290
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 5.562294329226294e-05,
+      "loss": 1.174,
+      "step": 169300
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 5.56217953623632e-05,
+      "loss": 1.1051,
+      "step": 169310
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 5.5620647432463455e-05,
+      "loss": 1.1708,
+      "step": 169320
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 5.5619499502563705e-05,
+      "loss": 1.1026,
+      "step": 169330
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 5.5618351572663955e-05,
+      "loss": 1.1879,
+      "step": 169340
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 5.561720364276421e-05,
+      "loss": 1.1733,
+      "step": 169350
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 5.561605571286446e-05,
+      "loss": 1.2819,
+      "step": 169360
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 5.561490778296472e-05,
+      "loss": 1.1807,
+      "step": 169370
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 5.561375985306497e-05,
+      "loss": 1.2076,
+      "step": 169380
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 5.561261192316522e-05,
+      "loss": 1.2629,
+      "step": 169390
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 5.5611463993265474e-05,
+      "loss": 1.2297,
+      "step": 169400
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 5.561031606336573e-05,
+      "loss": 1.2256,
+      "step": 169410
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 5.5609168133465974e-05,
+      "loss": 1.2865,
+      "step": 169420
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 5.560802020356623e-05,
+      "loss": 1.3494,
+      "step": 169430
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 5.560687227366649e-05,
+      "loss": 1.2412,
+      "step": 169440
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 5.560572434376673e-05,
+      "loss": 1.3212,
+      "step": 169450
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 5.560457641386699e-05,
+      "loss": 1.1216,
+      "step": 169460
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 5.560342848396724e-05,
+      "loss": 1.1859,
+      "step": 169470
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 5.560228055406749e-05,
+      "loss": 1.1517,
+      "step": 169480
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 5.560113262416774e-05,
+      "loss": 1.1789,
+      "step": 169490
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 5.5599984694268e-05,
+      "loss": 1.3004,
+      "step": 169500
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 5.559883676436825e-05,
+      "loss": 1.1569,
+      "step": 169510
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 5.5597688834468506e-05,
+      "loss": 1.1781,
+      "step": 169520
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 5.5596540904568756e-05,
+      "loss": 1.161,
+      "step": 169530
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 5.5595392974669006e-05,
+      "loss": 1.2244,
+      "step": 169540
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 5.559424504476926e-05,
+      "loss": 1.1413,
+      "step": 169550
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 5.559309711486952e-05,
+      "loss": 1.2397,
+      "step": 169560
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 5.559194918496976e-05,
+      "loss": 1.2693,
+      "step": 169570
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 5.559080125507002e-05,
+      "loss": 1.1871,
+      "step": 169580
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 5.5589653325170276e-05,
+      "loss": 1.1017,
+      "step": 169590
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 5.558850539527052e-05,
+      "loss": 1.1374,
+      "step": 169600
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 5.5587357465370775e-05,
+      "loss": 1.3162,
+      "step": 169610
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 5.558620953547103e-05,
+      "loss": 1.1876,
+      "step": 169620
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 5.558506160557128e-05,
+      "loss": 1.18,
+      "step": 169630
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 5.558391367567153e-05,
+      "loss": 1.257,
+      "step": 169640
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 5.558276574577179e-05,
+      "loss": 1.0267,
+      "step": 169650
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 5.558161781587204e-05,
+      "loss": 1.1529,
+      "step": 169660
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 5.5580469885972295e-05,
+      "loss": 1.2642,
+      "step": 169670
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.5579321956072545e-05,
+      "loss": 1.1316,
+      "step": 169680
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.5578174026172794e-05,
+      "loss": 1.079,
+      "step": 169690
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.557702609627305e-05,
+      "loss": 1.2038,
+      "step": 169700
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.557587816637331e-05,
+      "loss": 1.1825,
+      "step": 169710
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.557473023647355e-05,
+      "loss": 1.2078,
+      "step": 169720
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.557358230657381e-05,
+      "loss": 1.1225,
+      "step": 169730
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.5572434376674064e-05,
+      "loss": 1.1779,
+      "step": 169740
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.557128644677431e-05,
+      "loss": 1.1069,
+      "step": 169750
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.5570138516874564e-05,
+      "loss": 1.1793,
+      "step": 169760
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.556899058697482e-05,
+      "loss": 1.1638,
+      "step": 169770
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.556784265707507e-05,
+      "loss": 1.2077,
+      "step": 169780
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.556669472717532e-05,
+      "loss": 1.194,
+      "step": 169790
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.556554679727558e-05,
+      "loss": 1.1934,
+      "step": 169800
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.5564398867375827e-05,
+      "loss": 1.1668,
+      "step": 169810
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.556325093747608e-05,
+      "loss": 1.0955,
+      "step": 169820
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.556210300757633e-05,
+      "loss": 1.2369,
+      "step": 169830
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.556095507767658e-05,
+      "loss": 1.307,
+      "step": 169840
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.555980714777684e-05,
+      "loss": 1.0814,
+      "step": 169850
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.5558659217877096e-05,
+      "loss": 1.1622,
+      "step": 169860
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.555751128797734e-05,
+      "loss": 1.0782,
+      "step": 169870
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.5556363358077596e-05,
+      "loss": 1.2592,
+      "step": 169880
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 5.555521542817785e-05,
+      "loss": 1.112,
+      "step": 169890
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 5.5554067498278096e-05,
+      "loss": 1.2435,
+      "step": 169900
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 5.555291956837835e-05,
+      "loss": 1.1962,
+      "step": 169910
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 5.555177163847861e-05,
+      "loss": 1.2621,
+      "step": 169920
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 5.555062370857886e-05,
+      "loss": 1.0823,
+      "step": 169930
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 5.554947577867911e-05,
+      "loss": 1.262,
+      "step": 169940
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 5.5548327848779365e-05,
+      "loss": 1.2682,
+      "step": 169950
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 5.5547179918879615e-05,
+      "loss": 1.15,
+      "step": 169960
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 5.554603198897987e-05,
+      "loss": 1.2647,
+      "step": 169970
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 5.554488405908012e-05,
+      "loss": 1.1915,
+      "step": 169980
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 5.554373612918037e-05,
+      "loss": 1.1741,
+      "step": 169990
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 5.554258819928063e-05,
+      "loss": 1.2129,
+      "step": 170000
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 5.5541440269380885e-05,
+      "loss": 1.1749,
+      "step": 170010
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 5.554029233948113e-05,
+      "loss": 1.2962,
+      "step": 170020
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 5.5539144409581384e-05,
+      "loss": 1.1597,
+      "step": 170030
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 5.5537996479681634e-05,
+      "loss": 1.1601,
+      "step": 170040
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 5.5536848549781884e-05,
+      "loss": 1.148,
+      "step": 170050
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 5.553570061988214e-05,
+      "loss": 1.1609,
+      "step": 170060
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 5.553455268998239e-05,
+      "loss": 1.0733,
+      "step": 170070
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 5.553340476008265e-05,
+      "loss": 1.1331,
+      "step": 170080
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 5.55322568301829e-05,
+      "loss": 1.1128,
+      "step": 170090
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 5.553110890028315e-05,
+      "loss": 1.2807,
+      "step": 170100
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 5.5529960970383403e-05,
+      "loss": 1.0674,
+      "step": 170110
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 5.552881304048366e-05,
+      "loss": 1.1754,
+      "step": 170120
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 5.55276651105839e-05,
+      "loss": 1.2425,
+      "step": 170130
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 5.552651718068416e-05,
+      "loss": 1.3239,
+      "step": 170140
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 5.5525369250784416e-05,
+      "loss": 1.2061,
+      "step": 170150
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 5.552422132088466e-05,
+      "loss": 1.1947,
+      "step": 170160
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 5.5523073390984916e-05,
+      "loss": 1.1356,
+      "step": 170170
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 5.552192546108517e-05,
+      "loss": 1.1521,
+      "step": 170180
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 5.552077753118542e-05,
+      "loss": 1.1393,
+      "step": 170190
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 5.551962960128567e-05,
+      "loss": 1.0753,
+      "step": 170200
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 5.551848167138593e-05,
+      "loss": 1.2416,
+      "step": 170210
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 5.551733374148618e-05,
+      "loss": 1.1604,
+      "step": 170220
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 5.5516185811586436e-05,
+      "loss": 1.2293,
+      "step": 170230
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 5.5515037881686685e-05,
+      "loss": 1.0821,
+      "step": 170240
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 5.5513889951786935e-05,
+      "loss": 1.089,
+      "step": 170250
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 5.551274202188719e-05,
+      "loss": 1.1432,
+      "step": 170260
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 5.551159409198745e-05,
+      "loss": 1.0324,
+      "step": 170270
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 5.551044616208769e-05,
+      "loss": 1.1869,
+      "step": 170280
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 5.550929823218795e-05,
+      "loss": 1.2658,
+      "step": 170290
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 5.5508150302288205e-05,
+      "loss": 1.1552,
+      "step": 170300
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 5.550700237238845e-05,
+      "loss": 1.2631,
+      "step": 170310
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 5.5505854442488705e-05,
+      "loss": 1.1966,
+      "step": 170320
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 5.550470651258896e-05,
+      "loss": 1.0623,
+      "step": 170330
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 5.550355858268921e-05,
+      "loss": 1.2892,
+      "step": 170340
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 5.550241065278946e-05,
+      "loss": 1.2456,
+      "step": 170350
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 5.550126272288972e-05,
+      "loss": 1.2308,
+      "step": 170360
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 5.550011479298997e-05,
+      "loss": 1.1055,
+      "step": 170370
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 5.5498966863090224e-05,
+      "loss": 1.2671,
+      "step": 170380
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 5.5497818933190474e-05,
+      "loss": 1.175,
+      "step": 170390
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 5.5496671003290724e-05,
+      "loss": 1.2285,
+      "step": 170400
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 5.549552307339098e-05,
+      "loss": 1.1159,
+      "step": 170410
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 5.549437514349124e-05,
+      "loss": 1.1567,
+      "step": 170420
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 5.549322721359148e-05,
+      "loss": 1.2267,
+      "step": 170430
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 5.549207928369174e-05,
+      "loss": 1.122,
+      "step": 170440
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 5.549093135379199e-05,
+      "loss": 1.1784,
+      "step": 170450
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 5.5489783423892236e-05,
+      "loss": 1.2208,
+      "step": 170460
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 5.548863549399249e-05,
+      "loss": 1.1179,
+      "step": 170470
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 5.548748756409275e-05,
+      "loss": 1.1133,
+      "step": 170480
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 5.5486339634193e-05,
+      "loss": 1.1416,
+      "step": 170490
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 5.548519170429325e-05,
+      "loss": 1.1416,
+      "step": 170500
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 5.5484043774393506e-05,
+      "loss": 1.1526,
+      "step": 170510
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 5.5482895844493756e-05,
+      "loss": 1.3077,
+      "step": 170520
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 5.548174791459401e-05,
+      "loss": 1.2801,
+      "step": 170530
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 5.548059998469426e-05,
+      "loss": 1.306,
+      "step": 170540
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 5.547945205479451e-05,
+      "loss": 1.0825,
+      "step": 170550
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 5.547830412489477e-05,
+      "loss": 1.1472,
+      "step": 170560
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 5.5477156194995026e-05,
+      "loss": 1.175,
+      "step": 170570
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 5.547600826509527e-05,
+      "loss": 1.0712,
+      "step": 170580
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 5.5474860335195525e-05,
+      "loss": 1.1034,
+      "step": 170590
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 5.547371240529578e-05,
+      "loss": 1.0821,
+      "step": 170600
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 5.5472564475396025e-05,
+      "loss": 1.1713,
+      "step": 170610
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 5.547141654549628e-05,
+      "loss": 1.1792,
+      "step": 170620
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 5.547026861559654e-05,
+      "loss": 1.1013,
+      "step": 170630
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 5.546912068569679e-05,
+      "loss": 1.1127,
+      "step": 170640
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 5.546797275579704e-05,
+      "loss": 1.284,
+      "step": 170650
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 5.5466824825897295e-05,
+      "loss": 1.0958,
+      "step": 170660
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 5.5465676895997544e-05,
+      "loss": 1.1876,
+      "step": 170670
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 5.54645289660978e-05,
+      "loss": 1.1557,
+      "step": 170680
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 5.546338103619805e-05,
+      "loss": 1.1605,
+      "step": 170690
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 5.54622331062983e-05,
+      "loss": 1.0737,
+      "step": 170700
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 5.546108517639856e-05,
+      "loss": 1.385,
+      "step": 170710
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 5.5459937246498814e-05,
+      "loss": 1.2028,
+      "step": 170720
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 5.545878931659906e-05,
+      "loss": 1.1694,
+      "step": 170730
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 5.5457641386699314e-05,
+      "loss": 1.261,
+      "step": 170740
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 5.545649345679957e-05,
+      "loss": 1.1457,
+      "step": 170750
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 5.5455345526899813e-05,
+      "loss": 1.155,
+      "step": 170760
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 5.545419759700007e-05,
+      "loss": 1.2735,
+      "step": 170770
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 5.545304966710033e-05,
+      "loss": 1.3104,
+      "step": 170780
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 5.5451901737200577e-05,
+      "loss": 1.2231,
+      "step": 170790
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 5.5450753807300826e-05,
+      "loss": 1.183,
+      "step": 170800
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 5.544960587740108e-05,
+      "loss": 1.1018,
+      "step": 170810
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 5.544845794750133e-05,
+      "loss": 1.1596,
+      "step": 170820
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 5.544731001760159e-05,
+      "loss": 1.2396,
+      "step": 170830
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 5.544616208770184e-05,
+      "loss": 1.2036,
+      "step": 170840
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 5.544501415780209e-05,
+      "loss": 1.1197,
+      "step": 170850
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 5.5443866227902346e-05,
+      "loss": 1.1453,
+      "step": 170860
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 5.54427182980026e-05,
+      "loss": 1.3273,
+      "step": 170870
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 5.5441570368102846e-05,
+      "loss": 1.1198,
+      "step": 170880
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 5.54404224382031e-05,
+      "loss": 1.1575,
+      "step": 170890
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 5.543927450830336e-05,
+      "loss": 1.1341,
+      "step": 170900
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 5.54381265784036e-05,
+      "loss": 1.0309,
+      "step": 170910
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 5.543697864850386e-05,
+      "loss": 1.2349,
+      "step": 170920
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 5.5435830718604115e-05,
+      "loss": 1.1499,
+      "step": 170930
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 5.5434682788704365e-05,
+      "loss": 1.2093,
+      "step": 170940
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 5.5433534858804615e-05,
+      "loss": 1.1811,
+      "step": 170950
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 5.543238692890487e-05,
+      "loss": 1.148,
+      "step": 170960
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 5.543123899900512e-05,
+      "loss": 1.1334,
+      "step": 170970
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 5.543009106910538e-05,
+      "loss": 1.0485,
+      "step": 170980
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 5.542894313920563e-05,
+      "loss": 1.1687,
+      "step": 170990
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 5.542779520930588e-05,
+      "loss": 1.1017,
+      "step": 171000
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 5.5426647279406134e-05,
+      "loss": 1.1304,
+      "step": 171010
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 5.542549934950639e-05,
+      "loss": 1.2452,
+      "step": 171020
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 5.5424351419606634e-05,
+      "loss": 1.1378,
+      "step": 171030
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 5.542320348970689e-05,
+      "loss": 1.1921,
+      "step": 171040
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 5.542205555980715e-05,
+      "loss": 1.1618,
+      "step": 171050
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 5.542090762990739e-05,
+      "loss": 1.301,
+      "step": 171060
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 5.541975970000765e-05,
+      "loss": 1.1113,
+      "step": 171070
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 5.5418611770107904e-05,
+      "loss": 1.2617,
+      "step": 171080
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 5.5417463840208153e-05,
+      "loss": 1.1548,
+      "step": 171090
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 5.54163159103084e-05,
+      "loss": 1.1074,
+      "step": 171100
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 5.541516798040866e-05,
+      "loss": 1.2126,
+      "step": 171110
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 5.541402005050891e-05,
+      "loss": 1.3106,
+      "step": 171120
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 5.5412872120609166e-05,
+      "loss": 1.1705,
+      "step": 171130
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 5.5411724190709416e-05,
+      "loss": 1.2384,
+      "step": 171140
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 5.5410576260809666e-05,
+      "loss": 1.2025,
+      "step": 171150
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 5.540942833090992e-05,
+      "loss": 1.257,
+      "step": 171160
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 5.540828040101018e-05,
+      "loss": 1.2967,
+      "step": 171170
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 5.540713247111042e-05,
+      "loss": 1.1757,
+      "step": 171180
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 5.540598454121068e-05,
+      "loss": 1.1922,
+      "step": 171190
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 5.5404836611310936e-05,
+      "loss": 1.213,
+      "step": 171200
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 5.540368868141118e-05,
+      "loss": 1.2202,
+      "step": 171210
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 5.5402540751511435e-05,
+      "loss": 0.9663,
+      "step": 171220
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 5.540139282161169e-05,
+      "loss": 1.1839,
+      "step": 171230
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 5.540024489171194e-05,
+      "loss": 1.2826,
+      "step": 171240
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 5.539909696181219e-05,
+      "loss": 1.2536,
+      "step": 171250
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 5.539794903191245e-05,
+      "loss": 1.2943,
+      "step": 171260
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 5.53968011020127e-05,
+      "loss": 1.2819,
+      "step": 171270
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 5.5395653172112955e-05,
+      "loss": 1.0564,
+      "step": 171280
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 5.5394505242213205e-05,
+      "loss": 1.2488,
+      "step": 171290
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 5.5393357312313455e-05,
+      "loss": 1.1403,
+      "step": 171300
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 5.539220938241371e-05,
+      "loss": 1.2301,
+      "step": 171310
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 5.539106145251397e-05,
+      "loss": 1.2093,
+      "step": 171320
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 5.538991352261421e-05,
+      "loss": 1.0951,
+      "step": 171330
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 5.538876559271447e-05,
+      "loss": 1.1734,
+      "step": 171340
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 5.5387617662814724e-05,
+      "loss": 1.2596,
+      "step": 171350
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 5.538646973291497e-05,
+      "loss": 1.1958,
+      "step": 171360
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 5.5385321803015224e-05,
+      "loss": 1.1727,
+      "step": 171370
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 5.538417387311548e-05,
+      "loss": 1.1848,
+      "step": 171380
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 5.538302594321573e-05,
+      "loss": 1.2222,
+      "step": 171390
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 5.538187801331598e-05,
+      "loss": 1.1043,
+      "step": 171400
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 5.538073008341623e-05,
+      "loss": 1.2103,
+      "step": 171410
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 5.537958215351649e-05,
+      "loss": 1.2405,
+      "step": 171420
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 5.537843422361674e-05,
+      "loss": 1.0574,
+      "step": 171430
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 5.5377286293716986e-05,
+      "loss": 1.1222,
+      "step": 171440
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 5.537613836381724e-05,
+      "loss": 1.1842,
+      "step": 171450
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 5.53749904339175e-05,
+      "loss": 1.0837,
+      "step": 171460
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 5.537384250401774e-05,
+      "loss": 1.2441,
+      "step": 171470
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 5.5372694574118e-05,
+      "loss": 1.2529,
+      "step": 171480
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 5.5371546644218256e-05,
+      "loss": 1.1676,
+      "step": 171490
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 5.5370398714318506e-05,
+      "loss": 1.2366,
+      "step": 171500
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 5.5369250784418756e-05,
+      "loss": 1.2396,
+      "step": 171510
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 5.536810285451901e-05,
+      "loss": 1.1422,
+      "step": 171520
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 5.536695492461926e-05,
+      "loss": 1.1916,
+      "step": 171530
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 5.536580699471952e-05,
+      "loss": 1.2377,
+      "step": 171540
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 5.536465906481977e-05,
+      "loss": 1.2423,
+      "step": 171550
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 5.536351113492002e-05,
+      "loss": 1.1582,
+      "step": 171560
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 5.5362363205020275e-05,
+      "loss": 1.2061,
+      "step": 171570
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 5.536121527512053e-05,
+      "loss": 1.0882,
+      "step": 171580
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 5.5360067345220775e-05,
+      "loss": 1.1866,
+      "step": 171590
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 5.535891941532103e-05,
+      "loss": 1.1492,
+      "step": 171600
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 5.535777148542129e-05,
+      "loss": 1.1164,
+      "step": 171610
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 5.535662355552153e-05,
+      "loss": 1.1,
+      "step": 171620
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 5.535547562562179e-05,
+      "loss": 1.1092,
+      "step": 171630
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 5.5354327695722045e-05,
+      "loss": 1.0702,
+      "step": 171640
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 5.5353179765822294e-05,
+      "loss": 1.1221,
+      "step": 171650
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 5.5352031835922544e-05,
+      "loss": 1.1939,
+      "step": 171660
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 5.53508839060228e-05,
+      "loss": 1.2166,
+      "step": 171670
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 5.534973597612305e-05,
+      "loss": 1.2442,
+      "step": 171680
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 5.534858804622331e-05,
+      "loss": 1.0944,
+      "step": 171690
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 5.534744011632356e-05,
+      "loss": 1.281,
+      "step": 171700
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 5.534629218642381e-05,
+      "loss": 1.1791,
+      "step": 171710
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 5.5345144256524064e-05,
+      "loss": 1.1064,
+      "step": 171720
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 5.534399632662432e-05,
+      "loss": 1.1326,
+      "step": 171730
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 5.5342848396724563e-05,
+      "loss": 1.2956,
+      "step": 171740
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 5.534170046682482e-05,
+      "loss": 1.3017,
+      "step": 171750
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 5.534055253692508e-05,
+      "loss": 1.1531,
+      "step": 171760
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 5.533940460702532e-05,
+      "loss": 1.149,
+      "step": 171770
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 5.5338256677125576e-05,
+      "loss": 1.1458,
+      "step": 171780
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 5.533710874722583e-05,
+      "loss": 1.0556,
+      "step": 171790
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 5.533596081732608e-05,
+      "loss": 1.2465,
+      "step": 171800
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 5.533481288742633e-05,
+      "loss": 1.1797,
+      "step": 171810
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 5.533366495752659e-05,
+      "loss": 1.2214,
+      "step": 171820
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 5.533251702762684e-05,
+      "loss": 1.1693,
+      "step": 171830
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 5.5331369097727096e-05,
+      "loss": 1.14,
+      "step": 171840
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 5.5330221167827346e-05,
+      "loss": 1.1947,
+      "step": 171850
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 5.5329073237927596e-05,
+      "loss": 1.1854,
+      "step": 171860
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 5.532792530802785e-05,
+      "loss": 1.2307,
+      "step": 171870
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 5.532677737812811e-05,
+      "loss": 1.1604,
+      "step": 171880
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 5.532562944822835e-05,
+      "loss": 1.0893,
+      "step": 171890
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 5.532448151832861e-05,
+      "loss": 1.2057,
+      "step": 171900
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 5.5323333588428865e-05,
+      "loss": 1.2026,
+      "step": 171910
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 5.532218565852911e-05,
+      "loss": 1.3231,
+      "step": 171920
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 5.5321037728629365e-05,
+      "loss": 1.3484,
+      "step": 171930
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 5.531988979872962e-05,
+      "loss": 1.203,
+      "step": 171940
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 5.531874186882987e-05,
+      "loss": 1.1668,
+      "step": 171950
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 5.531759393893012e-05,
+      "loss": 1.2803,
+      "step": 171960
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 5.531644600903038e-05,
+      "loss": 1.0963,
+      "step": 171970
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 5.531529807913063e-05,
+      "loss": 1.1488,
+      "step": 171980
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 5.5314150149230884e-05,
+      "loss": 1.3579,
+      "step": 171990
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 5.5313002219331134e-05,
+      "loss": 1.1275,
+      "step": 172000
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 5.5311854289431384e-05,
+      "loss": 1.2293,
+      "step": 172010
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 5.531070635953164e-05,
+      "loss": 1.2916,
+      "step": 172020
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 5.53095584296319e-05,
+      "loss": 1.2774,
+      "step": 172030
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 5.530841049973214e-05,
+      "loss": 1.3114,
+      "step": 172040
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 5.53072625698324e-05,
+      "loss": 1.3019,
+      "step": 172050
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 5.5306114639932654e-05,
+      "loss": 1.1006,
+      "step": 172060
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 5.53049667100329e-05,
+      "loss": 1.0829,
+      "step": 172070
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 5.530381878013315e-05,
+      "loss": 1.3275,
+      "step": 172080
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 5.530267085023341e-05,
+      "loss": 1.1521,
+      "step": 172090
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 5.530152292033366e-05,
+      "loss": 1.3806,
+      "step": 172100
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 5.530037499043391e-05,
+      "loss": 1.1087,
+      "step": 172110
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 5.5299227060534166e-05,
+      "loss": 1.1884,
+      "step": 172120
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 5.5298079130634416e-05,
+      "loss": 1.1981,
+      "step": 172130
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 5.529693120073467e-05,
+      "loss": 1.2596,
+      "step": 172140
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 5.529578327083492e-05,
+      "loss": 1.1656,
+      "step": 172150
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 5.529463534093517e-05,
+      "loss": 1.1137,
+      "step": 172160
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 5.529348741103543e-05,
+      "loss": 1.2786,
+      "step": 172170
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 5.5292339481135686e-05,
+      "loss": 1.3008,
+      "step": 172180
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 5.529119155123593e-05,
+      "loss": 1.2655,
+      "step": 172190
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 5.5290043621336185e-05,
+      "loss": 1.2015,
+      "step": 172200
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 5.528889569143644e-05,
+      "loss": 1.1554,
+      "step": 172210
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 5.5287747761536685e-05,
+      "loss": 1.2993,
+      "step": 172220
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 5.528659983163694e-05,
+      "loss": 1.2396,
+      "step": 172230
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 5.52854519017372e-05,
+      "loss": 1.167,
+      "step": 172240
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 5.528430397183745e-05,
+      "loss": 1.2023,
+      "step": 172250
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 5.52831560419377e-05,
+      "loss": 1.2823,
+      "step": 172260
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 5.5282008112037955e-05,
+      "loss": 1.1863,
+      "step": 172270
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 5.5280860182138205e-05,
+      "loss": 1.2527,
+      "step": 172280
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 5.527971225223846e-05,
+      "loss": 1.1188,
+      "step": 172290
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 5.527856432233871e-05,
+      "loss": 1.2276,
+      "step": 172300
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 5.527741639243896e-05,
+      "loss": 1.2598,
+      "step": 172310
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 5.527626846253922e-05,
+      "loss": 1.1201,
+      "step": 172320
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 5.5275120532639474e-05,
+      "loss": 1.2085,
+      "step": 172330
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 5.527397260273972e-05,
+      "loss": 1.212,
+      "step": 172340
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 5.5272824672839974e-05,
+      "loss": 1.2419,
+      "step": 172350
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 5.527167674294023e-05,
+      "loss": 1.1218,
+      "step": 172360
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 5.5270528813040474e-05,
+      "loss": 1.123,
+      "step": 172370
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 5.526938088314073e-05,
+      "loss": 1.2747,
+      "step": 172380
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 5.526823295324099e-05,
+      "loss": 1.0849,
+      "step": 172390
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 5.526708502334124e-05,
+      "loss": 1.2026,
+      "step": 172400
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 5.5265937093441487e-05,
+      "loss": 1.2177,
+      "step": 172410
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 5.526478916354174e-05,
+      "loss": 1.2813,
+      "step": 172420
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 5.526364123364199e-05,
+      "loss": 1.1459,
+      "step": 172430
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 5.526249330374225e-05,
+      "loss": 1.221,
+      "step": 172440
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 5.52613453738425e-05,
+      "loss": 1.2364,
+      "step": 172450
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 5.526019744394275e-05,
+      "loss": 1.1227,
+      "step": 172460
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 5.5259049514043006e-05,
+      "loss": 1.1938,
+      "step": 172470
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 5.525790158414326e-05,
+      "loss": 1.0839,
+      "step": 172480
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 5.5256753654243506e-05,
+      "loss": 1.2899,
+      "step": 172490
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 5.525560572434376e-05,
+      "loss": 1.1837,
+      "step": 172500
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 5.525445779444402e-05,
+      "loss": 1.2171,
+      "step": 172510
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 5.525330986454426e-05,
+      "loss": 1.1387,
+      "step": 172520
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 5.525216193464452e-05,
+      "loss": 1.0786,
+      "step": 172530
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 5.5251014004744775e-05,
+      "loss": 1.1131,
+      "step": 172540
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 5.5249866074845025e-05,
+      "loss": 1.1869,
+      "step": 172550
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 5.5248718144945275e-05,
+      "loss": 1.1154,
+      "step": 172560
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 5.524757021504553e-05,
+      "loss": 1.1569,
+      "step": 172570
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 5.524642228514578e-05,
+      "loss": 1.1118,
+      "step": 172580
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 5.524527435524604e-05,
+      "loss": 1.1734,
+      "step": 172590
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 5.524412642534629e-05,
+      "loss": 1.0557,
+      "step": 172600
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 5.524297849544654e-05,
+      "loss": 1.2367,
+      "step": 172610
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 5.5241830565546795e-05,
+      "loss": 1.2221,
+      "step": 172620
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 5.524068263564705e-05,
+      "loss": 1.128,
+      "step": 172630
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 5.5239534705747294e-05,
+      "loss": 1.3339,
+      "step": 172640
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 5.523838677584755e-05,
+      "loss": 1.2816,
+      "step": 172650
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 5.523723884594781e-05,
+      "loss": 1.2595,
+      "step": 172660
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 5.523609091604805e-05,
+      "loss": 1.0853,
+      "step": 172670
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 5.523494298614831e-05,
+      "loss": 1.2315,
+      "step": 172680
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 5.5233795056248564e-05,
+      "loss": 1.1255,
+      "step": 172690
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 5.5232647126348814e-05,
+      "loss": 1.2974,
+      "step": 172700
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 5.5231499196449064e-05,
+      "loss": 1.2326,
+      "step": 172710
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 5.523035126654932e-05,
+      "loss": 1.1498,
+      "step": 172720
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 5.522920333664957e-05,
+      "loss": 1.261,
+      "step": 172730
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 5.522805540674983e-05,
+      "loss": 1.195,
+      "step": 172740
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 5.5226907476850077e-05,
+      "loss": 1.3446,
+      "step": 172750
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 5.5225759546950326e-05,
+      "loss": 1.1462,
+      "step": 172760
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 5.522461161705058e-05,
+      "loss": 1.1953,
+      "step": 172770
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 5.5223463687150826e-05,
+      "loss": 1.0837,
+      "step": 172780
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 5.522231575725108e-05,
+      "loss": 1.2314,
+      "step": 172790
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 5.522116782735134e-05,
+      "loss": 1.2828,
+      "step": 172800
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 5.522001989745159e-05,
+      "loss": 1.1296,
+      "step": 172810
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 5.521887196755184e-05,
+      "loss": 1.1297,
+      "step": 172820
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 5.5217724037652096e-05,
+      "loss": 1.086,
+      "step": 172830
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 5.5216576107752346e-05,
+      "loss": 1.2291,
+      "step": 172840
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 5.52154281778526e-05,
+      "loss": 1.129,
+      "step": 172850
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 5.521428024795285e-05,
+      "loss": 1.1618,
+      "step": 172860
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 5.52131323180531e-05,
+      "loss": 1.2648,
+      "step": 172870
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 5.521198438815336e-05,
+      "loss": 1.0713,
+      "step": 172880
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 5.5210836458253615e-05,
+      "loss": 1.1567,
+      "step": 172890
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 5.520968852835386e-05,
+      "loss": 1.2307,
+      "step": 172900
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 5.5208540598454115e-05,
+      "loss": 1.1968,
+      "step": 172910
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 5.520739266855437e-05,
+      "loss": 1.177,
+      "step": 172920
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 5.5206244738654615e-05,
+      "loss": 1.165,
+      "step": 172930
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 5.520509680875487e-05,
+      "loss": 1.11,
+      "step": 172940
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.520394887885513e-05,
+      "loss": 1.1565,
+      "step": 172950
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.520280094895538e-05,
+      "loss": 1.2451,
+      "step": 172960
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.520165301905563e-05,
+      "loss": 1.2345,
+      "step": 172970
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.5200505089155884e-05,
+      "loss": 1.0685,
+      "step": 172980
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.5199357159256134e-05,
+      "loss": 1.1635,
+      "step": 172990
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.519820922935639e-05,
+      "loss": 1.1365,
+      "step": 173000
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.519706129945664e-05,
+      "loss": 1.0988,
+      "step": 173010
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.519591336955689e-05,
+      "loss": 1.1971,
+      "step": 173020
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.519476543965715e-05,
+      "loss": 1.1513,
+      "step": 173030
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.5193617509757404e-05,
+      "loss": 1.1728,
+      "step": 173040
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.519246957985765e-05,
+      "loss": 1.2042,
+      "step": 173050
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.51913216499579e-05,
+      "loss": 1.0937,
+      "step": 173060
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.519017372005816e-05,
+      "loss": 1.2011,
+      "step": 173070
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.51890257901584e-05,
+      "loss": 1.166,
+      "step": 173080
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.518787786025866e-05,
+      "loss": 1.1232,
+      "step": 173090
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.5186729930358916e-05,
+      "loss": 1.1226,
+      "step": 173100
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.5185582000459166e-05,
+      "loss": 1.1431,
+      "step": 173110
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.5184434070559416e-05,
+      "loss": 1.1606,
+      "step": 173120
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.518328614065967e-05,
+      "loss": 1.1286,
+      "step": 173130
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.518213821075992e-05,
+      "loss": 1.1332,
+      "step": 173140
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.518099028086018e-05,
+      "loss": 1.0698,
+      "step": 173150
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 5.517984235096043e-05,
+      "loss": 1.1796,
+      "step": 173160
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.517869442106068e-05,
+      "loss": 1.2914,
+      "step": 173170
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.5177546491160935e-05,
+      "loss": 1.0736,
+      "step": 173180
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.517639856126119e-05,
+      "loss": 1.1651,
+      "step": 173190
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.5175250631361435e-05,
+      "loss": 1.1635,
+      "step": 173200
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.517410270146169e-05,
+      "loss": 1.319,
+      "step": 173210
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.517295477156195e-05,
+      "loss": 1.0961,
+      "step": 173220
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.517180684166219e-05,
+      "loss": 1.2682,
+      "step": 173230
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.517065891176245e-05,
+      "loss": 1.2306,
+      "step": 173240
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.5169510981862705e-05,
+      "loss": 1.0834,
+      "step": 173250
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.5168363051962955e-05,
+      "loss": 1.173,
+      "step": 173260
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.5167215122063204e-05,
+      "loss": 1.0878,
+      "step": 173270
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.516606719216346e-05,
+      "loss": 1.0783,
+      "step": 173280
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.516491926226371e-05,
+      "loss": 1.357,
+      "step": 173290
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.516377133236397e-05,
+      "loss": 1.2085,
+      "step": 173300
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.516262340246422e-05,
+      "loss": 1.1454,
+      "step": 173310
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.516147547256447e-05,
+      "loss": 1.2504,
+      "step": 173320
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.5160327542664724e-05,
+      "loss": 1.1588,
+      "step": 173330
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.515917961276498e-05,
+      "loss": 1.0466,
+      "step": 173340
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.5158031682865224e-05,
+      "loss": 1.1778,
+      "step": 173350
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.515688375296548e-05,
+      "loss": 1.3244,
+      "step": 173360
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.515573582306574e-05,
+      "loss": 1.0986,
+      "step": 173370
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 5.515458789316598e-05,
+      "loss": 1.1336,
+      "step": 173380
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 5.5153439963266237e-05,
+      "loss": 1.2124,
+      "step": 173390
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 5.515229203336649e-05,
+      "loss": 1.311,
+      "step": 173400
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 5.515114410346674e-05,
+      "loss": 1.0693,
+      "step": 173410
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 5.514999617356699e-05,
+      "loss": 1.1535,
+      "step": 173420
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 5.514884824366725e-05,
+      "loss": 1.1414,
+      "step": 173430
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 5.51477003137675e-05,
+      "loss": 1.2855,
+      "step": 173440
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 5.5146552383867756e-05,
+      "loss": 1.1856,
+      "step": 173450
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 5.5145404453968006e-05,
+      "loss": 1.1446,
+      "step": 173460
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 5.5144256524068256e-05,
+      "loss": 1.1849,
+      "step": 173470
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 5.514310859416851e-05,
+      "loss": 1.1141,
+      "step": 173480
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 5.514196066426877e-05,
+      "loss": 1.2445,
+      "step": 173490
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 5.514081273436901e-05,
+      "loss": 1.1922,
+      "step": 173500
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 5.513966480446927e-05,
+      "loss": 1.1545,
+      "step": 173510
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 5.5138516874569525e-05,
+      "loss": 1.2672,
+      "step": 173520
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 5.513736894466977e-05,
+      "loss": 1.1177,
+      "step": 173530
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 5.5136221014770025e-05,
+      "loss": 1.1303,
+      "step": 173540
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 5.513507308487028e-05,
+      "loss": 1.1879,
+      "step": 173550
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 5.513392515497053e-05,
+      "loss": 1.3017,
+      "step": 173560
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 5.513277722507078e-05,
+      "loss": 1.1731,
+      "step": 173570
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 5.513162929517104e-05,
+      "loss": 1.0935,
+      "step": 173580
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 5.513048136527129e-05,
+      "loss": 1.1502,
+      "step": 173590
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 5.5129333435371545e-05,
+      "loss": 1.1784,
+      "step": 173600
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 5.5128185505471794e-05,
+      "loss": 1.1158,
+      "step": 173610
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 5.5127037575572044e-05,
+      "loss": 1.1432,
+      "step": 173620
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 5.51258896456723e-05,
+      "loss": 1.1086,
+      "step": 173630
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 5.512474171577256e-05,
+      "loss": 1.2446,
+      "step": 173640
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 5.51235937858728e-05,
+      "loss": 1.2621,
+      "step": 173650
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 5.512244585597306e-05,
+      "loss": 1.1986,
+      "step": 173660
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 5.5121297926073314e-05,
+      "loss": 1.2498,
+      "step": 173670
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 5.512014999617356e-05,
+      "loss": 1.2435,
+      "step": 173680
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 5.5119002066273814e-05,
+      "loss": 1.0631,
+      "step": 173690
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 5.511785413637407e-05,
+      "loss": 1.0199,
+      "step": 173700
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 5.511670620647432e-05,
+      "loss": 1.1571,
+      "step": 173710
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 5.511555827657457e-05,
+      "loss": 1.1689,
+      "step": 173720
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 5.5114410346674827e-05,
+      "loss": 1.2893,
+      "step": 173730
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 5.5113262416775076e-05,
+      "loss": 1.25,
+      "step": 173740
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 5.511211448687533e-05,
+      "loss": 1.1788,
+      "step": 173750
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 5.511096655697558e-05,
+      "loss": 1.0244,
+      "step": 173760
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 5.510981862707583e-05,
+      "loss": 1.2043,
+      "step": 173770
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 5.510867069717609e-05,
+      "loss": 1.1961,
+      "step": 173780
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 5.5107522767276346e-05,
+      "loss": 1.0633,
+      "step": 173790
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 5.510637483737659e-05,
+      "loss": 1.1663,
+      "step": 173800
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 5.5105226907476846e-05,
+      "loss": 1.1495,
+      "step": 173810
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 5.51040789775771e-05,
+      "loss": 1.1194,
+      "step": 173820
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 5.5102931047677345e-05,
+      "loss": 1.2103,
+      "step": 173830
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 5.51017831177776e-05,
+      "loss": 1.2711,
+      "step": 173840
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 5.510063518787786e-05,
+      "loss": 1.166,
+      "step": 173850
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 5.509948725797811e-05,
+      "loss": 1.2571,
+      "step": 173860
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 5.509833932807836e-05,
+      "loss": 1.1326,
+      "step": 173870
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 5.5097191398178615e-05,
+      "loss": 1.2638,
+      "step": 173880
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 5.5096043468278865e-05,
+      "loss": 1.2219,
+      "step": 173890
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 5.509489553837912e-05,
+      "loss": 1.2227,
+      "step": 173900
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 5.509374760847937e-05,
+      "loss": 1.0984,
+      "step": 173910
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 5.509259967857962e-05,
+      "loss": 1.1462,
+      "step": 173920
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 5.509145174867988e-05,
+      "loss": 1.2168,
+      "step": 173930
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 5.5090303818780134e-05,
+      "loss": 1.0672,
+      "step": 173940
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 5.508915588888038e-05,
+      "loss": 1.0931,
+      "step": 173950
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 5.5088007958980634e-05,
+      "loss": 1.0464,
+      "step": 173960
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 5.508686002908089e-05,
+      "loss": 1.1776,
+      "step": 173970
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 5.5085712099181134e-05,
+      "loss": 1.1976,
+      "step": 173980
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 5.508456416928139e-05,
+      "loss": 1.1734,
+      "step": 173990
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 5.508341623938165e-05,
+      "loss": 1.2851,
+      "step": 174000
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 5.50822683094819e-05,
+      "loss": 1.2228,
+      "step": 174010
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 5.508112037958215e-05,
+      "loss": 1.2211,
+      "step": 174020
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 5.5079972449682403e-05,
+      "loss": 1.1663,
+      "step": 174030
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 5.507882451978265e-05,
+      "loss": 1.2334,
+      "step": 174040
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 5.507767658988291e-05,
+      "loss": 1.2591,
+      "step": 174050
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 5.507652865998316e-05,
+      "loss": 1.1839,
+      "step": 174060
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 5.507538073008341e-05,
+      "loss": 1.2419,
+      "step": 174070
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 5.5074232800183666e-05,
+      "loss": 1.2836,
+      "step": 174080
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 5.507308487028392e-05,
+      "loss": 1.184,
+      "step": 174090
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 5.5071936940384166e-05,
+      "loss": 1.1147,
+      "step": 174100
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 5.507078901048442e-05,
+      "loss": 1.2092,
+      "step": 174110
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 5.506964108058468e-05,
+      "loss": 1.1652,
+      "step": 174120
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 5.506849315068492e-05,
+      "loss": 1.1255,
+      "step": 174130
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 5.506734522078518e-05,
+      "loss": 1.1752,
+      "step": 174140
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 5.506619729088543e-05,
+      "loss": 1.1514,
+      "step": 174150
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 5.5065049360985685e-05,
+      "loss": 1.0643,
+      "step": 174160
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 5.5063901431085935e-05,
+      "loss": 1.0425,
+      "step": 174170
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 5.5062753501186185e-05,
+      "loss": 1.1856,
+      "step": 174180
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 5.506160557128644e-05,
+      "loss": 1.1753,
+      "step": 174190
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 5.50604576413867e-05,
+      "loss": 1.2854,
+      "step": 174200
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 5.505930971148694e-05,
+      "loss": 1.3656,
+      "step": 174210
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 5.50581617815872e-05,
+      "loss": 1.2905,
+      "step": 174220
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 5.5057013851687455e-05,
+      "loss": 1.2065,
+      "step": 174230
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 5.50558659217877e-05,
+      "loss": 1.2709,
+      "step": 174240
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 5.5054717991887954e-05,
+      "loss": 1.2708,
+      "step": 174250
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 5.505357006198821e-05,
+      "loss": 1.1669,
+      "step": 174260
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 5.505242213208846e-05,
+      "loss": 1.2136,
+      "step": 174270
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 5.505127420218871e-05,
+      "loss": 1.1585,
+      "step": 174280
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 5.505012627228897e-05,
+      "loss": 1.0077,
+      "step": 174290
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 5.504897834238922e-05,
+      "loss": 1.1209,
+      "step": 174300
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 5.5047830412489474e-05,
+      "loss": 1.1178,
+      "step": 174310
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 5.5046682482589724e-05,
+      "loss": 1.2815,
+      "step": 174320
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 5.5045534552689974e-05,
+      "loss": 1.2429,
+      "step": 174330
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 5.504438662279023e-05,
+      "loss": 1.2351,
+      "step": 174340
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 5.504323869289049e-05,
+      "loss": 1.0933,
+      "step": 174350
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 5.504209076299073e-05,
+      "loss": 1.1894,
+      "step": 174360
+    },
+    {
+      "epoch": 8.0,
+      "eval_cer": 0.6695213154589411,
+      "eval_em": 0.007357212003872217,
+      "eval_f1": 0.007357212003872217,
+      "eval_loss": 1.200652837753296,
+      "eval_runtime": 1727.5263,
+      "eval_samples_per_second": 5.98,
+      "eval_steps_per_second": 0.748,
+      "eval_wer": 0.9926427879961278,
+      "step": 174362
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 5.5040942833090987e-05,
+      "loss": 1.1423,
+      "step": 174370
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 5.503979490319124e-05,
+      "loss": 1.0582,
+      "step": 174380
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 5.5038646973291486e-05,
+      "loss": 1.1873,
+      "step": 174390
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 5.503749904339174e-05,
+      "loss": 1.1066,
+      "step": 174400
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 5.5036351113492e-05,
+      "loss": 1.1161,
+      "step": 174410
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 5.503520318359225e-05,
+      "loss": 1.2272,
+      "step": 174420
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 5.50340552536925e-05,
+      "loss": 1.1305,
+      "step": 174430
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 5.5032907323792756e-05,
+      "loss": 1.0432,
+      "step": 174440
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 5.5031759393893006e-05,
+      "loss": 1.1827,
+      "step": 174450
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 5.503061146399326e-05,
+      "loss": 1.0783,
+      "step": 174460
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 5.502946353409351e-05,
+      "loss": 1.0285,
+      "step": 174470
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.502831560419376e-05,
+      "loss": 1.1372,
+      "step": 174480
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.502716767429402e-05,
+      "loss": 1.1907,
+      "step": 174490
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.5026019744394275e-05,
+      "loss": 1.1759,
+      "step": 174500
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.502487181449452e-05,
+      "loss": 1.1651,
+      "step": 174510
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.5023723884594775e-05,
+      "loss": 1.1865,
+      "step": 174520
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.502257595469503e-05,
+      "loss": 1.1363,
+      "step": 174530
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.5021428024795275e-05,
+      "loss": 1.15,
+      "step": 174540
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.502028009489553e-05,
+      "loss": 1.0573,
+      "step": 174550
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.501913216499579e-05,
+      "loss": 1.1364,
+      "step": 174560
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.501798423509604e-05,
+      "loss": 1.1398,
+      "step": 174570
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.501683630519629e-05,
+      "loss": 1.019,
+      "step": 174580
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.5015688375296544e-05,
+      "loss": 1.1291,
+      "step": 174590
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.5014540445396794e-05,
+      "loss": 1.069,
+      "step": 174600
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.501339251549705e-05,
+      "loss": 1.1329,
+      "step": 174610
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.50122445855973e-05,
+      "loss": 1.1264,
+      "step": 174620
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.501109665569755e-05,
+      "loss": 1.2719,
+      "step": 174630
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.500994872579781e-05,
+      "loss": 1.39,
+      "step": 174640
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.5008800795898064e-05,
+      "loss": 1.0668,
+      "step": 174650
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.500765286599831e-05,
+      "loss": 1.1558,
+      "step": 174660
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.5006504936098564e-05,
+      "loss": 1.1367,
+      "step": 174670
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 5.500535700619882e-05,
+      "loss": 1.0942,
+      "step": 174680
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.500420907629906e-05,
+      "loss": 1.1947,
+      "step": 174690
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.500306114639932e-05,
+      "loss": 1.1128,
+      "step": 174700
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.5001913216499576e-05,
+      "loss": 1.197,
+      "step": 174710
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.5000765286599826e-05,
+      "loss": 1.0361,
+      "step": 174720
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.4999617356700076e-05,
+      "loss": 1.1209,
+      "step": 174730
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.499846942680033e-05,
+      "loss": 1.0069,
+      "step": 174740
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.499732149690058e-05,
+      "loss": 1.0239,
+      "step": 174750
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.499617356700084e-05,
+      "loss": 1.0096,
+      "step": 174760
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.499502563710109e-05,
+      "loss": 1.1212,
+      "step": 174770
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.499387770720134e-05,
+      "loss": 1.1375,
+      "step": 174780
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.4992729777301596e-05,
+      "loss": 1.1411,
+      "step": 174790
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.499158184740185e-05,
+      "loss": 1.1021,
+      "step": 174800
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.4990433917502095e-05,
+      "loss": 1.1548,
+      "step": 174810
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.498928598760235e-05,
+      "loss": 1.078,
+      "step": 174820
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.498813805770261e-05,
+      "loss": 1.1527,
+      "step": 174830
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.498699012780285e-05,
+      "loss": 1.0301,
+      "step": 174840
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.498584219790311e-05,
+      "loss": 0.9162,
+      "step": 174850
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.4984694268003365e-05,
+      "loss": 1.1086,
+      "step": 174860
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.4983546338103615e-05,
+      "loss": 1.1384,
+      "step": 174870
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.4982398408203865e-05,
+      "loss": 1.1609,
+      "step": 174880
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.498125047830412e-05,
+      "loss": 1.2054,
+      "step": 174890
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 5.498010254840437e-05,
+      "loss": 0.986,
+      "step": 174900
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.497895461850463e-05,
+      "loss": 1.2671,
+      "step": 174910
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.497780668860488e-05,
+      "loss": 1.1238,
+      "step": 174920
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.497665875870513e-05,
+      "loss": 1.1177,
+      "step": 174930
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.4975510828805384e-05,
+      "loss": 1.0864,
+      "step": 174940
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.497436289890564e-05,
+      "loss": 1.3269,
+      "step": 174950
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.4973214969005884e-05,
+      "loss": 1.0658,
+      "step": 174960
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.497206703910614e-05,
+      "loss": 1.2169,
+      "step": 174970
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.49709191092064e-05,
+      "loss": 1.1429,
+      "step": 174980
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.496977117930664e-05,
+      "loss": 1.3177,
+      "step": 174990
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.49686232494069e-05,
+      "loss": 1.2773,
+      "step": 175000
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.4967475319507153e-05,
+      "loss": 1.0649,
+      "step": 175010
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.49663273896074e-05,
+      "loss": 0.9843,
+      "step": 175020
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.496517945970765e-05,
+      "loss": 1.2938,
+      "step": 175030
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.496403152980791e-05,
+      "loss": 1.0956,
+      "step": 175040
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.496288359990816e-05,
+      "loss": 1.1326,
+      "step": 175050
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.4961735670008416e-05,
+      "loss": 1.0586,
+      "step": 175060
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.4960587740108666e-05,
+      "loss": 1.1377,
+      "step": 175070
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.4959439810208916e-05,
+      "loss": 1.1363,
+      "step": 175080
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.495829188030917e-05,
+      "loss": 1.1105,
+      "step": 175090
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.495714395040943e-05,
+      "loss": 1.2101,
+      "step": 175100
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.495599602050967e-05,
+      "loss": 1.0968,
+      "step": 175110
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 5.495484809060993e-05,
+      "loss": 1.0928,
+      "step": 175120
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.4953700160710186e-05,
+      "loss": 1.0036,
+      "step": 175130
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.495255223081043e-05,
+      "loss": 1.1335,
+      "step": 175140
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.4951404300910685e-05,
+      "loss": 1.0435,
+      "step": 175150
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.495025637101094e-05,
+      "loss": 1.1907,
+      "step": 175160
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.494910844111119e-05,
+      "loss": 1.1851,
+      "step": 175170
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.494796051121144e-05,
+      "loss": 1.0044,
+      "step": 175180
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.49468125813117e-05,
+      "loss": 1.1796,
+      "step": 175190
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.494566465141195e-05,
+      "loss": 1.1727,
+      "step": 175200
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.4944516721512205e-05,
+      "loss": 1.0785,
+      "step": 175210
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.4943368791612455e-05,
+      "loss": 1.1374,
+      "step": 175220
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.4942220861712704e-05,
+      "loss": 1.0428,
+      "step": 175230
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.494107293181296e-05,
+      "loss": 1.1177,
+      "step": 175240
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.493992500191322e-05,
+      "loss": 1.1542,
+      "step": 175250
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.493877707201346e-05,
+      "loss": 1.0905,
+      "step": 175260
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.493762914211372e-05,
+      "loss": 1.2396,
+      "step": 175270
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.4936481212213974e-05,
+      "loss": 1.0999,
+      "step": 175280
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.493533328231422e-05,
+      "loss": 1.09,
+      "step": 175290
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.4934185352414474e-05,
+      "loss": 1.1187,
+      "step": 175300
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.493303742251473e-05,
+      "loss": 1.0394,
+      "step": 175310
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.493188949261498e-05,
+      "loss": 1.2293,
+      "step": 175320
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.493074156271523e-05,
+      "loss": 1.102,
+      "step": 175330
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.492959363281549e-05,
+      "loss": 1.1322,
+      "step": 175340
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.4928445702915737e-05,
+      "loss": 1.042,
+      "step": 175350
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.492729777301599e-05,
+      "loss": 1.0791,
+      "step": 175360
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.492614984311624e-05,
+      "loss": 1.1798,
+      "step": 175370
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.492500191321649e-05,
+      "loss": 1.0936,
+      "step": 175380
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.492385398331675e-05,
+      "loss": 1.0904,
+      "step": 175390
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.4922706053417006e-05,
+      "loss": 1.0948,
+      "step": 175400
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.492155812351725e-05,
+      "loss": 1.1631,
+      "step": 175410
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.4920410193617506e-05,
+      "loss": 1.0932,
+      "step": 175420
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.491926226371776e-05,
+      "loss": 1.128,
+      "step": 175430
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.4918114333818006e-05,
+      "loss": 1.1448,
+      "step": 175440
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.491696640391826e-05,
+      "loss": 1.1635,
+      "step": 175450
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.491581847401852e-05,
+      "loss": 1.1946,
+      "step": 175460
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.491467054411877e-05,
+      "loss": 1.1555,
+      "step": 175470
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.491352261421902e-05,
+      "loss": 1.0458,
+      "step": 175480
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.4912374684319275e-05,
+      "loss": 1.1606,
+      "step": 175490
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.4911226754419525e-05,
+      "loss": 1.0799,
+      "step": 175500
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.491007882451978e-05,
+      "loss": 1.2527,
+      "step": 175510
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.4908930894620025e-05,
+      "loss": 1.1421,
+      "step": 175520
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.490778296472028e-05,
+      "loss": 1.0984,
+      "step": 175530
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.490663503482054e-05,
+      "loss": 1.1593,
+      "step": 175540
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.490548710492078e-05,
+      "loss": 1.193,
+      "step": 175550
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 5.490433917502104e-05,
+      "loss": 0.9741,
+      "step": 175560
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.4903191245121294e-05,
+      "loss": 1.2368,
+      "step": 175570
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.4902043315221544e-05,
+      "loss": 1.0231,
+      "step": 175580
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.4900895385321794e-05,
+      "loss": 1.2176,
+      "step": 175590
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.489974745542205e-05,
+      "loss": 1.116,
+      "step": 175600
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.48985995255223e-05,
+      "loss": 1.221,
+      "step": 175610
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.489745159562256e-05,
+      "loss": 1.0713,
+      "step": 175620
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.489630366572281e-05,
+      "loss": 1.1568,
+      "step": 175630
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.489515573582306e-05,
+      "loss": 1.1755,
+      "step": 175640
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.4894007805923314e-05,
+      "loss": 1.074,
+      "step": 175650
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.489285987602357e-05,
+      "loss": 1.0465,
+      "step": 175660
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.489171194612381e-05,
+      "loss": 1.2169,
+      "step": 175670
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.489056401622407e-05,
+      "loss": 1.1048,
+      "step": 175680
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.4889416086324326e-05,
+      "loss": 1.1569,
+      "step": 175690
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.488826815642457e-05,
+      "loss": 1.1506,
+      "step": 175700
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.4887120226524826e-05,
+      "loss": 1.1805,
+      "step": 175710
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.488597229662508e-05,
+      "loss": 1.0644,
+      "step": 175720
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.488482436672533e-05,
+      "loss": 1.0889,
+      "step": 175730
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.488367643682558e-05,
+      "loss": 0.9821,
+      "step": 175740
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.488252850692584e-05,
+      "loss": 1.1009,
+      "step": 175750
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.488138057702609e-05,
+      "loss": 1.0918,
+      "step": 175760
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 5.4880232647126346e-05,
+      "loss": 1.2457,
+      "step": 175770
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.4879084717226596e-05,
+      "loss": 1.1027,
+      "step": 175780
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.4877936787326845e-05,
+      "loss": 1.1541,
+      "step": 175790
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.48767888574271e-05,
+      "loss": 1.0434,
+      "step": 175800
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.487564092752736e-05,
+      "loss": 1.133,
+      "step": 175810
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.48744929976276e-05,
+      "loss": 1.1673,
+      "step": 175820
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.487334506772786e-05,
+      "loss": 1.0881,
+      "step": 175830
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.4872197137828115e-05,
+      "loss": 1.175,
+      "step": 175840
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.487104920792836e-05,
+      "loss": 1.0858,
+      "step": 175850
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.4869901278028615e-05,
+      "loss": 1.1148,
+      "step": 175860
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.486875334812887e-05,
+      "loss": 1.21,
+      "step": 175870
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.486760541822912e-05,
+      "loss": 1.0812,
+      "step": 175880
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.486645748832937e-05,
+      "loss": 1.1517,
+      "step": 175890
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.486530955842963e-05,
+      "loss": 1.1287,
+      "step": 175900
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.486416162852988e-05,
+      "loss": 1.0648,
+      "step": 175910
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.4863013698630134e-05,
+      "loss": 1.0886,
+      "step": 175920
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.4861865768730384e-05,
+      "loss": 1.1655,
+      "step": 175930
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.4860717838830634e-05,
+      "loss": 1.0901,
+      "step": 175940
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.485956990893089e-05,
+      "loss": 1.1134,
+      "step": 175950
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.485842197903115e-05,
+      "loss": 1.1726,
+      "step": 175960
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.485727404913139e-05,
+      "loss": 1.2598,
+      "step": 175970
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.485612611923165e-05,
+      "loss": 1.0969,
+      "step": 175980
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 5.4854978189331903e-05,
+      "loss": 1.0406,
+      "step": 175990
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.4853830259432147e-05,
+      "loss": 1.0039,
+      "step": 176000
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.48526823295324e-05,
+      "loss": 1.0681,
+      "step": 176010
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.485153439963266e-05,
+      "loss": 1.117,
+      "step": 176020
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.485038646973291e-05,
+      "loss": 1.1104,
+      "step": 176030
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.484923853983316e-05,
+      "loss": 1.1659,
+      "step": 176040
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.4848090609933416e-05,
+      "loss": 1.0321,
+      "step": 176050
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.4846942680033666e-05,
+      "loss": 1.0127,
+      "step": 176060
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.484579475013392e-05,
+      "loss": 1.091,
+      "step": 176070
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.484464682023417e-05,
+      "loss": 1.1228,
+      "step": 176080
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.484349889033442e-05,
+      "loss": 1.236,
+      "step": 176090
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.484235096043468e-05,
+      "loss": 1.1198,
+      "step": 176100
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.4841203030534936e-05,
+      "loss": 1.1222,
+      "step": 176110
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.484005510063518e-05,
+      "loss": 1.0552,
+      "step": 176120
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.4838907170735435e-05,
+      "loss": 1.121,
+      "step": 176130
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.483775924083569e-05,
+      "loss": 1.1535,
+      "step": 176140
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.4836611310935935e-05,
+      "loss": 1.1053,
+      "step": 176150
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.483546338103619e-05,
+      "loss": 1.1193,
+      "step": 176160
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.483431545113645e-05,
+      "loss": 0.9765,
+      "step": 176170
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.48331675212367e-05,
+      "loss": 1.2255,
+      "step": 176180
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.483201959133695e-05,
+      "loss": 1.0872,
+      "step": 176190
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.4830871661437205e-05,
+      "loss": 1.1513,
+      "step": 176200
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.4829723731537454e-05,
+      "loss": 1.1549,
+      "step": 176210
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.482857580163771e-05,
+      "loss": 1.203,
+      "step": 176220
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.482742787173796e-05,
+      "loss": 1.1164,
+      "step": 176230
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.482627994183821e-05,
+      "loss": 0.9886,
+      "step": 176240
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.482513201193847e-05,
+      "loss": 1.1627,
+      "step": 176250
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.4823984082038724e-05,
+      "loss": 1.2839,
+      "step": 176260
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.482283615213897e-05,
+      "loss": 1.1653,
+      "step": 176270
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.4821688222239224e-05,
+      "loss": 1.0995,
+      "step": 176280
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.482054029233948e-05,
+      "loss": 1.1449,
+      "step": 176290
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.4819392362439723e-05,
+      "loss": 1.0503,
+      "step": 176300
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.481824443253998e-05,
+      "loss": 1.0461,
+      "step": 176310
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.481709650264024e-05,
+      "loss": 1.0071,
+      "step": 176320
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.4815948572740487e-05,
+      "loss": 1.19,
+      "step": 176330
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.4814800642840736e-05,
+      "loss": 1.1649,
+      "step": 176340
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.481365271294099e-05,
+      "loss": 1.0579,
+      "step": 176350
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.481250478304124e-05,
+      "loss": 1.0709,
+      "step": 176360
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.48113568531415e-05,
+      "loss": 1.0948,
+      "step": 176370
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.481020892324175e-05,
+      "loss": 1.0261,
+      "step": 176380
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.4809060993342e-05,
+      "loss": 1.2379,
+      "step": 176390
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.4807913063442256e-05,
+      "loss": 1.0895,
+      "step": 176400
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.480676513354251e-05,
+      "loss": 1.0617,
+      "step": 176410
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.4805617203642756e-05,
+      "loss": 1.1646,
+      "step": 176420
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.480446927374301e-05,
+      "loss": 1.1387,
+      "step": 176430
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.480332134384327e-05,
+      "loss": 1.2122,
+      "step": 176440
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.480217341394351e-05,
+      "loss": 1.0647,
+      "step": 176450
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.480102548404377e-05,
+      "loss": 1.2578,
+      "step": 176460
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.4799877554144025e-05,
+      "loss": 1.132,
+      "step": 176470
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.4798729624244275e-05,
+      "loss": 1.221,
+      "step": 176480
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.4797581694344525e-05,
+      "loss": 1.0488,
+      "step": 176490
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.479643376444478e-05,
+      "loss": 1.1205,
+      "step": 176500
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.479528583454503e-05,
+      "loss": 1.0078,
+      "step": 176510
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.479413790464529e-05,
+      "loss": 1.1323,
+      "step": 176520
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.479298997474554e-05,
+      "loss": 1.1351,
+      "step": 176530
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.479184204484579e-05,
+      "loss": 1.1335,
+      "step": 176540
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.4790694114946044e-05,
+      "loss": 1.1596,
+      "step": 176550
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.47895461850463e-05,
+      "loss": 1.2192,
+      "step": 176560
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.4788398255146544e-05,
+      "loss": 1.0814,
+      "step": 176570
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.47872503252468e-05,
+      "loss": 0.9764,
+      "step": 176580
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.478610239534706e-05,
+      "loss": 1.0871,
+      "step": 176590
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.47849544654473e-05,
+      "loss": 1.1064,
+      "step": 176600
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.478380653554756e-05,
+      "loss": 1.146,
+      "step": 176610
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.4782658605647814e-05,
+      "loss": 1.0667,
+      "step": 176620
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.4781510675748064e-05,
+      "loss": 1.0125,
+      "step": 176630
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.478036274584831e-05,
+      "loss": 1.1703,
+      "step": 176640
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 5.477921481594857e-05,
+      "loss": 1.1384,
+      "step": 176650
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.477806688604882e-05,
+      "loss": 1.0885,
+      "step": 176660
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.4776918956149076e-05,
+      "loss": 1.2089,
+      "step": 176670
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.4775771026249326e-05,
+      "loss": 1.1507,
+      "step": 176680
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.4774623096349576e-05,
+      "loss": 1.2066,
+      "step": 176690
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.477347516644983e-05,
+      "loss": 1.1208,
+      "step": 176700
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.477232723655009e-05,
+      "loss": 1.1919,
+      "step": 176710
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.477117930665033e-05,
+      "loss": 1.1244,
+      "step": 176720
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.477003137675059e-05,
+      "loss": 1.1283,
+      "step": 176730
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.4768883446850846e-05,
+      "loss": 1.1331,
+      "step": 176740
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.476773551695109e-05,
+      "loss": 1.3055,
+      "step": 176750
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.4766587587051346e-05,
+      "loss": 1.075,
+      "step": 176760
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.47654396571516e-05,
+      "loss": 1.1431,
+      "step": 176770
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.476429172725185e-05,
+      "loss": 1.1028,
+      "step": 176780
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.47631437973521e-05,
+      "loss": 1.1494,
+      "step": 176790
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.476199586745236e-05,
+      "loss": 1.0325,
+      "step": 176800
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.476084793755261e-05,
+      "loss": 1.149,
+      "step": 176810
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.4759700007652865e-05,
+      "loss": 1.1798,
+      "step": 176820
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.4758552077753115e-05,
+      "loss": 1.0813,
+      "step": 176830
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.4757404147853365e-05,
+      "loss": 1.2573,
+      "step": 176840
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.475625621795362e-05,
+      "loss": 1.1707,
+      "step": 176850
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.475510828805388e-05,
+      "loss": 1.1459,
+      "step": 176860
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.475396035815412e-05,
+      "loss": 1.0624,
+      "step": 176870
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.475281242825438e-05,
+      "loss": 1.0693,
+      "step": 176880
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.475166449835463e-05,
+      "loss": 1.1145,
+      "step": 176890
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.475051656845488e-05,
+      "loss": 1.1294,
+      "step": 176900
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.4749368638555134e-05,
+      "loss": 1.0045,
+      "step": 176910
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.4748220708655384e-05,
+      "loss": 1.2217,
+      "step": 176920
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.474707277875564e-05,
+      "loss": 1.0715,
+      "step": 176930
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.474592484885589e-05,
+      "loss": 1.2492,
+      "step": 176940
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.474477691895614e-05,
+      "loss": 1.1382,
+      "step": 176950
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.47436289890564e-05,
+      "loss": 1.2782,
+      "step": 176960
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.4742481059156653e-05,
+      "loss": 1.0513,
+      "step": 176970
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.4741333129256897e-05,
+      "loss": 1.1199,
+      "step": 176980
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.474018519935715e-05,
+      "loss": 1.1256,
+      "step": 176990
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.473903726945741e-05,
+      "loss": 1.1227,
+      "step": 177000
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.473788933955765e-05,
+      "loss": 1.0712,
+      "step": 177010
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.473674140965791e-05,
+      "loss": 1.0814,
+      "step": 177020
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.4735593479758166e-05,
+      "loss": 1.102,
+      "step": 177030
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.4734445549858416e-05,
+      "loss": 1.2606,
+      "step": 177040
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.4733297619958666e-05,
+      "loss": 1.2162,
+      "step": 177050
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.473214969005892e-05,
+      "loss": 1.133,
+      "step": 177060
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.473100176015917e-05,
+      "loss": 1.1503,
+      "step": 177070
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.472985383025943e-05,
+      "loss": 1.0656,
+      "step": 177080
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.472870590035968e-05,
+      "loss": 1.2599,
+      "step": 177090
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.472755797045993e-05,
+      "loss": 1.2336,
+      "step": 177100
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.4726410040560185e-05,
+      "loss": 1.0214,
+      "step": 177110
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.472526211066044e-05,
+      "loss": 1.2364,
+      "step": 177120
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.4724114180760685e-05,
+      "loss": 1.1638,
+      "step": 177130
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.472296625086094e-05,
+      "loss": 1.2671,
+      "step": 177140
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.47218183209612e-05,
+      "loss": 1.154,
+      "step": 177150
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.472067039106144e-05,
+      "loss": 0.9837,
+      "step": 177160
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.47195224611617e-05,
+      "loss": 1.0918,
+      "step": 177170
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.4718374531261955e-05,
+      "loss": 1.1478,
+      "step": 177180
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.4717226601362204e-05,
+      "loss": 1.1675,
+      "step": 177190
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.4716078671462454e-05,
+      "loss": 1.1575,
+      "step": 177200
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.471493074156271e-05,
+      "loss": 1.196,
+      "step": 177210
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.471378281166296e-05,
+      "loss": 1.0836,
+      "step": 177220
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.471263488176322e-05,
+      "loss": 1.1207,
+      "step": 177230
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.471148695186347e-05,
+      "loss": 1.1862,
+      "step": 177240
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.471033902196372e-05,
+      "loss": 1.1656,
+      "step": 177250
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.4709191092063974e-05,
+      "loss": 1.0212,
+      "step": 177260
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.470804316216423e-05,
+      "loss": 1.1077,
+      "step": 177270
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.4706895232264473e-05,
+      "loss": 1.0332,
+      "step": 177280
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.470574730236473e-05,
+      "loss": 1.1096,
+      "step": 177290
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 5.470459937246499e-05,
+      "loss": 0.992,
+      "step": 177300
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.470345144256523e-05,
+      "loss": 1.109,
+      "step": 177310
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.4702303512665486e-05,
+      "loss": 1.153,
+      "step": 177320
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.470115558276574e-05,
+      "loss": 1.0369,
+      "step": 177330
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.470000765286599e-05,
+      "loss": 1.0952,
+      "step": 177340
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.469885972296624e-05,
+      "loss": 1.1014,
+      "step": 177350
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.46977117930665e-05,
+      "loss": 1.1796,
+      "step": 177360
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.469656386316675e-05,
+      "loss": 1.0104,
+      "step": 177370
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.4695415933267006e-05,
+      "loss": 1.2288,
+      "step": 177380
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.4694268003367256e-05,
+      "loss": 1.125,
+      "step": 177390
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.4693120073467506e-05,
+      "loss": 1.1466,
+      "step": 177400
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.469197214356776e-05,
+      "loss": 1.2279,
+      "step": 177410
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.469082421366802e-05,
+      "loss": 1.0762,
+      "step": 177420
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.468967628376826e-05,
+      "loss": 1.1687,
+      "step": 177430
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.468852835386852e-05,
+      "loss": 1.0398,
+      "step": 177440
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.4687380423968775e-05,
+      "loss": 1.1474,
+      "step": 177450
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.468623249406902e-05,
+      "loss": 1.0578,
+      "step": 177460
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.4685084564169275e-05,
+      "loss": 1.0617,
+      "step": 177470
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.468393663426953e-05,
+      "loss": 1.1816,
+      "step": 177480
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.468278870436978e-05,
+      "loss": 1.125,
+      "step": 177490
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.468164077447003e-05,
+      "loss": 1.0375,
+      "step": 177500
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.468049284457029e-05,
+      "loss": 1.0656,
+      "step": 177510
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.467934491467054e-05,
+      "loss": 1.0062,
+      "step": 177520
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.4678196984770794e-05,
+      "loss": 1.1025,
+      "step": 177530
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.4677049054871044e-05,
+      "loss": 1.1372,
+      "step": 177540
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.4675901124971294e-05,
+      "loss": 1.1179,
+      "step": 177550
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.467475319507155e-05,
+      "loss": 1.1218,
+      "step": 177560
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.467360526517181e-05,
+      "loss": 1.1671,
+      "step": 177570
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.467245733527205e-05,
+      "loss": 1.016,
+      "step": 177580
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.467130940537231e-05,
+      "loss": 1.0541,
+      "step": 177590
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.4670161475472564e-05,
+      "loss": 1.1014,
+      "step": 177600
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.466901354557281e-05,
+      "loss": 1.1797,
+      "step": 177610
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.466786561567306e-05,
+      "loss": 1.094,
+      "step": 177620
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.466671768577332e-05,
+      "loss": 1.1847,
+      "step": 177630
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.466556975587357e-05,
+      "loss": 1.0805,
+      "step": 177640
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.466442182597382e-05,
+      "loss": 1.1166,
+      "step": 177650
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.4663273896074076e-05,
+      "loss": 1.0878,
+      "step": 177660
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.4662125966174326e-05,
+      "loss": 1.1178,
+      "step": 177670
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.466097803627458e-05,
+      "loss": 1.1716,
+      "step": 177680
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.465983010637483e-05,
+      "loss": 1.1168,
+      "step": 177690
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.465868217647508e-05,
+      "loss": 1.0967,
+      "step": 177700
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.465753424657534e-05,
+      "loss": 1.1253,
+      "step": 177710
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.4656386316675596e-05,
+      "loss": 1.0278,
+      "step": 177720
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.465523838677584e-05,
+      "loss": 1.2095,
+      "step": 177730
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 5.4654090456876096e-05,
+      "loss": 1.1229,
+      "step": 177740
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.465294252697635e-05,
+      "loss": 1.1381,
+      "step": 177750
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.4651794597076595e-05,
+      "loss": 1.0607,
+      "step": 177760
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.465064666717685e-05,
+      "loss": 1.2342,
+      "step": 177770
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.464949873727711e-05,
+      "loss": 1.0804,
+      "step": 177780
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.464835080737736e-05,
+      "loss": 1.0961,
+      "step": 177790
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.464720287747761e-05,
+      "loss": 1.0326,
+      "step": 177800
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.4646054947577865e-05,
+      "loss": 1.0446,
+      "step": 177810
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.4644907017678115e-05,
+      "loss": 1.1518,
+      "step": 177820
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.464375908777837e-05,
+      "loss": 1.159,
+      "step": 177830
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.464261115787862e-05,
+      "loss": 1.2112,
+      "step": 177840
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.464146322797887e-05,
+      "loss": 1.1164,
+      "step": 177850
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.464031529807913e-05,
+      "loss": 1.0518,
+      "step": 177860
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.4639167368179384e-05,
+      "loss": 1.2021,
+      "step": 177870
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.463801943827963e-05,
+      "loss": 1.1187,
+      "step": 177880
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.4636871508379884e-05,
+      "loss": 1.2119,
+      "step": 177890
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.463572357848014e-05,
+      "loss": 1.1207,
+      "step": 177900
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.4634575648580384e-05,
+      "loss": 1.2505,
+      "step": 177910
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.463342771868064e-05,
+      "loss": 1.1926,
+      "step": 177920
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.46322797887809e-05,
+      "loss": 1.0705,
+      "step": 177930
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.463113185888115e-05,
+      "loss": 1.1075,
+      "step": 177940
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.46299839289814e-05,
+      "loss": 1.0232,
+      "step": 177950
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.462883599908165e-05,
+      "loss": 1.154,
+      "step": 177960
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.46276880691819e-05,
+      "loss": 1.0607,
+      "step": 177970
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.462654013928216e-05,
+      "loss": 1.1351,
+      "step": 177980
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.462539220938241e-05,
+      "loss": 1.1333,
+      "step": 177990
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.462424427948266e-05,
+      "loss": 1.1408,
+      "step": 178000
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.4623096349582916e-05,
+      "loss": 1.2301,
+      "step": 178010
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.462194841968317e-05,
+      "loss": 1.0933,
+      "step": 178020
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.4620800489783416e-05,
+      "loss": 1.1738,
+      "step": 178030
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.461965255988367e-05,
+      "loss": 1.1772,
+      "step": 178040
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.461850462998393e-05,
+      "loss": 1.1541,
+      "step": 178050
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.461735670008417e-05,
+      "loss": 1.2197,
+      "step": 178060
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.461620877018443e-05,
+      "loss": 1.1569,
+      "step": 178070
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.4615060840284685e-05,
+      "loss": 0.9738,
+      "step": 178080
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.4613912910384935e-05,
+      "loss": 1.1648,
+      "step": 178090
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.4612764980485185e-05,
+      "loss": 1.1565,
+      "step": 178100
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.461161705058544e-05,
+      "loss": 1.2244,
+      "step": 178110
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.461046912068569e-05,
+      "loss": 1.2047,
+      "step": 178120
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.460932119078595e-05,
+      "loss": 1.1062,
+      "step": 178130
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.46081732608862e-05,
+      "loss": 1.0966,
+      "step": 178140
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.460702533098645e-05,
+      "loss": 1.1035,
+      "step": 178150
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.4605877401086705e-05,
+      "loss": 1.2121,
+      "step": 178160
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.460472947118696e-05,
+      "loss": 1.0482,
+      "step": 178170
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.4603581541287204e-05,
+      "loss": 1.0608,
+      "step": 178180
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.460243361138746e-05,
+      "loss": 1.1049,
+      "step": 178190
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.460128568148772e-05,
+      "loss": 1.0722,
+      "step": 178200
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.460013775158796e-05,
+      "loss": 1.1307,
+      "step": 178210
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.459898982168822e-05,
+      "loss": 1.0854,
+      "step": 178220
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.4597841891788474e-05,
+      "loss": 1.1378,
+      "step": 178230
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.4596693961888724e-05,
+      "loss": 1.2051,
+      "step": 178240
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.4595546031988974e-05,
+      "loss": 1.0314,
+      "step": 178250
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.4594398102089223e-05,
+      "loss": 1.0699,
+      "step": 178260
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.459325017218948e-05,
+      "loss": 1.1305,
+      "step": 178270
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.459210224228974e-05,
+      "loss": 1.1565,
+      "step": 178280
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.459095431238998e-05,
+      "loss": 1.076,
+      "step": 178290
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.4589806382490236e-05,
+      "loss": 1.155,
+      "step": 178300
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.458865845259049e-05,
+      "loss": 1.1395,
+      "step": 178310
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.4587510522690736e-05,
+      "loss": 1.127,
+      "step": 178320
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.458636259279099e-05,
+      "loss": 1.0818,
+      "step": 178330
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.458521466289125e-05,
+      "loss": 1.13,
+      "step": 178340
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.45840667329915e-05,
+      "loss": 1.0716,
+      "step": 178350
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.458291880309175e-05,
+      "loss": 1.1275,
+      "step": 178360
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.4581770873192006e-05,
+      "loss": 1.0981,
+      "step": 178370
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.4580622943292256e-05,
+      "loss": 1.0891,
+      "step": 178380
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.457947501339251e-05,
+      "loss": 1.0847,
+      "step": 178390
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.457832708349276e-05,
+      "loss": 1.0715,
+      "step": 178400
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.457717915359301e-05,
+      "loss": 1.0937,
+      "step": 178410
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.457603122369327e-05,
+      "loss": 1.0836,
+      "step": 178420
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.4574883293793525e-05,
+      "loss": 1.0958,
+      "step": 178430
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.457373536389377e-05,
+      "loss": 1.1353,
+      "step": 178440
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.4572587433994025e-05,
+      "loss": 1.2691,
+      "step": 178450
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.457143950409428e-05,
+      "loss": 1.1736,
+      "step": 178460
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.4570291574194525e-05,
+      "loss": 1.034,
+      "step": 178470
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.456914364429478e-05,
+      "loss": 1.1555,
+      "step": 178480
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.456799571439504e-05,
+      "loss": 1.0683,
+      "step": 178490
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.456684778449529e-05,
+      "loss": 1.1107,
+      "step": 178500
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.456569985459554e-05,
+      "loss": 1.1618,
+      "step": 178510
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.4564551924695794e-05,
+      "loss": 1.016,
+      "step": 178520
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.4563403994796044e-05,
+      "loss": 1.1368,
+      "step": 178530
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.45622560648963e-05,
+      "loss": 1.0921,
+      "step": 178540
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.456110813499655e-05,
+      "loss": 1.1626,
+      "step": 178550
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.45599602050968e-05,
+      "loss": 1.1549,
+      "step": 178560
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.455881227519706e-05,
+      "loss": 1.0976,
+      "step": 178570
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.4557664345297314e-05,
+      "loss": 0.9982,
+      "step": 178580
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.455651641539756e-05,
+      "loss": 1.2394,
+      "step": 178590
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.455536848549781e-05,
+      "loss": 1.0031,
+      "step": 178600
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.455422055559807e-05,
+      "loss": 1.2146,
+      "step": 178610
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.455307262569831e-05,
+      "loss": 1.2243,
+      "step": 178620
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.455192469579857e-05,
+      "loss": 1.1253,
+      "step": 178630
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.4550776765898826e-05,
+      "loss": 1.1867,
+      "step": 178640
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.4549628835999076e-05,
+      "loss": 1.1803,
+      "step": 178650
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.4548480906099326e-05,
+      "loss": 1.1333,
+      "step": 178660
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.454733297619958e-05,
+      "loss": 1.1924,
+      "step": 178670
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.454618504629983e-05,
+      "loss": 1.1286,
+      "step": 178680
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.454503711640009e-05,
+      "loss": 1.1417,
+      "step": 178690
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.454388918650034e-05,
+      "loss": 1.1696,
+      "step": 178700
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.454274125660059e-05,
+      "loss": 1.0282,
+      "step": 178710
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.4541593326700845e-05,
+      "loss": 1.1055,
+      "step": 178720
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.45404453968011e-05,
+      "loss": 1.0186,
+      "step": 178730
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.4539297466901345e-05,
+      "loss": 1.292,
+      "step": 178740
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.45381495370016e-05,
+      "loss": 1.047,
+      "step": 178750
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.453700160710186e-05,
+      "loss": 1.0862,
+      "step": 178760
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.45358536772021e-05,
+      "loss": 1.1858,
+      "step": 178770
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.453470574730236e-05,
+      "loss": 1.1016,
+      "step": 178780
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.4533557817402615e-05,
+      "loss": 1.2315,
+      "step": 178790
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.4532409887502865e-05,
+      "loss": 1.2401,
+      "step": 178800
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.4531261957603115e-05,
+      "loss": 1.2003,
+      "step": 178810
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.453011402770337e-05,
+      "loss": 0.9979,
+      "step": 178820
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.452896609780362e-05,
+      "loss": 1.0801,
+      "step": 178830
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.452781816790388e-05,
+      "loss": 1.1117,
+      "step": 178840
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.452667023800413e-05,
+      "loss": 1.1307,
+      "step": 178850
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.452552230810438e-05,
+      "loss": 1.2907,
+      "step": 178860
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.4524374378204634e-05,
+      "loss": 1.1772,
+      "step": 178870
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.452322644830489e-05,
+      "loss": 1.1784,
+      "step": 178880
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.4522078518405134e-05,
+      "loss": 1.0938,
+      "step": 178890
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.452093058850539e-05,
+      "loss": 1.1296,
+      "step": 178900
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.451978265860565e-05,
+      "loss": 1.1487,
+      "step": 178910
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.451863472870589e-05,
+      "loss": 1.1415,
+      "step": 178920
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.451748679880615e-05,
+      "loss": 1.1509,
+      "step": 178930
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.45163388689064e-05,
+      "loss": 1.249,
+      "step": 178940
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.451519093900665e-05,
+      "loss": 1.2261,
+      "step": 178950
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.45140430091069e-05,
+      "loss": 1.1519,
+      "step": 178960
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.451289507920716e-05,
+      "loss": 1.0925,
+      "step": 178970
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.451174714930741e-05,
+      "loss": 1.1962,
+      "step": 178980
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.4510599219407666e-05,
+      "loss": 1.1089,
+      "step": 178990
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.4509451289507916e-05,
+      "loss": 1.0152,
+      "step": 179000
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.4508303359608166e-05,
+      "loss": 1.0616,
+      "step": 179010
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.450715542970842e-05,
+      "loss": 1.1614,
+      "step": 179020
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.450600749980868e-05,
+      "loss": 1.105,
+      "step": 179030
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.450485956990892e-05,
+      "loss": 1.0172,
+      "step": 179040
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.450371164000918e-05,
+      "loss": 1.0872,
+      "step": 179050
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.4502563710109435e-05,
+      "loss": 1.1113,
+      "step": 179060
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.450141578020968e-05,
+      "loss": 1.273,
+      "step": 179070
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.4500267850309935e-05,
+      "loss": 1.1991,
+      "step": 179080
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.449911992041019e-05,
+      "loss": 1.1339,
+      "step": 179090
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.449797199051044e-05,
+      "loss": 1.1915,
+      "step": 179100
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.449682406061069e-05,
+      "loss": 1.0541,
+      "step": 179110
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.449567613071095e-05,
+      "loss": 1.0577,
+      "step": 179120
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.44945282008112e-05,
+      "loss": 1.1018,
+      "step": 179130
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.4493380270911455e-05,
+      "loss": 1.1396,
+      "step": 179140
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.4492232341011704e-05,
+      "loss": 1.1384,
+      "step": 179150
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.4491084411111954e-05,
+      "loss": 1.1142,
+      "step": 179160
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.448993648121221e-05,
+      "loss": 1.1938,
+      "step": 179170
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.448878855131247e-05,
+      "loss": 1.1147,
+      "step": 179180
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.448764062141271e-05,
+      "loss": 1.0486,
+      "step": 179190
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.448649269151297e-05,
+      "loss": 1.2114,
+      "step": 179200
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.4485344761613224e-05,
+      "loss": 1.0549,
+      "step": 179210
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.448419683171347e-05,
+      "loss": 1.2151,
+      "step": 179220
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.4483048901813724e-05,
+      "loss": 1.2118,
+      "step": 179230
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.448190097191398e-05,
+      "loss": 1.1352,
+      "step": 179240
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.448075304201423e-05,
+      "loss": 1.0824,
+      "step": 179250
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.447960511211448e-05,
+      "loss": 1.1517,
+      "step": 179260
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.4478457182214737e-05,
+      "loss": 1.1412,
+      "step": 179270
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.4477309252314986e-05,
+      "loss": 1.0997,
+      "step": 179280
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.447616132241524e-05,
+      "loss": 1.0203,
+      "step": 179290
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.447501339251549e-05,
+      "loss": 1.1278,
+      "step": 179300
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.447386546261574e-05,
+      "loss": 1.1431,
+      "step": 179310
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.4472717532716e-05,
+      "loss": 1.1934,
+      "step": 179320
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.4471569602816256e-05,
+      "loss": 1.0498,
+      "step": 179330
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.44704216729165e-05,
+      "loss": 1.1571,
+      "step": 179340
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.4469273743016756e-05,
+      "loss": 1.0604,
+      "step": 179350
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.446812581311701e-05,
+      "loss": 1.2421,
+      "step": 179360
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.4466977883217255e-05,
+      "loss": 1.0796,
+      "step": 179370
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.446582995331751e-05,
+      "loss": 1.0485,
+      "step": 179380
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.446468202341777e-05,
+      "loss": 1.1388,
+      "step": 179390
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.446353409351802e-05,
+      "loss": 1.041,
+      "step": 179400
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.446238616361827e-05,
+      "loss": 1.1562,
+      "step": 179410
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.4461238233718525e-05,
+      "loss": 1.1891,
+      "step": 179420
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.4460090303818775e-05,
+      "loss": 1.1783,
+      "step": 179430
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.445894237391903e-05,
+      "loss": 1.1894,
+      "step": 179440
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.445779444401928e-05,
+      "loss": 1.0278,
+      "step": 179450
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.445664651411953e-05,
+      "loss": 1.1321,
+      "step": 179460
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.445549858421979e-05,
+      "loss": 1.1886,
+      "step": 179470
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.4454350654320044e-05,
+      "loss": 1.0151,
+      "step": 179480
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.445320272442029e-05,
+      "loss": 1.0069,
+      "step": 179490
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.4452054794520544e-05,
+      "loss": 1.144,
+      "step": 179500
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.44509068646208e-05,
+      "loss": 1.1402,
+      "step": 179510
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.4449758934721044e-05,
+      "loss": 1.0971,
+      "step": 179520
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.44486110048213e-05,
+      "loss": 1.1619,
+      "step": 179530
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.444746307492156e-05,
+      "loss": 1.1187,
+      "step": 179540
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.444631514502181e-05,
+      "loss": 1.1335,
+      "step": 179550
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.444516721512206e-05,
+      "loss": 1.0271,
+      "step": 179560
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.4444019285222314e-05,
+      "loss": 1.047,
+      "step": 179570
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.444287135532256e-05,
+      "loss": 1.2051,
+      "step": 179580
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.444172342542282e-05,
+      "loss": 1.1616,
+      "step": 179590
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.444057549552307e-05,
+      "loss": 1.2687,
+      "step": 179600
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.443942756562332e-05,
+      "loss": 1.0827,
+      "step": 179610
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.4438279635723576e-05,
+      "loss": 0.96,
+      "step": 179620
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.443713170582382e-05,
+      "loss": 1.0655,
+      "step": 179630
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.4435983775924076e-05,
+      "loss": 1.1211,
+      "step": 179640
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.443483584602433e-05,
+      "loss": 1.1253,
+      "step": 179650
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.443368791612458e-05,
+      "loss": 1.058,
+      "step": 179660
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.443253998622483e-05,
+      "loss": 1.1679,
+      "step": 179670
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.443139205632509e-05,
+      "loss": 1.0787,
+      "step": 179680
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.443024412642534e-05,
+      "loss": 1.1548,
+      "step": 179690
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.4429096196525595e-05,
+      "loss": 1.0091,
+      "step": 179700
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.4427948266625845e-05,
+      "loss": 1.1136,
+      "step": 179710
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.4426800336726095e-05,
+      "loss": 1.1481,
+      "step": 179720
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.442565240682635e-05,
+      "loss": 1.1452,
+      "step": 179730
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.442450447692661e-05,
+      "loss": 1.169,
+      "step": 179740
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.442335654702685e-05,
+      "loss": 1.2295,
+      "step": 179750
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.442220861712711e-05,
+      "loss": 1.0621,
+      "step": 179760
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.4421060687227365e-05,
+      "loss": 1.1788,
+      "step": 179770
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.441991275732761e-05,
+      "loss": 1.0825,
+      "step": 179780
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.4418764827427865e-05,
+      "loss": 1.191,
+      "step": 179790
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.441761689752812e-05,
+      "loss": 1.0755,
+      "step": 179800
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.441646896762837e-05,
+      "loss": 1.1132,
+      "step": 179810
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.441532103772862e-05,
+      "loss": 1.2053,
+      "step": 179820
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.441417310782888e-05,
+      "loss": 1.1067,
+      "step": 179830
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.441302517792913e-05,
+      "loss": 1.1699,
+      "step": 179840
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.4411877248029384e-05,
+      "loss": 1.1443,
+      "step": 179850
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.4410729318129634e-05,
+      "loss": 1.1748,
+      "step": 179860
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.4409581388229884e-05,
+      "loss": 1.1557,
+      "step": 179870
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.440843345833014e-05,
+      "loss": 1.1492,
+      "step": 179880
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.44072855284304e-05,
+      "loss": 1.0928,
+      "step": 179890
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.440613759853064e-05,
+      "loss": 1.1081,
+      "step": 179900
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.44049896686309e-05,
+      "loss": 1.1414,
+      "step": 179910
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.440384173873115e-05,
+      "loss": 1.1414,
+      "step": 179920
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.4402693808831396e-05,
+      "loss": 1.1534,
+      "step": 179930
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.440154587893165e-05,
+      "loss": 1.1345,
+      "step": 179940
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.440039794903191e-05,
+      "loss": 1.2399,
+      "step": 179950
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.439925001913216e-05,
+      "loss": 0.9802,
+      "step": 179960
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.439810208923241e-05,
+      "loss": 1.1764,
+      "step": 179970
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.4396954159332666e-05,
+      "loss": 1.1241,
+      "step": 179980
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.4395806229432916e-05,
+      "loss": 1.1023,
+      "step": 179990
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.439465829953317e-05,
+      "loss": 1.066,
+      "step": 180000
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.439351036963342e-05,
+      "loss": 1.212,
+      "step": 180010
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.439236243973367e-05,
+      "loss": 1.0391,
+      "step": 180020
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.439121450983393e-05,
+      "loss": 1.162,
+      "step": 180030
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.4390066579934185e-05,
+      "loss": 1.0908,
+      "step": 180040
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.438891865003443e-05,
+      "loss": 1.0726,
+      "step": 180050
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.4387770720134685e-05,
+      "loss": 1.2355,
+      "step": 180060
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.438662279023494e-05,
+      "loss": 1.2034,
+      "step": 180070
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.4385474860335185e-05,
+      "loss": 1.1485,
+      "step": 180080
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.438432693043544e-05,
+      "loss": 1.1048,
+      "step": 180090
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.43831790005357e-05,
+      "loss": 1.0809,
+      "step": 180100
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.438203107063595e-05,
+      "loss": 1.0328,
+      "step": 180110
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.43808831407362e-05,
+      "loss": 1.1305,
+      "step": 180120
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.4379735210836454e-05,
+      "loss": 1.086,
+      "step": 180130
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.4378587280936704e-05,
+      "loss": 1.2434,
+      "step": 180140
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.437743935103696e-05,
+      "loss": 1.1212,
+      "step": 180150
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.437629142113721e-05,
+      "loss": 1.0945,
+      "step": 180160
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.437514349123746e-05,
+      "loss": 0.9894,
+      "step": 180170
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.437399556133772e-05,
+      "loss": 1.2132,
+      "step": 180180
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.4372847631437974e-05,
+      "loss": 1.0667,
+      "step": 180190
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.437169970153822e-05,
+      "loss": 1.1658,
+      "step": 180200
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.4370551771638474e-05,
+      "loss": 1.1654,
+      "step": 180210
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.436940384173873e-05,
+      "loss": 1.2492,
+      "step": 180220
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.436825591183897e-05,
+      "loss": 1.1619,
+      "step": 180230
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.436710798193923e-05,
+      "loss": 1.1831,
+      "step": 180240
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.4365960052039487e-05,
+      "loss": 1.0684,
+      "step": 180250
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.4364812122139736e-05,
+      "loss": 1.0618,
+      "step": 180260
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.4363664192239986e-05,
+      "loss": 1.1198,
+      "step": 180270
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.436251626234024e-05,
+      "loss": 1.1337,
+      "step": 180280
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.436136833244049e-05,
+      "loss": 1.0807,
+      "step": 180290
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.436022040254075e-05,
+      "loss": 1.0544,
+      "step": 180300
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.4359072472641e-05,
+      "loss": 1.1786,
+      "step": 180310
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.435792454274125e-05,
+      "loss": 1.2626,
+      "step": 180320
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.4356776612841506e-05,
+      "loss": 1.056,
+      "step": 180330
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.435562868294176e-05,
+      "loss": 1.0613,
+      "step": 180340
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.4354480753042005e-05,
+      "loss": 1.1773,
+      "step": 180350
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.435333282314226e-05,
+      "loss": 1.1624,
+      "step": 180360
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.435218489324252e-05,
+      "loss": 1.1164,
+      "step": 180370
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.435103696334276e-05,
+      "loss": 1.1498,
+      "step": 180380
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.434988903344302e-05,
+      "loss": 1.1996,
+      "step": 180390
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.4348741103543275e-05,
+      "loss": 1.1642,
+      "step": 180400
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.4347593173643525e-05,
+      "loss": 1.0977,
+      "step": 180410
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.4346445243743775e-05,
+      "loss": 1.1524,
+      "step": 180420
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.434529731384403e-05,
+      "loss": 1.0835,
+      "step": 180430
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.434414938394428e-05,
+      "loss": 1.1339,
+      "step": 180440
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.434300145404454e-05,
+      "loss": 1.0782,
+      "step": 180450
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.434185352414479e-05,
+      "loss": 1.2207,
+      "step": 180460
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.434070559424504e-05,
+      "loss": 1.0639,
+      "step": 180470
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.4339557664345294e-05,
+      "loss": 1.1538,
+      "step": 180480
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.433840973444555e-05,
+      "loss": 1.1848,
+      "step": 180490
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.4337261804545794e-05,
+      "loss": 1.0818,
+      "step": 180500
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.433611387464605e-05,
+      "loss": 1.0726,
+      "step": 180510
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.433496594474631e-05,
+      "loss": 1.1299,
+      "step": 180520
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.433381801484655e-05,
+      "loss": 1.0837,
+      "step": 180530
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.433267008494681e-05,
+      "loss": 1.1249,
+      "step": 180540
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.4331522155047064e-05,
+      "loss": 1.1573,
+      "step": 180550
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.433037422514731e-05,
+      "loss": 1.1277,
+      "step": 180560
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.432922629524756e-05,
+      "loss": 1.1921,
+      "step": 180570
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.432807836534782e-05,
+      "loss": 1.1455,
+      "step": 180580
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.432693043544807e-05,
+      "loss": 1.1976,
+      "step": 180590
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.4325782505548326e-05,
+      "loss": 1.026,
+      "step": 180600
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.4324634575648576e-05,
+      "loss": 1.1654,
+      "step": 180610
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.4323486645748826e-05,
+      "loss": 1.2022,
+      "step": 180620
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.432233871584908e-05,
+      "loss": 1.1517,
+      "step": 180630
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.432119078594934e-05,
+      "loss": 1.0682,
+      "step": 180640
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.432004285604958e-05,
+      "loss": 1.1377,
+      "step": 180650
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.431889492614984e-05,
+      "loss": 1.1446,
+      "step": 180660
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.4317746996250096e-05,
+      "loss": 1.2265,
+      "step": 180670
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.431659906635034e-05,
+      "loss": 1.2061,
+      "step": 180680
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.4315451136450595e-05,
+      "loss": 1.1706,
+      "step": 180690
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.431430320655085e-05,
+      "loss": 1.1365,
+      "step": 180700
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.43131552766511e-05,
+      "loss": 1.1555,
+      "step": 180710
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.431200734675135e-05,
+      "loss": 1.1373,
+      "step": 180720
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.431085941685161e-05,
+      "loss": 1.145,
+      "step": 180730
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.430971148695186e-05,
+      "loss": 1.1254,
+      "step": 180740
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.4308563557052115e-05,
+      "loss": 1.1164,
+      "step": 180750
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.4307415627152365e-05,
+      "loss": 1.0959,
+      "step": 180760
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.4306267697252615e-05,
+      "loss": 1.1638,
+      "step": 180770
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.430511976735287e-05,
+      "loss": 1.2184,
+      "step": 180780
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.430397183745313e-05,
+      "loss": 1.0732,
+      "step": 180790
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.430282390755337e-05,
+      "loss": 1.2131,
+      "step": 180800
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.430167597765363e-05,
+      "loss": 1.1828,
+      "step": 180810
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.4300528047753884e-05,
+      "loss": 1.1578,
+      "step": 180820
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.429938011785413e-05,
+      "loss": 1.2319,
+      "step": 180830
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.4298232187954384e-05,
+      "loss": 1.3767,
+      "step": 180840
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.429708425805464e-05,
+      "loss": 1.2883,
+      "step": 180850
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.429593632815489e-05,
+      "loss": 1.1305,
+      "step": 180860
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.429478839825514e-05,
+      "loss": 1.1267,
+      "step": 180870
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.42936404683554e-05,
+      "loss": 1.1965,
+      "step": 180880
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.429249253845565e-05,
+      "loss": 1.1904,
+      "step": 180890
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.42913446085559e-05,
+      "loss": 1.1454,
+      "step": 180900
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.429019667865615e-05,
+      "loss": 1.186,
+      "step": 180910
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.42890487487564e-05,
+      "loss": 1.1198,
+      "step": 180920
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.428790081885666e-05,
+      "loss": 1.1004,
+      "step": 180930
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.4286752888956916e-05,
+      "loss": 1.1754,
+      "step": 180940
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.428560495905716e-05,
+      "loss": 1.1142,
+      "step": 180950
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.4284457029157416e-05,
+      "loss": 1.0577,
+      "step": 180960
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.428330909925767e-05,
+      "loss": 1.1582,
+      "step": 180970
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.4282161169357916e-05,
+      "loss": 1.1288,
+      "step": 180980
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.428101323945817e-05,
+      "loss": 1.1196,
+      "step": 180990
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.427986530955842e-05,
+      "loss": 1.0756,
+      "step": 181000
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.427871737965868e-05,
+      "loss": 1.0746,
+      "step": 181010
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.427756944975893e-05,
+      "loss": 1.1954,
+      "step": 181020
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.427642151985918e-05,
+      "loss": 1.133,
+      "step": 181030
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.4275273589959435e-05,
+      "loss": 1.1153,
+      "step": 181040
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.427412566005969e-05,
+      "loss": 1.0602,
+      "step": 181050
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.4272977730159935e-05,
+      "loss": 1.0792,
+      "step": 181060
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.427182980026019e-05,
+      "loss": 1.0877,
+      "step": 181070
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.427068187036045e-05,
+      "loss": 1.0972,
+      "step": 181080
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.426953394046069e-05,
+      "loss": 1.2284,
+      "step": 181090
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.426838601056095e-05,
+      "loss": 1.2556,
+      "step": 181100
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.4267238080661204e-05,
+      "loss": 1.1507,
+      "step": 181110
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.4266090150761454e-05,
+      "loss": 1.1025,
+      "step": 181120
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.4264942220861704e-05,
+      "loss": 1.1513,
+      "step": 181130
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.426379429096196e-05,
+      "loss": 1.1316,
+      "step": 181140
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.426264636106221e-05,
+      "loss": 1.1504,
+      "step": 181150
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.426149843116247e-05,
+      "loss": 1.1917,
+      "step": 181160
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.426035050126272e-05,
+      "loss": 1.0668,
+      "step": 181170
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.425920257136297e-05,
+      "loss": 1.2554,
+      "step": 181180
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.4258054641463224e-05,
+      "loss": 1.1198,
+      "step": 181190
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.425690671156348e-05,
+      "loss": 1.2576,
+      "step": 181200
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.425575878166372e-05,
+      "loss": 1.1361,
+      "step": 181210
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.425461085176398e-05,
+      "loss": 1.1088,
+      "step": 181220
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.4253462921864237e-05,
+      "loss": 1.1163,
+      "step": 181230
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.425231499196448e-05,
+      "loss": 1.1033,
+      "step": 181240
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.4251167062064736e-05,
+      "loss": 1.0418,
+      "step": 181250
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.425001913216499e-05,
+      "loss": 1.1095,
+      "step": 181260
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.424887120226524e-05,
+      "loss": 1.0919,
+      "step": 181270
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.424772327236549e-05,
+      "loss": 1.0799,
+      "step": 181280
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.424657534246575e-05,
+      "loss": 1.1099,
+      "step": 181290
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.4245427412566e-05,
+      "loss": 1.1793,
+      "step": 181300
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.4244279482666256e-05,
+      "loss": 1.0457,
+      "step": 181310
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.4243131552766506e-05,
+      "loss": 1.1317,
+      "step": 181320
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.4241983622866755e-05,
+      "loss": 1.0869,
+      "step": 181330
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.424083569296701e-05,
+      "loss": 1.1031,
+      "step": 181340
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.423968776306727e-05,
+      "loss": 1.3152,
+      "step": 181350
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.423853983316751e-05,
+      "loss": 1.1548,
+      "step": 181360
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.423739190326777e-05,
+      "loss": 1.1365,
+      "step": 181370
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.4236243973368025e-05,
+      "loss": 1.0471,
+      "step": 181380
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.423509604346827e-05,
+      "loss": 1.1005,
+      "step": 181390
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.4233948113568525e-05,
+      "loss": 1.1407,
+      "step": 181400
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.423280018366878e-05,
+      "loss": 1.1996,
+      "step": 181410
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.423165225376903e-05,
+      "loss": 1.1642,
+      "step": 181420
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.423050432386928e-05,
+      "loss": 1.062,
+      "step": 181430
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.422935639396954e-05,
+      "loss": 1.0352,
+      "step": 181440
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.422820846406979e-05,
+      "loss": 1.0338,
+      "step": 181450
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.4227060534170044e-05,
+      "loss": 1.1606,
+      "step": 181460
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.4225912604270294e-05,
+      "loss": 1.247,
+      "step": 181470
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.4224764674370544e-05,
+      "loss": 1.173,
+      "step": 181480
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.42236167444708e-05,
+      "loss": 1.1899,
+      "step": 181490
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.422246881457106e-05,
+      "loss": 1.0384,
+      "step": 181500
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.42213208846713e-05,
+      "loss": 1.0981,
+      "step": 181510
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.422017295477156e-05,
+      "loss": 1.2217,
+      "step": 181520
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.4219025024871813e-05,
+      "loss": 1.1539,
+      "step": 181530
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.4217877094972057e-05,
+      "loss": 1.2047,
+      "step": 181540
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.421672916507231e-05,
+      "loss": 1.2527,
+      "step": 181550
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.421558123517257e-05,
+      "loss": 1.1848,
+      "step": 181560
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.421443330527282e-05,
+      "loss": 1.0559,
+      "step": 181570
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.421328537537307e-05,
+      "loss": 1.0824,
+      "step": 181580
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.4212137445473326e-05,
+      "loss": 1.1435,
+      "step": 181590
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.4210989515573576e-05,
+      "loss": 1.1082,
+      "step": 181600
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.420984158567383e-05,
+      "loss": 1.2406,
+      "step": 181610
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.420869365577408e-05,
+      "loss": 1.1635,
+      "step": 181620
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.420754572587433e-05,
+      "loss": 1.239,
+      "step": 181630
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.420639779597459e-05,
+      "loss": 1.0947,
+      "step": 181640
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.4205249866074846e-05,
+      "loss": 1.1122,
+      "step": 181650
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.420410193617509e-05,
+      "loss": 1.0465,
+      "step": 181660
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.4202954006275345e-05,
+      "loss": 1.1624,
+      "step": 181670
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.42018060763756e-05,
+      "loss": 1.0271,
+      "step": 181680
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.4200658146475845e-05,
+      "loss": 1.3122,
+      "step": 181690
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.41995102165761e-05,
+      "loss": 1.1743,
+      "step": 181700
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.419836228667636e-05,
+      "loss": 1.0829,
+      "step": 181710
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.419721435677661e-05,
+      "loss": 1.233,
+      "step": 181720
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.419606642687686e-05,
+      "loss": 1.173,
+      "step": 181730
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.4194918496977115e-05,
+      "loss": 1.1561,
+      "step": 181740
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.4193770567077364e-05,
+      "loss": 1.2552,
+      "step": 181750
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.419262263717762e-05,
+      "loss": 1.1788,
+      "step": 181760
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.419147470727787e-05,
+      "loss": 1.1158,
+      "step": 181770
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.419032677737812e-05,
+      "loss": 1.1168,
+      "step": 181780
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.418917884747838e-05,
+      "loss": 1.1768,
+      "step": 181790
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.4188030917578634e-05,
+      "loss": 1.1493,
+      "step": 181800
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.418688298767888e-05,
+      "loss": 1.1707,
+      "step": 181810
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.4185735057779134e-05,
+      "loss": 1.1306,
+      "step": 181820
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.418458712787939e-05,
+      "loss": 1.1455,
+      "step": 181830
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.4183439197979634e-05,
+      "loss": 1.0818,
+      "step": 181840
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.418229126807989e-05,
+      "loss": 1.2259,
+      "step": 181850
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.418114333818015e-05,
+      "loss": 1.1614,
+      "step": 181860
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.41799954082804e-05,
+      "loss": 1.1473,
+      "step": 181870
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.4178847478380646e-05,
+      "loss": 1.0745,
+      "step": 181880
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.41776995484809e-05,
+      "loss": 1.0734,
+      "step": 181890
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.417655161858115e-05,
+      "loss": 1.0645,
+      "step": 181900
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.417540368868141e-05,
+      "loss": 1.0921,
+      "step": 181910
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.417425575878166e-05,
+      "loss": 1.1208,
+      "step": 181920
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.417310782888191e-05,
+      "loss": 1.224,
+      "step": 181930
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.4171959898982166e-05,
+      "loss": 1.1366,
+      "step": 181940
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.417081196908242e-05,
+      "loss": 1.1599,
+      "step": 181950
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.4169664039182666e-05,
+      "loss": 1.1657,
+      "step": 181960
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.416851610928292e-05,
+      "loss": 1.0262,
+      "step": 181970
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.416736817938318e-05,
+      "loss": 1.0445,
+      "step": 181980
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.416622024948342e-05,
+      "loss": 1.1639,
+      "step": 181990
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.416507231958368e-05,
+      "loss": 1.2065,
+      "step": 182000
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.4163924389683935e-05,
+      "loss": 1.2326,
+      "step": 182010
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.4162776459784185e-05,
+      "loss": 1.1114,
+      "step": 182020
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.4161628529884435e-05,
+      "loss": 1.2028,
+      "step": 182030
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.416048059998469e-05,
+      "loss": 1.0156,
+      "step": 182040
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.415933267008494e-05,
+      "loss": 1.1663,
+      "step": 182050
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.41581847401852e-05,
+      "loss": 1.1244,
+      "step": 182060
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.415703681028545e-05,
+      "loss": 1.1474,
+      "step": 182070
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.41558888803857e-05,
+      "loss": 1.1874,
+      "step": 182080
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.4154740950485954e-05,
+      "loss": 1.1366,
+      "step": 182090
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.415359302058621e-05,
+      "loss": 1.2293,
+      "step": 182100
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.4152445090686454e-05,
+      "loss": 1.3157,
+      "step": 182110
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.415129716078671e-05,
+      "loss": 1.2068,
+      "step": 182120
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.415014923088697e-05,
+      "loss": 1.0491,
+      "step": 182130
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.414900130098721e-05,
+      "loss": 0.9962,
+      "step": 182140
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.414785337108747e-05,
+      "loss": 1.2657,
+      "step": 182150
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.4146705441187724e-05,
+      "loss": 1.1245,
+      "step": 182160
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.4145557511287974e-05,
+      "loss": 1.1501,
+      "step": 182170
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.4144409581388223e-05,
+      "loss": 1.1518,
+      "step": 182180
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.414326165148848e-05,
+      "loss": 1.1869,
+      "step": 182190
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.414211372158873e-05,
+      "loss": 1.1307,
+      "step": 182200
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.4140965791688987e-05,
+      "loss": 1.0805,
+      "step": 182210
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.4139817861789236e-05,
+      "loss": 1.1557,
+      "step": 182220
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.4138669931889486e-05,
+      "loss": 1.1557,
+      "step": 182230
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.413752200198974e-05,
+      "loss": 1.2128,
+      "step": 182240
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.413637407209e-05,
+      "loss": 1.1313,
+      "step": 182250
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.413522614219024e-05,
+      "loss": 1.0992,
+      "step": 182260
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.41340782122905e-05,
+      "loss": 1.1243,
+      "step": 182270
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.4132930282390756e-05,
+      "loss": 1.0564,
+      "step": 182280
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.4131782352491e-05,
+      "loss": 1.0489,
+      "step": 182290
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.4130634422591256e-05,
+      "loss": 1.109,
+      "step": 182300
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.412948649269151e-05,
+      "loss": 1.0514,
+      "step": 182310
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.412833856279176e-05,
+      "loss": 1.0971,
+      "step": 182320
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.412719063289201e-05,
+      "loss": 1.0773,
+      "step": 182330
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.412604270299227e-05,
+      "loss": 1.0905,
+      "step": 182340
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.412489477309252e-05,
+      "loss": 1.0895,
+      "step": 182350
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.4123746843192775e-05,
+      "loss": 1.1626,
+      "step": 182360
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.412259891329302e-05,
+      "loss": 1.0368,
+      "step": 182370
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.4121450983393275e-05,
+      "loss": 1.074,
+      "step": 182380
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.412030305349353e-05,
+      "loss": 1.0256,
+      "step": 182390
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.4119155123593774e-05,
+      "loss": 1.1456,
+      "step": 182400
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.411800719369403e-05,
+      "loss": 1.0379,
+      "step": 182410
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.411685926379429e-05,
+      "loss": 1.1184,
+      "step": 182420
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.411571133389454e-05,
+      "loss": 1.2474,
+      "step": 182430
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.411456340399479e-05,
+      "loss": 1.1261,
+      "step": 182440
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.4113415474095044e-05,
+      "loss": 1.2089,
+      "step": 182450
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.4112267544195294e-05,
+      "loss": 1.1988,
+      "step": 182460
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.411111961429555e-05,
+      "loss": 1.199,
+      "step": 182470
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.41099716843958e-05,
+      "loss": 1.1179,
+      "step": 182480
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.410882375449605e-05,
+      "loss": 1.0818,
+      "step": 182490
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.410767582459631e-05,
+      "loss": 1.1254,
+      "step": 182500
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.4106527894696563e-05,
+      "loss": 1.2627,
+      "step": 182510
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.4105379964796807e-05,
+      "loss": 1.1884,
+      "step": 182520
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.410423203489706e-05,
+      "loss": 1.2087,
+      "step": 182530
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.410308410499732e-05,
+      "loss": 1.2063,
+      "step": 182540
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.410193617509756e-05,
+      "loss": 1.1938,
+      "step": 182550
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.410078824519782e-05,
+      "loss": 1.1558,
+      "step": 182560
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.4099640315298076e-05,
+      "loss": 1.1103,
+      "step": 182570
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.4098492385398326e-05,
+      "loss": 1.1918,
+      "step": 182580
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.4097344455498576e-05,
+      "loss": 0.9926,
+      "step": 182590
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.409619652559883e-05,
+      "loss": 1.0594,
+      "step": 182600
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.409504859569908e-05,
+      "loss": 1.0657,
+      "step": 182610
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.409390066579934e-05,
+      "loss": 1.1743,
+      "step": 182620
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.409275273589959e-05,
+      "loss": 1.0167,
+      "step": 182630
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.409160480599984e-05,
+      "loss": 1.2105,
+      "step": 182640
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.4090456876100095e-05,
+      "loss": 1.1793,
+      "step": 182650
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.408930894620035e-05,
+      "loss": 1.2378,
+      "step": 182660
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.4088161016300595e-05,
+      "loss": 1.005,
+      "step": 182670
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.408701308640085e-05,
+      "loss": 1.0276,
+      "step": 182680
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.408586515650111e-05,
+      "loss": 1.057,
+      "step": 182690
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.408471722660135e-05,
+      "loss": 1.1775,
+      "step": 182700
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.408356929670161e-05,
+      "loss": 1.1543,
+      "step": 182710
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.4082421366801865e-05,
+      "loss": 1.032,
+      "step": 182720
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.4081273436902114e-05,
+      "loss": 1.0624,
+      "step": 182730
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.4080125507002364e-05,
+      "loss": 1.1304,
+      "step": 182740
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.407897757710262e-05,
+      "loss": 1.2198,
+      "step": 182750
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.407782964720287e-05,
+      "loss": 1.0924,
+      "step": 182760
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.407668171730313e-05,
+      "loss": 1.1925,
+      "step": 182770
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.407553378740338e-05,
+      "loss": 1.2525,
+      "step": 182780
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.407438585750363e-05,
+      "loss": 1.0358,
+      "step": 182790
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.4073237927603884e-05,
+      "loss": 1.1224,
+      "step": 182800
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.407208999770414e-05,
+      "loss": 1.0865,
+      "step": 182810
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.4070942067804384e-05,
+      "loss": 1.2132,
+      "step": 182820
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.406979413790464e-05,
+      "loss": 1.2395,
+      "step": 182830
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.40686462080049e-05,
+      "loss": 1.2321,
+      "step": 182840
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.406749827810514e-05,
+      "loss": 1.1376,
+      "step": 182850
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.4066350348205396e-05,
+      "loss": 1.1546,
+      "step": 182860
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.406520241830565e-05,
+      "loss": 1.0628,
+      "step": 182870
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.40640544884059e-05,
+      "loss": 1.2728,
+      "step": 182880
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.406290655850615e-05,
+      "loss": 1.0417,
+      "step": 182890
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.406175862860641e-05,
+      "loss": 1.1849,
+      "step": 182900
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.406061069870666e-05,
+      "loss": 1.1182,
+      "step": 182910
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.4059462768806916e-05,
+      "loss": 1.1244,
+      "step": 182920
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.4058314838907166e-05,
+      "loss": 1.1871,
+      "step": 182930
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.4057166909007416e-05,
+      "loss": 1.0968,
+      "step": 182940
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.405601897910767e-05,
+      "loss": 1.007,
+      "step": 182950
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.405487104920793e-05,
+      "loss": 1.2585,
+      "step": 182960
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.405372311930817e-05,
+      "loss": 1.1091,
+      "step": 182970
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.405257518940843e-05,
+      "loss": 1.0706,
+      "step": 182980
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.4051427259508685e-05,
+      "loss": 1.0298,
+      "step": 182990
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.405027932960893e-05,
+      "loss": 1.1991,
+      "step": 183000
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.4049131399709185e-05,
+      "loss": 1.1286,
+      "step": 183010
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.404798346980944e-05,
+      "loss": 1.3005,
+      "step": 183020
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.404683553990969e-05,
+      "loss": 1.1497,
+      "step": 183030
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.404568761000994e-05,
+      "loss": 1.1554,
+      "step": 183040
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.40445396801102e-05,
+      "loss": 1.1823,
+      "step": 183050
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.404339175021045e-05,
+      "loss": 1.1418,
+      "step": 183060
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.4042243820310704e-05,
+      "loss": 1.2003,
+      "step": 183070
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.4041095890410954e-05,
+      "loss": 1.0986,
+      "step": 183080
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.4039947960511204e-05,
+      "loss": 1.178,
+      "step": 183090
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.403880003061146e-05,
+      "loss": 1.2066,
+      "step": 183100
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.403765210071172e-05,
+      "loss": 1.1445,
+      "step": 183110
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.403650417081196e-05,
+      "loss": 1.1004,
+      "step": 183120
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.403535624091222e-05,
+      "loss": 1.0672,
+      "step": 183130
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.4034208311012474e-05,
+      "loss": 1.2034,
+      "step": 183140
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.403306038111272e-05,
+      "loss": 1.1611,
+      "step": 183150
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.4031912451212973e-05,
+      "loss": 1.097,
+      "step": 183160
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.403076452131323e-05,
+      "loss": 1.1475,
+      "step": 183170
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.402961659141348e-05,
+      "loss": 1.0375,
+      "step": 183180
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.402846866151373e-05,
+      "loss": 1.1466,
+      "step": 183190
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.4027320731613986e-05,
+      "loss": 1.296,
+      "step": 183200
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.4026172801714236e-05,
+      "loss": 1.2657,
+      "step": 183210
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.402502487181449e-05,
+      "loss": 1.1086,
+      "step": 183220
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.402387694191474e-05,
+      "loss": 1.091,
+      "step": 183230
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.402272901201499e-05,
+      "loss": 1.1355,
+      "step": 183240
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.402158108211525e-05,
+      "loss": 1.3591,
+      "step": 183250
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.4020433152215506e-05,
+      "loss": 0.9956,
+      "step": 183260
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.401928522231575e-05,
+      "loss": 1.0466,
+      "step": 183270
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.4018137292416006e-05,
+      "loss": 1.0907,
+      "step": 183280
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.401698936251626e-05,
+      "loss": 0.9905,
+      "step": 183290
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.4015841432616505e-05,
+      "loss": 1.1124,
+      "step": 183300
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.401469350271676e-05,
+      "loss": 1.0848,
+      "step": 183310
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.401354557281702e-05,
+      "loss": 1.1593,
+      "step": 183320
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.401239764291727e-05,
+      "loss": 1.1575,
+      "step": 183330
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.401124971301752e-05,
+      "loss": 1.1814,
+      "step": 183340
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.4010101783117775e-05,
+      "loss": 1.1151,
+      "step": 183350
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.4008953853218025e-05,
+      "loss": 1.1492,
+      "step": 183360
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.400780592331828e-05,
+      "loss": 1.2267,
+      "step": 183370
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.400665799341853e-05,
+      "loss": 1.1466,
+      "step": 183380
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.400551006351878e-05,
+      "loss": 1.1051,
+      "step": 183390
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.400436213361904e-05,
+      "loss": 1.245,
+      "step": 183400
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.4003214203719294e-05,
+      "loss": 1.2641,
+      "step": 183410
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.400206627381954e-05,
+      "loss": 1.1154,
+      "step": 183420
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.4000918343919794e-05,
+      "loss": 1.148,
+      "step": 183430
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.399977041402005e-05,
+      "loss": 1.1818,
+      "step": 183440
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.3998622484120294e-05,
+      "loss": 1.0332,
+      "step": 183450
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.399747455422055e-05,
+      "loss": 1.0105,
+      "step": 183460
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.399632662432081e-05,
+      "loss": 1.0584,
+      "step": 183470
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.399517869442106e-05,
+      "loss": 1.1994,
+      "step": 183480
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.399403076452131e-05,
+      "loss": 1.1539,
+      "step": 183490
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.399288283462156e-05,
+      "loss": 1.1576,
+      "step": 183500
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.399173490472181e-05,
+      "loss": 1.1262,
+      "step": 183510
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.399058697482207e-05,
+      "loss": 1.1744,
+      "step": 183520
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.398943904492232e-05,
+      "loss": 1.2448,
+      "step": 183530
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.398829111502257e-05,
+      "loss": 1.1109,
+      "step": 183540
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.3987143185122826e-05,
+      "loss": 1.2285,
+      "step": 183550
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.398599525522308e-05,
+      "loss": 1.1024,
+      "step": 183560
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.3984847325323326e-05,
+      "loss": 1.1478,
+      "step": 183570
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.398369939542358e-05,
+      "loss": 1.1937,
+      "step": 183580
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.398255146552384e-05,
+      "loss": 1.2185,
+      "step": 183590
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.398140353562408e-05,
+      "loss": 1.1548,
+      "step": 183600
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.398025560572434e-05,
+      "loss": 1.0745,
+      "step": 183610
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.3979107675824595e-05,
+      "loss": 1.0945,
+      "step": 183620
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.3977959745924845e-05,
+      "loss": 1.0878,
+      "step": 183630
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.3976811816025095e-05,
+      "loss": 1.1415,
+      "step": 183640
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.397566388612535e-05,
+      "loss": 1.2028,
+      "step": 183650
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.39745159562256e-05,
+      "loss": 1.1936,
+      "step": 183660
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.397336802632586e-05,
+      "loss": 1.1842,
+      "step": 183670
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.397222009642611e-05,
+      "loss": 1.1876,
+      "step": 183680
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.397107216652636e-05,
+      "loss": 1.0811,
+      "step": 183690
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.3969924236626615e-05,
+      "loss": 1.3524,
+      "step": 183700
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.396877630672687e-05,
+      "loss": 1.1754,
+      "step": 183710
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.3967628376827114e-05,
+      "loss": 1.2429,
+      "step": 183720
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.396648044692737e-05,
+      "loss": 1.1217,
+      "step": 183730
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.396533251702762e-05,
+      "loss": 1.2499,
+      "step": 183740
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.396418458712787e-05,
+      "loss": 1.2111,
+      "step": 183750
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.396303665722813e-05,
+      "loss": 1.0864,
+      "step": 183760
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.396188872732838e-05,
+      "loss": 1.1468,
+      "step": 183770
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.3960740797428634e-05,
+      "loss": 1.1637,
+      "step": 183780
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.3959592867528884e-05,
+      "loss": 1.0203,
+      "step": 183790
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.3958444937629134e-05,
+      "loss": 1.0773,
+      "step": 183800
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.395729700772939e-05,
+      "loss": 1.1185,
+      "step": 183810
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.395614907782965e-05,
+      "loss": 1.1557,
+      "step": 183820
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.395500114792989e-05,
+      "loss": 1.0339,
+      "step": 183830
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.3953853218030146e-05,
+      "loss": 1.1476,
+      "step": 183840
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.39527052881304e-05,
+      "loss": 1.0236,
+      "step": 183850
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.3951557358230646e-05,
+      "loss": 1.2389,
+      "step": 183860
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.39504094283309e-05,
+      "loss": 1.2552,
+      "step": 183870
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.394926149843116e-05,
+      "loss": 1.0814,
+      "step": 183880
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.394811356853141e-05,
+      "loss": 1.1495,
+      "step": 183890
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.394696563863166e-05,
+      "loss": 1.1348,
+      "step": 183900
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.3945817708731916e-05,
+      "loss": 1.1871,
+      "step": 183910
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.3944669778832166e-05,
+      "loss": 1.1369,
+      "step": 183920
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.394352184893242e-05,
+      "loss": 1.0423,
+      "step": 183930
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.394237391903267e-05,
+      "loss": 1.1282,
+      "step": 183940
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.394122598913292e-05,
+      "loss": 1.0897,
+      "step": 183950
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.394007805923318e-05,
+      "loss": 1.0721,
+      "step": 183960
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.3938930129333435e-05,
+      "loss": 1.1625,
+      "step": 183970
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.393778219943368e-05,
+      "loss": 1.0986,
+      "step": 183980
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.3936634269533935e-05,
+      "loss": 1.1273,
+      "step": 183990
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.393548633963419e-05,
+      "loss": 1.1768,
+      "step": 184000
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.3934338409734435e-05,
+      "loss": 1.1541,
+      "step": 184010
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.393319047983469e-05,
+      "loss": 1.127,
+      "step": 184020
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.393204254993495e-05,
+      "loss": 1.1419,
+      "step": 184030
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.39308946200352e-05,
+      "loss": 1.1579,
+      "step": 184040
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.392974669013545e-05,
+      "loss": 1.2203,
+      "step": 184050
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.3928598760235704e-05,
+      "loss": 1.1217,
+      "step": 184060
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.3927450830335954e-05,
+      "loss": 1.1655,
+      "step": 184070
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.392630290043621e-05,
+      "loss": 1.2134,
+      "step": 184080
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.392515497053646e-05,
+      "loss": 1.0756,
+      "step": 184090
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.392400704063671e-05,
+      "loss": 1.1499,
+      "step": 184100
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.392285911073697e-05,
+      "loss": 1.1813,
+      "step": 184110
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.3921711180837224e-05,
+      "loss": 1.0932,
+      "step": 184120
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.392056325093747e-05,
+      "loss": 1.3875,
+      "step": 184130
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.3919415321037723e-05,
+      "loss": 1.0787,
+      "step": 184140
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.391826739113798e-05,
+      "loss": 1.0583,
+      "step": 184150
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.391711946123822e-05,
+      "loss": 1.1234,
+      "step": 184160
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.391597153133848e-05,
+      "loss": 1.1362,
+      "step": 184170
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.3914823601438736e-05,
+      "loss": 1.1248,
+      "step": 184180
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.3913675671538986e-05,
+      "loss": 1.0382,
+      "step": 184190
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.3912527741639236e-05,
+      "loss": 1.23,
+      "step": 184200
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.391137981173949e-05,
+      "loss": 1.18,
+      "step": 184210
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.391023188183974e-05,
+      "loss": 1.2431,
+      "step": 184220
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.390908395194e-05,
+      "loss": 1.0796,
+      "step": 184230
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.390793602204025e-05,
+      "loss": 1.1712,
+      "step": 184240
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.39067880921405e-05,
+      "loss": 1.0618,
+      "step": 184250
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.3905640162240756e-05,
+      "loss": 1.0973,
+      "step": 184260
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.390449223234101e-05,
+      "loss": 1.1223,
+      "step": 184270
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.3903344302441255e-05,
+      "loss": 1.0562,
+      "step": 184280
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.390219637254151e-05,
+      "loss": 1.0273,
+      "step": 184290
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.390104844264177e-05,
+      "loss": 1.2914,
+      "step": 184300
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.389990051274201e-05,
+      "loss": 1.052,
+      "step": 184310
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.389875258284227e-05,
+      "loss": 1.2022,
+      "step": 184320
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.3897604652942525e-05,
+      "loss": 1.0596,
+      "step": 184330
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.3896456723042775e-05,
+      "loss": 1.2977,
+      "step": 184340
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.3895308793143025e-05,
+      "loss": 1.1285,
+      "step": 184350
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.389416086324328e-05,
+      "loss": 0.9753,
+      "step": 184360
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.389301293334353e-05,
+      "loss": 1.076,
+      "step": 184370
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.389186500344379e-05,
+      "loss": 1.1873,
+      "step": 184380
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.389071707354404e-05,
+      "loss": 1.2362,
+      "step": 184390
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.388956914364429e-05,
+      "loss": 1.0557,
+      "step": 184400
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.3888421213744544e-05,
+      "loss": 1.1015,
+      "step": 184410
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.38872732838448e-05,
+      "loss": 1.2032,
+      "step": 184420
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.3886125353945044e-05,
+      "loss": 1.173,
+      "step": 184430
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.38849774240453e-05,
+      "loss": 1.1011,
+      "step": 184440
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.388382949414556e-05,
+      "loss": 1.191,
+      "step": 184450
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.38826815642458e-05,
+      "loss": 1.1817,
+      "step": 184460
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.388153363434606e-05,
+      "loss": 1.0822,
+      "step": 184470
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.388038570444631e-05,
+      "loss": 1.189,
+      "step": 184480
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.387923777454656e-05,
+      "loss": 1.0825,
+      "step": 184490
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.387808984464681e-05,
+      "loss": 1.1726,
+      "step": 184500
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.387694191474707e-05,
+      "loss": 1.1135,
+      "step": 184510
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.387579398484732e-05,
+      "loss": 1.0616,
+      "step": 184520
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.3874646054947576e-05,
+      "loss": 1.1179,
+      "step": 184530
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.3873498125047826e-05,
+      "loss": 1.2267,
+      "step": 184540
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.3872350195148076e-05,
+      "loss": 1.0773,
+      "step": 184550
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.387120226524833e-05,
+      "loss": 1.1946,
+      "step": 184560
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.387005433534859e-05,
+      "loss": 1.2298,
+      "step": 184570
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.386890640544883e-05,
+      "loss": 1.1468,
+      "step": 184580
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.386775847554909e-05,
+      "loss": 1.0625,
+      "step": 184590
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.3866610545649345e-05,
+      "loss": 1.1598,
+      "step": 184600
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.386546261574959e-05,
+      "loss": 1.0773,
+      "step": 184610
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.3864314685849845e-05,
+      "loss": 1.1817,
+      "step": 184620
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.38631667559501e-05,
+      "loss": 1.1706,
+      "step": 184630
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.386201882605035e-05,
+      "loss": 1.1451,
+      "step": 184640
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.38608708961506e-05,
+      "loss": 0.985,
+      "step": 184650
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.385972296625086e-05,
+      "loss": 1.1373,
+      "step": 184660
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.385857503635111e-05,
+      "loss": 1.0931,
+      "step": 184670
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.3857427106451365e-05,
+      "loss": 1.2012,
+      "step": 184680
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.3856279176551614e-05,
+      "loss": 1.1728,
+      "step": 184690
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.3855131246651864e-05,
+      "loss": 1.2237,
+      "step": 184700
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.385398331675212e-05,
+      "loss": 1.1363,
+      "step": 184710
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.385283538685238e-05,
+      "loss": 1.1028,
+      "step": 184720
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.385168745695262e-05,
+      "loss": 1.1542,
+      "step": 184730
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.385053952705288e-05,
+      "loss": 1.2372,
+      "step": 184740
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.3849391597153134e-05,
+      "loss": 1.1214,
+      "step": 184750
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.384824366725338e-05,
+      "loss": 1.1331,
+      "step": 184760
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.3847095737353634e-05,
+      "loss": 1.2455,
+      "step": 184770
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.384594780745389e-05,
+      "loss": 1.1791,
+      "step": 184780
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.384479987755414e-05,
+      "loss": 1.2149,
+      "step": 184790
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.384365194765439e-05,
+      "loss": 1.2684,
+      "step": 184800
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.384250401775465e-05,
+      "loss": 1.2515,
+      "step": 184810
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.3841356087854896e-05,
+      "loss": 1.0954,
+      "step": 184820
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.384020815795515e-05,
+      "loss": 1.2056,
+      "step": 184830
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.38390602280554e-05,
+      "loss": 1.1493,
+      "step": 184840
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.383791229815565e-05,
+      "loss": 1.2082,
+      "step": 184850
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.383676436825591e-05,
+      "loss": 1.1079,
+      "step": 184860
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.3835616438356166e-05,
+      "loss": 0.9768,
+      "step": 184870
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.383446850845641e-05,
+      "loss": 1.1815,
+      "step": 184880
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.3833320578556666e-05,
+      "loss": 1.2394,
+      "step": 184890
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.383217264865692e-05,
+      "loss": 1.0527,
+      "step": 184900
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.3831024718757165e-05,
+      "loss": 0.9801,
+      "step": 184910
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.382987678885742e-05,
+      "loss": 1.0602,
+      "step": 184920
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.382872885895768e-05,
+      "loss": 1.0339,
+      "step": 184930
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.382758092905793e-05,
+      "loss": 1.0174,
+      "step": 184940
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.382643299915818e-05,
+      "loss": 1.0943,
+      "step": 184950
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.3825285069258435e-05,
+      "loss": 1.1424,
+      "step": 184960
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.3824137139358685e-05,
+      "loss": 0.953,
+      "step": 184970
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.382298920945894e-05,
+      "loss": 1.1182,
+      "step": 184980
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.382184127955919e-05,
+      "loss": 1.1681,
+      "step": 184990
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.382069334965944e-05,
+      "loss": 1.0889,
+      "step": 185000
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.38195454197597e-05,
+      "loss": 1.2645,
+      "step": 185010
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.3818397489859955e-05,
+      "loss": 1.2488,
+      "step": 185020
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.38172495599602e-05,
+      "loss": 1.1024,
+      "step": 185030
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.3816101630060454e-05,
+      "loss": 1.0415,
+      "step": 185040
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.381495370016071e-05,
+      "loss": 1.2778,
+      "step": 185050
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.3813805770260954e-05,
+      "loss": 1.1058,
+      "step": 185060
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.381265784036121e-05,
+      "loss": 1.1783,
+      "step": 185070
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.381150991046147e-05,
+      "loss": 1.1541,
+      "step": 185080
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.381036198056172e-05,
+      "loss": 1.1286,
+      "step": 185090
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.380921405066197e-05,
+      "loss": 1.0976,
+      "step": 185100
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.3808066120762224e-05,
+      "loss": 1.2813,
+      "step": 185110
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.3806918190862473e-05,
+      "loss": 1.2049,
+      "step": 185120
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.380577026096273e-05,
+      "loss": 1.1013,
+      "step": 185130
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.380462233106297e-05,
+      "loss": 1.1175,
+      "step": 185140
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.380347440116323e-05,
+      "loss": 1.1221,
+      "step": 185150
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.3802326471263486e-05,
+      "loss": 1.2021,
+      "step": 185160
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.380117854136373e-05,
+      "loss": 1.2016,
+      "step": 185170
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.3800030611463986e-05,
+      "loss": 1.0879,
+      "step": 185180
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.379888268156424e-05,
+      "loss": 1.0737,
+      "step": 185190
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.379773475166449e-05,
+      "loss": 1.1962,
+      "step": 185200
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.379658682176474e-05,
+      "loss": 1.187,
+      "step": 185210
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.3795438891865e-05,
+      "loss": 1.076,
+      "step": 185220
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.379429096196525e-05,
+      "loss": 1.2085,
+      "step": 185230
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.3793143032065506e-05,
+      "loss": 1.1256,
+      "step": 185240
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.3791995102165755e-05,
+      "loss": 1.1641,
+      "step": 185250
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.3790847172266005e-05,
+      "loss": 1.0791,
+      "step": 185260
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.378969924236626e-05,
+      "loss": 1.1034,
+      "step": 185270
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.378855131246652e-05,
+      "loss": 1.1698,
+      "step": 185280
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.378740338256676e-05,
+      "loss": 1.0619,
+      "step": 185290
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.378625545266702e-05,
+      "loss": 1.1937,
+      "step": 185300
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.3785107522767275e-05,
+      "loss": 1.0981,
+      "step": 185310
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.378395959286752e-05,
+      "loss": 1.0932,
+      "step": 185320
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.3782811662967775e-05,
+      "loss": 1.0904,
+      "step": 185330
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.378166373306803e-05,
+      "loss": 1.1313,
+      "step": 185340
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.378051580316828e-05,
+      "loss": 1.1187,
+      "step": 185350
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.377936787326853e-05,
+      "loss": 1.0992,
+      "step": 185360
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.377821994336879e-05,
+      "loss": 1.0994,
+      "step": 185370
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.377707201346904e-05,
+      "loss": 1.0059,
+      "step": 185380
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.3775924083569294e-05,
+      "loss": 1.0889,
+      "step": 185390
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.3774776153669544e-05,
+      "loss": 1.1167,
+      "step": 185400
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.3773628223769794e-05,
+      "loss": 1.0952,
+      "step": 185410
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.377248029387005e-05,
+      "loss": 1.1313,
+      "step": 185420
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.377133236397031e-05,
+      "loss": 1.1041,
+      "step": 185430
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.377018443407055e-05,
+      "loss": 1.116,
+      "step": 185440
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.376903650417081e-05,
+      "loss": 1.2071,
+      "step": 185450
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.376788857427106e-05,
+      "loss": 1.0788,
+      "step": 185460
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.3766740644371306e-05,
+      "loss": 1.0704,
+      "step": 185470
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.376559271447156e-05,
+      "loss": 1.0902,
+      "step": 185480
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.376444478457182e-05,
+      "loss": 1.0489,
+      "step": 185490
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.376329685467207e-05,
+      "loss": 1.0572,
+      "step": 185500
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.376214892477232e-05,
+      "loss": 1.1079,
+      "step": 185510
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.3761000994872576e-05,
+      "loss": 1.1775,
+      "step": 185520
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.3759853064972826e-05,
+      "loss": 1.1709,
+      "step": 185530
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.375870513507308e-05,
+      "loss": 1.1194,
+      "step": 185540
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.375755720517333e-05,
+      "loss": 1.255,
+      "step": 185550
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.375640927527358e-05,
+      "loss": 1.1415,
+      "step": 185560
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.375526134537384e-05,
+      "loss": 1.0491,
+      "step": 185570
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 5.3754113415474095e-05,
+      "loss": 1.1547,
+      "step": 185580
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.375296548557434e-05,
+      "loss": 1.2001,
+      "step": 185590
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.3751817555674595e-05,
+      "loss": 1.1934,
+      "step": 185600
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.375066962577485e-05,
+      "loss": 1.077,
+      "step": 185610
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.3749521695875095e-05,
+      "loss": 1.165,
+      "step": 185620
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.374837376597535e-05,
+      "loss": 1.2335,
+      "step": 185630
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.374722583607561e-05,
+      "loss": 1.3215,
+      "step": 185640
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.374607790617586e-05,
+      "loss": 1.039,
+      "step": 185650
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.374492997627611e-05,
+      "loss": 1.0743,
+      "step": 185660
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.3743782046376364e-05,
+      "loss": 1.1031,
+      "step": 185670
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.3742634116476614e-05,
+      "loss": 1.2208,
+      "step": 185680
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.374148618657687e-05,
+      "loss": 1.1766,
+      "step": 185690
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.374033825667712e-05,
+      "loss": 1.1921,
+      "step": 185700
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.373919032677737e-05,
+      "loss": 1.1581,
+      "step": 185710
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.373804239687763e-05,
+      "loss": 1.0181,
+      "step": 185720
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.3736894466977884e-05,
+      "loss": 1.0843,
+      "step": 185730
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.373574653707813e-05,
+      "loss": 1.1365,
+      "step": 185740
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.3734598607178384e-05,
+      "loss": 1.133,
+      "step": 185750
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.373345067727864e-05,
+      "loss": 1.2313,
+      "step": 185760
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.373230274737888e-05,
+      "loss": 1.1399,
+      "step": 185770
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.373115481747914e-05,
+      "loss": 1.101,
+      "step": 185780
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.37300068875794e-05,
+      "loss": 1.0402,
+      "step": 185790
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.3728858957679646e-05,
+      "loss": 1.0541,
+      "step": 185800
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.3727711027779896e-05,
+      "loss": 1.0804,
+      "step": 185810
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.372656309788015e-05,
+      "loss": 1.0815,
+      "step": 185820
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.37254151679804e-05,
+      "loss": 1.127,
+      "step": 185830
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.372426723808066e-05,
+      "loss": 1.1153,
+      "step": 185840
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.372311930818091e-05,
+      "loss": 1.1476,
+      "step": 185850
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.372197137828116e-05,
+      "loss": 1.125,
+      "step": 185860
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.3720823448381416e-05,
+      "loss": 1.1341,
+      "step": 185870
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.371967551848167e-05,
+      "loss": 1.1921,
+      "step": 185880
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.3718527588581915e-05,
+      "loss": 1.1573,
+      "step": 185890
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.371737965868217e-05,
+      "loss": 1.2594,
+      "step": 185900
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.371623172878243e-05,
+      "loss": 1.0779,
+      "step": 185910
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.371508379888267e-05,
+      "loss": 1.0874,
+      "step": 185920
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.371393586898293e-05,
+      "loss": 1.2509,
+      "step": 185930
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.3712787939083185e-05,
+      "loss": 1.1725,
+      "step": 185940
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.3711640009183435e-05,
+      "loss": 1.2285,
+      "step": 185950
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.3710492079283685e-05,
+      "loss": 1.2007,
+      "step": 185960
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.370934414938394e-05,
+      "loss": 1.1496,
+      "step": 185970
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.370819621948419e-05,
+      "loss": 1.1458,
+      "step": 185980
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.370704828958445e-05,
+      "loss": 0.9819,
+      "step": 185990
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.37059003596847e-05,
+      "loss": 1.1644,
+      "step": 186000
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.370475242978495e-05,
+      "loss": 1.1302,
+      "step": 186010
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.3703604499885204e-05,
+      "loss": 1.0839,
+      "step": 186020
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.370245656998546e-05,
+      "loss": 1.1145,
+      "step": 186030
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.3701308640085704e-05,
+      "loss": 1.2785,
+      "step": 186040
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.370016071018596e-05,
+      "loss": 1.148,
+      "step": 186050
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.369901278028622e-05,
+      "loss": 1.291,
+      "step": 186060
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.369786485038646e-05,
+      "loss": 1.2301,
+      "step": 186070
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.369671692048672e-05,
+      "loss": 1.0132,
+      "step": 186080
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.3695568990586974e-05,
+      "loss": 1.0658,
+      "step": 186090
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.3694421060687223e-05,
+      "loss": 1.0717,
+      "step": 186100
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.369327313078747e-05,
+      "loss": 1.2177,
+      "step": 186110
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.369212520088773e-05,
+      "loss": 1.2571,
+      "step": 186120
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.369097727098798e-05,
+      "loss": 1.1485,
+      "step": 186130
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.3689829341088236e-05,
+      "loss": 1.1929,
+      "step": 186140
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.3688681411188486e-05,
+      "loss": 1.099,
+      "step": 186150
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.3687533481288736e-05,
+      "loss": 1.2146,
+      "step": 186160
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.368638555138899e-05,
+      "loss": 1.1136,
+      "step": 186170
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.368523762148925e-05,
+      "loss": 1.0843,
+      "step": 186180
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.368408969158949e-05,
+      "loss": 1.0919,
+      "step": 186190
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.368294176168975e-05,
+      "loss": 1.1613,
+      "step": 186200
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.3681793831790006e-05,
+      "loss": 1.1242,
+      "step": 186210
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.368064590189025e-05,
+      "loss": 1.1648,
+      "step": 186220
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.3679497971990505e-05,
+      "loss": 1.1734,
+      "step": 186230
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 5.367835004209076e-05,
+      "loss": 1.2565,
+      "step": 186240
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.367720211219101e-05,
+      "loss": 1.2102,
+      "step": 186250
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.367605418229126e-05,
+      "loss": 1.0484,
+      "step": 186260
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.367490625239152e-05,
+      "loss": 1.0962,
+      "step": 186270
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.367375832249177e-05,
+      "loss": 1.1274,
+      "step": 186280
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.3672610392592025e-05,
+      "loss": 1.0882,
+      "step": 186290
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.3671462462692275e-05,
+      "loss": 1.1671,
+      "step": 186300
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.3670314532792525e-05,
+      "loss": 1.004,
+      "step": 186310
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.366916660289278e-05,
+      "loss": 1.1288,
+      "step": 186320
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.366801867299304e-05,
+      "loss": 1.0681,
+      "step": 186330
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.366687074309328e-05,
+      "loss": 1.1198,
+      "step": 186340
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.366572281319354e-05,
+      "loss": 1.1995,
+      "step": 186350
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.3664574883293794e-05,
+      "loss": 1.0517,
+      "step": 186360
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.366342695339404e-05,
+      "loss": 1.1775,
+      "step": 186370
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.3662279023494294e-05,
+      "loss": 1.2118,
+      "step": 186380
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.366113109359455e-05,
+      "loss": 1.1607,
+      "step": 186390
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.36599831636948e-05,
+      "loss": 1.0806,
+      "step": 186400
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.365883523379505e-05,
+      "loss": 1.1618,
+      "step": 186410
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.365768730389531e-05,
+      "loss": 1.2561,
+      "step": 186420
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.365653937399556e-05,
+      "loss": 1.1503,
+      "step": 186430
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.365539144409581e-05,
+      "loss": 1.184,
+      "step": 186440
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.365424351419606e-05,
+      "loss": 0.9903,
+      "step": 186450
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.365309558429631e-05,
+      "loss": 1.1917,
+      "step": 186460
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.365194765439657e-05,
+      "loss": 1.2508,
+      "step": 186470
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.3650799724496826e-05,
+      "loss": 1.0775,
+      "step": 186480
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.364965179459707e-05,
+      "loss": 1.0157,
+      "step": 186490
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.3648503864697326e-05,
+      "loss": 1.0544,
+      "step": 186500
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.3647355934797576e-05,
+      "loss": 1.1268,
+      "step": 186510
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.3646208004897826e-05,
+      "loss": 1.1706,
+      "step": 186520
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.364506007499808e-05,
+      "loss": 1.1003,
+      "step": 186530
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.364391214509833e-05,
+      "loss": 1.035,
+      "step": 186540
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.364276421519859e-05,
+      "loss": 1.1184,
+      "step": 186550
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.364161628529884e-05,
+      "loss": 1.0919,
+      "step": 186560
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.364046835539909e-05,
+      "loss": 1.1394,
+      "step": 186570
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.3639320425499345e-05,
+      "loss": 1.0495,
+      "step": 186580
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.36381724955996e-05,
+      "loss": 1.1663,
+      "step": 186590
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.3637024565699845e-05,
+      "loss": 1.1405,
+      "step": 186600
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.36358766358001e-05,
+      "loss": 1.0723,
+      "step": 186610
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.363472870590036e-05,
+      "loss": 1.1864,
+      "step": 186620
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.36335807760006e-05,
+      "loss": 1.1486,
+      "step": 186630
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.363243284610086e-05,
+      "loss": 1.1193,
+      "step": 186640
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.3631284916201114e-05,
+      "loss": 1.0659,
+      "step": 186650
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.3630136986301364e-05,
+      "loss": 1.087,
+      "step": 186660
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.3628989056401614e-05,
+      "loss": 1.1564,
+      "step": 186670
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.362784112650187e-05,
+      "loss": 1.2899,
+      "step": 186680
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.362669319660212e-05,
+      "loss": 1.1084,
+      "step": 186690
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.362554526670238e-05,
+      "loss": 1.1068,
+      "step": 186700
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.362439733680263e-05,
+      "loss": 1.1873,
+      "step": 186710
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.362324940690288e-05,
+      "loss": 1.0453,
+      "step": 186720
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.3622101477003134e-05,
+      "loss": 1.219,
+      "step": 186730
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.362095354710339e-05,
+      "loss": 1.3006,
+      "step": 186740
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.361980561720363e-05,
+      "loss": 1.0972,
+      "step": 186750
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.361865768730389e-05,
+      "loss": 1.19,
+      "step": 186760
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.3617509757404147e-05,
+      "loss": 1.3066,
+      "step": 186770
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.361636182750439e-05,
+      "loss": 1.0688,
+      "step": 186780
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.3615213897604646e-05,
+      "loss": 1.1755,
+      "step": 186790
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.36140659677049e-05,
+      "loss": 1.1937,
+      "step": 186800
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.361291803780515e-05,
+      "loss": 1.193,
+      "step": 186810
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.36117701079054e-05,
+      "loss": 1.1679,
+      "step": 186820
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.361062217800566e-05,
+      "loss": 1.0925,
+      "step": 186830
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.360947424810591e-05,
+      "loss": 1.2542,
+      "step": 186840
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.3608326318206166e-05,
+      "loss": 1.2422,
+      "step": 186850
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.3607178388306416e-05,
+      "loss": 1.1657,
+      "step": 186860
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.3606030458406665e-05,
+      "loss": 1.2311,
+      "step": 186870
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.360488252850692e-05,
+      "loss": 1.2505,
+      "step": 186880
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 5.360373459860718e-05,
+      "loss": 1.1721,
+      "step": 186890
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.360258666870742e-05,
+      "loss": 1.1867,
+      "step": 186900
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.360143873880768e-05,
+      "loss": 1.1133,
+      "step": 186910
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.3600290808907935e-05,
+      "loss": 1.078,
+      "step": 186920
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.359914287900818e-05,
+      "loss": 1.2274,
+      "step": 186930
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.3597994949108435e-05,
+      "loss": 1.1807,
+      "step": 186940
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.359684701920869e-05,
+      "loss": 1.1224,
+      "step": 186950
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.359569908930894e-05,
+      "loss": 1.1656,
+      "step": 186960
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.359455115940919e-05,
+      "loss": 1.135,
+      "step": 186970
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.359340322950945e-05,
+      "loss": 1.1804,
+      "step": 186980
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.35922552996097e-05,
+      "loss": 1.199,
+      "step": 186990
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.3591107369709954e-05,
+      "loss": 1.1516,
+      "step": 187000
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.3589959439810204e-05,
+      "loss": 1.238,
+      "step": 187010
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.3588811509910454e-05,
+      "loss": 1.1444,
+      "step": 187020
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.358766358001071e-05,
+      "loss": 1.1055,
+      "step": 187030
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.358651565011097e-05,
+      "loss": 1.0325,
+      "step": 187040
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.358536772021121e-05,
+      "loss": 1.0193,
+      "step": 187050
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.358421979031147e-05,
+      "loss": 1.2155,
+      "step": 187060
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.3583071860411724e-05,
+      "loss": 1.2142,
+      "step": 187070
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.358192393051197e-05,
+      "loss": 1.2858,
+      "step": 187080
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.358077600061222e-05,
+      "loss": 1.0798,
+      "step": 187090
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.357962807071248e-05,
+      "loss": 1.1299,
+      "step": 187100
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.357848014081273e-05,
+      "loss": 1.1661,
+      "step": 187110
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.357733221091298e-05,
+      "loss": 1.0825,
+      "step": 187120
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.3576184281013236e-05,
+      "loss": 1.1643,
+      "step": 187130
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.3575036351113486e-05,
+      "loss": 1.0169,
+      "step": 187140
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.357388842121374e-05,
+      "loss": 1.3911,
+      "step": 187150
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.357274049131399e-05,
+      "loss": 1.2509,
+      "step": 187160
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.357159256141424e-05,
+      "loss": 1.1388,
+      "step": 187170
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.35704446315145e-05,
+      "loss": 1.1916,
+      "step": 187180
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.3569296701614756e-05,
+      "loss": 1.1885,
+      "step": 187190
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.3568148771715e-05,
+      "loss": 1.2531,
+      "step": 187200
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.3567000841815255e-05,
+      "loss": 1.1093,
+      "step": 187210
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.356585291191551e-05,
+      "loss": 1.1076,
+      "step": 187220
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.3564704982015755e-05,
+      "loss": 1.2194,
+      "step": 187230
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.356355705211601e-05,
+      "loss": 1.0849,
+      "step": 187240
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.356240912221627e-05,
+      "loss": 1.1587,
+      "step": 187250
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.356126119231652e-05,
+      "loss": 1.1478,
+      "step": 187260
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.356011326241677e-05,
+      "loss": 1.1816,
+      "step": 187270
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.3558965332517025e-05,
+      "loss": 1.2579,
+      "step": 187280
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.3557817402617275e-05,
+      "loss": 1.0944,
+      "step": 187290
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.355666947271753e-05,
+      "loss": 1.0963,
+      "step": 187300
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.355552154281778e-05,
+      "loss": 1.166,
+      "step": 187310
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.355437361291803e-05,
+      "loss": 1.1661,
+      "step": 187320
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.355322568301829e-05,
+      "loss": 1.2322,
+      "step": 187330
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.3552077753118544e-05,
+      "loss": 1.0906,
+      "step": 187340
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.355092982321879e-05,
+      "loss": 1.1184,
+      "step": 187350
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.3549781893319044e-05,
+      "loss": 1.1942,
+      "step": 187360
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.35486339634193e-05,
+      "loss": 1.1547,
+      "step": 187370
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.3547486033519544e-05,
+      "loss": 1.2156,
+      "step": 187380
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.35463381036198e-05,
+      "loss": 1.1492,
+      "step": 187390
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.354519017372006e-05,
+      "loss": 1.1226,
+      "step": 187400
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.354404224382031e-05,
+      "loss": 1.0624,
+      "step": 187410
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.3542894313920557e-05,
+      "loss": 1.1542,
+      "step": 187420
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.354174638402081e-05,
+      "loss": 1.1528,
+      "step": 187430
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.354059845412106e-05,
+      "loss": 1.134,
+      "step": 187440
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.353945052422132e-05,
+      "loss": 1.2351,
+      "step": 187450
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.353830259432157e-05,
+      "loss": 1.1903,
+      "step": 187460
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.353715466442182e-05,
+      "loss": 1.1613,
+      "step": 187470
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.3536006734522076e-05,
+      "loss": 1.2378,
+      "step": 187480
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.353485880462233e-05,
+      "loss": 1.1755,
+      "step": 187490
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.3533710874722576e-05,
+      "loss": 1.1398,
+      "step": 187500
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.353256294482283e-05,
+      "loss": 1.131,
+      "step": 187510
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.353141501492309e-05,
+      "loss": 1.0086,
+      "step": 187520
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.353026708502333e-05,
+      "loss": 1.1544,
+      "step": 187530
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 5.352911915512359e-05,
+      "loss": 1.0195,
+      "step": 187540
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.3527971225223845e-05,
+      "loss": 1.1643,
+      "step": 187550
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.3526823295324095e-05,
+      "loss": 1.2347,
+      "step": 187560
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.3525675365424345e-05,
+      "loss": 1.1404,
+      "step": 187570
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.35245274355246e-05,
+      "loss": 1.08,
+      "step": 187580
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.352337950562485e-05,
+      "loss": 1.0649,
+      "step": 187590
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.352223157572511e-05,
+      "loss": 1.11,
+      "step": 187600
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.352108364582536e-05,
+      "loss": 1.2528,
+      "step": 187610
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.351993571592561e-05,
+      "loss": 1.2209,
+      "step": 187620
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.3518787786025864e-05,
+      "loss": 1.0763,
+      "step": 187630
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.351763985612612e-05,
+      "loss": 1.0902,
+      "step": 187640
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.3516491926226364e-05,
+      "loss": 1.0793,
+      "step": 187650
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.351534399632662e-05,
+      "loss": 1.2232,
+      "step": 187660
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.351419606642688e-05,
+      "loss": 1.0824,
+      "step": 187670
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.351304813652712e-05,
+      "loss": 1.0512,
+      "step": 187680
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.351190020662738e-05,
+      "loss": 1.2199,
+      "step": 187690
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.3510752276727634e-05,
+      "loss": 1.1281,
+      "step": 187700
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.3509604346827884e-05,
+      "loss": 1.2442,
+      "step": 187710
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.3508456416928133e-05,
+      "loss": 1.1426,
+      "step": 187720
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.350730848702839e-05,
+      "loss": 1.06,
+      "step": 187730
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.350616055712864e-05,
+      "loss": 1.086,
+      "step": 187740
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.3505012627228897e-05,
+      "loss": 1.1683,
+      "step": 187750
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.3503864697329146e-05,
+      "loss": 1.074,
+      "step": 187760
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.3502716767429396e-05,
+      "loss": 1.1398,
+      "step": 187770
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.350156883752965e-05,
+      "loss": 0.9823,
+      "step": 187780
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.350042090762991e-05,
+      "loss": 1.2176,
+      "step": 187790
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.349927297773015e-05,
+      "loss": 1.1132,
+      "step": 187800
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.349812504783041e-05,
+      "loss": 1.1842,
+      "step": 187810
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.3496977117930666e-05,
+      "loss": 1.1068,
+      "step": 187820
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.349582918803091e-05,
+      "loss": 1.1027,
+      "step": 187830
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.3494681258131166e-05,
+      "loss": 1.0249,
+      "step": 187840
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.349353332823142e-05,
+      "loss": 1.1684,
+      "step": 187850
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.349238539833167e-05,
+      "loss": 1.1759,
+      "step": 187860
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.349123746843192e-05,
+      "loss": 1.1319,
+      "step": 187870
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.349008953853217e-05,
+      "loss": 1.0863,
+      "step": 187880
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.348894160863243e-05,
+      "loss": 1.2268,
+      "step": 187890
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.3487793678732685e-05,
+      "loss": 1.0342,
+      "step": 187900
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.348664574883293e-05,
+      "loss": 1.1049,
+      "step": 187910
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.3485497818933185e-05,
+      "loss": 0.9965,
+      "step": 187920
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.348434988903344e-05,
+      "loss": 1.0571,
+      "step": 187930
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.3483201959133684e-05,
+      "loss": 1.1123,
+      "step": 187940
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.348205402923394e-05,
+      "loss": 1.0247,
+      "step": 187950
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.34809060993342e-05,
+      "loss": 1.0621,
+      "step": 187960
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.347975816943445e-05,
+      "loss": 1.1105,
+      "step": 187970
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.34786102395347e-05,
+      "loss": 1.1301,
+      "step": 187980
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.3477462309634954e-05,
+      "loss": 1.1112,
+      "step": 187990
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.3476314379735204e-05,
+      "loss": 1.1821,
+      "step": 188000
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.347516644983546e-05,
+      "loss": 1.2256,
+      "step": 188010
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.347401851993571e-05,
+      "loss": 1.0075,
+      "step": 188020
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.347287059003596e-05,
+      "loss": 1.0698,
+      "step": 188030
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.347172266013622e-05,
+      "loss": 1.0792,
+      "step": 188040
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.3470574730236474e-05,
+      "loss": 1.2448,
+      "step": 188050
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.346942680033672e-05,
+      "loss": 1.1272,
+      "step": 188060
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.346827887043697e-05,
+      "loss": 1.1141,
+      "step": 188070
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.346713094053723e-05,
+      "loss": 1.1754,
+      "step": 188080
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.346598301063747e-05,
+      "loss": 1.1525,
+      "step": 188090
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.346483508073773e-05,
+      "loss": 1.1273,
+      "step": 188100
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.3463687150837986e-05,
+      "loss": 1.1379,
+      "step": 188110
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.3462539220938236e-05,
+      "loss": 1.1408,
+      "step": 188120
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.3461391291038486e-05,
+      "loss": 1.0899,
+      "step": 188130
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.346024336113874e-05,
+      "loss": 1.2852,
+      "step": 188140
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.345909543123899e-05,
+      "loss": 1.0989,
+      "step": 188150
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.345794750133925e-05,
+      "loss": 1.2681,
+      "step": 188160
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.34567995714395e-05,
+      "loss": 1.167,
+      "step": 188170
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.345565164153975e-05,
+      "loss": 1.0339,
+      "step": 188180
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.3454503711640005e-05,
+      "loss": 1.0024,
+      "step": 188190
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 5.345335578174026e-05,
+      "loss": 1.2342,
+      "step": 188200
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.3452207851840505e-05,
+      "loss": 1.166,
+      "step": 188210
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.345105992194076e-05,
+      "loss": 1.2459,
+      "step": 188220
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.344991199204102e-05,
+      "loss": 1.1647,
+      "step": 188230
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.344876406214126e-05,
+      "loss": 1.1333,
+      "step": 188240
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.344761613224152e-05,
+      "loss": 0.9859,
+      "step": 188250
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.3446468202341775e-05,
+      "loss": 1.2137,
+      "step": 188260
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.3445320272442025e-05,
+      "loss": 1.1519,
+      "step": 188270
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.3444172342542274e-05,
+      "loss": 1.1882,
+      "step": 188280
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.344302441264253e-05,
+      "loss": 1.1439,
+      "step": 188290
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.344187648274278e-05,
+      "loss": 1.0642,
+      "step": 188300
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.344072855284304e-05,
+      "loss": 1.0983,
+      "step": 188310
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.343958062294329e-05,
+      "loss": 1.0629,
+      "step": 188320
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.343843269304354e-05,
+      "loss": 1.2002,
+      "step": 188330
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.3437284763143794e-05,
+      "loss": 1.1682,
+      "step": 188340
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.343613683324405e-05,
+      "loss": 1.2388,
+      "step": 188350
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.3434988903344294e-05,
+      "loss": 1.0759,
+      "step": 188360
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.343384097344455e-05,
+      "loss": 1.1167,
+      "step": 188370
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.343269304354481e-05,
+      "loss": 1.13,
+      "step": 188380
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.343154511364505e-05,
+      "loss": 1.1716,
+      "step": 188390
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.3430397183745307e-05,
+      "loss": 1.2234,
+      "step": 188400
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.342924925384556e-05,
+      "loss": 0.9993,
+      "step": 188410
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.342810132394581e-05,
+      "loss": 1.1456,
+      "step": 188420
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.342695339404606e-05,
+      "loss": 1.1019,
+      "step": 188430
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.342580546414632e-05,
+      "loss": 1.077,
+      "step": 188440
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.342465753424657e-05,
+      "loss": 1.0417,
+      "step": 188450
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.3423509604346826e-05,
+      "loss": 1.1849,
+      "step": 188460
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.3422361674447076e-05,
+      "loss": 1.2481,
+      "step": 188470
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.3421213744547326e-05,
+      "loss": 1.279,
+      "step": 188480
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.342006581464758e-05,
+      "loss": 1.263,
+      "step": 188490
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.341891788474784e-05,
+      "loss": 1.1775,
+      "step": 188500
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.341776995484808e-05,
+      "loss": 1.1322,
+      "step": 188510
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.341662202494834e-05,
+      "loss": 1.2566,
+      "step": 188520
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.3415474095048595e-05,
+      "loss": 1.1841,
+      "step": 188530
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.341432616514884e-05,
+      "loss": 1.0779,
+      "step": 188540
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.3413178235249095e-05,
+      "loss": 1.1531,
+      "step": 188550
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.341203030534935e-05,
+      "loss": 1.0099,
+      "step": 188560
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.34108823754496e-05,
+      "loss": 1.1761,
+      "step": 188570
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.340973444554985e-05,
+      "loss": 1.0479,
+      "step": 188580
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.340858651565011e-05,
+      "loss": 1.0973,
+      "step": 188590
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.340743858575036e-05,
+      "loss": 1.1344,
+      "step": 188600
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.3406290655850614e-05,
+      "loss": 1.133,
+      "step": 188610
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.3405142725950864e-05,
+      "loss": 1.1532,
+      "step": 188620
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 5.3403994796051114e-05,
+      "loss": 1.0519,
+      "step": 188630
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.340284686615137e-05,
+      "loss": 1.0957,
+      "step": 188640
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.340169893625163e-05,
+      "loss": 1.1245,
+      "step": 188650
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.340055100635187e-05,
+      "loss": 1.1055,
+      "step": 188660
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.339940307645213e-05,
+      "loss": 1.2471,
+      "step": 188670
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.3398255146552384e-05,
+      "loss": 1.2529,
+      "step": 188680
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.339710721665263e-05,
+      "loss": 1.1698,
+      "step": 188690
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.3395959286752883e-05,
+      "loss": 1.1796,
+      "step": 188700
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.339481135685314e-05,
+      "loss": 1.0354,
+      "step": 188710
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.339366342695339e-05,
+      "loss": 1.0497,
+      "step": 188720
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.339251549705364e-05,
+      "loss": 1.3282,
+      "step": 188730
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.3391367567153896e-05,
+      "loss": 1.2781,
+      "step": 188740
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.3390219637254146e-05,
+      "loss": 1.1144,
+      "step": 188750
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.33890717073544e-05,
+      "loss": 1.1758,
+      "step": 188760
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.338792377745465e-05,
+      "loss": 1.2352,
+      "step": 188770
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.33867758475549e-05,
+      "loss": 1.1421,
+      "step": 188780
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.338562791765516e-05,
+      "loss": 1.1844,
+      "step": 188790
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.3384479987755416e-05,
+      "loss": 1.0827,
+      "step": 188800
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.338333205785566e-05,
+      "loss": 1.0023,
+      "step": 188810
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.3382184127955916e-05,
+      "loss": 1.1389,
+      "step": 188820
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.338103619805617e-05,
+      "loss": 1.1251,
+      "step": 188830
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.3379888268156415e-05,
+      "loss": 1.1183,
+      "step": 188840
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.337874033825667e-05,
+      "loss": 1.3229,
+      "step": 188850
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.337759240835693e-05,
+      "loss": 1.1876,
+      "step": 188860
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.337644447845718e-05,
+      "loss": 1.1324,
+      "step": 188870
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.337529654855743e-05,
+      "loss": 1.0869,
+      "step": 188880
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.3374148618657685e-05,
+      "loss": 1.1757,
+      "step": 188890
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.3373000688757935e-05,
+      "loss": 1.1102,
+      "step": 188900
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.337185275885819e-05,
+      "loss": 1.0123,
+      "step": 188910
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.337070482895844e-05,
+      "loss": 1.062,
+      "step": 188920
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.336955689905869e-05,
+      "loss": 1.1322,
+      "step": 188930
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.336840896915895e-05,
+      "loss": 1.201,
+      "step": 188940
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.3367261039259204e-05,
+      "loss": 1.0635,
+      "step": 188950
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.336611310935945e-05,
+      "loss": 1.1879,
+      "step": 188960
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.3364965179459704e-05,
+      "loss": 1.0932,
+      "step": 188970
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.336381724955996e-05,
+      "loss": 1.1507,
+      "step": 188980
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.3362669319660204e-05,
+      "loss": 1.2487,
+      "step": 188990
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.336152138976046e-05,
+      "loss": 1.0191,
+      "step": 189000
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.336037345986072e-05,
+      "loss": 1.1212,
+      "step": 189010
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.335922552996097e-05,
+      "loss": 1.2873,
+      "step": 189020
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.335807760006122e-05,
+      "loss": 1.2403,
+      "step": 189030
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.3356929670161473e-05,
+      "loss": 1.2543,
+      "step": 189040
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.335578174026172e-05,
+      "loss": 1.0093,
+      "step": 189050
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.335463381036198e-05,
+      "loss": 1.0758,
+      "step": 189060
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.335348588046223e-05,
+      "loss": 1.1177,
+      "step": 189070
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.335233795056248e-05,
+      "loss": 1.2133,
+      "step": 189080
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.3351190020662736e-05,
+      "loss": 1.192,
+      "step": 189090
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.335004209076299e-05,
+      "loss": 1.1627,
+      "step": 189100
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.3348894160863236e-05,
+      "loss": 1.1091,
+      "step": 189110
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.334774623096349e-05,
+      "loss": 1.1911,
+      "step": 189120
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.334659830106375e-05,
+      "loss": 1.1061,
+      "step": 189130
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.334545037116399e-05,
+      "loss": 1.1165,
+      "step": 189140
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.334430244126425e-05,
+      "loss": 1.2764,
+      "step": 189150
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.3343154511364506e-05,
+      "loss": 1.07,
+      "step": 189160
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.3342006581464755e-05,
+      "loss": 1.1029,
+      "step": 189170
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.3340858651565005e-05,
+      "loss": 1.0235,
+      "step": 189180
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.333971072166526e-05,
+      "loss": 1.1604,
+      "step": 189190
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.333856279176551e-05,
+      "loss": 1.2011,
+      "step": 189200
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.333741486186577e-05,
+      "loss": 1.3337,
+      "step": 189210
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.333626693196602e-05,
+      "loss": 1.0841,
+      "step": 189220
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.333511900206627e-05,
+      "loss": 1.0884,
+      "step": 189230
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.3333971072166525e-05,
+      "loss": 1.121,
+      "step": 189240
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.333282314226677e-05,
+      "loss": 1.2313,
+      "step": 189250
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.3331675212367024e-05,
+      "loss": 1.1658,
+      "step": 189260
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.333052728246728e-05,
+      "loss": 1.1561,
+      "step": 189270
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.332937935256753e-05,
+      "loss": 1.0022,
+      "step": 189280
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.332823142266778e-05,
+      "loss": 1.2119,
+      "step": 189290
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.332708349276804e-05,
+      "loss": 1.0841,
+      "step": 189300
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.332593556286829e-05,
+      "loss": 1.1131,
+      "step": 189310
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.3324787632968544e-05,
+      "loss": 1.2175,
+      "step": 189320
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.3323639703068794e-05,
+      "loss": 1.1378,
+      "step": 189330
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.3322491773169044e-05,
+      "loss": 1.0523,
+      "step": 189340
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.33213438432693e-05,
+      "loss": 1.2052,
+      "step": 189350
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.332019591336956e-05,
+      "loss": 1.1151,
+      "step": 189360
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.33190479834698e-05,
+      "loss": 1.1047,
+      "step": 189370
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.3317900053570057e-05,
+      "loss": 1.2362,
+      "step": 189380
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.331675212367031e-05,
+      "loss": 1.22,
+      "step": 189390
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.3315604193770556e-05,
+      "loss": 1.1434,
+      "step": 189400
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.331445626387081e-05,
+      "loss": 1.1448,
+      "step": 189410
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.331330833397107e-05,
+      "loss": 1.156,
+      "step": 189420
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.331216040407132e-05,
+      "loss": 1.1155,
+      "step": 189430
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.331101247417157e-05,
+      "loss": 1.2509,
+      "step": 189440
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.3309864544271826e-05,
+      "loss": 1.2863,
+      "step": 189450
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.3308716614372076e-05,
+      "loss": 1.3616,
+      "step": 189460
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.330756868447233e-05,
+      "loss": 1.173,
+      "step": 189470
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.330642075457258e-05,
+      "loss": 1.096,
+      "step": 189480
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.330527282467283e-05,
+      "loss": 1.2018,
+      "step": 189490
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.330412489477309e-05,
+      "loss": 1.1257,
+      "step": 189500
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.3302976964873345e-05,
+      "loss": 1.1645,
+      "step": 189510
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.330182903497359e-05,
+      "loss": 1.0511,
+      "step": 189520
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.3300681105073845e-05,
+      "loss": 1.1721,
+      "step": 189530
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.32995331751741e-05,
+      "loss": 1.1996,
+      "step": 189540
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.3298385245274345e-05,
+      "loss": 1.1241,
+      "step": 189550
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.32972373153746e-05,
+      "loss": 1.0573,
+      "step": 189560
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.329608938547486e-05,
+      "loss": 1.068,
+      "step": 189570
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.329494145557511e-05,
+      "loss": 1.2251,
+      "step": 189580
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.329379352567536e-05,
+      "loss": 1.0955,
+      "step": 189590
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.3292645595775614e-05,
+      "loss": 1.0356,
+      "step": 189600
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.3291497665875864e-05,
+      "loss": 1.0671,
+      "step": 189610
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.329034973597612e-05,
+      "loss": 1.161,
+      "step": 189620
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.328920180607637e-05,
+      "loss": 1.2613,
+      "step": 189630
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.328805387617662e-05,
+      "loss": 1.2628,
+      "step": 189640
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.328690594627688e-05,
+      "loss": 1.0898,
+      "step": 189650
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.3285758016377134e-05,
+      "loss": 1.2222,
+      "step": 189660
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.328461008647738e-05,
+      "loss": 1.2248,
+      "step": 189670
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.3283462156577633e-05,
+      "loss": 1.1109,
+      "step": 189680
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.328231422667789e-05,
+      "loss": 1.1831,
+      "step": 189690
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.328116629677813e-05,
+      "loss": 1.0873,
+      "step": 189700
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.328001836687839e-05,
+      "loss": 1.1394,
+      "step": 189710
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.3278870436978646e-05,
+      "loss": 1.037,
+      "step": 189720
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.3277722507078896e-05,
+      "loss": 1.0305,
+      "step": 189730
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.3276574577179146e-05,
+      "loss": 1.0436,
+      "step": 189740
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.32754266472794e-05,
+      "loss": 1.365,
+      "step": 189750
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.327427871737965e-05,
+      "loss": 1.2659,
+      "step": 189760
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.327313078747991e-05,
+      "loss": 1.1593,
+      "step": 189770
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.327198285758016e-05,
+      "loss": 1.0967,
+      "step": 189780
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.327083492768041e-05,
+      "loss": 1.279,
+      "step": 189790
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.3269686997780666e-05,
+      "loss": 1.1756,
+      "step": 189800
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.326853906788092e-05,
+      "loss": 1.0772,
+      "step": 189810
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.3267391137981165e-05,
+      "loss": 1.1506,
+      "step": 189820
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.326624320808142e-05,
+      "loss": 1.1472,
+      "step": 189830
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.326509527818168e-05,
+      "loss": 1.1826,
+      "step": 189840
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.326394734828192e-05,
+      "loss": 1.3126,
+      "step": 189850
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.326279941838218e-05,
+      "loss": 1.2221,
+      "step": 189860
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.3261651488482435e-05,
+      "loss": 1.135,
+      "step": 189870
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.3260503558582685e-05,
+      "loss": 1.1895,
+      "step": 189880
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.3259355628682935e-05,
+      "loss": 1.1447,
+      "step": 189890
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.325820769878319e-05,
+      "loss": 1.2415,
+      "step": 189900
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.325705976888344e-05,
+      "loss": 1.1084,
+      "step": 189910
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.32559118389837e-05,
+      "loss": 1.1772,
+      "step": 189920
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.325476390908395e-05,
+      "loss": 1.2486,
+      "step": 189930
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.32536159791842e-05,
+      "loss": 1.0765,
+      "step": 189940
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.3252468049284454e-05,
+      "loss": 1.21,
+      "step": 189950
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.325132011938471e-05,
+      "loss": 1.1268,
+      "step": 189960
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.3250172189484954e-05,
+      "loss": 1.1122,
+      "step": 189970
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.324902425958521e-05,
+      "loss": 1.0897,
+      "step": 189980
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.324787632968547e-05,
+      "loss": 1.2077,
+      "step": 189990
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.324672839978571e-05,
+      "loss": 1.146,
+      "step": 190000
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.324558046988597e-05,
+      "loss": 1.1006,
+      "step": 190010
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.3244432539986223e-05,
+      "loss": 1.0984,
+      "step": 190020
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.324328461008647e-05,
+      "loss": 1.1149,
+      "step": 190030
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.324213668018672e-05,
+      "loss": 1.2445,
+      "step": 190040
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.324098875028698e-05,
+      "loss": 1.2047,
+      "step": 190050
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.323984082038723e-05,
+      "loss": 1.111,
+      "step": 190060
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.3238692890487486e-05,
+      "loss": 1.1599,
+      "step": 190070
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.3237544960587736e-05,
+      "loss": 1.1145,
+      "step": 190080
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.3236397030687986e-05,
+      "loss": 1.2506,
+      "step": 190090
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.323524910078824e-05,
+      "loss": 1.0612,
+      "step": 190100
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.32341011708885e-05,
+      "loss": 1.2463,
+      "step": 190110
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.323295324098874e-05,
+      "loss": 1.1215,
+      "step": 190120
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.3231805311089e-05,
+      "loss": 1.0823,
+      "step": 190130
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.3230657381189256e-05,
+      "loss": 1.0961,
+      "step": 190140
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.32295094512895e-05,
+      "loss": 1.0669,
+      "step": 190150
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.3228361521389755e-05,
+      "loss": 1.1709,
+      "step": 190160
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.322721359149001e-05,
+      "loss": 1.0499,
+      "step": 190170
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.322606566159026e-05,
+      "loss": 1.0639,
+      "step": 190180
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.322491773169051e-05,
+      "loss": 1.0791,
+      "step": 190190
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.322376980179077e-05,
+      "loss": 1.3099,
+      "step": 190200
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.322262187189102e-05,
+      "loss": 1.1361,
+      "step": 190210
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.3221473941991275e-05,
+      "loss": 1.2594,
+      "step": 190220
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.3220326012091525e-05,
+      "loss": 1.0459,
+      "step": 190230
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.3219178082191774e-05,
+      "loss": 1.0266,
+      "step": 190240
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.321803015229203e-05,
+      "loss": 1.1226,
+      "step": 190250
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.321688222239229e-05,
+      "loss": 1.2538,
+      "step": 190260
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.321573429249253e-05,
+      "loss": 1.0034,
+      "step": 190270
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.321458636259279e-05,
+      "loss": 1.1266,
+      "step": 190280
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.3213438432693044e-05,
+      "loss": 1.1186,
+      "step": 190290
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.321229050279329e-05,
+      "loss": 1.1237,
+      "step": 190300
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.3211142572893544e-05,
+      "loss": 1.1071,
+      "step": 190310
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.32099946429938e-05,
+      "loss": 1.1812,
+      "step": 190320
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.320884671309405e-05,
+      "loss": 1.2077,
+      "step": 190330
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.32076987831943e-05,
+      "loss": 1.2692,
+      "step": 190340
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.320655085329456e-05,
+      "loss": 1.114,
+      "step": 190350
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.3205402923394807e-05,
+      "loss": 1.2311,
+      "step": 190360
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.320425499349506e-05,
+      "loss": 1.0308,
+      "step": 190370
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.320310706359531e-05,
+      "loss": 1.105,
+      "step": 190380
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.320195913369556e-05,
+      "loss": 1.2439,
+      "step": 190390
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.320081120379582e-05,
+      "loss": 1.2705,
+      "step": 190400
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.3199663273896076e-05,
+      "loss": 1.0506,
+      "step": 190410
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.319851534399632e-05,
+      "loss": 1.2127,
+      "step": 190420
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.3197367414096576e-05,
+      "loss": 1.0151,
+      "step": 190430
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.319621948419683e-05,
+      "loss": 1.1969,
+      "step": 190440
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.3195071554297076e-05,
+      "loss": 1.0515,
+      "step": 190450
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.319392362439733e-05,
+      "loss": 1.1477,
+      "step": 190460
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.319277569449759e-05,
+      "loss": 1.2131,
+      "step": 190470
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.319162776459784e-05,
+      "loss": 1.2858,
+      "step": 190480
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.319047983469809e-05,
+      "loss": 1.0163,
+      "step": 190490
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.3189331904798345e-05,
+      "loss": 1.116,
+      "step": 190500
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.3188183974898595e-05,
+      "loss": 1.139,
+      "step": 190510
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.318703604499885e-05,
+      "loss": 1.1329,
+      "step": 190520
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.31858881150991e-05,
+      "loss": 1.0647,
+      "step": 190530
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.318474018519935e-05,
+      "loss": 1.0936,
+      "step": 190540
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.318359225529961e-05,
+      "loss": 1.1839,
+      "step": 190550
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.3182444325399865e-05,
+      "loss": 1.1191,
+      "step": 190560
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.318129639550011e-05,
+      "loss": 1.0891,
+      "step": 190570
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.3180148465600364e-05,
+      "loss": 1.2399,
+      "step": 190580
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.317900053570062e-05,
+      "loss": 1.1349,
+      "step": 190590
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.3177852605800864e-05,
+      "loss": 1.1152,
+      "step": 190600
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.317670467590112e-05,
+      "loss": 0.9396,
+      "step": 190610
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.317555674600137e-05,
+      "loss": 1.2619,
+      "step": 190620
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.317440881610163e-05,
+      "loss": 1.1612,
+      "step": 190630
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.317326088620188e-05,
+      "loss": 1.1467,
+      "step": 190640
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.317211295630213e-05,
+      "loss": 1.0731,
+      "step": 190650
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.3170965026402383e-05,
+      "loss": 1.1001,
+      "step": 190660
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.316981709650264e-05,
+      "loss": 1.1133,
+      "step": 190670
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.316866916660288e-05,
+      "loss": 1.1162,
+      "step": 190680
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.316752123670314e-05,
+      "loss": 1.271,
+      "step": 190690
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.3166373306803396e-05,
+      "loss": 1.0661,
+      "step": 190700
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.316522537690364e-05,
+      "loss": 0.9841,
+      "step": 190710
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.3164077447003896e-05,
+      "loss": 1.0472,
+      "step": 190720
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.316292951710415e-05,
+      "loss": 1.1599,
+      "step": 190730
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.31617815872044e-05,
+      "loss": 1.113,
+      "step": 190740
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.316063365730465e-05,
+      "loss": 1.0768,
+      "step": 190750
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.315948572740491e-05,
+      "loss": 1.1427,
+      "step": 190760
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.315833779750516e-05,
+      "loss": 1.0348,
+      "step": 190770
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.3157189867605416e-05,
+      "loss": 1.0589,
+      "step": 190780
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.3156041937705665e-05,
+      "loss": 1.2313,
+      "step": 190790
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.3154894007805915e-05,
+      "loss": 1.1925,
+      "step": 190800
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.315374607790617e-05,
+      "loss": 1.1384,
+      "step": 190810
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.315259814800643e-05,
+      "loss": 1.0974,
+      "step": 190820
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.315145021810667e-05,
+      "loss": 1.207,
+      "step": 190830
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.315030228820693e-05,
+      "loss": 1.1912,
+      "step": 190840
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.3149154358307185e-05,
+      "loss": 1.2359,
+      "step": 190850
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.314800642840743e-05,
+      "loss": 1.2457,
+      "step": 190860
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.3146858498507685e-05,
+      "loss": 1.0649,
+      "step": 190870
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.314571056860794e-05,
+      "loss": 1.1485,
+      "step": 190880
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.314456263870819e-05,
+      "loss": 1.1767,
+      "step": 190890
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.314341470880844e-05,
+      "loss": 1.238,
+      "step": 190900
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.31422667789087e-05,
+      "loss": 1.2187,
+      "step": 190910
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.314111884900895e-05,
+      "loss": 1.1392,
+      "step": 190920
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.3139970919109204e-05,
+      "loss": 1.2181,
+      "step": 190930
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.3138822989209454e-05,
+      "loss": 1.156,
+      "step": 190940
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.3137675059309704e-05,
+      "loss": 1.1167,
+      "step": 190950
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.313652712940996e-05,
+      "loss": 1.0856,
+      "step": 190960
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.313537919951022e-05,
+      "loss": 1.1525,
+      "step": 190970
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.313423126961046e-05,
+      "loss": 1.252,
+      "step": 190980
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.313308333971072e-05,
+      "loss": 1.1728,
+      "step": 190990
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.313193540981097e-05,
+      "loss": 1.1538,
+      "step": 191000
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.3130787479911216e-05,
+      "loss": 1.1476,
+      "step": 191010
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.312963955001147e-05,
+      "loss": 1.173,
+      "step": 191020
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.312849162011173e-05,
+      "loss": 1.1778,
+      "step": 191030
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.312734369021198e-05,
+      "loss": 1.2317,
+      "step": 191040
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.312619576031223e-05,
+      "loss": 1.2082,
+      "step": 191050
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.3125047830412486e-05,
+      "loss": 1.1344,
+      "step": 191060
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.3123899900512736e-05,
+      "loss": 1.1665,
+      "step": 191070
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.312275197061299e-05,
+      "loss": 1.1821,
+      "step": 191080
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.312160404071324e-05,
+      "loss": 1.0705,
+      "step": 191090
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.312045611081349e-05,
+      "loss": 1.1948,
+      "step": 191100
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.311930818091375e-05,
+      "loss": 1.1276,
+      "step": 191110
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.3118160251014006e-05,
+      "loss": 1.1398,
+      "step": 191120
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.311701232111425e-05,
+      "loss": 1.2065,
+      "step": 191130
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.3115864391214505e-05,
+      "loss": 1.1837,
+      "step": 191140
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.311471646131476e-05,
+      "loss": 1.1157,
+      "step": 191150
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.3113568531415005e-05,
+      "loss": 1.1526,
+      "step": 191160
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.311242060151526e-05,
+      "loss": 1.1101,
+      "step": 191170
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.311127267161552e-05,
+      "loss": 1.0394,
+      "step": 191180
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.311012474171577e-05,
+      "loss": 1.1622,
+      "step": 191190
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.310897681181602e-05,
+      "loss": 1.0929,
+      "step": 191200
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.3107828881916275e-05,
+      "loss": 1.1204,
+      "step": 191210
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.3106680952016524e-05,
+      "loss": 1.1288,
+      "step": 191220
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.310553302211678e-05,
+      "loss": 1.193,
+      "step": 191230
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.310438509221703e-05,
+      "loss": 1.1666,
+      "step": 191240
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.310323716231728e-05,
+      "loss": 1.1598,
+      "step": 191250
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.310208923241754e-05,
+      "loss": 1.0921,
+      "step": 191260
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.3100941302517794e-05,
+      "loss": 1.2315,
+      "step": 191270
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.309979337261804e-05,
+      "loss": 1.1148,
+      "step": 191280
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.3098645442718294e-05,
+      "loss": 1.1963,
+      "step": 191290
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.309749751281855e-05,
+      "loss": 1.1571,
+      "step": 191300
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.3096349582918793e-05,
+      "loss": 1.0442,
+      "step": 191310
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.309520165301905e-05,
+      "loss": 1.1879,
+      "step": 191320
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.309405372311931e-05,
+      "loss": 1.1042,
+      "step": 191330
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.3092905793219557e-05,
+      "loss": 1.1835,
+      "step": 191340
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.3091757863319806e-05,
+      "loss": 1.0829,
+      "step": 191350
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.309060993342006e-05,
+      "loss": 1.1263,
+      "step": 191360
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.308946200352031e-05,
+      "loss": 1.1128,
+      "step": 191370
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.308831407362057e-05,
+      "loss": 1.1794,
+      "step": 191380
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.308716614372082e-05,
+      "loss": 1.1654,
+      "step": 191390
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.308601821382107e-05,
+      "loss": 1.2954,
+      "step": 191400
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.3084870283921326e-05,
+      "loss": 1.0541,
+      "step": 191410
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.308372235402158e-05,
+      "loss": 1.1707,
+      "step": 191420
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.3082574424121826e-05,
+      "loss": 1.2158,
+      "step": 191430
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.308142649422208e-05,
+      "loss": 1.2402,
+      "step": 191440
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.308027856432234e-05,
+      "loss": 1.1591,
+      "step": 191450
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.307913063442258e-05,
+      "loss": 1.1502,
+      "step": 191460
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.307798270452284e-05,
+      "loss": 1.1488,
+      "step": 191470
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.3076834774623095e-05,
+      "loss": 1.2023,
+      "step": 191480
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.3075686844723345e-05,
+      "loss": 1.1287,
+      "step": 191490
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.3074538914823595e-05,
+      "loss": 1.1491,
+      "step": 191500
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.307339098492385e-05,
+      "loss": 1.1847,
+      "step": 191510
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.30722430550241e-05,
+      "loss": 1.1233,
+      "step": 191520
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.307109512512436e-05,
+      "loss": 1.0969,
+      "step": 191530
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.306994719522461e-05,
+      "loss": 1.1618,
+      "step": 191540
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.306879926532486e-05,
+      "loss": 1.1712,
+      "step": 191550
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.3067651335425114e-05,
+      "loss": 1.129,
+      "step": 191560
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.306650340552537e-05,
+      "loss": 1.1551,
+      "step": 191570
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.3065355475625614e-05,
+      "loss": 1.1795,
+      "step": 191580
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.306420754572587e-05,
+      "loss": 0.9655,
+      "step": 191590
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.306305961582613e-05,
+      "loss": 1.1897,
+      "step": 191600
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.306191168592637e-05,
+      "loss": 1.2341,
+      "step": 191610
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.306076375602663e-05,
+      "loss": 1.1785,
+      "step": 191620
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.3059615826126884e-05,
+      "loss": 1.1191,
+      "step": 191630
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.3058467896227133e-05,
+      "loss": 1.2705,
+      "step": 191640
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.305731996632738e-05,
+      "loss": 1.1272,
+      "step": 191650
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.305617203642764e-05,
+      "loss": 1.18,
+      "step": 191660
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.305502410652789e-05,
+      "loss": 1.2009,
+      "step": 191670
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 5.3053876176628146e-05,
+      "loss": 1.2438,
+      "step": 191680
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.3052728246728396e-05,
+      "loss": 1.1239,
+      "step": 191690
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.3051580316828646e-05,
+      "loss": 1.1766,
+      "step": 191700
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.30504323869289e-05,
+      "loss": 1.1381,
+      "step": 191710
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.304928445702916e-05,
+      "loss": 1.1139,
+      "step": 191720
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.30481365271294e-05,
+      "loss": 1.0927,
+      "step": 191730
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.304698859722966e-05,
+      "loss": 1.1964,
+      "step": 191740
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.3045840667329916e-05,
+      "loss": 1.1848,
+      "step": 191750
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.304469273743016e-05,
+      "loss": 1.2278,
+      "step": 191760
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.3043544807530415e-05,
+      "loss": 1.2211,
+      "step": 191770
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.304239687763067e-05,
+      "loss": 1.1241,
+      "step": 191780
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.304124894773092e-05,
+      "loss": 1.0477,
+      "step": 191790
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.304010101783117e-05,
+      "loss": 1.1851,
+      "step": 191800
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.303895308793143e-05,
+      "loss": 1.18,
+      "step": 191810
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.303780515803168e-05,
+      "loss": 1.1541,
+      "step": 191820
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.3036657228131935e-05,
+      "loss": 1.0791,
+      "step": 191830
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.3035509298232185e-05,
+      "loss": 1.2547,
+      "step": 191840
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.3034361368332435e-05,
+      "loss": 1.0933,
+      "step": 191850
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.303321343843269e-05,
+      "loss": 1.0015,
+      "step": 191860
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.303206550853295e-05,
+      "loss": 1.1612,
+      "step": 191870
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.303091757863319e-05,
+      "loss": 1.116,
+      "step": 191880
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.302976964873345e-05,
+      "loss": 1.2155,
+      "step": 191890
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.3028621718833704e-05,
+      "loss": 1.0819,
+      "step": 191900
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.302747378893395e-05,
+      "loss": 1.0463,
+      "step": 191910
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.3026325859034204e-05,
+      "loss": 1.1671,
+      "step": 191920
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.302517792913446e-05,
+      "loss": 1.137,
+      "step": 191930
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.302402999923471e-05,
+      "loss": 1.19,
+      "step": 191940
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.302288206933496e-05,
+      "loss": 1.2024,
+      "step": 191950
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.302173413943522e-05,
+      "loss": 1.1844,
+      "step": 191960
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.302058620953547e-05,
+      "loss": 1.0845,
+      "step": 191970
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.301943827963572e-05,
+      "loss": 1.0149,
+      "step": 191980
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.3018290349735966e-05,
+      "loss": 1.1428,
+      "step": 191990
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.301714241983622e-05,
+      "loss": 1.062,
+      "step": 192000
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.301599448993648e-05,
+      "loss": 1.1375,
+      "step": 192010
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.301484656003672e-05,
+      "loss": 1.1512,
+      "step": 192020
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.301369863013698e-05,
+      "loss": 1.2122,
+      "step": 192030
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.3012550700237236e-05,
+      "loss": 1.1062,
+      "step": 192040
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.3011402770337486e-05,
+      "loss": 1.2601,
+      "step": 192050
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.3010254840437736e-05,
+      "loss": 1.0309,
+      "step": 192060
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.300910691053799e-05,
+      "loss": 1.1324,
+      "step": 192070
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.300795898063824e-05,
+      "loss": 1.2168,
+      "step": 192080
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.30068110507385e-05,
+      "loss": 1.112,
+      "step": 192090
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.300566312083875e-05,
+      "loss": 1.0737,
+      "step": 192100
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.3004515190939e-05,
+      "loss": 1.0787,
+      "step": 192110
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.3003367261039255e-05,
+      "loss": 1.1121,
+      "step": 192120
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.300221933113951e-05,
+      "loss": 1.1863,
+      "step": 192130
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.3001071401239755e-05,
+      "loss": 1.0158,
+      "step": 192140
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.299992347134001e-05,
+      "loss": 1.2571,
+      "step": 192150
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.299877554144027e-05,
+      "loss": 1.0947,
+      "step": 192160
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.299762761154051e-05,
+      "loss": 1.1056,
+      "step": 192170
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.299647968164077e-05,
+      "loss": 1.1233,
+      "step": 192180
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.2995331751741025e-05,
+      "loss": 1.0598,
+      "step": 192190
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.2994183821841274e-05,
+      "loss": 1.141,
+      "step": 192200
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.2993035891941524e-05,
+      "loss": 1.1527,
+      "step": 192210
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.299188796204178e-05,
+      "loss": 1.2869,
+      "step": 192220
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.299074003214203e-05,
+      "loss": 1.0645,
+      "step": 192230
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.298959210224229e-05,
+      "loss": 1.2212,
+      "step": 192240
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.298844417234254e-05,
+      "loss": 1.1484,
+      "step": 192250
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.298729624244279e-05,
+      "loss": 1.2231,
+      "step": 192260
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.2986148312543044e-05,
+      "loss": 1.2275,
+      "step": 192270
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.29850003826433e-05,
+      "loss": 1.2235,
+      "step": 192280
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.2983852452743543e-05,
+      "loss": 1.1841,
+      "step": 192290
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.29827045228438e-05,
+      "loss": 1.1619,
+      "step": 192300
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.298155659294406e-05,
+      "loss": 1.321,
+      "step": 192310
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.29804086630443e-05,
+      "loss": 1.3258,
+      "step": 192320
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.2979260733144556e-05,
+      "loss": 1.134,
+      "step": 192330
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.297811280324481e-05,
+      "loss": 1.2261,
+      "step": 192340
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.297696487334506e-05,
+      "loss": 1.068,
+      "step": 192350
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.297581694344531e-05,
+      "loss": 1.1566,
+      "step": 192360
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.297466901354557e-05,
+      "loss": 1.2701,
+      "step": 192370
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.297352108364582e-05,
+      "loss": 1.362,
+      "step": 192380
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.2972373153746076e-05,
+      "loss": 1.1626,
+      "step": 192390
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.2971225223846326e-05,
+      "loss": 1.2142,
+      "step": 192400
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.2970077293946576e-05,
+      "loss": 1.2332,
+      "step": 192410
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.296892936404683e-05,
+      "loss": 1.1154,
+      "step": 192420
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.296778143414709e-05,
+      "loss": 1.2475,
+      "step": 192430
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.296663350424733e-05,
+      "loss": 1.1953,
+      "step": 192440
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.296548557434759e-05,
+      "loss": 1.1743,
+      "step": 192450
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.2964337644447845e-05,
+      "loss": 1.1566,
+      "step": 192460
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.296318971454809e-05,
+      "loss": 1.1522,
+      "step": 192470
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.2962041784648345e-05,
+      "loss": 1.1514,
+      "step": 192480
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.29608938547486e-05,
+      "loss": 1.2201,
+      "step": 192490
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.295974592484885e-05,
+      "loss": 1.0561,
+      "step": 192500
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.29585979949491e-05,
+      "loss": 1.0755,
+      "step": 192510
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.295745006504936e-05,
+      "loss": 1.1842,
+      "step": 192520
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.295630213514961e-05,
+      "loss": 1.1222,
+      "step": 192530
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.2955154205249864e-05,
+      "loss": 1.1367,
+      "step": 192540
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.2954006275350114e-05,
+      "loss": 1.2087,
+      "step": 192550
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.2952858345450364e-05,
+      "loss": 1.275,
+      "step": 192560
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.295171041555062e-05,
+      "loss": 1.2358,
+      "step": 192570
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.295056248565088e-05,
+      "loss": 1.1441,
+      "step": 192580
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.294941455575112e-05,
+      "loss": 1.2043,
+      "step": 192590
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.294826662585138e-05,
+      "loss": 1.1339,
+      "step": 192600
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.2947118695951634e-05,
+      "loss": 1.2166,
+      "step": 192610
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.294597076605188e-05,
+      "loss": 1.0969,
+      "step": 192620
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.294482283615213e-05,
+      "loss": 1.0853,
+      "step": 192630
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.294367490625239e-05,
+      "loss": 1.1266,
+      "step": 192640
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.294252697635264e-05,
+      "loss": 1.0636,
+      "step": 192650
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.294137904645289e-05,
+      "loss": 1.1377,
+      "step": 192660
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.2940231116553146e-05,
+      "loss": 1.1559,
+      "step": 192670
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.2939083186653396e-05,
+      "loss": 1.1281,
+      "step": 192680
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.293793525675365e-05,
+      "loss": 1.1012,
+      "step": 192690
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.29367873268539e-05,
+      "loss": 1.1386,
+      "step": 192700
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.293563939695415e-05,
+      "loss": 1.2315,
+      "step": 192710
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.293449146705441e-05,
+      "loss": 1.1856,
+      "step": 192720
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.2933343537154666e-05,
+      "loss": 1.068,
+      "step": 192730
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.293219560725491e-05,
+      "loss": 1.0818,
+      "step": 192740
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.2931047677355165e-05,
+      "loss": 1.1084,
+      "step": 192750
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.292989974745542e-05,
+      "loss": 1.0532,
+      "step": 192760
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.2928751817555665e-05,
+      "loss": 1.2323,
+      "step": 192770
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.292760388765592e-05,
+      "loss": 1.1066,
+      "step": 192780
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.292645595775618e-05,
+      "loss": 1.2751,
+      "step": 192790
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.292530802785643e-05,
+      "loss": 1.1226,
+      "step": 192800
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.292416009795668e-05,
+      "loss": 1.2141,
+      "step": 192810
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.2923012168056935e-05,
+      "loss": 1.1066,
+      "step": 192820
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.2921864238157185e-05,
+      "loss": 1.2987,
+      "step": 192830
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.292071630825744e-05,
+      "loss": 1.1964,
+      "step": 192840
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.291956837835769e-05,
+      "loss": 1.1741,
+      "step": 192850
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.291842044845794e-05,
+      "loss": 1.0849,
+      "step": 192860
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.29172725185582e-05,
+      "loss": 1.0562,
+      "step": 192870
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.2916124588658454e-05,
+      "loss": 1.1458,
+      "step": 192880
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.29149766587587e-05,
+      "loss": 1.2532,
+      "step": 192890
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.2913828728858954e-05,
+      "loss": 1.071,
+      "step": 192900
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.291268079895921e-05,
+      "loss": 1.2934,
+      "step": 192910
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.2911532869059454e-05,
+      "loss": 1.1446,
+      "step": 192920
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.291038493915971e-05,
+      "loss": 1.1032,
+      "step": 192930
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.290923700925997e-05,
+      "loss": 1.1438,
+      "step": 192940
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.290808907936022e-05,
+      "loss": 1.1714,
+      "step": 192950
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.2906941149460467e-05,
+      "loss": 1.1558,
+      "step": 192960
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.290579321956072e-05,
+      "loss": 1.0347,
+      "step": 192970
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.290464528966097e-05,
+      "loss": 1.1895,
+      "step": 192980
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.290349735976123e-05,
+      "loss": 1.0904,
+      "step": 192990
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.290234942986148e-05,
+      "loss": 1.4132,
+      "step": 193000
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.290120149996173e-05,
+      "loss": 1.1722,
+      "step": 193010
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.2900053570061986e-05,
+      "loss": 1.1113,
+      "step": 193020
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.289890564016224e-05,
+      "loss": 1.2671,
+      "step": 193030
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.2897757710262486e-05,
+      "loss": 1.162,
+      "step": 193040
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.289660978036274e-05,
+      "loss": 1.0474,
+      "step": 193050
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.2895461850463e-05,
+      "loss": 1.1342,
+      "step": 193060
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.289431392056324e-05,
+      "loss": 1.1322,
+      "step": 193070
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.28931659906635e-05,
+      "loss": 1.1588,
+      "step": 193080
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.2892018060763755e-05,
+      "loss": 1.3911,
+      "step": 193090
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.2890870130864005e-05,
+      "loss": 1.1515,
+      "step": 193100
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.2889722200964255e-05,
+      "loss": 1.0285,
+      "step": 193110
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.288857427106451e-05,
+      "loss": 1.1451,
+      "step": 193120
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.288742634116476e-05,
+      "loss": 1.1099,
+      "step": 193130
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.288627841126502e-05,
+      "loss": 1.2016,
+      "step": 193140
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.288513048136527e-05,
+      "loss": 1.1621,
+      "step": 193150
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.288398255146552e-05,
+      "loss": 1.0692,
+      "step": 193160
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.2882834621565775e-05,
+      "loss": 1.0175,
+      "step": 193170
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.288168669166603e-05,
+      "loss": 1.1438,
+      "step": 193180
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.2880538761766274e-05,
+      "loss": 1.1217,
+      "step": 193190
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.287939083186653e-05,
+      "loss": 1.2256,
+      "step": 193200
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.287824290196679e-05,
+      "loss": 1.3592,
+      "step": 193210
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.287709497206703e-05,
+      "loss": 1.0623,
+      "step": 193220
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.287594704216729e-05,
+      "loss": 1.2127,
+      "step": 193230
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.2874799112267544e-05,
+      "loss": 1.1593,
+      "step": 193240
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.2873651182367794e-05,
+      "loss": 1.0474,
+      "step": 193250
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.2872503252468044e-05,
+      "loss": 1.1446,
+      "step": 193260
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.28713553225683e-05,
+      "loss": 1.2487,
+      "step": 193270
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.287020739266855e-05,
+      "loss": 1.0887,
+      "step": 193280
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.286905946276881e-05,
+      "loss": 1.0812,
+      "step": 193290
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.2867911532869057e-05,
+      "loss": 1.1877,
+      "step": 193300
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.2866763602969306e-05,
+      "loss": 1.0885,
+      "step": 193310
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.286561567306956e-05,
+      "loss": 1.1311,
+      "step": 193320
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.286446774316982e-05,
+      "loss": 1.1752,
+      "step": 193330
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.286331981327006e-05,
+      "loss": 1.2143,
+      "step": 193340
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.286217188337032e-05,
+      "loss": 1.08,
+      "step": 193350
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.286102395347057e-05,
+      "loss": 1.1643,
+      "step": 193360
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.285987602357082e-05,
+      "loss": 1.0716,
+      "step": 193370
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.2858728093671076e-05,
+      "loss": 1.1357,
+      "step": 193380
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.2857580163771326e-05,
+      "loss": 1.1475,
+      "step": 193390
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.285643223387158e-05,
+      "loss": 1.2231,
+      "step": 193400
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.285528430397183e-05,
+      "loss": 1.2634,
+      "step": 193410
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.285413637407208e-05,
+      "loss": 1.2129,
+      "step": 193420
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.285298844417234e-05,
+      "loss": 1.2529,
+      "step": 193430
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.2851840514272595e-05,
+      "loss": 1.0524,
+      "step": 193440
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.285069258437284e-05,
+      "loss": 1.1446,
+      "step": 193450
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.2849544654473095e-05,
+      "loss": 1.1433,
+      "step": 193460
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.284839672457335e-05,
+      "loss": 1.253,
+      "step": 193470
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.2847248794673595e-05,
+      "loss": 1.0729,
+      "step": 193480
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.284610086477385e-05,
+      "loss": 1.0779,
+      "step": 193490
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.284495293487411e-05,
+      "loss": 1.1416,
+      "step": 193500
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.284380500497436e-05,
+      "loss": 1.1563,
+      "step": 193510
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.284265707507461e-05,
+      "loss": 1.1613,
+      "step": 193520
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.2841509145174864e-05,
+      "loss": 1.1132,
+      "step": 193530
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.2840361215275114e-05,
+      "loss": 1.2104,
+      "step": 193540
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.283921328537537e-05,
+      "loss": 1.2935,
+      "step": 193550
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.283806535547562e-05,
+      "loss": 1.1298,
+      "step": 193560
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.283691742557587e-05,
+      "loss": 1.036,
+      "step": 193570
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.283576949567613e-05,
+      "loss": 1.1325,
+      "step": 193580
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.2834621565776384e-05,
+      "loss": 1.1246,
+      "step": 193590
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.283347363587663e-05,
+      "loss": 1.1894,
+      "step": 193600
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.283232570597688e-05,
+      "loss": 1.1166,
+      "step": 193610
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.283117777607714e-05,
+      "loss": 1.1624,
+      "step": 193620
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.283002984617738e-05,
+      "loss": 1.1293,
+      "step": 193630
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.282888191627764e-05,
+      "loss": 1.2038,
+      "step": 193640
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.2827733986377896e-05,
+      "loss": 1.0107,
+      "step": 193650
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.2826586056478146e-05,
+      "loss": 1.2049,
+      "step": 193660
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.2825438126578396e-05,
+      "loss": 1.1302,
+      "step": 193670
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.282429019667865e-05,
+      "loss": 1.2203,
+      "step": 193680
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.28231422667789e-05,
+      "loss": 1.1439,
+      "step": 193690
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.282199433687916e-05,
+      "loss": 1.1873,
+      "step": 193700
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.282084640697941e-05,
+      "loss": 1.0841,
+      "step": 193710
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.281969847707966e-05,
+      "loss": 1.1215,
+      "step": 193720
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.2818550547179915e-05,
+      "loss": 1.1164,
+      "step": 193730
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.281740261728017e-05,
+      "loss": 1.096,
+      "step": 193740
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.2816254687380415e-05,
+      "loss": 1.2198,
+      "step": 193750
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.281510675748067e-05,
+      "loss": 1.1385,
+      "step": 193760
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.281395882758093e-05,
+      "loss": 1.0119,
+      "step": 193770
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.281281089768117e-05,
+      "loss": 1.1715,
+      "step": 193780
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.281166296778143e-05,
+      "loss": 1.2218,
+      "step": 193790
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.2810515037881685e-05,
+      "loss": 1.1158,
+      "step": 193800
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.2809367107981935e-05,
+      "loss": 1.1001,
+      "step": 193810
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.2808219178082184e-05,
+      "loss": 1.1175,
+      "step": 193820
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.280707124818244e-05,
+      "loss": 1.19,
+      "step": 193830
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.280592331828269e-05,
+      "loss": 1.1038,
+      "step": 193840
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.280477538838295e-05,
+      "loss": 1.2648,
+      "step": 193850
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.28036274584832e-05,
+      "loss": 1.063,
+      "step": 193860
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.280247952858345e-05,
+      "loss": 1.0998,
+      "step": 193870
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.2801331598683704e-05,
+      "loss": 1.0687,
+      "step": 193880
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.280018366878396e-05,
+      "loss": 1.2513,
+      "step": 193890
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.2799035738884204e-05,
+      "loss": 1.1779,
+      "step": 193900
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.279788780898446e-05,
+      "loss": 1.1608,
+      "step": 193910
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.279673987908472e-05,
+      "loss": 1.0485,
+      "step": 193920
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.279559194918496e-05,
+      "loss": 1.2963,
+      "step": 193930
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.2794444019285217e-05,
+      "loss": 1.1491,
+      "step": 193940
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.279329608938547e-05,
+      "loss": 1.1777,
+      "step": 193950
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.279214815948572e-05,
+      "loss": 1.1066,
+      "step": 193960
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.279100022958597e-05,
+      "loss": 1.1051,
+      "step": 193970
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.278985229968623e-05,
+      "loss": 1.2474,
+      "step": 193980
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.278870436978648e-05,
+      "loss": 1.3375,
+      "step": 193990
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.2787556439886736e-05,
+      "loss": 1.1469,
+      "step": 194000
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.2786408509986986e-05,
+      "loss": 1.1323,
+      "step": 194010
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.2785260580087236e-05,
+      "loss": 1.2311,
+      "step": 194020
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.278411265018749e-05,
+      "loss": 1.142,
+      "step": 194030
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.278296472028775e-05,
+      "loss": 1.0197,
+      "step": 194040
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.278181679038799e-05,
+      "loss": 1.1072,
+      "step": 194050
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.278066886048825e-05,
+      "loss": 1.2312,
+      "step": 194060
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.2779520930588505e-05,
+      "loss": 1.1509,
+      "step": 194070
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.277837300068875e-05,
+      "loss": 1.1742,
+      "step": 194080
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.2777225070789005e-05,
+      "loss": 1.138,
+      "step": 194090
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.277607714088926e-05,
+      "loss": 1.1465,
+      "step": 194100
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.277492921098951e-05,
+      "loss": 1.3127,
+      "step": 194110
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.277378128108976e-05,
+      "loss": 1.167,
+      "step": 194120
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.277263335119002e-05,
+      "loss": 1.1347,
+      "step": 194130
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.277148542129027e-05,
+      "loss": 1.061,
+      "step": 194140
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.2770337491390525e-05,
+      "loss": 1.1498,
+      "step": 194150
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.2769189561490774e-05,
+      "loss": 1.172,
+      "step": 194160
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.2768041631591024e-05,
+      "loss": 1.0589,
+      "step": 194170
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.276689370169128e-05,
+      "loss": 1.1557,
+      "step": 194180
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.276574577179154e-05,
+      "loss": 1.2232,
+      "step": 194190
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.276459784189178e-05,
+      "loss": 1.2238,
+      "step": 194200
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.276344991199204e-05,
+      "loss": 1.0358,
+      "step": 194210
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.2762301982092294e-05,
+      "loss": 1.0471,
+      "step": 194220
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.276115405219254e-05,
+      "loss": 1.1836,
+      "step": 194230
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.2760006122292794e-05,
+      "loss": 1.1668,
+      "step": 194240
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.275885819239305e-05,
+      "loss": 1.0426,
+      "step": 194250
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.27577102624933e-05,
+      "loss": 1.1625,
+      "step": 194260
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.275656233259355e-05,
+      "loss": 1.2047,
+      "step": 194270
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.2755414402693807e-05,
+      "loss": 1.195,
+      "step": 194280
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.2754266472794056e-05,
+      "loss": 1.1181,
+      "step": 194290
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.275311854289431e-05,
+      "loss": 1.3066,
+      "step": 194300
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.275197061299456e-05,
+      "loss": 1.1203,
+      "step": 194310
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.275082268309481e-05,
+      "loss": 1.2267,
+      "step": 194320
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.274967475319507e-05,
+      "loss": 1.0801,
+      "step": 194330
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.2748526823295326e-05,
+      "loss": 0.9787,
+      "step": 194340
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.274737889339557e-05,
+      "loss": 1.0876,
+      "step": 194350
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.2746230963495826e-05,
+      "loss": 1.1687,
+      "step": 194360
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.274508303359608e-05,
+      "loss": 1.0307,
+      "step": 194370
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.2743935103696325e-05,
+      "loss": 1.1723,
+      "step": 194380
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.274278717379658e-05,
+      "loss": 1.1714,
+      "step": 194390
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.274163924389684e-05,
+      "loss": 1.2213,
+      "step": 194400
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.274049131399709e-05,
+      "loss": 1.0429,
+      "step": 194410
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.273934338409734e-05,
+      "loss": 1.0782,
+      "step": 194420
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.2738195454197595e-05,
+      "loss": 1.2197,
+      "step": 194430
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.2737047524297845e-05,
+      "loss": 1.0967,
+      "step": 194440
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.27358995943981e-05,
+      "loss": 1.1069,
+      "step": 194450
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.273475166449835e-05,
+      "loss": 1.1021,
+      "step": 194460
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.27336037345986e-05,
+      "loss": 1.1333,
+      "step": 194470
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.273245580469886e-05,
+      "loss": 1.1942,
+      "step": 194480
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.2731307874799114e-05,
+      "loss": 0.9902,
+      "step": 194490
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.273015994489936e-05,
+      "loss": 1.0398,
+      "step": 194500
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.2729012014999614e-05,
+      "loss": 1.033,
+      "step": 194510
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.272786408509987e-05,
+      "loss": 1.2383,
+      "step": 194520
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.2726716155200114e-05,
+      "loss": 1.0725,
+      "step": 194530
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.272556822530037e-05,
+      "loss": 1.0781,
+      "step": 194540
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.272442029540063e-05,
+      "loss": 1.1907,
+      "step": 194550
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.272327236550088e-05,
+      "loss": 1.1359,
+      "step": 194560
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.272212443560113e-05,
+      "loss": 1.1489,
+      "step": 194570
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.2720976505701383e-05,
+      "loss": 1.092,
+      "step": 194580
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.271982857580163e-05,
+      "loss": 1.1372,
+      "step": 194590
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.271868064590189e-05,
+      "loss": 1.1926,
+      "step": 194600
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.271753271600214e-05,
+      "loss": 1.1935,
+      "step": 194610
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.271638478610239e-05,
+      "loss": 1.1522,
+      "step": 194620
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.2715236856202646e-05,
+      "loss": 1.2256,
+      "step": 194630
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.27140889263029e-05,
+      "loss": 1.1813,
+      "step": 194640
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.2712940996403146e-05,
+      "loss": 1.2013,
+      "step": 194650
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.27117930665034e-05,
+      "loss": 1.0452,
+      "step": 194660
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.271064513660366e-05,
+      "loss": 1.0457,
+      "step": 194670
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.27094972067039e-05,
+      "loss": 1.2434,
+      "step": 194680
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.270834927680416e-05,
+      "loss": 1.1797,
+      "step": 194690
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.2707201346904416e-05,
+      "loss": 1.3281,
+      "step": 194700
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.2706053417004665e-05,
+      "loss": 1.151,
+      "step": 194710
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.2704905487104915e-05,
+      "loss": 1.1505,
+      "step": 194720
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.2703757557205165e-05,
+      "loss": 1.1378,
+      "step": 194730
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.270260962730542e-05,
+      "loss": 1.0437,
+      "step": 194740
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.270146169740568e-05,
+      "loss": 1.1347,
+      "step": 194750
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.270031376750592e-05,
+      "loss": 1.0743,
+      "step": 194760
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.269916583760618e-05,
+      "loss": 1.1139,
+      "step": 194770
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.2698017907706435e-05,
+      "loss": 1.1798,
+      "step": 194780
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.269686997780668e-05,
+      "loss": 1.1784,
+      "step": 194790
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.2695722047906934e-05,
+      "loss": 1.1776,
+      "step": 194800
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.269457411800719e-05,
+      "loss": 1.0736,
+      "step": 194810
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.269342618810744e-05,
+      "loss": 1.2328,
+      "step": 194820
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.269227825820769e-05,
+      "loss": 1.0427,
+      "step": 194830
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.269113032830795e-05,
+      "loss": 1.1456,
+      "step": 194840
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.26899823984082e-05,
+      "loss": 1.1756,
+      "step": 194850
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.2688834468508454e-05,
+      "loss": 1.2372,
+      "step": 194860
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.2687686538608704e-05,
+      "loss": 1.111,
+      "step": 194870
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.2686538608708954e-05,
+      "loss": 1.2198,
+      "step": 194880
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.268539067880921e-05,
+      "loss": 1.044,
+      "step": 194890
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.268424274890947e-05,
+      "loss": 1.185,
+      "step": 194900
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.268309481900971e-05,
+      "loss": 1.1459,
+      "step": 194910
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.2681946889109967e-05,
+      "loss": 1.1898,
+      "step": 194920
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.268079895921022e-05,
+      "loss": 1.0298,
+      "step": 194930
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.2679651029310466e-05,
+      "loss": 1.2081,
+      "step": 194940
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.267850309941072e-05,
+      "loss": 1.0363,
+      "step": 194950
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.267735516951098e-05,
+      "loss": 1.0963,
+      "step": 194960
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.267620723961123e-05,
+      "loss": 0.9722,
+      "step": 194970
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.267505930971148e-05,
+      "loss": 1.0717,
+      "step": 194980
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.2673911379811736e-05,
+      "loss": 1.0787,
+      "step": 194990
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.2672763449911986e-05,
+      "loss": 1.1592,
+      "step": 195000
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.267161552001224e-05,
+      "loss": 1.1376,
+      "step": 195010
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.267046759011249e-05,
+      "loss": 1.0816,
+      "step": 195020
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.266931966021274e-05,
+      "loss": 1.1223,
+      "step": 195030
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.2668171730313e-05,
+      "loss": 1.1576,
+      "step": 195040
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.2667023800413255e-05,
+      "loss": 0.9937,
+      "step": 195050
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.26658758705135e-05,
+      "loss": 1.1059,
+      "step": 195060
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.2664727940613755e-05,
+      "loss": 1.1331,
+      "step": 195070
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.266358001071401e-05,
+      "loss": 1.0757,
+      "step": 195080
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.2662432080814255e-05,
+      "loss": 1.1277,
+      "step": 195090
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.266128415091451e-05,
+      "loss": 1.2488,
+      "step": 195100
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.266013622101477e-05,
+      "loss": 1.2411,
+      "step": 195110
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.265898829111502e-05,
+      "loss": 1.0142,
+      "step": 195120
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.265784036121527e-05,
+      "loss": 1.0986,
+      "step": 195130
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.2656692431315524e-05,
+      "loss": 1.088,
+      "step": 195140
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.2655544501415774e-05,
+      "loss": 1.1452,
+      "step": 195150
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.265439657151603e-05,
+      "loss": 1.1097,
+      "step": 195160
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.265324864161628e-05,
+      "loss": 1.171,
+      "step": 195170
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.265210071171653e-05,
+      "loss": 1.1836,
+      "step": 195180
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.265095278181679e-05,
+      "loss": 1.0898,
+      "step": 195190
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.2649804851917044e-05,
+      "loss": 1.1638,
+      "step": 195200
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.264865692201729e-05,
+      "loss": 1.1545,
+      "step": 195210
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.2647508992117544e-05,
+      "loss": 1.1855,
+      "step": 195220
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.26463610622178e-05,
+      "loss": 1.1185,
+      "step": 195230
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.264521313231804e-05,
+      "loss": 1.2039,
+      "step": 195240
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.26440652024183e-05,
+      "loss": 1.1491,
+      "step": 195250
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.2642917272518557e-05,
+      "loss": 0.9952,
+      "step": 195260
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.2641769342618806e-05,
+      "loss": 1.1127,
+      "step": 195270
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.2640621412719056e-05,
+      "loss": 1.2862,
+      "step": 195280
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.263947348281931e-05,
+      "loss": 1.2085,
+      "step": 195290
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.263832555291956e-05,
+      "loss": 1.1734,
+      "step": 195300
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.263717762301982e-05,
+      "loss": 1.2242,
+      "step": 195310
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.263602969312007e-05,
+      "loss": 1.2148,
+      "step": 195320
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.263488176322032e-05,
+      "loss": 1.0163,
+      "step": 195330
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.2633733833320576e-05,
+      "loss": 1.0658,
+      "step": 195340
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.263258590342083e-05,
+      "loss": 1.0476,
+      "step": 195350
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.2631437973521075e-05,
+      "loss": 1.1762,
+      "step": 195360
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.263029004362133e-05,
+      "loss": 1.0512,
+      "step": 195370
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.262914211372159e-05,
+      "loss": 1.1371,
+      "step": 195380
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.262799418382183e-05,
+      "loss": 1.1204,
+      "step": 195390
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.262684625392209e-05,
+      "loss": 1.2931,
+      "step": 195400
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.2625698324022345e-05,
+      "loss": 1.1708,
+      "step": 195410
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.2624550394122595e-05,
+      "loss": 1.1123,
+      "step": 195420
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.2623402464222845e-05,
+      "loss": 1.1001,
+      "step": 195430
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.26222545343231e-05,
+      "loss": 1.1106,
+      "step": 195440
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.262110660442335e-05,
+      "loss": 1.2195,
+      "step": 195450
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.261995867452361e-05,
+      "loss": 1.2364,
+      "step": 195460
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.261881074462386e-05,
+      "loss": 1.1586,
+      "step": 195470
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.261766281472411e-05,
+      "loss": 1.2242,
+      "step": 195480
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.2616514884824364e-05,
+      "loss": 1.2916,
+      "step": 195490
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.261536695492462e-05,
+      "loss": 1.275,
+      "step": 195500
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.2614219025024864e-05,
+      "loss": 1.0693,
+      "step": 195510
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.261307109512512e-05,
+      "loss": 1.2413,
+      "step": 195520
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.261192316522538e-05,
+      "loss": 1.1407,
+      "step": 195530
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.261077523532562e-05,
+      "loss": 1.164,
+      "step": 195540
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.260962730542588e-05,
+      "loss": 1.2133,
+      "step": 195550
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.2608479375526133e-05,
+      "loss": 1.2829,
+      "step": 195560
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.260733144562638e-05,
+      "loss": 1.1155,
+      "step": 195570
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.260618351572663e-05,
+      "loss": 0.993,
+      "step": 195580
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.260503558582689e-05,
+      "loss": 1.1618,
+      "step": 195590
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.260388765592714e-05,
+      "loss": 1.1983,
+      "step": 195600
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.2602739726027396e-05,
+      "loss": 1.1209,
+      "step": 195610
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.2601591796127646e-05,
+      "loss": 1.1974,
+      "step": 195620
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.2600443866227896e-05,
+      "loss": 1.1875,
+      "step": 195630
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.259929593632815e-05,
+      "loss": 1.2561,
+      "step": 195640
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.259814800642841e-05,
+      "loss": 1.1564,
+      "step": 195650
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.259700007652865e-05,
+      "loss": 1.1068,
+      "step": 195660
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.259585214662891e-05,
+      "loss": 1.0962,
+      "step": 195670
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.2594704216729166e-05,
+      "loss": 0.9893,
+      "step": 195680
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.259355628682941e-05,
+      "loss": 1.153,
+      "step": 195690
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.2592408356929665e-05,
+      "loss": 1.1819,
+      "step": 195700
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.259126042702992e-05,
+      "loss": 1.1828,
+      "step": 195710
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.259011249713017e-05,
+      "loss": 1.0936,
+      "step": 195720
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.258896456723042e-05,
+      "loss": 1.1603,
+      "step": 195730
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.258781663733068e-05,
+      "loss": 0.9965,
+      "step": 195740
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.258666870743093e-05,
+      "loss": 1.1107,
+      "step": 195750
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.2585520777531185e-05,
+      "loss": 1.1826,
+      "step": 195760
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.2584372847631435e-05,
+      "loss": 1.1722,
+      "step": 195770
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.2583224917731684e-05,
+      "loss": 1.1897,
+      "step": 195780
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.258207698783194e-05,
+      "loss": 1.1548,
+      "step": 195790
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.25809290579322e-05,
+      "loss": 1.1551,
+      "step": 195800
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.257978112803244e-05,
+      "loss": 1.1524,
+      "step": 195810
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.25786331981327e-05,
+      "loss": 1.1319,
+      "step": 195820
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.2577485268232954e-05,
+      "loss": 1.1803,
+      "step": 195830
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.25763373383332e-05,
+      "loss": 1.1111,
+      "step": 195840
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.2575189408433454e-05,
+      "loss": 1.1217,
+      "step": 195850
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.257404147853371e-05,
+      "loss": 1.2046,
+      "step": 195860
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.257289354863396e-05,
+      "loss": 1.1923,
+      "step": 195870
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.257174561873421e-05,
+      "loss": 1.0953,
+      "step": 195880
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.257059768883447e-05,
+      "loss": 1.1964,
+      "step": 195890
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.2569449758934717e-05,
+      "loss": 1.2265,
+      "step": 195900
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.256830182903497e-05,
+      "loss": 1.1653,
+      "step": 195910
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.256715389913522e-05,
+      "loss": 1.1418,
+      "step": 195920
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.256600596923547e-05,
+      "loss": 1.0887,
+      "step": 195930
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.256485803933573e-05,
+      "loss": 1.2,
+      "step": 195940
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.2563710109435986e-05,
+      "loss": 1.2322,
+      "step": 195950
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.256256217953623e-05,
+      "loss": 1.0323,
+      "step": 195960
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.2561414249636486e-05,
+      "loss": 1.2248,
+      "step": 195970
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.256026631973674e-05,
+      "loss": 1.02,
+      "step": 195980
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.2559118389836986e-05,
+      "loss": 1.2182,
+      "step": 195990
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.255797045993724e-05,
+      "loss": 1.0527,
+      "step": 196000
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.25568225300375e-05,
+      "loss": 1.1332,
+      "step": 196010
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.255567460013775e-05,
+      "loss": 1.055,
+      "step": 196020
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.2554526670238e-05,
+      "loss": 1.106,
+      "step": 196030
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.2553378740338255e-05,
+      "loss": 1.3516,
+      "step": 196040
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.2552230810438505e-05,
+      "loss": 1.1862,
+      "step": 196050
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.255108288053876e-05,
+      "loss": 1.1756,
+      "step": 196060
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.254993495063901e-05,
+      "loss": 1.1923,
+      "step": 196070
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.254878702073926e-05,
+      "loss": 1.2106,
+      "step": 196080
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.254763909083952e-05,
+      "loss": 1.131,
+      "step": 196090
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.254649116093976e-05,
+      "loss": 1.1899,
+      "step": 196100
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.254534323104002e-05,
+      "loss": 1.0332,
+      "step": 196110
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.2544195301140274e-05,
+      "loss": 1.0972,
+      "step": 196120
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.2543047371240524e-05,
+      "loss": 1.2018,
+      "step": 196130
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.2541899441340774e-05,
+      "loss": 1.2294,
+      "step": 196140
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.254075151144103e-05,
+      "loss": 1.0953,
+      "step": 196150
+    },
+    {
+      "epoch": 9.0,
+      "eval_cer": 0.6690775439168468,
+      "eval_em": 0.007357212003872217,
+      "eval_f1": 0.007357212003872217,
+      "eval_loss": 1.1978390216827393,
+      "eval_runtime": 1711.2229,
+      "eval_samples_per_second": 6.037,
+      "eval_steps_per_second": 0.755,
+      "eval_wer": 0.9926427879961278,
+      "step": 196158
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.253960358154128e-05,
+      "loss": 1.1502,
+      "step": 196160
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.253845565164154e-05,
+      "loss": 1.1081,
+      "step": 196170
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.253730772174179e-05,
+      "loss": 1.1554,
+      "step": 196180
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.253615979184204e-05,
+      "loss": 1.0357,
+      "step": 196190
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.2535011861942294e-05,
+      "loss": 1.0739,
+      "step": 196200
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.253386393204255e-05,
+      "loss": 1.0441,
+      "step": 196210
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.253271600214279e-05,
+      "loss": 1.2067,
+      "step": 196220
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.253156807224305e-05,
+      "loss": 1.0286,
+      "step": 196230
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.2530420142343306e-05,
+      "loss": 1.0334,
+      "step": 196240
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.252927221244355e-05,
+      "loss": 1.1295,
+      "step": 196250
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.2528124282543806e-05,
+      "loss": 1.0763,
+      "step": 196260
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.252697635264406e-05,
+      "loss": 1.1171,
+      "step": 196270
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.252582842274431e-05,
+      "loss": 1.0579,
+      "step": 196280
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.252468049284456e-05,
+      "loss": 1.0619,
+      "step": 196290
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.252353256294482e-05,
+      "loss": 1.0755,
+      "step": 196300
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.252238463304507e-05,
+      "loss": 1.0291,
+      "step": 196310
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.2521236703145326e-05,
+      "loss": 1.082,
+      "step": 196320
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.2520088773245576e-05,
+      "loss": 1.1235,
+      "step": 196330
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.2518940843345825e-05,
+      "loss": 1.0232,
+      "step": 196340
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.251779291344608e-05,
+      "loss": 1.2298,
+      "step": 196350
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.251664498354634e-05,
+      "loss": 1.0455,
+      "step": 196360
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.251549705364658e-05,
+      "loss": 1.069,
+      "step": 196370
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.251434912374684e-05,
+      "loss": 1.1098,
+      "step": 196380
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.2513201193847095e-05,
+      "loss": 1.1715,
+      "step": 196390
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.251205326394734e-05,
+      "loss": 1.2227,
+      "step": 196400
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.2510905334047595e-05,
+      "loss": 1.0719,
+      "step": 196410
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.250975740414785e-05,
+      "loss": 1.0784,
+      "step": 196420
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.25086094742481e-05,
+      "loss": 1.1132,
+      "step": 196430
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.250746154434835e-05,
+      "loss": 1.0547,
+      "step": 196440
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.250631361444861e-05,
+      "loss": 1.1159,
+      "step": 196450
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.250516568454886e-05,
+      "loss": 1.0508,
+      "step": 196460
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.2504017754649114e-05,
+      "loss": 1.2599,
+      "step": 196470
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.2502869824749364e-05,
+      "loss": 1.0008,
+      "step": 196480
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.2501721894849614e-05,
+      "loss": 1.1973,
+      "step": 196490
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.250057396494987e-05,
+      "loss": 1.0268,
+      "step": 196500
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.249942603505013e-05,
+      "loss": 1.1795,
+      "step": 196510
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.249827810515037e-05,
+      "loss": 1.0152,
+      "step": 196520
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.249713017525063e-05,
+      "loss": 1.1552,
+      "step": 196530
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.2495982245350883e-05,
+      "loss": 1.0722,
+      "step": 196540
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.2494834315451127e-05,
+      "loss": 0.9909,
+      "step": 196550
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.249368638555138e-05,
+      "loss": 0.9972,
+      "step": 196560
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.249253845565164e-05,
+      "loss": 0.9872,
+      "step": 196570
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.249139052575189e-05,
+      "loss": 1.1143,
+      "step": 196580
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.249024259585214e-05,
+      "loss": 1.1002,
+      "step": 196590
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.2489094665952396e-05,
+      "loss": 1.1312,
+      "step": 196600
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.2487946736052646e-05,
+      "loss": 1.0658,
+      "step": 196610
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.24867988061529e-05,
+      "loss": 1.1016,
+      "step": 196620
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.248565087625315e-05,
+      "loss": 1.0736,
+      "step": 196630
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.24845029463534e-05,
+      "loss": 1.1415,
+      "step": 196640
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.248335501645366e-05,
+      "loss": 1.1422,
+      "step": 196650
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.2482207086553916e-05,
+      "loss": 1.0079,
+      "step": 196660
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.248105915665416e-05,
+      "loss": 1.0791,
+      "step": 196670
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.2479911226754415e-05,
+      "loss": 1.0117,
+      "step": 196680
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.247876329685467e-05,
+      "loss": 0.9716,
+      "step": 196690
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.2477615366954915e-05,
+      "loss": 1.073,
+      "step": 196700
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.247646743705517e-05,
+      "loss": 1.2545,
+      "step": 196710
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.247531950715543e-05,
+      "loss": 1.0644,
+      "step": 196720
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.247417157725568e-05,
+      "loss": 1.2138,
+      "step": 196730
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.247302364735593e-05,
+      "loss": 1.2087,
+      "step": 196740
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.2471875717456185e-05,
+      "loss": 1.0983,
+      "step": 196750
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.2470727787556434e-05,
+      "loss": 1.0505,
+      "step": 196760
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.246957985765669e-05,
+      "loss": 1.1055,
+      "step": 196770
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.246843192775694e-05,
+      "loss": 1.115,
+      "step": 196780
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.246728399785719e-05,
+      "loss": 1.063,
+      "step": 196790
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.246613606795745e-05,
+      "loss": 1.0968,
+      "step": 196800
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.2464988138057704e-05,
+      "loss": 1.0211,
+      "step": 196810
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.246384020815795e-05,
+      "loss": 0.9732,
+      "step": 196820
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.2462692278258204e-05,
+      "loss": 1.1271,
+      "step": 196830
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.246154434835846e-05,
+      "loss": 1.0978,
+      "step": 196840
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.2460396418458703e-05,
+      "loss": 1.0694,
+      "step": 196850
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.245924848855896e-05,
+      "loss": 1.0413,
+      "step": 196860
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.245810055865922e-05,
+      "loss": 1.0599,
+      "step": 196870
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.2456952628759467e-05,
+      "loss": 1.0432,
+      "step": 196880
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.2455804698859716e-05,
+      "loss": 1.0414,
+      "step": 196890
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.245465676895997e-05,
+      "loss": 1.0387,
+      "step": 196900
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.245350883906022e-05,
+      "loss": 1.1246,
+      "step": 196910
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.245236090916048e-05,
+      "loss": 1.0629,
+      "step": 196920
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.245121297926073e-05,
+      "loss": 0.9677,
+      "step": 196930
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.245006504936098e-05,
+      "loss": 1.1955,
+      "step": 196940
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.2448917119461236e-05,
+      "loss": 1.1281,
+      "step": 196950
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.244776918956149e-05,
+      "loss": 1.0276,
+      "step": 196960
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.2446621259661736e-05,
+      "loss": 1.0237,
+      "step": 196970
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.244547332976199e-05,
+      "loss": 1.0513,
+      "step": 196980
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.244432539986225e-05,
+      "loss": 1.0749,
+      "step": 196990
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.244317746996249e-05,
+      "loss": 1.1341,
+      "step": 197000
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.244202954006275e-05,
+      "loss": 1.1477,
+      "step": 197010
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.2440881610163005e-05,
+      "loss": 1.1767,
+      "step": 197020
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.2439733680263255e-05,
+      "loss": 1.1655,
+      "step": 197030
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.2438585750363505e-05,
+      "loss": 1.0594,
+      "step": 197040
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.243743782046376e-05,
+      "loss": 1.1051,
+      "step": 197050
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.243628989056401e-05,
+      "loss": 1.1117,
+      "step": 197060
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.243514196066427e-05,
+      "loss": 1.0499,
+      "step": 197070
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.243399403076452e-05,
+      "loss": 1.1173,
+      "step": 197080
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.243284610086477e-05,
+      "loss": 1.1178,
+      "step": 197090
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.2431698170965024e-05,
+      "loss": 1.1079,
+      "step": 197100
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.243055024106528e-05,
+      "loss": 1.0833,
+      "step": 197110
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.2429402311165524e-05,
+      "loss": 1.2315,
+      "step": 197120
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.242825438126578e-05,
+      "loss": 1.1308,
+      "step": 197130
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.242710645136604e-05,
+      "loss": 1.136,
+      "step": 197140
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.242595852146628e-05,
+      "loss": 1.0478,
+      "step": 197150
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.242481059156654e-05,
+      "loss": 1.1022,
+      "step": 197160
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.2423662661666794e-05,
+      "loss": 1.0514,
+      "step": 197170
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.2422514731767044e-05,
+      "loss": 1.2043,
+      "step": 197180
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.242136680186729e-05,
+      "loss": 1.0359,
+      "step": 197190
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.242021887196755e-05,
+      "loss": 0.9983,
+      "step": 197200
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.24190709420678e-05,
+      "loss": 1.1742,
+      "step": 197210
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.2417923012168056e-05,
+      "loss": 1.2195,
+      "step": 197220
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.2416775082268306e-05,
+      "loss": 1.068,
+      "step": 197230
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.2415627152368556e-05,
+      "loss": 1.0378,
+      "step": 197240
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.241447922246881e-05,
+      "loss": 1.1925,
+      "step": 197250
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.241333129256907e-05,
+      "loss": 1.0004,
+      "step": 197260
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.241218336266931e-05,
+      "loss": 0.9963,
+      "step": 197270
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.241103543276957e-05,
+      "loss": 1.085,
+      "step": 197280
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.2409887502869826e-05,
+      "loss": 1.1178,
+      "step": 197290
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.240873957297007e-05,
+      "loss": 1.0553,
+      "step": 197300
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.2407591643070326e-05,
+      "loss": 1.1204,
+      "step": 197310
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.240644371317058e-05,
+      "loss": 1.0677,
+      "step": 197320
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.240529578327083e-05,
+      "loss": 1.1695,
+      "step": 197330
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.240414785337108e-05,
+      "loss": 0.9762,
+      "step": 197340
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.240299992347134e-05,
+      "loss": 1.077,
+      "step": 197350
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.240185199357159e-05,
+      "loss": 1.3576,
+      "step": 197360
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.2400704063671845e-05,
+      "loss": 1.0656,
+      "step": 197370
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.2399556133772095e-05,
+      "loss": 1.0932,
+      "step": 197380
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.2398408203872345e-05,
+      "loss": 1.0085,
+      "step": 197390
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.23972602739726e-05,
+      "loss": 1.1422,
+      "step": 197400
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.239611234407286e-05,
+      "loss": 1.1521,
+      "step": 197410
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.23949644141731e-05,
+      "loss": 1.0506,
+      "step": 197420
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.239381648427336e-05,
+      "loss": 1.1442,
+      "step": 197430
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.2392668554373614e-05,
+      "loss": 1.0132,
+      "step": 197440
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.239152062447386e-05,
+      "loss": 1.0684,
+      "step": 197450
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.2390372694574114e-05,
+      "loss": 1.0499,
+      "step": 197460
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.2389224764674364e-05,
+      "loss": 1.0558,
+      "step": 197470
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.238807683477462e-05,
+      "loss": 1.0134,
+      "step": 197480
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.238692890487487e-05,
+      "loss": 1.0846,
+      "step": 197490
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.238578097497512e-05,
+      "loss": 1.0401,
+      "step": 197500
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.238463304507538e-05,
+      "loss": 1.12,
+      "step": 197510
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.2383485115175633e-05,
+      "loss": 0.9414,
+      "step": 197520
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.2382337185275877e-05,
+      "loss": 1.0177,
+      "step": 197530
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.238118925537613e-05,
+      "loss": 1.1648,
+      "step": 197540
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.238004132547639e-05,
+      "loss": 1.1239,
+      "step": 197550
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.237889339557663e-05,
+      "loss": 1.1447,
+      "step": 197560
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.237774546567689e-05,
+      "loss": 1.0255,
+      "step": 197570
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.2376597535777146e-05,
+      "loss": 1.1371,
+      "step": 197580
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.2375449605877396e-05,
+      "loss": 1.0982,
+      "step": 197590
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.2374301675977646e-05,
+      "loss": 1.1388,
+      "step": 197600
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.23731537460779e-05,
+      "loss": 1.061,
+      "step": 197610
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.237200581617815e-05,
+      "loss": 1.1736,
+      "step": 197620
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.237085788627841e-05,
+      "loss": 1.1408,
+      "step": 197630
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.236970995637866e-05,
+      "loss": 1.0099,
+      "step": 197640
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.236856202647891e-05,
+      "loss": 1.1666,
+      "step": 197650
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.2367414096579165e-05,
+      "loss": 1.0993,
+      "step": 197660
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.236626616667942e-05,
+      "loss": 1.1828,
+      "step": 197670
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.2365118236779665e-05,
+      "loss": 1.2177,
+      "step": 197680
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.236397030687992e-05,
+      "loss": 1.2415,
+      "step": 197690
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.236282237698018e-05,
+      "loss": 1.0468,
+      "step": 197700
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.236167444708042e-05,
+      "loss": 1.2425,
+      "step": 197710
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.236052651718068e-05,
+      "loss": 1.0326,
+      "step": 197720
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.2359378587280935e-05,
+      "loss": 1.0889,
+      "step": 197730
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.2358230657381184e-05,
+      "loss": 1.0148,
+      "step": 197740
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.2357082727481434e-05,
+      "loss": 1.0343,
+      "step": 197750
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.235593479758169e-05,
+      "loss": 1.1284,
+      "step": 197760
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.235478686768194e-05,
+      "loss": 1.1299,
+      "step": 197770
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.23536389377822e-05,
+      "loss": 1.1828,
+      "step": 197780
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.235249100788245e-05,
+      "loss": 0.988,
+      "step": 197790
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.23513430779827e-05,
+      "loss": 1.1176,
+      "step": 197800
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.2350195148082954e-05,
+      "loss": 1.0015,
+      "step": 197810
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.234904721818321e-05,
+      "loss": 1.0155,
+      "step": 197820
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.2347899288283453e-05,
+      "loss": 1.1713,
+      "step": 197830
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.234675135838371e-05,
+      "loss": 1.1229,
+      "step": 197840
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.234560342848397e-05,
+      "loss": 1.1613,
+      "step": 197850
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.234445549858421e-05,
+      "loss": 1.0465,
+      "step": 197860
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.2343307568684466e-05,
+      "loss": 1.0708,
+      "step": 197870
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.234215963878472e-05,
+      "loss": 1.1403,
+      "step": 197880
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.234101170888497e-05,
+      "loss": 1.0814,
+      "step": 197890
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.233986377898522e-05,
+      "loss": 1.0153,
+      "step": 197900
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.233871584908548e-05,
+      "loss": 1.084,
+      "step": 197910
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.233756791918573e-05,
+      "loss": 0.9983,
+      "step": 197920
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.2336419989285986e-05,
+      "loss": 1.0148,
+      "step": 197930
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.2335272059386236e-05,
+      "loss": 1.2058,
+      "step": 197940
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.2334124129486486e-05,
+      "loss": 1.0909,
+      "step": 197950
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.233297619958674e-05,
+      "loss": 1.1677,
+      "step": 197960
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.2331828269687e-05,
+      "loss": 1.1516,
+      "step": 197970
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.233068033978724e-05,
+      "loss": 1.2656,
+      "step": 197980
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.23295324098875e-05,
+      "loss": 1.148,
+      "step": 197990
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.2328384479987755e-05,
+      "loss": 1.0787,
+      "step": 198000
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.2327236550088e-05,
+      "loss": 1.0758,
+      "step": 198010
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.2326088620188255e-05,
+      "loss": 1.1367,
+      "step": 198020
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.232494069028851e-05,
+      "loss": 1.076,
+      "step": 198030
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.232379276038876e-05,
+      "loss": 1.0778,
+      "step": 198040
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.232264483048901e-05,
+      "loss": 1.0937,
+      "step": 198050
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.232149690058927e-05,
+      "loss": 1.0121,
+      "step": 198060
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.232034897068952e-05,
+      "loss": 0.9863,
+      "step": 198070
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.2319201040789774e-05,
+      "loss": 1.0863,
+      "step": 198080
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.2318053110890024e-05,
+      "loss": 1.125,
+      "step": 198090
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.2316905180990274e-05,
+      "loss": 1.0784,
+      "step": 198100
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.231575725109053e-05,
+      "loss": 1.179,
+      "step": 198110
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.231460932119079e-05,
+      "loss": 1.0536,
+      "step": 198120
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.231346139129103e-05,
+      "loss": 1.0116,
+      "step": 198130
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.231231346139129e-05,
+      "loss": 1.0945,
+      "step": 198140
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.2311165531491544e-05,
+      "loss": 1.1041,
+      "step": 198150
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.231001760159179e-05,
+      "loss": 1.0686,
+      "step": 198160
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.230886967169204e-05,
+      "loss": 1.067,
+      "step": 198170
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.23077217417923e-05,
+      "loss": 1.0632,
+      "step": 198180
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.230657381189255e-05,
+      "loss": 1.1818,
+      "step": 198190
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.23054258819928e-05,
+      "loss": 1.1883,
+      "step": 198200
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.2304277952093056e-05,
+      "loss": 0.9657,
+      "step": 198210
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.2303130022193306e-05,
+      "loss": 1.0854,
+      "step": 198220
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.230198209229356e-05,
+      "loss": 1.1154,
+      "step": 198230
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.230083416239381e-05,
+      "loss": 1.0318,
+      "step": 198240
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.229968623249406e-05,
+      "loss": 1.1098,
+      "step": 198250
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.229853830259432e-05,
+      "loss": 1.1493,
+      "step": 198260
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.2297390372694576e-05,
+      "loss": 1.1634,
+      "step": 198270
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.229624244279482e-05,
+      "loss": 1.11,
+      "step": 198280
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.2295094512895076e-05,
+      "loss": 1.1616,
+      "step": 198290
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.229394658299533e-05,
+      "loss": 1.1124,
+      "step": 198300
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.2292798653095575e-05,
+      "loss": 1.1409,
+      "step": 198310
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.229165072319583e-05,
+      "loss": 1.0569,
+      "step": 198320
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.229050279329609e-05,
+      "loss": 1.033,
+      "step": 198330
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.228935486339634e-05,
+      "loss": 1.1458,
+      "step": 198340
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.228820693349659e-05,
+      "loss": 1.1216,
+      "step": 198350
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.2287059003596845e-05,
+      "loss": 1.0827,
+      "step": 198360
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.2285911073697095e-05,
+      "loss": 1.0881,
+      "step": 198370
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.228476314379735e-05,
+      "loss": 1.1258,
+      "step": 198380
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.22836152138976e-05,
+      "loss": 1.0234,
+      "step": 198390
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.228246728399785e-05,
+      "loss": 1.0893,
+      "step": 198400
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.228131935409811e-05,
+      "loss": 1.0975,
+      "step": 198410
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.2280171424198364e-05,
+      "loss": 1.178,
+      "step": 198420
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.227902349429861e-05,
+      "loss": 0.9992,
+      "step": 198430
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.2277875564398864e-05,
+      "loss": 1.1738,
+      "step": 198440
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.227672763449912e-05,
+      "loss": 1.0303,
+      "step": 198450
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.2275579704599364e-05,
+      "loss": 1.0739,
+      "step": 198460
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.227443177469962e-05,
+      "loss": 1.0556,
+      "step": 198470
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.227328384479988e-05,
+      "loss": 1.0513,
+      "step": 198480
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.227213591490013e-05,
+      "loss": 1.0008,
+      "step": 198490
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.227098798500038e-05,
+      "loss": 1.1524,
+      "step": 198500
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.226984005510063e-05,
+      "loss": 1.0613,
+      "step": 198510
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.226869212520088e-05,
+      "loss": 1.0977,
+      "step": 198520
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.226754419530114e-05,
+      "loss": 1.1065,
+      "step": 198530
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.226639626540139e-05,
+      "loss": 1.0761,
+      "step": 198540
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.226524833550164e-05,
+      "loss": 1.1132,
+      "step": 198550
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.2264100405601896e-05,
+      "loss": 1.044,
+      "step": 198560
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.226295247570215e-05,
+      "loss": 1.2427,
+      "step": 198570
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.2261804545802396e-05,
+      "loss": 1.3023,
+      "step": 198580
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.226065661590265e-05,
+      "loss": 1.0911,
+      "step": 198590
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.225950868600291e-05,
+      "loss": 1.0903,
+      "step": 198600
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.225836075610315e-05,
+      "loss": 1.1479,
+      "step": 198610
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.225721282620341e-05,
+      "loss": 1.1389,
+      "step": 198620
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.2256064896303665e-05,
+      "loss": 1.1419,
+      "step": 198630
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.2254916966403915e-05,
+      "loss": 1.0337,
+      "step": 198640
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.2253769036504165e-05,
+      "loss": 1.0,
+      "step": 198650
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.225262110660442e-05,
+      "loss": 1.0542,
+      "step": 198660
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.225147317670467e-05,
+      "loss": 1.1169,
+      "step": 198670
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.225032524680493e-05,
+      "loss": 1.1172,
+      "step": 198680
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.224917731690518e-05,
+      "loss": 1.0564,
+      "step": 198690
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.224802938700543e-05,
+      "loss": 1.1255,
+      "step": 198700
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.2246881457105685e-05,
+      "loss": 1.0494,
+      "step": 198710
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.224573352720594e-05,
+      "loss": 1.0306,
+      "step": 198720
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.2244585597306184e-05,
+      "loss": 1.1212,
+      "step": 198730
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.224343766740644e-05,
+      "loss": 1.2712,
+      "step": 198740
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.22422897375067e-05,
+      "loss": 0.9982,
+      "step": 198750
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.224114180760694e-05,
+      "loss": 1.1045,
+      "step": 198760
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.22399938777072e-05,
+      "loss": 1.1535,
+      "step": 198770
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.2238845947807454e-05,
+      "loss": 1.187,
+      "step": 198780
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.2237698017907704e-05,
+      "loss": 1.1372,
+      "step": 198790
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.2236550088007954e-05,
+      "loss": 1.2148,
+      "step": 198800
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.223540215810821e-05,
+      "loss": 1.0311,
+      "step": 198810
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.223425422820846e-05,
+      "loss": 1.1422,
+      "step": 198820
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.223310629830872e-05,
+      "loss": 1.036,
+      "step": 198830
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.223195836840896e-05,
+      "loss": 1.1042,
+      "step": 198840
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.2230810438509216e-05,
+      "loss": 1.0226,
+      "step": 198850
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.222966250860947e-05,
+      "loss": 1.0,
+      "step": 198860
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.2228514578709716e-05,
+      "loss": 1.1063,
+      "step": 198870
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.222736664880997e-05,
+      "loss": 1.1155,
+      "step": 198880
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.222621871891023e-05,
+      "loss": 1.0783,
+      "step": 198890
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.222507078901047e-05,
+      "loss": 1.0757,
+      "step": 198900
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.222392285911073e-05,
+      "loss": 1.1615,
+      "step": 198910
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.2222774929210986e-05,
+      "loss": 1.1469,
+      "step": 198920
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.2221626999311236e-05,
+      "loss": 1.2127,
+      "step": 198930
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.222047906941149e-05,
+      "loss": 1.0364,
+      "step": 198940
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.221933113951174e-05,
+      "loss": 1.1163,
+      "step": 198950
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.221818320961199e-05,
+      "loss": 1.1233,
+      "step": 198960
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.221703527971225e-05,
+      "loss": 1.1288,
+      "step": 198970
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.2215887349812505e-05,
+      "loss": 1.0879,
+      "step": 198980
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.221473941991275e-05,
+      "loss": 1.0134,
+      "step": 198990
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.2213591490013005e-05,
+      "loss": 0.8891,
+      "step": 199000
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.221244356011326e-05,
+      "loss": 1.1455,
+      "step": 199010
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.2211295630213505e-05,
+      "loss": 1.1868,
+      "step": 199020
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.221014770031376e-05,
+      "loss": 1.1374,
+      "step": 199030
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.220899977041402e-05,
+      "loss": 1.0996,
+      "step": 199040
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.220785184051426e-05,
+      "loss": 1.0224,
+      "step": 199050
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.220670391061452e-05,
+      "loss": 1.1139,
+      "step": 199060
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.2205555980714774e-05,
+      "loss": 1.139,
+      "step": 199070
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.2204408050815024e-05,
+      "loss": 1.1248,
+      "step": 199080
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.220326012091528e-05,
+      "loss": 1.1306,
+      "step": 199090
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.220211219101553e-05,
+      "loss": 1.105,
+      "step": 199100
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.220096426111578e-05,
+      "loss": 1.0982,
+      "step": 199110
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.219981633121604e-05,
+      "loss": 1.119,
+      "step": 199120
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.2198668401316294e-05,
+      "loss": 0.9453,
+      "step": 199130
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.219752047141654e-05,
+      "loss": 1.024,
+      "step": 199140
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.219637254151679e-05,
+      "loss": 1.1399,
+      "step": 199150
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.219522461161705e-05,
+      "loss": 1.1419,
+      "step": 199160
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.219407668171729e-05,
+      "loss": 1.1252,
+      "step": 199170
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.219292875181755e-05,
+      "loss": 1.0917,
+      "step": 199180
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.2191780821917806e-05,
+      "loss": 1.0238,
+      "step": 199190
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.219063289201805e-05,
+      "loss": 1.0198,
+      "step": 199200
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.2189484962118306e-05,
+      "loss": 1.069,
+      "step": 199210
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.218833703221856e-05,
+      "loss": 1.0196,
+      "step": 199220
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.218718910231881e-05,
+      "loss": 1.1066,
+      "step": 199230
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.218604117241907e-05,
+      "loss": 1.1151,
+      "step": 199240
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.218489324251932e-05,
+      "loss": 1.0384,
+      "step": 199250
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.218374531261957e-05,
+      "loss": 0.9814,
+      "step": 199260
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.2182597382719826e-05,
+      "loss": 1.0745,
+      "step": 199270
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.218144945282008e-05,
+      "loss": 1.1116,
+      "step": 199280
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.2180301522920325e-05,
+      "loss": 1.1811,
+      "step": 199290
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.217915359302058e-05,
+      "loss": 0.9809,
+      "step": 199300
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.217800566312084e-05,
+      "loss": 1.2537,
+      "step": 199310
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.217685773322108e-05,
+      "loss": 1.0893,
+      "step": 199320
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.217570980332134e-05,
+      "loss": 1.016,
+      "step": 199330
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.2174561873421595e-05,
+      "loss": 1.0453,
+      "step": 199340
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.217341394352184e-05,
+      "loss": 1.1529,
+      "step": 199350
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.2172266013622095e-05,
+      "loss": 1.0251,
+      "step": 199360
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.217111808372235e-05,
+      "loss": 1.0763,
+      "step": 199370
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.21699701538226e-05,
+      "loss": 1.1644,
+      "step": 199380
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.216882222392286e-05,
+      "loss": 1.0654,
+      "step": 199390
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.216767429402311e-05,
+      "loss": 1.1578,
+      "step": 199400
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.216652636412336e-05,
+      "loss": 1.0975,
+      "step": 199410
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.2165378434223614e-05,
+      "loss": 1.138,
+      "step": 199420
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.216423050432387e-05,
+      "loss": 1.1451,
+      "step": 199430
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.2163082574424114e-05,
+      "loss": 1.083,
+      "step": 199440
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.216193464452437e-05,
+      "loss": 1.0641,
+      "step": 199450
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.216078671462463e-05,
+      "loss": 1.0424,
+      "step": 199460
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.215963878472487e-05,
+      "loss": 1.054,
+      "step": 199470
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.215849085482513e-05,
+      "loss": 1.0746,
+      "step": 199480
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.215734292492538e-05,
+      "loss": 1.1427,
+      "step": 199490
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.2156194995025626e-05,
+      "loss": 1.0847,
+      "step": 199500
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.215504706512588e-05,
+      "loss": 1.1414,
+      "step": 199510
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.215389913522614e-05,
+      "loss": 1.1164,
+      "step": 199520
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.215275120532639e-05,
+      "loss": 1.054,
+      "step": 199530
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.2151603275426646e-05,
+      "loss": 1.173,
+      "step": 199540
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.2150455345526896e-05,
+      "loss": 0.975,
+      "step": 199550
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.2149307415627146e-05,
+      "loss": 1.0027,
+      "step": 199560
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.21481594857274e-05,
+      "loss": 1.0896,
+      "step": 199570
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.214701155582766e-05,
+      "loss": 1.1607,
+      "step": 199580
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.21458636259279e-05,
+      "loss": 1.0616,
+      "step": 199590
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.214471569602816e-05,
+      "loss": 1.0296,
+      "step": 199600
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.2143567766128415e-05,
+      "loss": 1.09,
+      "step": 199610
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.214241983622866e-05,
+      "loss": 1.1802,
+      "step": 199620
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.2141271906328915e-05,
+      "loss": 1.0999,
+      "step": 199630
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.214012397642917e-05,
+      "loss": 0.9298,
+      "step": 199640
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.2138976046529415e-05,
+      "loss": 1.0085,
+      "step": 199650
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.213782811662967e-05,
+      "loss": 1.0776,
+      "step": 199660
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.213668018672993e-05,
+      "loss": 1.0095,
+      "step": 199670
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.213553225683018e-05,
+      "loss": 1.1694,
+      "step": 199680
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.2134384326930435e-05,
+      "loss": 1.0882,
+      "step": 199690
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.2133236397030684e-05,
+      "loss": 1.0918,
+      "step": 199700
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.2132088467130934e-05,
+      "loss": 1.0613,
+      "step": 199710
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.213094053723119e-05,
+      "loss": 1.0557,
+      "step": 199720
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.212979260733145e-05,
+      "loss": 1.1639,
+      "step": 199730
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.212864467743169e-05,
+      "loss": 1.1643,
+      "step": 199740
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.212749674753195e-05,
+      "loss": 0.9999,
+      "step": 199750
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.2126348817632204e-05,
+      "loss": 1.1195,
+      "step": 199760
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.212520088773245e-05,
+      "loss": 1.0592,
+      "step": 199770
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.2124052957832704e-05,
+      "loss": 1.0763,
+      "step": 199780
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.212290502793296e-05,
+      "loss": 1.1891,
+      "step": 199790
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.21217570980332e-05,
+      "loss": 1.0067,
+      "step": 199800
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.212060916813346e-05,
+      "loss": 1.087,
+      "step": 199810
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.2119461238233717e-05,
+      "loss": 1.0644,
+      "step": 199820
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.2118313308333966e-05,
+      "loss": 1.1308,
+      "step": 199830
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.211716537843422e-05,
+      "loss": 1.0366,
+      "step": 199840
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.211601744853447e-05,
+      "loss": 1.0835,
+      "step": 199850
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.211486951863472e-05,
+      "loss": 1.1426,
+      "step": 199860
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.211372158873498e-05,
+      "loss": 1.0425,
+      "step": 199870
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.2112573658835236e-05,
+      "loss": 1.0072,
+      "step": 199880
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.211142572893548e-05,
+      "loss": 1.0342,
+      "step": 199890
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.2110277799035736e-05,
+      "loss": 1.0689,
+      "step": 199900
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.210912986913599e-05,
+      "loss": 1.1206,
+      "step": 199910
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.2107981939236235e-05,
+      "loss": 1.1343,
+      "step": 199920
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.210683400933649e-05,
+      "loss": 1.1494,
+      "step": 199930
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.210568607943675e-05,
+      "loss": 0.9853,
+      "step": 199940
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.210453814953699e-05,
+      "loss": 1.1955,
+      "step": 199950
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.210339021963725e-05,
+      "loss": 1.0362,
+      "step": 199960
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.2102242289737505e-05,
+      "loss": 1.0318,
+      "step": 199970
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.2101094359837755e-05,
+      "loss": 0.9717,
+      "step": 199980
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.209994642993801e-05,
+      "loss": 1.2271,
+      "step": 199990
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.209879850003826e-05,
+      "loss": 1.1562,
+      "step": 200000
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.209765057013851e-05,
+      "loss": 1.1929,
+      "step": 200010
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.209650264023877e-05,
+      "loss": 1.018,
+      "step": 200020
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.2095354710339024e-05,
+      "loss": 1.0792,
+      "step": 200030
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.209420678043927e-05,
+      "loss": 1.123,
+      "step": 200040
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.2093058850539524e-05,
+      "loss": 1.0586,
+      "step": 200050
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.209191092063978e-05,
+      "loss": 1.2179,
+      "step": 200060
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.2090762990740024e-05,
+      "loss": 1.0623,
+      "step": 200070
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.208961506084028e-05,
+      "loss": 0.9684,
+      "step": 200080
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.208846713094054e-05,
+      "loss": 1.1145,
+      "step": 200090
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.208731920104078e-05,
+      "loss": 1.0728,
+      "step": 200100
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.208617127114104e-05,
+      "loss": 1.1602,
+      "step": 200110
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.2085023341241294e-05,
+      "loss": 1.0721,
+      "step": 200120
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.208387541134154e-05,
+      "loss": 1.2139,
+      "step": 200130
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.20827274814418e-05,
+      "loss": 1.0823,
+      "step": 200140
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.208157955154205e-05,
+      "loss": 1.0232,
+      "step": 200150
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.20804316216423e-05,
+      "loss": 1.0399,
+      "step": 200160
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.2079283691742556e-05,
+      "loss": 1.1192,
+      "step": 200170
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.207813576184281e-05,
+      "loss": 1.0143,
+      "step": 200180
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.2076987831943056e-05,
+      "loss": 1.046,
+      "step": 200190
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.207583990204331e-05,
+      "loss": 1.0553,
+      "step": 200200
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.207469197214357e-05,
+      "loss": 1.0364,
+      "step": 200210
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.207354404224381e-05,
+      "loss": 1.1162,
+      "step": 200220
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.207239611234407e-05,
+      "loss": 1.1679,
+      "step": 200230
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.207124818244432e-05,
+      "loss": 1.1818,
+      "step": 200240
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.207010025254457e-05,
+      "loss": 1.0879,
+      "step": 200250
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.2068952322644825e-05,
+      "loss": 1.0039,
+      "step": 200260
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.2067804392745075e-05,
+      "loss": 1.0024,
+      "step": 200270
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.206665646284533e-05,
+      "loss": 1.0736,
+      "step": 200280
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.206550853294559e-05,
+      "loss": 1.1516,
+      "step": 200290
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.206436060304583e-05,
+      "loss": 1.0844,
+      "step": 200300
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.206321267314609e-05,
+      "loss": 1.0282,
+      "step": 200310
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.2062064743246345e-05,
+      "loss": 1.0845,
+      "step": 200320
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.206091681334659e-05,
+      "loss": 1.1604,
+      "step": 200330
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.2059768883446845e-05,
+      "loss": 1.2126,
+      "step": 200340
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.20586209535471e-05,
+      "loss": 1.1807,
+      "step": 200350
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.2057473023647344e-05,
+      "loss": 1.0498,
+      "step": 200360
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.20563250937476e-05,
+      "loss": 1.0633,
+      "step": 200370
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.205517716384786e-05,
+      "loss": 1.0792,
+      "step": 200380
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.205402923394811e-05,
+      "loss": 1.0827,
+      "step": 200390
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.205288130404836e-05,
+      "loss": 1.1038,
+      "step": 200400
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.2051733374148614e-05,
+      "loss": 1.0375,
+      "step": 200410
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.2050585444248864e-05,
+      "loss": 1.046,
+      "step": 200420
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.204943751434912e-05,
+      "loss": 1.1178,
+      "step": 200430
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.204828958444938e-05,
+      "loss": 1.1125,
+      "step": 200440
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.204714165454962e-05,
+      "loss": 1.1166,
+      "step": 200450
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.204599372464988e-05,
+      "loss": 1.1849,
+      "step": 200460
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.204484579475013e-05,
+      "loss": 1.1024,
+      "step": 200470
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.2043697864850376e-05,
+      "loss": 0.9635,
+      "step": 200480
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.204254993495063e-05,
+      "loss": 1.0601,
+      "step": 200490
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.204140200505089e-05,
+      "loss": 1.2203,
+      "step": 200500
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.204025407515113e-05,
+      "loss": 0.9964,
+      "step": 200510
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.203910614525139e-05,
+      "loss": 1.1264,
+      "step": 200520
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.2037958215351646e-05,
+      "loss": 1.0482,
+      "step": 200530
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.2036810285451896e-05,
+      "loss": 1.0065,
+      "step": 200540
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.2035662355552146e-05,
+      "loss": 1.0401,
+      "step": 200550
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.20345144256524e-05,
+      "loss": 1.2936,
+      "step": 200560
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.203336649575265e-05,
+      "loss": 1.0503,
+      "step": 200570
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.203221856585291e-05,
+      "loss": 1.0552,
+      "step": 200580
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.2031070635953165e-05,
+      "loss": 1.1429,
+      "step": 200590
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.202992270605341e-05,
+      "loss": 1.0357,
+      "step": 200600
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.2028774776153665e-05,
+      "loss": 1.0742,
+      "step": 200610
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.202762684625392e-05,
+      "loss": 1.0479,
+      "step": 200620
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.2026478916354165e-05,
+      "loss": 1.1281,
+      "step": 200630
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.202533098645442e-05,
+      "loss": 1.0795,
+      "step": 200640
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.202418305655468e-05,
+      "loss": 1.1546,
+      "step": 200650
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.202303512665492e-05,
+      "loss": 1.071,
+      "step": 200660
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.202188719675518e-05,
+      "loss": 1.0627,
+      "step": 200670
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.2020739266855434e-05,
+      "loss": 1.0871,
+      "step": 200680
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.2019591336955684e-05,
+      "loss": 1.0393,
+      "step": 200690
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.2018443407055934e-05,
+      "loss": 1.0083,
+      "step": 200700
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.201729547715619e-05,
+      "loss": 0.9263,
+      "step": 200710
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.201614754725644e-05,
+      "loss": 1.0826,
+      "step": 200720
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.20149996173567e-05,
+      "loss": 0.9742,
+      "step": 200730
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.2013851687456954e-05,
+      "loss": 1.1075,
+      "step": 200740
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.20127037575572e-05,
+      "loss": 1.0407,
+      "step": 200750
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.2011555827657454e-05,
+      "loss": 1.1554,
+      "step": 200760
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.201040789775771e-05,
+      "loss": 1.1978,
+      "step": 200770
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.200925996785795e-05,
+      "loss": 1.0875,
+      "step": 200780
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.200811203795821e-05,
+      "loss": 1.171,
+      "step": 200790
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.2006964108058467e-05,
+      "loss": 1.1527,
+      "step": 200800
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.200581617815871e-05,
+      "loss": 1.1703,
+      "step": 200810
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.2004668248258966e-05,
+      "loss": 1.2086,
+      "step": 200820
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.200352031835922e-05,
+      "loss": 1.119,
+      "step": 200830
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.200237238845947e-05,
+      "loss": 1.1186,
+      "step": 200840
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.200122445855972e-05,
+      "loss": 1.0211,
+      "step": 200850
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.200007652865998e-05,
+      "loss": 1.0375,
+      "step": 200860
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.199892859876023e-05,
+      "loss": 1.1875,
+      "step": 200870
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.1997780668860486e-05,
+      "loss": 1.0631,
+      "step": 200880
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.199663273896074e-05,
+      "loss": 1.0101,
+      "step": 200890
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.1995484809060985e-05,
+      "loss": 1.0909,
+      "step": 200900
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.199433687916124e-05,
+      "loss": 1.1032,
+      "step": 200910
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.19931889492615e-05,
+      "loss": 1.0619,
+      "step": 200920
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.199204101936174e-05,
+      "loss": 1.1138,
+      "step": 200930
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.1990893089462e-05,
+      "loss": 1.0767,
+      "step": 200940
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.1989745159562255e-05,
+      "loss": 1.1484,
+      "step": 200950
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.19885972296625e-05,
+      "loss": 1.0138,
+      "step": 200960
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.1987449299762755e-05,
+      "loss": 1.0108,
+      "step": 200970
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.198630136986301e-05,
+      "loss": 1.0345,
+      "step": 200980
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.198515343996326e-05,
+      "loss": 1.1793,
+      "step": 200990
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.198400551006351e-05,
+      "loss": 1.072,
+      "step": 201000
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.198285758016377e-05,
+      "loss": 1.2,
+      "step": 201010
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.198170965026402e-05,
+      "loss": 1.1021,
+      "step": 201020
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.1980561720364274e-05,
+      "loss": 1.0547,
+      "step": 201030
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.197941379046453e-05,
+      "loss": 0.9999,
+      "step": 201040
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.1978265860564774e-05,
+      "loss": 0.9685,
+      "step": 201050
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.197711793066503e-05,
+      "loss": 1.076,
+      "step": 201060
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.197597000076529e-05,
+      "loss": 1.1674,
+      "step": 201070
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.197482207086553e-05,
+      "loss": 1.0561,
+      "step": 201080
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.197367414096579e-05,
+      "loss": 1.072,
+      "step": 201090
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.1972526211066044e-05,
+      "loss": 1.0607,
+      "step": 201100
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.1971378281166287e-05,
+      "loss": 1.1733,
+      "step": 201110
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.197023035126654e-05,
+      "loss": 1.1756,
+      "step": 201120
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.19690824213668e-05,
+      "loss": 0.9673,
+      "step": 201130
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.196793449146705e-05,
+      "loss": 1.0532,
+      "step": 201140
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.19667865615673e-05,
+      "loss": 1.0841,
+      "step": 201150
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.1965638631667556e-05,
+      "loss": 1.0814,
+      "step": 201160
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.1964490701767806e-05,
+      "loss": 1.0785,
+      "step": 201170
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.196334277186806e-05,
+      "loss": 1.024,
+      "step": 201180
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.196219484196832e-05,
+      "loss": 1.1388,
+      "step": 201190
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.196104691206856e-05,
+      "loss": 0.9638,
+      "step": 201200
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.195989898216882e-05,
+      "loss": 1.0587,
+      "step": 201210
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.1958751052269076e-05,
+      "loss": 1.0703,
+      "step": 201220
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.195760312236932e-05,
+      "loss": 0.9673,
+      "step": 201230
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.1956455192469575e-05,
+      "loss": 1.067,
+      "step": 201240
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.195530726256983e-05,
+      "loss": 1.0697,
+      "step": 201250
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.1954159332670075e-05,
+      "loss": 1.1091,
+      "step": 201260
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.195301140277033e-05,
+      "loss": 1.066,
+      "step": 201270
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.195186347287059e-05,
+      "loss": 1.0902,
+      "step": 201280
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.195071554297084e-05,
+      "loss": 1.1702,
+      "step": 201290
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.194956761307109e-05,
+      "loss": 1.0954,
+      "step": 201300
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.1948419683171345e-05,
+      "loss": 1.088,
+      "step": 201310
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.1947271753271595e-05,
+      "loss": 1.2294,
+      "step": 201320
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.194612382337185e-05,
+      "loss": 1.0437,
+      "step": 201330
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.194497589347211e-05,
+      "loss": 1.137,
+      "step": 201340
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.194382796357235e-05,
+      "loss": 1.0312,
+      "step": 201350
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.194268003367261e-05,
+      "loss": 1.0623,
+      "step": 201360
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.1941532103772864e-05,
+      "loss": 1.0953,
+      "step": 201370
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.194038417387311e-05,
+      "loss": 1.1395,
+      "step": 201380
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.1939236243973364e-05,
+      "loss": 1.0557,
+      "step": 201390
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.193808831407362e-05,
+      "loss": 0.9973,
+      "step": 201400
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.1936940384173864e-05,
+      "loss": 1.1093,
+      "step": 201410
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.193579245427412e-05,
+      "loss": 1.0452,
+      "step": 201420
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.193464452437438e-05,
+      "loss": 1.0715,
+      "step": 201430
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.193349659447463e-05,
+      "loss": 1.1138,
+      "step": 201440
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.1932348664574876e-05,
+      "loss": 1.0169,
+      "step": 201450
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.193120073467513e-05,
+      "loss": 1.176,
+      "step": 201460
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.193005280477538e-05,
+      "loss": 1.0164,
+      "step": 201470
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.192890487487564e-05,
+      "loss": 1.1407,
+      "step": 201480
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.1927756944975896e-05,
+      "loss": 1.2282,
+      "step": 201490
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.192660901507614e-05,
+      "loss": 1.1238,
+      "step": 201500
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.1925461085176396e-05,
+      "loss": 1.0632,
+      "step": 201510
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.192431315527665e-05,
+      "loss": 1.1278,
+      "step": 201520
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.1923165225376896e-05,
+      "loss": 1.1238,
+      "step": 201530
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.192201729547715e-05,
+      "loss": 1.1257,
+      "step": 201540
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.192086936557741e-05,
+      "loss": 1.1483,
+      "step": 201550
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.191972143567765e-05,
+      "loss": 1.0745,
+      "step": 201560
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.191857350577791e-05,
+      "loss": 1.0908,
+      "step": 201570
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.1917425575878165e-05,
+      "loss": 1.1839,
+      "step": 201580
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.1916277645978415e-05,
+      "loss": 1.1606,
+      "step": 201590
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.1915129716078665e-05,
+      "loss": 1.0541,
+      "step": 201600
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.1913981786178915e-05,
+      "loss": 1.0377,
+      "step": 201610
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.191283385627917e-05,
+      "loss": 0.9658,
+      "step": 201620
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.191168592637943e-05,
+      "loss": 1.0422,
+      "step": 201630
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.191053799647967e-05,
+      "loss": 0.9945,
+      "step": 201640
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.190939006657993e-05,
+      "loss": 1.1756,
+      "step": 201650
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.1908242136680184e-05,
+      "loss": 1.0867,
+      "step": 201660
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.190709420678043e-05,
+      "loss": 1.0526,
+      "step": 201670
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.1905946276880684e-05,
+      "loss": 1.0058,
+      "step": 201680
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.190479834698094e-05,
+      "loss": 1.1838,
+      "step": 201690
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.190365041708119e-05,
+      "loss": 1.1077,
+      "step": 201700
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.190250248718144e-05,
+      "loss": 1.0944,
+      "step": 201710
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.19013545572817e-05,
+      "loss": 1.0717,
+      "step": 201720
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.190020662738195e-05,
+      "loss": 1.057,
+      "step": 201730
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.1899058697482204e-05,
+      "loss": 1.07,
+      "step": 201740
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.1897910767582453e-05,
+      "loss": 1.0352,
+      "step": 201750
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.18967628376827e-05,
+      "loss": 1.0504,
+      "step": 201760
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.189561490778296e-05,
+      "loss": 1.0575,
+      "step": 201770
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.1894466977883217e-05,
+      "loss": 1.1655,
+      "step": 201780
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.189331904798346e-05,
+      "loss": 1.1205,
+      "step": 201790
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.1892171118083716e-05,
+      "loss": 1.1614,
+      "step": 201800
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.189102318818397e-05,
+      "loss": 1.0151,
+      "step": 201810
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.1889875258284216e-05,
+      "loss": 1.0451,
+      "step": 201820
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.188872732838447e-05,
+      "loss": 1.1251,
+      "step": 201830
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.188757939848473e-05,
+      "loss": 1.2401,
+      "step": 201840
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.188643146858498e-05,
+      "loss": 1.1231,
+      "step": 201850
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.188528353868523e-05,
+      "loss": 1.0845,
+      "step": 201860
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.1884135608785486e-05,
+      "loss": 1.0931,
+      "step": 201870
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.1882987678885735e-05,
+      "loss": 1.2273,
+      "step": 201880
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.188183974898599e-05,
+      "loss": 1.0672,
+      "step": 201890
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.188069181908624e-05,
+      "loss": 1.1939,
+      "step": 201900
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.187954388918649e-05,
+      "loss": 1.1354,
+      "step": 201910
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.187839595928675e-05,
+      "loss": 1.1256,
+      "step": 201920
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.1877248029387005e-05,
+      "loss": 1.0146,
+      "step": 201930
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.187610009948725e-05,
+      "loss": 1.268,
+      "step": 201940
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.1874952169587505e-05,
+      "loss": 1.0806,
+      "step": 201950
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.187380423968776e-05,
+      "loss": 1.1681,
+      "step": 201960
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.1872656309788004e-05,
+      "loss": 1.0477,
+      "step": 201970
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.187150837988826e-05,
+      "loss": 1.0877,
+      "step": 201980
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.187036044998852e-05,
+      "loss": 1.0842,
+      "step": 201990
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.186921252008877e-05,
+      "loss": 1.0522,
+      "step": 202000
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.186806459018902e-05,
+      "loss": 1.0662,
+      "step": 202010
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.1866916660289274e-05,
+      "loss": 1.0234,
+      "step": 202020
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.1865768730389524e-05,
+      "loss": 1.1887,
+      "step": 202030
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.186462080048978e-05,
+      "loss": 1.28,
+      "step": 202040
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.186347287059003e-05,
+      "loss": 1.1226,
+      "step": 202050
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.186232494069028e-05,
+      "loss": 0.965,
+      "step": 202060
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.186117701079054e-05,
+      "loss": 1.129,
+      "step": 202070
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.1860029080890794e-05,
+      "loss": 1.0724,
+      "step": 202080
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.1858881150991037e-05,
+      "loss": 1.0694,
+      "step": 202090
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.185773322109129e-05,
+      "loss": 1.3199,
+      "step": 202100
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.185658529119155e-05,
+      "loss": 1.067,
+      "step": 202110
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.185543736129179e-05,
+      "loss": 1.0006,
+      "step": 202120
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.185428943139205e-05,
+      "loss": 1.1368,
+      "step": 202130
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.1853141501492306e-05,
+      "loss": 1.0809,
+      "step": 202140
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.1851993571592556e-05,
+      "loss": 1.0994,
+      "step": 202150
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.1850845641692806e-05,
+      "loss": 1.0414,
+      "step": 202160
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.184969771179306e-05,
+      "loss": 1.0917,
+      "step": 202170
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.184854978189331e-05,
+      "loss": 1.1349,
+      "step": 202180
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.184740185199357e-05,
+      "loss": 1.083,
+      "step": 202190
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.184625392209382e-05,
+      "loss": 1.1125,
+      "step": 202200
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.184510599219407e-05,
+      "loss": 1.073,
+      "step": 202210
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.1843958062294325e-05,
+      "loss": 1.0651,
+      "step": 202220
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.184281013239458e-05,
+      "loss": 1.1302,
+      "step": 202230
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.1841662202494825e-05,
+      "loss": 1.0809,
+      "step": 202240
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.184051427259508e-05,
+      "loss": 1.1151,
+      "step": 202250
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.183936634269534e-05,
+      "loss": 1.0488,
+      "step": 202260
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.183821841279558e-05,
+      "loss": 1.1611,
+      "step": 202270
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.183707048289584e-05,
+      "loss": 1.1608,
+      "step": 202280
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.1835922552996095e-05,
+      "loss": 0.9822,
+      "step": 202290
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.1834774623096345e-05,
+      "loss": 1.0529,
+      "step": 202300
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.1833626693196594e-05,
+      "loss": 1.0865,
+      "step": 202310
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.183247876329685e-05,
+      "loss": 1.1117,
+      "step": 202320
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.18313308333971e-05,
+      "loss": 1.2608,
+      "step": 202330
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.183018290349736e-05,
+      "loss": 1.0611,
+      "step": 202340
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.182903497359761e-05,
+      "loss": 1.0043,
+      "step": 202350
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.182788704369786e-05,
+      "loss": 1.318,
+      "step": 202360
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.1826739113798114e-05,
+      "loss": 1.1658,
+      "step": 202370
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.182559118389837e-05,
+      "loss": 1.0974,
+      "step": 202380
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.1824443253998614e-05,
+      "loss": 1.0564,
+      "step": 202390
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.182329532409887e-05,
+      "loss": 1.1099,
+      "step": 202400
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.182214739419913e-05,
+      "loss": 1.0863,
+      "step": 202410
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.182099946429937e-05,
+      "loss": 1.0958,
+      "step": 202420
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.1819851534399626e-05,
+      "loss": 1.0689,
+      "step": 202430
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.181870360449988e-05,
+      "loss": 1.0412,
+      "step": 202440
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.181755567460013e-05,
+      "loss": 1.0915,
+      "step": 202450
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.181640774470038e-05,
+      "loss": 1.084,
+      "step": 202460
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.181525981480064e-05,
+      "loss": 1.0686,
+      "step": 202470
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.181411188490089e-05,
+      "loss": 1.1393,
+      "step": 202480
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.1812963955001146e-05,
+      "loss": 1.2493,
+      "step": 202490
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.1811816025101396e-05,
+      "loss": 1.1126,
+      "step": 202500
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.1810668095201646e-05,
+      "loss": 1.0713,
+      "step": 202510
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.18095201653019e-05,
+      "loss": 1.1632,
+      "step": 202520
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.180837223540216e-05,
+      "loss": 1.1492,
+      "step": 202530
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.18072243055024e-05,
+      "loss": 1.1496,
+      "step": 202540
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.180607637560266e-05,
+      "loss": 1.1,
+      "step": 202550
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.1804928445702915e-05,
+      "loss": 1.1391,
+      "step": 202560
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.180378051580316e-05,
+      "loss": 0.9851,
+      "step": 202570
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.1802632585903415e-05,
+      "loss": 1.1389,
+      "step": 202580
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.180148465600367e-05,
+      "loss": 1.1547,
+      "step": 202590
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.180033672610392e-05,
+      "loss": 1.0474,
+      "step": 202600
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.179918879620417e-05,
+      "loss": 1.0948,
+      "step": 202610
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.179804086630443e-05,
+      "loss": 0.9747,
+      "step": 202620
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.179689293640468e-05,
+      "loss": 1.0933,
+      "step": 202630
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.1795745006504934e-05,
+      "loss": 1.0204,
+      "step": 202640
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.1794597076605184e-05,
+      "loss": 1.0792,
+      "step": 202650
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.1793449146705434e-05,
+      "loss": 1.0688,
+      "step": 202660
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.179230121680569e-05,
+      "loss": 1.1431,
+      "step": 202670
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.179115328690595e-05,
+      "loss": 1.125,
+      "step": 202680
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.179000535700619e-05,
+      "loss": 1.083,
+      "step": 202690
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.178885742710645e-05,
+      "loss": 1.0269,
+      "step": 202700
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.1787709497206704e-05,
+      "loss": 1.0364,
+      "step": 202710
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.178656156730695e-05,
+      "loss": 1.082,
+      "step": 202720
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.1785413637407203e-05,
+      "loss": 1.264,
+      "step": 202730
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.178426570750746e-05,
+      "loss": 1.0156,
+      "step": 202740
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.178311777760771e-05,
+      "loss": 1.0676,
+      "step": 202750
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.178196984770796e-05,
+      "loss": 1.0276,
+      "step": 202760
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.1780821917808216e-05,
+      "loss": 1.1593,
+      "step": 202770
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.1779673987908466e-05,
+      "loss": 1.1386,
+      "step": 202780
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.177852605800872e-05,
+      "loss": 1.2282,
+      "step": 202790
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.177737812810897e-05,
+      "loss": 1.1155,
+      "step": 202800
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.177623019820922e-05,
+      "loss": 1.0472,
+      "step": 202810
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.177508226830948e-05,
+      "loss": 1.1336,
+      "step": 202820
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.1773934338409736e-05,
+      "loss": 1.0711,
+      "step": 202830
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.177278640850998e-05,
+      "loss": 1.0698,
+      "step": 202840
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.1771638478610236e-05,
+      "loss": 1.0929,
+      "step": 202850
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.177049054871049e-05,
+      "loss": 0.9648,
+      "step": 202860
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.1769342618810735e-05,
+      "loss": 1.0598,
+      "step": 202870
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.176819468891099e-05,
+      "loss": 1.1529,
+      "step": 202880
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.176704675901125e-05,
+      "loss": 1.1015,
+      "step": 202890
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.17658988291115e-05,
+      "loss": 1.0624,
+      "step": 202900
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.176475089921175e-05,
+      "loss": 1.103,
+      "step": 202910
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.1763602969312005e-05,
+      "loss": 1.055,
+      "step": 202920
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.1762455039412255e-05,
+      "loss": 0.9898,
+      "step": 202930
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.176130710951251e-05,
+      "loss": 1.1525,
+      "step": 202940
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.176015917961276e-05,
+      "loss": 1.1414,
+      "step": 202950
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.175901124971301e-05,
+      "loss": 1.1316,
+      "step": 202960
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.175786331981327e-05,
+      "loss": 1.0405,
+      "step": 202970
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.175671538991351e-05,
+      "loss": 1.1195,
+      "step": 202980
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.175556746001377e-05,
+      "loss": 1.0352,
+      "step": 202990
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.1754419530114024e-05,
+      "loss": 1.0475,
+      "step": 203000
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.1753271600214274e-05,
+      "loss": 1.2296,
+      "step": 203010
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.1752123670314524e-05,
+      "loss": 1.1745,
+      "step": 203020
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.175097574041478e-05,
+      "loss": 1.1592,
+      "step": 203030
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.174982781051503e-05,
+      "loss": 1.1139,
+      "step": 203040
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.174867988061529e-05,
+      "loss": 1.142,
+      "step": 203050
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.174753195071554e-05,
+      "loss": 1.1749,
+      "step": 203060
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.1746384020815787e-05,
+      "loss": 1.1293,
+      "step": 203070
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.174523609091604e-05,
+      "loss": 0.9982,
+      "step": 203080
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.17440881610163e-05,
+      "loss": 1.1441,
+      "step": 203090
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.174294023111654e-05,
+      "loss": 1.1874,
+      "step": 203100
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.17417923012168e-05,
+      "loss": 1.1592,
+      "step": 203110
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.1740644371317056e-05,
+      "loss": 1.153,
+      "step": 203120
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.17394964414173e-05,
+      "loss": 1.0391,
+      "step": 203130
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.1738348511517556e-05,
+      "loss": 1.1488,
+      "step": 203140
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.173720058161781e-05,
+      "loss": 1.0663,
+      "step": 203150
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.173605265171806e-05,
+      "loss": 1.0083,
+      "step": 203160
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.173490472181831e-05,
+      "loss": 1.1048,
+      "step": 203170
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.173375679191857e-05,
+      "loss": 1.0117,
+      "step": 203180
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.173260886201882e-05,
+      "loss": 1.0523,
+      "step": 203190
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.1731460932119075e-05,
+      "loss": 1.0784,
+      "step": 203200
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.1730313002219325e-05,
+      "loss": 1.0664,
+      "step": 203210
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.1729165072319575e-05,
+      "loss": 1.1422,
+      "step": 203220
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.172801714241983e-05,
+      "loss": 0.9855,
+      "step": 203230
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.172686921252009e-05,
+      "loss": 1.0645,
+      "step": 203240
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.172572128262033e-05,
+      "loss": 1.0461,
+      "step": 203250
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.172457335272059e-05,
+      "loss": 1.1177,
+      "step": 203260
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.1723425422820845e-05,
+      "loss": 1.0991,
+      "step": 203270
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.172227749292109e-05,
+      "loss": 1.1383,
+      "step": 203280
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.1721129563021344e-05,
+      "loss": 1.0311,
+      "step": 203290
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.17199816331216e-05,
+      "loss": 1.0039,
+      "step": 203300
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.171883370322185e-05,
+      "loss": 1.1909,
+      "step": 203310
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.17176857733221e-05,
+      "loss": 1.1354,
+      "step": 203320
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.171653784342236e-05,
+      "loss": 1.1392,
+      "step": 203330
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.171538991352261e-05,
+      "loss": 0.9917,
+      "step": 203340
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.1714241983622864e-05,
+      "loss": 1.0,
+      "step": 203350
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.1713094053723114e-05,
+      "loss": 1.0603,
+      "step": 203360
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.1711946123823364e-05,
+      "loss": 1.153,
+      "step": 203370
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.171079819392362e-05,
+      "loss": 1.1427,
+      "step": 203380
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.170965026402388e-05,
+      "loss": 1.2016,
+      "step": 203390
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.170850233412412e-05,
+      "loss": 1.1697,
+      "step": 203400
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.1707354404224376e-05,
+      "loss": 1.1004,
+      "step": 203410
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.170620647432463e-05,
+      "loss": 1.0994,
+      "step": 203420
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.1705058544424876e-05,
+      "loss": 1.0837,
+      "step": 203430
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.170391061452513e-05,
+      "loss": 1.0883,
+      "step": 203440
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.170276268462539e-05,
+      "loss": 1.1662,
+      "step": 203450
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.170161475472564e-05,
+      "loss": 1.1258,
+      "step": 203460
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.170046682482589e-05,
+      "loss": 1.1651,
+      "step": 203470
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.1699318894926146e-05,
+      "loss": 1.1552,
+      "step": 203480
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.1698170965026396e-05,
+      "loss": 0.9764,
+      "step": 203490
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.169702303512665e-05,
+      "loss": 1.1567,
+      "step": 203500
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.16958751052269e-05,
+      "loss": 1.1668,
+      "step": 203510
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.169472717532715e-05,
+      "loss": 1.1655,
+      "step": 203520
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.169357924542741e-05,
+      "loss": 1.0834,
+      "step": 203530
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.1692431315527665e-05,
+      "loss": 1.0439,
+      "step": 203540
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.169128338562791e-05,
+      "loss": 1.0707,
+      "step": 203550
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.1690135455728165e-05,
+      "loss": 1.0234,
+      "step": 203560
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.168898752582842e-05,
+      "loss": 1.0679,
+      "step": 203570
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.1687839595928665e-05,
+      "loss": 1.1411,
+      "step": 203580
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.168669166602892e-05,
+      "loss": 1.2161,
+      "step": 203590
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.168554373612918e-05,
+      "loss": 1.0775,
+      "step": 203600
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.168439580622943e-05,
+      "loss": 1.2959,
+      "step": 203610
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.168324787632968e-05,
+      "loss": 1.0013,
+      "step": 203620
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.1682099946429934e-05,
+      "loss": 1.1309,
+      "step": 203630
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.1680952016530184e-05,
+      "loss": 1.1142,
+      "step": 203640
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.167980408663044e-05,
+      "loss": 1.1607,
+      "step": 203650
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.167865615673069e-05,
+      "loss": 1.0562,
+      "step": 203660
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.167750822683094e-05,
+      "loss": 1.0493,
+      "step": 203670
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.16763602969312e-05,
+      "loss": 1.22,
+      "step": 203680
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.1675212367031454e-05,
+      "loss": 1.1353,
+      "step": 203690
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.16740644371317e-05,
+      "loss": 0.9504,
+      "step": 203700
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.1672916507231953e-05,
+      "loss": 1.168,
+      "step": 203710
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.167176857733221e-05,
+      "loss": 1.1809,
+      "step": 203720
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.167062064743245e-05,
+      "loss": 1.1105,
+      "step": 203730
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.166947271753271e-05,
+      "loss": 1.0986,
+      "step": 203740
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.1668324787632966e-05,
+      "loss": 0.9866,
+      "step": 203750
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.1667176857733216e-05,
+      "loss": 1.1121,
+      "step": 203760
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.1666028927833466e-05,
+      "loss": 1.1015,
+      "step": 203770
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.166488099793372e-05,
+      "loss": 1.1301,
+      "step": 203780
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.166373306803397e-05,
+      "loss": 1.1154,
+      "step": 203790
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.166258513813423e-05,
+      "loss": 0.9976,
+      "step": 203800
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.166143720823448e-05,
+      "loss": 1.0438,
+      "step": 203810
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.166028927833473e-05,
+      "loss": 1.1564,
+      "step": 203820
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.1659141348434986e-05,
+      "loss": 1.2202,
+      "step": 203830
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.165799341853524e-05,
+      "loss": 1.0787,
+      "step": 203840
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.1656845488635485e-05,
+      "loss": 1.1275,
+      "step": 203850
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.165569755873574e-05,
+      "loss": 1.0826,
+      "step": 203860
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.1654549628836e-05,
+      "loss": 1.1588,
+      "step": 203870
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.165340169893624e-05,
+      "loss": 1.1439,
+      "step": 203880
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.16522537690365e-05,
+      "loss": 1.1639,
+      "step": 203890
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.1651105839136755e-05,
+      "loss": 1.1096,
+      "step": 203900
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.1649957909237005e-05,
+      "loss": 1.0402,
+      "step": 203910
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.1648809979337255e-05,
+      "loss": 0.9695,
+      "step": 203920
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.164766204943751e-05,
+      "loss": 1.0388,
+      "step": 203930
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.164651411953776e-05,
+      "loss": 1.0506,
+      "step": 203940
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.164536618963802e-05,
+      "loss": 1.0374,
+      "step": 203950
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.164421825973827e-05,
+      "loss": 1.1732,
+      "step": 203960
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.164307032983852e-05,
+      "loss": 1.0108,
+      "step": 203970
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.1641922399938774e-05,
+      "loss": 1.2013,
+      "step": 203980
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.164077447003903e-05,
+      "loss": 1.1265,
+      "step": 203990
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.1639626540139274e-05,
+      "loss": 1.0863,
+      "step": 204000
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.163847861023953e-05,
+      "loss": 1.1484,
+      "step": 204010
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.163733068033979e-05,
+      "loss": 1.0531,
+      "step": 204020
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.163618275044003e-05,
+      "loss": 1.0371,
+      "step": 204030
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.163503482054029e-05,
+      "loss": 1.0792,
+      "step": 204040
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.163388689064054e-05,
+      "loss": 1.0005,
+      "step": 204050
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.163273896074079e-05,
+      "loss": 0.9657,
+      "step": 204060
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.163159103084104e-05,
+      "loss": 1.0154,
+      "step": 204070
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.16304431009413e-05,
+      "loss": 1.1077,
+      "step": 204080
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.162929517104155e-05,
+      "loss": 1.1121,
+      "step": 204090
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.1628147241141806e-05,
+      "loss": 1.0775,
+      "step": 204100
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.1626999311242056e-05,
+      "loss": 0.9859,
+      "step": 204110
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.1625851381342306e-05,
+      "loss": 1.0096,
+      "step": 204120
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.162470345144256e-05,
+      "loss": 1.1696,
+      "step": 204130
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.162355552154282e-05,
+      "loss": 1.0995,
+      "step": 204140
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.162240759164306e-05,
+      "loss": 1.0621,
+      "step": 204150
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.162125966174332e-05,
+      "loss": 1.1592,
+      "step": 204160
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.1620111731843575e-05,
+      "loss": 1.1088,
+      "step": 204170
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.161896380194382e-05,
+      "loss": 1.0454,
+      "step": 204180
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.1617815872044075e-05,
+      "loss": 1.0965,
+      "step": 204190
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.161666794214433e-05,
+      "loss": 1.1609,
+      "step": 204200
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.161552001224458e-05,
+      "loss": 1.0763,
+      "step": 204210
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.161437208234483e-05,
+      "loss": 0.9228,
+      "step": 204220
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.161322415244509e-05,
+      "loss": 1.148,
+      "step": 204230
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.161207622254534e-05,
+      "loss": 0.9804,
+      "step": 204240
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.1610928292645595e-05,
+      "loss": 1.1976,
+      "step": 204250
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.1609780362745844e-05,
+      "loss": 1.1175,
+      "step": 204260
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.1608632432846094e-05,
+      "loss": 0.8894,
+      "step": 204270
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.160748450294635e-05,
+      "loss": 1.1255,
+      "step": 204280
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.160633657304661e-05,
+      "loss": 0.9757,
+      "step": 204290
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.160518864314685e-05,
+      "loss": 1.0761,
+      "step": 204300
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.160404071324711e-05,
+      "loss": 1.1114,
+      "step": 204310
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.1602892783347364e-05,
+      "loss": 1.0247,
+      "step": 204320
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.160174485344761e-05,
+      "loss": 1.0501,
+      "step": 204330
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.1600596923547864e-05,
+      "loss": 1.1128,
+      "step": 204340
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.1599448993648114e-05,
+      "loss": 1.198,
+      "step": 204350
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.159830106374837e-05,
+      "loss": 1.0906,
+      "step": 204360
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.159715313384862e-05,
+      "loss": 1.1328,
+      "step": 204370
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.159600520394887e-05,
+      "loss": 1.1966,
+      "step": 204380
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.1594857274049126e-05,
+      "loss": 1.0601,
+      "step": 204390
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.159370934414938e-05,
+      "loss": 1.074,
+      "step": 204400
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.1592561414249626e-05,
+      "loss": 1.1514,
+      "step": 204410
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.159141348434988e-05,
+      "loss": 1.1117,
+      "step": 204420
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.159026555445014e-05,
+      "loss": 0.9768,
+      "step": 204430
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.158911762455038e-05,
+      "loss": 1.0879,
+      "step": 204440
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.158796969465064e-05,
+      "loss": 1.1422,
+      "step": 204450
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.1586821764750896e-05,
+      "loss": 1.1098,
+      "step": 204460
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.1585673834851146e-05,
+      "loss": 0.9585,
+      "step": 204470
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.1584525904951395e-05,
+      "loss": 1.1804,
+      "step": 204480
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.158337797505165e-05,
+      "loss": 1.0745,
+      "step": 204490
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.15822300451519e-05,
+      "loss": 1.2367,
+      "step": 204500
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.158108211525216e-05,
+      "loss": 1.1474,
+      "step": 204510
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.157993418535241e-05,
+      "loss": 1.1956,
+      "step": 204520
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.157878625545266e-05,
+      "loss": 1.1967,
+      "step": 204530
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.1577638325552915e-05,
+      "loss": 1.1106,
+      "step": 204540
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.157649039565317e-05,
+      "loss": 1.135,
+      "step": 204550
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.1575342465753415e-05,
+      "loss": 1.0987,
+      "step": 204560
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.157419453585367e-05,
+      "loss": 1.0489,
+      "step": 204570
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.157304660595393e-05,
+      "loss": 1.018,
+      "step": 204580
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.157189867605417e-05,
+      "loss": 1.1532,
+      "step": 204590
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.157075074615443e-05,
+      "loss": 1.0987,
+      "step": 204600
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.1569602816254684e-05,
+      "loss": 1.0532,
+      "step": 204610
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.1568454886354934e-05,
+      "loss": 1.1294,
+      "step": 204620
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.1567306956455184e-05,
+      "loss": 1.0364,
+      "step": 204630
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.156615902655544e-05,
+      "loss": 1.1846,
+      "step": 204640
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.156501109665569e-05,
+      "loss": 1.1624,
+      "step": 204650
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.156386316675595e-05,
+      "loss": 1.1563,
+      "step": 204660
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.15627152368562e-05,
+      "loss": 1.1467,
+      "step": 204670
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.156156730695645e-05,
+      "loss": 1.1113,
+      "step": 204680
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.1560419377056703e-05,
+      "loss": 1.1005,
+      "step": 204690
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.155927144715696e-05,
+      "loss": 1.0538,
+      "step": 204700
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.15581235172572e-05,
+      "loss": 1.1107,
+      "step": 204710
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.155697558735746e-05,
+      "loss": 1.1293,
+      "step": 204720
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.1555827657457716e-05,
+      "loss": 1.0238,
+      "step": 204730
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.155467972755796e-05,
+      "loss": 1.0611,
+      "step": 204740
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.1553531797658216e-05,
+      "loss": 1.0407,
+      "step": 204750
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.155238386775847e-05,
+      "loss": 1.3005,
+      "step": 204760
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.155123593785872e-05,
+      "loss": 1.1814,
+      "step": 204770
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.155008800795897e-05,
+      "loss": 1.1495,
+      "step": 204780
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.154894007805923e-05,
+      "loss": 1.2238,
+      "step": 204790
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.154779214815948e-05,
+      "loss": 1.1368,
+      "step": 204800
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.1546644218259736e-05,
+      "loss": 1.0302,
+      "step": 204810
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.1545496288359985e-05,
+      "loss": 1.1915,
+      "step": 204820
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.1544348358460235e-05,
+      "loss": 1.1184,
+      "step": 204830
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.154320042856049e-05,
+      "loss": 1.1395,
+      "step": 204840
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.154205249866075e-05,
+      "loss": 1.1525,
+      "step": 204850
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.154090456876099e-05,
+      "loss": 1.0606,
+      "step": 204860
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.153975663886125e-05,
+      "loss": 1.0418,
+      "step": 204870
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.1538608708961505e-05,
+      "loss": 1.1287,
+      "step": 204880
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.153746077906175e-05,
+      "loss": 1.0837,
+      "step": 204890
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.1536312849162005e-05,
+      "loss": 1.1232,
+      "step": 204900
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.153516491926226e-05,
+      "loss": 1.12,
+      "step": 204910
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.153401698936251e-05,
+      "loss": 1.1188,
+      "step": 204920
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.153286905946276e-05,
+      "loss": 1.0222,
+      "step": 204930
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.153172112956302e-05,
+      "loss": 1.1319,
+      "step": 204940
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.153057319966327e-05,
+      "loss": 1.1003,
+      "step": 204950
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.1529425269763524e-05,
+      "loss": 1.262,
+      "step": 204960
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.1528277339863774e-05,
+      "loss": 1.0517,
+      "step": 204970
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.1527129409964024e-05,
+      "loss": 1.1331,
+      "step": 204980
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.152598148006428e-05,
+      "loss": 1.2954,
+      "step": 204990
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.152483355016454e-05,
+      "loss": 1.0654,
+      "step": 205000
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.152368562026478e-05,
+      "loss": 1.1114,
+      "step": 205010
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.152253769036504e-05,
+      "loss": 1.1772,
+      "step": 205020
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.152138976046529e-05,
+      "loss": 1.0961,
+      "step": 205030
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.1520241830565536e-05,
+      "loss": 1.0654,
+      "step": 205040
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.151909390066579e-05,
+      "loss": 1.0714,
+      "step": 205050
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.151794597076605e-05,
+      "loss": 1.0797,
+      "step": 205060
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.15167980408663e-05,
+      "loss": 1.0696,
+      "step": 205070
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.151565011096655e-05,
+      "loss": 1.0347,
+      "step": 205080
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.1514502181066806e-05,
+      "loss": 1.0928,
+      "step": 205090
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.1513354251167056e-05,
+      "loss": 0.9951,
+      "step": 205100
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.151220632126731e-05,
+      "loss": 1.0593,
+      "step": 205110
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.151105839136756e-05,
+      "loss": 1.0033,
+      "step": 205120
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.150991046146781e-05,
+      "loss": 1.0964,
+      "step": 205130
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.150876253156807e-05,
+      "loss": 1.0665,
+      "step": 205140
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.1507614601668325e-05,
+      "loss": 1.1758,
+      "step": 205150
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.150646667176857e-05,
+      "loss": 1.159,
+      "step": 205160
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.1505318741868825e-05,
+      "loss": 1.0657,
+      "step": 205170
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.150417081196908e-05,
+      "loss": 1.2006,
+      "step": 205180
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.1503022882069325e-05,
+      "loss": 1.0647,
+      "step": 205190
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.150187495216958e-05,
+      "loss": 1.1236,
+      "step": 205200
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.150072702226984e-05,
+      "loss": 1.2265,
+      "step": 205210
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.149957909237009e-05,
+      "loss": 1.0529,
+      "step": 205220
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.149843116247034e-05,
+      "loss": 0.9279,
+      "step": 205230
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.1497283232570594e-05,
+      "loss": 1.104,
+      "step": 205240
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.1496135302670844e-05,
+      "loss": 1.1415,
+      "step": 205250
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.14949873727711e-05,
+      "loss": 1.1208,
+      "step": 205260
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.149383944287135e-05,
+      "loss": 1.0682,
+      "step": 205270
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.14926915129716e-05,
+      "loss": 1.1252,
+      "step": 205280
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.149154358307186e-05,
+      "loss": 1.0988,
+      "step": 205290
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.1490395653172114e-05,
+      "loss": 1.0328,
+      "step": 205300
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.148924772327236e-05,
+      "loss": 1.0001,
+      "step": 205310
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.1488099793372614e-05,
+      "loss": 1.1269,
+      "step": 205320
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.148695186347287e-05,
+      "loss": 1.1335,
+      "step": 205330
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.148580393357311e-05,
+      "loss": 1.092,
+      "step": 205340
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.148465600367337e-05,
+      "loss": 1.0019,
+      "step": 205350
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.148350807377363e-05,
+      "loss": 0.9995,
+      "step": 205360
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.1482360143873876e-05,
+      "loss": 1.1422,
+      "step": 205370
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.1481212213974126e-05,
+      "loss": 1.1609,
+      "step": 205380
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.148006428407438e-05,
+      "loss": 1.2736,
+      "step": 205390
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.147891635417463e-05,
+      "loss": 1.1123,
+      "step": 205400
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.147776842427489e-05,
+      "loss": 1.0525,
+      "step": 205410
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.147662049437514e-05,
+      "loss": 1.117,
+      "step": 205420
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.147547256447539e-05,
+      "loss": 1.1071,
+      "step": 205430
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.1474324634575646e-05,
+      "loss": 1.0688,
+      "step": 205440
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.14731767046759e-05,
+      "loss": 1.1318,
+      "step": 205450
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.1472028774776145e-05,
+      "loss": 1.1062,
+      "step": 205460
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.14708808448764e-05,
+      "loss": 0.9812,
+      "step": 205470
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.146973291497666e-05,
+      "loss": 1.2306,
+      "step": 205480
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.14685849850769e-05,
+      "loss": 1.1113,
+      "step": 205490
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.146743705517716e-05,
+      "loss": 1.1389,
+      "step": 205500
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.1466289125277415e-05,
+      "loss": 1.2135,
+      "step": 205510
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.1465141195377665e-05,
+      "loss": 1.0488,
+      "step": 205520
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.1463993265477915e-05,
+      "loss": 1.1131,
+      "step": 205530
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.146284533557817e-05,
+      "loss": 1.0959,
+      "step": 205540
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.146169740567842e-05,
+      "loss": 1.1016,
+      "step": 205550
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.146054947577868e-05,
+      "loss": 1.0491,
+      "step": 205560
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.145940154587893e-05,
+      "loss": 1.1112,
+      "step": 205570
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.145825361597918e-05,
+      "loss": 1.1358,
+      "step": 205580
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.1457105686079434e-05,
+      "loss": 1.0628,
+      "step": 205590
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.145595775617969e-05,
+      "loss": 1.0503,
+      "step": 205600
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.1454809826279934e-05,
+      "loss": 1.04,
+      "step": 205610
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.145366189638019e-05,
+      "loss": 1.2369,
+      "step": 205620
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.145251396648045e-05,
+      "loss": 1.1795,
+      "step": 205630
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.145136603658069e-05,
+      "loss": 1.0317,
+      "step": 205640
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.145021810668095e-05,
+      "loss": 1.0655,
+      "step": 205650
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.1449070176781204e-05,
+      "loss": 1.1127,
+      "step": 205660
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.1447922246881453e-05,
+      "loss": 1.1282,
+      "step": 205670
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.14467743169817e-05,
+      "loss": 1.1163,
+      "step": 205680
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.144562638708196e-05,
+      "loss": 1.1722,
+      "step": 205690
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.144447845718221e-05,
+      "loss": 1.07,
+      "step": 205700
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.1443330527282466e-05,
+      "loss": 1.1597,
+      "step": 205710
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.144218259738271e-05,
+      "loss": 1.0256,
+      "step": 205720
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.1441034667482966e-05,
+      "loss": 1.2024,
+      "step": 205730
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.143988673758322e-05,
+      "loss": 1.2334,
+      "step": 205740
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.1438738807683466e-05,
+      "loss": 1.1276,
+      "step": 205750
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.143759087778372e-05,
+      "loss": 1.0167,
+      "step": 205760
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.143644294788398e-05,
+      "loss": 1.1787,
+      "step": 205770
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.143529501798423e-05,
+      "loss": 1.1997,
+      "step": 205780
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.143414708808448e-05,
+      "loss": 1.1232,
+      "step": 205790
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.1432999158184735e-05,
+      "loss": 0.9811,
+      "step": 205800
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.1431851228284985e-05,
+      "loss": 0.9983,
+      "step": 205810
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.143070329838524e-05,
+      "loss": 1.1224,
+      "step": 205820
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.142955536848549e-05,
+      "loss": 1.011,
+      "step": 205830
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.142840743858574e-05,
+      "loss": 1.052,
+      "step": 205840
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.1427259508686e-05,
+      "loss": 1.0181,
+      "step": 205850
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.1426111578786255e-05,
+      "loss": 1.1933,
+      "step": 205860
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.14249636488865e-05,
+      "loss": 1.1578,
+      "step": 205870
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.1423815718986755e-05,
+      "loss": 1.2404,
+      "step": 205880
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.142266778908701e-05,
+      "loss": 1.0593,
+      "step": 205890
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.1421519859187254e-05,
+      "loss": 1.1349,
+      "step": 205900
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.142037192928751e-05,
+      "loss": 1.1496,
+      "step": 205910
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.141922399938777e-05,
+      "loss": 1.1926,
+      "step": 205920
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.141807606948802e-05,
+      "loss": 1.1182,
+      "step": 205930
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.141692813958827e-05,
+      "loss": 0.9835,
+      "step": 205940
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.1415780209688524e-05,
+      "loss": 1.1193,
+      "step": 205950
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.1414632279788774e-05,
+      "loss": 1.1118,
+      "step": 205960
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.141348434988903e-05,
+      "loss": 1.1898,
+      "step": 205970
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.141233641998928e-05,
+      "loss": 1.062,
+      "step": 205980
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.141118849008953e-05,
+      "loss": 1.0122,
+      "step": 205990
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.141004056018979e-05,
+      "loss": 1.1693,
+      "step": 206000
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.140889263029004e-05,
+      "loss": 1.0539,
+      "step": 206010
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.1407744700390286e-05,
+      "loss": 1.0782,
+      "step": 206020
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.140659677049054e-05,
+      "loss": 0.9977,
+      "step": 206030
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.14054488405908e-05,
+      "loss": 0.988,
+      "step": 206040
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.140430091069104e-05,
+      "loss": 1.0501,
+      "step": 206050
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.14031529807913e-05,
+      "loss": 1.0768,
+      "step": 206060
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.1402005050891556e-05,
+      "loss": 0.9633,
+      "step": 206070
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.1400857120991806e-05,
+      "loss": 1.1356,
+      "step": 206080
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.1399709191092056e-05,
+      "loss": 1.0486,
+      "step": 206090
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.139856126119231e-05,
+      "loss": 1.2308,
+      "step": 206100
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.139741333129256e-05,
+      "loss": 1.06,
+      "step": 206110
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.139626540139282e-05,
+      "loss": 1.0555,
+      "step": 206120
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.139511747149307e-05,
+      "loss": 1.1556,
+      "step": 206130
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.139396954159332e-05,
+      "loss": 1.0886,
+      "step": 206140
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.1392821611693575e-05,
+      "loss": 1.0641,
+      "step": 206150
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.139167368179383e-05,
+      "loss": 1.164,
+      "step": 206160
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.1390525751894075e-05,
+      "loss": 1.19,
+      "step": 206170
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.138937782199433e-05,
+      "loss": 1.2057,
+      "step": 206180
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.138822989209459e-05,
+      "loss": 1.0888,
+      "step": 206190
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.138708196219483e-05,
+      "loss": 1.0671,
+      "step": 206200
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.138593403229509e-05,
+      "loss": 1.0692,
+      "step": 206210
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.1384786102395344e-05,
+      "loss": 1.1699,
+      "step": 206220
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.1383638172495594e-05,
+      "loss": 1.0748,
+      "step": 206230
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.1382490242595844e-05,
+      "loss": 1.1164,
+      "step": 206240
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.13813423126961e-05,
+      "loss": 1.0084,
+      "step": 206250
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.138019438279635e-05,
+      "loss": 0.9726,
+      "step": 206260
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.137904645289661e-05,
+      "loss": 1.1715,
+      "step": 206270
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.137789852299686e-05,
+      "loss": 1.1647,
+      "step": 206280
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.137675059309711e-05,
+      "loss": 1.0141,
+      "step": 206290
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.1375602663197364e-05,
+      "loss": 1.187,
+      "step": 206300
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.137445473329762e-05,
+      "loss": 1.1649,
+      "step": 206310
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.137330680339786e-05,
+      "loss": 1.125,
+      "step": 206320
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.137215887349812e-05,
+      "loss": 1.0399,
+      "step": 206330
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.137101094359838e-05,
+      "loss": 1.0487,
+      "step": 206340
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.136986301369862e-05,
+      "loss": 1.1112,
+      "step": 206350
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.1368715083798876e-05,
+      "loss": 1.1271,
+      "step": 206360
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.136756715389913e-05,
+      "loss": 1.109,
+      "step": 206370
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.136641922399938e-05,
+      "loss": 1.1259,
+      "step": 206380
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.136527129409963e-05,
+      "loss": 1.1939,
+      "step": 206390
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.136412336419989e-05,
+      "loss": 0.9679,
+      "step": 206400
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.136297543430014e-05,
+      "loss": 1.046,
+      "step": 206410
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.1361827504400396e-05,
+      "loss": 1.2551,
+      "step": 206420
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.1360679574500646e-05,
+      "loss": 1.1487,
+      "step": 206430
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.1359531644600895e-05,
+      "loss": 1.2188,
+      "step": 206440
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.135838371470115e-05,
+      "loss": 1.0009,
+      "step": 206450
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.135723578480141e-05,
+      "loss": 1.0726,
+      "step": 206460
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.135608785490165e-05,
+      "loss": 1.1334,
+      "step": 206470
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.135493992500191e-05,
+      "loss": 1.1746,
+      "step": 206480
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.1353791995102165e-05,
+      "loss": 1.1136,
+      "step": 206490
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.135264406520241e-05,
+      "loss": 1.0916,
+      "step": 206500
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.1351496135302665e-05,
+      "loss": 1.1152,
+      "step": 206510
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.135034820540292e-05,
+      "loss": 1.1208,
+      "step": 206520
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.134920027550317e-05,
+      "loss": 1.1011,
+      "step": 206530
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.134805234560342e-05,
+      "loss": 1.1397,
+      "step": 206540
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.134690441570368e-05,
+      "loss": 1.1019,
+      "step": 206550
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.134575648580393e-05,
+      "loss": 1.075,
+      "step": 206560
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.1344608555904184e-05,
+      "loss": 1.082,
+      "step": 206570
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.1343460626004434e-05,
+      "loss": 1.1818,
+      "step": 206580
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.1342312696104684e-05,
+      "loss": 1.1761,
+      "step": 206590
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.134116476620494e-05,
+      "loss": 1.1147,
+      "step": 206600
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.13400168363052e-05,
+      "loss": 1.1148,
+      "step": 206610
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.133886890640544e-05,
+      "loss": 1.1992,
+      "step": 206620
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.13377209765057e-05,
+      "loss": 1.1087,
+      "step": 206630
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.1336573046605954e-05,
+      "loss": 0.9853,
+      "step": 206640
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.13354251167062e-05,
+      "loss": 1.0652,
+      "step": 206650
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.133427718680645e-05,
+      "loss": 1.2165,
+      "step": 206660
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.133312925690671e-05,
+      "loss": 1.1077,
+      "step": 206670
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.133198132700696e-05,
+      "loss": 0.9497,
+      "step": 206680
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.133083339710721e-05,
+      "loss": 1.0973,
+      "step": 206690
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.1329685467207466e-05,
+      "loss": 1.0509,
+      "step": 206700
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.1328537537307716e-05,
+      "loss": 1.1384,
+      "step": 206710
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.132738960740797e-05,
+      "loss": 1.2341,
+      "step": 206720
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.132624167750822e-05,
+      "loss": 1.1059,
+      "step": 206730
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.132509374760847e-05,
+      "loss": 1.0105,
+      "step": 206740
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.132394581770873e-05,
+      "loss": 1.082,
+      "step": 206750
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.1322797887808986e-05,
+      "loss": 1.0277,
+      "step": 206760
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.132164995790923e-05,
+      "loss": 1.2294,
+      "step": 206770
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.1320502028009485e-05,
+      "loss": 1.1903,
+      "step": 206780
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.131935409810974e-05,
+      "loss": 1.0422,
+      "step": 206790
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.1318206168209985e-05,
+      "loss": 1.1202,
+      "step": 206800
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.131705823831024e-05,
+      "loss": 1.1158,
+      "step": 206810
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.13159103084105e-05,
+      "loss": 1.0891,
+      "step": 206820
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.131476237851075e-05,
+      "loss": 1.0882,
+      "step": 206830
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.1313614448611e-05,
+      "loss": 1.086,
+      "step": 206840
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.1312466518711255e-05,
+      "loss": 1.0726,
+      "step": 206850
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.1311318588811505e-05,
+      "loss": 1.0539,
+      "step": 206860
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.131017065891176e-05,
+      "loss": 1.0426,
+      "step": 206870
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.130902272901201e-05,
+      "loss": 1.0534,
+      "step": 206880
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.130787479911226e-05,
+      "loss": 1.0663,
+      "step": 206890
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.130672686921252e-05,
+      "loss": 1.1264,
+      "step": 206900
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.1305578939312774e-05,
+      "loss": 1.0758,
+      "step": 206910
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.130443100941302e-05,
+      "loss": 0.9808,
+      "step": 206920
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.1303283079513274e-05,
+      "loss": 0.9797,
+      "step": 206930
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.130213514961353e-05,
+      "loss": 1.0596,
+      "step": 206940
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.1300987219713774e-05,
+      "loss": 1.1561,
+      "step": 206950
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.129983928981403e-05,
+      "loss": 1.0832,
+      "step": 206960
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.129869135991429e-05,
+      "loss": 1.0767,
+      "step": 206970
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.129754343001454e-05,
+      "loss": 1.14,
+      "step": 206980
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.1296395500114787e-05,
+      "loss": 1.1109,
+      "step": 206990
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.129524757021504e-05,
+      "loss": 1.1064,
+      "step": 207000
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.129409964031529e-05,
+      "loss": 1.1563,
+      "step": 207010
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.129295171041555e-05,
+      "loss": 1.112,
+      "step": 207020
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.12918037805158e-05,
+      "loss": 1.1731,
+      "step": 207030
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.129065585061605e-05,
+      "loss": 1.0633,
+      "step": 207040
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.1289507920716306e-05,
+      "loss": 1.2442,
+      "step": 207050
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.128835999081656e-05,
+      "loss": 1.036,
+      "step": 207060
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.1287212060916806e-05,
+      "loss": 1.076,
+      "step": 207070
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.128606413101706e-05,
+      "loss": 1.095,
+      "step": 207080
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.128491620111731e-05,
+      "loss": 1.1561,
+      "step": 207090
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.128376827121756e-05,
+      "loss": 1.1272,
+      "step": 207100
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.128262034131782e-05,
+      "loss": 1.059,
+      "step": 207110
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.128147241141807e-05,
+      "loss": 1.0657,
+      "step": 207120
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.1280324481518325e-05,
+      "loss": 1.1242,
+      "step": 207130
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.1279176551618575e-05,
+      "loss": 1.1133,
+      "step": 207140
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.1278028621718825e-05,
+      "loss": 1.2243,
+      "step": 207150
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.127688069181908e-05,
+      "loss": 1.0056,
+      "step": 207160
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.127573276191934e-05,
+      "loss": 1.1448,
+      "step": 207170
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.127458483201958e-05,
+      "loss": 1.1171,
+      "step": 207180
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.127343690211984e-05,
+      "loss": 1.1067,
+      "step": 207190
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.1272288972220094e-05,
+      "loss": 1.1121,
+      "step": 207200
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.127114104232034e-05,
+      "loss": 1.1175,
+      "step": 207210
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.1269993112420594e-05,
+      "loss": 1.0798,
+      "step": 207220
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.126884518252085e-05,
+      "loss": 1.0892,
+      "step": 207230
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.12676972526211e-05,
+      "loss": 1.1675,
+      "step": 207240
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.126654932272135e-05,
+      "loss": 1.0792,
+      "step": 207250
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.126540139282161e-05,
+      "loss": 1.0195,
+      "step": 207260
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.126425346292186e-05,
+      "loss": 1.0475,
+      "step": 207270
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.1263105533022114e-05,
+      "loss": 1.1366,
+      "step": 207280
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.1261957603122363e-05,
+      "loss": 0.9739,
+      "step": 207290
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.126080967322261e-05,
+      "loss": 1.0372,
+      "step": 207300
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.125966174332287e-05,
+      "loss": 1.1103,
+      "step": 207310
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.125851381342313e-05,
+      "loss": 1.0109,
+      "step": 207320
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.125736588352337e-05,
+      "loss": 1.0506,
+      "step": 207330
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.1256217953623626e-05,
+      "loss": 1.2036,
+      "step": 207340
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.125507002372388e-05,
+      "loss": 1.0235,
+      "step": 207350
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.1253922093824126e-05,
+      "loss": 1.1387,
+      "step": 207360
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.125277416392438e-05,
+      "loss": 1.0367,
+      "step": 207370
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.125162623402464e-05,
+      "loss": 1.0867,
+      "step": 207380
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.125047830412489e-05,
+      "loss": 1.026,
+      "step": 207390
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.124933037422514e-05,
+      "loss": 1.0714,
+      "step": 207400
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.1248182444325396e-05,
+      "loss": 1.1632,
+      "step": 207410
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.1247034514425645e-05,
+      "loss": 1.0755,
+      "step": 207420
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.12458865845259e-05,
+      "loss": 1.0451,
+      "step": 207430
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.124473865462615e-05,
+      "loss": 1.1359,
+      "step": 207440
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.12435907247264e-05,
+      "loss": 1.1263,
+      "step": 207450
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.124244279482666e-05,
+      "loss": 1.0827,
+      "step": 207460
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.1241294864926915e-05,
+      "loss": 1.0722,
+      "step": 207470
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.124014693502716e-05,
+      "loss": 1.0695,
+      "step": 207480
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.1238999005127415e-05,
+      "loss": 1.0802,
+      "step": 207490
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.123785107522767e-05,
+      "loss": 1.0735,
+      "step": 207500
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.1236703145327915e-05,
+      "loss": 1.3179,
+      "step": 207510
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.123555521542817e-05,
+      "loss": 1.0259,
+      "step": 207520
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.123440728552843e-05,
+      "loss": 1.171,
+      "step": 207530
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.123325935562868e-05,
+      "loss": 1.0945,
+      "step": 207540
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.123211142572893e-05,
+      "loss": 1.1511,
+      "step": 207550
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.1230963495829184e-05,
+      "loss": 1.0866,
+      "step": 207560
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.1229815565929434e-05,
+      "loss": 1.1566,
+      "step": 207570
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.122866763602969e-05,
+      "loss": 1.128,
+      "step": 207580
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.122751970612994e-05,
+      "loss": 1.2009,
+      "step": 207590
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.122637177623019e-05,
+      "loss": 1.0405,
+      "step": 207600
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.122522384633045e-05,
+      "loss": 1.2521,
+      "step": 207610
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.1224075916430704e-05,
+      "loss": 1.0429,
+      "step": 207620
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.122292798653095e-05,
+      "loss": 1.0561,
+      "step": 207630
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.12217800566312e-05,
+      "loss": 1.1469,
+      "step": 207640
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.122063212673146e-05,
+      "loss": 1.1883,
+      "step": 207650
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.12194841968317e-05,
+      "loss": 1.1815,
+      "step": 207660
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.121833626693196e-05,
+      "loss": 1.1169,
+      "step": 207670
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.1217188337032216e-05,
+      "loss": 0.9716,
+      "step": 207680
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.1216040407132466e-05,
+      "loss": 1.2128,
+      "step": 207690
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.1214892477232716e-05,
+      "loss": 1.0236,
+      "step": 207700
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.121374454733297e-05,
+      "loss": 1.0731,
+      "step": 207710
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.121259661743322e-05,
+      "loss": 1.1791,
+      "step": 207720
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.121144868753348e-05,
+      "loss": 1.0586,
+      "step": 207730
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.121030075763373e-05,
+      "loss": 1.0563,
+      "step": 207740
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.120915282773398e-05,
+      "loss": 1.123,
+      "step": 207750
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.1208004897834235e-05,
+      "loss": 1.1466,
+      "step": 207760
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.120685696793449e-05,
+      "loss": 1.0043,
+      "step": 207770
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.1205709038034735e-05,
+      "loss": 1.0638,
+      "step": 207780
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.120456110813499e-05,
+      "loss": 1.2132,
+      "step": 207790
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.120341317823525e-05,
+      "loss": 1.1108,
+      "step": 207800
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.120226524833549e-05,
+      "loss": 1.1039,
+      "step": 207810
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.120111731843575e-05,
+      "loss": 1.2144,
+      "step": 207820
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.1199969388536005e-05,
+      "loss": 1.0118,
+      "step": 207830
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.1198821458636255e-05,
+      "loss": 1.117,
+      "step": 207840
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.1197673528736504e-05,
+      "loss": 1.2398,
+      "step": 207850
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.119652559883676e-05,
+      "loss": 1.2072,
+      "step": 207860
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.119537766893701e-05,
+      "loss": 1.058,
+      "step": 207870
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.119422973903727e-05,
+      "loss": 1.2165,
+      "step": 207880
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.119308180913752e-05,
+      "loss": 1.1242,
+      "step": 207890
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.119193387923777e-05,
+      "loss": 1.3369,
+      "step": 207900
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.1190785949338024e-05,
+      "loss": 1.0836,
+      "step": 207910
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.118963801943828e-05,
+      "loss": 1.0272,
+      "step": 207920
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.1188490089538524e-05,
+      "loss": 1.184,
+      "step": 207930
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.118734215963878e-05,
+      "loss": 1.2239,
+      "step": 207940
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.118619422973904e-05,
+      "loss": 1.1461,
+      "step": 207950
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.118504629983928e-05,
+      "loss": 0.9931,
+      "step": 207960
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.1183898369939537e-05,
+      "loss": 1.1495,
+      "step": 207970
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.118275044003979e-05,
+      "loss": 1.0397,
+      "step": 207980
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.118160251014004e-05,
+      "loss": 1.0623,
+      "step": 207990
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.118045458024029e-05,
+      "loss": 1.1974,
+      "step": 208000
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.117930665034055e-05,
+      "loss": 1.0196,
+      "step": 208010
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.11781587204408e-05,
+      "loss": 1.1242,
+      "step": 208020
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.1177010790541056e-05,
+      "loss": 1.0611,
+      "step": 208030
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.1175862860641306e-05,
+      "loss": 1.1916,
+      "step": 208040
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.1174714930741556e-05,
+      "loss": 1.0965,
+      "step": 208050
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.117356700084181e-05,
+      "loss": 1.1342,
+      "step": 208060
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.117241907094207e-05,
+      "loss": 1.2465,
+      "step": 208070
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.117127114104231e-05,
+      "loss": 1.1168,
+      "step": 208080
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.117012321114257e-05,
+      "loss": 1.126,
+      "step": 208090
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.1168975281242825e-05,
+      "loss": 1.1092,
+      "step": 208100
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.116782735134307e-05,
+      "loss": 1.2347,
+      "step": 208110
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.1166679421443325e-05,
+      "loss": 1.0904,
+      "step": 208120
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.116553149154358e-05,
+      "loss": 0.9819,
+      "step": 208130
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.116438356164383e-05,
+      "loss": 1.0917,
+      "step": 208140
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.116323563174408e-05,
+      "loss": 1.1543,
+      "step": 208150
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.116208770184434e-05,
+      "loss": 1.2165,
+      "step": 208160
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.116093977194459e-05,
+      "loss": 1.1887,
+      "step": 208170
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.1159791842044844e-05,
+      "loss": 1.0633,
+      "step": 208180
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.1158643912145094e-05,
+      "loss": 1.1053,
+      "step": 208190
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.1157495982245344e-05,
+      "loss": 1.1821,
+      "step": 208200
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.11563480523456e-05,
+      "loss": 1.1172,
+      "step": 208210
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.115520012244586e-05,
+      "loss": 1.2906,
+      "step": 208220
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.11540521925461e-05,
+      "loss": 1.1692,
+      "step": 208230
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.115290426264636e-05,
+      "loss": 1.1127,
+      "step": 208240
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.1151756332746614e-05,
+      "loss": 1.1168,
+      "step": 208250
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.115060840284686e-05,
+      "loss": 1.1697,
+      "step": 208260
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.1149460472947113e-05,
+      "loss": 1.1252,
+      "step": 208270
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.114831254304737e-05,
+      "loss": 1.0788,
+      "step": 208280
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.114716461314762e-05,
+      "loss": 1.2096,
+      "step": 208290
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.114601668324787e-05,
+      "loss": 1.0953,
+      "step": 208300
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.1144868753348126e-05,
+      "loss": 1.0705,
+      "step": 208310
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.1143720823448376e-05,
+      "loss": 1.1826,
+      "step": 208320
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.114257289354863e-05,
+      "loss": 1.1222,
+      "step": 208330
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.114142496364888e-05,
+      "loss": 1.1281,
+      "step": 208340
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.114027703374913e-05,
+      "loss": 1.1674,
+      "step": 208350
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.113912910384939e-05,
+      "loss": 1.0489,
+      "step": 208360
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.1137981173949646e-05,
+      "loss": 0.9473,
+      "step": 208370
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.113683324404989e-05,
+      "loss": 1.0902,
+      "step": 208380
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.1135685314150146e-05,
+      "loss": 1.078,
+      "step": 208390
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.11345373842504e-05,
+      "loss": 1.0643,
+      "step": 208400
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.1133389454350645e-05,
+      "loss": 1.1394,
+      "step": 208410
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.11322415244509e-05,
+      "loss": 1.0564,
+      "step": 208420
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.113109359455116e-05,
+      "loss": 1.1172,
+      "step": 208430
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.112994566465141e-05,
+      "loss": 1.0169,
+      "step": 208440
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.112879773475166e-05,
+      "loss": 1.1805,
+      "step": 208450
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.112764980485191e-05,
+      "loss": 1.1085,
+      "step": 208460
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.1126501874952165e-05,
+      "loss": 1.0412,
+      "step": 208470
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.112535394505242e-05,
+      "loss": 0.9227,
+      "step": 208480
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.1124206015152664e-05,
+      "loss": 1.2366,
+      "step": 208490
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.112305808525292e-05,
+      "loss": 1.1893,
+      "step": 208500
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.112191015535318e-05,
+      "loss": 1.2202,
+      "step": 208510
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.112076222545342e-05,
+      "loss": 1.1314,
+      "step": 208520
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.111961429555368e-05,
+      "loss": 1.1152,
+      "step": 208530
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.1118466365653934e-05,
+      "loss": 1.1531,
+      "step": 208540
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.1117318435754184e-05,
+      "loss": 1.0524,
+      "step": 208550
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.1116170505854434e-05,
+      "loss": 1.106,
+      "step": 208560
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.111502257595469e-05,
+      "loss": 1.1321,
+      "step": 208570
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.111387464605494e-05,
+      "loss": 1.0726,
+      "step": 208580
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.11127267161552e-05,
+      "loss": 1.1226,
+      "step": 208590
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.111157878625545e-05,
+      "loss": 1.1804,
+      "step": 208600
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.11104308563557e-05,
+      "loss": 1.0111,
+      "step": 208610
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.110928292645595e-05,
+      "loss": 1.0821,
+      "step": 208620
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.110813499655621e-05,
+      "loss": 1.0769,
+      "step": 208630
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.110698706665645e-05,
+      "loss": 1.1092,
+      "step": 208640
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.110583913675671e-05,
+      "loss": 1.0287,
+      "step": 208650
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.1104691206856966e-05,
+      "loss": 1.1565,
+      "step": 208660
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.110354327695721e-05,
+      "loss": 1.1158,
+      "step": 208670
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.1102395347057466e-05,
+      "loss": 1.139,
+      "step": 208680
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.110124741715772e-05,
+      "loss": 1.2003,
+      "step": 208690
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.110009948725797e-05,
+      "loss": 1.0724,
+      "step": 208700
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.109895155735822e-05,
+      "loss": 1.1556,
+      "step": 208710
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.109780362745848e-05,
+      "loss": 1.3014,
+      "step": 208720
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.109665569755873e-05,
+      "loss": 1.1609,
+      "step": 208730
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.1095507767658985e-05,
+      "loss": 1.1326,
+      "step": 208740
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.1094359837759235e-05,
+      "loss": 1.0898,
+      "step": 208750
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.1093211907859485e-05,
+      "loss": 1.224,
+      "step": 208760
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.109206397795974e-05,
+      "loss": 1.0763,
+      "step": 208770
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.109091604806e-05,
+      "loss": 1.0335,
+      "step": 208780
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.108976811816024e-05,
+      "loss": 1.1289,
+      "step": 208790
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.10886201882605e-05,
+      "loss": 1.2112,
+      "step": 208800
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.1087472258360755e-05,
+      "loss": 1.0375,
+      "step": 208810
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.1086324328461e-05,
+      "loss": 1.2443,
+      "step": 208820
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.1085176398561254e-05,
+      "loss": 1.1082,
+      "step": 208830
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.108402846866151e-05,
+      "loss": 1.1442,
+      "step": 208840
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.108288053876176e-05,
+      "loss": 1.0433,
+      "step": 208850
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.108173260886201e-05,
+      "loss": 1.3057,
+      "step": 208860
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.108058467896227e-05,
+      "loss": 0.9645,
+      "step": 208870
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.107943674906252e-05,
+      "loss": 1.1236,
+      "step": 208880
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.1078288819162774e-05,
+      "loss": 1.0118,
+      "step": 208890
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.1077140889263024e-05,
+      "loss": 1.1154,
+      "step": 208900
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.1075992959363274e-05,
+      "loss": 1.1851,
+      "step": 208910
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.107484502946353e-05,
+      "loss": 1.139,
+      "step": 208920
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.107369709956379e-05,
+      "loss": 0.9913,
+      "step": 208930
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.107254916966403e-05,
+      "loss": 0.982,
+      "step": 208940
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.1071401239764287e-05,
+      "loss": 1.0889,
+      "step": 208950
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.107025330986454e-05,
+      "loss": 1.1721,
+      "step": 208960
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.1069105379964786e-05,
+      "loss": 1.1317,
+      "step": 208970
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.106795745006504e-05,
+      "loss": 1.0678,
+      "step": 208980
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.10668095201653e-05,
+      "loss": 1.1504,
+      "step": 208990
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.106566159026555e-05,
+      "loss": 1.1181,
+      "step": 209000
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.10645136603658e-05,
+      "loss": 1.026,
+      "step": 209010
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.1063365730466056e-05,
+      "loss": 1.053,
+      "step": 209020
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.1062217800566306e-05,
+      "loss": 1.0249,
+      "step": 209030
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.106106987066656e-05,
+      "loss": 1.1374,
+      "step": 209040
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.105992194076681e-05,
+      "loss": 1.0412,
+      "step": 209050
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.105877401086706e-05,
+      "loss": 1.1156,
+      "step": 209060
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.105762608096732e-05,
+      "loss": 1.1105,
+      "step": 209070
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.1056478151067575e-05,
+      "loss": 1.0388,
+      "step": 209080
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.105533022116782e-05,
+      "loss": 1.1958,
+      "step": 209090
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.1054182291268075e-05,
+      "loss": 1.0753,
+      "step": 209100
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.105303436136833e-05,
+      "loss": 1.0362,
+      "step": 209110
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.1051886431468575e-05,
+      "loss": 1.0698,
+      "step": 209120
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.105073850156883e-05,
+      "loss": 1.1524,
+      "step": 209130
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.104959057166909e-05,
+      "loss": 1.0779,
+      "step": 209140
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.104844264176934e-05,
+      "loss": 1.1615,
+      "step": 209150
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.104729471186959e-05,
+      "loss": 1.0418,
+      "step": 209160
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.1046146781969844e-05,
+      "loss": 1.1228,
+      "step": 209170
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.1044998852070094e-05,
+      "loss": 1.0801,
+      "step": 209180
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.104385092217035e-05,
+      "loss": 0.9374,
+      "step": 209190
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.10427029922706e-05,
+      "loss": 1.0518,
+      "step": 209200
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.104155506237085e-05,
+      "loss": 1.0005,
+      "step": 209210
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.104040713247111e-05,
+      "loss": 1.093,
+      "step": 209220
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.1039259202571364e-05,
+      "loss": 1.1551,
+      "step": 209230
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.103811127267161e-05,
+      "loss": 1.0694,
+      "step": 209240
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.1036963342771863e-05,
+      "loss": 1.2114,
+      "step": 209250
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.103581541287212e-05,
+      "loss": 1.2186,
+      "step": 209260
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.103466748297236e-05,
+      "loss": 1.0797,
+      "step": 209270
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.103351955307262e-05,
+      "loss": 1.226,
+      "step": 209280
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.1032371623172876e-05,
+      "loss": 1.034,
+      "step": 209290
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.1031223693273126e-05,
+      "loss": 1.1437,
+      "step": 209300
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.1030075763373376e-05,
+      "loss": 1.0877,
+      "step": 209310
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.102892783347363e-05,
+      "loss": 1.2211,
+      "step": 209320
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.102777990357388e-05,
+      "loss": 1.1009,
+      "step": 209330
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.102663197367414e-05,
+      "loss": 1.0272,
+      "step": 209340
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.102548404377439e-05,
+      "loss": 1.1338,
+      "step": 209350
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.102433611387464e-05,
+      "loss": 1.0969,
+      "step": 209360
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.1023188183974896e-05,
+      "loss": 1.1315,
+      "step": 209370
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.102204025407515e-05,
+      "loss": 1.1011,
+      "step": 209380
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.1020892324175395e-05,
+      "loss": 1.1055,
+      "step": 209390
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.101974439427565e-05,
+      "loss": 1.0547,
+      "step": 209400
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.101859646437591e-05,
+      "loss": 1.1232,
+      "step": 209410
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.101744853447615e-05,
+      "loss": 1.2199,
+      "step": 209420
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.101630060457641e-05,
+      "loss": 1.1936,
+      "step": 209430
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.1015152674676665e-05,
+      "loss": 1.0494,
+      "step": 209440
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.1014004744776915e-05,
+      "loss": 1.2354,
+      "step": 209450
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.1012856814877165e-05,
+      "loss": 1.1525,
+      "step": 209460
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.101170888497742e-05,
+      "loss": 1.0408,
+      "step": 209470
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.101056095507767e-05,
+      "loss": 1.1068,
+      "step": 209480
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.100941302517793e-05,
+      "loss": 1.2116,
+      "step": 209490
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.100826509527818e-05,
+      "loss": 1.2395,
+      "step": 209500
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.100711716537843e-05,
+      "loss": 1.0799,
+      "step": 209510
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.1005969235478684e-05,
+      "loss": 0.9255,
+      "step": 209520
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.100482130557894e-05,
+      "loss": 1.0575,
+      "step": 209530
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.1003673375679184e-05,
+      "loss": 1.1654,
+      "step": 209540
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.100252544577944e-05,
+      "loss": 1.0813,
+      "step": 209550
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.10013775158797e-05,
+      "loss": 1.031,
+      "step": 209560
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.100022958597994e-05,
+      "loss": 1.0493,
+      "step": 209570
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.09990816560802e-05,
+      "loss": 1.0867,
+      "step": 209580
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.0997933726180453e-05,
+      "loss": 0.9919,
+      "step": 209590
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.09967857962807e-05,
+      "loss": 1.149,
+      "step": 209600
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.099563786638095e-05,
+      "loss": 1.0485,
+      "step": 209610
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.099448993648121e-05,
+      "loss": 1.1274,
+      "step": 209620
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.099334200658146e-05,
+      "loss": 1.1175,
+      "step": 209630
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.0992194076681716e-05,
+      "loss": 1.1155,
+      "step": 209640
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.0991046146781966e-05,
+      "loss": 1.0927,
+      "step": 209650
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.0989898216882216e-05,
+      "loss": 1.1726,
+      "step": 209660
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.098875028698247e-05,
+      "loss": 1.1578,
+      "step": 209670
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.098760235708273e-05,
+      "loss": 1.1679,
+      "step": 209680
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.098645442718297e-05,
+      "loss": 1.1866,
+      "step": 209690
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.098530649728323e-05,
+      "loss": 1.1785,
+      "step": 209700
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.0984158567383486e-05,
+      "loss": 1.212,
+      "step": 209710
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.098301063748373e-05,
+      "loss": 1.1443,
+      "step": 209720
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.0981862707583985e-05,
+      "loss": 1.1167,
+      "step": 209730
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.098071477768424e-05,
+      "loss": 1.1503,
+      "step": 209740
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.097956684778449e-05,
+      "loss": 1.1987,
+      "step": 209750
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.097841891788474e-05,
+      "loss": 1.0063,
+      "step": 209760
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.0977270987985e-05,
+      "loss": 1.1043,
+      "step": 209770
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.097612305808525e-05,
+      "loss": 1.0494,
+      "step": 209780
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.0974975128185505e-05,
+      "loss": 1.0995,
+      "step": 209790
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.0973827198285755e-05,
+      "loss": 1.0855,
+      "step": 209800
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.0972679268386004e-05,
+      "loss": 1.1116,
+      "step": 209810
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.097153133848626e-05,
+      "loss": 1.0545,
+      "step": 209820
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.0970383408586504e-05,
+      "loss": 0.9688,
+      "step": 209830
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.096923547868676e-05,
+      "loss": 1.0525,
+      "step": 209840
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.096808754878702e-05,
+      "loss": 1.1025,
+      "step": 209850
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.096693961888727e-05,
+      "loss": 1.1171,
+      "step": 209860
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.096579168898752e-05,
+      "loss": 1.1631,
+      "step": 209870
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.0964643759087774e-05,
+      "loss": 1.1172,
+      "step": 209880
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.0963495829188024e-05,
+      "loss": 1.0933,
+      "step": 209890
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.096234789928828e-05,
+      "loss": 1.125,
+      "step": 209900
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.096119996938853e-05,
+      "loss": 1.1064,
+      "step": 209910
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.096005203948878e-05,
+      "loss": 1.2232,
+      "step": 209920
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.0958904109589037e-05,
+      "loss": 1.2158,
+      "step": 209930
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.095775617968929e-05,
+      "loss": 1.0755,
+      "step": 209940
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.0956608249789536e-05,
+      "loss": 1.2008,
+      "step": 209950
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.095546031988979e-05,
+      "loss": 1.0773,
+      "step": 209960
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.095431238999005e-05,
+      "loss": 1.0723,
+      "step": 209970
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.095316446009029e-05,
+      "loss": 1.0298,
+      "step": 209980
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.095201653019055e-05,
+      "loss": 1.12,
+      "step": 209990
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.0950868600290806e-05,
+      "loss": 1.2705,
+      "step": 210000
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.0949720670391056e-05,
+      "loss": 1.1371,
+      "step": 210010
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.0948572740491306e-05,
+      "loss": 1.0734,
+      "step": 210020
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.094742481059156e-05,
+      "loss": 1.0795,
+      "step": 210030
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.094627688069181e-05,
+      "loss": 1.0914,
+      "step": 210040
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.094512895079207e-05,
+      "loss": 1.1581,
+      "step": 210050
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.094398102089232e-05,
+      "loss": 1.1521,
+      "step": 210060
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.094283309099257e-05,
+      "loss": 1.1842,
+      "step": 210070
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.0941685161092825e-05,
+      "loss": 1.2227,
+      "step": 210080
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.094053723119308e-05,
+      "loss": 1.0736,
+      "step": 210090
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.0939389301293325e-05,
+      "loss": 1.1715,
+      "step": 210100
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.093824137139358e-05,
+      "loss": 1.1474,
+      "step": 210110
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.093709344149384e-05,
+      "loss": 1.117,
+      "step": 210120
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.093594551159408e-05,
+      "loss": 1.0686,
+      "step": 210130
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.093479758169434e-05,
+      "loss": 1.0109,
+      "step": 210140
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.0933649651794594e-05,
+      "loss": 1.1695,
+      "step": 210150
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.0932501721894844e-05,
+      "loss": 1.0406,
+      "step": 210160
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.0931353791995094e-05,
+      "loss": 1.1192,
+      "step": 210170
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.093020586209535e-05,
+      "loss": 1.0393,
+      "step": 210180
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.09290579321956e-05,
+      "loss": 1.0669,
+      "step": 210190
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.092791000229586e-05,
+      "loss": 1.235,
+      "step": 210200
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.092676207239611e-05,
+      "loss": 1.1476,
+      "step": 210210
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.092561414249636e-05,
+      "loss": 1.0056,
+      "step": 210220
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.0924466212596613e-05,
+      "loss": 0.9416,
+      "step": 210230
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.092331828269687e-05,
+      "loss": 1.0211,
+      "step": 210240
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.092217035279711e-05,
+      "loss": 1.1866,
+      "step": 210250
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.092102242289737e-05,
+      "loss": 1.0556,
+      "step": 210260
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.0919874492997626e-05,
+      "loss": 1.0748,
+      "step": 210270
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.091872656309787e-05,
+      "loss": 1.043,
+      "step": 210280
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.0917578633198126e-05,
+      "loss": 1.1769,
+      "step": 210290
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.091643070329838e-05,
+      "loss": 1.1208,
+      "step": 210300
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.091528277339863e-05,
+      "loss": 1.0295,
+      "step": 210310
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.091413484349888e-05,
+      "loss": 1.1093,
+      "step": 210320
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.091298691359914e-05,
+      "loss": 1.1362,
+      "step": 210330
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.091183898369939e-05,
+      "loss": 1.0783,
+      "step": 210340
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.0910691053799646e-05,
+      "loss": 1.1351,
+      "step": 210350
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.0909543123899895e-05,
+      "loss": 1.0961,
+      "step": 210360
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.0908395194000145e-05,
+      "loss": 1.2001,
+      "step": 210370
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.09072472641004e-05,
+      "loss": 1.0222,
+      "step": 210380
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.090609933420066e-05,
+      "loss": 1.0588,
+      "step": 210390
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.09049514043009e-05,
+      "loss": 1.1652,
+      "step": 210400
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.090380347440116e-05,
+      "loss": 1.1362,
+      "step": 210410
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.0902655544501415e-05,
+      "loss": 1.1025,
+      "step": 210420
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.090150761460166e-05,
+      "loss": 1.1257,
+      "step": 210430
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.0900359684701915e-05,
+      "loss": 1.1101,
+      "step": 210440
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.089921175480217e-05,
+      "loss": 1.0985,
+      "step": 210450
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.089806382490242e-05,
+      "loss": 1.0895,
+      "step": 210460
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.089691589500267e-05,
+      "loss": 1.1182,
+      "step": 210470
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.089576796510293e-05,
+      "loss": 1.0569,
+      "step": 210480
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.089462003520318e-05,
+      "loss": 1.1658,
+      "step": 210490
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.0893472105303434e-05,
+      "loss": 1.1644,
+      "step": 210500
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.0892324175403684e-05,
+      "loss": 1.0089,
+      "step": 210510
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.0891176245503934e-05,
+      "loss": 1.0776,
+      "step": 210520
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.089002831560419e-05,
+      "loss": 1.1051,
+      "step": 210530
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.088888038570445e-05,
+      "loss": 1.1405,
+      "step": 210540
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.088773245580469e-05,
+      "loss": 1.1229,
+      "step": 210550
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.088658452590495e-05,
+      "loss": 1.166,
+      "step": 210560
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.0885436596005203e-05,
+      "loss": 1.2049,
+      "step": 210570
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.0884288666105446e-05,
+      "loss": 1.1955,
+      "step": 210580
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.08831407362057e-05,
+      "loss": 1.1484,
+      "step": 210590
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.088199280630596e-05,
+      "loss": 1.1762,
+      "step": 210600
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.088084487640621e-05,
+      "loss": 1.081,
+      "step": 210610
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.087969694650646e-05,
+      "loss": 1.1968,
+      "step": 210620
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.0878549016606716e-05,
+      "loss": 1.1173,
+      "step": 210630
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.0877401086706966e-05,
+      "loss": 1.1731,
+      "step": 210640
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.087625315680722e-05,
+      "loss": 1.1056,
+      "step": 210650
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.087510522690747e-05,
+      "loss": 1.0895,
+      "step": 210660
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.087395729700772e-05,
+      "loss": 1.177,
+      "step": 210670
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.087280936710798e-05,
+      "loss": 1.0198,
+      "step": 210680
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.0871661437208236e-05,
+      "loss": 1.0466,
+      "step": 210690
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.087051350730848e-05,
+      "loss": 1.1501,
+      "step": 210700
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.0869365577408735e-05,
+      "loss": 1.082,
+      "step": 210710
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.086821764750899e-05,
+      "loss": 1.1263,
+      "step": 210720
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.0867069717609235e-05,
+      "loss": 1.1212,
+      "step": 210730
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.086592178770949e-05,
+      "loss": 1.099,
+      "step": 210740
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.086477385780975e-05,
+      "loss": 1.0496,
+      "step": 210750
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.086362592791e-05,
+      "loss": 1.1047,
+      "step": 210760
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.086247799801025e-05,
+      "loss": 1.0842,
+      "step": 210770
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.0861330068110505e-05,
+      "loss": 1.102,
+      "step": 210780
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.0860182138210754e-05,
+      "loss": 1.0366,
+      "step": 210790
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.085903420831101e-05,
+      "loss": 1.0636,
+      "step": 210800
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.085788627841126e-05,
+      "loss": 1.0567,
+      "step": 210810
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.085673834851151e-05,
+      "loss": 1.16,
+      "step": 210820
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.085559041861177e-05,
+      "loss": 1.0786,
+      "step": 210830
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.0854442488712024e-05,
+      "loss": 1.0542,
+      "step": 210840
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.085329455881227e-05,
+      "loss": 1.0397,
+      "step": 210850
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.0852146628912524e-05,
+      "loss": 1.2057,
+      "step": 210860
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.085099869901278e-05,
+      "loss": 1.1104,
+      "step": 210870
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.0849850769113023e-05,
+      "loss": 1.1143,
+      "step": 210880
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.084870283921328e-05,
+      "loss": 1.206,
+      "step": 210890
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.084755490931354e-05,
+      "loss": 1.1355,
+      "step": 210900
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.0846406979413787e-05,
+      "loss": 1.1847,
+      "step": 210910
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.0845259049514036e-05,
+      "loss": 1.0452,
+      "step": 210920
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.084411111961429e-05,
+      "loss": 1.1117,
+      "step": 210930
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.084296318971454e-05,
+      "loss": 1.1747,
+      "step": 210940
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.08418152598148e-05,
+      "loss": 1.0826,
+      "step": 210950
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.084066732991505e-05,
+      "loss": 1.124,
+      "step": 210960
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.08395194000153e-05,
+      "loss": 1.118,
+      "step": 210970
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.0838371470115556e-05,
+      "loss": 1.1386,
+      "step": 210980
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.083722354021581e-05,
+      "loss": 1.1454,
+      "step": 210990
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.0836075610316056e-05,
+      "loss": 1.1051,
+      "step": 211000
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.083492768041631e-05,
+      "loss": 1.09,
+      "step": 211010
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.083377975051657e-05,
+      "loss": 1.0527,
+      "step": 211020
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.083263182061681e-05,
+      "loss": 1.1607,
+      "step": 211030
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.083148389071707e-05,
+      "loss": 1.1595,
+      "step": 211040
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.0830335960817325e-05,
+      "loss": 1.1642,
+      "step": 211050
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.0829188030917575e-05,
+      "loss": 1.1156,
+      "step": 211060
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.0828040101017825e-05,
+      "loss": 1.1576,
+      "step": 211070
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.082689217111808e-05,
+      "loss": 1.1202,
+      "step": 211080
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.082574424121833e-05,
+      "loss": 1.1596,
+      "step": 211090
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.082459631131859e-05,
+      "loss": 1.1188,
+      "step": 211100
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.082344838141884e-05,
+      "loss": 1.2164,
+      "step": 211110
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.082230045151909e-05,
+      "loss": 1.053,
+      "step": 211120
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.0821152521619344e-05,
+      "loss": 0.9637,
+      "step": 211130
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.08200045917196e-05,
+      "loss": 1.1346,
+      "step": 211140
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.0818856661819844e-05,
+      "loss": 1.2535,
+      "step": 211150
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.08177087319201e-05,
+      "loss": 1.2185,
+      "step": 211160
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.081656080202036e-05,
+      "loss": 1.0213,
+      "step": 211170
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.08154128721206e-05,
+      "loss": 1.157,
+      "step": 211180
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.081426494222086e-05,
+      "loss": 1.0652,
+      "step": 211190
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.081311701232111e-05,
+      "loss": 1.1322,
+      "step": 211200
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.0811969082421363e-05,
+      "loss": 0.9717,
+      "step": 211210
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.081082115252161e-05,
+      "loss": 1.1811,
+      "step": 211220
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.080967322262186e-05,
+      "loss": 1.17,
+      "step": 211230
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.080852529272212e-05,
+      "loss": 1.1705,
+      "step": 211240
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.0807377362822376e-05,
+      "loss": 1.1651,
+      "step": 211250
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.080622943292262e-05,
+      "loss": 1.0793,
+      "step": 211260
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.0805081503022876e-05,
+      "loss": 1.0101,
+      "step": 211270
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.080393357312313e-05,
+      "loss": 1.0374,
+      "step": 211280
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.0802785643223376e-05,
+      "loss": 1.1542,
+      "step": 211290
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.080163771332363e-05,
+      "loss": 1.1336,
+      "step": 211300
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.080048978342389e-05,
+      "loss": 0.9399,
+      "step": 211310
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.079934185352414e-05,
+      "loss": 1.0608,
+      "step": 211320
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.079819392362439e-05,
+      "loss": 1.2101,
+      "step": 211330
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.0797045993724645e-05,
+      "loss": 1.1199,
+      "step": 211340
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.0795898063824895e-05,
+      "loss": 1.0917,
+      "step": 211350
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.079475013392515e-05,
+      "loss": 1.0946,
+      "step": 211360
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.07936022040254e-05,
+      "loss": 1.019,
+      "step": 211370
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.079245427412565e-05,
+      "loss": 1.1606,
+      "step": 211380
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.079130634422591e-05,
+      "loss": 0.9898,
+      "step": 211390
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.0790158414326165e-05,
+      "loss": 1.1032,
+      "step": 211400
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.078901048442641e-05,
+      "loss": 1.2301,
+      "step": 211410
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.0787862554526665e-05,
+      "loss": 1.0517,
+      "step": 211420
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.078671462462692e-05,
+      "loss": 1.076,
+      "step": 211430
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.0785566694727164e-05,
+      "loss": 1.1504,
+      "step": 211440
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.078441876482742e-05,
+      "loss": 1.0664,
+      "step": 211450
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.078327083492768e-05,
+      "loss": 1.0669,
+      "step": 211460
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.078212290502793e-05,
+      "loss": 1.1448,
+      "step": 211470
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.078097497512818e-05,
+      "loss": 1.0608,
+      "step": 211480
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.0779827045228434e-05,
+      "loss": 1.1622,
+      "step": 211490
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.0778679115328684e-05,
+      "loss": 1.0904,
+      "step": 211500
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.077753118542894e-05,
+      "loss": 1.0372,
+      "step": 211510
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.077638325552919e-05,
+      "loss": 1.0922,
+      "step": 211520
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.077523532562944e-05,
+      "loss": 1.1868,
+      "step": 211530
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.07740873957297e-05,
+      "loss": 1.0941,
+      "step": 211540
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.0772939465829953e-05,
+      "loss": 1.0412,
+      "step": 211550
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.0771791535930196e-05,
+      "loss": 1.1376,
+      "step": 211560
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.077064360603045e-05,
+      "loss": 1.0893,
+      "step": 211570
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.076949567613071e-05,
+      "loss": 1.2247,
+      "step": 211580
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.076834774623095e-05,
+      "loss": 1.2144,
+      "step": 211590
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.076719981633121e-05,
+      "loss": 1.0069,
+      "step": 211600
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.0766051886431466e-05,
+      "loss": 1.0063,
+      "step": 211610
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.0764903956531716e-05,
+      "loss": 1.1416,
+      "step": 211620
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.0763756026631966e-05,
+      "loss": 1.2714,
+      "step": 211630
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.076260809673222e-05,
+      "loss": 1.1097,
+      "step": 211640
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.076146016683247e-05,
+      "loss": 1.1061,
+      "step": 211650
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.076031223693273e-05,
+      "loss": 1.1624,
+      "step": 211660
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.075916430703298e-05,
+      "loss": 1.0658,
+      "step": 211670
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.075801637713323e-05,
+      "loss": 1.0446,
+      "step": 211680
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.0756868447233485e-05,
+      "loss": 1.0534,
+      "step": 211690
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.075572051733374e-05,
+      "loss": 1.3344,
+      "step": 211700
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.0754572587433985e-05,
+      "loss": 1.0417,
+      "step": 211710
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.075342465753424e-05,
+      "loss": 1.0722,
+      "step": 211720
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.07522767276345e-05,
+      "loss": 1.0189,
+      "step": 211730
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.075112879773474e-05,
+      "loss": 1.1864,
+      "step": 211740
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.0749980867835e-05,
+      "loss": 1.2129,
+      "step": 211750
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.0748832937935255e-05,
+      "loss": 1.0595,
+      "step": 211760
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.0747685008035504e-05,
+      "loss": 1.1923,
+      "step": 211770
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.0746537078135754e-05,
+      "loss": 1.1724,
+      "step": 211780
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.074538914823601e-05,
+      "loss": 1.11,
+      "step": 211790
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.074424121833626e-05,
+      "loss": 1.0848,
+      "step": 211800
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.074309328843652e-05,
+      "loss": 1.1099,
+      "step": 211810
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.074194535853677e-05,
+      "loss": 1.1353,
+      "step": 211820
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.074079742863702e-05,
+      "loss": 1.0395,
+      "step": 211830
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.0739649498737274e-05,
+      "loss": 1.1375,
+      "step": 211840
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.073850156883753e-05,
+      "loss": 1.363,
+      "step": 211850
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.0737353638937773e-05,
+      "loss": 1.0934,
+      "step": 211860
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.073620570903803e-05,
+      "loss": 1.073,
+      "step": 211870
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.073505777913829e-05,
+      "loss": 1.0581,
+      "step": 211880
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.073390984923853e-05,
+      "loss": 1.1515,
+      "step": 211890
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.0732761919338786e-05,
+      "loss": 1.0613,
+      "step": 211900
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.073161398943904e-05,
+      "loss": 1.0791,
+      "step": 211910
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.073046605953929e-05,
+      "loss": 1.0192,
+      "step": 211920
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.072931812963954e-05,
+      "loss": 1.0984,
+      "step": 211930
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.07281701997398e-05,
+      "loss": 0.9914,
+      "step": 211940
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.072702226984005e-05,
+      "loss": 1.2083,
+      "step": 211950
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.0725874339940306e-05,
+      "loss": 1.1119,
+      "step": 211960
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.0724726410040556e-05,
+      "loss": 1.2258,
+      "step": 211970
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.0723578480140806e-05,
+      "loss": 1.1551,
+      "step": 211980
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.072243055024106e-05,
+      "loss": 1.1057,
+      "step": 211990
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.072128262034132e-05,
+      "loss": 1.129,
+      "step": 212000
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.072013469044156e-05,
+      "loss": 1.0782,
+      "step": 212010
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.071898676054182e-05,
+      "loss": 1.2311,
+      "step": 212020
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.0717838830642075e-05,
+      "loss": 1.0107,
+      "step": 212030
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.071669090074232e-05,
+      "loss": 0.9566,
+      "step": 212040
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.0715542970842575e-05,
+      "loss": 0.9863,
+      "step": 212050
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.071439504094283e-05,
+      "loss": 1.0811,
+      "step": 212060
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.071324711104308e-05,
+      "loss": 1.0654,
+      "step": 212070
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.071209918114333e-05,
+      "loss": 1.1448,
+      "step": 212080
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.071095125124359e-05,
+      "loss": 1.1424,
+      "step": 212090
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.070980332134384e-05,
+      "loss": 1.0613,
+      "step": 212100
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.0708655391444094e-05,
+      "loss": 1.3247,
+      "step": 212110
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.0707507461544344e-05,
+      "loss": 1.0686,
+      "step": 212120
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.0706359531644594e-05,
+      "loss": 1.2008,
+      "step": 212130
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.070521160174485e-05,
+      "loss": 1.1826,
+      "step": 212140
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.070406367184511e-05,
+      "loss": 0.9952,
+      "step": 212150
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.070291574194535e-05,
+      "loss": 1.1047,
+      "step": 212160
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.070176781204561e-05,
+      "loss": 0.976,
+      "step": 212170
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.0700619882145864e-05,
+      "loss": 1.145,
+      "step": 212180
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.069947195224611e-05,
+      "loss": 1.0962,
+      "step": 212190
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.069832402234636e-05,
+      "loss": 1.15,
+      "step": 212200
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.069717609244662e-05,
+      "loss": 1.0106,
+      "step": 212210
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.069602816254687e-05,
+      "loss": 1.1071,
+      "step": 212220
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.069488023264712e-05,
+      "loss": 1.0524,
+      "step": 212230
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.0693732302747376e-05,
+      "loss": 1.2009,
+      "step": 212240
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.0692584372847626e-05,
+      "loss": 1.1342,
+      "step": 212250
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.069143644294788e-05,
+      "loss": 1.0989,
+      "step": 212260
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.069028851304813e-05,
+      "loss": 1.1481,
+      "step": 212270
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.068914058314838e-05,
+      "loss": 1.1007,
+      "step": 212280
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.068799265324864e-05,
+      "loss": 1.0707,
+      "step": 212290
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.0686844723348896e-05,
+      "loss": 1.0675,
+      "step": 212300
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.068569679344914e-05,
+      "loss": 1.0174,
+      "step": 212310
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.0684548863549395e-05,
+      "loss": 1.0936,
+      "step": 212320
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.068340093364965e-05,
+      "loss": 1.2637,
+      "step": 212330
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.0682253003749895e-05,
+      "loss": 1.0498,
+      "step": 212340
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.068110507385015e-05,
+      "loss": 1.0662,
+      "step": 212350
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.067995714395041e-05,
+      "loss": 1.1454,
+      "step": 212360
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.067880921405066e-05,
+      "loss": 1.017,
+      "step": 212370
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.067766128415091e-05,
+      "loss": 1.1887,
+      "step": 212380
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.0676513354251165e-05,
+      "loss": 1.0902,
+      "step": 212390
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.0675365424351415e-05,
+      "loss": 1.0888,
+      "step": 212400
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.067421749445167e-05,
+      "loss": 1.1196,
+      "step": 212410
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.067306956455192e-05,
+      "loss": 1.0102,
+      "step": 212420
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.067192163465217e-05,
+      "loss": 1.1269,
+      "step": 212430
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.067077370475243e-05,
+      "loss": 1.051,
+      "step": 212440
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.0669625774852684e-05,
+      "loss": 1.1927,
+      "step": 212450
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.066847784495293e-05,
+      "loss": 1.1625,
+      "step": 212460
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.0667329915053184e-05,
+      "loss": 1.1343,
+      "step": 212470
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.066618198515344e-05,
+      "loss": 1.0389,
+      "step": 212480
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.0665034055253684e-05,
+      "loss": 1.1129,
+      "step": 212490
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.066388612535394e-05,
+      "loss": 1.076,
+      "step": 212500
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.06627381954542e-05,
+      "loss": 1.1445,
+      "step": 212510
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.066159026555445e-05,
+      "loss": 1.0363,
+      "step": 212520
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.0660442335654697e-05,
+      "loss": 1.1929,
+      "step": 212530
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.065929440575495e-05,
+      "loss": 1.0717,
+      "step": 212540
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.06581464758552e-05,
+      "loss": 1.19,
+      "step": 212550
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.065699854595546e-05,
+      "loss": 1.0595,
+      "step": 212560
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.06558506160557e-05,
+      "loss": 1.0955,
+      "step": 212570
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.065470268615596e-05,
+      "loss": 1.1683,
+      "step": 212580
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.0653554756256216e-05,
+      "loss": 1.1977,
+      "step": 212590
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.065240682635646e-05,
+      "loss": 1.1793,
+      "step": 212600
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.0651258896456716e-05,
+      "loss": 1.0952,
+      "step": 212610
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.065011096655697e-05,
+      "loss": 1.0064,
+      "step": 212620
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.064896303665722e-05,
+      "loss": 1.0342,
+      "step": 212630
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.064781510675747e-05,
+      "loss": 1.1996,
+      "step": 212640
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.064666717685773e-05,
+      "loss": 1.1357,
+      "step": 212650
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.064551924695798e-05,
+      "loss": 1.11,
+      "step": 212660
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.0644371317058235e-05,
+      "loss": 1.0643,
+      "step": 212670
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.0643223387158485e-05,
+      "loss": 1.1146,
+      "step": 212680
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.0642075457258735e-05,
+      "loss": 1.1632,
+      "step": 212690
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.064092752735899e-05,
+      "loss": 1.0092,
+      "step": 212700
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.063977959745925e-05,
+      "loss": 1.1317,
+      "step": 212710
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.063863166755949e-05,
+      "loss": 1.12,
+      "step": 212720
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.063748373765975e-05,
+      "loss": 1.0079,
+      "step": 212730
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.0636335807760005e-05,
+      "loss": 1.0269,
+      "step": 212740
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.063518787786025e-05,
+      "loss": 1.0015,
+      "step": 212750
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.0634039947960504e-05,
+      "loss": 1.1145,
+      "step": 212760
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.063289201806076e-05,
+      "loss": 1.1596,
+      "step": 212770
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.063174408816101e-05,
+      "loss": 1.1046,
+      "step": 212780
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.063059615826126e-05,
+      "loss": 1.2518,
+      "step": 212790
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.062944822836152e-05,
+      "loss": 1.0914,
+      "step": 212800
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.062830029846177e-05,
+      "loss": 1.0764,
+      "step": 212810
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.0627152368562024e-05,
+      "loss": 0.9954,
+      "step": 212820
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.0626004438662274e-05,
+      "loss": 1.0439,
+      "step": 212830
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.0624856508762523e-05,
+      "loss": 1.1228,
+      "step": 212840
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.062370857886278e-05,
+      "loss": 1.1739,
+      "step": 212850
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.062256064896304e-05,
+      "loss": 1.1516,
+      "step": 212860
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.062141271906328e-05,
+      "loss": 1.1801,
+      "step": 212870
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.0620264789163536e-05,
+      "loss": 1.2233,
+      "step": 212880
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.061911685926379e-05,
+      "loss": 1.1339,
+      "step": 212890
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.0617968929364036e-05,
+      "loss": 1.0565,
+      "step": 212900
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.061682099946429e-05,
+      "loss": 1.0083,
+      "step": 212910
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.061567306956455e-05,
+      "loss": 1.1699,
+      "step": 212920
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.06145251396648e-05,
+      "loss": 1.0854,
+      "step": 212930
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.061337720976505e-05,
+      "loss": 1.1152,
+      "step": 212940
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.0612229279865306e-05,
+      "loss": 1.16,
+      "step": 212950
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.0611081349965556e-05,
+      "loss": 1.1392,
+      "step": 212960
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.060993342006581e-05,
+      "loss": 1.1492,
+      "step": 212970
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.060878549016606e-05,
+      "loss": 0.9563,
+      "step": 212980
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.060763756026631e-05,
+      "loss": 1.0267,
+      "step": 212990
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.060648963036657e-05,
+      "loss": 1.0585,
+      "step": 213000
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.0605341700466825e-05,
+      "loss": 1.1417,
+      "step": 213010
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.060419377056707e-05,
+      "loss": 0.8928,
+      "step": 213020
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.0603045840667325e-05,
+      "loss": 1.2793,
+      "step": 213030
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.060189791076758e-05,
+      "loss": 1.1118,
+      "step": 213040
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.0600749980867825e-05,
+      "loss": 1.2141,
+      "step": 213050
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.059960205096808e-05,
+      "loss": 1.0546,
+      "step": 213060
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.059845412106834e-05,
+      "loss": 1.0847,
+      "step": 213070
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.059730619116859e-05,
+      "loss": 1.0684,
+      "step": 213080
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.059615826126884e-05,
+      "loss": 1.1135,
+      "step": 213090
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.0595010331369094e-05,
+      "loss": 1.0573,
+      "step": 213100
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.0593862401469344e-05,
+      "loss": 1.0709,
+      "step": 213110
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.05927144715696e-05,
+      "loss": 1.1331,
+      "step": 213120
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.059156654166985e-05,
+      "loss": 1.1654,
+      "step": 213130
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.05904186117701e-05,
+      "loss": 1.2127,
+      "step": 213140
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.058927068187036e-05,
+      "loss": 0.9861,
+      "step": 213150
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.0588122751970614e-05,
+      "loss": 1.1567,
+      "step": 213160
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.058697482207086e-05,
+      "loss": 1.1714,
+      "step": 213170
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.058582689217111e-05,
+      "loss": 1.0342,
+      "step": 213180
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.058467896227137e-05,
+      "loss": 1.1299,
+      "step": 213190
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.058353103237161e-05,
+      "loss": 1.1305,
+      "step": 213200
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.058238310247187e-05,
+      "loss": 1.0976,
+      "step": 213210
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.0581235172572126e-05,
+      "loss": 1.0704,
+      "step": 213220
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.0580087242672376e-05,
+      "loss": 1.0654,
+      "step": 213230
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.0578939312772626e-05,
+      "loss": 1.2307,
+      "step": 213240
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.057779138287288e-05,
+      "loss": 1.0836,
+      "step": 213250
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.057664345297313e-05,
+      "loss": 1.2355,
+      "step": 213260
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.057549552307339e-05,
+      "loss": 1.2083,
+      "step": 213270
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.057434759317364e-05,
+      "loss": 1.1363,
+      "step": 213280
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.057319966327389e-05,
+      "loss": 1.0668,
+      "step": 213290
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.0572051733374145e-05,
+      "loss": 1.1612,
+      "step": 213300
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.05709038034744e-05,
+      "loss": 1.1598,
+      "step": 213310
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.0569755873574645e-05,
+      "loss": 1.0474,
+      "step": 213320
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.05686079436749e-05,
+      "loss": 1.065,
+      "step": 213330
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.056746001377516e-05,
+      "loss": 1.08,
+      "step": 213340
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.05663120838754e-05,
+      "loss": 1.0807,
+      "step": 213350
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.056516415397566e-05,
+      "loss": 1.1134,
+      "step": 213360
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.0564016224075915e-05,
+      "loss": 1.1146,
+      "step": 213370
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.0562868294176165e-05,
+      "loss": 1.0839,
+      "step": 213380
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.0561720364276414e-05,
+      "loss": 1.1691,
+      "step": 213390
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.056057243437667e-05,
+      "loss": 1.1198,
+      "step": 213400
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.055942450447692e-05,
+      "loss": 1.12,
+      "step": 213410
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.055827657457718e-05,
+      "loss": 1.0985,
+      "step": 213420
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.055712864467743e-05,
+      "loss": 1.1072,
+      "step": 213430
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.055598071477768e-05,
+      "loss": 1.1327,
+      "step": 213440
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.0554832784877934e-05,
+      "loss": 1.1212,
+      "step": 213450
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.055368485497819e-05,
+      "loss": 1.0444,
+      "step": 213460
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.0552536925078434e-05,
+      "loss": 1.1315,
+      "step": 213470
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.055138899517869e-05,
+      "loss": 1.0529,
+      "step": 213480
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.055024106527895e-05,
+      "loss": 1.1918,
+      "step": 213490
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.054909313537919e-05,
+      "loss": 1.1716,
+      "step": 213500
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.0547945205479447e-05,
+      "loss": 1.054,
+      "step": 213510
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.05467972755797e-05,
+      "loss": 1.2976,
+      "step": 213520
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.054564934567995e-05,
+      "loss": 0.9878,
+      "step": 213530
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.05445014157802e-05,
+      "loss": 1.2274,
+      "step": 213540
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.054335348588046e-05,
+      "loss": 1.0633,
+      "step": 213550
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.054220555598071e-05,
+      "loss": 0.9965,
+      "step": 213560
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.0541057626080966e-05,
+      "loss": 1.1383,
+      "step": 213570
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.0539909696181216e-05,
+      "loss": 1.1633,
+      "step": 213580
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.0538761766281466e-05,
+      "loss": 1.1319,
+      "step": 213590
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.053761383638172e-05,
+      "loss": 1.0678,
+      "step": 213600
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.053646590648198e-05,
+      "loss": 1.1502,
+      "step": 213610
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.053531797658222e-05,
+      "loss": 1.0775,
+      "step": 213620
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.053417004668248e-05,
+      "loss": 1.1867,
+      "step": 213630
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.0533022116782735e-05,
+      "loss": 1.0819,
+      "step": 213640
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.053187418688298e-05,
+      "loss": 1.0739,
+      "step": 213650
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.0530726256983235e-05,
+      "loss": 1.0953,
+      "step": 213660
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.052957832708349e-05,
+      "loss": 1.0585,
+      "step": 213670
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.052843039718374e-05,
+      "loss": 1.0942,
+      "step": 213680
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.052728246728399e-05,
+      "loss": 1.0342,
+      "step": 213690
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.052613453738425e-05,
+      "loss": 1.1309,
+      "step": 213700
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.05249866074845e-05,
+      "loss": 1.1742,
+      "step": 213710
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.0523838677584755e-05,
+      "loss": 1.1139,
+      "step": 213720
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.0522690747685004e-05,
+      "loss": 1.1114,
+      "step": 213730
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.0521542817785254e-05,
+      "loss": 1.1287,
+      "step": 213740
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.052039488788551e-05,
+      "loss": 1.118,
+      "step": 213750
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.051924695798577e-05,
+      "loss": 1.1739,
+      "step": 213760
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.051809902808601e-05,
+      "loss": 1.2683,
+      "step": 213770
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.051695109818627e-05,
+      "loss": 1.0829,
+      "step": 213780
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.0515803168286524e-05,
+      "loss": 0.9652,
+      "step": 213790
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.051465523838677e-05,
+      "loss": 1.0664,
+      "step": 213800
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.0513507308487024e-05,
+      "loss": 1.0223,
+      "step": 213810
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.051235937858728e-05,
+      "loss": 1.0398,
+      "step": 213820
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.051121144868753e-05,
+      "loss": 1.0492,
+      "step": 213830
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.051006351878778e-05,
+      "loss": 1.0781,
+      "step": 213840
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.0508915588888037e-05,
+      "loss": 1.2397,
+      "step": 213850
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.0507767658988286e-05,
+      "loss": 1.0881,
+      "step": 213860
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.050661972908854e-05,
+      "loss": 1.0955,
+      "step": 213870
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.050547179918879e-05,
+      "loss": 1.0863,
+      "step": 213880
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.050432386928904e-05,
+      "loss": 1.0044,
+      "step": 213890
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.05031759393893e-05,
+      "loss": 1.1384,
+      "step": 213900
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.0502028009489556e-05,
+      "loss": 1.1075,
+      "step": 213910
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.05008800795898e-05,
+      "loss": 1.0966,
+      "step": 213920
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.0499732149690056e-05,
+      "loss": 1.1758,
+      "step": 213930
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.0498584219790306e-05,
+      "loss": 0.9945,
+      "step": 213940
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.0497436289890555e-05,
+      "loss": 1.1191,
+      "step": 213950
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.049628835999081e-05,
+      "loss": 1.086,
+      "step": 213960
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.049514043009106e-05,
+      "loss": 1.0987,
+      "step": 213970
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.049399250019132e-05,
+      "loss": 1.0315,
+      "step": 213980
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.049284457029157e-05,
+      "loss": 1.1332,
+      "step": 213990
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.049169664039182e-05,
+      "loss": 1.0487,
+      "step": 214000
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.0490548710492075e-05,
+      "loss": 1.104,
+      "step": 214010
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.048940078059233e-05,
+      "loss": 1.0419,
+      "step": 214020
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.0488252850692575e-05,
+      "loss": 1.2055,
+      "step": 214030
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.048710492079283e-05,
+      "loss": 1.0642,
+      "step": 214040
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.048595699089309e-05,
+      "loss": 1.1166,
+      "step": 214050
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.048480906099333e-05,
+      "loss": 1.0949,
+      "step": 214060
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.048366113109359e-05,
+      "loss": 1.0926,
+      "step": 214070
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.0482513201193844e-05,
+      "loss": 1.1628,
+      "step": 214080
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.0481365271294094e-05,
+      "loss": 1.1203,
+      "step": 214090
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.0480217341394344e-05,
+      "loss": 1.0388,
+      "step": 214100
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.04790694114946e-05,
+      "loss": 1.0539,
+      "step": 214110
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.047792148159485e-05,
+      "loss": 1.0986,
+      "step": 214120
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.047677355169511e-05,
+      "loss": 1.1973,
+      "step": 214130
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.047562562179536e-05,
+      "loss": 1.1669,
+      "step": 214140
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.047447769189561e-05,
+      "loss": 1.113,
+      "step": 214150
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.047332976199586e-05,
+      "loss": 0.9995,
+      "step": 214160
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.047218183209612e-05,
+      "loss": 1.0832,
+      "step": 214170
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.047103390219636e-05,
+      "loss": 1.1395,
+      "step": 214180
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.046988597229662e-05,
+      "loss": 1.0861,
+      "step": 214190
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.0468738042396876e-05,
+      "loss": 1.0854,
+      "step": 214200
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.046759011249712e-05,
+      "loss": 1.1658,
+      "step": 214210
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.0466442182597376e-05,
+      "loss": 1.1063,
+      "step": 214220
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.046529425269763e-05,
+      "loss": 1.128,
+      "step": 214230
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.046414632279788e-05,
+      "loss": 1.1471,
+      "step": 214240
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.046299839289813e-05,
+      "loss": 0.9638,
+      "step": 214250
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.046185046299839e-05,
+      "loss": 1.0528,
+      "step": 214260
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.046070253309864e-05,
+      "loss": 1.1269,
+      "step": 214270
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.0459554603198895e-05,
+      "loss": 1.1574,
+      "step": 214280
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.0458406673299145e-05,
+      "loss": 1.191,
+      "step": 214290
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.0457258743399395e-05,
+      "loss": 1.0956,
+      "step": 214300
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.045611081349965e-05,
+      "loss": 1.0307,
+      "step": 214310
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.045496288359991e-05,
+      "loss": 1.141,
+      "step": 214320
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.045381495370015e-05,
+      "loss": 1.1483,
+      "step": 214330
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.045266702380041e-05,
+      "loss": 1.044,
+      "step": 214340
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.0451519093900665e-05,
+      "loss": 1.2218,
+      "step": 214350
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.045037116400091e-05,
+      "loss": 1.1749,
+      "step": 214360
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.0449223234101164e-05,
+      "loss": 1.1477,
+      "step": 214370
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.044807530420142e-05,
+      "loss": 1.3795,
+      "step": 214380
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.044692737430167e-05,
+      "loss": 1.2265,
+      "step": 214390
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.044577944440192e-05,
+      "loss": 1.1181,
+      "step": 214400
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.044463151450218e-05,
+      "loss": 1.1458,
+      "step": 214410
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.044348358460243e-05,
+      "loss": 1.1224,
+      "step": 214420
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.0442335654702684e-05,
+      "loss": 1.2686,
+      "step": 214430
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.0441187724802934e-05,
+      "loss": 1.0868,
+      "step": 214440
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.0440039794903184e-05,
+      "loss": 1.1985,
+      "step": 214450
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.043889186500344e-05,
+      "loss": 1.1005,
+      "step": 214460
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.04377439351037e-05,
+      "loss": 1.1656,
+      "step": 214470
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.043659600520394e-05,
+      "loss": 1.0961,
+      "step": 214480
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.0435448075304197e-05,
+      "loss": 1.0726,
+      "step": 214490
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.043430014540445e-05,
+      "loss": 1.2283,
+      "step": 214500
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.0433152215504696e-05,
+      "loss": 1.1833,
+      "step": 214510
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.043200428560495e-05,
+      "loss": 1.0278,
+      "step": 214520
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.043085635570521e-05,
+      "loss": 1.0854,
+      "step": 214530
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.042970842580546e-05,
+      "loss": 1.0753,
+      "step": 214540
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.042856049590571e-05,
+      "loss": 1.1568,
+      "step": 214550
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.0427412566005966e-05,
+      "loss": 1.1873,
+      "step": 214560
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.0426264636106216e-05,
+      "loss": 1.1223,
+      "step": 214570
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.042511670620647e-05,
+      "loss": 1.0395,
+      "step": 214580
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.042396877630672e-05,
+      "loss": 1.0366,
+      "step": 214590
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.042282084640697e-05,
+      "loss": 1.186,
+      "step": 214600
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.042167291650723e-05,
+      "loss": 0.9727,
+      "step": 214610
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.0420524986607485e-05,
+      "loss": 1.2299,
+      "step": 214620
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.041937705670773e-05,
+      "loss": 1.1061,
+      "step": 214630
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.0418229126807985e-05,
+      "loss": 1.1542,
+      "step": 214640
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.041708119690824e-05,
+      "loss": 1.1164,
+      "step": 214650
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.0415933267008485e-05,
+      "loss": 1.1046,
+      "step": 214660
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.041478533710874e-05,
+      "loss": 1.0532,
+      "step": 214670
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.0413637407209e-05,
+      "loss": 1.0404,
+      "step": 214680
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.041248947730925e-05,
+      "loss": 1.0818,
+      "step": 214690
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.04113415474095e-05,
+      "loss": 1.0766,
+      "step": 214700
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.0410193617509754e-05,
+      "loss": 1.0695,
+      "step": 214710
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.0409045687610004e-05,
+      "loss": 1.1677,
+      "step": 214720
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.040789775771026e-05,
+      "loss": 1.0075,
+      "step": 214730
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.040674982781051e-05,
+      "loss": 1.19,
+      "step": 214740
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.040560189791076e-05,
+      "loss": 1.229,
+      "step": 214750
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.040445396801102e-05,
+      "loss": 1.1392,
+      "step": 214760
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.0403306038111274e-05,
+      "loss": 1.1403,
+      "step": 214770
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.040215810821152e-05,
+      "loss": 1.1442,
+      "step": 214780
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.0401010178311774e-05,
+      "loss": 1.0815,
+      "step": 214790
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.039986224841203e-05,
+      "loss": 1.0908,
+      "step": 214800
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.039871431851227e-05,
+      "loss": 0.9997,
+      "step": 214810
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.039756638861253e-05,
+      "loss": 1.0376,
+      "step": 214820
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.0396418458712787e-05,
+      "loss": 1.2111,
+      "step": 214830
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.0395270528813036e-05,
+      "loss": 1.0337,
+      "step": 214840
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.0394122598913286e-05,
+      "loss": 1.1429,
+      "step": 214850
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.039297466901354e-05,
+      "loss": 1.2945,
+      "step": 214860
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.039182673911379e-05,
+      "loss": 1.0815,
+      "step": 214870
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.039067880921405e-05,
+      "loss": 1.0691,
+      "step": 214880
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.03895308793143e-05,
+      "loss": 1.2106,
+      "step": 214890
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.038838294941455e-05,
+      "loss": 1.1522,
+      "step": 214900
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.0387235019514806e-05,
+      "loss": 1.1965,
+      "step": 214910
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.038608708961506e-05,
+      "loss": 1.0635,
+      "step": 214920
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.0384939159715305e-05,
+      "loss": 1.0677,
+      "step": 214930
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.038379122981556e-05,
+      "loss": 1.0912,
+      "step": 214940
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.038264329991582e-05,
+      "loss": 1.0845,
+      "step": 214950
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.038149537001606e-05,
+      "loss": 1.1122,
+      "step": 214960
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.038034744011632e-05,
+      "loss": 1.0028,
+      "step": 214970
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.0379199510216575e-05,
+      "loss": 1.0786,
+      "step": 214980
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.0378051580316825e-05,
+      "loss": 1.0895,
+      "step": 214990
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.0376903650417075e-05,
+      "loss": 1.0485,
+      "step": 215000
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.037575572051733e-05,
+      "loss": 1.0808,
+      "step": 215010
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.037460779061758e-05,
+      "loss": 1.1439,
+      "step": 215020
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.037345986071784e-05,
+      "loss": 1.1878,
+      "step": 215030
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.037231193081809e-05,
+      "loss": 0.951,
+      "step": 215040
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.037116400091834e-05,
+      "loss": 1.0149,
+      "step": 215050
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.0370016071018594e-05,
+      "loss": 1.0716,
+      "step": 215060
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.036886814111885e-05,
+      "loss": 1.0673,
+      "step": 215070
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.0367720211219094e-05,
+      "loss": 1.219,
+      "step": 215080
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.036657228131935e-05,
+      "loss": 1.0656,
+      "step": 215090
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.036542435141961e-05,
+      "loss": 1.1372,
+      "step": 215100
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.036427642151985e-05,
+      "loss": 1.1396,
+      "step": 215110
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.036312849162011e-05,
+      "loss": 1.0717,
+      "step": 215120
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.0361980561720363e-05,
+      "loss": 1.1588,
+      "step": 215130
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.036083263182061e-05,
+      "loss": 1.0891,
+      "step": 215140
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.035968470192086e-05,
+      "loss": 1.1048,
+      "step": 215150
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.035853677202112e-05,
+      "loss": 1.0571,
+      "step": 215160
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.035738884212137e-05,
+      "loss": 1.0695,
+      "step": 215170
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.0356240912221626e-05,
+      "loss": 1.0771,
+      "step": 215180
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.0355092982321876e-05,
+      "loss": 1.0452,
+      "step": 215190
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.0353945052422126e-05,
+      "loss": 1.0592,
+      "step": 215200
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.035279712252238e-05,
+      "loss": 1.0717,
+      "step": 215210
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.035164919262264e-05,
+      "loss": 1.0505,
+      "step": 215220
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.035050126272288e-05,
+      "loss": 1.1506,
+      "step": 215230
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.034935333282314e-05,
+      "loss": 1.0389,
+      "step": 215240
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.0348205402923396e-05,
+      "loss": 1.2684,
+      "step": 215250
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.034705747302364e-05,
+      "loss": 1.2176,
+      "step": 215260
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.0345909543123895e-05,
+      "loss": 1.0387,
+      "step": 215270
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.034476161322415e-05,
+      "loss": 1.0543,
+      "step": 215280
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.03436136833244e-05,
+      "loss": 1.0066,
+      "step": 215290
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.034246575342465e-05,
+      "loss": 1.0621,
+      "step": 215300
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.03413178235249e-05,
+      "loss": 1.0318,
+      "step": 215310
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.034016989362516e-05,
+      "loss": 1.0698,
+      "step": 215320
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.0339021963725415e-05,
+      "loss": 1.0607,
+      "step": 215330
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.033787403382566e-05,
+      "loss": 1.1638,
+      "step": 215340
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.0336726103925914e-05,
+      "loss": 1.0844,
+      "step": 215350
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.033557817402617e-05,
+      "loss": 1.1056,
+      "step": 215360
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.0334430244126414e-05,
+      "loss": 1.2405,
+      "step": 215370
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.033328231422667e-05,
+      "loss": 1.1011,
+      "step": 215380
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.033213438432693e-05,
+      "loss": 1.0288,
+      "step": 215390
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.033098645442718e-05,
+      "loss": 1.0607,
+      "step": 215400
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.032983852452743e-05,
+      "loss": 1.121,
+      "step": 215410
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.0328690594627684e-05,
+      "loss": 1.0502,
+      "step": 215420
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.0327542664727934e-05,
+      "loss": 1.0187,
+      "step": 215430
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.032639473482819e-05,
+      "loss": 1.0932,
+      "step": 215440
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.032524680492844e-05,
+      "loss": 1.1608,
+      "step": 215450
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.032409887502869e-05,
+      "loss": 1.112,
+      "step": 215460
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.0322950945128947e-05,
+      "loss": 1.1341,
+      "step": 215470
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.03218030152292e-05,
+      "loss": 1.2142,
+      "step": 215480
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.0320655085329446e-05,
+      "loss": 1.2685,
+      "step": 215490
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.03195071554297e-05,
+      "loss": 1.1201,
+      "step": 215500
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.031835922552996e-05,
+      "loss": 1.1326,
+      "step": 215510
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.03172112956302e-05,
+      "loss": 1.1371,
+      "step": 215520
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.031606336573046e-05,
+      "loss": 1.0577,
+      "step": 215530
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.0314915435830716e-05,
+      "loss": 1.1037,
+      "step": 215540
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.0313767505930966e-05,
+      "loss": 0.9848,
+      "step": 215550
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.0312619576031216e-05,
+      "loss": 1.2491,
+      "step": 215560
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.031147164613147e-05,
+      "loss": 1.1752,
+      "step": 215570
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.031032371623172e-05,
+      "loss": 1.0355,
+      "step": 215580
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.030917578633198e-05,
+      "loss": 1.1485,
+      "step": 215590
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.030802785643223e-05,
+      "loss": 1.2027,
+      "step": 215600
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.030687992653248e-05,
+      "loss": 1.1178,
+      "step": 215610
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.0305731996632735e-05,
+      "loss": 1.0916,
+      "step": 215620
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.030458406673299e-05,
+      "loss": 1.1553,
+      "step": 215630
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.0303436136833235e-05,
+      "loss": 1.1324,
+      "step": 215640
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.030228820693349e-05,
+      "loss": 1.1113,
+      "step": 215650
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.030114027703375e-05,
+      "loss": 1.1202,
+      "step": 215660
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.029999234713399e-05,
+      "loss": 1.1026,
+      "step": 215670
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.029884441723425e-05,
+      "loss": 1.1535,
+      "step": 215680
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.0297696487334504e-05,
+      "loss": 1.0916,
+      "step": 215690
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.0296548557434754e-05,
+      "loss": 1.1145,
+      "step": 215700
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.0295400627535004e-05,
+      "loss": 1.0844,
+      "step": 215710
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.029425269763526e-05,
+      "loss": 1.2559,
+      "step": 215720
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.029310476773551e-05,
+      "loss": 1.2401,
+      "step": 215730
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.029195683783577e-05,
+      "loss": 1.0319,
+      "step": 215740
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.029080890793602e-05,
+      "loss": 1.1684,
+      "step": 215750
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.028966097803627e-05,
+      "loss": 1.0757,
+      "step": 215760
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.0288513048136524e-05,
+      "loss": 1.2377,
+      "step": 215770
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.028736511823678e-05,
+      "loss": 1.0431,
+      "step": 215780
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.028621718833702e-05,
+      "loss": 1.1238,
+      "step": 215790
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.028506925843728e-05,
+      "loss": 1.0816,
+      "step": 215800
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.0283921328537537e-05,
+      "loss": 1.0218,
+      "step": 215810
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.028277339863778e-05,
+      "loss": 1.1273,
+      "step": 215820
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.0281625468738036e-05,
+      "loss": 1.2072,
+      "step": 215830
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.028047753883829e-05,
+      "loss": 1.1423,
+      "step": 215840
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.027932960893854e-05,
+      "loss": 1.0733,
+      "step": 215850
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.027818167903879e-05,
+      "loss": 1.1125,
+      "step": 215860
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.027703374913905e-05,
+      "loss": 1.1987,
+      "step": 215870
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.02758858192393e-05,
+      "loss": 0.9162,
+      "step": 215880
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.0274737889339556e-05,
+      "loss": 1.0492,
+      "step": 215890
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.0273589959439806e-05,
+      "loss": 1.0905,
+      "step": 215900
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.0272442029540055e-05,
+      "loss": 1.1573,
+      "step": 215910
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.027129409964031e-05,
+      "loss": 1.1099,
+      "step": 215920
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.027014616974057e-05,
+      "loss": 1.142,
+      "step": 215930
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.026899823984081e-05,
+      "loss": 1.0744,
+      "step": 215940
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.026785030994107e-05,
+      "loss": 1.2323,
+      "step": 215950
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.0266702380041325e-05,
+      "loss": 1.1458,
+      "step": 215960
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.026555445014157e-05,
+      "loss": 1.1054,
+      "step": 215970
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.0264406520241825e-05,
+      "loss": 1.0689,
+      "step": 215980
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.026325859034208e-05,
+      "loss": 1.07,
+      "step": 215990
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.026211066044233e-05,
+      "loss": 1.0534,
+      "step": 216000
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.026096273054258e-05,
+      "loss": 1.1686,
+      "step": 216010
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.025981480064284e-05,
+      "loss": 1.0017,
+      "step": 216020
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.025866687074309e-05,
+      "loss": 1.1794,
+      "step": 216030
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.0257518940843344e-05,
+      "loss": 1.2087,
+      "step": 216040
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.0256371010943594e-05,
+      "loss": 1.098,
+      "step": 216050
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.0255223081043844e-05,
+      "loss": 1.0381,
+      "step": 216060
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.02540751511441e-05,
+      "loss": 1.135,
+      "step": 216070
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.025292722124436e-05,
+      "loss": 1.1211,
+      "step": 216080
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.02517792913446e-05,
+      "loss": 1.0473,
+      "step": 216090
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.025063136144486e-05,
+      "loss": 1.1879,
+      "step": 216100
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.0249483431545113e-05,
+      "loss": 1.1545,
+      "step": 216110
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.0248335501645357e-05,
+      "loss": 1.2644,
+      "step": 216120
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.024718757174561e-05,
+      "loss": 1.1402,
+      "step": 216130
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.024603964184587e-05,
+      "loss": 1.1278,
+      "step": 216140
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.024489171194612e-05,
+      "loss": 1.1422,
+      "step": 216150
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.024374378204637e-05,
+      "loss": 1.035,
+      "step": 216160
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.0242595852146626e-05,
+      "loss": 1.0847,
+      "step": 216170
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.0241447922246876e-05,
+      "loss": 1.1192,
+      "step": 216180
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.024029999234713e-05,
+      "loss": 1.0435,
+      "step": 216190
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.023915206244738e-05,
+      "loss": 1.1496,
+      "step": 216200
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.023800413254763e-05,
+      "loss": 1.1178,
+      "step": 216210
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.023685620264789e-05,
+      "loss": 1.1375,
+      "step": 216220
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.0235708272748146e-05,
+      "loss": 0.9954,
+      "step": 216230
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.023456034284839e-05,
+      "loss": 1.1207,
+      "step": 216240
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.0233412412948645e-05,
+      "loss": 1.1242,
+      "step": 216250
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.02322644830489e-05,
+      "loss": 1.1439,
+      "step": 216260
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.0231116553149145e-05,
+      "loss": 0.9895,
+      "step": 216270
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.02299686232494e-05,
+      "loss": 1.32,
+      "step": 216280
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.022882069334966e-05,
+      "loss": 1.0899,
+      "step": 216290
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.022767276344991e-05,
+      "loss": 1.18,
+      "step": 216300
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.022652483355016e-05,
+      "loss": 1.1985,
+      "step": 216310
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.0225376903650415e-05,
+      "loss": 1.1175,
+      "step": 216320
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.0224228973750664e-05,
+      "loss": 1.0278,
+      "step": 216330
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.022308104385092e-05,
+      "loss": 1.0731,
+      "step": 216340
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.022193311395117e-05,
+      "loss": 1.1273,
+      "step": 216350
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.022078518405142e-05,
+      "loss": 1.1878,
+      "step": 216360
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.021963725415168e-05,
+      "loss": 1.2383,
+      "step": 216370
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.0218489324251934e-05,
+      "loss": 1.0804,
+      "step": 216380
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.021734139435218e-05,
+      "loss": 1.1712,
+      "step": 216390
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.0216193464452434e-05,
+      "loss": 1.159,
+      "step": 216400
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.021504553455269e-05,
+      "loss": 1.0911,
+      "step": 216410
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.0213897604652933e-05,
+      "loss": 1.0877,
+      "step": 216420
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.021274967475319e-05,
+      "loss": 1.1024,
+      "step": 216430
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.021160174485345e-05,
+      "loss": 1.1563,
+      "step": 216440
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.0210453814953697e-05,
+      "loss": 1.1152,
+      "step": 216450
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.0209305885053946e-05,
+      "loss": 1.2377,
+      "step": 216460
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.02081579551542e-05,
+      "loss": 1.0895,
+      "step": 216470
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.020701002525445e-05,
+      "loss": 1.1081,
+      "step": 216480
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.020586209535471e-05,
+      "loss": 1.1454,
+      "step": 216490
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.020471416545496e-05,
+      "loss": 1.1322,
+      "step": 216500
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.020356623555521e-05,
+      "loss": 1.0931,
+      "step": 216510
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.0202418305655466e-05,
+      "loss": 1.1833,
+      "step": 216520
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.020127037575572e-05,
+      "loss": 1.0601,
+      "step": 216530
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.0200122445855966e-05,
+      "loss": 1.1576,
+      "step": 216540
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.019897451595622e-05,
+      "loss": 1.1494,
+      "step": 216550
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.019782658605648e-05,
+      "loss": 1.1358,
+      "step": 216560
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.019667865615672e-05,
+      "loss": 1.1168,
+      "step": 216570
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.019553072625698e-05,
+      "loss": 0.9903,
+      "step": 216580
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.0194382796357235e-05,
+      "loss": 1.1573,
+      "step": 216590
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.0193234866457485e-05,
+      "loss": 1.0069,
+      "step": 216600
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.0192086936557735e-05,
+      "loss": 1.1096,
+      "step": 216610
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.019093900665799e-05,
+      "loss": 1.1925,
+      "step": 216620
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.018979107675824e-05,
+      "loss": 1.0963,
+      "step": 216630
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.01886431468585e-05,
+      "loss": 1.0584,
+      "step": 216640
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.018749521695875e-05,
+      "loss": 1.1532,
+      "step": 216650
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.0186347287059e-05,
+      "loss": 1.1533,
+      "step": 216660
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.0185199357159254e-05,
+      "loss": 1.1329,
+      "step": 216670
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.018405142725951e-05,
+      "loss": 1.1104,
+      "step": 216680
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.0182903497359754e-05,
+      "loss": 1.1714,
+      "step": 216690
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.018175556746001e-05,
+      "loss": 1.1775,
+      "step": 216700
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.018060763756026e-05,
+      "loss": 1.0929,
+      "step": 216710
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.017945970766051e-05,
+      "loss": 1.0931,
+      "step": 216720
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.017831177776077e-05,
+      "loss": 1.2499,
+      "step": 216730
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.017716384786102e-05,
+      "loss": 1.1403,
+      "step": 216740
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.0176015917961274e-05,
+      "loss": 1.0311,
+      "step": 216750
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.0174867988061523e-05,
+      "loss": 1.1663,
+      "step": 216760
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.017372005816177e-05,
+      "loss": 1.1928,
+      "step": 216770
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.017257212826203e-05,
+      "loss": 1.1588,
+      "step": 216780
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.0171424198362287e-05,
+      "loss": 1.0426,
+      "step": 216790
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.017027626846253e-05,
+      "loss": 1.087,
+      "step": 216800
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.0169128338562786e-05,
+      "loss": 1.1698,
+      "step": 216810
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.016798040866304e-05,
+      "loss": 1.325,
+      "step": 216820
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.0166832478763286e-05,
+      "loss": 1.1857,
+      "step": 216830
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.016568454886354e-05,
+      "loss": 1.2234,
+      "step": 216840
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.01645366189638e-05,
+      "loss": 1.1015,
+      "step": 216850
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.016338868906405e-05,
+      "loss": 0.9545,
+      "step": 216860
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.01622407591643e-05,
+      "loss": 1.081,
+      "step": 216870
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.0161092829264556e-05,
+      "loss": 1.1496,
+      "step": 216880
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.0159944899364805e-05,
+      "loss": 1.091,
+      "step": 216890
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.015879696946506e-05,
+      "loss": 1.0561,
+      "step": 216900
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.015764903956531e-05,
+      "loss": 1.1747,
+      "step": 216910
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.015650110966556e-05,
+      "loss": 1.2164,
+      "step": 216920
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.015535317976582e-05,
+      "loss": 0.9955,
+      "step": 216930
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.0154205249866075e-05,
+      "loss": 1.0103,
+      "step": 216940
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.015305731996632e-05,
+      "loss": 1.0946,
+      "step": 216950
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.0151909390066575e-05,
+      "loss": 1.1047,
+      "step": 216960
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.015076146016683e-05,
+      "loss": 1.0925,
+      "step": 216970
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.0149613530267074e-05,
+      "loss": 1.2866,
+      "step": 216980
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.014846560036733e-05,
+      "loss": 1.0954,
+      "step": 216990
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.014731767046759e-05,
+      "loss": 1.092,
+      "step": 217000
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.014616974056784e-05,
+      "loss": 1.1672,
+      "step": 217010
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.014502181066809e-05,
+      "loss": 1.085,
+      "step": 217020
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.0143873880768344e-05,
+      "loss": 1.1078,
+      "step": 217030
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.0142725950868594e-05,
+      "loss": 1.1382,
+      "step": 217040
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.014157802096885e-05,
+      "loss": 1.1465,
+      "step": 217050
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.01404300910691e-05,
+      "loss": 1.0981,
+      "step": 217060
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.013928216116935e-05,
+      "loss": 1.133,
+      "step": 217070
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.013813423126961e-05,
+      "loss": 1.1611,
+      "step": 217080
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.0136986301369863e-05,
+      "loss": 1.1772,
+      "step": 217090
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.0135838371470107e-05,
+      "loss": 1.2209,
+      "step": 217100
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.013469044157036e-05,
+      "loss": 1.1758,
+      "step": 217110
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.013354251167062e-05,
+      "loss": 1.0171,
+      "step": 217120
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.013239458177086e-05,
+      "loss": 0.9891,
+      "step": 217130
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.013124665187112e-05,
+      "loss": 1.1515,
+      "step": 217140
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.0130098721971376e-05,
+      "loss": 1.0895,
+      "step": 217150
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.0128950792071626e-05,
+      "loss": 1.0241,
+      "step": 217160
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.0127802862171876e-05,
+      "loss": 1.1604,
+      "step": 217170
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.012665493227213e-05,
+      "loss": 1.0392,
+      "step": 217180
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.012550700237238e-05,
+      "loss": 1.0345,
+      "step": 217190
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.012435907247264e-05,
+      "loss": 1.0257,
+      "step": 217200
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.012321114257289e-05,
+      "loss": 0.9965,
+      "step": 217210
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.012206321267314e-05,
+      "loss": 0.9953,
+      "step": 217220
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.0120915282773395e-05,
+      "loss": 1.2429,
+      "step": 217230
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.011976735287365e-05,
+      "loss": 1.015,
+      "step": 217240
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.0118619422973895e-05,
+      "loss": 1.0662,
+      "step": 217250
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.011747149307415e-05,
+      "loss": 1.1119,
+      "step": 217260
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.011632356317441e-05,
+      "loss": 1.1533,
+      "step": 217270
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.011517563327465e-05,
+      "loss": 1.1753,
+      "step": 217280
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.011402770337491e-05,
+      "loss": 1.2292,
+      "step": 217290
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.0112879773475165e-05,
+      "loss": 1.1351,
+      "step": 217300
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.0111731843575414e-05,
+      "loss": 1.1509,
+      "step": 217310
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.0110583913675664e-05,
+      "loss": 1.1398,
+      "step": 217320
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.010943598377592e-05,
+      "loss": 1.1914,
+      "step": 217330
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.010828805387617e-05,
+      "loss": 1.0838,
+      "step": 217340
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.010714012397643e-05,
+      "loss": 1.0069,
+      "step": 217350
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.010599219407668e-05,
+      "loss": 1.0996,
+      "step": 217360
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.010484426417693e-05,
+      "loss": 0.9909,
+      "step": 217370
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.0103696334277184e-05,
+      "loss": 1.0288,
+      "step": 217380
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.010254840437744e-05,
+      "loss": 1.0136,
+      "step": 217390
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.0101400474477683e-05,
+      "loss": 1.0074,
+      "step": 217400
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.010025254457794e-05,
+      "loss": 1.2015,
+      "step": 217410
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.00991046146782e-05,
+      "loss": 1.0951,
+      "step": 217420
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.009795668477844e-05,
+      "loss": 1.0083,
+      "step": 217430
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.0096808754878696e-05,
+      "loss": 1.2789,
+      "step": 217440
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.009566082497895e-05,
+      "loss": 1.1407,
+      "step": 217450
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.00945128950792e-05,
+      "loss": 1.1285,
+      "step": 217460
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.009336496517945e-05,
+      "loss": 1.1771,
+      "step": 217470
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.009221703527971e-05,
+      "loss": 0.8843,
+      "step": 217480
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.009106910537996e-05,
+      "loss": 1.054,
+      "step": 217490
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.0089921175480216e-05,
+      "loss": 1.084,
+      "step": 217500
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.0088773245580466e-05,
+      "loss": 1.0635,
+      "step": 217510
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.0087625315680716e-05,
+      "loss": 0.9943,
+      "step": 217520
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.008647738578097e-05,
+      "loss": 1.091,
+      "step": 217530
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.008532945588123e-05,
+      "loss": 1.0637,
+      "step": 217540
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.008418152598147e-05,
+      "loss": 1.1245,
+      "step": 217550
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.008303359608173e-05,
+      "loss": 1.1205,
+      "step": 217560
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.0081885666181985e-05,
+      "loss": 1.1391,
+      "step": 217570
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.008073773628223e-05,
+      "loss": 1.1686,
+      "step": 217580
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.0079589806382485e-05,
+      "loss": 1.2012,
+      "step": 217590
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.007844187648274e-05,
+      "loss": 1.0838,
+      "step": 217600
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.007729394658299e-05,
+      "loss": 1.1701,
+      "step": 217610
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.007614601668324e-05,
+      "loss": 1.1019,
+      "step": 217620
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.00749980867835e-05,
+      "loss": 1.0813,
+      "step": 217630
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.007385015688375e-05,
+      "loss": 1.1569,
+      "step": 217640
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.0072702226984004e-05,
+      "loss": 1.2067,
+      "step": 217650
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.0071554297084254e-05,
+      "loss": 1.0881,
+      "step": 217660
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.0070406367184504e-05,
+      "loss": 1.0413,
+      "step": 217670
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.006925843728476e-05,
+      "loss": 1.1421,
+      "step": 217680
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.006811050738502e-05,
+      "loss": 1.0747,
+      "step": 217690
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.006696257748526e-05,
+      "loss": 1.242,
+      "step": 217700
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.006581464758552e-05,
+      "loss": 1.2564,
+      "step": 217710
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.0064666717685774e-05,
+      "loss": 1.1055,
+      "step": 217720
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.006351878778602e-05,
+      "loss": 1.0738,
+      "step": 217730
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.006237085788627e-05,
+      "loss": 1.1845,
+      "step": 217740
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.006122292798653e-05,
+      "loss": 1.1587,
+      "step": 217750
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.006007499808678e-05,
+      "loss": 1.2857,
+      "step": 217760
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.005892706818703e-05,
+      "loss": 1.0472,
+      "step": 217770
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.0057779138287286e-05,
+      "loss": 1.0902,
+      "step": 217780
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.0056631208387536e-05,
+      "loss": 1.0874,
+      "step": 217790
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.005548327848779e-05,
+      "loss": 1.0134,
+      "step": 217800
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.005433534858804e-05,
+      "loss": 1.0659,
+      "step": 217810
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.005318741868829e-05,
+      "loss": 1.0695,
+      "step": 217820
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.005203948878855e-05,
+      "loss": 1.0778,
+      "step": 217830
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.0050891558888806e-05,
+      "loss": 1.1639,
+      "step": 217840
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.004974362898905e-05,
+      "loss": 1.1156,
+      "step": 217850
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.0048595699089306e-05,
+      "loss": 1.1361,
+      "step": 217860
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.004744776918956e-05,
+      "loss": 1.0635,
+      "step": 217870
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.0046299839289805e-05,
+      "loss": 1.1061,
+      "step": 217880
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.004515190939006e-05,
+      "loss": 1.1507,
+      "step": 217890
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.004400397949032e-05,
+      "loss": 1.1077,
+      "step": 217900
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.004285604959057e-05,
+      "loss": 1.1033,
+      "step": 217910
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.004170811969082e-05,
+      "loss": 1.1052,
+      "step": 217920
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.0040560189791075e-05,
+      "loss": 0.9986,
+      "step": 217930
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.0039412259891325e-05,
+      "loss": 1.029,
+      "step": 217940
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.003826432999158e-05,
+      "loss": 1.1175,
+      "step": 217950
+    },
+    {
+      "epoch": 10.0,
+      "eval_cer": 0.6684071955726049,
+      "eval_em": 0.007357212003872217,
+      "eval_f1": 0.007357212003872217,
+      "eval_loss": 1.2018513679504395,
+      "eval_runtime": 1718.2171,
+      "eval_samples_per_second": 6.012,
+      "eval_steps_per_second": 0.752,
+      "eval_wer": 0.9926427879961278,
+      "step": 217953
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.003711640009183e-05,
+      "loss": 1.1708,
+      "step": 217960
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.003596847019208e-05,
+      "loss": 1.1888,
+      "step": 217970
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.003482054029234e-05,
+      "loss": 0.998,
+      "step": 217980
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.0033672610392594e-05,
+      "loss": 1.0463,
+      "step": 217990
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.003252468049284e-05,
+      "loss": 1.086,
+      "step": 218000
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.0031376750593094e-05,
+      "loss": 1.0895,
+      "step": 218010
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.003022882069335e-05,
+      "loss": 1.0312,
+      "step": 218020
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.0029080890793594e-05,
+      "loss": 1.0401,
+      "step": 218030
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.002793296089385e-05,
+      "loss": 1.0338,
+      "step": 218040
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.002678503099411e-05,
+      "loss": 0.9859,
+      "step": 218050
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.002563710109436e-05,
+      "loss": 1.0,
+      "step": 218060
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.002448917119461e-05,
+      "loss": 0.9528,
+      "step": 218070
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.0023341241294857e-05,
+      "loss": 1.2502,
+      "step": 218080
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.002219331139511e-05,
+      "loss": 0.921,
+      "step": 218090
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.002104538149537e-05,
+      "loss": 1.1013,
+      "step": 218100
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.001989745159561e-05,
+      "loss": 1.1938,
+      "step": 218110
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.001874952169587e-05,
+      "loss": 1.1729,
+      "step": 218120
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.0017601591796126e-05,
+      "loss": 1.0069,
+      "step": 218130
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.001645366189637e-05,
+      "loss": 0.9861,
+      "step": 218140
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.0015305731996626e-05,
+      "loss": 1.0407,
+      "step": 218150
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.001415780209688e-05,
+      "loss": 1.0772,
+      "step": 218160
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.001300987219713e-05,
+      "loss": 1.1009,
+      "step": 218170
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.001186194229738e-05,
+      "loss": 1.0017,
+      "step": 218180
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.001071401239764e-05,
+      "loss": 1.072,
+      "step": 218190
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.000956608249789e-05,
+      "loss": 1.1578,
+      "step": 218200
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.0008418152598145e-05,
+      "loss": 1.1335,
+      "step": 218210
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.0007270222698395e-05,
+      "loss": 1.0333,
+      "step": 218220
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.0006122292798645e-05,
+      "loss": 1.0032,
+      "step": 218230
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.00049743628989e-05,
+      "loss": 1.0088,
+      "step": 218240
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.000382643299916e-05,
+      "loss": 0.9495,
+      "step": 218250
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.00026785030994e-05,
+      "loss": 0.9809,
+      "step": 218260
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.000153057319966e-05,
+      "loss": 1.0771,
+      "step": 218270
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.0000382643299915e-05,
+      "loss": 1.0481,
+      "step": 218280
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.999923471340016e-05,
+      "loss": 1.1154,
+      "step": 218290
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.9998086783500414e-05,
+      "loss": 1.0848,
+      "step": 218300
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.999693885360067e-05,
+      "loss": 0.9925,
+      "step": 218310
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.999579092370092e-05,
+      "loss": 1.0072,
+      "step": 218320
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.999464299380117e-05,
+      "loss": 1.0843,
+      "step": 218330
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.999349506390143e-05,
+      "loss": 1.0726,
+      "step": 218340
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.999234713400168e-05,
+      "loss": 1.0546,
+      "step": 218350
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.9991199204101934e-05,
+      "loss": 1.0803,
+      "step": 218360
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.9990051274202184e-05,
+      "loss": 1.0056,
+      "step": 218370
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.9988903344302433e-05,
+      "loss": 1.1497,
+      "step": 218380
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.998775541440269e-05,
+      "loss": 1.066,
+      "step": 218390
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.998660748450295e-05,
+      "loss": 1.1099,
+      "step": 218400
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.998545955460319e-05,
+      "loss": 1.0912,
+      "step": 218410
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.9984311624703446e-05,
+      "loss": 0.9011,
+      "step": 218420
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.99831636948037e-05,
+      "loss": 1.0761,
+      "step": 218430
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.9982015764903946e-05,
+      "loss": 1.0389,
+      "step": 218440
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.99808678350042e-05,
+      "loss": 1.0897,
+      "step": 218450
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.997971990510446e-05,
+      "loss": 1.2549,
+      "step": 218460
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.997857197520471e-05,
+      "loss": 1.0055,
+      "step": 218470
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.997742404530496e-05,
+      "loss": 1.1397,
+      "step": 218480
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.9976276115405216e-05,
+      "loss": 1.1682,
+      "step": 218490
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.9975128185505466e-05,
+      "loss": 1.0212,
+      "step": 218500
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.997398025560572e-05,
+      "loss": 1.0324,
+      "step": 218510
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.997283232570597e-05,
+      "loss": 1.0318,
+      "step": 218520
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.997168439580622e-05,
+      "loss": 1.1362,
+      "step": 218530
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.997053646590648e-05,
+      "loss": 1.1131,
+      "step": 218540
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.9969388536006735e-05,
+      "loss": 1.1434,
+      "step": 218550
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.996824060610698e-05,
+      "loss": 1.0019,
+      "step": 218560
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.9967092676207235e-05,
+      "loss": 1.0515,
+      "step": 218570
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.996594474630749e-05,
+      "loss": 1.0847,
+      "step": 218580
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.9964796816407735e-05,
+      "loss": 0.9405,
+      "step": 218590
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.996364888650799e-05,
+      "loss": 1.0104,
+      "step": 218600
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.996250095660825e-05,
+      "loss": 1.0308,
+      "step": 218610
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.99613530267085e-05,
+      "loss": 1.1515,
+      "step": 218620
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.996020509680875e-05,
+      "loss": 1.0047,
+      "step": 218630
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.9959057166909004e-05,
+      "loss": 1.0147,
+      "step": 218640
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.9957909237009254e-05,
+      "loss": 0.998,
+      "step": 218650
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.995676130710951e-05,
+      "loss": 1.0747,
+      "step": 218660
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.995561337720976e-05,
+      "loss": 1.1312,
+      "step": 218670
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.995446544731001e-05,
+      "loss": 1.1095,
+      "step": 218680
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.995331751741027e-05,
+      "loss": 0.9828,
+      "step": 218690
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.9952169587510524e-05,
+      "loss": 1.0238,
+      "step": 218700
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.995102165761077e-05,
+      "loss": 1.0839,
+      "step": 218710
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.994987372771102e-05,
+      "loss": 1.0013,
+      "step": 218720
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.994872579781128e-05,
+      "loss": 1.1216,
+      "step": 218730
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.994757786791152e-05,
+      "loss": 1.0937,
+      "step": 218740
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.994642993801178e-05,
+      "loss": 0.9639,
+      "step": 218750
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.9945282008112036e-05,
+      "loss": 1.004,
+      "step": 218760
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.9944134078212286e-05,
+      "loss": 1.0188,
+      "step": 218770
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.9942986148312536e-05,
+      "loss": 1.0275,
+      "step": 218780
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.994183821841279e-05,
+      "loss": 0.9935,
+      "step": 218790
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.994069028851304e-05,
+      "loss": 1.1675,
+      "step": 218800
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.99395423586133e-05,
+      "loss": 1.0744,
+      "step": 218810
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.993839442871355e-05,
+      "loss": 0.9901,
+      "step": 218820
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.99372464988138e-05,
+      "loss": 1.0631,
+      "step": 218830
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.9936098568914056e-05,
+      "loss": 0.9765,
+      "step": 218840
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.993495063901431e-05,
+      "loss": 1.0782,
+      "step": 218850
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.9933802709114555e-05,
+      "loss": 0.9637,
+      "step": 218860
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.993265477921481e-05,
+      "loss": 1.0093,
+      "step": 218870
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.993150684931507e-05,
+      "loss": 1.0345,
+      "step": 218880
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.993035891941531e-05,
+      "loss": 0.9805,
+      "step": 218890
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.992921098951557e-05,
+      "loss": 1.0817,
+      "step": 218900
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.9928063059615825e-05,
+      "loss": 1.1661,
+      "step": 218910
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.9926915129716075e-05,
+      "loss": 0.9294,
+      "step": 218920
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.9925767199816325e-05,
+      "loss": 1.0321,
+      "step": 218930
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.992461926991658e-05,
+      "loss": 0.9928,
+      "step": 218940
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.992347134001683e-05,
+      "loss": 0.99,
+      "step": 218950
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.992232341011709e-05,
+      "loss": 1.0444,
+      "step": 218960
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.992117548021734e-05,
+      "loss": 1.0706,
+      "step": 218970
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.992002755031759e-05,
+      "loss": 0.9742,
+      "step": 218980
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.9918879620417844e-05,
+      "loss": 1.052,
+      "step": 218990
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.99177316905181e-05,
+      "loss": 1.1484,
+      "step": 219000
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.9916583760618344e-05,
+      "loss": 0.9766,
+      "step": 219010
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.99154358307186e-05,
+      "loss": 1.0559,
+      "step": 219020
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.991428790081886e-05,
+      "loss": 1.1295,
+      "step": 219030
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.99131399709191e-05,
+      "loss": 1.0199,
+      "step": 219040
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.991199204101936e-05,
+      "loss": 0.9713,
+      "step": 219050
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.991084411111961e-05,
+      "loss": 1.0351,
+      "step": 219060
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.990969618121986e-05,
+      "loss": 0.9085,
+      "step": 219070
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.990854825132011e-05,
+      "loss": 1.0651,
+      "step": 219080
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.990740032142037e-05,
+      "loss": 1.0697,
+      "step": 219090
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.990625239152062e-05,
+      "loss": 0.9657,
+      "step": 219100
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.9905104461620876e-05,
+      "loss": 1.1128,
+      "step": 219110
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.9903956531721126e-05,
+      "loss": 1.099,
+      "step": 219120
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.9902808601821376e-05,
+      "loss": 1.0109,
+      "step": 219130
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.990166067192163e-05,
+      "loss": 0.9991,
+      "step": 219140
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.990051274202189e-05,
+      "loss": 1.0292,
+      "step": 219150
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.989936481212213e-05,
+      "loss": 1.0252,
+      "step": 219160
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.989821688222239e-05,
+      "loss": 1.0081,
+      "step": 219170
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.9897068952322645e-05,
+      "loss": 1.0272,
+      "step": 219180
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.989592102242289e-05,
+      "loss": 1.0908,
+      "step": 219190
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.9894773092523145e-05,
+      "loss": 1.0788,
+      "step": 219200
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.98936251626234e-05,
+      "loss": 1.2326,
+      "step": 219210
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.989247723272365e-05,
+      "loss": 1.062,
+      "step": 219220
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.98913293028239e-05,
+      "loss": 1.0463,
+      "step": 219230
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.989018137292416e-05,
+      "loss": 1.0863,
+      "step": 219240
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.988903344302441e-05,
+      "loss": 1.1453,
+      "step": 219250
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.9887885513124665e-05,
+      "loss": 1.0305,
+      "step": 219260
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.9886737583224914e-05,
+      "loss": 1.0438,
+      "step": 219270
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.9885589653325164e-05,
+      "loss": 1.0214,
+      "step": 219280
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.988444172342542e-05,
+      "loss": 1.0728,
+      "step": 219290
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.988329379352568e-05,
+      "loss": 1.0086,
+      "step": 219300
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.988214586362592e-05,
+      "loss": 1.1162,
+      "step": 219310
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.988099793372618e-05,
+      "loss": 1.0812,
+      "step": 219320
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.9879850003826434e-05,
+      "loss": 1.1397,
+      "step": 219330
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.987870207392668e-05,
+      "loss": 1.047,
+      "step": 219340
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.9877554144026934e-05,
+      "loss": 1.1039,
+      "step": 219350
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.987640621412719e-05,
+      "loss": 1.1384,
+      "step": 219360
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.987525828422744e-05,
+      "loss": 1.0366,
+      "step": 219370
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.987411035432769e-05,
+      "loss": 0.9785,
+      "step": 219380
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.9872962424427947e-05,
+      "loss": 1.0263,
+      "step": 219390
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.9871814494528196e-05,
+      "loss": 1.0206,
+      "step": 219400
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.987066656462845e-05,
+      "loss": 0.9993,
+      "step": 219410
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.98695186347287e-05,
+      "loss": 1.2107,
+      "step": 219420
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.986837070482895e-05,
+      "loss": 1.1026,
+      "step": 219430
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.986722277492921e-05,
+      "loss": 1.0026,
+      "step": 219440
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.986607484502945e-05,
+      "loss": 0.9493,
+      "step": 219450
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.986492691512971e-05,
+      "loss": 0.9992,
+      "step": 219460
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.9863778985229966e-05,
+      "loss": 0.9887,
+      "step": 219470
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.9862631055330216e-05,
+      "loss": 1.0227,
+      "step": 219480
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.9861483125430465e-05,
+      "loss": 0.988,
+      "step": 219490
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.986033519553072e-05,
+      "loss": 1.0442,
+      "step": 219500
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.985918726563097e-05,
+      "loss": 1.0543,
+      "step": 219510
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.985803933573123e-05,
+      "loss": 1.0185,
+      "step": 219520
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.985689140583148e-05,
+      "loss": 1.0459,
+      "step": 219530
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.985574347593173e-05,
+      "loss": 0.9603,
+      "step": 219540
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.9854595546031985e-05,
+      "loss": 1.0676,
+      "step": 219550
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.985344761613224e-05,
+      "loss": 1.0786,
+      "step": 219560
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.9852299686232485e-05,
+      "loss": 1.2014,
+      "step": 219570
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.985115175633274e-05,
+      "loss": 1.1105,
+      "step": 219580
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.9850003826433e-05,
+      "loss": 1.0243,
+      "step": 219590
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.984885589653324e-05,
+      "loss": 1.0012,
+      "step": 219600
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.98477079666335e-05,
+      "loss": 0.9878,
+      "step": 219610
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.9846560036733754e-05,
+      "loss": 1.1275,
+      "step": 219620
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.9845412106834004e-05,
+      "loss": 1.1077,
+      "step": 219630
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.9844264176934254e-05,
+      "loss": 0.9904,
+      "step": 219640
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.984311624703451e-05,
+      "loss": 1.083,
+      "step": 219650
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.984196831713476e-05,
+      "loss": 1.0424,
+      "step": 219660
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.984082038723502e-05,
+      "loss": 1.0356,
+      "step": 219670
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.983967245733527e-05,
+      "loss": 1.0054,
+      "step": 219680
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.983852452743552e-05,
+      "loss": 1.0628,
+      "step": 219690
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.983737659753577e-05,
+      "loss": 0.9751,
+      "step": 219700
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.983622866763603e-05,
+      "loss": 1.062,
+      "step": 219710
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.983508073773627e-05,
+      "loss": 0.9548,
+      "step": 219720
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.983393280783653e-05,
+      "loss": 0.9275,
+      "step": 219730
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.9832784877936786e-05,
+      "loss": 1.105,
+      "step": 219740
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.983163694803703e-05,
+      "loss": 0.9491,
+      "step": 219750
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.9830489018137286e-05,
+      "loss": 1.1724,
+      "step": 219760
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.982934108823754e-05,
+      "loss": 0.9978,
+      "step": 219770
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.982819315833779e-05,
+      "loss": 1.0268,
+      "step": 219780
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.982704522843804e-05,
+      "loss": 1.2055,
+      "step": 219790
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.98258972985383e-05,
+      "loss": 1.2506,
+      "step": 219800
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.982474936863855e-05,
+      "loss": 1.1365,
+      "step": 219810
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.9823601438738806e-05,
+      "loss": 1.0814,
+      "step": 219820
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.9822453508839055e-05,
+      "loss": 1.0356,
+      "step": 219830
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.9821305578939305e-05,
+      "loss": 1.1079,
+      "step": 219840
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.982015764903956e-05,
+      "loss": 1.0609,
+      "step": 219850
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.981900971913982e-05,
+      "loss": 1.036,
+      "step": 219860
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.981786178924006e-05,
+      "loss": 0.997,
+      "step": 219870
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.981671385934032e-05,
+      "loss": 1.0117,
+      "step": 219880
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.9815565929440575e-05,
+      "loss": 1.0716,
+      "step": 219890
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.981441799954082e-05,
+      "loss": 1.0875,
+      "step": 219900
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.9813270069641075e-05,
+      "loss": 1.0978,
+      "step": 219910
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.981212213974133e-05,
+      "loss": 1.0291,
+      "step": 219920
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.981097420984158e-05,
+      "loss": 1.013,
+      "step": 219930
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.980982627994183e-05,
+      "loss": 1.1004,
+      "step": 219940
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.980867835004209e-05,
+      "loss": 1.119,
+      "step": 219950
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.980753042014234e-05,
+      "loss": 1.0127,
+      "step": 219960
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.9806382490242594e-05,
+      "loss": 1.1079,
+      "step": 219970
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.9805234560342844e-05,
+      "loss": 0.964,
+      "step": 219980
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.9804086630443094e-05,
+      "loss": 1.0771,
+      "step": 219990
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.980293870054335e-05,
+      "loss": 1.0398,
+      "step": 220000
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.980179077064361e-05,
+      "loss": 1.016,
+      "step": 220010
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.980064284074385e-05,
+      "loss": 0.973,
+      "step": 220020
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.979949491084411e-05,
+      "loss": 1.0794,
+      "step": 220030
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.979834698094436e-05,
+      "loss": 1.0651,
+      "step": 220040
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.9797199051044606e-05,
+      "loss": 1.0554,
+      "step": 220050
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.979605112114486e-05,
+      "loss": 1.0362,
+      "step": 220060
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.979490319124512e-05,
+      "loss": 1.0406,
+      "step": 220070
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.979375526134537e-05,
+      "loss": 1.149,
+      "step": 220080
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.979260733144562e-05,
+      "loss": 1.1299,
+      "step": 220090
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.9791459401545876e-05,
+      "loss": 1.0304,
+      "step": 220100
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.9790311471646126e-05,
+      "loss": 0.978,
+      "step": 220110
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.978916354174638e-05,
+      "loss": 1.0711,
+      "step": 220120
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.978801561184663e-05,
+      "loss": 1.0592,
+      "step": 220130
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.978686768194688e-05,
+      "loss": 1.1206,
+      "step": 220140
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.978571975204714e-05,
+      "loss": 0.9747,
+      "step": 220150
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.9784571822147395e-05,
+      "loss": 1.0075,
+      "step": 220160
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.978342389224764e-05,
+      "loss": 1.0699,
+      "step": 220170
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.9782275962347895e-05,
+      "loss": 1.011,
+      "step": 220180
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.978112803244815e-05,
+      "loss": 1.0852,
+      "step": 220190
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.9779980102548395e-05,
+      "loss": 1.0728,
+      "step": 220200
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.977883217264865e-05,
+      "loss": 0.9685,
+      "step": 220210
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.977768424274891e-05,
+      "loss": 1.0361,
+      "step": 220220
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.977653631284916e-05,
+      "loss": 1.027,
+      "step": 220230
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.977538838294941e-05,
+      "loss": 1.0781,
+      "step": 220240
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.9774240453049664e-05,
+      "loss": 1.1144,
+      "step": 220250
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.9773092523149914e-05,
+      "loss": 1.0865,
+      "step": 220260
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.977194459325017e-05,
+      "loss": 1.1541,
+      "step": 220270
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.977079666335042e-05,
+      "loss": 0.9939,
+      "step": 220280
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.976964873345067e-05,
+      "loss": 1.0993,
+      "step": 220290
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.976850080355093e-05,
+      "loss": 1.0958,
+      "step": 220300
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.9767352873651184e-05,
+      "loss": 1.1044,
+      "step": 220310
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.976620494375143e-05,
+      "loss": 1.1186,
+      "step": 220320
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.9765057013851684e-05,
+      "loss": 0.9573,
+      "step": 220330
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.976390908395194e-05,
+      "loss": 1.0283,
+      "step": 220340
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.976276115405218e-05,
+      "loss": 0.9909,
+      "step": 220350
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.976161322415244e-05,
+      "loss": 1.1657,
+      "step": 220360
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.9760465294252697e-05,
+      "loss": 0.9792,
+      "step": 220370
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.9759317364352946e-05,
+      "loss": 1.0959,
+      "step": 220380
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.9758169434453196e-05,
+      "loss": 0.9865,
+      "step": 220390
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.975702150455345e-05,
+      "loss": 1.04,
+      "step": 220400
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.97558735746537e-05,
+      "loss": 1.0685,
+      "step": 220410
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.975472564475396e-05,
+      "loss": 1.1267,
+      "step": 220420
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.975357771485421e-05,
+      "loss": 1.0598,
+      "step": 220430
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.975242978495446e-05,
+      "loss": 1.0747,
+      "step": 220440
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.9751281855054716e-05,
+      "loss": 1.0357,
+      "step": 220450
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.975013392515497e-05,
+      "loss": 1.1299,
+      "step": 220460
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.9748985995255215e-05,
+      "loss": 1.0494,
+      "step": 220470
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.974783806535547e-05,
+      "loss": 0.9691,
+      "step": 220480
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.974669013545573e-05,
+      "loss": 1.0781,
+      "step": 220490
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.974554220555597e-05,
+      "loss": 0.9976,
+      "step": 220500
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.974439427565623e-05,
+      "loss": 1.0957,
+      "step": 220510
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.9743246345756485e-05,
+      "loss": 1.1281,
+      "step": 220520
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.9742098415856735e-05,
+      "loss": 1.0072,
+      "step": 220530
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.9740950485956985e-05,
+      "loss": 1.0348,
+      "step": 220540
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.973980255605724e-05,
+      "loss": 1.122,
+      "step": 220550
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.973865462615749e-05,
+      "loss": 1.1283,
+      "step": 220560
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.973750669625775e-05,
+      "loss": 0.9827,
+      "step": 220570
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.9736358766358e-05,
+      "loss": 1.0143,
+      "step": 220580
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.973521083645825e-05,
+      "loss": 0.9516,
+      "step": 220590
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.9734062906558504e-05,
+      "loss": 1.0873,
+      "step": 220600
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.973291497665876e-05,
+      "loss": 1.0266,
+      "step": 220610
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.9731767046759004e-05,
+      "loss": 1.0568,
+      "step": 220620
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.973061911685926e-05,
+      "loss": 1.029,
+      "step": 220630
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.972947118695952e-05,
+      "loss": 1.1958,
+      "step": 220640
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.972832325705976e-05,
+      "loss": 1.0037,
+      "step": 220650
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.972717532716002e-05,
+      "loss": 0.9902,
+      "step": 220660
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.9726027397260274e-05,
+      "loss": 1.1598,
+      "step": 220670
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.972487946736052e-05,
+      "loss": 1.0512,
+      "step": 220680
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.972373153746077e-05,
+      "loss": 0.9937,
+      "step": 220690
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.972258360756103e-05,
+      "loss": 1.067,
+      "step": 220700
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.972143567766128e-05,
+      "loss": 1.0962,
+      "step": 220710
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.9720287747761536e-05,
+      "loss": 1.042,
+      "step": 220720
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.9719139817861786e-05,
+      "loss": 1.0124,
+      "step": 220730
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.9717991887962036e-05,
+      "loss": 1.1417,
+      "step": 220740
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.971684395806229e-05,
+      "loss": 1.0989,
+      "step": 220750
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.971569602816255e-05,
+      "loss": 1.0188,
+      "step": 220760
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.971454809826279e-05,
+      "loss": 0.8915,
+      "step": 220770
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.971340016836305e-05,
+      "loss": 1.0039,
+      "step": 220780
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.9712252238463306e-05,
+      "loss": 1.0611,
+      "step": 220790
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.971110430856355e-05,
+      "loss": 1.1353,
+      "step": 220800
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.9709956378663805e-05,
+      "loss": 1.11,
+      "step": 220810
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.9708808448764055e-05,
+      "loss": 1.2074,
+      "step": 220820
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.970766051886431e-05,
+      "loss": 1.0734,
+      "step": 220830
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.970651258896456e-05,
+      "loss": 1.0713,
+      "step": 220840
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.970536465906481e-05,
+      "loss": 1.0261,
+      "step": 220850
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.970421672916507e-05,
+      "loss": 0.8515,
+      "step": 220860
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.9703068799265325e-05,
+      "loss": 0.9737,
+      "step": 220870
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.970192086936557e-05,
+      "loss": 1.0885,
+      "step": 220880
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.9700772939465825e-05,
+      "loss": 1.0572,
+      "step": 220890
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.969962500956608e-05,
+      "loss": 1.0981,
+      "step": 220900
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.9698477079666324e-05,
+      "loss": 1.0182,
+      "step": 220910
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.969732914976658e-05,
+      "loss": 0.9847,
+      "step": 220920
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.969618121986684e-05,
+      "loss": 0.9917,
+      "step": 220930
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.969503328996709e-05,
+      "loss": 1.035,
+      "step": 220940
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.969388536006734e-05,
+      "loss": 1.0861,
+      "step": 220950
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.9692737430167594e-05,
+      "loss": 1.0566,
+      "step": 220960
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.9691589500267844e-05,
+      "loss": 0.9405,
+      "step": 220970
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.96904415703681e-05,
+      "loss": 1.0127,
+      "step": 220980
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.968929364046835e-05,
+      "loss": 0.951,
+      "step": 220990
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.96881457105686e-05,
+      "loss": 1.071,
+      "step": 221000
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.968699778066886e-05,
+      "loss": 1.1488,
+      "step": 221010
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.968584985076911e-05,
+      "loss": 1.0147,
+      "step": 221020
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.9684701920869356e-05,
+      "loss": 0.9852,
+      "step": 221030
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.968355399096961e-05,
+      "loss": 1.101,
+      "step": 221040
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.968240606106987e-05,
+      "loss": 1.0257,
+      "step": 221050
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.968125813117011e-05,
+      "loss": 1.1618,
+      "step": 221060
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.968011020127037e-05,
+      "loss": 1.0283,
+      "step": 221070
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.9678962271370626e-05,
+      "loss": 1.0269,
+      "step": 221080
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.9677814341470876e-05,
+      "loss": 0.9822,
+      "step": 221090
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.9676666411571126e-05,
+      "loss": 1.0992,
+      "step": 221100
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.967551848167138e-05,
+      "loss": 1.1123,
+      "step": 221110
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.967437055177163e-05,
+      "loss": 1.1472,
+      "step": 221120
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.967322262187189e-05,
+      "loss": 1.1757,
+      "step": 221130
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.967207469197214e-05,
+      "loss": 1.038,
+      "step": 221140
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.967092676207239e-05,
+      "loss": 1.1158,
+      "step": 221150
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.9669778832172645e-05,
+      "loss": 1.0479,
+      "step": 221160
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.96686309022729e-05,
+      "loss": 1.0796,
+      "step": 221170
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.9667482972373145e-05,
+      "loss": 1.1392,
+      "step": 221180
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.96663350424734e-05,
+      "loss": 1.0178,
+      "step": 221190
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.966518711257366e-05,
+      "loss": 0.9537,
+      "step": 221200
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.96640391826739e-05,
+      "loss": 1.0559,
+      "step": 221210
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.966289125277416e-05,
+      "loss": 1.0163,
+      "step": 221220
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.9661743322874414e-05,
+      "loss": 1.1094,
+      "step": 221230
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.9660595392974664e-05,
+      "loss": 1.2027,
+      "step": 221240
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.9659447463074914e-05,
+      "loss": 1.0963,
+      "step": 221250
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.965829953317517e-05,
+      "loss": 1.0499,
+      "step": 221260
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.965715160327542e-05,
+      "loss": 1.0645,
+      "step": 221270
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.965600367337568e-05,
+      "loss": 1.0412,
+      "step": 221280
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.965485574347593e-05,
+      "loss": 1.1676,
+      "step": 221290
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.965370781357618e-05,
+      "loss": 1.0078,
+      "step": 221300
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.9652559883676434e-05,
+      "loss": 1.0441,
+      "step": 221310
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.965141195377669e-05,
+      "loss": 1.0409,
+      "step": 221320
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.965026402387693e-05,
+      "loss": 1.0056,
+      "step": 221330
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.964911609397719e-05,
+      "loss": 1.1673,
+      "step": 221340
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.9647968164077447e-05,
+      "loss": 0.9997,
+      "step": 221350
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.964682023417769e-05,
+      "loss": 1.0277,
+      "step": 221360
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.9645672304277946e-05,
+      "loss": 1.0909,
+      "step": 221370
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.96445243743782e-05,
+      "loss": 0.9829,
+      "step": 221380
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.964337644447845e-05,
+      "loss": 0.9656,
+      "step": 221390
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.96422285145787e-05,
+      "loss": 1.0002,
+      "step": 221400
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.964108058467896e-05,
+      "loss": 0.9636,
+      "step": 221410
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.963993265477921e-05,
+      "loss": 1.1139,
+      "step": 221420
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.9638784724879466e-05,
+      "loss": 0.9522,
+      "step": 221430
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.9637636794979716e-05,
+      "loss": 1.0141,
+      "step": 221440
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.9636488865079965e-05,
+      "loss": 1.0732,
+      "step": 221450
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.963534093518022e-05,
+      "loss": 0.9934,
+      "step": 221460
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.963419300528048e-05,
+      "loss": 1.1653,
+      "step": 221470
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.963304507538072e-05,
+      "loss": 1.1055,
+      "step": 221480
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.963189714548098e-05,
+      "loss": 1.0891,
+      "step": 221490
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.9630749215581235e-05,
+      "loss": 0.9949,
+      "step": 221500
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.962960128568148e-05,
+      "loss": 1.0279,
+      "step": 221510
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.9628453355781735e-05,
+      "loss": 0.9392,
+      "step": 221520
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.962730542588199e-05,
+      "loss": 0.9494,
+      "step": 221530
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.962615749598224e-05,
+      "loss": 1.0619,
+      "step": 221540
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.962500956608249e-05,
+      "loss": 1.0767,
+      "step": 221550
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.962386163618275e-05,
+      "loss": 0.9735,
+      "step": 221560
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.9622713706283e-05,
+      "loss": 1.091,
+      "step": 221570
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.9621565776383254e-05,
+      "loss": 1.0977,
+      "step": 221580
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.9620417846483504e-05,
+      "loss": 1.0267,
+      "step": 221590
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.9619269916583754e-05,
+      "loss": 1.0514,
+      "step": 221600
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.961812198668401e-05,
+      "loss": 1.187,
+      "step": 221610
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.961697405678427e-05,
+      "loss": 0.9877,
+      "step": 221620
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.961582612688451e-05,
+      "loss": 1.0905,
+      "step": 221630
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.961467819698477e-05,
+      "loss": 0.9913,
+      "step": 221640
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.9613530267085024e-05,
+      "loss": 1.0745,
+      "step": 221650
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.9612382337185267e-05,
+      "loss": 1.0297,
+      "step": 221660
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.961123440728552e-05,
+      "loss": 1.1237,
+      "step": 221670
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.961008647738578e-05,
+      "loss": 1.1122,
+      "step": 221680
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.960893854748603e-05,
+      "loss": 1.0003,
+      "step": 221690
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.960779061758628e-05,
+      "loss": 0.9493,
+      "step": 221700
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.9606642687686536e-05,
+      "loss": 1.0383,
+      "step": 221710
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.9605494757786786e-05,
+      "loss": 0.9826,
+      "step": 221720
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.960434682788704e-05,
+      "loss": 0.9762,
+      "step": 221730
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.960319889798729e-05,
+      "loss": 1.0161,
+      "step": 221740
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.960205096808754e-05,
+      "loss": 0.9536,
+      "step": 221750
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.96009030381878e-05,
+      "loss": 1.0734,
+      "step": 221760
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.9599755108288056e-05,
+      "loss": 1.1005,
+      "step": 221770
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.95986071783883e-05,
+      "loss": 1.086,
+      "step": 221780
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.9597459248488555e-05,
+      "loss": 1.0275,
+      "step": 221790
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.959631131858881e-05,
+      "loss": 1.0844,
+      "step": 221800
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.9595163388689055e-05,
+      "loss": 1.1084,
+      "step": 221810
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.959401545878931e-05,
+      "loss": 1.0636,
+      "step": 221820
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.959286752888957e-05,
+      "loss": 1.0386,
+      "step": 221830
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.959171959898982e-05,
+      "loss": 0.9901,
+      "step": 221840
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.959057166909007e-05,
+      "loss": 1.0754,
+      "step": 221850
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.9589423739190325e-05,
+      "loss": 1.0762,
+      "step": 221860
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.9588275809290575e-05,
+      "loss": 0.9789,
+      "step": 221870
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.958712787939083e-05,
+      "loss": 1.1078,
+      "step": 221880
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.958597994949108e-05,
+      "loss": 1.0801,
+      "step": 221890
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.958483201959133e-05,
+      "loss": 1.1014,
+      "step": 221900
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.958368408969159e-05,
+      "loss": 1.0814,
+      "step": 221910
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.9582536159791844e-05,
+      "loss": 0.98,
+      "step": 221920
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.958138822989209e-05,
+      "loss": 1.0298,
+      "step": 221930
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.9580240299992344e-05,
+      "loss": 1.0408,
+      "step": 221940
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.95790923700926e-05,
+      "loss": 1.0565,
+      "step": 221950
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.9577944440192844e-05,
+      "loss": 1.0274,
+      "step": 221960
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.95767965102931e-05,
+      "loss": 1.0104,
+      "step": 221970
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.957564858039336e-05,
+      "loss": 1.0797,
+      "step": 221980
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.957450065049361e-05,
+      "loss": 1.1013,
+      "step": 221990
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.9573352720593857e-05,
+      "loss": 1.0037,
+      "step": 222000
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.957220479069411e-05,
+      "loss": 0.9906,
+      "step": 222010
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.957105686079436e-05,
+      "loss": 1.1012,
+      "step": 222020
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.956990893089462e-05,
+      "loss": 1.0172,
+      "step": 222030
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.956876100099487e-05,
+      "loss": 1.0094,
+      "step": 222040
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.956761307109512e-05,
+      "loss": 1.0132,
+      "step": 222050
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.9566465141195376e-05,
+      "loss": 1.0739,
+      "step": 222060
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.956531721129563e-05,
+      "loss": 1.144,
+      "step": 222070
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.9564169281395876e-05,
+      "loss": 0.9289,
+      "step": 222080
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.956302135149613e-05,
+      "loss": 1.0216,
+      "step": 222090
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.956187342159639e-05,
+      "loss": 0.9769,
+      "step": 222100
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.956072549169663e-05,
+      "loss": 0.9205,
+      "step": 222110
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.955957756179689e-05,
+      "loss": 1.1497,
+      "step": 222120
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.9558429631897145e-05,
+      "loss": 1.0581,
+      "step": 222130
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.9557281701997395e-05,
+      "loss": 1.005,
+      "step": 222140
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.9556133772097645e-05,
+      "loss": 1.1296,
+      "step": 222150
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.95549858421979e-05,
+      "loss": 1.038,
+      "step": 222160
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.955383791229815e-05,
+      "loss": 1.0383,
+      "step": 222170
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.955268998239841e-05,
+      "loss": 1.0076,
+      "step": 222180
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.955154205249865e-05,
+      "loss": 1.0405,
+      "step": 222190
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.955039412259891e-05,
+      "loss": 1.0884,
+      "step": 222200
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.9549246192699164e-05,
+      "loss": 1.0552,
+      "step": 222210
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.954809826279941e-05,
+      "loss": 1.0053,
+      "step": 222220
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.9546950332899664e-05,
+      "loss": 1.0716,
+      "step": 222230
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.954580240299992e-05,
+      "loss": 1.058,
+      "step": 222240
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.954465447310017e-05,
+      "loss": 1.0703,
+      "step": 222250
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.954350654320042e-05,
+      "loss": 1.0378,
+      "step": 222260
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.954235861330068e-05,
+      "loss": 1.0406,
+      "step": 222270
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.954121068340093e-05,
+      "loss": 1.0307,
+      "step": 222280
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.9540062753501184e-05,
+      "loss": 1.1953,
+      "step": 222290
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.9538914823601433e-05,
+      "loss": 1.0803,
+      "step": 222300
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.953776689370168e-05,
+      "loss": 1.0577,
+      "step": 222310
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.953661896380194e-05,
+      "loss": 1.0281,
+      "step": 222320
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.9535471033902197e-05,
+      "loss": 0.9879,
+      "step": 222330
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.953432310400244e-05,
+      "loss": 1.1025,
+      "step": 222340
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.9533175174102696e-05,
+      "loss": 1.0411,
+      "step": 222350
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.953202724420295e-05,
+      "loss": 0.9937,
+      "step": 222360
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.9530879314303196e-05,
+      "loss": 1.0847,
+      "step": 222370
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.952973138440345e-05,
+      "loss": 1.0034,
+      "step": 222380
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.952858345450371e-05,
+      "loss": 1.0957,
+      "step": 222390
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.952743552460396e-05,
+      "loss": 0.9751,
+      "step": 222400
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.952628759470421e-05,
+      "loss": 1.0362,
+      "step": 222410
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.9525139664804466e-05,
+      "loss": 1.0842,
+      "step": 222420
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.9523991734904715e-05,
+      "loss": 1.0559,
+      "step": 222430
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.952284380500497e-05,
+      "loss": 1.0598,
+      "step": 222440
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.952169587510522e-05,
+      "loss": 1.0629,
+      "step": 222450
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.952054794520547e-05,
+      "loss": 1.1846,
+      "step": 222460
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.951940001530573e-05,
+      "loss": 1.0824,
+      "step": 222470
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.9518252085405985e-05,
+      "loss": 0.9803,
+      "step": 222480
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.951710415550623e-05,
+      "loss": 1.0337,
+      "step": 222490
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.9515956225606485e-05,
+      "loss": 1.1503,
+      "step": 222500
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.951480829570674e-05,
+      "loss": 1.0196,
+      "step": 222510
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.9513660365806984e-05,
+      "loss": 1.0306,
+      "step": 222520
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.951251243590724e-05,
+      "loss": 1.0306,
+      "step": 222530
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.95113645060075e-05,
+      "loss": 1.0949,
+      "step": 222540
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.951021657610775e-05,
+      "loss": 1.0119,
+      "step": 222550
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.9509068646208e-05,
+      "loss": 1.1302,
+      "step": 222560
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.9507920716308254e-05,
+      "loss": 1.0909,
+      "step": 222570
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.9506772786408504e-05,
+      "loss": 1.027,
+      "step": 222580
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.950562485650876e-05,
+      "loss": 1.1336,
+      "step": 222590
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.950447692660901e-05,
+      "loss": 1.0072,
+      "step": 222600
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.950332899670926e-05,
+      "loss": 1.0964,
+      "step": 222610
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.950218106680952e-05,
+      "loss": 1.0578,
+      "step": 222620
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.9501033136909774e-05,
+      "loss": 1.1621,
+      "step": 222630
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.9499885207010017e-05,
+      "loss": 0.9883,
+      "step": 222640
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.949873727711027e-05,
+      "loss": 1.1309,
+      "step": 222650
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.949758934721053e-05,
+      "loss": 1.1029,
+      "step": 222660
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.949644141731077e-05,
+      "loss": 1.0271,
+      "step": 222670
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.949529348741103e-05,
+      "loss": 1.0806,
+      "step": 222680
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.9494145557511286e-05,
+      "loss": 1.1049,
+      "step": 222690
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.9492997627611536e-05,
+      "loss": 1.1157,
+      "step": 222700
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.9491849697711786e-05,
+      "loss": 0.9986,
+      "step": 222710
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.949070176781204e-05,
+      "loss": 1.1306,
+      "step": 222720
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.948955383791229e-05,
+      "loss": 1.1102,
+      "step": 222730
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.948840590801255e-05,
+      "loss": 1.144,
+      "step": 222740
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.94872579781128e-05,
+      "loss": 1.1192,
+      "step": 222750
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.948611004821305e-05,
+      "loss": 0.9577,
+      "step": 222760
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.9484962118313305e-05,
+      "loss": 1.0968,
+      "step": 222770
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.948381418841356e-05,
+      "loss": 1.0382,
+      "step": 222780
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.9482666258513805e-05,
+      "loss": 1.0367,
+      "step": 222790
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.948151832861406e-05,
+      "loss": 1.0974,
+      "step": 222800
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.948037039871432e-05,
+      "loss": 1.1159,
+      "step": 222810
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.947922246881456e-05,
+      "loss": 1.0368,
+      "step": 222820
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.947807453891482e-05,
+      "loss": 1.1291,
+      "step": 222830
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.9476926609015075e-05,
+      "loss": 0.9627,
+      "step": 222840
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.9475778679115325e-05,
+      "loss": 1.1765,
+      "step": 222850
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.9474630749215574e-05,
+      "loss": 1.0371,
+      "step": 222860
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.947348281931583e-05,
+      "loss": 1.129,
+      "step": 222870
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.947233488941608e-05,
+      "loss": 1.1197,
+      "step": 222880
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.947118695951634e-05,
+      "loss": 1.0546,
+      "step": 222890
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.947003902961659e-05,
+      "loss": 1.0215,
+      "step": 222900
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.946889109971684e-05,
+      "loss": 1.1558,
+      "step": 222910
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.9467743169817094e-05,
+      "loss": 0.9945,
+      "step": 222920
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.946659523991735e-05,
+      "loss": 1.0074,
+      "step": 222930
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.9465447310017594e-05,
+      "loss": 1.084,
+      "step": 222940
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.946429938011785e-05,
+      "loss": 1.0711,
+      "step": 222950
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.946315145021811e-05,
+      "loss": 1.0033,
+      "step": 222960
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.946200352031835e-05,
+      "loss": 1.0451,
+      "step": 222970
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.9460855590418606e-05,
+      "loss": 1.0233,
+      "step": 222980
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.945970766051886e-05,
+      "loss": 1.0548,
+      "step": 222990
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.945855973061911e-05,
+      "loss": 0.9822,
+      "step": 223000
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.945741180071936e-05,
+      "loss": 1.1751,
+      "step": 223010
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.945626387081962e-05,
+      "loss": 1.1188,
+      "step": 223020
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.945511594091987e-05,
+      "loss": 1.0848,
+      "step": 223030
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.9453968011020126e-05,
+      "loss": 1.0694,
+      "step": 223040
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.9452820081120376e-05,
+      "loss": 0.9651,
+      "step": 223050
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.9451672151220626e-05,
+      "loss": 1.1247,
+      "step": 223060
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.945052422132088e-05,
+      "loss": 1.0282,
+      "step": 223070
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.944937629142114e-05,
+      "loss": 0.9707,
+      "step": 223080
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.944822836152138e-05,
+      "loss": 0.9951,
+      "step": 223090
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.944708043162164e-05,
+      "loss": 1.0499,
+      "step": 223100
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.9445932501721895e-05,
+      "loss": 1.0502,
+      "step": 223110
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.944478457182214e-05,
+      "loss": 1.1248,
+      "step": 223120
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.9443636641922395e-05,
+      "loss": 1.098,
+      "step": 223130
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.944248871202265e-05,
+      "loss": 1.101,
+      "step": 223140
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.94413407821229e-05,
+      "loss": 1.0535,
+      "step": 223150
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.944019285222315e-05,
+      "loss": 1.0365,
+      "step": 223160
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.943904492232341e-05,
+      "loss": 1.1856,
+      "step": 223170
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.943789699242366e-05,
+      "loss": 0.9733,
+      "step": 223180
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.9436749062523914e-05,
+      "loss": 1.049,
+      "step": 223190
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.9435601132624164e-05,
+      "loss": 0.9943,
+      "step": 223200
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.9434453202724414e-05,
+      "loss": 1.0021,
+      "step": 223210
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.943330527282467e-05,
+      "loss": 1.0904,
+      "step": 223220
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.943215734292493e-05,
+      "loss": 1.0737,
+      "step": 223230
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.943100941302517e-05,
+      "loss": 1.1073,
+      "step": 223240
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.942986148312543e-05,
+      "loss": 1.1416,
+      "step": 223250
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.9428713553225684e-05,
+      "loss": 1.1058,
+      "step": 223260
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.942756562332593e-05,
+      "loss": 1.1241,
+      "step": 223270
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.9426417693426183e-05,
+      "loss": 1.1535,
+      "step": 223280
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.942526976352644e-05,
+      "loss": 1.1249,
+      "step": 223290
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.942412183362669e-05,
+      "loss": 1.0683,
+      "step": 223300
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.942297390372694e-05,
+      "loss": 1.1359,
+      "step": 223310
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.9421825973827196e-05,
+      "loss": 1.1542,
+      "step": 223320
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.9420678043927446e-05,
+      "loss": 1.1225,
+      "step": 223330
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.94195301140277e-05,
+      "loss": 1.1105,
+      "step": 223340
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.941838218412795e-05,
+      "loss": 1.0183,
+      "step": 223350
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.94172342542282e-05,
+      "loss": 1.0135,
+      "step": 223360
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.941608632432846e-05,
+      "loss": 1.0421,
+      "step": 223370
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.9414938394428716e-05,
+      "loss": 0.9612,
+      "step": 223380
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.941379046452896e-05,
+      "loss": 0.9775,
+      "step": 223390
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.9412642534629216e-05,
+      "loss": 1.1265,
+      "step": 223400
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.941149460472947e-05,
+      "loss": 1.0168,
+      "step": 223410
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.9410346674829715e-05,
+      "loss": 1.0445,
+      "step": 223420
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.940919874492997e-05,
+      "loss": 0.9731,
+      "step": 223430
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.940805081503023e-05,
+      "loss": 1.1094,
+      "step": 223440
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.940690288513048e-05,
+      "loss": 1.1976,
+      "step": 223450
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.940575495523073e-05,
+      "loss": 1.0992,
+      "step": 223460
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.9404607025330985e-05,
+      "loss": 1.0614,
+      "step": 223470
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.9403459095431235e-05,
+      "loss": 1.0245,
+      "step": 223480
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.940231116553149e-05,
+      "loss": 1.0388,
+      "step": 223490
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.940116323563174e-05,
+      "loss": 1.0801,
+      "step": 223500
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.940001530573199e-05,
+      "loss": 1.0348,
+      "step": 223510
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.939886737583225e-05,
+      "loss": 1.1383,
+      "step": 223520
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.9397719445932504e-05,
+      "loss": 1.0037,
+      "step": 223530
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.939657151603275e-05,
+      "loss": 1.1332,
+      "step": 223540
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.9395423586133004e-05,
+      "loss": 1.0562,
+      "step": 223550
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.9394275656233254e-05,
+      "loss": 1.0645,
+      "step": 223560
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.9393127726333504e-05,
+      "loss": 1.0153,
+      "step": 223570
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.939197979643376e-05,
+      "loss": 1.1876,
+      "step": 223580
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.939083186653401e-05,
+      "loss": 1.0597,
+      "step": 223590
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.938968393663427e-05,
+      "loss": 1.1669,
+      "step": 223600
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.938853600673452e-05,
+      "loss": 1.0073,
+      "step": 223610
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.9387388076834767e-05,
+      "loss": 1.0709,
+      "step": 223620
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.938624014693502e-05,
+      "loss": 0.9905,
+      "step": 223630
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.938509221703528e-05,
+      "loss": 1.0513,
+      "step": 223640
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.938394428713552e-05,
+      "loss": 1.1332,
+      "step": 223650
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.938279635723578e-05,
+      "loss": 1.0205,
+      "step": 223660
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.9381648427336036e-05,
+      "loss": 1.0365,
+      "step": 223670
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.938050049743628e-05,
+      "loss": 1.0711,
+      "step": 223680
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.9379352567536536e-05,
+      "loss": 1.0,
+      "step": 223690
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.937820463763679e-05,
+      "loss": 1.033,
+      "step": 223700
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.937705670773704e-05,
+      "loss": 1.0159,
+      "step": 223710
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.937590877783729e-05,
+      "loss": 1.0309,
+      "step": 223720
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.937476084793755e-05,
+      "loss": 1.0336,
+      "step": 223730
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.93736129180378e-05,
+      "loss": 1.0606,
+      "step": 223740
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.9372464988138055e-05,
+      "loss": 0.9709,
+      "step": 223750
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.9371317058238305e-05,
+      "loss": 1.09,
+      "step": 223760
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.9370169128338555e-05,
+      "loss": 1.0722,
+      "step": 223770
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.936902119843881e-05,
+      "loss": 1.1259,
+      "step": 223780
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.936787326853907e-05,
+      "loss": 1.0193,
+      "step": 223790
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.936672533863931e-05,
+      "loss": 1.0087,
+      "step": 223800
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.936557740873957e-05,
+      "loss": 1.0933,
+      "step": 223810
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.9364429478839825e-05,
+      "loss": 1.0958,
+      "step": 223820
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.936328154894007e-05,
+      "loss": 1.0104,
+      "step": 223830
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.9362133619040324e-05,
+      "loss": 1.0199,
+      "step": 223840
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.936098568914058e-05,
+      "loss": 0.9629,
+      "step": 223850
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.935983775924083e-05,
+      "loss": 1.2364,
+      "step": 223860
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.935868982934108e-05,
+      "loss": 1.0919,
+      "step": 223870
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.935754189944134e-05,
+      "loss": 1.1121,
+      "step": 223880
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.935639396954159e-05,
+      "loss": 1.0386,
+      "step": 223890
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.9355246039641844e-05,
+      "loss": 1.0311,
+      "step": 223900
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.9354098109742094e-05,
+      "loss": 1.0423,
+      "step": 223910
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.9352950179842344e-05,
+      "loss": 1.0195,
+      "step": 223920
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.93518022499426e-05,
+      "loss": 1.0054,
+      "step": 223930
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.935065432004286e-05,
+      "loss": 0.9917,
+      "step": 223940
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.93495063901431e-05,
+      "loss": 1.1182,
+      "step": 223950
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.9348358460243356e-05,
+      "loss": 1.009,
+      "step": 223960
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.934721053034361e-05,
+      "loss": 1.0333,
+      "step": 223970
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.9346062600443856e-05,
+      "loss": 1.0935,
+      "step": 223980
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.934491467054411e-05,
+      "loss": 1.01,
+      "step": 223990
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.934376674064437e-05,
+      "loss": 1.0588,
+      "step": 224000
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.934261881074462e-05,
+      "loss": 1.1289,
+      "step": 224010
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.934147088084487e-05,
+      "loss": 1.2625,
+      "step": 224020
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.9340322950945126e-05,
+      "loss": 1.0674,
+      "step": 224030
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.9339175021045376e-05,
+      "loss": 1.1214,
+      "step": 224040
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.933802709114563e-05,
+      "loss": 1.0033,
+      "step": 224050
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.933687916124588e-05,
+      "loss": 1.0219,
+      "step": 224060
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.933573123134613e-05,
+      "loss": 1.0756,
+      "step": 224070
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.933458330144639e-05,
+      "loss": 1.0987,
+      "step": 224080
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.9333435371546645e-05,
+      "loss": 1.0425,
+      "step": 224090
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.933228744164689e-05,
+      "loss": 0.9962,
+      "step": 224100
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.9331139511747145e-05,
+      "loss": 1.0654,
+      "step": 224110
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.93299915818474e-05,
+      "loss": 1.0341,
+      "step": 224120
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.9328843651947645e-05,
+      "loss": 1.0259,
+      "step": 224130
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.93276957220479e-05,
+      "loss": 1.0687,
+      "step": 224140
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.932654779214816e-05,
+      "loss": 1.0465,
+      "step": 224150
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.932539986224841e-05,
+      "loss": 1.0759,
+      "step": 224160
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.932425193234866e-05,
+      "loss": 1.0603,
+      "step": 224170
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.9323104002448914e-05,
+      "loss": 1.0255,
+      "step": 224180
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.9321956072549164e-05,
+      "loss": 1.0265,
+      "step": 224190
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.932080814264942e-05,
+      "loss": 1.2638,
+      "step": 224200
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.931966021274967e-05,
+      "loss": 1.0679,
+      "step": 224210
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.931851228284992e-05,
+      "loss": 1.1711,
+      "step": 224220
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.931736435295018e-05,
+      "loss": 1.1106,
+      "step": 224230
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.9316216423050434e-05,
+      "loss": 1.1611,
+      "step": 224240
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.931506849315068e-05,
+      "loss": 1.0822,
+      "step": 224250
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.9313920563250933e-05,
+      "loss": 1.1064,
+      "step": 224260
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.931277263335119e-05,
+      "loss": 0.9619,
+      "step": 224270
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.931162470345143e-05,
+      "loss": 1.0905,
+      "step": 224280
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.931047677355169e-05,
+      "loss": 1.1788,
+      "step": 224290
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.9309328843651946e-05,
+      "loss": 1.1338,
+      "step": 224300
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.9308180913752196e-05,
+      "loss": 1.007,
+      "step": 224310
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.9307032983852446e-05,
+      "loss": 0.9905,
+      "step": 224320
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.93058850539527e-05,
+      "loss": 0.9544,
+      "step": 224330
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.930473712405295e-05,
+      "loss": 0.9874,
+      "step": 224340
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.930358919415321e-05,
+      "loss": 1.1057,
+      "step": 224350
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.930244126425346e-05,
+      "loss": 0.8887,
+      "step": 224360
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.930129333435371e-05,
+      "loss": 1.0196,
+      "step": 224370
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.9300145404453966e-05,
+      "loss": 1.1415,
+      "step": 224380
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.929899747455422e-05,
+      "loss": 1.0262,
+      "step": 224390
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.9297849544654465e-05,
+      "loss": 1.0043,
+      "step": 224400
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.929670161475472e-05,
+      "loss": 1.2048,
+      "step": 224410
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.929555368485498e-05,
+      "loss": 1.1322,
+      "step": 224420
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.929440575495522e-05,
+      "loss": 1.0598,
+      "step": 224430
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.929325782505548e-05,
+      "loss": 1.0535,
+      "step": 224440
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.9292109895155735e-05,
+      "loss": 1.05,
+      "step": 224450
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.9290961965255985e-05,
+      "loss": 1.0342,
+      "step": 224460
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.9289814035356235e-05,
+      "loss": 0.9792,
+      "step": 224470
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.928866610545649e-05,
+      "loss": 1.0366,
+      "step": 224480
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.928751817555674e-05,
+      "loss": 0.922,
+      "step": 224490
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.9286370245657e-05,
+      "loss": 1.1228,
+      "step": 224500
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.928522231575725e-05,
+      "loss": 1.0282,
+      "step": 224510
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.92840743858575e-05,
+      "loss": 1.215,
+      "step": 224520
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.9282926455957754e-05,
+      "loss": 1.0608,
+      "step": 224530
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.928177852605801e-05,
+      "loss": 0.9638,
+      "step": 224540
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.9280630596158254e-05,
+      "loss": 1.0446,
+      "step": 224550
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.927948266625851e-05,
+      "loss": 1.0867,
+      "step": 224560
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.927833473635877e-05,
+      "loss": 1.0975,
+      "step": 224570
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.927718680645901e-05,
+      "loss": 1.1334,
+      "step": 224580
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.927603887655927e-05,
+      "loss": 0.9916,
+      "step": 224590
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.927489094665952e-05,
+      "loss": 1.0254,
+      "step": 224600
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.927374301675977e-05,
+      "loss": 1.0477,
+      "step": 224610
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.927259508686002e-05,
+      "loss": 1.0137,
+      "step": 224620
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.927144715696028e-05,
+      "loss": 1.1246,
+      "step": 224630
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.927029922706053e-05,
+      "loss": 1.0683,
+      "step": 224640
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.9269151297160786e-05,
+      "loss": 1.0741,
+      "step": 224650
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.9268003367261036e-05,
+      "loss": 1.0005,
+      "step": 224660
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.9266855437361286e-05,
+      "loss": 1.1496,
+      "step": 224670
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.926570750746154e-05,
+      "loss": 1.0144,
+      "step": 224680
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.92645595775618e-05,
+      "loss": 1.1282,
+      "step": 224690
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.926341164766204e-05,
+      "loss": 1.0307,
+      "step": 224700
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.92622637177623e-05,
+      "loss": 0.9979,
+      "step": 224710
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.9261115787862555e-05,
+      "loss": 1.0355,
+      "step": 224720
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.92599678579628e-05,
+      "loss": 1.0549,
+      "step": 224730
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.9258819928063055e-05,
+      "loss": 1.1823,
+      "step": 224740
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.925767199816331e-05,
+      "loss": 1.0475,
+      "step": 224750
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.925652406826356e-05,
+      "loss": 1.1469,
+      "step": 224760
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.925537613836381e-05,
+      "loss": 1.0955,
+      "step": 224770
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.925422820846407e-05,
+      "loss": 1.0543,
+      "step": 224780
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.925308027856432e-05,
+      "loss": 1.0307,
+      "step": 224790
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.9251932348664575e-05,
+      "loss": 1.0842,
+      "step": 224800
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.9250784418764825e-05,
+      "loss": 1.1166,
+      "step": 224810
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.9249636488865074e-05,
+      "loss": 1.0828,
+      "step": 224820
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.924848855896533e-05,
+      "loss": 1.055,
+      "step": 224830
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.924734062906559e-05,
+      "loss": 1.0432,
+      "step": 224840
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.924619269916583e-05,
+      "loss": 1.082,
+      "step": 224850
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.924504476926609e-05,
+      "loss": 1.0279,
+      "step": 224860
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.9243896839366344e-05,
+      "loss": 1.1094,
+      "step": 224870
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.924274890946659e-05,
+      "loss": 1.0612,
+      "step": 224880
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.9241600979566844e-05,
+      "loss": 1.0517,
+      "step": 224890
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.92404530496671e-05,
+      "loss": 1.0183,
+      "step": 224900
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.923930511976735e-05,
+      "loss": 1.1135,
+      "step": 224910
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.92381571898676e-05,
+      "loss": 0.9963,
+      "step": 224920
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.923700925996785e-05,
+      "loss": 1.0895,
+      "step": 224930
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.9235861330068106e-05,
+      "loss": 1.0249,
+      "step": 224940
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.923471340016836e-05,
+      "loss": 0.9946,
+      "step": 224950
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.9233565470268606e-05,
+      "loss": 1.0323,
+      "step": 224960
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.923241754036886e-05,
+      "loss": 1.11,
+      "step": 224970
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.923126961046912e-05,
+      "loss": 1.003,
+      "step": 224980
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.923012168056936e-05,
+      "loss": 1.1204,
+      "step": 224990
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.922897375066962e-05,
+      "loss": 1.0946,
+      "step": 225000
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.9227825820769876e-05,
+      "loss": 1.1978,
+      "step": 225010
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.9226677890870126e-05,
+      "loss": 1.0612,
+      "step": 225020
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.9225529960970376e-05,
+      "loss": 0.9969,
+      "step": 225030
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.922438203107063e-05,
+      "loss": 1.016,
+      "step": 225040
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.922323410117088e-05,
+      "loss": 1.1897,
+      "step": 225050
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.922208617127114e-05,
+      "loss": 1.0424,
+      "step": 225060
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.922093824137139e-05,
+      "loss": 1.1401,
+      "step": 225070
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.921979031147164e-05,
+      "loss": 1.1454,
+      "step": 225080
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.9218642381571895e-05,
+      "loss": 1.1068,
+      "step": 225090
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.921749445167215e-05,
+      "loss": 1.1651,
+      "step": 225100
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.9216346521772395e-05,
+      "loss": 0.9859,
+      "step": 225110
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.921519859187265e-05,
+      "loss": 1.0298,
+      "step": 225120
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.921405066197291e-05,
+      "loss": 1.1643,
+      "step": 225130
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.921290273207315e-05,
+      "loss": 1.1221,
+      "step": 225140
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.921175480217341e-05,
+      "loss": 1.0178,
+      "step": 225150
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.9210606872273664e-05,
+      "loss": 1.064,
+      "step": 225160
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.9209458942373914e-05,
+      "loss": 0.9604,
+      "step": 225170
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.9208311012474164e-05,
+      "loss": 0.9699,
+      "step": 225180
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.920716308257442e-05,
+      "loss": 1.1184,
+      "step": 225190
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.920601515267467e-05,
+      "loss": 1.1103,
+      "step": 225200
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.920486722277493e-05,
+      "loss": 1.1014,
+      "step": 225210
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.920371929287518e-05,
+      "loss": 1.0676,
+      "step": 225220
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.920257136297543e-05,
+      "loss": 1.0305,
+      "step": 225230
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.9201423433075683e-05,
+      "loss": 0.9879,
+      "step": 225240
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.920027550317594e-05,
+      "loss": 0.9787,
+      "step": 225250
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.919912757327618e-05,
+      "loss": 1.001,
+      "step": 225260
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.919797964337644e-05,
+      "loss": 1.2263,
+      "step": 225270
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.9196831713476696e-05,
+      "loss": 1.0258,
+      "step": 225280
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.919568378357694e-05,
+      "loss": 1.1038,
+      "step": 225290
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.9194535853677196e-05,
+      "loss": 1.1585,
+      "step": 225300
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.919338792377745e-05,
+      "loss": 1.1669,
+      "step": 225310
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.91922399938777e-05,
+      "loss": 1.1233,
+      "step": 225320
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.919109206397795e-05,
+      "loss": 1.1058,
+      "step": 225330
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.918994413407821e-05,
+      "loss": 0.9717,
+      "step": 225340
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.918879620417846e-05,
+      "loss": 1.1161,
+      "step": 225350
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.9187648274278716e-05,
+      "loss": 1.1809,
+      "step": 225360
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.9186500344378965e-05,
+      "loss": 1.0494,
+      "step": 225370
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.9185352414479215e-05,
+      "loss": 1.1479,
+      "step": 225380
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.918420448457947e-05,
+      "loss": 1.0579,
+      "step": 225390
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.918305655467973e-05,
+      "loss": 1.1173,
+      "step": 225400
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.918190862477997e-05,
+      "loss": 1.1437,
+      "step": 225410
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.918076069488023e-05,
+      "loss": 1.0963,
+      "step": 225420
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.9179612764980485e-05,
+      "loss": 1.0724,
+      "step": 225430
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.917846483508073e-05,
+      "loss": 1.0808,
+      "step": 225440
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.9177316905180985e-05,
+      "loss": 1.016,
+      "step": 225450
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.917616897528124e-05,
+      "loss": 0.941,
+      "step": 225460
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.917502104538149e-05,
+      "loss": 1.0428,
+      "step": 225470
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.917387311548174e-05,
+      "loss": 1.0903,
+      "step": 225480
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.9172725185582e-05,
+      "loss": 0.9965,
+      "step": 225490
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.917157725568225e-05,
+      "loss": 1.0874,
+      "step": 225500
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.9170429325782504e-05,
+      "loss": 1.047,
+      "step": 225510
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.9169281395882754e-05,
+      "loss": 1.17,
+      "step": 225520
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.9168133465983004e-05,
+      "loss": 1.0027,
+      "step": 225530
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.916698553608326e-05,
+      "loss": 1.0559,
+      "step": 225540
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.916583760618352e-05,
+      "loss": 1.0989,
+      "step": 225550
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.916468967628376e-05,
+      "loss": 1.0189,
+      "step": 225560
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.916354174638402e-05,
+      "loss": 1.13,
+      "step": 225570
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.916239381648427e-05,
+      "loss": 1.0519,
+      "step": 225580
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.9161245886584516e-05,
+      "loss": 1.235,
+      "step": 225590
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.916009795668477e-05,
+      "loss": 1.0368,
+      "step": 225600
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.915895002678503e-05,
+      "loss": 1.0318,
+      "step": 225610
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.915780209688528e-05,
+      "loss": 1.0388,
+      "step": 225620
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.915665416698553e-05,
+      "loss": 1.0303,
+      "step": 225630
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.9155506237085786e-05,
+      "loss": 0.9705,
+      "step": 225640
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.9154358307186036e-05,
+      "loss": 0.9822,
+      "step": 225650
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.915321037728629e-05,
+      "loss": 1.0203,
+      "step": 225660
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.915206244738654e-05,
+      "loss": 1.0849,
+      "step": 225670
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.915091451748679e-05,
+      "loss": 1.1945,
+      "step": 225680
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.914976658758705e-05,
+      "loss": 1.09,
+      "step": 225690
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.9148618657687305e-05,
+      "loss": 1.1261,
+      "step": 225700
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.914747072778755e-05,
+      "loss": 1.024,
+      "step": 225710
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.9146322797887805e-05,
+      "loss": 0.9891,
+      "step": 225720
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.914517486798806e-05,
+      "loss": 1.1527,
+      "step": 225730
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.9144026938088305e-05,
+      "loss": 1.147,
+      "step": 225740
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.914287900818856e-05,
+      "loss": 0.9742,
+      "step": 225750
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.914173107828882e-05,
+      "loss": 1.19,
+      "step": 225760
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.914058314838907e-05,
+      "loss": 1.012,
+      "step": 225770
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.913943521848932e-05,
+      "loss": 1.1163,
+      "step": 225780
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.9138287288589574e-05,
+      "loss": 1.072,
+      "step": 225790
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.9137139358689824e-05,
+      "loss": 1.019,
+      "step": 225800
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.913599142879008e-05,
+      "loss": 0.9988,
+      "step": 225810
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.913484349889033e-05,
+      "loss": 1.0702,
+      "step": 225820
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.913369556899058e-05,
+      "loss": 1.0758,
+      "step": 225830
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.913254763909084e-05,
+      "loss": 1.1639,
+      "step": 225840
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.9131399709191094e-05,
+      "loss": 1.0742,
+      "step": 225850
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.913025177929134e-05,
+      "loss": 1.027,
+      "step": 225860
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.9129103849391594e-05,
+      "loss": 1.0944,
+      "step": 225870
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.912795591949185e-05,
+      "loss": 1.2122,
+      "step": 225880
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.912680798959209e-05,
+      "loss": 1.0427,
+      "step": 225890
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.912566005969235e-05,
+      "loss": 0.9655,
+      "step": 225900
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.912451212979261e-05,
+      "loss": 1.0166,
+      "step": 225910
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.9123364199892856e-05,
+      "loss": 0.9374,
+      "step": 225920
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.9122216269993106e-05,
+      "loss": 1.0921,
+      "step": 225930
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.912106834009336e-05,
+      "loss": 1.07,
+      "step": 225940
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.911992041019361e-05,
+      "loss": 1.1523,
+      "step": 225950
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.911877248029387e-05,
+      "loss": 1.0392,
+      "step": 225960
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.911762455039412e-05,
+      "loss": 0.9728,
+      "step": 225970
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.911647662049437e-05,
+      "loss": 1.0156,
+      "step": 225980
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.9115328690594626e-05,
+      "loss": 1.2142,
+      "step": 225990
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.911418076069488e-05,
+      "loss": 0.9773,
+      "step": 226000
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.9113032830795125e-05,
+      "loss": 1.0269,
+      "step": 226010
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.911188490089538e-05,
+      "loss": 1.1498,
+      "step": 226020
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.911073697099564e-05,
+      "loss": 0.9117,
+      "step": 226030
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.910958904109588e-05,
+      "loss": 1.0452,
+      "step": 226040
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.910844111119614e-05,
+      "loss": 1.0169,
+      "step": 226050
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.9107293181296395e-05,
+      "loss": 0.9893,
+      "step": 226060
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.9106145251396645e-05,
+      "loss": 1.1467,
+      "step": 226070
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.9104997321496895e-05,
+      "loss": 1.0751,
+      "step": 226080
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.910384939159715e-05,
+      "loss": 1.1916,
+      "step": 226090
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.91027014616974e-05,
+      "loss": 1.1278,
+      "step": 226100
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.910155353179766e-05,
+      "loss": 1.0133,
+      "step": 226110
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.910040560189791e-05,
+      "loss": 1.0109,
+      "step": 226120
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.909925767199816e-05,
+      "loss": 1.109,
+      "step": 226130
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.9098109742098414e-05,
+      "loss": 0.9993,
+      "step": 226140
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.909696181219867e-05,
+      "loss": 1.2046,
+      "step": 226150
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.9095813882298914e-05,
+      "loss": 1.124,
+      "step": 226160
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.909466595239917e-05,
+      "loss": 1.041,
+      "step": 226170
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.909351802249943e-05,
+      "loss": 1.1591,
+      "step": 226180
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.909237009259967e-05,
+      "loss": 0.9291,
+      "step": 226190
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.909122216269993e-05,
+      "loss": 1.1434,
+      "step": 226200
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.9090074232800184e-05,
+      "loss": 1.0629,
+      "step": 226210
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.9088926302900433e-05,
+      "loss": 1.084,
+      "step": 226220
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.908777837300068e-05,
+      "loss": 1.0472,
+      "step": 226230
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.908663044310094e-05,
+      "loss": 1.1368,
+      "step": 226240
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.908548251320119e-05,
+      "loss": 1.0433,
+      "step": 226250
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.9084334583301446e-05,
+      "loss": 1.0073,
+      "step": 226260
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.9083186653401696e-05,
+      "loss": 0.9304,
+      "step": 226270
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.9082038723501946e-05,
+      "loss": 1.1264,
+      "step": 226280
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.90808907936022e-05,
+      "loss": 1.0623,
+      "step": 226290
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.9079742863702446e-05,
+      "loss": 1.0171,
+      "step": 226300
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.90785949338027e-05,
+      "loss": 1.091,
+      "step": 226310
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.907744700390296e-05,
+      "loss": 1.186,
+      "step": 226320
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.907629907400321e-05,
+      "loss": 1.0366,
+      "step": 226330
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.907515114410346e-05,
+      "loss": 1.1127,
+      "step": 226340
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.9074003214203715e-05,
+      "loss": 1.1304,
+      "step": 226350
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.9072855284303965e-05,
+      "loss": 1.0974,
+      "step": 226360
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.907170735440422e-05,
+      "loss": 1.149,
+      "step": 226370
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.907055942450447e-05,
+      "loss": 1.0287,
+      "step": 226380
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.906941149460472e-05,
+      "loss": 1.0327,
+      "step": 226390
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.906826356470498e-05,
+      "loss": 1.0328,
+      "step": 226400
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.9067115634805235e-05,
+      "loss": 1.093,
+      "step": 226410
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.906596770490548e-05,
+      "loss": 0.9385,
+      "step": 226420
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.9064819775005735e-05,
+      "loss": 1.1254,
+      "step": 226430
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.906367184510599e-05,
+      "loss": 1.0306,
+      "step": 226440
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.9062523915206234e-05,
+      "loss": 1.0253,
+      "step": 226450
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.906137598530649e-05,
+      "loss": 1.0351,
+      "step": 226460
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.906022805540675e-05,
+      "loss": 1.0909,
+      "step": 226470
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.9059080125507e-05,
+      "loss": 1.01,
+      "step": 226480
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.905793219560725e-05,
+      "loss": 1.1217,
+      "step": 226490
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.9056784265707504e-05,
+      "loss": 1.0208,
+      "step": 226500
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.9055636335807754e-05,
+      "loss": 1.0165,
+      "step": 226510
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.905448840590801e-05,
+      "loss": 1.0557,
+      "step": 226520
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.905334047600826e-05,
+      "loss": 1.1085,
+      "step": 226530
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.905219254610851e-05,
+      "loss": 1.0701,
+      "step": 226540
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.905104461620877e-05,
+      "loss": 0.9236,
+      "step": 226550
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.904989668630902e-05,
+      "loss": 1.0198,
+      "step": 226560
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.9048748756409266e-05,
+      "loss": 1.2089,
+      "step": 226570
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.904760082650952e-05,
+      "loss": 1.077,
+      "step": 226580
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.904645289660978e-05,
+      "loss": 0.9927,
+      "step": 226590
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.904530496671002e-05,
+      "loss": 0.9511,
+      "step": 226600
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.904415703681028e-05,
+      "loss": 0.9809,
+      "step": 226610
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.9043009106910536e-05,
+      "loss": 1.1224,
+      "step": 226620
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.9041861177010786e-05,
+      "loss": 1.023,
+      "step": 226630
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.9040713247111036e-05,
+      "loss": 1.0288,
+      "step": 226640
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.903956531721129e-05,
+      "loss": 1.0379,
+      "step": 226650
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.903841738731154e-05,
+      "loss": 1.1319,
+      "step": 226660
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.90372694574118e-05,
+      "loss": 1.0996,
+      "step": 226670
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.903612152751205e-05,
+      "loss": 0.9263,
+      "step": 226680
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.90349735976123e-05,
+      "loss": 1.158,
+      "step": 226690
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.9033825667712555e-05,
+      "loss": 0.9123,
+      "step": 226700
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.903267773781281e-05,
+      "loss": 1.062,
+      "step": 226710
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.9031529807913055e-05,
+      "loss": 1.0049,
+      "step": 226720
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.903038187801331e-05,
+      "loss": 0.9549,
+      "step": 226730
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.902923394811357e-05,
+      "loss": 1.2043,
+      "step": 226740
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.902808601821381e-05,
+      "loss": 1.0289,
+      "step": 226750
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.902693808831407e-05,
+      "loss": 1.0558,
+      "step": 226760
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.9025790158414324e-05,
+      "loss": 1.048,
+      "step": 226770
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.9024642228514574e-05,
+      "loss": 1.0114,
+      "step": 226780
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.9023494298614824e-05,
+      "loss": 1.0575,
+      "step": 226790
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.902234636871508e-05,
+      "loss": 0.9776,
+      "step": 226800
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.902119843881533e-05,
+      "loss": 1.1206,
+      "step": 226810
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.902005050891559e-05,
+      "loss": 1.1704,
+      "step": 226820
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.901890257901584e-05,
+      "loss": 0.9475,
+      "step": 226830
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.901775464911609e-05,
+      "loss": 0.9881,
+      "step": 226840
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.9016606719216344e-05,
+      "loss": 0.9468,
+      "step": 226850
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.90154587893166e-05,
+      "loss": 1.0572,
+      "step": 226860
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.901431085941684e-05,
+      "loss": 1.0276,
+      "step": 226870
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.90131629295171e-05,
+      "loss": 1.0393,
+      "step": 226880
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.901201499961736e-05,
+      "loss": 1.1446,
+      "step": 226890
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.90108670697176e-05,
+      "loss": 1.1794,
+      "step": 226900
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.9009719139817856e-05,
+      "loss": 1.1025,
+      "step": 226910
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.900857120991811e-05,
+      "loss": 1.0983,
+      "step": 226920
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.900742328001836e-05,
+      "loss": 1.003,
+      "step": 226930
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.900627535011861e-05,
+      "loss": 1.0473,
+      "step": 226940
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.900512742021887e-05,
+      "loss": 1.0758,
+      "step": 226950
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.900397949031912e-05,
+      "loss": 1.165,
+      "step": 226960
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.9002831560419376e-05,
+      "loss": 1.1504,
+      "step": 226970
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.9001683630519626e-05,
+      "loss": 1.0342,
+      "step": 226980
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.9000535700619875e-05,
+      "loss": 1.037,
+      "step": 226990
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.899938777072013e-05,
+      "loss": 1.0173,
+      "step": 227000
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.899823984082039e-05,
+      "loss": 1.0299,
+      "step": 227010
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.899709191092063e-05,
+      "loss": 0.9838,
+      "step": 227020
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.899594398102089e-05,
+      "loss": 1.0196,
+      "step": 227030
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.8994796051121145e-05,
+      "loss": 1.0491,
+      "step": 227040
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.899364812122139e-05,
+      "loss": 1.0347,
+      "step": 227050
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.8992500191321645e-05,
+      "loss": 1.0617,
+      "step": 227060
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.89913522614219e-05,
+      "loss": 1.1814,
+      "step": 227070
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.899020433152215e-05,
+      "loss": 1.0312,
+      "step": 227080
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.89890564016224e-05,
+      "loss": 1.1069,
+      "step": 227090
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.898790847172266e-05,
+      "loss": 1.0563,
+      "step": 227100
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.898676054182291e-05,
+      "loss": 1.0037,
+      "step": 227110
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.8985612611923164e-05,
+      "loss": 1.0966,
+      "step": 227120
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.8984464682023414e-05,
+      "loss": 1.0545,
+      "step": 227130
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.8983316752123664e-05,
+      "loss": 1.1635,
+      "step": 227140
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.898216882222392e-05,
+      "loss": 1.0714,
+      "step": 227150
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.898102089232418e-05,
+      "loss": 1.0994,
+      "step": 227160
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.897987296242442e-05,
+      "loss": 1.0804,
+      "step": 227170
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.897872503252468e-05,
+      "loss": 1.1362,
+      "step": 227180
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.8977577102624934e-05,
+      "loss": 1.2078,
+      "step": 227190
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.897642917272518e-05,
+      "loss": 1.0246,
+      "step": 227200
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.897528124282543e-05,
+      "loss": 1.1949,
+      "step": 227210
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.897413331292569e-05,
+      "loss": 1.0479,
+      "step": 227220
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.897298538302594e-05,
+      "loss": 0.989,
+      "step": 227230
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.897183745312619e-05,
+      "loss": 1.0521,
+      "step": 227240
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.8970689523226446e-05,
+      "loss": 1.0526,
+      "step": 227250
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.8969541593326696e-05,
+      "loss": 1.0052,
+      "step": 227260
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.896839366342695e-05,
+      "loss": 1.1001,
+      "step": 227270
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.89672457335272e-05,
+      "loss": 1.1377,
+      "step": 227280
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.896609780362745e-05,
+      "loss": 0.9702,
+      "step": 227290
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.896494987372771e-05,
+      "loss": 1.0473,
+      "step": 227300
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.8963801943827966e-05,
+      "loss": 1.0369,
+      "step": 227310
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.896265401392821e-05,
+      "loss": 1.0373,
+      "step": 227320
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.8961506084028465e-05,
+      "loss": 1.1117,
+      "step": 227330
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.896035815412872e-05,
+      "loss": 0.8689,
+      "step": 227340
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.8959210224228965e-05,
+      "loss": 1.0994,
+      "step": 227350
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.895806229432922e-05,
+      "loss": 1.0882,
+      "step": 227360
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.895691436442948e-05,
+      "loss": 1.1329,
+      "step": 227370
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.895576643452973e-05,
+      "loss": 1.0415,
+      "step": 227380
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.895461850462998e-05,
+      "loss": 1.0369,
+      "step": 227390
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.8953470574730235e-05,
+      "loss": 1.1029,
+      "step": 227400
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.8952322644830485e-05,
+      "loss": 1.1368,
+      "step": 227410
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.895117471493074e-05,
+      "loss": 0.9712,
+      "step": 227420
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.895002678503099e-05,
+      "loss": 1.1918,
+      "step": 227430
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.894887885513124e-05,
+      "loss": 1.01,
+      "step": 227440
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.89477309252315e-05,
+      "loss": 1.0296,
+      "step": 227450
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.8946582995331754e-05,
+      "loss": 0.9878,
+      "step": 227460
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.8945435065432e-05,
+      "loss": 1.1144,
+      "step": 227470
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.8944287135532254e-05,
+      "loss": 1.0329,
+      "step": 227480
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.894313920563251e-05,
+      "loss": 1.1002,
+      "step": 227490
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.8941991275732754e-05,
+      "loss": 1.0927,
+      "step": 227500
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.894084334583301e-05,
+      "loss": 1.1494,
+      "step": 227510
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.893969541593327e-05,
+      "loss": 1.0779,
+      "step": 227520
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.893854748603352e-05,
+      "loss": 0.9666,
+      "step": 227530
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.8937399556133767e-05,
+      "loss": 1.0825,
+      "step": 227540
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.893625162623402e-05,
+      "loss": 0.9952,
+      "step": 227550
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.893510369633427e-05,
+      "loss": 1.0039,
+      "step": 227560
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.893395576643453e-05,
+      "loss": 1.0624,
+      "step": 227570
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.893280783653478e-05,
+      "loss": 1.0173,
+      "step": 227580
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.893165990663503e-05,
+      "loss": 1.0508,
+      "step": 227590
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.8930511976735286e-05,
+      "loss": 0.9963,
+      "step": 227600
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.892936404683554e-05,
+      "loss": 1.1072,
+      "step": 227610
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.8928216116935786e-05,
+      "loss": 1.024,
+      "step": 227620
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.892706818703604e-05,
+      "loss": 1.0668,
+      "step": 227630
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.89259202571363e-05,
+      "loss": 1.1395,
+      "step": 227640
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.892477232723654e-05,
+      "loss": 1.0296,
+      "step": 227650
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.89236243973368e-05,
+      "loss": 1.1421,
+      "step": 227660
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.892247646743705e-05,
+      "loss": 1.0365,
+      "step": 227670
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.8921328537537305e-05,
+      "loss": 1.1296,
+      "step": 227680
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.8920180607637555e-05,
+      "loss": 1.1315,
+      "step": 227690
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.8919032677737805e-05,
+      "loss": 1.0262,
+      "step": 227700
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.891788474783806e-05,
+      "loss": 0.9688,
+      "step": 227710
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.891673681793832e-05,
+      "loss": 1.1342,
+      "step": 227720
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.891558888803856e-05,
+      "loss": 1.0771,
+      "step": 227730
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.891444095813882e-05,
+      "loss": 0.9932,
+      "step": 227740
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.8913293028239074e-05,
+      "loss": 1.0527,
+      "step": 227750
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.891214509833932e-05,
+      "loss": 1.0514,
+      "step": 227760
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.8910997168439574e-05,
+      "loss": 0.995,
+      "step": 227770
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.890984923853983e-05,
+      "loss": 1.059,
+      "step": 227780
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.890870130864008e-05,
+      "loss": 1.1543,
+      "step": 227790
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.890755337874033e-05,
+      "loss": 1.0007,
+      "step": 227800
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.890640544884059e-05,
+      "loss": 1.0798,
+      "step": 227810
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.890525751894084e-05,
+      "loss": 1.0734,
+      "step": 227820
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.8904109589041094e-05,
+      "loss": 0.9188,
+      "step": 227830
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.8902961659141344e-05,
+      "loss": 1.1501,
+      "step": 227840
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.890181372924159e-05,
+      "loss": 1.0197,
+      "step": 227850
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.890066579934185e-05,
+      "loss": 1.0924,
+      "step": 227860
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.889951786944211e-05,
+      "loss": 1.2052,
+      "step": 227870
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.889836993954235e-05,
+      "loss": 0.9843,
+      "step": 227880
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.8897222009642606e-05,
+      "loss": 1.1138,
+      "step": 227890
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.889607407974286e-05,
+      "loss": 1.1464,
+      "step": 227900
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.8894926149843106e-05,
+      "loss": 1.0125,
+      "step": 227910
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.889377821994336e-05,
+      "loss": 1.0459,
+      "step": 227920
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.889263029004362e-05,
+      "loss": 0.945,
+      "step": 227930
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.889148236014387e-05,
+      "loss": 1.1506,
+      "step": 227940
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.889033443024412e-05,
+      "loss": 1.1118,
+      "step": 227950
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.8889186500344376e-05,
+      "loss": 1.1382,
+      "step": 227960
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.8888038570444625e-05,
+      "loss": 1.0864,
+      "step": 227970
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.888689064054488e-05,
+      "loss": 1.1516,
+      "step": 227980
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.888574271064513e-05,
+      "loss": 1.0442,
+      "step": 227990
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.888459478074538e-05,
+      "loss": 1.0395,
+      "step": 228000
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.888344685084564e-05,
+      "loss": 1.0122,
+      "step": 228010
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.8882298920945895e-05,
+      "loss": 1.0684,
+      "step": 228020
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.888115099104614e-05,
+      "loss": 1.2126,
+      "step": 228030
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.8880003061146395e-05,
+      "loss": 1.0492,
+      "step": 228040
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.887885513124665e-05,
+      "loss": 0.9358,
+      "step": 228050
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.8877707201346895e-05,
+      "loss": 0.9457,
+      "step": 228060
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.887655927144715e-05,
+      "loss": 0.9668,
+      "step": 228070
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.887541134154741e-05,
+      "loss": 0.981,
+      "step": 228080
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.887426341164766e-05,
+      "loss": 1.0602,
+      "step": 228090
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.887311548174791e-05,
+      "loss": 0.9987,
+      "step": 228100
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.8871967551848164e-05,
+      "loss": 1.0957,
+      "step": 228110
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.8870819621948414e-05,
+      "loss": 1.0947,
+      "step": 228120
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.886967169204867e-05,
+      "loss": 1.1417,
+      "step": 228130
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.886852376214892e-05,
+      "loss": 1.0492,
+      "step": 228140
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.886737583224917e-05,
+      "loss": 1.0182,
+      "step": 228150
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.886622790234943e-05,
+      "loss": 1.162,
+      "step": 228160
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.8865079972449684e-05,
+      "loss": 1.0073,
+      "step": 228170
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.886393204254993e-05,
+      "loss": 1.0465,
+      "step": 228180
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.886278411265018e-05,
+      "loss": 1.1014,
+      "step": 228190
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.886163618275044e-05,
+      "loss": 1.1347,
+      "step": 228200
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.886048825285068e-05,
+      "loss": 1.0055,
+      "step": 228210
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.885934032295094e-05,
+      "loss": 1.0767,
+      "step": 228220
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.8858192393051196e-05,
+      "loss": 1.0076,
+      "step": 228230
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.8857044463151446e-05,
+      "loss": 1.0825,
+      "step": 228240
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.8855896533251696e-05,
+      "loss": 1.0728,
+      "step": 228250
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.885474860335195e-05,
+      "loss": 1.1586,
+      "step": 228260
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.88536006734522e-05,
+      "loss": 1.0201,
+      "step": 228270
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.885245274355246e-05,
+      "loss": 1.098,
+      "step": 228280
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.885130481365271e-05,
+      "loss": 1.0497,
+      "step": 228290
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.885015688375296e-05,
+      "loss": 1.116,
+      "step": 228300
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.8849008953853215e-05,
+      "loss": 1.0768,
+      "step": 228310
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.884786102395347e-05,
+      "loss": 1.1013,
+      "step": 228320
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.8846713094053715e-05,
+      "loss": 1.0502,
+      "step": 228330
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.884556516415397e-05,
+      "loss": 1.0517,
+      "step": 228340
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.884441723425423e-05,
+      "loss": 1.1163,
+      "step": 228350
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.884326930435447e-05,
+      "loss": 1.0574,
+      "step": 228360
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.884212137445473e-05,
+      "loss": 1.236,
+      "step": 228370
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.8840973444554985e-05,
+      "loss": 1.0829,
+      "step": 228380
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.8839825514655235e-05,
+      "loss": 1.1298,
+      "step": 228390
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.8838677584755484e-05,
+      "loss": 1.1348,
+      "step": 228400
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.883752965485574e-05,
+      "loss": 1.1082,
+      "step": 228410
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.883638172495599e-05,
+      "loss": 1.0403,
+      "step": 228420
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.883523379505625e-05,
+      "loss": 0.9749,
+      "step": 228430
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.88340858651565e-05,
+      "loss": 1.0007,
+      "step": 228440
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.883293793525675e-05,
+      "loss": 1.0771,
+      "step": 228450
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.8831790005357004e-05,
+      "loss": 0.968,
+      "step": 228460
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.883064207545726e-05,
+      "loss": 1.1404,
+      "step": 228470
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.8829494145557504e-05,
+      "loss": 1.0047,
+      "step": 228480
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.882834621565776e-05,
+      "loss": 1.076,
+      "step": 228490
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.882719828575802e-05,
+      "loss": 1.076,
+      "step": 228500
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.882605035585826e-05,
+      "loss": 1.1356,
+      "step": 228510
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.8824902425958517e-05,
+      "loss": 1.139,
+      "step": 228520
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.882375449605877e-05,
+      "loss": 1.0094,
+      "step": 228530
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.882260656615902e-05,
+      "loss": 1.056,
+      "step": 228540
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.882145863625927e-05,
+      "loss": 1.0185,
+      "step": 228550
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.882031070635953e-05,
+      "loss": 1.0846,
+      "step": 228560
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.881916277645978e-05,
+      "loss": 1.0481,
+      "step": 228570
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.8818014846560036e-05,
+      "loss": 1.1432,
+      "step": 228580
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.8816866916660286e-05,
+      "loss": 0.999,
+      "step": 228590
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.8815718986760536e-05,
+      "loss": 1.1361,
+      "step": 228600
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.881457105686079e-05,
+      "loss": 1.1289,
+      "step": 228610
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.881342312696105e-05,
+      "loss": 0.9812,
+      "step": 228620
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.881227519706129e-05,
+      "loss": 1.0704,
+      "step": 228630
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.881112726716155e-05,
+      "loss": 1.0229,
+      "step": 228640
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.8809979337261805e-05,
+      "loss": 0.9906,
+      "step": 228650
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.880883140736205e-05,
+      "loss": 1.0766,
+      "step": 228660
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.8807683477462305e-05,
+      "loss": 1.076,
+      "step": 228670
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.880653554756256e-05,
+      "loss": 1.1636,
+      "step": 228680
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.880538761766281e-05,
+      "loss": 1.0965,
+      "step": 228690
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.880423968776306e-05,
+      "loss": 1.0796,
+      "step": 228700
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.880309175786332e-05,
+      "loss": 1.0972,
+      "step": 228710
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.880194382796357e-05,
+      "loss": 1.0456,
+      "step": 228720
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.8800795898063824e-05,
+      "loss": 1.0578,
+      "step": 228730
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.8799647968164074e-05,
+      "loss": 1.0024,
+      "step": 228740
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.8798500038264324e-05,
+      "loss": 1.0922,
+      "step": 228750
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.879735210836458e-05,
+      "loss": 1.1914,
+      "step": 228760
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.879620417846484e-05,
+      "loss": 0.9989,
+      "step": 228770
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.879505624856508e-05,
+      "loss": 1.2365,
+      "step": 228780
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.879390831866534e-05,
+      "loss": 1.0245,
+      "step": 228790
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.8792760388765594e-05,
+      "loss": 1.1173,
+      "step": 228800
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.879161245886584e-05,
+      "loss": 1.0972,
+      "step": 228810
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.8790464528966093e-05,
+      "loss": 0.9974,
+      "step": 228820
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.878931659906635e-05,
+      "loss": 1.0585,
+      "step": 228830
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.87881686691666e-05,
+      "loss": 1.0598,
+      "step": 228840
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.878702073926685e-05,
+      "loss": 1.0697,
+      "step": 228850
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.8785872809367106e-05,
+      "loss": 1.0105,
+      "step": 228860
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.8784724879467356e-05,
+      "loss": 1.1019,
+      "step": 228870
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.878357694956761e-05,
+      "loss": 1.0436,
+      "step": 228880
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.878242901966786e-05,
+      "loss": 1.0401,
+      "step": 228890
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.878128108976811e-05,
+      "loss": 1.1027,
+      "step": 228900
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.878013315986837e-05,
+      "loss": 1.0303,
+      "step": 228910
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.8778985229968626e-05,
+      "loss": 1.0467,
+      "step": 228920
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.877783730006887e-05,
+      "loss": 1.2667,
+      "step": 228930
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.8776689370169126e-05,
+      "loss": 1.0721,
+      "step": 228940
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.877554144026938e-05,
+      "loss": 1.0354,
+      "step": 228950
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.8774393510369625e-05,
+      "loss": 1.0441,
+      "step": 228960
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.877324558046988e-05,
+      "loss": 1.0621,
+      "step": 228970
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.877209765057014e-05,
+      "loss": 1.0339,
+      "step": 228980
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.877094972067039e-05,
+      "loss": 1.0629,
+      "step": 228990
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.876980179077064e-05,
+      "loss": 0.9855,
+      "step": 229000
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.8768653860870895e-05,
+      "loss": 1.044,
+      "step": 229010
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.8767505930971145e-05,
+      "loss": 1.0266,
+      "step": 229020
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.87663580010714e-05,
+      "loss": 1.1215,
+      "step": 229030
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.8765210071171645e-05,
+      "loss": 1.1209,
+      "step": 229040
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.87640621412719e-05,
+      "loss": 1.0411,
+      "step": 229050
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.876291421137216e-05,
+      "loss": 1.1155,
+      "step": 229060
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.87617662814724e-05,
+      "loss": 0.9788,
+      "step": 229070
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.876061835157266e-05,
+      "loss": 1.0477,
+      "step": 229080
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.8759470421672914e-05,
+      "loss": 1.0662,
+      "step": 229090
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.8758322491773164e-05,
+      "loss": 1.0621,
+      "step": 229100
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.8757174561873414e-05,
+      "loss": 1.0163,
+      "step": 229110
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.875602663197367e-05,
+      "loss": 1.1066,
+      "step": 229120
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.875487870207392e-05,
+      "loss": 1.1332,
+      "step": 229130
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.875373077217418e-05,
+      "loss": 1.0225,
+      "step": 229140
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.875258284227443e-05,
+      "loss": 1.0769,
+      "step": 229150
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.875143491237468e-05,
+      "loss": 1.0337,
+      "step": 229160
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.875028698247493e-05,
+      "loss": 0.9745,
+      "step": 229170
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.874913905257519e-05,
+      "loss": 1.0956,
+      "step": 229180
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.874799112267543e-05,
+      "loss": 1.0326,
+      "step": 229190
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.874684319277569e-05,
+      "loss": 1.0305,
+      "step": 229200
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.8745695262875946e-05,
+      "loss": 1.0497,
+      "step": 229210
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.874454733297619e-05,
+      "loss": 1.1839,
+      "step": 229220
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.8743399403076446e-05,
+      "loss": 0.998,
+      "step": 229230
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.87422514731767e-05,
+      "loss": 1.1375,
+      "step": 229240
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.874110354327695e-05,
+      "loss": 1.1356,
+      "step": 229250
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.87399556133772e-05,
+      "loss": 0.943,
+      "step": 229260
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.873880768347746e-05,
+      "loss": 1.1895,
+      "step": 229270
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.873765975357771e-05,
+      "loss": 1.036,
+      "step": 229280
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.8736511823677965e-05,
+      "loss": 1.0936,
+      "step": 229290
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.8735363893778215e-05,
+      "loss": 1.0245,
+      "step": 229300
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.8734215963878465e-05,
+      "loss": 1.128,
+      "step": 229310
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.873306803397872e-05,
+      "loss": 1.0952,
+      "step": 229320
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.873192010407898e-05,
+      "loss": 1.0023,
+      "step": 229330
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.873077217417922e-05,
+      "loss": 1.1832,
+      "step": 229340
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.872962424427948e-05,
+      "loss": 0.9876,
+      "step": 229350
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.8728476314379735e-05,
+      "loss": 1.1068,
+      "step": 229360
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.872732838447998e-05,
+      "loss": 1.0183,
+      "step": 229370
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.8726180454580234e-05,
+      "loss": 1.0428,
+      "step": 229380
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.872503252468049e-05,
+      "loss": 1.1297,
+      "step": 229390
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.872388459478074e-05,
+      "loss": 1.0643,
+      "step": 229400
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.872273666488099e-05,
+      "loss": 1.1321,
+      "step": 229410
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.872158873498125e-05,
+      "loss": 1.0691,
+      "step": 229420
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.87204408050815e-05,
+      "loss": 1.1612,
+      "step": 229430
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.8719292875181754e-05,
+      "loss": 1.147,
+      "step": 229440
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.8718144945282004e-05,
+      "loss": 1.0257,
+      "step": 229450
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.8716997015382254e-05,
+      "loss": 1.0775,
+      "step": 229460
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.871584908548251e-05,
+      "loss": 0.9529,
+      "step": 229470
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.871470115558277e-05,
+      "loss": 0.9419,
+      "step": 229480
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.871355322568301e-05,
+      "loss": 1.0573,
+      "step": 229490
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.8712405295783267e-05,
+      "loss": 1.1889,
+      "step": 229500
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.871125736588352e-05,
+      "loss": 1.1947,
+      "step": 229510
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.8710109435983766e-05,
+      "loss": 1.1573,
+      "step": 229520
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.870896150608402e-05,
+      "loss": 1.0968,
+      "step": 229530
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.870781357618428e-05,
+      "loss": 1.1978,
+      "step": 229540
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.870666564628453e-05,
+      "loss": 1.0289,
+      "step": 229550
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.870551771638478e-05,
+      "loss": 1.0245,
+      "step": 229560
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.8704369786485036e-05,
+      "loss": 1.0556,
+      "step": 229570
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.8703221856585286e-05,
+      "loss": 1.056,
+      "step": 229580
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.870207392668554e-05,
+      "loss": 1.0498,
+      "step": 229590
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.870092599678579e-05,
+      "loss": 1.0107,
+      "step": 229600
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.869977806688604e-05,
+      "loss": 0.9762,
+      "step": 229610
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.86986301369863e-05,
+      "loss": 0.9717,
+      "step": 229620
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.8697482207086555e-05,
+      "loss": 1.1015,
+      "step": 229630
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.86963342771868e-05,
+      "loss": 1.0543,
+      "step": 229640
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.8695186347287055e-05,
+      "loss": 1.0898,
+      "step": 229650
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.869403841738731e-05,
+      "loss": 0.9513,
+      "step": 229660
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.8692890487487555e-05,
+      "loss": 1.101,
+      "step": 229670
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.869174255758781e-05,
+      "loss": 1.0333,
+      "step": 229680
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.869059462768807e-05,
+      "loss": 1.0537,
+      "step": 229690
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.868944669778832e-05,
+      "loss": 1.0832,
+      "step": 229700
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.868829876788857e-05,
+      "loss": 1.1232,
+      "step": 229710
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.8687150837988824e-05,
+      "loss": 1.0099,
+      "step": 229720
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.8686002908089074e-05,
+      "loss": 1.2155,
+      "step": 229730
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.868485497818933e-05,
+      "loss": 1.0906,
+      "step": 229740
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.868370704828958e-05,
+      "loss": 1.0615,
+      "step": 229750
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.868255911838983e-05,
+      "loss": 1.0232,
+      "step": 229760
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.868141118849009e-05,
+      "loss": 1.0816,
+      "step": 229770
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.8680263258590344e-05,
+      "loss": 1.0853,
+      "step": 229780
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.867911532869059e-05,
+      "loss": 1.0242,
+      "step": 229790
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.8677967398790843e-05,
+      "loss": 1.0556,
+      "step": 229800
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.86768194688911e-05,
+      "loss": 1.0258,
+      "step": 229810
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.867567153899134e-05,
+      "loss": 1.1614,
+      "step": 229820
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.86745236090916e-05,
+      "loss": 1.1283,
+      "step": 229830
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.8673375679191856e-05,
+      "loss": 1.069,
+      "step": 229840
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.8672227749292106e-05,
+      "loss": 1.1875,
+      "step": 229850
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.8671079819392356e-05,
+      "loss": 1.0834,
+      "step": 229860
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.866993188949261e-05,
+      "loss": 1.0255,
+      "step": 229870
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.866878395959286e-05,
+      "loss": 1.1543,
+      "step": 229880
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.866763602969312e-05,
+      "loss": 0.9833,
+      "step": 229890
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.866648809979337e-05,
+      "loss": 1.0063,
+      "step": 229900
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.866534016989362e-05,
+      "loss": 1.122,
+      "step": 229910
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.8664192239993876e-05,
+      "loss": 1.111,
+      "step": 229920
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.866304431009413e-05,
+      "loss": 1.1728,
+      "step": 229930
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.8661896380194375e-05,
+      "loss": 1.0368,
+      "step": 229940
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.866074845029463e-05,
+      "loss": 1.0683,
+      "step": 229950
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.865960052039489e-05,
+      "loss": 1.1032,
+      "step": 229960
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.865845259049513e-05,
+      "loss": 1.1424,
+      "step": 229970
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.865730466059539e-05,
+      "loss": 0.9913,
+      "step": 229980
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.8656156730695645e-05,
+      "loss": 1.1032,
+      "step": 229990
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.8655008800795895e-05,
+      "loss": 1.1073,
+      "step": 230000
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.8653860870896145e-05,
+      "loss": 1.1691,
+      "step": 230010
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.86527129409964e-05,
+      "loss": 1.1185,
+      "step": 230020
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.865156501109665e-05,
+      "loss": 1.1163,
+      "step": 230030
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.865041708119691e-05,
+      "loss": 1.0797,
+      "step": 230040
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.864926915129716e-05,
+      "loss": 1.1306,
+      "step": 230050
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.864812122139741e-05,
+      "loss": 1.0522,
+      "step": 230060
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.8646973291497664e-05,
+      "loss": 0.9618,
+      "step": 230070
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.864582536159792e-05,
+      "loss": 1.1247,
+      "step": 230080
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.8644677431698164e-05,
+      "loss": 0.9593,
+      "step": 230090
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.864352950179842e-05,
+      "loss": 1.1279,
+      "step": 230100
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.864238157189868e-05,
+      "loss": 1.1391,
+      "step": 230110
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.864123364199892e-05,
+      "loss": 1.0148,
+      "step": 230120
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.864008571209918e-05,
+      "loss": 1.0447,
+      "step": 230130
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.8638937782199433e-05,
+      "loss": 1.1142,
+      "step": 230140
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.863778985229968e-05,
+      "loss": 0.9794,
+      "step": 230150
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.863664192239993e-05,
+      "loss": 1.0416,
+      "step": 230160
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.863549399250019e-05,
+      "loss": 1.0615,
+      "step": 230170
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.863434606260044e-05,
+      "loss": 1.1092,
+      "step": 230180
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.8633198132700696e-05,
+      "loss": 1.1256,
+      "step": 230190
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.8632050202800946e-05,
+      "loss": 1.0673,
+      "step": 230200
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.8630902272901196e-05,
+      "loss": 0.9729,
+      "step": 230210
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.862975434300145e-05,
+      "loss": 1.0699,
+      "step": 230220
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.862860641310171e-05,
+      "loss": 1.0433,
+      "step": 230230
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.862745848320195e-05,
+      "loss": 1.1808,
+      "step": 230240
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.862631055330221e-05,
+      "loss": 1.1182,
+      "step": 230250
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.8625162623402466e-05,
+      "loss": 1.1778,
+      "step": 230260
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.862401469350271e-05,
+      "loss": 1.0764,
+      "step": 230270
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.8622866763602965e-05,
+      "loss": 1.0829,
+      "step": 230280
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.862171883370322e-05,
+      "loss": 1.0187,
+      "step": 230290
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.862057090380347e-05,
+      "loss": 1.0983,
+      "step": 230300
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.861942297390372e-05,
+      "loss": 1.0668,
+      "step": 230310
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.861827504400398e-05,
+      "loss": 1.0722,
+      "step": 230320
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.861712711410423e-05,
+      "loss": 1.0876,
+      "step": 230330
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.8615979184204485e-05,
+      "loss": 0.9815,
+      "step": 230340
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.8614831254304735e-05,
+      "loss": 1.0559,
+      "step": 230350
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.8613683324404984e-05,
+      "loss": 1.0532,
+      "step": 230360
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.861253539450524e-05,
+      "loss": 1.043,
+      "step": 230370
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.86113874646055e-05,
+      "loss": 1.0224,
+      "step": 230380
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.861023953470574e-05,
+      "loss": 1.057,
+      "step": 230390
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.8609091604806e-05,
+      "loss": 0.9579,
+      "step": 230400
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.860794367490625e-05,
+      "loss": 1.0954,
+      "step": 230410
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.86067957450065e-05,
+      "loss": 1.1009,
+      "step": 230420
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.8605647815106754e-05,
+      "loss": 1.1232,
+      "step": 230430
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.8604499885207004e-05,
+      "loss": 1.1483,
+      "step": 230440
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.860335195530726e-05,
+      "loss": 0.9644,
+      "step": 230450
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.860220402540751e-05,
+      "loss": 1.0662,
+      "step": 230460
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.860105609550776e-05,
+      "loss": 1.0917,
+      "step": 230470
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.8599908165608017e-05,
+      "loss": 1.1235,
+      "step": 230480
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.859876023570827e-05,
+      "loss": 1.0237,
+      "step": 230490
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.8597612305808516e-05,
+      "loss": 0.9358,
+      "step": 230500
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.859646437590877e-05,
+      "loss": 0.8883,
+      "step": 230510
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.859531644600903e-05,
+      "loss": 1.0052,
+      "step": 230520
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.859416851610927e-05,
+      "loss": 1.1327,
+      "step": 230530
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.859302058620953e-05,
+      "loss": 1.0797,
+      "step": 230540
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.8591872656309786e-05,
+      "loss": 1.0556,
+      "step": 230550
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.8590724726410036e-05,
+      "loss": 0.9884,
+      "step": 230560
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.8589576796510286e-05,
+      "loss": 1.2575,
+      "step": 230570
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.858842886661054e-05,
+      "loss": 1.07,
+      "step": 230580
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.858728093671079e-05,
+      "loss": 1.1349,
+      "step": 230590
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.858613300681105e-05,
+      "loss": 1.0226,
+      "step": 230600
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.85849850769113e-05,
+      "loss": 1.0971,
+      "step": 230610
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.858383714701155e-05,
+      "loss": 1.0577,
+      "step": 230620
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.8582689217111805e-05,
+      "loss": 1.0039,
+      "step": 230630
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.858154128721206e-05,
+      "loss": 1.0627,
+      "step": 230640
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.8580393357312305e-05,
+      "loss": 1.0835,
+      "step": 230650
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.857924542741256e-05,
+      "loss": 1.094,
+      "step": 230660
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.857809749751282e-05,
+      "loss": 1.1025,
+      "step": 230670
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.857694956761306e-05,
+      "loss": 1.1706,
+      "step": 230680
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.857580163771332e-05,
+      "loss": 1.0271,
+      "step": 230690
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.8574653707813574e-05,
+      "loss": 1.0769,
+      "step": 230700
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.8573505777913824e-05,
+      "loss": 1.0395,
+      "step": 230710
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.8572357848014074e-05,
+      "loss": 1.0468,
+      "step": 230720
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.857120991811433e-05,
+      "loss": 1.0445,
+      "step": 230730
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.857006198821458e-05,
+      "loss": 1.0823,
+      "step": 230740
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.856891405831484e-05,
+      "loss": 1.1154,
+      "step": 230750
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.856776612841509e-05,
+      "loss": 1.1336,
+      "step": 230760
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.856661819851534e-05,
+      "loss": 0.9865,
+      "step": 230770
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.8565470268615593e-05,
+      "loss": 1.0664,
+      "step": 230780
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.856432233871585e-05,
+      "loss": 1.1303,
+      "step": 230790
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.856317440881609e-05,
+      "loss": 1.0905,
+      "step": 230800
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.856202647891635e-05,
+      "loss": 1.1119,
+      "step": 230810
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.8560878549016606e-05,
+      "loss": 1.0698,
+      "step": 230820
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.855973061911685e-05,
+      "loss": 1.0364,
+      "step": 230830
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.8558582689217106e-05,
+      "loss": 1.0666,
+      "step": 230840
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.855743475931736e-05,
+      "loss": 1.0909,
+      "step": 230850
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.855628682941761e-05,
+      "loss": 1.0949,
+      "step": 230860
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.855513889951786e-05,
+      "loss": 0.9745,
+      "step": 230870
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.855399096961812e-05,
+      "loss": 1.1616,
+      "step": 230880
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.855284303971837e-05,
+      "loss": 1.1139,
+      "step": 230890
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.8551695109818626e-05,
+      "loss": 1.1671,
+      "step": 230900
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.8550547179918875e-05,
+      "loss": 1.0867,
+      "step": 230910
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.8549399250019125e-05,
+      "loss": 1.0851,
+      "step": 230920
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.854825132011938e-05,
+      "loss": 1.0605,
+      "step": 230930
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.854710339021964e-05,
+      "loss": 0.9877,
+      "step": 230940
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.854595546031988e-05,
+      "loss": 1.1855,
+      "step": 230950
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.854480753042014e-05,
+      "loss": 1.082,
+      "step": 230960
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.8543659600520395e-05,
+      "loss": 1.1122,
+      "step": 230970
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.854251167062064e-05,
+      "loss": 1.1887,
+      "step": 230980
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.8541363740720895e-05,
+      "loss": 1.0381,
+      "step": 230990
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.854021581082115e-05,
+      "loss": 1.1125,
+      "step": 231000
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.85390678809214e-05,
+      "loss": 1.1247,
+      "step": 231010
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.853791995102165e-05,
+      "loss": 1.0471,
+      "step": 231020
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.853677202112191e-05,
+      "loss": 1.2007,
+      "step": 231030
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.853562409122216e-05,
+      "loss": 1.0458,
+      "step": 231040
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.8534476161322414e-05,
+      "loss": 1.0789,
+      "step": 231050
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.8533328231422664e-05,
+      "loss": 1.0464,
+      "step": 231060
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.8532180301522914e-05,
+      "loss": 1.1138,
+      "step": 231070
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.853103237162317e-05,
+      "loss": 1.2575,
+      "step": 231080
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.852988444172343e-05,
+      "loss": 1.0446,
+      "step": 231090
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.852873651182367e-05,
+      "loss": 1.1304,
+      "step": 231100
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.852758858192393e-05,
+      "loss": 1.0509,
+      "step": 231110
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.8526440652024183e-05,
+      "loss": 1.1818,
+      "step": 231120
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.8525292722124426e-05,
+      "loss": 1.0349,
+      "step": 231130
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.852414479222468e-05,
+      "loss": 1.1464,
+      "step": 231140
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.852299686232494e-05,
+      "loss": 1.0212,
+      "step": 231150
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.852184893242519e-05,
+      "loss": 1.0395,
+      "step": 231160
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.852070100252544e-05,
+      "loss": 1.0377,
+      "step": 231170
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.8519553072625696e-05,
+      "loss": 1.1628,
+      "step": 231180
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.8518405142725946e-05,
+      "loss": 1.1043,
+      "step": 231190
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.85172572128262e-05,
+      "loss": 1.0517,
+      "step": 231200
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.851610928292645e-05,
+      "loss": 1.131,
+      "step": 231210
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.85149613530267e-05,
+      "loss": 1.1452,
+      "step": 231220
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.851381342312696e-05,
+      "loss": 1.0623,
+      "step": 231230
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.8512665493227216e-05,
+      "loss": 1.15,
+      "step": 231240
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.851151756332746e-05,
+      "loss": 0.9611,
+      "step": 231250
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.8510369633427715e-05,
+      "loss": 1.0079,
+      "step": 231260
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.850922170352797e-05,
+      "loss": 1.0032,
+      "step": 231270
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.8508073773628215e-05,
+      "loss": 1.1145,
+      "step": 231280
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.850692584372847e-05,
+      "loss": 1.1408,
+      "step": 231290
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.850577791382873e-05,
+      "loss": 1.0515,
+      "step": 231300
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.850462998392898e-05,
+      "loss": 0.9945,
+      "step": 231310
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.850348205402923e-05,
+      "loss": 1.0781,
+      "step": 231320
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.8502334124129485e-05,
+      "loss": 1.0267,
+      "step": 231330
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.8501186194229734e-05,
+      "loss": 1.1046,
+      "step": 231340
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.850003826432999e-05,
+      "loss": 0.9885,
+      "step": 231350
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.849889033443024e-05,
+      "loss": 1.129,
+      "step": 231360
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.849774240453049e-05,
+      "loss": 1.0976,
+      "step": 231370
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.849659447463075e-05,
+      "loss": 1.1091,
+      "step": 231380
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.8495446544731004e-05,
+      "loss": 1.2014,
+      "step": 231390
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.849429861483125e-05,
+      "loss": 1.0516,
+      "step": 231400
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.8493150684931504e-05,
+      "loss": 1.1241,
+      "step": 231410
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.849200275503176e-05,
+      "loss": 1.1027,
+      "step": 231420
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.8490854825132003e-05,
+      "loss": 1.0279,
+      "step": 231430
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.848970689523226e-05,
+      "loss": 1.1487,
+      "step": 231440
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.848855896533252e-05,
+      "loss": 1.0933,
+      "step": 231450
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.8487411035432767e-05,
+      "loss": 1.103,
+      "step": 231460
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.8486263105533016e-05,
+      "loss": 1.0749,
+      "step": 231470
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.848511517563327e-05,
+      "loss": 1.024,
+      "step": 231480
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.848396724573352e-05,
+      "loss": 1.0573,
+      "step": 231490
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.848281931583378e-05,
+      "loss": 1.2734,
+      "step": 231500
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.848167138593403e-05,
+      "loss": 1.1037,
+      "step": 231510
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.848052345603428e-05,
+      "loss": 1.0793,
+      "step": 231520
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.8479375526134536e-05,
+      "loss": 1.0709,
+      "step": 231530
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.847822759623479e-05,
+      "loss": 1.1699,
+      "step": 231540
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.8477079666335036e-05,
+      "loss": 1.0797,
+      "step": 231550
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.847593173643529e-05,
+      "loss": 1.0605,
+      "step": 231560
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.847478380653555e-05,
+      "loss": 1.003,
+      "step": 231570
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.847363587663579e-05,
+      "loss": 0.9862,
+      "step": 231580
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.847248794673605e-05,
+      "loss": 1.1473,
+      "step": 231590
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.8471340016836305e-05,
+      "loss": 1.1289,
+      "step": 231600
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.8470192086936555e-05,
+      "loss": 1.1717,
+      "step": 231610
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.8469044157036805e-05,
+      "loss": 1.1199,
+      "step": 231620
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.846789622713706e-05,
+      "loss": 1.0695,
+      "step": 231630
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.846674829723731e-05,
+      "loss": 1.0379,
+      "step": 231640
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.846560036733757e-05,
+      "loss": 1.1171,
+      "step": 231650
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.846445243743782e-05,
+      "loss": 1.0232,
+      "step": 231660
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.846330450753807e-05,
+      "loss": 1.0621,
+      "step": 231670
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.8462156577638324e-05,
+      "loss": 1.0277,
+      "step": 231680
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.846100864773858e-05,
+      "loss": 1.1438,
+      "step": 231690
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.8459860717838824e-05,
+      "loss": 1.074,
+      "step": 231700
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.845871278793908e-05,
+      "loss": 1.0095,
+      "step": 231710
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.845756485803934e-05,
+      "loss": 1.1011,
+      "step": 231720
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.845641692813958e-05,
+      "loss": 1.0934,
+      "step": 231730
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.845526899823984e-05,
+      "loss": 1.1876,
+      "step": 231740
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.8454121068340094e-05,
+      "loss": 1.0587,
+      "step": 231750
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.8452973138440343e-05,
+      "loss": 1.0658,
+      "step": 231760
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.845182520854059e-05,
+      "loss": 1.0486,
+      "step": 231770
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.845067727864085e-05,
+      "loss": 1.0954,
+      "step": 231780
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.84495293487411e-05,
+      "loss": 1.0674,
+      "step": 231790
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.8448381418841356e-05,
+      "loss": 1.1605,
+      "step": 231800
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.84472334889416e-05,
+      "loss": 1.0204,
+      "step": 231810
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.8446085559041856e-05,
+      "loss": 1.0679,
+      "step": 231820
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.844493762914211e-05,
+      "loss": 1.0848,
+      "step": 231830
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.8443789699242356e-05,
+      "loss": 1.022,
+      "step": 231840
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.844264176934261e-05,
+      "loss": 1.0077,
+      "step": 231850
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.844149383944287e-05,
+      "loss": 1.2121,
+      "step": 231860
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.844034590954312e-05,
+      "loss": 1.0277,
+      "step": 231870
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.843919797964337e-05,
+      "loss": 0.9707,
+      "step": 231880
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.8438050049743625e-05,
+      "loss": 0.9844,
+      "step": 231890
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.8436902119843875e-05,
+      "loss": 1.0723,
+      "step": 231900
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.843575418994413e-05,
+      "loss": 1.0968,
+      "step": 231910
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.843460626004438e-05,
+      "loss": 1.0924,
+      "step": 231920
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.843345833014463e-05,
+      "loss": 1.2227,
+      "step": 231930
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.843231040024489e-05,
+      "loss": 1.1597,
+      "step": 231940
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.8431162470345145e-05,
+      "loss": 1.0934,
+      "step": 231950
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.843001454044539e-05,
+      "loss": 1.0603,
+      "step": 231960
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.8428866610545645e-05,
+      "loss": 0.9802,
+      "step": 231970
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.84277186806459e-05,
+      "loss": 0.943,
+      "step": 231980
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.8426570750746144e-05,
+      "loss": 1.0723,
+      "step": 231990
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.84254228208464e-05,
+      "loss": 1.1326,
+      "step": 232000
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.842427489094666e-05,
+      "loss": 1.0479,
+      "step": 232010
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.842312696104691e-05,
+      "loss": 0.9422,
+      "step": 232020
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.842197903114716e-05,
+      "loss": 1.0681,
+      "step": 232030
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.8420831101247414e-05,
+      "loss": 1.0694,
+      "step": 232040
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.8419683171347664e-05,
+      "loss": 1.0682,
+      "step": 232050
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.841853524144792e-05,
+      "loss": 1.0604,
+      "step": 232060
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.841738731154817e-05,
+      "loss": 1.1187,
+      "step": 232070
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.841623938164842e-05,
+      "loss": 1.1213,
+      "step": 232080
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.841509145174868e-05,
+      "loss": 0.9607,
+      "step": 232090
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.8413943521848933e-05,
+      "loss": 1.0691,
+      "step": 232100
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.8412795591949176e-05,
+      "loss": 1.0795,
+      "step": 232110
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.841164766204943e-05,
+      "loss": 1.0296,
+      "step": 232120
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.841049973214969e-05,
+      "loss": 1.0947,
+      "step": 232130
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.840935180224993e-05,
+      "loss": 1.1467,
+      "step": 232140
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.840820387235019e-05,
+      "loss": 1.0456,
+      "step": 232150
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.8407055942450446e-05,
+      "loss": 1.1569,
+      "step": 232160
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.8405908012550696e-05,
+      "loss": 1.1167,
+      "step": 232170
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.8404760082650946e-05,
+      "loss": 1.0186,
+      "step": 232180
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.84036121527512e-05,
+      "loss": 0.9007,
+      "step": 232190
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.840246422285145e-05,
+      "loss": 1.0868,
+      "step": 232200
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.840131629295171e-05,
+      "loss": 1.1623,
+      "step": 232210
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.840016836305196e-05,
+      "loss": 1.0116,
+      "step": 232220
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.839902043315221e-05,
+      "loss": 0.9355,
+      "step": 232230
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.8397872503252465e-05,
+      "loss": 1.0911,
+      "step": 232240
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.839672457335272e-05,
+      "loss": 1.1577,
+      "step": 232250
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.8395576643452965e-05,
+      "loss": 1.0548,
+      "step": 232260
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.839442871355322e-05,
+      "loss": 1.0391,
+      "step": 232270
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.839328078365348e-05,
+      "loss": 1.1,
+      "step": 232280
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.839213285375372e-05,
+      "loss": 1.1913,
+      "step": 232290
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.839098492385398e-05,
+      "loss": 1.0178,
+      "step": 232300
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.8389836993954235e-05,
+      "loss": 0.9602,
+      "step": 232310
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.8388689064054484e-05,
+      "loss": 1.1596,
+      "step": 232320
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.8387541134154734e-05,
+      "loss": 0.9158,
+      "step": 232330
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.838639320425499e-05,
+      "loss": 1.1061,
+      "step": 232340
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.838524527435524e-05,
+      "loss": 1.1038,
+      "step": 232350
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.83840973444555e-05,
+      "loss": 1.0075,
+      "step": 232360
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.838294941455575e-05,
+      "loss": 0.9816,
+      "step": 232370
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.8381801484656e-05,
+      "loss": 1.0786,
+      "step": 232380
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.8380653554756254e-05,
+      "loss": 1.0536,
+      "step": 232390
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.837950562485651e-05,
+      "loss": 1.0551,
+      "step": 232400
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.8378357694956753e-05,
+      "loss": 0.9319,
+      "step": 232410
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.837720976505701e-05,
+      "loss": 1.0418,
+      "step": 232420
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.837606183515727e-05,
+      "loss": 1.207,
+      "step": 232430
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.837491390525751e-05,
+      "loss": 1.0255,
+      "step": 232440
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.8373765975357766e-05,
+      "loss": 1.1961,
+      "step": 232450
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.837261804545802e-05,
+      "loss": 1.1576,
+      "step": 232460
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.837147011555827e-05,
+      "loss": 1.1923,
+      "step": 232470
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.837032218565852e-05,
+      "loss": 1.1763,
+      "step": 232480
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.836917425575878e-05,
+      "loss": 1.0562,
+      "step": 232490
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.836802632585903e-05,
+      "loss": 1.0979,
+      "step": 232500
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.8366878395959286e-05,
+      "loss": 1.036,
+      "step": 232510
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.8365730466059536e-05,
+      "loss": 1.1314,
+      "step": 232520
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.8364582536159786e-05,
+      "loss": 1.0139,
+      "step": 232530
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.836343460626004e-05,
+      "loss": 1.0956,
+      "step": 232540
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.83622866763603e-05,
+      "loss": 1.1004,
+      "step": 232550
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.836113874646054e-05,
+      "loss": 1.0633,
+      "step": 232560
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.83599908165608e-05,
+      "loss": 1.1791,
+      "step": 232570
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.8358842886661055e-05,
+      "loss": 1.02,
+      "step": 232580
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.83576949567613e-05,
+      "loss": 1.0613,
+      "step": 232590
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.8356547026861555e-05,
+      "loss": 1.0875,
+      "step": 232600
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.835539909696181e-05,
+      "loss": 1.1555,
+      "step": 232610
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.835425116706206e-05,
+      "loss": 0.9861,
+      "step": 232620
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.835310323716231e-05,
+      "loss": 1.2016,
+      "step": 232630
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.835195530726257e-05,
+      "loss": 1.0072,
+      "step": 232640
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.835080737736282e-05,
+      "loss": 0.9438,
+      "step": 232650
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.8349659447463074e-05,
+      "loss": 0.9539,
+      "step": 232660
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.8348511517563324e-05,
+      "loss": 1.0041,
+      "step": 232670
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.8347363587663574e-05,
+      "loss": 1.0657,
+      "step": 232680
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.834621565776383e-05,
+      "loss": 1.0025,
+      "step": 232690
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.834506772786409e-05,
+      "loss": 1.0342,
+      "step": 232700
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.834391979796433e-05,
+      "loss": 0.963,
+      "step": 232710
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.834277186806459e-05,
+      "loss": 1.1236,
+      "step": 232720
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.8341623938164844e-05,
+      "loss": 1.1136,
+      "step": 232730
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.834047600826509e-05,
+      "loss": 1.1507,
+      "step": 232740
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.833932807836534e-05,
+      "loss": 1.0505,
+      "step": 232750
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.83381801484656e-05,
+      "loss": 0.961,
+      "step": 232760
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.833703221856585e-05,
+      "loss": 1.1188,
+      "step": 232770
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.83358842886661e-05,
+      "loss": 1.1006,
+      "step": 232780
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.8334736358766356e-05,
+      "loss": 1.1053,
+      "step": 232790
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.8333588428866606e-05,
+      "loss": 0.8903,
+      "step": 232800
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.833244049896686e-05,
+      "loss": 1.0669,
+      "step": 232810
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.833129256906711e-05,
+      "loss": 0.9845,
+      "step": 232820
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.833014463916736e-05,
+      "loss": 1.1174,
+      "step": 232830
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.832899670926762e-05,
+      "loss": 1.1014,
+      "step": 232840
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.8327848779367876e-05,
+      "loss": 1.1011,
+      "step": 232850
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.832670084946812e-05,
+      "loss": 1.1187,
+      "step": 232860
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.8325552919568375e-05,
+      "loss": 0.9845,
+      "step": 232870
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.832440498966863e-05,
+      "loss": 1.181,
+      "step": 232880
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.8323257059768875e-05,
+      "loss": 0.9359,
+      "step": 232890
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.832210912986913e-05,
+      "loss": 1.0693,
+      "step": 232900
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.832096119996939e-05,
+      "loss": 1.06,
+      "step": 232910
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.831981327006964e-05,
+      "loss": 0.9437,
+      "step": 232920
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.831866534016989e-05,
+      "loss": 1.0527,
+      "step": 232930
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.8317517410270145e-05,
+      "loss": 1.009,
+      "step": 232940
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.8316369480370395e-05,
+      "loss": 1.1268,
+      "step": 232950
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.831522155047065e-05,
+      "loss": 1.1698,
+      "step": 232960
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.83140736205709e-05,
+      "loss": 1.0888,
+      "step": 232970
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.831292569067115e-05,
+      "loss": 1.1304,
+      "step": 232980
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.831177776077141e-05,
+      "loss": 1.1287,
+      "step": 232990
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.8310629830871664e-05,
+      "loss": 1.0477,
+      "step": 233000
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.830948190097191e-05,
+      "loss": 1.0323,
+      "step": 233010
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.8308333971072164e-05,
+      "loss": 1.1239,
+      "step": 233020
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.830718604117242e-05,
+      "loss": 1.1485,
+      "step": 233030
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.8306038111272664e-05,
+      "loss": 1.169,
+      "step": 233040
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.830489018137292e-05,
+      "loss": 1.1152,
+      "step": 233050
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.830374225147318e-05,
+      "loss": 1.2071,
+      "step": 233060
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.830259432157343e-05,
+      "loss": 1.1517,
+      "step": 233070
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.830144639167368e-05,
+      "loss": 1.0226,
+      "step": 233080
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.830029846177393e-05,
+      "loss": 1.1882,
+      "step": 233090
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.829915053187418e-05,
+      "loss": 1.1982,
+      "step": 233100
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.829800260197444e-05,
+      "loss": 1.0948,
+      "step": 233110
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.829685467207469e-05,
+      "loss": 0.9905,
+      "step": 233120
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.829570674217494e-05,
+      "loss": 1.1609,
+      "step": 233130
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.8294558812275196e-05,
+      "loss": 1.1405,
+      "step": 233140
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.829341088237545e-05,
+      "loss": 1.0814,
+      "step": 233150
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.8292262952475696e-05,
+      "loss": 1.1348,
+      "step": 233160
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.829111502257595e-05,
+      "loss": 1.1269,
+      "step": 233170
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.82899670926762e-05,
+      "loss": 0.9903,
+      "step": 233180
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.828881916277645e-05,
+      "loss": 1.1509,
+      "step": 233190
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.828767123287671e-05,
+      "loss": 0.9859,
+      "step": 233200
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.828652330297696e-05,
+      "loss": 1.0975,
+      "step": 233210
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.8285375373077215e-05,
+      "loss": 1.1644,
+      "step": 233220
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.8284227443177465e-05,
+      "loss": 1.2454,
+      "step": 233230
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.8283079513277715e-05,
+      "loss": 0.9971,
+      "step": 233240
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.828193158337797e-05,
+      "loss": 1.0132,
+      "step": 233250
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.828078365347823e-05,
+      "loss": 1.0979,
+      "step": 233260
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.827963572357847e-05,
+      "loss": 0.9587,
+      "step": 233270
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.827848779367873e-05,
+      "loss": 1.0754,
+      "step": 233280
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.8277339863778985e-05,
+      "loss": 1.0059,
+      "step": 233290
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.827619193387923e-05,
+      "loss": 1.0562,
+      "step": 233300
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.8275044003979484e-05,
+      "loss": 1.1222,
+      "step": 233310
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.827389607407974e-05,
+      "loss": 1.1284,
+      "step": 233320
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.827274814417999e-05,
+      "loss": 1.0811,
+      "step": 233330
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.827160021428024e-05,
+      "loss": 1.09,
+      "step": 233340
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.82704522843805e-05,
+      "loss": 1.0227,
+      "step": 233350
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.826930435448075e-05,
+      "loss": 1.0237,
+      "step": 233360
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.8268156424581004e-05,
+      "loss": 1.0299,
+      "step": 233370
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.8267008494681254e-05,
+      "loss": 1.0846,
+      "step": 233380
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.8265860564781503e-05,
+      "loss": 1.1203,
+      "step": 233390
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.826471263488176e-05,
+      "loss": 1.0811,
+      "step": 233400
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.826356470498202e-05,
+      "loss": 1.0972,
+      "step": 233410
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.826241677508226e-05,
+      "loss": 1.0082,
+      "step": 233420
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.8261268845182516e-05,
+      "loss": 1.0472,
+      "step": 233430
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.826012091528277e-05,
+      "loss": 1.1002,
+      "step": 233440
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.8258972985383016e-05,
+      "loss": 1.0153,
+      "step": 233450
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.825782505548327e-05,
+      "loss": 1.0533,
+      "step": 233460
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.825667712558353e-05,
+      "loss": 1.1419,
+      "step": 233470
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.825552919568378e-05,
+      "loss": 1.0407,
+      "step": 233480
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.825438126578403e-05,
+      "loss": 1.0804,
+      "step": 233490
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.8253233335884286e-05,
+      "loss": 1.0,
+      "step": 233500
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.8252085405984536e-05,
+      "loss": 1.0256,
+      "step": 233510
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.825093747608479e-05,
+      "loss": 1.164,
+      "step": 233520
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.824978954618504e-05,
+      "loss": 1.2037,
+      "step": 233530
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.824864161628529e-05,
+      "loss": 1.1459,
+      "step": 233540
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.824749368638555e-05,
+      "loss": 1.0802,
+      "step": 233550
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.8246345756485805e-05,
+      "loss": 0.9939,
+      "step": 233560
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.824519782658605e-05,
+      "loss": 1.0961,
+      "step": 233570
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.8244049896686305e-05,
+      "loss": 1.0777,
+      "step": 233580
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.824290196678656e-05,
+      "loss": 1.0662,
+      "step": 233590
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.8241754036886805e-05,
+      "loss": 1.009,
+      "step": 233600
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.824060610698706e-05,
+      "loss": 1.0629,
+      "step": 233610
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.823945817708732e-05,
+      "loss": 1.0185,
+      "step": 233620
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.823831024718757e-05,
+      "loss": 1.0995,
+      "step": 233630
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.823716231728782e-05,
+      "loss": 1.0897,
+      "step": 233640
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.8236014387388074e-05,
+      "loss": 1.0748,
+      "step": 233650
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.8234866457488324e-05,
+      "loss": 1.0655,
+      "step": 233660
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.823371852758858e-05,
+      "loss": 1.0084,
+      "step": 233670
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.823257059768883e-05,
+      "loss": 1.0298,
+      "step": 233680
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.823142266778908e-05,
+      "loss": 0.9862,
+      "step": 233690
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.823027473788934e-05,
+      "loss": 1.0155,
+      "step": 233700
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.8229126807989594e-05,
+      "loss": 1.0673,
+      "step": 233710
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.822797887808984e-05,
+      "loss": 1.048,
+      "step": 233720
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.822683094819009e-05,
+      "loss": 1.018,
+      "step": 233730
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.822568301829035e-05,
+      "loss": 1.1266,
+      "step": 233740
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.822453508839059e-05,
+      "loss": 1.1372,
+      "step": 233750
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.822338715849085e-05,
+      "loss": 1.146,
+      "step": 233760
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.8222239228591106e-05,
+      "loss": 1.1736,
+      "step": 233770
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.8221091298691356e-05,
+      "loss": 1.1404,
+      "step": 233780
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.8219943368791606e-05,
+      "loss": 1.0376,
+      "step": 233790
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.821879543889186e-05,
+      "loss": 1.0253,
+      "step": 233800
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.821764750899211e-05,
+      "loss": 1.1295,
+      "step": 233810
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.821649957909237e-05,
+      "loss": 1.1282,
+      "step": 233820
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.821535164919262e-05,
+      "loss": 1.0957,
+      "step": 233830
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.821420371929287e-05,
+      "loss": 1.019,
+      "step": 233840
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.8213055789393125e-05,
+      "loss": 1.1466,
+      "step": 233850
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.821190785949338e-05,
+      "loss": 1.0486,
+      "step": 233860
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.8210759929593625e-05,
+      "loss": 1.0653,
+      "step": 233870
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.820961199969388e-05,
+      "loss": 1.0845,
+      "step": 233880
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.820846406979414e-05,
+      "loss": 1.1464,
+      "step": 233890
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.820731613989438e-05,
+      "loss": 1.0601,
+      "step": 233900
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.820616820999464e-05,
+      "loss": 1.0621,
+      "step": 233910
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.8205020280094895e-05,
+      "loss": 1.1332,
+      "step": 233920
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.8203872350195145e-05,
+      "loss": 1.0642,
+      "step": 233930
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.8202724420295394e-05,
+      "loss": 1.0452,
+      "step": 233940
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.820157649039565e-05,
+      "loss": 1.1029,
+      "step": 233950
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.82004285604959e-05,
+      "loss": 1.1299,
+      "step": 233960
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.819928063059616e-05,
+      "loss": 1.1559,
+      "step": 233970
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.819813270069641e-05,
+      "loss": 1.052,
+      "step": 233980
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.819698477079666e-05,
+      "loss": 1.1424,
+      "step": 233990
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.8195836840896914e-05,
+      "loss": 1.0545,
+      "step": 234000
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.819468891099717e-05,
+      "loss": 1.0279,
+      "step": 234010
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.8193540981097414e-05,
+      "loss": 1.125,
+      "step": 234020
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.819239305119767e-05,
+      "loss": 1.1228,
+      "step": 234030
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.819124512129793e-05,
+      "loss": 1.0214,
+      "step": 234040
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.819009719139817e-05,
+      "loss": 1.0205,
+      "step": 234050
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.8188949261498427e-05,
+      "loss": 1.1525,
+      "step": 234060
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.818780133159868e-05,
+      "loss": 1.0653,
+      "step": 234070
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.818665340169893e-05,
+      "loss": 1.0615,
+      "step": 234080
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.818550547179918e-05,
+      "loss": 1.0298,
+      "step": 234090
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.818435754189944e-05,
+      "loss": 1.0302,
+      "step": 234100
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.818320961199969e-05,
+      "loss": 1.0905,
+      "step": 234110
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.8182061682099946e-05,
+      "loss": 0.9046,
+      "step": 234120
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.8180913752200196e-05,
+      "loss": 1.156,
+      "step": 234130
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.8179765822300446e-05,
+      "loss": 1.0725,
+      "step": 234140
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.81786178924007e-05,
+      "loss": 0.9973,
+      "step": 234150
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.817746996250096e-05,
+      "loss": 1.1536,
+      "step": 234160
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.81763220326012e-05,
+      "loss": 0.9674,
+      "step": 234170
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.817517410270146e-05,
+      "loss": 1.1363,
+      "step": 234180
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.8174026172801715e-05,
+      "loss": 1.1462,
+      "step": 234190
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.817287824290196e-05,
+      "loss": 1.1576,
+      "step": 234200
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.8171730313002215e-05,
+      "loss": 1.1071,
+      "step": 234210
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.817058238310247e-05,
+      "loss": 1.0903,
+      "step": 234220
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.816943445320272e-05,
+      "loss": 1.1245,
+      "step": 234230
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.816828652330297e-05,
+      "loss": 1.0616,
+      "step": 234240
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.816713859340323e-05,
+      "loss": 1.1549,
+      "step": 234250
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.816599066350348e-05,
+      "loss": 1.1429,
+      "step": 234260
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.8164842733603735e-05,
+      "loss": 1.0826,
+      "step": 234270
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.8163694803703984e-05,
+      "loss": 1.0619,
+      "step": 234280
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.8162546873804234e-05,
+      "loss": 1.0978,
+      "step": 234290
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.816139894390449e-05,
+      "loss": 0.9808,
+      "step": 234300
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.816025101400475e-05,
+      "loss": 1.1637,
+      "step": 234310
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.815910308410499e-05,
+      "loss": 1.0821,
+      "step": 234320
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.815795515420525e-05,
+      "loss": 0.9454,
+      "step": 234330
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.8156807224305504e-05,
+      "loss": 1.0475,
+      "step": 234340
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.815565929440575e-05,
+      "loss": 1.095,
+      "step": 234350
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.8154511364506004e-05,
+      "loss": 0.9586,
+      "step": 234360
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.815336343460626e-05,
+      "loss": 1.1767,
+      "step": 234370
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.815221550470651e-05,
+      "loss": 1.1104,
+      "step": 234380
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.815106757480676e-05,
+      "loss": 1.095,
+      "step": 234390
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.8149919644907017e-05,
+      "loss": 1.1239,
+      "step": 234400
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.8148771715007266e-05,
+      "loss": 1.0497,
+      "step": 234410
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.814762378510752e-05,
+      "loss": 1.0773,
+      "step": 234420
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.814647585520777e-05,
+      "loss": 0.9413,
+      "step": 234430
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.814532792530802e-05,
+      "loss": 0.9512,
+      "step": 234440
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.814417999540828e-05,
+      "loss": 0.9293,
+      "step": 234450
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.8143032065508536e-05,
+      "loss": 1.1508,
+      "step": 234460
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.814188413560878e-05,
+      "loss": 1.0387,
+      "step": 234470
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.8140736205709036e-05,
+      "loss": 1.1212,
+      "step": 234480
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.813958827580929e-05,
+      "loss": 1.1173,
+      "step": 234490
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.8138440345909535e-05,
+      "loss": 1.07,
+      "step": 234500
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.813729241600979e-05,
+      "loss": 1.0501,
+      "step": 234510
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.813614448611005e-05,
+      "loss": 1.1705,
+      "step": 234520
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.81349965562103e-05,
+      "loss": 0.9793,
+      "step": 234530
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.813384862631055e-05,
+      "loss": 1.0792,
+      "step": 234540
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.81327006964108e-05,
+      "loss": 1.2538,
+      "step": 234550
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.8131552766511055e-05,
+      "loss": 1.0758,
+      "step": 234560
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.813040483661131e-05,
+      "loss": 1.037,
+      "step": 234570
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.8129256906711555e-05,
+      "loss": 1.1648,
+      "step": 234580
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.812810897681181e-05,
+      "loss": 1.088,
+      "step": 234590
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.812696104691207e-05,
+      "loss": 1.0682,
+      "step": 234600
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.812581311701231e-05,
+      "loss": 1.0958,
+      "step": 234610
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.812466518711257e-05,
+      "loss": 1.063,
+      "step": 234620
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.8123517257212824e-05,
+      "loss": 1.0413,
+      "step": 234630
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.812236932731307e-05,
+      "loss": 1.1442,
+      "step": 234640
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.8121221397413324e-05,
+      "loss": 1.1148,
+      "step": 234650
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.812007346751358e-05,
+      "loss": 1.2768,
+      "step": 234660
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.811892553761383e-05,
+      "loss": 1.06,
+      "step": 234670
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.811777760771409e-05,
+      "loss": 1.0965,
+      "step": 234680
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.811662967781434e-05,
+      "loss": 1.0608,
+      "step": 234690
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.811548174791459e-05,
+      "loss": 1.0367,
+      "step": 234700
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.811433381801484e-05,
+      "loss": 1.1323,
+      "step": 234710
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.81131858881151e-05,
+      "loss": 1.0274,
+      "step": 234720
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.811203795821534e-05,
+      "loss": 1.1366,
+      "step": 234730
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.81108900283156e-05,
+      "loss": 1.0032,
+      "step": 234740
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.8109742098415856e-05,
+      "loss": 1.0326,
+      "step": 234750
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.81085941685161e-05,
+      "loss": 1.0806,
+      "step": 234760
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.8107446238616356e-05,
+      "loss": 1.2896,
+      "step": 234770
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.810629830871661e-05,
+      "loss": 1.0412,
+      "step": 234780
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.8105150378816856e-05,
+      "loss": 1.1176,
+      "step": 234790
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.810400244891711e-05,
+      "loss": 1.0235,
+      "step": 234800
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.810285451901737e-05,
+      "loss": 1.1738,
+      "step": 234810
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.810170658911762e-05,
+      "loss": 1.0588,
+      "step": 234820
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.8100558659217875e-05,
+      "loss": 0.9782,
+      "step": 234830
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.8099410729318125e-05,
+      "loss": 1.1378,
+      "step": 234840
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.8098262799418375e-05,
+      "loss": 1.1064,
+      "step": 234850
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.809711486951863e-05,
+      "loss": 1.1497,
+      "step": 234860
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.809596693961889e-05,
+      "loss": 0.9608,
+      "step": 234870
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.809481900971913e-05,
+      "loss": 1.0843,
+      "step": 234880
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.809367107981939e-05,
+      "loss": 1.0032,
+      "step": 234890
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.8092523149919645e-05,
+      "loss": 1.0171,
+      "step": 234900
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.809137522001989e-05,
+      "loss": 1.0838,
+      "step": 234910
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.8090227290120144e-05,
+      "loss": 1.0599,
+      "step": 234920
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.80890793602204e-05,
+      "loss": 0.9747,
+      "step": 234930
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.8087931430320644e-05,
+      "loss": 1.0751,
+      "step": 234940
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.80867835004209e-05,
+      "loss": 1.2281,
+      "step": 234950
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.808563557052116e-05,
+      "loss": 1.0368,
+      "step": 234960
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.808448764062141e-05,
+      "loss": 1.0532,
+      "step": 234970
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.8083339710721664e-05,
+      "loss": 1.0277,
+      "step": 234980
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.8082191780821914e-05,
+      "loss": 1.1821,
+      "step": 234990
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.8081043850922164e-05,
+      "loss": 1.0131,
+      "step": 235000
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.807989592102242e-05,
+      "loss": 1.1281,
+      "step": 235010
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.807874799112268e-05,
+      "loss": 1.0661,
+      "step": 235020
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.807760006122292e-05,
+      "loss": 1.0916,
+      "step": 235030
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.8076452131323177e-05,
+      "loss": 1.1704,
+      "step": 235040
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.807530420142343e-05,
+      "loss": 1.0251,
+      "step": 235050
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.8074156271523676e-05,
+      "loss": 1.0367,
+      "step": 235060
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.807300834162393e-05,
+      "loss": 1.1018,
+      "step": 235070
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.807186041172419e-05,
+      "loss": 1.0229,
+      "step": 235080
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.807071248182443e-05,
+      "loss": 1.0492,
+      "step": 235090
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.806956455192469e-05,
+      "loss": 1.0755,
+      "step": 235100
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.8068416622024946e-05,
+      "loss": 1.0866,
+      "step": 235110
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.8067268692125196e-05,
+      "loss": 1.171,
+      "step": 235120
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.806612076222545e-05,
+      "loss": 1.0073,
+      "step": 235130
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.80649728323257e-05,
+      "loss": 1.0329,
+      "step": 235140
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.806382490242595e-05,
+      "loss": 1.029,
+      "step": 235150
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.806267697252621e-05,
+      "loss": 1.1332,
+      "step": 235160
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.8061529042626465e-05,
+      "loss": 1.0874,
+      "step": 235170
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.806038111272671e-05,
+      "loss": 1.0575,
+      "step": 235180
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.8059233182826965e-05,
+      "loss": 1.0261,
+      "step": 235190
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.805808525292722e-05,
+      "loss": 1.0679,
+      "step": 235200
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.8056937323027465e-05,
+      "loss": 1.2026,
+      "step": 235210
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.805578939312772e-05,
+      "loss": 1.0621,
+      "step": 235220
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.805464146322798e-05,
+      "loss": 1.114,
+      "step": 235230
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.805349353332822e-05,
+      "loss": 1.0385,
+      "step": 235240
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.805234560342848e-05,
+      "loss": 1.049,
+      "step": 235250
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.8051197673528734e-05,
+      "loss": 1.0861,
+      "step": 235260
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.8050049743628984e-05,
+      "loss": 1.0533,
+      "step": 235270
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.804890181372924e-05,
+      "loss": 1.0812,
+      "step": 235280
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.804775388382949e-05,
+      "loss": 1.1184,
+      "step": 235290
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.804660595392974e-05,
+      "loss": 1.0881,
+      "step": 235300
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.804545802403e-05,
+      "loss": 1.086,
+      "step": 235310
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.8044310094130254e-05,
+      "loss": 1.1199,
+      "step": 235320
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.80431621642305e-05,
+      "loss": 1.1572,
+      "step": 235330
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.8042014234330754e-05,
+      "loss": 0.9203,
+      "step": 235340
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.804086630443101e-05,
+      "loss": 1.0721,
+      "step": 235350
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.803971837453125e-05,
+      "loss": 1.095,
+      "step": 235360
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.803857044463151e-05,
+      "loss": 1.1195,
+      "step": 235370
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.8037422514731767e-05,
+      "loss": 1.199,
+      "step": 235380
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.803627458483201e-05,
+      "loss": 1.0231,
+      "step": 235390
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.8035126654932266e-05,
+      "loss": 1.021,
+      "step": 235400
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.803397872503252e-05,
+      "loss": 1.1027,
+      "step": 235410
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.803283079513277e-05,
+      "loss": 1.065,
+      "step": 235420
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.803168286523303e-05,
+      "loss": 1.1581,
+      "step": 235430
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.803053493533328e-05,
+      "loss": 1.0648,
+      "step": 235440
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.802938700543353e-05,
+      "loss": 1.0807,
+      "step": 235450
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.8028239075533786e-05,
+      "loss": 1.0647,
+      "step": 235460
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.802709114563404e-05,
+      "loss": 1.0947,
+      "step": 235470
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.8025943215734285e-05,
+      "loss": 1.0069,
+      "step": 235480
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.802479528583454e-05,
+      "loss": 1.1468,
+      "step": 235490
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.80236473559348e-05,
+      "loss": 1.0873,
+      "step": 235500
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.802249942603504e-05,
+      "loss": 1.0737,
+      "step": 235510
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.80213514961353e-05,
+      "loss": 1.1538,
+      "step": 235520
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.8020203566235555e-05,
+      "loss": 1.0606,
+      "step": 235530
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.80190556363358e-05,
+      "loss": 1.1519,
+      "step": 235540
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.8017907706436055e-05,
+      "loss": 1.0555,
+      "step": 235550
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.801675977653631e-05,
+      "loss": 1.0478,
+      "step": 235560
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.801561184663656e-05,
+      "loss": 1.1594,
+      "step": 235570
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.801446391673682e-05,
+      "loss": 1.1371,
+      "step": 235580
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.801331598683707e-05,
+      "loss": 1.0997,
+      "step": 235590
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.801216805693732e-05,
+      "loss": 1.1075,
+      "step": 235600
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.8011020127037574e-05,
+      "loss": 1.1215,
+      "step": 235610
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.800987219713783e-05,
+      "loss": 1.2186,
+      "step": 235620
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.8008724267238074e-05,
+      "loss": 1.0685,
+      "step": 235630
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.800757633733833e-05,
+      "loss": 1.0978,
+      "step": 235640
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.800642840743859e-05,
+      "loss": 1.0252,
+      "step": 235650
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.800528047753883e-05,
+      "loss": 1.1769,
+      "step": 235660
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.800413254763909e-05,
+      "loss": 1.0191,
+      "step": 235670
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.8002984617739343e-05,
+      "loss": 1.0214,
+      "step": 235680
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.8001836687839587e-05,
+      "loss": 1.1027,
+      "step": 235690
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.800068875793984e-05,
+      "loss": 1.033,
+      "step": 235700
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.79995408280401e-05,
+      "loss": 1.0388,
+      "step": 235710
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.799839289814035e-05,
+      "loss": 1.0927,
+      "step": 235720
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.7997244968240606e-05,
+      "loss": 1.0772,
+      "step": 235730
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.7996097038340856e-05,
+      "loss": 1.0875,
+      "step": 235740
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.7994949108441106e-05,
+      "loss": 1.1159,
+      "step": 235750
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.799380117854136e-05,
+      "loss": 1.0533,
+      "step": 235760
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.799265324864162e-05,
+      "loss": 1.0779,
+      "step": 235770
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.799150531874186e-05,
+      "loss": 1.0219,
+      "step": 235780
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.799035738884212e-05,
+      "loss": 1.2286,
+      "step": 235790
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.7989209458942376e-05,
+      "loss": 0.9879,
+      "step": 235800
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.798806152904262e-05,
+      "loss": 1.1018,
+      "step": 235810
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.7986913599142875e-05,
+      "loss": 1.0144,
+      "step": 235820
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.798576566924313e-05,
+      "loss": 1.0092,
+      "step": 235830
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.7984617739343375e-05,
+      "loss": 1.0764,
+      "step": 235840
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.798346980944363e-05,
+      "loss": 1.1108,
+      "step": 235850
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.798232187954389e-05,
+      "loss": 1.0835,
+      "step": 235860
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.798117394964414e-05,
+      "loss": 1.0991,
+      "step": 235870
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.7980026019744395e-05,
+      "loss": 1.1888,
+      "step": 235880
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.7978878089844645e-05,
+      "loss": 1.1095,
+      "step": 235890
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.7977730159944894e-05,
+      "loss": 1.0531,
+      "step": 235900
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.797658223004515e-05,
+      "loss": 1.0545,
+      "step": 235910
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.7975434300145394e-05,
+      "loss": 1.0557,
+      "step": 235920
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.797428637024565e-05,
+      "loss": 1.0427,
+      "step": 235930
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.797313844034591e-05,
+      "loss": 1.1379,
+      "step": 235940
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.797199051044615e-05,
+      "loss": 0.9543,
+      "step": 235950
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.797084258054641e-05,
+      "loss": 1.0841,
+      "step": 235960
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.7969694650646664e-05,
+      "loss": 1.0309,
+      "step": 235970
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.7968546720746914e-05,
+      "loss": 1.104,
+      "step": 235980
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.7967398790847163e-05,
+      "loss": 1.0451,
+      "step": 235990
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.796625086094742e-05,
+      "loss": 1.099,
+      "step": 236000
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.796510293104767e-05,
+      "loss": 0.9997,
+      "step": 236010
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.7963955001147927e-05,
+      "loss": 1.0497,
+      "step": 236020
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.796280707124818e-05,
+      "loss": 1.057,
+      "step": 236030
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.7961659141348426e-05,
+      "loss": 1.0964,
+      "step": 236040
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.796051121144868e-05,
+      "loss": 1.0728,
+      "step": 236050
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.795936328154894e-05,
+      "loss": 1.117,
+      "step": 236060
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.795821535164918e-05,
+      "loss": 1.0763,
+      "step": 236070
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.795706742174944e-05,
+      "loss": 1.203,
+      "step": 236080
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.7955919491849696e-05,
+      "loss": 1.0173,
+      "step": 236090
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.795477156194994e-05,
+      "loss": 1.2005,
+      "step": 236100
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.7953623632050196e-05,
+      "loss": 1.0331,
+      "step": 236110
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.795247570215045e-05,
+      "loss": 1.0997,
+      "step": 236120
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.79513277722507e-05,
+      "loss": 1.2006,
+      "step": 236130
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.795017984235095e-05,
+      "loss": 1.1198,
+      "step": 236140
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.794903191245121e-05,
+      "loss": 1.2124,
+      "step": 236150
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.794788398255146e-05,
+      "loss": 0.9982,
+      "step": 236160
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.7946736052651715e-05,
+      "loss": 1.0574,
+      "step": 236170
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.794558812275197e-05,
+      "loss": 1.0426,
+      "step": 236180
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.7944440192852215e-05,
+      "loss": 1.1508,
+      "step": 236190
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.794329226295247e-05,
+      "loss": 1.0916,
+      "step": 236200
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.794214433305273e-05,
+      "loss": 1.0444,
+      "step": 236210
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.794099640315297e-05,
+      "loss": 1.1552,
+      "step": 236220
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.793984847325323e-05,
+      "loss": 1.0896,
+      "step": 236230
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.7938700543353484e-05,
+      "loss": 1.0836,
+      "step": 236240
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.793755261345373e-05,
+      "loss": 1.0943,
+      "step": 236250
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.7936404683553984e-05,
+      "loss": 1.042,
+      "step": 236260
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.793525675365424e-05,
+      "loss": 1.0718,
+      "step": 236270
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.793410882375449e-05,
+      "loss": 0.9905,
+      "step": 236280
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.793296089385474e-05,
+      "loss": 1.143,
+      "step": 236290
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.7931812963955e-05,
+      "loss": 1.0367,
+      "step": 236300
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.793066503405525e-05,
+      "loss": 1.0766,
+      "step": 236310
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.7929517104155504e-05,
+      "loss": 1.3029,
+      "step": 236320
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.792836917425576e-05,
+      "loss": 1.1593,
+      "step": 236330
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.7927221244356e-05,
+      "loss": 1.1382,
+      "step": 236340
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.792607331445626e-05,
+      "loss": 1.1699,
+      "step": 236350
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.7924925384556517e-05,
+      "loss": 1.1344,
+      "step": 236360
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.792377745465676e-05,
+      "loss": 1.0149,
+      "step": 236370
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.7922629524757016e-05,
+      "loss": 1.098,
+      "step": 236380
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.792148159485727e-05,
+      "loss": 1.08,
+      "step": 236390
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.7920333664957516e-05,
+      "loss": 1.0936,
+      "step": 236400
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.791918573505777e-05,
+      "loss": 1.0213,
+      "step": 236410
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.791803780515803e-05,
+      "loss": 1.0643,
+      "step": 236420
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.791688987525828e-05,
+      "loss": 1.1282,
+      "step": 236430
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.791574194535853e-05,
+      "loss": 1.1551,
+      "step": 236440
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.7914594015458786e-05,
+      "loss": 0.9992,
+      "step": 236450
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.7913446085559035e-05,
+      "loss": 0.9641,
+      "step": 236460
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.791229815565929e-05,
+      "loss": 1.0588,
+      "step": 236470
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.791115022575955e-05,
+      "loss": 1.166,
+      "step": 236480
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.791000229585979e-05,
+      "loss": 1.0957,
+      "step": 236490
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.790885436596005e-05,
+      "loss": 1.1373,
+      "step": 236500
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.7907706436060305e-05,
+      "loss": 1.1006,
+      "step": 236510
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.790655850616055e-05,
+      "loss": 1.131,
+      "step": 236520
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.7905410576260805e-05,
+      "loss": 1.1667,
+      "step": 236530
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.790426264636106e-05,
+      "loss": 1.1667,
+      "step": 236540
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.7903114716461304e-05,
+      "loss": 1.1197,
+      "step": 236550
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.790196678656156e-05,
+      "loss": 1.0926,
+      "step": 236560
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.790081885666182e-05,
+      "loss": 0.9705,
+      "step": 236570
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.789967092676207e-05,
+      "loss": 0.999,
+      "step": 236580
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.789852299686232e-05,
+      "loss": 1.1856,
+      "step": 236590
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.7897375066962574e-05,
+      "loss": 1.0519,
+      "step": 236600
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.7896227137062824e-05,
+      "loss": 1.1382,
+      "step": 236610
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.789507920716308e-05,
+      "loss": 1.1031,
+      "step": 236620
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.789393127726334e-05,
+      "loss": 1.1899,
+      "step": 236630
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.789278334736358e-05,
+      "loss": 0.9997,
+      "step": 236640
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.789163541746384e-05,
+      "loss": 1.0523,
+      "step": 236650
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.7890487487564093e-05,
+      "loss": 1.182,
+      "step": 236660
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.7889339557664337e-05,
+      "loss": 1.0162,
+      "step": 236670
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.788819162776459e-05,
+      "loss": 1.0652,
+      "step": 236680
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.788704369786485e-05,
+      "loss": 1.1573,
+      "step": 236690
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.788589576796509e-05,
+      "loss": 1.2049,
+      "step": 236700
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.788474783806535e-05,
+      "loss": 1.0655,
+      "step": 236710
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.7883599908165606e-05,
+      "loss": 1.1688,
+      "step": 236720
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.7882451978265856e-05,
+      "loss": 1.0358,
+      "step": 236730
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.7881304048366106e-05,
+      "loss": 1.0636,
+      "step": 236740
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.788015611846636e-05,
+      "loss": 1.1651,
+      "step": 236750
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.787900818856661e-05,
+      "loss": 1.1107,
+      "step": 236760
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.787786025866687e-05,
+      "loss": 0.9946,
+      "step": 236770
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.7876712328767126e-05,
+      "loss": 1.145,
+      "step": 236780
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.787556439886737e-05,
+      "loss": 1.1249,
+      "step": 236790
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.7874416468967625e-05,
+      "loss": 1.048,
+      "step": 236800
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.787326853906788e-05,
+      "loss": 1.0816,
+      "step": 236810
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.7872120609168125e-05,
+      "loss": 1.1457,
+      "step": 236820
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.787097267926838e-05,
+      "loss": 1.1685,
+      "step": 236830
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.786982474936864e-05,
+      "loss": 1.075,
+      "step": 236840
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.786867681946888e-05,
+      "loss": 1.0673,
+      "step": 236850
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.786752888956914e-05,
+      "loss": 0.8524,
+      "step": 236860
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.7866380959669395e-05,
+      "loss": 1.0876,
+      "step": 236870
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.7865233029769644e-05,
+      "loss": 1.0919,
+      "step": 236880
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.7864085099869894e-05,
+      "loss": 1.1235,
+      "step": 236890
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.786293716997015e-05,
+      "loss": 1.17,
+      "step": 236900
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.78617892400704e-05,
+      "loss": 1.086,
+      "step": 236910
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.786064131017066e-05,
+      "loss": 0.9256,
+      "step": 236920
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.7859493380270914e-05,
+      "loss": 1.0793,
+      "step": 236930
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.785834545037116e-05,
+      "loss": 0.9697,
+      "step": 236940
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.7857197520471414e-05,
+      "loss": 1.1329,
+      "step": 236950
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.785604959057167e-05,
+      "loss": 0.992,
+      "step": 236960
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.7854901660671913e-05,
+      "loss": 1.0488,
+      "step": 236970
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.785375373077217e-05,
+      "loss": 1.0767,
+      "step": 236980
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.785260580087243e-05,
+      "loss": 1.0798,
+      "step": 236990
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.785145787097267e-05,
+      "loss": 1.148,
+      "step": 237000
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.7850309941072926e-05,
+      "loss": 1.0148,
+      "step": 237010
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.784916201117318e-05,
+      "loss": 1.1333,
+      "step": 237020
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.784801408127343e-05,
+      "loss": 1.1457,
+      "step": 237030
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.784686615137368e-05,
+      "loss": 1.2602,
+      "step": 237040
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.784571822147394e-05,
+      "loss": 1.1587,
+      "step": 237050
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.784457029157419e-05,
+      "loss": 1.1146,
+      "step": 237060
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.7843422361674446e-05,
+      "loss": 1.0295,
+      "step": 237070
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.78422744317747e-05,
+      "loss": 1.0285,
+      "step": 237080
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.7841126501874946e-05,
+      "loss": 1.1011,
+      "step": 237090
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.78399785719752e-05,
+      "loss": 1.0271,
+      "step": 237100
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.783883064207546e-05,
+      "loss": 1.0975,
+      "step": 237110
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.78376827121757e-05,
+      "loss": 0.9936,
+      "step": 237120
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.783653478227596e-05,
+      "loss": 0.9849,
+      "step": 237130
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.7835386852376215e-05,
+      "loss": 1.0097,
+      "step": 237140
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.783423892247646e-05,
+      "loss": 1.1347,
+      "step": 237150
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.7833090992576715e-05,
+      "loss": 1.0321,
+      "step": 237160
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.783194306267697e-05,
+      "loss": 1.0956,
+      "step": 237170
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.783079513277722e-05,
+      "loss": 1.0526,
+      "step": 237180
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.782964720287748e-05,
+      "loss": 0.9908,
+      "step": 237190
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.782849927297773e-05,
+      "loss": 1.0731,
+      "step": 237200
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.782735134307798e-05,
+      "loss": 1.2259,
+      "step": 237210
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.7826203413178234e-05,
+      "loss": 1.0581,
+      "step": 237220
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.782505548327849e-05,
+      "loss": 0.9787,
+      "step": 237230
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.7823907553378734e-05,
+      "loss": 1.2502,
+      "step": 237240
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.782275962347899e-05,
+      "loss": 1.1081,
+      "step": 237250
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.782161169357925e-05,
+      "loss": 1.1257,
+      "step": 237260
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.782046376367949e-05,
+      "loss": 1.1845,
+      "step": 237270
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.781931583377975e-05,
+      "loss": 1.0886,
+      "step": 237280
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.781816790388e-05,
+      "loss": 1.0361,
+      "step": 237290
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.781701997398025e-05,
+      "loss": 1.0826,
+      "step": 237300
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.7815872044080503e-05,
+      "loss": 1.0542,
+      "step": 237310
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.781472411418075e-05,
+      "loss": 1.0801,
+      "step": 237320
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.781357618428101e-05,
+      "loss": 1.1092,
+      "step": 237330
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.7812428254381267e-05,
+      "loss": 1.1216,
+      "step": 237340
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.781128032448151e-05,
+      "loss": 1.0685,
+      "step": 237350
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.7810132394581766e-05,
+      "loss": 0.9571,
+      "step": 237360
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.780898446468202e-05,
+      "loss": 1.1682,
+      "step": 237370
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.7807836534782266e-05,
+      "loss": 0.9984,
+      "step": 237380
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.780668860488252e-05,
+      "loss": 1.0691,
+      "step": 237390
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.780554067498278e-05,
+      "loss": 1.1124,
+      "step": 237400
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.780439274508302e-05,
+      "loss": 1.1642,
+      "step": 237410
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.780324481518328e-05,
+      "loss": 1.1283,
+      "step": 237420
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.7802096885283536e-05,
+      "loss": 1.0975,
+      "step": 237430
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.7800948955383785e-05,
+      "loss": 0.9973,
+      "step": 237440
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.7799801025484035e-05,
+      "loss": 0.9902,
+      "step": 237450
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.779865309558429e-05,
+      "loss": 1.1378,
+      "step": 237460
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.779750516568454e-05,
+      "loss": 1.111,
+      "step": 237470
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.77963572357848e-05,
+      "loss": 1.017,
+      "step": 237480
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.7795209305885055e-05,
+      "loss": 1.0306,
+      "step": 237490
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.77940613759853e-05,
+      "loss": 1.103,
+      "step": 237500
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.7792913446085555e-05,
+      "loss": 1.061,
+      "step": 237510
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.779176551618581e-05,
+      "loss": 1.2059,
+      "step": 237520
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.7790617586286054e-05,
+      "loss": 1.1318,
+      "step": 237530
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.778946965638631e-05,
+      "loss": 1.0637,
+      "step": 237540
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.778832172648657e-05,
+      "loss": 1.1686,
+      "step": 237550
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.778717379658681e-05,
+      "loss": 1.0819,
+      "step": 237560
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.778602586668707e-05,
+      "loss": 0.9807,
+      "step": 237570
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.7784877936787324e-05,
+      "loss": 1.1049,
+      "step": 237580
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.7783730006887574e-05,
+      "loss": 1.0098,
+      "step": 237590
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.7782582076987824e-05,
+      "loss": 1.1404,
+      "step": 237600
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.778143414708808e-05,
+      "loss": 1.102,
+      "step": 237610
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.778028621718833e-05,
+      "loss": 1.0955,
+      "step": 237620
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.777913828728859e-05,
+      "loss": 1.0668,
+      "step": 237630
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.7777990357388843e-05,
+      "loss": 1.0993,
+      "step": 237640
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.7776842427489087e-05,
+      "loss": 0.9545,
+      "step": 237650
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.777569449758934e-05,
+      "loss": 0.963,
+      "step": 237660
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.77745465676896e-05,
+      "loss": 1.1578,
+      "step": 237670
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.777339863778984e-05,
+      "loss": 1.0566,
+      "step": 237680
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.77722507078901e-05,
+      "loss": 1.0193,
+      "step": 237690
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.7771102777990356e-05,
+      "loss": 1.0101,
+      "step": 237700
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.77699548480906e-05,
+      "loss": 1.0944,
+      "step": 237710
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.7768806918190856e-05,
+      "loss": 1.0822,
+      "step": 237720
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.776765898829111e-05,
+      "loss": 1.2085,
+      "step": 237730
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.776651105839136e-05,
+      "loss": 1.1693,
+      "step": 237740
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.776536312849161e-05,
+      "loss": 1.0588,
+      "step": 237750
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.776421519859187e-05,
+      "loss": 1.0933,
+      "step": 237760
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.776306726869212e-05,
+      "loss": 1.0702,
+      "step": 237770
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.7761919338792375e-05,
+      "loss": 1.1714,
+      "step": 237780
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.776077140889263e-05,
+      "loss": 1.1166,
+      "step": 237790
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.7759623478992875e-05,
+      "loss": 1.029,
+      "step": 237800
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.775847554909313e-05,
+      "loss": 1.0402,
+      "step": 237810
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.775732761919339e-05,
+      "loss": 0.9892,
+      "step": 237820
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.775617968929363e-05,
+      "loss": 1.0023,
+      "step": 237830
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.775503175939389e-05,
+      "loss": 1.054,
+      "step": 237840
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.7753883829494145e-05,
+      "loss": 1.0938,
+      "step": 237850
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.775273589959439e-05,
+      "loss": 1.1857,
+      "step": 237860
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.7751587969694644e-05,
+      "loss": 1.1781,
+      "step": 237870
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.77504400397949e-05,
+      "loss": 1.2102,
+      "step": 237880
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.774929210989515e-05,
+      "loss": 1.0624,
+      "step": 237890
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.77481441799954e-05,
+      "loss": 0.9913,
+      "step": 237900
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.774699625009566e-05,
+      "loss": 1.0721,
+      "step": 237910
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.774584832019591e-05,
+      "loss": 1.0673,
+      "step": 237920
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.7744700390296164e-05,
+      "loss": 1.0234,
+      "step": 237930
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.774355246039642e-05,
+      "loss": 1.1169,
+      "step": 237940
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.7742404530496663e-05,
+      "loss": 1.0995,
+      "step": 237950
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.774125660059692e-05,
+      "loss": 1.1528,
+      "step": 237960
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.774010867069718e-05,
+      "loss": 1.1143,
+      "step": 237970
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.773896074079742e-05,
+      "loss": 1.0355,
+      "step": 237980
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.7737812810897676e-05,
+      "loss": 1.1404,
+      "step": 237990
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.773666488099793e-05,
+      "loss": 1.0985,
+      "step": 238000
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.7735516951098176e-05,
+      "loss": 1.0598,
+      "step": 238010
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.773436902119843e-05,
+      "loss": 1.0178,
+      "step": 238020
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.773322109129869e-05,
+      "loss": 1.2043,
+      "step": 238030
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.773207316139894e-05,
+      "loss": 1.0148,
+      "step": 238040
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.773092523149919e-05,
+      "loss": 1.143,
+      "step": 238050
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.7729777301599446e-05,
+      "loss": 1.1073,
+      "step": 238060
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.7728629371699696e-05,
+      "loss": 1.1372,
+      "step": 238070
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.772748144179995e-05,
+      "loss": 1.0668,
+      "step": 238080
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.772633351190021e-05,
+      "loss": 1.1898,
+      "step": 238090
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.772518558200045e-05,
+      "loss": 1.0917,
+      "step": 238100
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.772403765210071e-05,
+      "loss": 1.1073,
+      "step": 238110
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.7722889722200965e-05,
+      "loss": 1.1294,
+      "step": 238120
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.772174179230121e-05,
+      "loss": 1.1372,
+      "step": 238130
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.7720593862401465e-05,
+      "loss": 1.0788,
+      "step": 238140
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.771944593250172e-05,
+      "loss": 1.1688,
+      "step": 238150
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.7718298002601965e-05,
+      "loss": 1.0058,
+      "step": 238160
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.771715007270222e-05,
+      "loss": 1.0832,
+      "step": 238170
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.771600214280248e-05,
+      "loss": 1.0577,
+      "step": 238180
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.771485421290273e-05,
+      "loss": 1.1453,
+      "step": 238190
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.771370628300298e-05,
+      "loss": 1.0462,
+      "step": 238200
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.7712558353103234e-05,
+      "loss": 1.0505,
+      "step": 238210
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.7711410423203484e-05,
+      "loss": 1.1095,
+      "step": 238220
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.771026249330374e-05,
+      "loss": 1.0385,
+      "step": 238230
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.7709114563404e-05,
+      "loss": 1.1048,
+      "step": 238240
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.770796663350424e-05,
+      "loss": 1.0632,
+      "step": 238250
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.77068187036045e-05,
+      "loss": 1.0679,
+      "step": 238260
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.7705670773704754e-05,
+      "loss": 1.0739,
+      "step": 238270
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.7704522843805e-05,
+      "loss": 1.0765,
+      "step": 238280
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.7703374913905253e-05,
+      "loss": 1.1297,
+      "step": 238290
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.770222698400551e-05,
+      "loss": 1.1556,
+      "step": 238300
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.770107905410575e-05,
+      "loss": 1.1457,
+      "step": 238310
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.769993112420601e-05,
+      "loss": 1.0465,
+      "step": 238320
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.7698783194306266e-05,
+      "loss": 1.093,
+      "step": 238330
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.7697635264406516e-05,
+      "loss": 1.1001,
+      "step": 238340
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.7696487334506766e-05,
+      "loss": 1.0296,
+      "step": 238350
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.769533940460702e-05,
+      "loss": 1.1535,
+      "step": 238360
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.769419147470727e-05,
+      "loss": 1.188,
+      "step": 238370
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.769304354480753e-05,
+      "loss": 0.9691,
+      "step": 238380
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.7691895614907786e-05,
+      "loss": 1.0452,
+      "step": 238390
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.769074768500803e-05,
+      "loss": 1.1497,
+      "step": 238400
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.7689599755108286e-05,
+      "loss": 1.0247,
+      "step": 238410
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.768845182520854e-05,
+      "loss": 1.0414,
+      "step": 238420
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.7687303895308785e-05,
+      "loss": 1.003,
+      "step": 238430
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.768615596540904e-05,
+      "loss": 0.9983,
+      "step": 238440
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.76850080355093e-05,
+      "loss": 1.009,
+      "step": 238450
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.768386010560954e-05,
+      "loss": 1.0709,
+      "step": 238460
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.76827121757098e-05,
+      "loss": 1.1556,
+      "step": 238470
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.7681564245810055e-05,
+      "loss": 1.1366,
+      "step": 238480
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.7680416315910305e-05,
+      "loss": 1.0517,
+      "step": 238490
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.7679268386010555e-05,
+      "loss": 1.1371,
+      "step": 238500
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.767812045611081e-05,
+      "loss": 0.9873,
+      "step": 238510
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.767697252621106e-05,
+      "loss": 0.9867,
+      "step": 238520
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.767582459631132e-05,
+      "loss": 1.1045,
+      "step": 238530
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.7674676666411574e-05,
+      "loss": 1.1873,
+      "step": 238540
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.767352873651182e-05,
+      "loss": 1.0227,
+      "step": 238550
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.7672380806612074e-05,
+      "loss": 1.0693,
+      "step": 238560
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.767123287671233e-05,
+      "loss": 1.0731,
+      "step": 238570
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.7670084946812574e-05,
+      "loss": 1.0451,
+      "step": 238580
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.766893701691283e-05,
+      "loss": 1.1634,
+      "step": 238590
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.766778908701309e-05,
+      "loss": 1.2479,
+      "step": 238600
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.766664115711333e-05,
+      "loss": 1.0863,
+      "step": 238610
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.766549322721359e-05,
+      "loss": 1.1224,
+      "step": 238620
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.766434529731384e-05,
+      "loss": 1.075,
+      "step": 238630
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.766319736741409e-05,
+      "loss": 1.0563,
+      "step": 238640
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.766204943751434e-05,
+      "loss": 1.0695,
+      "step": 238650
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.766090150761459e-05,
+      "loss": 1.0909,
+      "step": 238660
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.765975357771485e-05,
+      "loss": 1.065,
+      "step": 238670
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.7658605647815106e-05,
+      "loss": 1.1966,
+      "step": 238680
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.765745771791535e-05,
+      "loss": 1.0773,
+      "step": 238690
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.7656309788015606e-05,
+      "loss": 1.0921,
+      "step": 238700
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.765516185811586e-05,
+      "loss": 1.1296,
+      "step": 238710
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.7654013928216106e-05,
+      "loss": 1.1889,
+      "step": 238720
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.765286599831636e-05,
+      "loss": 1.159,
+      "step": 238730
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.765171806841662e-05,
+      "loss": 1.1278,
+      "step": 238740
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.765057013851687e-05,
+      "loss": 1.1383,
+      "step": 238750
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.764942220861712e-05,
+      "loss": 1.0005,
+      "step": 238760
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.7648274278717375e-05,
+      "loss": 1.1166,
+      "step": 238770
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.7647126348817625e-05,
+      "loss": 1.1516,
+      "step": 238780
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.764597841891788e-05,
+      "loss": 1.0986,
+      "step": 238790
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.764483048901813e-05,
+      "loss": 1.0851,
+      "step": 238800
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.764368255911838e-05,
+      "loss": 1.0654,
+      "step": 238810
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.764253462921864e-05,
+      "loss": 1.0058,
+      "step": 238820
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.7641386699318895e-05,
+      "loss": 1.225,
+      "step": 238830
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.764023876941914e-05,
+      "loss": 1.0881,
+      "step": 238840
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.7639090839519394e-05,
+      "loss": 1.0455,
+      "step": 238850
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.763794290961965e-05,
+      "loss": 1.0352,
+      "step": 238860
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.7636794979719894e-05,
+      "loss": 1.2094,
+      "step": 238870
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.763564704982015e-05,
+      "loss": 1.0943,
+      "step": 238880
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.763449911992041e-05,
+      "loss": 1.1367,
+      "step": 238890
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.763335119002066e-05,
+      "loss": 1.0529,
+      "step": 238900
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.763220326012091e-05,
+      "loss": 0.9447,
+      "step": 238910
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.7631055330221164e-05,
+      "loss": 1.1002,
+      "step": 238920
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.7629907400321413e-05,
+      "loss": 1.1605,
+      "step": 238930
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.762875947042167e-05,
+      "loss": 1.1257,
+      "step": 238940
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.762761154052192e-05,
+      "loss": 1.1049,
+      "step": 238950
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.762646361062217e-05,
+      "loss": 0.9828,
+      "step": 238960
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.7625315680722426e-05,
+      "loss": 1.1863,
+      "step": 238970
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.762416775082268e-05,
+      "loss": 1.1526,
+      "step": 238980
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.7623019820922926e-05,
+      "loss": 1.0285,
+      "step": 238990
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.762187189102318e-05,
+      "loss": 1.1261,
+      "step": 239000
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.762072396112344e-05,
+      "loss": 1.0937,
+      "step": 239010
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.761957603122368e-05,
+      "loss": 0.9375,
+      "step": 239020
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.761842810132394e-05,
+      "loss": 1.1266,
+      "step": 239030
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.7617280171424196e-05,
+      "loss": 0.8946,
+      "step": 239040
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.7616132241524446e-05,
+      "loss": 1.009,
+      "step": 239050
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.7614984311624695e-05,
+      "loss": 1.0013,
+      "step": 239060
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.761383638172495e-05,
+      "loss": 1.1335,
+      "step": 239070
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.76126884518252e-05,
+      "loss": 1.0746,
+      "step": 239080
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.761154052192546e-05,
+      "loss": 1.0744,
+      "step": 239090
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.761039259202571e-05,
+      "loss": 1.1883,
+      "step": 239100
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.760924466212596e-05,
+      "loss": 0.9828,
+      "step": 239110
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.7608096732226215e-05,
+      "loss": 1.0478,
+      "step": 239120
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.760694880232647e-05,
+      "loss": 1.0644,
+      "step": 239130
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.7605800872426715e-05,
+      "loss": 1.123,
+      "step": 239140
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.760465294252697e-05,
+      "loss": 1.086,
+      "step": 239150
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.760350501262723e-05,
+      "loss": 1.2197,
+      "step": 239160
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.760235708272747e-05,
+      "loss": 1.1364,
+      "step": 239170
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.760120915282773e-05,
+      "loss": 0.9965,
+      "step": 239180
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.7600061222927984e-05,
+      "loss": 1.1073,
+      "step": 239190
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.7598913293028234e-05,
+      "loss": 1.0945,
+      "step": 239200
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.7597765363128484e-05,
+      "loss": 1.0141,
+      "step": 239210
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.759661743322874e-05,
+      "loss": 1.1055,
+      "step": 239220
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.759546950332899e-05,
+      "loss": 1.0598,
+      "step": 239230
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.759432157342925e-05,
+      "loss": 1.0261,
+      "step": 239240
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.75931736435295e-05,
+      "loss": 1.0215,
+      "step": 239250
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.759202571362975e-05,
+      "loss": 1.0101,
+      "step": 239260
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.759087778373e-05,
+      "loss": 1.138,
+      "step": 239270
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.758972985383026e-05,
+      "loss": 1.1091,
+      "step": 239280
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.75885819239305e-05,
+      "loss": 0.9925,
+      "step": 239290
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.758743399403076e-05,
+      "loss": 1.0679,
+      "step": 239300
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.7586286064131016e-05,
+      "loss": 1.0176,
+      "step": 239310
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.758513813423126e-05,
+      "loss": 1.1501,
+      "step": 239320
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.7583990204331516e-05,
+      "loss": 1.1386,
+      "step": 239330
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.758284227443177e-05,
+      "loss": 1.1537,
+      "step": 239340
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.758169434453202e-05,
+      "loss": 1.1096,
+      "step": 239350
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.758054641463227e-05,
+      "loss": 1.1733,
+      "step": 239360
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.757939848473253e-05,
+      "loss": 1.1037,
+      "step": 239370
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.757825055483278e-05,
+      "loss": 1.0736,
+      "step": 239380
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.7577102624933036e-05,
+      "loss": 1.0986,
+      "step": 239390
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.7575954695033285e-05,
+      "loss": 1.0245,
+      "step": 239400
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.7574806765133535e-05,
+      "loss": 0.9933,
+      "step": 239410
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.757365883523379e-05,
+      "loss": 1.1521,
+      "step": 239420
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.757251090533405e-05,
+      "loss": 1.0329,
+      "step": 239430
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.757136297543429e-05,
+      "loss": 1.0516,
+      "step": 239440
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.757021504553455e-05,
+      "loss": 1.1207,
+      "step": 239450
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.7569067115634805e-05,
+      "loss": 0.9346,
+      "step": 239460
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.756791918573505e-05,
+      "loss": 1.0346,
+      "step": 239470
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.7566771255835305e-05,
+      "loss": 1.1121,
+      "step": 239480
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.756562332593556e-05,
+      "loss": 1.1753,
+      "step": 239490
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.756447539603581e-05,
+      "loss": 1.0705,
+      "step": 239500
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.756332746613606e-05,
+      "loss": 1.0098,
+      "step": 239510
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.756217953623632e-05,
+      "loss": 1.0886,
+      "step": 239520
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.756103160633657e-05,
+      "loss": 1.0775,
+      "step": 239530
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.7559883676436824e-05,
+      "loss": 0.9915,
+      "step": 239540
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.7558735746537074e-05,
+      "loss": 1.1073,
+      "step": 239550
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.7557587816637324e-05,
+      "loss": 1.0594,
+      "step": 239560
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.755643988673758e-05,
+      "loss": 1.0268,
+      "step": 239570
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.755529195683784e-05,
+      "loss": 1.0874,
+      "step": 239580
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.755414402693808e-05,
+      "loss": 0.9602,
+      "step": 239590
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.755299609703834e-05,
+      "loss": 1.257,
+      "step": 239600
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.755184816713859e-05,
+      "loss": 1.2022,
+      "step": 239610
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.7550700237238836e-05,
+      "loss": 1.0125,
+      "step": 239620
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.754955230733909e-05,
+      "loss": 0.9946,
+      "step": 239630
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.754840437743935e-05,
+      "loss": 1.1413,
+      "step": 239640
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.75472564475396e-05,
+      "loss": 1.0791,
+      "step": 239650
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.754610851763985e-05,
+      "loss": 1.1705,
+      "step": 239660
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.7544960587740106e-05,
+      "loss": 1.0485,
+      "step": 239670
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.7543812657840356e-05,
+      "loss": 1.0847,
+      "step": 239680
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.754266472794061e-05,
+      "loss": 1.136,
+      "step": 239690
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.754151679804086e-05,
+      "loss": 1.0142,
+      "step": 239700
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.754036886814111e-05,
+      "loss": 1.0999,
+      "step": 239710
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.753922093824137e-05,
+      "loss": 1.128,
+      "step": 239720
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.7538073008341625e-05,
+      "loss": 1.1347,
+      "step": 239730
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.753692507844187e-05,
+      "loss": 1.1142,
+      "step": 239740
+    },
+    {
+      "epoch": 11.0,
+      "eval_cer": 0.6689142960345541,
+      "eval_em": 0.007357212003872217,
+      "eval_f1": 0.007357212003872217,
+      "eval_loss": 1.2073729038238525,
+      "eval_runtime": 1302.9026,
+      "eval_samples_per_second": 7.928,
+      "eval_steps_per_second": 0.992,
+      "eval_wer": 0.9926427879961278,
+      "step": 239749
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.7535777148542125e-05,
+      "loss": 1.2098,
+      "step": 239750
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.753462921864238e-05,
+      "loss": 1.016,
+      "step": 239760
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.7533481288742625e-05,
+      "loss": 1.0891,
+      "step": 239770
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.753233335884288e-05,
+      "loss": 0.9627,
+      "step": 239780
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.753118542894314e-05,
+      "loss": 0.9768,
+      "step": 239790
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.753003749904339e-05,
+      "loss": 1.0564,
+      "step": 239800
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.752888956914364e-05,
+      "loss": 1.0898,
+      "step": 239810
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.7527741639243894e-05,
+      "loss": 0.9467,
+      "step": 239820
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.7526593709344144e-05,
+      "loss": 1.0435,
+      "step": 239830
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.75254457794444e-05,
+      "loss": 0.9693,
+      "step": 239840
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.752429784954465e-05,
+      "loss": 1.0163,
+      "step": 239850
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.75231499196449e-05,
+      "loss": 0.9726,
+      "step": 239860
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.752200198974516e-05,
+      "loss": 1.0211,
+      "step": 239870
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.7520854059845414e-05,
+      "loss": 1.07,
+      "step": 239880
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.751970612994566e-05,
+      "loss": 1.0405,
+      "step": 239890
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.7518558200045914e-05,
+      "loss": 1.1308,
+      "step": 239900
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.751741027014617e-05,
+      "loss": 1.0527,
+      "step": 239910
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.751626234024641e-05,
+      "loss": 0.9371,
+      "step": 239920
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.751511441034667e-05,
+      "loss": 0.9825,
+      "step": 239930
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.7513966480446927e-05,
+      "loss": 0.9965,
+      "step": 239940
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.7512818550547176e-05,
+      "loss": 1.1263,
+      "step": 239950
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.7511670620647426e-05,
+      "loss": 0.9826,
+      "step": 239960
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.751052269074768e-05,
+      "loss": 0.9911,
+      "step": 239970
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.750937476084793e-05,
+      "loss": 0.9909,
+      "step": 239980
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.750822683094819e-05,
+      "loss": 0.9623,
+      "step": 239990
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.750707890104844e-05,
+      "loss": 0.9966,
+      "step": 240000
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.750593097114869e-05,
+      "loss": 1.1017,
+      "step": 240010
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.7504783041248946e-05,
+      "loss": 1.2014,
+      "step": 240020
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.750363511134919e-05,
+      "loss": 1.0018,
+      "step": 240030
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.7502487181449445e-05,
+      "loss": 1.0921,
+      "step": 240040
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.75013392515497e-05,
+      "loss": 1.0152,
+      "step": 240050
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.750019132164995e-05,
+      "loss": 1.0259,
+      "step": 240060
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.74990433917502e-05,
+      "loss": 0.9992,
+      "step": 240070
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.749789546185046e-05,
+      "loss": 1.0318,
+      "step": 240080
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.749674753195071e-05,
+      "loss": 1.0693,
+      "step": 240090
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.7495599602050965e-05,
+      "loss": 0.932,
+      "step": 240100
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.7494451672151215e-05,
+      "loss": 0.9851,
+      "step": 240110
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.7493303742251465e-05,
+      "loss": 0.9458,
+      "step": 240120
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.749215581235172e-05,
+      "loss": 0.9517,
+      "step": 240130
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.749100788245198e-05,
+      "loss": 0.9709,
+      "step": 240140
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.748985995255222e-05,
+      "loss": 0.9815,
+      "step": 240150
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.748871202265248e-05,
+      "loss": 0.9603,
+      "step": 240160
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.7487564092752734e-05,
+      "loss": 1.1311,
+      "step": 240170
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.748641616285298e-05,
+      "loss": 0.989,
+      "step": 240180
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.7485268232953234e-05,
+      "loss": 1.0514,
+      "step": 240190
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.748412030305349e-05,
+      "loss": 1.0425,
+      "step": 240200
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.748297237315374e-05,
+      "loss": 1.0093,
+      "step": 240210
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.748182444325399e-05,
+      "loss": 0.9808,
+      "step": 240220
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.748067651335425e-05,
+      "loss": 0.88,
+      "step": 240230
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.74795285834545e-05,
+      "loss": 0.9511,
+      "step": 240240
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.747838065355475e-05,
+      "loss": 0.987,
+      "step": 240250
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.7477232723655e-05,
+      "loss": 1.0992,
+      "step": 240260
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.747608479375525e-05,
+      "loss": 1.0462,
+      "step": 240270
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.747493686385551e-05,
+      "loss": 0.9481,
+      "step": 240280
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.7473788933955766e-05,
+      "loss": 1.0356,
+      "step": 240290
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.747264100405601e-05,
+      "loss": 1.1064,
+      "step": 240300
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.7471493074156266e-05,
+      "loss": 1.0456,
+      "step": 240310
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.747034514425652e-05,
+      "loss": 0.9845,
+      "step": 240320
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.7469197214356766e-05,
+      "loss": 1.1773,
+      "step": 240330
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.746804928445702e-05,
+      "loss": 0.9878,
+      "step": 240340
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.746690135455728e-05,
+      "loss": 1.0805,
+      "step": 240350
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.746575342465753e-05,
+      "loss": 1.0589,
+      "step": 240360
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.746460549475778e-05,
+      "loss": 1.1305,
+      "step": 240370
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.7463457564858035e-05,
+      "loss": 1.2136,
+      "step": 240380
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.7462309634958285e-05,
+      "loss": 0.9954,
+      "step": 240390
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.746116170505854e-05,
+      "loss": 0.915,
+      "step": 240400
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.746001377515879e-05,
+      "loss": 1.1332,
+      "step": 240410
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.745886584525904e-05,
+      "loss": 1.0821,
+      "step": 240420
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.74577179153593e-05,
+      "loss": 1.0319,
+      "step": 240430
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.7456569985459555e-05,
+      "loss": 0.9389,
+      "step": 240440
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.74554220555598e-05,
+      "loss": 1.0409,
+      "step": 240450
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.7454274125660055e-05,
+      "loss": 1.0207,
+      "step": 240460
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.745312619576031e-05,
+      "loss": 1.0022,
+      "step": 240470
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.7451978265860554e-05,
+      "loss": 1.1908,
+      "step": 240480
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.745083033596081e-05,
+      "loss": 1.0157,
+      "step": 240490
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.744968240606107e-05,
+      "loss": 0.9729,
+      "step": 240500
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.744853447616132e-05,
+      "loss": 0.9798,
+      "step": 240510
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.744738654626157e-05,
+      "loss": 1.046,
+      "step": 240520
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.7446238616361824e-05,
+      "loss": 0.9825,
+      "step": 240530
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.7445090686462074e-05,
+      "loss": 0.9901,
+      "step": 240540
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.744394275656233e-05,
+      "loss": 1.186,
+      "step": 240550
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.744279482666258e-05,
+      "loss": 0.9189,
+      "step": 240560
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.744164689676283e-05,
+      "loss": 1.066,
+      "step": 240570
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.744049896686309e-05,
+      "loss": 1.0548,
+      "step": 240580
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.743935103696334e-05,
+      "loss": 1.0045,
+      "step": 240590
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.7438203107063586e-05,
+      "loss": 0.9475,
+      "step": 240600
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.743705517716384e-05,
+      "loss": 0.9972,
+      "step": 240610
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.74359072472641e-05,
+      "loss": 1.0485,
+      "step": 240620
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.743475931736434e-05,
+      "loss": 0.9759,
+      "step": 240630
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.74336113874646e-05,
+      "loss": 1.0425,
+      "step": 240640
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.7432463457564856e-05,
+      "loss": 1.016,
+      "step": 240650
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.7431315527665106e-05,
+      "loss": 1.1009,
+      "step": 240660
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.7430167597765356e-05,
+      "loss": 1.0242,
+      "step": 240670
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.742901966786561e-05,
+      "loss": 0.9972,
+      "step": 240680
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.742787173796586e-05,
+      "loss": 1.0171,
+      "step": 240690
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.742672380806612e-05,
+      "loss": 1.1094,
+      "step": 240700
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.742557587816637e-05,
+      "loss": 0.9357,
+      "step": 240710
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.742442794826662e-05,
+      "loss": 1.0761,
+      "step": 240720
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.7423280018366875e-05,
+      "loss": 0.9202,
+      "step": 240730
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.742213208846713e-05,
+      "loss": 0.9897,
+      "step": 240740
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.7420984158567375e-05,
+      "loss": 1.1098,
+      "step": 240750
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.741983622866763e-05,
+      "loss": 0.8957,
+      "step": 240760
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.741868829876789e-05,
+      "loss": 1.0077,
+      "step": 240770
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.741754036886813e-05,
+      "loss": 1.0162,
+      "step": 240780
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.741639243896839e-05,
+      "loss": 1.1075,
+      "step": 240790
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.7415244509068644e-05,
+      "loss": 0.9762,
+      "step": 240800
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.7414096579168894e-05,
+      "loss": 1.0447,
+      "step": 240810
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.7412948649269144e-05,
+      "loss": 0.977,
+      "step": 240820
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.74118007193694e-05,
+      "loss": 1.0408,
+      "step": 240830
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.741065278946965e-05,
+      "loss": 1.0703,
+      "step": 240840
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.740950485956991e-05,
+      "loss": 1.1357,
+      "step": 240850
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.740835692967016e-05,
+      "loss": 0.9185,
+      "step": 240860
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.740720899977041e-05,
+      "loss": 1.0165,
+      "step": 240870
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.7406061069870664e-05,
+      "loss": 0.9773,
+      "step": 240880
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.740491313997092e-05,
+      "loss": 1.0725,
+      "step": 240890
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.740376521007116e-05,
+      "loss": 1.0644,
+      "step": 240900
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.740261728017142e-05,
+      "loss": 1.0481,
+      "step": 240910
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.7401469350271677e-05,
+      "loss": 1.0381,
+      "step": 240920
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.740032142037192e-05,
+      "loss": 1.0839,
+      "step": 240930
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.7399173490472176e-05,
+      "loss": 0.9229,
+      "step": 240940
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.739802556057243e-05,
+      "loss": 1.081,
+      "step": 240950
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.739687763067268e-05,
+      "loss": 0.9446,
+      "step": 240960
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.739572970077293e-05,
+      "loss": 1.0961,
+      "step": 240970
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.739458177087319e-05,
+      "loss": 1.0105,
+      "step": 240980
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.739343384097344e-05,
+      "loss": 1.0582,
+      "step": 240990
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.7392285911073696e-05,
+      "loss": 0.9633,
+      "step": 241000
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.7391137981173946e-05,
+      "loss": 1.0461,
+      "step": 241010
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.7389990051274195e-05,
+      "loss": 1.1074,
+      "step": 241020
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.738884212137445e-05,
+      "loss": 0.9515,
+      "step": 241030
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.738769419147471e-05,
+      "loss": 0.9734,
+      "step": 241040
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.738654626157495e-05,
+      "loss": 1.1055,
+      "step": 241050
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.738539833167521e-05,
+      "loss": 1.0357,
+      "step": 241060
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.7384250401775465e-05,
+      "loss": 1.0279,
+      "step": 241070
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.738310247187571e-05,
+      "loss": 1.0755,
+      "step": 241080
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.7381954541975965e-05,
+      "loss": 0.958,
+      "step": 241090
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.738080661207622e-05,
+      "loss": 1.0183,
+      "step": 241100
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.737965868217647e-05,
+      "loss": 0.9861,
+      "step": 241110
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.737851075227672e-05,
+      "loss": 0.9344,
+      "step": 241120
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.737736282237698e-05,
+      "loss": 0.989,
+      "step": 241130
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.737621489247723e-05,
+      "loss": 0.9825,
+      "step": 241140
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.7375066962577484e-05,
+      "loss": 1.1522,
+      "step": 241150
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.7373919032677734e-05,
+      "loss": 1.0429,
+      "step": 241160
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.7372771102777984e-05,
+      "loss": 1.008,
+      "step": 241170
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.737162317287824e-05,
+      "loss": 1.0097,
+      "step": 241180
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.73704752429785e-05,
+      "loss": 1.0723,
+      "step": 241190
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.736932731307874e-05,
+      "loss": 1.0916,
+      "step": 241200
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.7368179383179e-05,
+      "loss": 0.9494,
+      "step": 241210
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.7367031453279254e-05,
+      "loss": 1.0581,
+      "step": 241220
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.7365883523379497e-05,
+      "loss": 0.967,
+      "step": 241230
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.736473559347975e-05,
+      "loss": 1.0279,
+      "step": 241240
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.736358766358001e-05,
+      "loss": 1.082,
+      "step": 241250
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.736243973368026e-05,
+      "loss": 1.0128,
+      "step": 241260
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.736129180378051e-05,
+      "loss": 1.031,
+      "step": 241270
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.7360143873880766e-05,
+      "loss": 1.0386,
+      "step": 241280
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.7358995943981016e-05,
+      "loss": 0.9307,
+      "step": 241290
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.735784801408127e-05,
+      "loss": 0.9391,
+      "step": 241300
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.735670008418152e-05,
+      "loss": 1.1037,
+      "step": 241310
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.735555215428177e-05,
+      "loss": 1.0354,
+      "step": 241320
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.735440422438203e-05,
+      "loss": 0.99,
+      "step": 241330
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.7353256294482286e-05,
+      "loss": 1.0989,
+      "step": 241340
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.735210836458253e-05,
+      "loss": 1.0513,
+      "step": 241350
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.7350960434682785e-05,
+      "loss": 1.0116,
+      "step": 241360
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.734981250478304e-05,
+      "loss": 0.914,
+      "step": 241370
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.7348664574883285e-05,
+      "loss": 0.9406,
+      "step": 241380
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.734751664498354e-05,
+      "loss": 0.9591,
+      "step": 241390
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.734636871508379e-05,
+      "loss": 1.0005,
+      "step": 241400
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.734522078518405e-05,
+      "loss": 1.055,
+      "step": 241410
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.73440728552843e-05,
+      "loss": 0.9898,
+      "step": 241420
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.734292492538455e-05,
+      "loss": 0.9206,
+      "step": 241430
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.7341776995484805e-05,
+      "loss": 0.9073,
+      "step": 241440
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.734062906558506e-05,
+      "loss": 0.9591,
+      "step": 241450
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.7339481135685304e-05,
+      "loss": 0.9985,
+      "step": 241460
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.733833320578556e-05,
+      "loss": 1.0393,
+      "step": 241470
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.733718527588582e-05,
+      "loss": 1.0898,
+      "step": 241480
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.733603734598606e-05,
+      "loss": 1.0425,
+      "step": 241490
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.733488941608632e-05,
+      "loss": 0.9732,
+      "step": 241500
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.7333741486186574e-05,
+      "loss": 0.8966,
+      "step": 241510
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.7332593556286824e-05,
+      "loss": 1.0842,
+      "step": 241520
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.7331445626387074e-05,
+      "loss": 0.99,
+      "step": 241530
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.733029769648733e-05,
+      "loss": 1.0567,
+      "step": 241540
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.732914976658758e-05,
+      "loss": 0.9654,
+      "step": 241550
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.732800183668784e-05,
+      "loss": 1.0897,
+      "step": 241560
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.7326853906788087e-05,
+      "loss": 1.0132,
+      "step": 241570
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.7325705976888336e-05,
+      "loss": 1.0401,
+      "step": 241580
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.732455804698859e-05,
+      "loss": 1.0305,
+      "step": 241590
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.732341011708885e-05,
+      "loss": 1.1003,
+      "step": 241600
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.732226218718909e-05,
+      "loss": 0.9939,
+      "step": 241610
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.732111425728935e-05,
+      "loss": 0.9584,
+      "step": 241620
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.7319966327389606e-05,
+      "loss": 1.0107,
+      "step": 241630
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.731881839748985e-05,
+      "loss": 1.1755,
+      "step": 241640
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.7317670467590106e-05,
+      "loss": 0.9903,
+      "step": 241650
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.731652253769036e-05,
+      "loss": 1.1334,
+      "step": 241660
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.731537460779061e-05,
+      "loss": 0.9774,
+      "step": 241670
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.731422667789086e-05,
+      "loss": 0.9834,
+      "step": 241680
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.731307874799112e-05,
+      "loss": 0.8933,
+      "step": 241690
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.731193081809137e-05,
+      "loss": 0.9368,
+      "step": 241700
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.7310782888191625e-05,
+      "loss": 1.0569,
+      "step": 241710
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.7309634958291875e-05,
+      "loss": 1.0731,
+      "step": 241720
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.7308487028392125e-05,
+      "loss": 0.9643,
+      "step": 241730
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.730733909849238e-05,
+      "loss": 0.9603,
+      "step": 241740
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.730619116859264e-05,
+      "loss": 0.9561,
+      "step": 241750
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.730504323869288e-05,
+      "loss": 0.983,
+      "step": 241760
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.730389530879314e-05,
+      "loss": 1.0598,
+      "step": 241770
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.7302747378893394e-05,
+      "loss": 1.0395,
+      "step": 241780
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.730159944899364e-05,
+      "loss": 0.9127,
+      "step": 241790
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.7300451519093894e-05,
+      "loss": 1.0113,
+      "step": 241800
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.729930358919415e-05,
+      "loss": 1.0431,
+      "step": 241810
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.72981556592944e-05,
+      "loss": 1.1017,
+      "step": 241820
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.729700772939465e-05,
+      "loss": 0.9698,
+      "step": 241830
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.729585979949491e-05,
+      "loss": 1.0743,
+      "step": 241840
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.729471186959516e-05,
+      "loss": 0.984,
+      "step": 241850
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.7293563939695414e-05,
+      "loss": 1.1164,
+      "step": 241860
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.7292416009795663e-05,
+      "loss": 1.0254,
+      "step": 241870
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.729126807989591e-05,
+      "loss": 0.8992,
+      "step": 241880
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.729012014999617e-05,
+      "loss": 0.9615,
+      "step": 241890
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.7288972220096427e-05,
+      "loss": 1.0266,
+      "step": 241900
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.728782429019667e-05,
+      "loss": 1.0609,
+      "step": 241910
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.7286676360296926e-05,
+      "loss": 0.9557,
+      "step": 241920
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.728552843039718e-05,
+      "loss": 1.0745,
+      "step": 241930
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.7284380500497426e-05,
+      "loss": 1.1072,
+      "step": 241940
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.728323257059768e-05,
+      "loss": 0.9855,
+      "step": 241950
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.728208464069794e-05,
+      "loss": 0.9227,
+      "step": 241960
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.728093671079819e-05,
+      "loss": 0.9678,
+      "step": 241970
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.727978878089844e-05,
+      "loss": 0.9729,
+      "step": 241980
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.7278640850998696e-05,
+      "loss": 0.9891,
+      "step": 241990
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.7277492921098945e-05,
+      "loss": 1.0178,
+      "step": 242000
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.72763449911992e-05,
+      "loss": 0.9273,
+      "step": 242010
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.727519706129945e-05,
+      "loss": 1.0288,
+      "step": 242020
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.72740491313997e-05,
+      "loss": 1.065,
+      "step": 242030
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.727290120149996e-05,
+      "loss": 1.0588,
+      "step": 242040
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.7271753271600215e-05,
+      "loss": 0.9604,
+      "step": 242050
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.727060534170046e-05,
+      "loss": 1.0689,
+      "step": 242060
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.7269457411800715e-05,
+      "loss": 1.1495,
+      "step": 242070
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.726830948190097e-05,
+      "loss": 1.0182,
+      "step": 242080
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.7267161552001214e-05,
+      "loss": 1.0098,
+      "step": 242090
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.726601362210147e-05,
+      "loss": 1.0255,
+      "step": 242100
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.726486569220173e-05,
+      "loss": 1.0648,
+      "step": 242110
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.726371776230198e-05,
+      "loss": 0.9848,
+      "step": 242120
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.726256983240223e-05,
+      "loss": 1.1847,
+      "step": 242130
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.7261421902502484e-05,
+      "loss": 0.9789,
+      "step": 242140
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.7260273972602734e-05,
+      "loss": 1.0859,
+      "step": 242150
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.725912604270299e-05,
+      "loss": 0.9823,
+      "step": 242160
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.725797811280324e-05,
+      "loss": 1.0014,
+      "step": 242170
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.725683018290349e-05,
+      "loss": 0.9308,
+      "step": 242180
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.725568225300375e-05,
+      "loss": 1.0772,
+      "step": 242190
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.7254534323104004e-05,
+      "loss": 1.0461,
+      "step": 242200
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.7253386393204247e-05,
+      "loss": 0.9914,
+      "step": 242210
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.72522384633045e-05,
+      "loss": 1.0651,
+      "step": 242220
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.725109053340476e-05,
+      "loss": 1.0715,
+      "step": 242230
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.7249942603505e-05,
+      "loss": 1.0705,
+      "step": 242240
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.724879467360526e-05,
+      "loss": 0.9919,
+      "step": 242250
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.7247646743705516e-05,
+      "loss": 0.9575,
+      "step": 242260
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.7246498813805766e-05,
+      "loss": 1.0367,
+      "step": 242270
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.7245350883906016e-05,
+      "loss": 1.0118,
+      "step": 242280
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.724420295400627e-05,
+      "loss": 0.9355,
+      "step": 242290
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.724305502410652e-05,
+      "loss": 1.0732,
+      "step": 242300
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.724190709420678e-05,
+      "loss": 1.0075,
+      "step": 242310
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.724075916430703e-05,
+      "loss": 1.047,
+      "step": 242320
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.723961123440728e-05,
+      "loss": 1.101,
+      "step": 242330
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.7238463304507535e-05,
+      "loss": 1.162,
+      "step": 242340
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.723731537460779e-05,
+      "loss": 0.9656,
+      "step": 242350
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.7236167444708035e-05,
+      "loss": 1.0109,
+      "step": 242360
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.723501951480829e-05,
+      "loss": 1.0459,
+      "step": 242370
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.723387158490855e-05,
+      "loss": 0.9518,
+      "step": 242380
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.723272365500879e-05,
+      "loss": 0.9945,
+      "step": 242390
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.723157572510905e-05,
+      "loss": 0.9817,
+      "step": 242400
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.7230427795209305e-05,
+      "loss": 1.0167,
+      "step": 242410
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.7229279865309555e-05,
+      "loss": 1.1341,
+      "step": 242420
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.7228131935409804e-05,
+      "loss": 1.1308,
+      "step": 242430
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.722698400551006e-05,
+      "loss": 1.0262,
+      "step": 242440
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.722583607561031e-05,
+      "loss": 1.0339,
+      "step": 242450
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.722468814571057e-05,
+      "loss": 0.9811,
+      "step": 242460
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.722354021581082e-05,
+      "loss": 1.0562,
+      "step": 242470
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.722239228591107e-05,
+      "loss": 1.1122,
+      "step": 242480
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.7221244356011324e-05,
+      "loss": 1.041,
+      "step": 242490
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.722009642611158e-05,
+      "loss": 1.0247,
+      "step": 242500
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.7218948496211824e-05,
+      "loss": 1.0237,
+      "step": 242510
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.721780056631208e-05,
+      "loss": 1.2438,
+      "step": 242520
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.721665263641234e-05,
+      "loss": 1.0115,
+      "step": 242530
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.721550470651258e-05,
+      "loss": 0.9548,
+      "step": 242540
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.7214356776612837e-05,
+      "loss": 0.9711,
+      "step": 242550
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.721320884671309e-05,
+      "loss": 1.0665,
+      "step": 242560
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.721206091681334e-05,
+      "loss": 0.9773,
+      "step": 242570
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.721091298691359e-05,
+      "loss": 1.0642,
+      "step": 242580
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.720976505701385e-05,
+      "loss": 1.1228,
+      "step": 242590
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.72086171271141e-05,
+      "loss": 1.0149,
+      "step": 242600
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.7207469197214356e-05,
+      "loss": 1.0476,
+      "step": 242610
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.7206321267314606e-05,
+      "loss": 1.0453,
+      "step": 242620
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.7205173337414856e-05,
+      "loss": 1.0316,
+      "step": 242630
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.720402540751511e-05,
+      "loss": 1.0263,
+      "step": 242640
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.720287747761537e-05,
+      "loss": 0.8906,
+      "step": 242650
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.720172954771561e-05,
+      "loss": 1.0399,
+      "step": 242660
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.720058161781587e-05,
+      "loss": 1.0129,
+      "step": 242670
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.7199433687916125e-05,
+      "loss": 0.9243,
+      "step": 242680
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.719828575801637e-05,
+      "loss": 0.9546,
+      "step": 242690
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.7197137828116625e-05,
+      "loss": 1.1222,
+      "step": 242700
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.719598989821688e-05,
+      "loss": 0.885,
+      "step": 242710
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.719484196831713e-05,
+      "loss": 0.9828,
+      "step": 242720
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.719369403841738e-05,
+      "loss": 1.0261,
+      "step": 242730
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.719254610851764e-05,
+      "loss": 1.0588,
+      "step": 242740
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.719139817861789e-05,
+      "loss": 0.9796,
+      "step": 242750
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.7190250248718144e-05,
+      "loss": 1.0727,
+      "step": 242760
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.718910231881839e-05,
+      "loss": 1.0391,
+      "step": 242770
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.7187954388918644e-05,
+      "loss": 1.042,
+      "step": 242780
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.71868064590189e-05,
+      "loss": 1.0665,
+      "step": 242790
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.7185658529119144e-05,
+      "loss": 1.0811,
+      "step": 242800
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.71845105992194e-05,
+      "loss": 1.0301,
+      "step": 242810
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.718336266931966e-05,
+      "loss": 0.9623,
+      "step": 242820
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.718221473941991e-05,
+      "loss": 1.0345,
+      "step": 242830
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.718106680952016e-05,
+      "loss": 1.028,
+      "step": 242840
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.7179918879620413e-05,
+      "loss": 0.9352,
+      "step": 242850
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.717877094972066e-05,
+      "loss": 1.0273,
+      "step": 242860
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.717762301982092e-05,
+      "loss": 1.0378,
+      "step": 242870
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.717647508992117e-05,
+      "loss": 0.9774,
+      "step": 242880
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.717532716002142e-05,
+      "loss": 0.9378,
+      "step": 242890
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.7174179230121676e-05,
+      "loss": 0.9223,
+      "step": 242900
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.717303130022193e-05,
+      "loss": 1.075,
+      "step": 242910
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.7171883370322176e-05,
+      "loss": 1.0012,
+      "step": 242920
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.717073544042243e-05,
+      "loss": 1.0023,
+      "step": 242930
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.716958751052269e-05,
+      "loss": 1.1094,
+      "step": 242940
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.716843958062293e-05,
+      "loss": 1.0155,
+      "step": 242950
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.716729165072319e-05,
+      "loss": 1.0316,
+      "step": 242960
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.7166143720823446e-05,
+      "loss": 0.8561,
+      "step": 242970
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.7164995790923695e-05,
+      "loss": 0.9942,
+      "step": 242980
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.7163847861023945e-05,
+      "loss": 1.0782,
+      "step": 242990
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.71626999311242e-05,
+      "loss": 0.9967,
+      "step": 243000
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.716155200122445e-05,
+      "loss": 1.1284,
+      "step": 243010
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.716040407132471e-05,
+      "loss": 0.9914,
+      "step": 243020
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.715925614142496e-05,
+      "loss": 1.0222,
+      "step": 243030
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.715810821152521e-05,
+      "loss": 0.979,
+      "step": 243040
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.7156960281625465e-05,
+      "loss": 0.9551,
+      "step": 243050
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.715581235172572e-05,
+      "loss": 1.0766,
+      "step": 243060
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.7154664421825964e-05,
+      "loss": 1.1177,
+      "step": 243070
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.715351649192622e-05,
+      "loss": 1.0127,
+      "step": 243080
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.715236856202648e-05,
+      "loss": 1.0381,
+      "step": 243090
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.715122063212672e-05,
+      "loss": 0.9419,
+      "step": 243100
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.715007270222698e-05,
+      "loss": 1.0888,
+      "step": 243110
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.7148924772327234e-05,
+      "loss": 1.0281,
+      "step": 243120
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.7147776842427484e-05,
+      "loss": 1.0558,
+      "step": 243130
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.7146628912527734e-05,
+      "loss": 0.9681,
+      "step": 243140
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.714548098262799e-05,
+      "loss": 1.0887,
+      "step": 243150
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.714433305272824e-05,
+      "loss": 1.053,
+      "step": 243160
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.71431851228285e-05,
+      "loss": 0.98,
+      "step": 243170
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.714203719292875e-05,
+      "loss": 0.9529,
+      "step": 243180
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.7140889263028997e-05,
+      "loss": 0.946,
+      "step": 243190
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.713974133312925e-05,
+      "loss": 1.0603,
+      "step": 243200
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.713859340322951e-05,
+      "loss": 1.064,
+      "step": 243210
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.713744547332975e-05,
+      "loss": 1.0723,
+      "step": 243220
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.713629754343001e-05,
+      "loss": 1.0147,
+      "step": 243230
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.7135149613530266e-05,
+      "loss": 1.0068,
+      "step": 243240
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.713400168363051e-05,
+      "loss": 1.0384,
+      "step": 243250
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.7132853753730766e-05,
+      "loss": 1.122,
+      "step": 243260
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.713170582383102e-05,
+      "loss": 0.9925,
+      "step": 243270
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.713055789393127e-05,
+      "loss": 1.053,
+      "step": 243280
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.712940996403152e-05,
+      "loss": 1.0896,
+      "step": 243290
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.712826203413178e-05,
+      "loss": 1.0993,
+      "step": 243300
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.712711410423203e-05,
+      "loss": 0.961,
+      "step": 243310
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.7125966174332285e-05,
+      "loss": 0.9742,
+      "step": 243320
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.7124818244432535e-05,
+      "loss": 0.9536,
+      "step": 243330
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.7123670314532785e-05,
+      "loss": 0.978,
+      "step": 243340
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.712252238463304e-05,
+      "loss": 0.9785,
+      "step": 243350
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.71213744547333e-05,
+      "loss": 1.0548,
+      "step": 243360
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.712022652483354e-05,
+      "loss": 1.0777,
+      "step": 243370
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.71190785949338e-05,
+      "loss": 0.9829,
+      "step": 243380
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.7117930665034055e-05,
+      "loss": 1.0925,
+      "step": 243390
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.71167827351343e-05,
+      "loss": 0.9743,
+      "step": 243400
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.7115634805234554e-05,
+      "loss": 1.0736,
+      "step": 243410
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.711448687533481e-05,
+      "loss": 1.0822,
+      "step": 243420
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.711333894543506e-05,
+      "loss": 1.0819,
+      "step": 243430
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.711219101553531e-05,
+      "loss": 1.0767,
+      "step": 243440
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.711104308563557e-05,
+      "loss": 1.0944,
+      "step": 243450
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.710989515573582e-05,
+      "loss": 0.8651,
+      "step": 243460
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.7108747225836074e-05,
+      "loss": 1.025,
+      "step": 243470
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.7107599295936324e-05,
+      "loss": 1.0679,
+      "step": 243480
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.7106451366036574e-05,
+      "loss": 1.1729,
+      "step": 243490
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.710530343613683e-05,
+      "loss": 1.01,
+      "step": 243500
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.710415550623709e-05,
+      "loss": 1.0077,
+      "step": 243510
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.710300757633733e-05,
+      "loss": 1.0399,
+      "step": 243520
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.7101859646437587e-05,
+      "loss": 0.95,
+      "step": 243530
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.710071171653784e-05,
+      "loss": 1.1256,
+      "step": 243540
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.7099563786638086e-05,
+      "loss": 1.0113,
+      "step": 243550
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.709841585673834e-05,
+      "loss": 1.0407,
+      "step": 243560
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.70972679268386e-05,
+      "loss": 0.99,
+      "step": 243570
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.709611999693885e-05,
+      "loss": 0.984,
+      "step": 243580
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.70949720670391e-05,
+      "loss": 0.9953,
+      "step": 243590
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.7093824137139356e-05,
+      "loss": 0.9579,
+      "step": 243600
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.7092676207239606e-05,
+      "loss": 1.0197,
+      "step": 243610
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.709152827733986e-05,
+      "loss": 1.1346,
+      "step": 243620
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.709038034744011e-05,
+      "loss": 0.9779,
+      "step": 243630
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.708923241754036e-05,
+      "loss": 0.9418,
+      "step": 243640
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.708808448764062e-05,
+      "loss": 0.9723,
+      "step": 243650
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.7086936557740875e-05,
+      "loss": 1.0353,
+      "step": 243660
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.708578862784112e-05,
+      "loss": 0.9429,
+      "step": 243670
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.7084640697941375e-05,
+      "loss": 0.9932,
+      "step": 243680
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.708349276804163e-05,
+      "loss": 1.0833,
+      "step": 243690
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.7082344838141875e-05,
+      "loss": 1.0841,
+      "step": 243700
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.708119690824213e-05,
+      "loss": 1.0171,
+      "step": 243710
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.708004897834239e-05,
+      "loss": 0.9599,
+      "step": 243720
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.707890104844264e-05,
+      "loss": 1.0041,
+      "step": 243730
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.707775311854289e-05,
+      "loss": 1.0339,
+      "step": 243740
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.7076605188643144e-05,
+      "loss": 1.0082,
+      "step": 243750
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.7075457258743394e-05,
+      "loss": 1.0222,
+      "step": 243760
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.707430932884365e-05,
+      "loss": 0.9727,
+      "step": 243770
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.70731613989439e-05,
+      "loss": 0.9477,
+      "step": 243780
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.707201346904415e-05,
+      "loss": 1.0148,
+      "step": 243790
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.707086553914441e-05,
+      "loss": 0.9883,
+      "step": 243800
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.7069717609244664e-05,
+      "loss": 1.0137,
+      "step": 243810
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.706856967934491e-05,
+      "loss": 1.0215,
+      "step": 243820
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.7067421749445163e-05,
+      "loss": 1.0705,
+      "step": 243830
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.706627381954542e-05,
+      "loss": 1.1266,
+      "step": 243840
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.706512588964566e-05,
+      "loss": 0.9945,
+      "step": 243850
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.706397795974592e-05,
+      "loss": 0.9588,
+      "step": 243860
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.7062830029846176e-05,
+      "loss": 1.023,
+      "step": 243870
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.7061682099946426e-05,
+      "loss": 0.9649,
+      "step": 243880
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.7060534170046676e-05,
+      "loss": 1.0832,
+      "step": 243890
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.705938624014693e-05,
+      "loss": 0.9415,
+      "step": 243900
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.705823831024718e-05,
+      "loss": 1.0145,
+      "step": 243910
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.705709038034744e-05,
+      "loss": 1.0364,
+      "step": 243920
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.705594245044769e-05,
+      "loss": 1.0356,
+      "step": 243930
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.705479452054794e-05,
+      "loss": 1.1103,
+      "step": 243940
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.7053646590648196e-05,
+      "loss": 1.0482,
+      "step": 243950
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.705249866074845e-05,
+      "loss": 0.9416,
+      "step": 243960
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.7051350730848695e-05,
+      "loss": 1.0472,
+      "step": 243970
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.705020280094895e-05,
+      "loss": 0.951,
+      "step": 243980
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.704905487104921e-05,
+      "loss": 1.0893,
+      "step": 243990
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.704790694114945e-05,
+      "loss": 1.0789,
+      "step": 244000
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.704675901124971e-05,
+      "loss": 0.9856,
+      "step": 244010
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.7045611081349965e-05,
+      "loss": 1.054,
+      "step": 244020
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.7044463151450215e-05,
+      "loss": 1.1167,
+      "step": 244030
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.7043315221550465e-05,
+      "loss": 1.0331,
+      "step": 244040
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.704216729165072e-05,
+      "loss": 1.0844,
+      "step": 244050
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.704101936175097e-05,
+      "loss": 0.9866,
+      "step": 244060
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.703987143185123e-05,
+      "loss": 1.0267,
+      "step": 244070
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.703872350195148e-05,
+      "loss": 1.0352,
+      "step": 244080
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.703757557205173e-05,
+      "loss": 0.9993,
+      "step": 244090
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.7036427642151984e-05,
+      "loss": 1.0125,
+      "step": 244100
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.703527971225224e-05,
+      "loss": 0.9616,
+      "step": 244110
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.7034131782352484e-05,
+      "loss": 1.1083,
+      "step": 244120
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.703298385245274e-05,
+      "loss": 0.9663,
+      "step": 244130
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.703183592255299e-05,
+      "loss": 1.0013,
+      "step": 244140
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.703068799265324e-05,
+      "loss": 1.0756,
+      "step": 244150
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.70295400627535e-05,
+      "loss": 1.0456,
+      "step": 244160
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.7028392132853747e-05,
+      "loss": 1.0452,
+      "step": 244170
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.7027244202954e-05,
+      "loss": 1.131,
+      "step": 244180
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.702609627305425e-05,
+      "loss": 1.1332,
+      "step": 244190
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.70249483431545e-05,
+      "loss": 0.952,
+      "step": 244200
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.702380041325476e-05,
+      "loss": 1.0102,
+      "step": 244210
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.7022652483355016e-05,
+      "loss": 0.9691,
+      "step": 244220
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.702150455345526e-05,
+      "loss": 0.9932,
+      "step": 244230
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.7020356623555516e-05,
+      "loss": 1.1955,
+      "step": 244240
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.701920869365577e-05,
+      "loss": 1.0157,
+      "step": 244250
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.7018060763756016e-05,
+      "loss": 1.0553,
+      "step": 244260
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.701691283385627e-05,
+      "loss": 1.1086,
+      "step": 244270
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.701576490395653e-05,
+      "loss": 0.9799,
+      "step": 244280
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.701461697405678e-05,
+      "loss": 1.0479,
+      "step": 244290
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.701346904415703e-05,
+      "loss": 1.1037,
+      "step": 244300
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.7012321114257285e-05,
+      "loss": 1.0768,
+      "step": 244310
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.7011173184357535e-05,
+      "loss": 1.0749,
+      "step": 244320
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.701002525445779e-05,
+      "loss": 1.1396,
+      "step": 244330
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.700887732455804e-05,
+      "loss": 0.9836,
+      "step": 244340
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.700772939465829e-05,
+      "loss": 1.0341,
+      "step": 244350
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.700658146475855e-05,
+      "loss": 1.0157,
+      "step": 244360
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.7005433534858805e-05,
+      "loss": 1.0327,
+      "step": 244370
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.700428560495905e-05,
+      "loss": 0.9732,
+      "step": 244380
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.7003137675059304e-05,
+      "loss": 0.9866,
+      "step": 244390
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.700198974515956e-05,
+      "loss": 1.0218,
+      "step": 244400
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.7000841815259804e-05,
+      "loss": 1.041,
+      "step": 244410
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.699969388536006e-05,
+      "loss": 0.9349,
+      "step": 244420
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.699854595546032e-05,
+      "loss": 0.9793,
+      "step": 244430
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.699739802556057e-05,
+      "loss": 1.0171,
+      "step": 244440
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.699625009566082e-05,
+      "loss": 1.1317,
+      "step": 244450
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.6995102165761074e-05,
+      "loss": 1.0934,
+      "step": 244460
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.6993954235861324e-05,
+      "loss": 1.066,
+      "step": 244470
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.699280630596158e-05,
+      "loss": 1.1246,
+      "step": 244480
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.699165837606183e-05,
+      "loss": 0.9666,
+      "step": 244490
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.699051044616208e-05,
+      "loss": 0.9526,
+      "step": 244500
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.6989362516262336e-05,
+      "loss": 1.0153,
+      "step": 244510
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.698821458636259e-05,
+      "loss": 1.0421,
+      "step": 244520
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.6987066656462836e-05,
+      "loss": 1.0041,
+      "step": 244530
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.698591872656309e-05,
+      "loss": 0.9902,
+      "step": 244540
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.698477079666335e-05,
+      "loss": 1.1089,
+      "step": 244550
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.698362286676359e-05,
+      "loss": 0.9967,
+      "step": 244560
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.698247493686385e-05,
+      "loss": 0.9911,
+      "step": 244570
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.6981327006964106e-05,
+      "loss": 1.1075,
+      "step": 244580
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.6980179077064356e-05,
+      "loss": 0.9203,
+      "step": 244590
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.6979031147164606e-05,
+      "loss": 1.1182,
+      "step": 244600
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.697788321726486e-05,
+      "loss": 1.0336,
+      "step": 244610
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.697673528736511e-05,
+      "loss": 1.087,
+      "step": 244620
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.697558735746537e-05,
+      "loss": 1.0448,
+      "step": 244630
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.697443942756562e-05,
+      "loss": 0.9835,
+      "step": 244640
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.697329149766587e-05,
+      "loss": 1.1416,
+      "step": 244650
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.6972143567766125e-05,
+      "loss": 0.9673,
+      "step": 244660
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.697099563786638e-05,
+      "loss": 0.9285,
+      "step": 244670
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.6969847707966625e-05,
+      "loss": 1.0068,
+      "step": 244680
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.696869977806688e-05,
+      "loss": 1.0323,
+      "step": 244690
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.696755184816714e-05,
+      "loss": 1.133,
+      "step": 244700
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.696640391826738e-05,
+      "loss": 0.9647,
+      "step": 244710
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.696525598836764e-05,
+      "loss": 1.0419,
+      "step": 244720
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.6964108058467894e-05,
+      "loss": 0.9549,
+      "step": 244730
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.6962960128568144e-05,
+      "loss": 1.11,
+      "step": 244740
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.6961812198668394e-05,
+      "loss": 1.0069,
+      "step": 244750
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.696066426876865e-05,
+      "loss": 0.9491,
+      "step": 244760
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.69595163388689e-05,
+      "loss": 0.99,
+      "step": 244770
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.695836840896916e-05,
+      "loss": 0.9698,
+      "step": 244780
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.695722047906941e-05,
+      "loss": 1.0634,
+      "step": 244790
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.695607254916966e-05,
+      "loss": 1.055,
+      "step": 244800
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.6954924619269913e-05,
+      "loss": 1.1043,
+      "step": 244810
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.695377668937017e-05,
+      "loss": 1.0357,
+      "step": 244820
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.695262875947041e-05,
+      "loss": 1.026,
+      "step": 244830
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.695148082957067e-05,
+      "loss": 0.9301,
+      "step": 244840
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.6950332899670926e-05,
+      "loss": 1.1556,
+      "step": 244850
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.694918496977117e-05,
+      "loss": 0.9044,
+      "step": 244860
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.6948037039871426e-05,
+      "loss": 0.9588,
+      "step": 244870
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.694688910997168e-05,
+      "loss": 1.0233,
+      "step": 244880
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.694574118007193e-05,
+      "loss": 1.0118,
+      "step": 244890
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.694459325017218e-05,
+      "loss": 1.0215,
+      "step": 244900
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.694344532027244e-05,
+      "loss": 1.0264,
+      "step": 244910
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.694229739037269e-05,
+      "loss": 1.0408,
+      "step": 244920
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.6941149460472946e-05,
+      "loss": 1.0389,
+      "step": 244930
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.6940001530573195e-05,
+      "loss": 0.9414,
+      "step": 244940
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.6938853600673445e-05,
+      "loss": 1.0335,
+      "step": 244950
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.69377056707737e-05,
+      "loss": 1.0172,
+      "step": 244960
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.693655774087396e-05,
+      "loss": 1.0718,
+      "step": 244970
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.69354098109742e-05,
+      "loss": 1.0377,
+      "step": 244980
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.693426188107446e-05,
+      "loss": 1.0753,
+      "step": 244990
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.6933113951174715e-05,
+      "loss": 0.9268,
+      "step": 245000
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.693196602127496e-05,
+      "loss": 0.9426,
+      "step": 245010
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.6930818091375215e-05,
+      "loss": 1.0642,
+      "step": 245020
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.692967016147547e-05,
+      "loss": 1.0344,
+      "step": 245030
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.692852223157572e-05,
+      "loss": 0.9098,
+      "step": 245040
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.692737430167597e-05,
+      "loss": 1.0674,
+      "step": 245050
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.692622637177623e-05,
+      "loss": 0.9705,
+      "step": 245060
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.692507844187648e-05,
+      "loss": 1.0097,
+      "step": 245070
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.6923930511976734e-05,
+      "loss": 0.9759,
+      "step": 245080
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.6922782582076984e-05,
+      "loss": 1.0009,
+      "step": 245090
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.6921634652177234e-05,
+      "loss": 1.0155,
+      "step": 245100
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.692048672227749e-05,
+      "loss": 0.9973,
+      "step": 245110
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.691933879237775e-05,
+      "loss": 1.0537,
+      "step": 245120
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.691819086247799e-05,
+      "loss": 1.0965,
+      "step": 245130
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.691704293257825e-05,
+      "loss": 1.0298,
+      "step": 245140
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.69158950026785e-05,
+      "loss": 1.0072,
+      "step": 245150
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.6914747072778746e-05,
+      "loss": 1.0288,
+      "step": 245160
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.6913599142879e-05,
+      "loss": 0.9988,
+      "step": 245170
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.691245121297926e-05,
+      "loss": 1.0643,
+      "step": 245180
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.691130328307951e-05,
+      "loss": 0.9949,
+      "step": 245190
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.691015535317976e-05,
+      "loss": 1.033,
+      "step": 245200
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.6909007423280016e-05,
+      "loss": 1.0316,
+      "step": 245210
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.6907859493380266e-05,
+      "loss": 1.0886,
+      "step": 245220
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.690671156348052e-05,
+      "loss": 1.0784,
+      "step": 245230
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.690556363358077e-05,
+      "loss": 1.0351,
+      "step": 245240
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.690441570368102e-05,
+      "loss": 1.0457,
+      "step": 245250
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.690326777378128e-05,
+      "loss": 1.051,
+      "step": 245260
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.6902119843881535e-05,
+      "loss": 1.0119,
+      "step": 245270
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.690097191398178e-05,
+      "loss": 0.9737,
+      "step": 245280
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.6899823984082035e-05,
+      "loss": 1.0871,
+      "step": 245290
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.689867605418229e-05,
+      "loss": 1.001,
+      "step": 245300
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.6897528124282535e-05,
+      "loss": 1.088,
+      "step": 245310
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.689638019438279e-05,
+      "loss": 1.0327,
+      "step": 245320
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.689523226448305e-05,
+      "loss": 1.146,
+      "step": 245330
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.68940843345833e-05,
+      "loss": 0.9005,
+      "step": 245340
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.689293640468355e-05,
+      "loss": 1.0905,
+      "step": 245350
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.6891788474783805e-05,
+      "loss": 0.9759,
+      "step": 245360
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.6890640544884054e-05,
+      "loss": 1.0452,
+      "step": 245370
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.688949261498431e-05,
+      "loss": 0.9825,
+      "step": 245380
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.688834468508456e-05,
+      "loss": 1.108,
+      "step": 245390
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.688719675518481e-05,
+      "loss": 0.9941,
+      "step": 245400
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.688604882528507e-05,
+      "loss": 1.0035,
+      "step": 245410
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.6884900895385324e-05,
+      "loss": 0.9488,
+      "step": 245420
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.688375296548557e-05,
+      "loss": 0.9723,
+      "step": 245430
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.6882605035585824e-05,
+      "loss": 1.0181,
+      "step": 245440
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.688145710568608e-05,
+      "loss": 1.1576,
+      "step": 245450
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.688030917578632e-05,
+      "loss": 1.0269,
+      "step": 245460
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.687916124588658e-05,
+      "loss": 1.0254,
+      "step": 245470
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.687801331598684e-05,
+      "loss": 1.0461,
+      "step": 245480
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.6876865386087086e-05,
+      "loss": 0.9474,
+      "step": 245490
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.6875717456187336e-05,
+      "loss": 1.0353,
+      "step": 245500
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.6874569526287586e-05,
+      "loss": 1.0325,
+      "step": 245510
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.687342159638784e-05,
+      "loss": 1.0542,
+      "step": 245520
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.68722736664881e-05,
+      "loss": 1.0645,
+      "step": 245530
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.687112573658834e-05,
+      "loss": 1.0158,
+      "step": 245540
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.68699778066886e-05,
+      "loss": 0.9515,
+      "step": 245550
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.6868829876788856e-05,
+      "loss": 1.0979,
+      "step": 245560
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.68676819468891e-05,
+      "loss": 0.9403,
+      "step": 245570
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.6866534016989356e-05,
+      "loss": 0.997,
+      "step": 245580
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.686538608708961e-05,
+      "loss": 1.0999,
+      "step": 245590
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.686423815718986e-05,
+      "loss": 1.1227,
+      "step": 245600
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.686309022729011e-05,
+      "loss": 1.0611,
+      "step": 245610
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.686194229739037e-05,
+      "loss": 1.1052,
+      "step": 245620
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.686079436749062e-05,
+      "loss": 1.0056,
+      "step": 245630
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.6859646437590875e-05,
+      "loss": 0.9423,
+      "step": 245640
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.6858498507691125e-05,
+      "loss": 1.0022,
+      "step": 245650
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.6857350577791375e-05,
+      "loss": 1.0437,
+      "step": 245660
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.685620264789163e-05,
+      "loss": 0.956,
+      "step": 245670
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.685505471799189e-05,
+      "loss": 0.9414,
+      "step": 245680
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.685390678809213e-05,
+      "loss": 1.0253,
+      "step": 245690
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.685275885819239e-05,
+      "loss": 1.1335,
+      "step": 245700
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.6851610928292644e-05,
+      "loss": 1.0072,
+      "step": 245710
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.685046299839289e-05,
+      "loss": 0.9178,
+      "step": 245720
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.6849315068493144e-05,
+      "loss": 1.0887,
+      "step": 245730
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.68481671385934e-05,
+      "loss": 1.0451,
+      "step": 245740
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.684701920869365e-05,
+      "loss": 0.9636,
+      "step": 245750
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.68458712787939e-05,
+      "loss": 1.1043,
+      "step": 245760
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.684472334889416e-05,
+      "loss": 1.0675,
+      "step": 245770
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.684357541899441e-05,
+      "loss": 1.0433,
+      "step": 245780
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.6842427489094663e-05,
+      "loss": 1.0356,
+      "step": 245790
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.684127955919491e-05,
+      "loss": 1.0523,
+      "step": 245800
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.684013162929516e-05,
+      "loss": 1.0265,
+      "step": 245810
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.683898369939542e-05,
+      "loss": 0.9137,
+      "step": 245820
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.6837835769495676e-05,
+      "loss": 1.0644,
+      "step": 245830
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.683668783959592e-05,
+      "loss": 1.0254,
+      "step": 245840
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.6835539909696176e-05,
+      "loss": 1.0235,
+      "step": 245850
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.683439197979643e-05,
+      "loss": 1.0457,
+      "step": 245860
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.6833244049896676e-05,
+      "loss": 1.0359,
+      "step": 245870
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.683209611999693e-05,
+      "loss": 1.0607,
+      "step": 245880
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.683094819009719e-05,
+      "loss": 0.9804,
+      "step": 245890
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.682980026019744e-05,
+      "loss": 1.0512,
+      "step": 245900
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.682865233029769e-05,
+      "loss": 0.9854,
+      "step": 245910
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.6827504400397945e-05,
+      "loss": 1.0041,
+      "step": 245920
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.6826356470498195e-05,
+      "loss": 0.9676,
+      "step": 245930
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.682520854059845e-05,
+      "loss": 1.128,
+      "step": 245940
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.68240606106987e-05,
+      "loss": 1.084,
+      "step": 245950
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.682291268079895e-05,
+      "loss": 1.0479,
+      "step": 245960
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.682176475089921e-05,
+      "loss": 1.0156,
+      "step": 245970
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.6820616820999465e-05,
+      "loss": 1.0121,
+      "step": 245980
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.681946889109971e-05,
+      "loss": 1.0256,
+      "step": 245990
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.6818320961199965e-05,
+      "loss": 1.0781,
+      "step": 246000
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.681717303130022e-05,
+      "loss": 1.0858,
+      "step": 246010
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.6816025101400464e-05,
+      "loss": 0.9478,
+      "step": 246020
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.681487717150072e-05,
+      "loss": 1.0195,
+      "step": 246030
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.681372924160098e-05,
+      "loss": 1.0025,
+      "step": 246040
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.681258131170123e-05,
+      "loss": 1.1525,
+      "step": 246050
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.681143338180148e-05,
+      "loss": 1.0489,
+      "step": 246060
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.6810285451901734e-05,
+      "loss": 1.135,
+      "step": 246070
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.6809137522001984e-05,
+      "loss": 1.049,
+      "step": 246080
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.680798959210224e-05,
+      "loss": 1.1145,
+      "step": 246090
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.680684166220249e-05,
+      "loss": 1.0326,
+      "step": 246100
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.680569373230274e-05,
+      "loss": 1.0176,
+      "step": 246110
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.6804545802403e-05,
+      "loss": 1.0135,
+      "step": 246120
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.680339787250325e-05,
+      "loss": 1.0462,
+      "step": 246130
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.6802249942603496e-05,
+      "loss": 1.0247,
+      "step": 246140
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.680110201270375e-05,
+      "loss": 0.9858,
+      "step": 246150
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.679995408280401e-05,
+      "loss": 1.1368,
+      "step": 246160
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.679880615290425e-05,
+      "loss": 1.0394,
+      "step": 246170
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.679765822300451e-05,
+      "loss": 1.0684,
+      "step": 246180
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.6796510293104766e-05,
+      "loss": 1.0587,
+      "step": 246190
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.6795362363205016e-05,
+      "loss": 1.0556,
+      "step": 246200
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.6794214433305266e-05,
+      "loss": 1.1157,
+      "step": 246210
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.679306650340552e-05,
+      "loss": 1.2879,
+      "step": 246220
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.679191857350577e-05,
+      "loss": 1.229,
+      "step": 246230
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.679077064360603e-05,
+      "loss": 0.9758,
+      "step": 246240
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.678962271370628e-05,
+      "loss": 1.0717,
+      "step": 246250
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.678847478380653e-05,
+      "loss": 1.0752,
+      "step": 246260
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.6787326853906785e-05,
+      "loss": 1.0717,
+      "step": 246270
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.678617892400704e-05,
+      "loss": 1.0155,
+      "step": 246280
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.6785030994107285e-05,
+      "loss": 1.0389,
+      "step": 246290
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.678388306420754e-05,
+      "loss": 1.0619,
+      "step": 246300
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.67827351343078e-05,
+      "loss": 1.0547,
+      "step": 246310
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.678158720440804e-05,
+      "loss": 0.9816,
+      "step": 246320
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.67804392745083e-05,
+      "loss": 0.9722,
+      "step": 246330
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.6779291344608555e-05,
+      "loss": 1.0353,
+      "step": 246340
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.6778143414708804e-05,
+      "loss": 1.0583,
+      "step": 246350
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.6776995484809054e-05,
+      "loss": 1.0807,
+      "step": 246360
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.677584755490931e-05,
+      "loss": 0.9451,
+      "step": 246370
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.677469962500956e-05,
+      "loss": 1.0303,
+      "step": 246380
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.677355169510982e-05,
+      "loss": 0.9519,
+      "step": 246390
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.677240376521007e-05,
+      "loss": 1.0831,
+      "step": 246400
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.677125583531032e-05,
+      "loss": 1.0141,
+      "step": 246410
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.6770107905410574e-05,
+      "loss": 0.9593,
+      "step": 246420
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.676895997551083e-05,
+      "loss": 1.0184,
+      "step": 246430
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.676781204561107e-05,
+      "loss": 1.004,
+      "step": 246440
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.676666411571133e-05,
+      "loss": 0.9584,
+      "step": 246450
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.676551618581159e-05,
+      "loss": 1.0528,
+      "step": 246460
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.676436825591183e-05,
+      "loss": 1.0195,
+      "step": 246470
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.6763220326012086e-05,
+      "loss": 1.1349,
+      "step": 246480
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.676207239611234e-05,
+      "loss": 1.0786,
+      "step": 246490
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.676092446621259e-05,
+      "loss": 0.9686,
+      "step": 246500
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.675977653631284e-05,
+      "loss": 1.0513,
+      "step": 246510
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.67586286064131e-05,
+      "loss": 1.0505,
+      "step": 246520
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.675748067651335e-05,
+      "loss": 1.0714,
+      "step": 246530
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.6756332746613606e-05,
+      "loss": 1.0472,
+      "step": 246540
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.6755184816713856e-05,
+      "loss": 0.9801,
+      "step": 246550
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.6754036886814106e-05,
+      "loss": 1.0463,
+      "step": 246560
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.675288895691436e-05,
+      "loss": 1.0399,
+      "step": 246570
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.675174102701462e-05,
+      "loss": 1.1726,
+      "step": 246580
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.675059309711486e-05,
+      "loss": 1.0269,
+      "step": 246590
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.674944516721512e-05,
+      "loss": 0.9637,
+      "step": 246600
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.6748297237315375e-05,
+      "loss": 0.9978,
+      "step": 246610
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.674714930741562e-05,
+      "loss": 1.0202,
+      "step": 246620
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.6746001377515875e-05,
+      "loss": 0.944,
+      "step": 246630
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.674485344761613e-05,
+      "loss": 1.0098,
+      "step": 246640
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.674370551771638e-05,
+      "loss": 1.0342,
+      "step": 246650
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.674255758781663e-05,
+      "loss": 0.9773,
+      "step": 246660
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.674140965791689e-05,
+      "loss": 1.0061,
+      "step": 246670
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.674026172801714e-05,
+      "loss": 1.0511,
+      "step": 246680
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.6739113798117394e-05,
+      "loss": 0.9398,
+      "step": 246690
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.6737965868217644e-05,
+      "loss": 1.0164,
+      "step": 246700
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.6736817938317894e-05,
+      "loss": 1.0238,
+      "step": 246710
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.673567000841815e-05,
+      "loss": 1.0148,
+      "step": 246720
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.673452207851841e-05,
+      "loss": 1.119,
+      "step": 246730
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.673337414861865e-05,
+      "loss": 1.0869,
+      "step": 246740
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.673222621871891e-05,
+      "loss": 1.0763,
+      "step": 246750
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.6731078288819164e-05,
+      "loss": 0.9814,
+      "step": 246760
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.672993035891941e-05,
+      "loss": 1.054,
+      "step": 246770
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.672878242901966e-05,
+      "loss": 1.0185,
+      "step": 246780
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.672763449911992e-05,
+      "loss": 1.0264,
+      "step": 246790
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.672648656922017e-05,
+      "loss": 1.0165,
+      "step": 246800
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.672533863932042e-05,
+      "loss": 0.9795,
+      "step": 246810
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.6724190709420676e-05,
+      "loss": 0.991,
+      "step": 246820
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.6723042779520926e-05,
+      "loss": 0.8847,
+      "step": 246830
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.672189484962118e-05,
+      "loss": 1.0715,
+      "step": 246840
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.672074691972143e-05,
+      "loss": 1.0803,
+      "step": 246850
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.671959898982168e-05,
+      "loss": 1.0321,
+      "step": 246860
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.671845105992194e-05,
+      "loss": 1.1779,
+      "step": 246870
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.671730313002218e-05,
+      "loss": 0.968,
+      "step": 246880
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.671615520012244e-05,
+      "loss": 0.9515,
+      "step": 246890
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.6715007270222695e-05,
+      "loss": 1.0753,
+      "step": 246900
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.6713859340322945e-05,
+      "loss": 1.0715,
+      "step": 246910
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.6712711410423195e-05,
+      "loss": 1.1417,
+      "step": 246920
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.671156348052345e-05,
+      "loss": 1.0895,
+      "step": 246930
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.67104155506237e-05,
+      "loss": 1.0967,
+      "step": 246940
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.670926762072396e-05,
+      "loss": 0.9528,
+      "step": 246950
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.670811969082421e-05,
+      "loss": 1.0314,
+      "step": 246960
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.670697176092446e-05,
+      "loss": 1.0544,
+      "step": 246970
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.6705823831024715e-05,
+      "loss": 0.9892,
+      "step": 246980
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.670467590112497e-05,
+      "loss": 1.0933,
+      "step": 246990
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.6703527971225214e-05,
+      "loss": 1.112,
+      "step": 247000
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.670238004132547e-05,
+      "loss": 1.0277,
+      "step": 247010
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.670123211142573e-05,
+      "loss": 1.0201,
+      "step": 247020
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.670008418152597e-05,
+      "loss": 1.0001,
+      "step": 247030
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.669893625162623e-05,
+      "loss": 0.9542,
+      "step": 247040
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.6697788321726484e-05,
+      "loss": 1.0376,
+      "step": 247050
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.6696640391826734e-05,
+      "loss": 0.9516,
+      "step": 247060
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.6695492461926984e-05,
+      "loss": 1.149,
+      "step": 247070
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.669434453202724e-05,
+      "loss": 1.0962,
+      "step": 247080
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.669319660212749e-05,
+      "loss": 1.0259,
+      "step": 247090
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.669204867222775e-05,
+      "loss": 1.089,
+      "step": 247100
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.6690900742327997e-05,
+      "loss": 1.1097,
+      "step": 247110
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.6689752812428246e-05,
+      "loss": 1.1006,
+      "step": 247120
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.66886048825285e-05,
+      "loss": 1.019,
+      "step": 247130
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.668745695262876e-05,
+      "loss": 1.1126,
+      "step": 247140
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.6686309022729e-05,
+      "loss": 1.0471,
+      "step": 247150
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.668516109282926e-05,
+      "loss": 0.9603,
+      "step": 247160
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.6684013162929516e-05,
+      "loss": 1.1016,
+      "step": 247170
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.668286523302976e-05,
+      "loss": 1.0011,
+      "step": 247180
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.6681717303130016e-05,
+      "loss": 1.0842,
+      "step": 247190
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.668056937323027e-05,
+      "loss": 1.0656,
+      "step": 247200
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.667942144333052e-05,
+      "loss": 1.0894,
+      "step": 247210
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.667827351343077e-05,
+      "loss": 0.941,
+      "step": 247220
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.667712558353103e-05,
+      "loss": 1.0982,
+      "step": 247230
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.667597765363128e-05,
+      "loss": 1.1169,
+      "step": 247240
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.6674829723731535e-05,
+      "loss": 1.0697,
+      "step": 247250
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.6673681793831785e-05,
+      "loss": 0.9389,
+      "step": 247260
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.6672533863932035e-05,
+      "loss": 1.0317,
+      "step": 247270
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.667138593403229e-05,
+      "loss": 0.9867,
+      "step": 247280
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.667023800413255e-05,
+      "loss": 1.0566,
+      "step": 247290
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.666909007423279e-05,
+      "loss": 0.985,
+      "step": 247300
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.666794214433305e-05,
+      "loss": 1.0842,
+      "step": 247310
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.6666794214433304e-05,
+      "loss": 1.009,
+      "step": 247320
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.666564628453355e-05,
+      "loss": 1.0647,
+      "step": 247330
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.6664498354633804e-05,
+      "loss": 1.0709,
+      "step": 247340
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.666335042473406e-05,
+      "loss": 1.0147,
+      "step": 247350
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.666220249483431e-05,
+      "loss": 0.9211,
+      "step": 247360
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.666105456493456e-05,
+      "loss": 1.0579,
+      "step": 247370
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.665990663503482e-05,
+      "loss": 1.0654,
+      "step": 247380
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.665875870513507e-05,
+      "loss": 1.1536,
+      "step": 247390
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.6657610775235324e-05,
+      "loss": 1.0576,
+      "step": 247400
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.6656462845335574e-05,
+      "loss": 1.0443,
+      "step": 247410
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.665531491543582e-05,
+      "loss": 0.9841,
+      "step": 247420
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.665416698553608e-05,
+      "loss": 1.0391,
+      "step": 247430
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.665301905563634e-05,
+      "loss": 1.0219,
+      "step": 247440
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.665187112573658e-05,
+      "loss": 1.0072,
+      "step": 247450
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.6650723195836836e-05,
+      "loss": 1.0717,
+      "step": 247460
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.664957526593709e-05,
+      "loss": 1.0697,
+      "step": 247470
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.6648427336037336e-05,
+      "loss": 1.0534,
+      "step": 247480
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.664727940613759e-05,
+      "loss": 1.2108,
+      "step": 247490
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.664613147623785e-05,
+      "loss": 1.1149,
+      "step": 247500
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.66449835463381e-05,
+      "loss": 1.0132,
+      "step": 247510
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.664383561643835e-05,
+      "loss": 0.9429,
+      "step": 247520
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.6642687686538606e-05,
+      "loss": 1.0933,
+      "step": 247530
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.6641539756638855e-05,
+      "loss": 1.0461,
+      "step": 247540
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.664039182673911e-05,
+      "loss": 1.0593,
+      "step": 247550
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.663924389683936e-05,
+      "loss": 1.1138,
+      "step": 247560
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.663809596693961e-05,
+      "loss": 1.0671,
+      "step": 247570
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.663694803703987e-05,
+      "loss": 1.0736,
+      "step": 247580
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.6635800107140125e-05,
+      "loss": 1.0092,
+      "step": 247590
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.663465217724037e-05,
+      "loss": 1.0079,
+      "step": 247600
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.6633504247340625e-05,
+      "loss": 1.093,
+      "step": 247610
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.663235631744088e-05,
+      "loss": 1.0473,
+      "step": 247620
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.6631208387541125e-05,
+      "loss": 1.0808,
+      "step": 247630
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.663006045764138e-05,
+      "loss": 1.0473,
+      "step": 247640
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.662891252774164e-05,
+      "loss": 0.9924,
+      "step": 247650
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.662776459784189e-05,
+      "loss": 0.912,
+      "step": 247660
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.662661666794214e-05,
+      "loss": 1.0077,
+      "step": 247670
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.6625468738042394e-05,
+      "loss": 1.0372,
+      "step": 247680
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.6624320808142644e-05,
+      "loss": 0.9782,
+      "step": 247690
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.66231728782429e-05,
+      "loss": 1.015,
+      "step": 247700
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.662202494834315e-05,
+      "loss": 0.9757,
+      "step": 247710
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.66208770184434e-05,
+      "loss": 0.9282,
+      "step": 247720
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.661972908854366e-05,
+      "loss": 1.0058,
+      "step": 247730
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.6618581158643914e-05,
+      "loss": 1.0315,
+      "step": 247740
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.661743322874416e-05,
+      "loss": 0.9287,
+      "step": 247750
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.661628529884441e-05,
+      "loss": 0.9614,
+      "step": 247760
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.661513736894467e-05,
+      "loss": 0.9435,
+      "step": 247770
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.661398943904491e-05,
+      "loss": 0.9941,
+      "step": 247780
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.661284150914517e-05,
+      "loss": 0.9526,
+      "step": 247790
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.6611693579245426e-05,
+      "loss": 1.0683,
+      "step": 247800
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.6610545649345676e-05,
+      "loss": 1.1296,
+      "step": 247810
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.6609397719445926e-05,
+      "loss": 0.9965,
+      "step": 247820
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.660824978954618e-05,
+      "loss": 1.1346,
+      "step": 247830
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.660710185964643e-05,
+      "loss": 1.1024,
+      "step": 247840
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.660595392974669e-05,
+      "loss": 1.0627,
+      "step": 247850
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.660480599984694e-05,
+      "loss": 1.0656,
+      "step": 247860
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.660365806994719e-05,
+      "loss": 1.0102,
+      "step": 247870
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.6602510140047445e-05,
+      "loss": 0.9909,
+      "step": 247880
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.66013622101477e-05,
+      "loss": 1.1812,
+      "step": 247890
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.6600214280247945e-05,
+      "loss": 1.0982,
+      "step": 247900
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.65990663503482e-05,
+      "loss": 1.1116,
+      "step": 247910
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.659791842044846e-05,
+      "loss": 1.0467,
+      "step": 247920
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.65967704905487e-05,
+      "loss": 1.1041,
+      "step": 247930
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.659562256064896e-05,
+      "loss": 1.1181,
+      "step": 247940
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.6594474630749215e-05,
+      "loss": 0.9832,
+      "step": 247950
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.6593326700849465e-05,
+      "loss": 1.053,
+      "step": 247960
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.6592178770949714e-05,
+      "loss": 0.9499,
+      "step": 247970
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.659103084104997e-05,
+      "loss": 0.9679,
+      "step": 247980
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.658988291115022e-05,
+      "loss": 1.0485,
+      "step": 247990
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.658873498125048e-05,
+      "loss": 0.973,
+      "step": 248000
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.658758705135073e-05,
+      "loss": 1.0114,
+      "step": 248010
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.658643912145098e-05,
+      "loss": 1.0293,
+      "step": 248020
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.6585291191551234e-05,
+      "loss": 1.0395,
+      "step": 248030
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.658414326165149e-05,
+      "loss": 1.1987,
+      "step": 248040
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.6582995331751734e-05,
+      "loss": 1.011,
+      "step": 248050
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.658184740185199e-05,
+      "loss": 0.9773,
+      "step": 248060
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.658069947195225e-05,
+      "loss": 0.8665,
+      "step": 248070
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.657955154205249e-05,
+      "loss": 1.0886,
+      "step": 248080
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.6578403612152747e-05,
+      "loss": 0.9478,
+      "step": 248090
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.6577255682253e-05,
+      "loss": 1.0678,
+      "step": 248100
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.657610775235325e-05,
+      "loss": 0.8841,
+      "step": 248110
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.65749598224535e-05,
+      "loss": 1.0483,
+      "step": 248120
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.657381189255376e-05,
+      "loss": 1.1707,
+      "step": 248130
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.657266396265401e-05,
+      "loss": 1.0451,
+      "step": 248140
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.6571516032754266e-05,
+      "loss": 0.9816,
+      "step": 248150
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.6570368102854516e-05,
+      "loss": 1.1019,
+      "step": 248160
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.6569220172954766e-05,
+      "loss": 1.0049,
+      "step": 248170
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.656807224305502e-05,
+      "loss": 1.0093,
+      "step": 248180
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.656692431315528e-05,
+      "loss": 1.0754,
+      "step": 248190
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.656577638325552e-05,
+      "loss": 1.0661,
+      "step": 248200
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.656462845335578e-05,
+      "loss": 1.1218,
+      "step": 248210
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.6563480523456035e-05,
+      "loss": 1.1446,
+      "step": 248220
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.656233259355628e-05,
+      "loss": 1.094,
+      "step": 248230
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.6561184663656535e-05,
+      "loss": 1.0016,
+      "step": 248240
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.656003673375679e-05,
+      "loss": 0.998,
+      "step": 248250
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.655888880385704e-05,
+      "loss": 1.1233,
+      "step": 248260
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.655774087395729e-05,
+      "loss": 1.0018,
+      "step": 248270
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.655659294405754e-05,
+      "loss": 0.9756,
+      "step": 248280
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.65554450141578e-05,
+      "loss": 1.0795,
+      "step": 248290
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.6554297084258054e-05,
+      "loss": 1.0673,
+      "step": 248300
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.65531491543583e-05,
+      "loss": 0.9641,
+      "step": 248310
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.6552001224458554e-05,
+      "loss": 1.0457,
+      "step": 248320
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.655085329455881e-05,
+      "loss": 0.9583,
+      "step": 248330
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.6549705364659054e-05,
+      "loss": 1.1514,
+      "step": 248340
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.654855743475931e-05,
+      "loss": 0.9962,
+      "step": 248350
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.654740950485957e-05,
+      "loss": 1.0388,
+      "step": 248360
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.654626157495982e-05,
+      "loss": 0.9604,
+      "step": 248370
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.654511364506007e-05,
+      "loss": 1.0085,
+      "step": 248380
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.6543965715160324e-05,
+      "loss": 0.9803,
+      "step": 248390
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.654281778526057e-05,
+      "loss": 1.1178,
+      "step": 248400
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.654166985536083e-05,
+      "loss": 1.1336,
+      "step": 248410
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.654052192546108e-05,
+      "loss": 1.0179,
+      "step": 248420
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.653937399556133e-05,
+      "loss": 1.0679,
+      "step": 248430
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.6538226065661586e-05,
+      "loss": 1.0112,
+      "step": 248440
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.653707813576184e-05,
+      "loss": 1.0703,
+      "step": 248450
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.6535930205862086e-05,
+      "loss": 1.0559,
+      "step": 248460
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.653478227596234e-05,
+      "loss": 1.0791,
+      "step": 248470
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.65336343460626e-05,
+      "loss": 1.0168,
+      "step": 248480
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.653248641616284e-05,
+      "loss": 1.1485,
+      "step": 248490
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.65313384862631e-05,
+      "loss": 0.9854,
+      "step": 248500
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.6530190556363356e-05,
+      "loss": 0.9631,
+      "step": 248510
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.6529042626463605e-05,
+      "loss": 1.0947,
+      "step": 248520
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.6527894696563855e-05,
+      "loss": 1.0281,
+      "step": 248530
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.652674676666411e-05,
+      "loss": 1.0813,
+      "step": 248540
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.652559883676436e-05,
+      "loss": 1.0204,
+      "step": 248550
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.652445090686462e-05,
+      "loss": 0.978,
+      "step": 248560
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.652330297696487e-05,
+      "loss": 1.0765,
+      "step": 248570
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.652215504706512e-05,
+      "loss": 1.1122,
+      "step": 248580
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.6521007117165375e-05,
+      "loss": 1.2206,
+      "step": 248590
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.651985918726563e-05,
+      "loss": 0.9909,
+      "step": 248600
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.6518711257365875e-05,
+      "loss": 1.0005,
+      "step": 248610
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.651756332746613e-05,
+      "loss": 1.0288,
+      "step": 248620
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.651641539756639e-05,
+      "loss": 1.2228,
+      "step": 248630
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.651526746766663e-05,
+      "loss": 0.9815,
+      "step": 248640
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.651411953776689e-05,
+      "loss": 0.9117,
+      "step": 248650
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.6512971607867144e-05,
+      "loss": 1.0262,
+      "step": 248660
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.6511823677967394e-05,
+      "loss": 0.9046,
+      "step": 248670
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.6510675748067644e-05,
+      "loss": 0.9863,
+      "step": 248680
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.65095278181679e-05,
+      "loss": 0.9439,
+      "step": 248690
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.650837988826815e-05,
+      "loss": 1.098,
+      "step": 248700
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.650723195836841e-05,
+      "loss": 1.0416,
+      "step": 248710
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.650608402846866e-05,
+      "loss": 1.0109,
+      "step": 248720
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.650493609856891e-05,
+      "loss": 1.0225,
+      "step": 248730
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.650378816866916e-05,
+      "loss": 1.0219,
+      "step": 248740
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.650264023876942e-05,
+      "loss": 1.1018,
+      "step": 248750
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.650149230886966e-05,
+      "loss": 1.0963,
+      "step": 248760
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.650034437896992e-05,
+      "loss": 1.0447,
+      "step": 248770
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.6499196449070176e-05,
+      "loss": 1.0384,
+      "step": 248780
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.649804851917042e-05,
+      "loss": 1.2046,
+      "step": 248790
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.6496900589270676e-05,
+      "loss": 1.0393,
+      "step": 248800
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.649575265937093e-05,
+      "loss": 0.9917,
+      "step": 248810
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.649460472947118e-05,
+      "loss": 1.0417,
+      "step": 248820
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.649345679957143e-05,
+      "loss": 1.0226,
+      "step": 248830
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.649230886967169e-05,
+      "loss": 0.9382,
+      "step": 248840
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.649116093977194e-05,
+      "loss": 0.9175,
+      "step": 248850
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.6490013009872195e-05,
+      "loss": 1.1186,
+      "step": 248860
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.6488865079972445e-05,
+      "loss": 0.9866,
+      "step": 248870
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.6487717150072695e-05,
+      "loss": 1.0202,
+      "step": 248880
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.648656922017295e-05,
+      "loss": 1.062,
+      "step": 248890
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.648542129027321e-05,
+      "loss": 1.1192,
+      "step": 248900
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.648427336037345e-05,
+      "loss": 1.0762,
+      "step": 248910
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.648312543047371e-05,
+      "loss": 1.0617,
+      "step": 248920
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.6481977500573965e-05,
+      "loss": 1.0999,
+      "step": 248930
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.648082957067421e-05,
+      "loss": 0.9532,
+      "step": 248940
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.6479681640774464e-05,
+      "loss": 1.0902,
+      "step": 248950
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.647853371087472e-05,
+      "loss": 1.1146,
+      "step": 248960
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.647738578097497e-05,
+      "loss": 1.0702,
+      "step": 248970
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.647623785107522e-05,
+      "loss": 1.0596,
+      "step": 248980
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.647508992117548e-05,
+      "loss": 0.9789,
+      "step": 248990
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.647394199127573e-05,
+      "loss": 0.9456,
+      "step": 249000
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.6472794061375984e-05,
+      "loss": 0.9607,
+      "step": 249010
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.6471646131476234e-05,
+      "loss": 1.0304,
+      "step": 249020
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.6470498201576484e-05,
+      "loss": 1.0495,
+      "step": 249030
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.646935027167674e-05,
+      "loss": 1.1171,
+      "step": 249040
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.6468202341777e-05,
+      "loss": 1.0329,
+      "step": 249050
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.646705441187724e-05,
+      "loss": 1.1309,
+      "step": 249060
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.6465906481977497e-05,
+      "loss": 0.9282,
+      "step": 249070
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.646475855207775e-05,
+      "loss": 1.1752,
+      "step": 249080
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.6463610622177996e-05,
+      "loss": 1.0883,
+      "step": 249090
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.646246269227825e-05,
+      "loss": 1.1669,
+      "step": 249100
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.646131476237851e-05,
+      "loss": 1.0526,
+      "step": 249110
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.646016683247876e-05,
+      "loss": 1.127,
+      "step": 249120
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.645901890257901e-05,
+      "loss": 1.0188,
+      "step": 249130
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.6457870972679266e-05,
+      "loss": 1.007,
+      "step": 249140
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.6456723042779516e-05,
+      "loss": 1.0619,
+      "step": 249150
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.645557511287977e-05,
+      "loss": 1.0856,
+      "step": 249160
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.645442718298002e-05,
+      "loss": 0.8782,
+      "step": 249170
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.645327925308027e-05,
+      "loss": 0.953,
+      "step": 249180
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.645213132318053e-05,
+      "loss": 1.0279,
+      "step": 249190
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.6450983393280785e-05,
+      "loss": 1.0325,
+      "step": 249200
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.644983546338103e-05,
+      "loss": 0.9681,
+      "step": 249210
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.6448687533481285e-05,
+      "loss": 1.0259,
+      "step": 249220
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.644753960358154e-05,
+      "loss": 0.9739,
+      "step": 249230
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.6446391673681785e-05,
+      "loss": 1.0253,
+      "step": 249240
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.644524374378204e-05,
+      "loss": 1.1522,
+      "step": 249250
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.64440958138823e-05,
+      "loss": 1.0007,
+      "step": 249260
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.644294788398255e-05,
+      "loss": 1.0454,
+      "step": 249270
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.64417999540828e-05,
+      "loss": 0.9628,
+      "step": 249280
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.6440652024183054e-05,
+      "loss": 1.1171,
+      "step": 249290
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.6439504094283304e-05,
+      "loss": 0.9634,
+      "step": 249300
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.643835616438356e-05,
+      "loss": 0.9853,
+      "step": 249310
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.643720823448381e-05,
+      "loss": 0.9422,
+      "step": 249320
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.643606030458406e-05,
+      "loss": 1.0878,
+      "step": 249330
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.643491237468432e-05,
+      "loss": 0.9061,
+      "step": 249340
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.6433764444784574e-05,
+      "loss": 1.0563,
+      "step": 249350
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.643261651488482e-05,
+      "loss": 1.0564,
+      "step": 249360
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.6431468584985074e-05,
+      "loss": 1.0094,
+      "step": 249370
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.643032065508533e-05,
+      "loss": 1.0579,
+      "step": 249380
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.642917272518557e-05,
+      "loss": 1.0678,
+      "step": 249390
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.642802479528583e-05,
+      "loss": 1.0656,
+      "step": 249400
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.6426876865386086e-05,
+      "loss": 0.9686,
+      "step": 249410
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.6425728935486336e-05,
+      "loss": 1.0203,
+      "step": 249420
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.6424581005586586e-05,
+      "loss": 1.0906,
+      "step": 249430
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.642343307568684e-05,
+      "loss": 1.053,
+      "step": 249440
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.642228514578709e-05,
+      "loss": 1.0484,
+      "step": 249450
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.642113721588735e-05,
+      "loss": 1.0776,
+      "step": 249460
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.64199892859876e-05,
+      "loss": 0.984,
+      "step": 249470
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.641884135608785e-05,
+      "loss": 1.0349,
+      "step": 249480
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.6417693426188106e-05,
+      "loss": 1.0791,
+      "step": 249490
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.641654549628836e-05,
+      "loss": 1.0029,
+      "step": 249500
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.6415397566388605e-05,
+      "loss": 1.1109,
+      "step": 249510
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.641424963648886e-05,
+      "loss": 1.073,
+      "step": 249520
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.641310170658912e-05,
+      "loss": 0.9114,
+      "step": 249530
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.641195377668936e-05,
+      "loss": 0.9699,
+      "step": 249540
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.641080584678962e-05,
+      "loss": 0.9762,
+      "step": 249550
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.6409657916889875e-05,
+      "loss": 1.0779,
+      "step": 249560
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.6408509986990125e-05,
+      "loss": 1.0197,
+      "step": 249570
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.6407362057090375e-05,
+      "loss": 1.0471,
+      "step": 249580
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.640621412719063e-05,
+      "loss": 1.1184,
+      "step": 249590
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.640506619729088e-05,
+      "loss": 1.097,
+      "step": 249600
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.640391826739114e-05,
+      "loss": 1.0204,
+      "step": 249610
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.640277033749139e-05,
+      "loss": 1.0566,
+      "step": 249620
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.640162240759164e-05,
+      "loss": 0.9713,
+      "step": 249630
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.6400474477691894e-05,
+      "loss": 1.0083,
+      "step": 249640
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.639932654779214e-05,
+      "loss": 0.9522,
+      "step": 249650
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.6398178617892394e-05,
+      "loss": 0.9878,
+      "step": 249660
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.639703068799265e-05,
+      "loss": 0.8995,
+      "step": 249670
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.63958827580929e-05,
+      "loss": 1.1375,
+      "step": 249680
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.639473482819315e-05,
+      "loss": 1.044,
+      "step": 249690
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.639358689829341e-05,
+      "loss": 0.9903,
+      "step": 249700
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.639243896839366e-05,
+      "loss": 0.9989,
+      "step": 249710
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.639129103849391e-05,
+      "loss": 1.1432,
+      "step": 249720
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.639014310859416e-05,
+      "loss": 1.0529,
+      "step": 249730
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.638899517869441e-05,
+      "loss": 0.9432,
+      "step": 249740
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.638784724879467e-05,
+      "loss": 1.0092,
+      "step": 249750
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.6386699318894926e-05,
+      "loss": 1.0216,
+      "step": 249760
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.638555138899517e-05,
+      "loss": 1.0783,
+      "step": 249770
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.6384403459095426e-05,
+      "loss": 1.0431,
+      "step": 249780
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.638325552919568e-05,
+      "loss": 0.9687,
+      "step": 249790
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.6382107599295926e-05,
+      "loss": 1.1428,
+      "step": 249800
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.638095966939618e-05,
+      "loss": 0.9977,
+      "step": 249810
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.637981173949644e-05,
+      "loss": 1.046,
+      "step": 249820
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.637866380959669e-05,
+      "loss": 1.0114,
+      "step": 249830
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.637751587969694e-05,
+      "loss": 1.1002,
+      "step": 249840
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.6376367949797195e-05,
+      "loss": 0.9938,
+      "step": 249850
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.6375220019897445e-05,
+      "loss": 1.092,
+      "step": 249860
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.63740720899977e-05,
+      "loss": 1.0269,
+      "step": 249870
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.637292416009795e-05,
+      "loss": 1.0342,
+      "step": 249880
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.63717762301982e-05,
+      "loss": 1.0041,
+      "step": 249890
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.637062830029846e-05,
+      "loss": 0.9971,
+      "step": 249900
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.6369480370398715e-05,
+      "loss": 0.9456,
+      "step": 249910
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.636833244049896e-05,
+      "loss": 1.1582,
+      "step": 249920
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.6367184510599214e-05,
+      "loss": 1.0312,
+      "step": 249930
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.636603658069947e-05,
+      "loss": 1.0514,
+      "step": 249940
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.6364888650799714e-05,
+      "loss": 1.0145,
+      "step": 249950
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.636374072089997e-05,
+      "loss": 1.0574,
+      "step": 249960
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.636259279100023e-05,
+      "loss": 0.977,
+      "step": 249970
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.636144486110048e-05,
+      "loss": 1.0289,
+      "step": 249980
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.636029693120073e-05,
+      "loss": 0.9786,
+      "step": 249990
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.6359149001300984e-05,
+      "loss": 1.1131,
+      "step": 250000
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.6358001071401234e-05,
+      "loss": 1.1202,
+      "step": 250010
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.635685314150149e-05,
+      "loss": 1.0212,
+      "step": 250020
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.635570521160174e-05,
+      "loss": 0.9602,
+      "step": 250030
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.635455728170199e-05,
+      "loss": 1.0024,
+      "step": 250040
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.6353409351802247e-05,
+      "loss": 1.035,
+      "step": 250050
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.63522614219025e-05,
+      "loss": 1.1059,
+      "step": 250060
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.6351113492002746e-05,
+      "loss": 1.0307,
+      "step": 250070
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.6349965562103e-05,
+      "loss": 1.1014,
+      "step": 250080
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.634881763220326e-05,
+      "loss": 1.0856,
+      "step": 250090
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.63476697023035e-05,
+      "loss": 1.0214,
+      "step": 250100
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.634652177240376e-05,
+      "loss": 1.041,
+      "step": 250110
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.6345373842504016e-05,
+      "loss": 1.1376,
+      "step": 250120
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.6344225912604266e-05,
+      "loss": 0.9863,
+      "step": 250130
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.6343077982704516e-05,
+      "loss": 1.0058,
+      "step": 250140
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.634193005280477e-05,
+      "loss": 1.0632,
+      "step": 250150
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.634078212290502e-05,
+      "loss": 1.0686,
+      "step": 250160
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.633963419300528e-05,
+      "loss": 1.0908,
+      "step": 250170
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.633848626310553e-05,
+      "loss": 1.1286,
+      "step": 250180
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.633733833320578e-05,
+      "loss": 1.1519,
+      "step": 250190
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.6336190403306035e-05,
+      "loss": 1.0256,
+      "step": 250200
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.633504247340629e-05,
+      "loss": 1.1356,
+      "step": 250210
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.6333894543506535e-05,
+      "loss": 0.9987,
+      "step": 250220
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.633274661360679e-05,
+      "loss": 1.1146,
+      "step": 250230
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.633159868370705e-05,
+      "loss": 1.064,
+      "step": 250240
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.633045075380729e-05,
+      "loss": 0.8903,
+      "step": 250250
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.632930282390755e-05,
+      "loss": 1.1006,
+      "step": 250260
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.6328154894007804e-05,
+      "loss": 1.0307,
+      "step": 250270
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.6327006964108054e-05,
+      "loss": 1.0236,
+      "step": 250280
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.6325859034208304e-05,
+      "loss": 0.9397,
+      "step": 250290
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.632471110430856e-05,
+      "loss": 1.009,
+      "step": 250300
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.632356317440881e-05,
+      "loss": 0.8815,
+      "step": 250310
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.632241524450907e-05,
+      "loss": 0.9523,
+      "step": 250320
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.632126731460932e-05,
+      "loss": 0.8971,
+      "step": 250330
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.632011938470957e-05,
+      "loss": 1.0487,
+      "step": 250340
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.6318971454809823e-05,
+      "loss": 0.9539,
+      "step": 250350
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.631782352491008e-05,
+      "loss": 0.9749,
+      "step": 250360
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.631667559501032e-05,
+      "loss": 1.013,
+      "step": 250370
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.631552766511058e-05,
+      "loss": 1.0694,
+      "step": 250380
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.6314379735210836e-05,
+      "loss": 1.1047,
+      "step": 250390
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.631323180531108e-05,
+      "loss": 1.1452,
+      "step": 250400
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.6312083875411336e-05,
+      "loss": 0.9933,
+      "step": 250410
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.631093594551159e-05,
+      "loss": 1.0278,
+      "step": 250420
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.630978801561184e-05,
+      "loss": 1.101,
+      "step": 250430
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.630864008571209e-05,
+      "loss": 1.0609,
+      "step": 250440
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.630749215581235e-05,
+      "loss": 0.9639,
+      "step": 250450
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.63063442259126e-05,
+      "loss": 1.0665,
+      "step": 250460
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.6305196296012856e-05,
+      "loss": 1.0036,
+      "step": 250470
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.6304048366113105e-05,
+      "loss": 1.0052,
+      "step": 250480
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.6302900436213355e-05,
+      "loss": 1.0746,
+      "step": 250490
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.630175250631361e-05,
+      "loss": 1.1228,
+      "step": 250500
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.630060457641387e-05,
+      "loss": 1.0269,
+      "step": 250510
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.629945664651411e-05,
+      "loss": 0.9618,
+      "step": 250520
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.629830871661437e-05,
+      "loss": 1.0342,
+      "step": 250530
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.6297160786714625e-05,
+      "loss": 1.0663,
+      "step": 250540
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.629601285681487e-05,
+      "loss": 1.0897,
+      "step": 250550
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.6294864926915125e-05,
+      "loss": 1.0594,
+      "step": 250560
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.629371699701538e-05,
+      "loss": 0.9249,
+      "step": 250570
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.629256906711563e-05,
+      "loss": 1.0384,
+      "step": 250580
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.629142113721588e-05,
+      "loss": 1.1003,
+      "step": 250590
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.629027320731614e-05,
+      "loss": 0.9405,
+      "step": 250600
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.628912527741639e-05,
+      "loss": 1.0794,
+      "step": 250610
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.6287977347516644e-05,
+      "loss": 1.0589,
+      "step": 250620
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.6286829417616894e-05,
+      "loss": 1.0717,
+      "step": 250630
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.6285681487717144e-05,
+      "loss": 1.0118,
+      "step": 250640
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.62845335578174e-05,
+      "loss": 1.0197,
+      "step": 250650
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.628338562791766e-05,
+      "loss": 1.0298,
+      "step": 250660
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.62822376980179e-05,
+      "loss": 0.9903,
+      "step": 250670
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.628108976811816e-05,
+      "loss": 1.0465,
+      "step": 250680
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.6279941838218413e-05,
+      "loss": 1.0544,
+      "step": 250690
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.6278793908318656e-05,
+      "loss": 0.9273,
+      "step": 250700
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.627764597841891e-05,
+      "loss": 0.9774,
+      "step": 250710
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.627649804851917e-05,
+      "loss": 1.0643,
+      "step": 250720
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.627535011861942e-05,
+      "loss": 0.9463,
+      "step": 250730
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.627420218871967e-05,
+      "loss": 1.0437,
+      "step": 250740
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.6273054258819926e-05,
+      "loss": 1.001,
+      "step": 250750
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.6271906328920176e-05,
+      "loss": 0.9535,
+      "step": 250760
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.627075839902043e-05,
+      "loss": 0.9682,
+      "step": 250770
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.626961046912068e-05,
+      "loss": 1.0048,
+      "step": 250780
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.626846253922093e-05,
+      "loss": 1.0174,
+      "step": 250790
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.626731460932119e-05,
+      "loss": 1.0353,
+      "step": 250800
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.6266166679421446e-05,
+      "loss": 1.0272,
+      "step": 250810
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.626501874952169e-05,
+      "loss": 0.9888,
+      "step": 250820
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.6263870819621945e-05,
+      "loss": 1.0514,
+      "step": 250830
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.62627228897222e-05,
+      "loss": 1.0216,
+      "step": 250840
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.6261574959822445e-05,
+      "loss": 0.9636,
+      "step": 250850
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.62604270299227e-05,
+      "loss": 0.97,
+      "step": 250860
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.625927910002296e-05,
+      "loss": 1.0015,
+      "step": 250870
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.625813117012321e-05,
+      "loss": 0.9673,
+      "step": 250880
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.625698324022346e-05,
+      "loss": 0.9621,
+      "step": 250890
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.6255835310323715e-05,
+      "loss": 1.071,
+      "step": 250900
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.6254687380423964e-05,
+      "loss": 1.024,
+      "step": 250910
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.625353945052422e-05,
+      "loss": 1.0802,
+      "step": 250920
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.625239152062447e-05,
+      "loss": 1.1276,
+      "step": 250930
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.625124359072472e-05,
+      "loss": 1.0563,
+      "step": 250940
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.625009566082498e-05,
+      "loss": 0.9623,
+      "step": 250950
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.6248947730925234e-05,
+      "loss": 1.0353,
+      "step": 250960
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.624779980102548e-05,
+      "loss": 1.0681,
+      "step": 250970
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.6246651871125734e-05,
+      "loss": 1.0099,
+      "step": 250980
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.624550394122599e-05,
+      "loss": 1.0491,
+      "step": 250990
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.6244356011326233e-05,
+      "loss": 0.9869,
+      "step": 251000
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.624320808142649e-05,
+      "loss": 1.1405,
+      "step": 251010
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.624206015152674e-05,
+      "loss": 1.1679,
+      "step": 251020
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.6240912221626997e-05,
+      "loss": 1.0933,
+      "step": 251030
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.6239764291727246e-05,
+      "loss": 0.9995,
+      "step": 251040
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.6238616361827496e-05,
+      "loss": 0.9954,
+      "step": 251050
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.623746843192775e-05,
+      "loss": 1.0963,
+      "step": 251060
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.623632050202801e-05,
+      "loss": 1.0377,
+      "step": 251070
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.623517257212825e-05,
+      "loss": 1.0353,
+      "step": 251080
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.623402464222851e-05,
+      "loss": 1.0805,
+      "step": 251090
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.6232876712328766e-05,
+      "loss": 0.9734,
+      "step": 251100
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.623172878242901e-05,
+      "loss": 1.0286,
+      "step": 251110
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.6230580852529266e-05,
+      "loss": 1.0331,
+      "step": 251120
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.622943292262952e-05,
+      "loss": 1.0608,
+      "step": 251130
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.622828499272977e-05,
+      "loss": 1.0591,
+      "step": 251140
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.622713706283002e-05,
+      "loss": 1.0432,
+      "step": 251150
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.622598913293028e-05,
+      "loss": 1.0009,
+      "step": 251160
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.622484120303053e-05,
+      "loss": 0.9795,
+      "step": 251170
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.6223693273130785e-05,
+      "loss": 0.9625,
+      "step": 251180
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.6222545343231035e-05,
+      "loss": 0.9695,
+      "step": 251190
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.6221397413331285e-05,
+      "loss": 1.0066,
+      "step": 251200
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.622024948343154e-05,
+      "loss": 1.0524,
+      "step": 251210
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.62191015535318e-05,
+      "loss": 0.9962,
+      "step": 251220
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.621795362363204e-05,
+      "loss": 1.046,
+      "step": 251230
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.62168056937323e-05,
+      "loss": 1.0039,
+      "step": 251240
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.6215657763832554e-05,
+      "loss": 1.025,
+      "step": 251250
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.62145098339328e-05,
+      "loss": 1.0718,
+      "step": 251260
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.6213361904033054e-05,
+      "loss": 1.0233,
+      "step": 251270
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.621221397413331e-05,
+      "loss": 1.0965,
+      "step": 251280
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.621106604423356e-05,
+      "loss": 1.0378,
+      "step": 251290
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.620991811433381e-05,
+      "loss": 0.9896,
+      "step": 251300
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.620877018443407e-05,
+      "loss": 1.1157,
+      "step": 251310
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.620762225453432e-05,
+      "loss": 1.1048,
+      "step": 251320
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.6206474324634573e-05,
+      "loss": 1.0961,
+      "step": 251330
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.620532639473482e-05,
+      "loss": 1.0193,
+      "step": 251340
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.620417846483507e-05,
+      "loss": 1.12,
+      "step": 251350
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.620303053493533e-05,
+      "loss": 0.9537,
+      "step": 251360
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.6201882605035586e-05,
+      "loss": 0.9864,
+      "step": 251370
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.620073467513583e-05,
+      "loss": 1.0988,
+      "step": 251380
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.6199586745236086e-05,
+      "loss": 0.9536,
+      "step": 251390
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.619843881533634e-05,
+      "loss": 1.0071,
+      "step": 251400
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.6197290885436586e-05,
+      "loss": 0.993,
+      "step": 251410
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.619614295553684e-05,
+      "loss": 1.0447,
+      "step": 251420
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.61949950256371e-05,
+      "loss": 1.0786,
+      "step": 251430
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.619384709573735e-05,
+      "loss": 1.1531,
+      "step": 251440
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.61926991658376e-05,
+      "loss": 1.0941,
+      "step": 251450
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.6191551235937855e-05,
+      "loss": 0.9337,
+      "step": 251460
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.6190403306038105e-05,
+      "loss": 0.9519,
+      "step": 251470
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.618925537613836e-05,
+      "loss": 1.0066,
+      "step": 251480
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.618810744623861e-05,
+      "loss": 1.0416,
+      "step": 251490
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.618695951633886e-05,
+      "loss": 1.1756,
+      "step": 251500
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.618581158643912e-05,
+      "loss": 1.0065,
+      "step": 251510
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.6184663656539375e-05,
+      "loss": 1.1581,
+      "step": 251520
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.618351572663962e-05,
+      "loss": 0.9993,
+      "step": 251530
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.6182367796739875e-05,
+      "loss": 1.0504,
+      "step": 251540
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.618121986684013e-05,
+      "loss": 1.0496,
+      "step": 251550
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.6180071936940374e-05,
+      "loss": 0.9789,
+      "step": 251560
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.617892400704063e-05,
+      "loss": 1.0421,
+      "step": 251570
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.617777607714089e-05,
+      "loss": 1.088,
+      "step": 251580
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.617662814724114e-05,
+      "loss": 1.0863,
+      "step": 251590
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.617548021734139e-05,
+      "loss": 1.037,
+      "step": 251600
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.6174332287441644e-05,
+      "loss": 1.1236,
+      "step": 251610
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.6173184357541894e-05,
+      "loss": 1.1058,
+      "step": 251620
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.617203642764215e-05,
+      "loss": 1.1058,
+      "step": 251630
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.61708884977424e-05,
+      "loss": 0.9145,
+      "step": 251640
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.616974056784265e-05,
+      "loss": 1.0423,
+      "step": 251650
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.616859263794291e-05,
+      "loss": 1.0438,
+      "step": 251660
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.6167444708043163e-05,
+      "loss": 0.96,
+      "step": 251670
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.6166296778143406e-05,
+      "loss": 1.0343,
+      "step": 251680
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.616514884824366e-05,
+      "loss": 0.9921,
+      "step": 251690
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.616400091834392e-05,
+      "loss": 1.0309,
+      "step": 251700
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.616285298844416e-05,
+      "loss": 1.0055,
+      "step": 251710
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.616170505854442e-05,
+      "loss": 0.96,
+      "step": 251720
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.6160557128644676e-05,
+      "loss": 1.1252,
+      "step": 251730
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.6159409198744926e-05,
+      "loss": 0.9848,
+      "step": 251740
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.6158261268845176e-05,
+      "loss": 1.0185,
+      "step": 251750
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.615711333894543e-05,
+      "loss": 1.1815,
+      "step": 251760
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.615596540904568e-05,
+      "loss": 1.0004,
+      "step": 251770
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.615481747914594e-05,
+      "loss": 1.1063,
+      "step": 251780
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.615366954924619e-05,
+      "loss": 0.981,
+      "step": 251790
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.615252161934644e-05,
+      "loss": 1.0833,
+      "step": 251800
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.6151373689446695e-05,
+      "loss": 1.0625,
+      "step": 251810
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.615022575954695e-05,
+      "loss": 1.0801,
+      "step": 251820
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.6149077829647195e-05,
+      "loss": 0.9799,
+      "step": 251830
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.614792989974745e-05,
+      "loss": 1.0156,
+      "step": 251840
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.614678196984771e-05,
+      "loss": 1.1598,
+      "step": 251850
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.614563403994795e-05,
+      "loss": 0.9826,
+      "step": 251860
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.614448611004821e-05,
+      "loss": 0.9715,
+      "step": 251870
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.6143338180148465e-05,
+      "loss": 0.8777,
+      "step": 251880
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.6142190250248714e-05,
+      "loss": 1.0348,
+      "step": 251890
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.6141042320348964e-05,
+      "loss": 1.0566,
+      "step": 251900
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.613989439044922e-05,
+      "loss": 1.0068,
+      "step": 251910
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.613874646054947e-05,
+      "loss": 0.9531,
+      "step": 251920
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.613759853064973e-05,
+      "loss": 1.0408,
+      "step": 251930
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.613645060074998e-05,
+      "loss": 0.9261,
+      "step": 251940
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.613530267085023e-05,
+      "loss": 1.0384,
+      "step": 251950
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.6134154740950484e-05,
+      "loss": 0.9956,
+      "step": 251960
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.613300681105074e-05,
+      "loss": 1.1184,
+      "step": 251970
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.6131858881150983e-05,
+      "loss": 0.9802,
+      "step": 251980
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.613071095125124e-05,
+      "loss": 0.9866,
+      "step": 251990
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.61295630213515e-05,
+      "loss": 1.0644,
+      "step": 252000
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.612841509145174e-05,
+      "loss": 1.0339,
+      "step": 252010
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.6127267161551996e-05,
+      "loss": 0.9947,
+      "step": 252020
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.612611923165225e-05,
+      "loss": 1.0477,
+      "step": 252030
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.61249713017525e-05,
+      "loss": 0.9996,
+      "step": 252040
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.612382337185275e-05,
+      "loss": 1.0741,
+      "step": 252050
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.612267544195301e-05,
+      "loss": 1.2244,
+      "step": 252060
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.612152751205326e-05,
+      "loss": 0.9979,
+      "step": 252070
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.6120379582153516e-05,
+      "loss": 1.027,
+      "step": 252080
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.6119231652253766e-05,
+      "loss": 1.0125,
+      "step": 252090
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.6118083722354016e-05,
+      "loss": 0.9989,
+      "step": 252100
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.611693579245427e-05,
+      "loss": 1.1096,
+      "step": 252110
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.611578786255453e-05,
+      "loss": 1.1765,
+      "step": 252120
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.611463993265477e-05,
+      "loss": 1.0259,
+      "step": 252130
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.611349200275503e-05,
+      "loss": 1.0302,
+      "step": 252140
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.6112344072855285e-05,
+      "loss": 1.1664,
+      "step": 252150
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.611119614295553e-05,
+      "loss": 1.0887,
+      "step": 252160
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.6110048213055785e-05,
+      "loss": 1.0461,
+      "step": 252170
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.610890028315604e-05,
+      "loss": 1.0361,
+      "step": 252180
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.610775235325629e-05,
+      "loss": 1.1608,
+      "step": 252190
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.610660442335654e-05,
+      "loss": 1.1288,
+      "step": 252200
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.61054564934568e-05,
+      "loss": 0.9587,
+      "step": 252210
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.610430856355705e-05,
+      "loss": 1.1637,
+      "step": 252220
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.6103160633657304e-05,
+      "loss": 1.0696,
+      "step": 252230
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.6102012703757554e-05,
+      "loss": 1.1347,
+      "step": 252240
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.6100864773857804e-05,
+      "loss": 1.0285,
+      "step": 252250
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.609971684395806e-05,
+      "loss": 1.1238,
+      "step": 252260
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.609856891405832e-05,
+      "loss": 1.1269,
+      "step": 252270
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.609742098415856e-05,
+      "loss": 1.0697,
+      "step": 252280
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.609627305425882e-05,
+      "loss": 1.0241,
+      "step": 252290
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.6095125124359074e-05,
+      "loss": 0.9961,
+      "step": 252300
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.609397719445932e-05,
+      "loss": 1.0779,
+      "step": 252310
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.609282926455957e-05,
+      "loss": 1.1666,
+      "step": 252320
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.609168133465983e-05,
+      "loss": 1.0213,
+      "step": 252330
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.609053340476008e-05,
+      "loss": 1.0346,
+      "step": 252340
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.608938547486033e-05,
+      "loss": 1.0933,
+      "step": 252350
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.6088237544960586e-05,
+      "loss": 1.0318,
+      "step": 252360
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.6087089615060836e-05,
+      "loss": 1.1505,
+      "step": 252370
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.608594168516109e-05,
+      "loss": 1.0146,
+      "step": 252380
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.6084793755261336e-05,
+      "loss": 1.1489,
+      "step": 252390
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.608364582536159e-05,
+      "loss": 1.0303,
+      "step": 252400
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.608249789546185e-05,
+      "loss": 1.0167,
+      "step": 252410
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.608134996556209e-05,
+      "loss": 0.9719,
+      "step": 252420
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.608020203566235e-05,
+      "loss": 0.9902,
+      "step": 252430
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.6079054105762605e-05,
+      "loss": 1.0528,
+      "step": 252440
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.6077906175862855e-05,
+      "loss": 1.0893,
+      "step": 252450
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.6076758245963105e-05,
+      "loss": 1.1423,
+      "step": 252460
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.607561031606336e-05,
+      "loss": 1.0616,
+      "step": 252470
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.607446238616361e-05,
+      "loss": 0.8957,
+      "step": 252480
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.607331445626387e-05,
+      "loss": 1.0589,
+      "step": 252490
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.607216652636412e-05,
+      "loss": 1.0104,
+      "step": 252500
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.607101859646437e-05,
+      "loss": 1.0485,
+      "step": 252510
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.6069870666564625e-05,
+      "loss": 1.0021,
+      "step": 252520
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.606872273666488e-05,
+      "loss": 1.1871,
+      "step": 252530
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.6067574806765124e-05,
+      "loss": 1.1485,
+      "step": 252540
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.606642687686538e-05,
+      "loss": 1.092,
+      "step": 252550
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.606527894696564e-05,
+      "loss": 1.0931,
+      "step": 252560
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.606413101706588e-05,
+      "loss": 1.0803,
+      "step": 252570
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.606298308716614e-05,
+      "loss": 1.0831,
+      "step": 252580
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.6061835157266394e-05,
+      "loss": 1.0456,
+      "step": 252590
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.6060687227366644e-05,
+      "loss": 1.011,
+      "step": 252600
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.6059539297466894e-05,
+      "loss": 1.0753,
+      "step": 252610
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.605839136756715e-05,
+      "loss": 1.0254,
+      "step": 252620
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.60572434376674e-05,
+      "loss": 1.0131,
+      "step": 252630
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.605609550776766e-05,
+      "loss": 1.1109,
+      "step": 252640
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.605494757786791e-05,
+      "loss": 1.079,
+      "step": 252650
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.6053799647968156e-05,
+      "loss": 1.0353,
+      "step": 252660
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.605265171806841e-05,
+      "loss": 1.1546,
+      "step": 252670
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.605150378816867e-05,
+      "loss": 0.9755,
+      "step": 252680
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.605035585826891e-05,
+      "loss": 1.0378,
+      "step": 252690
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.604920792836917e-05,
+      "loss": 1.0489,
+      "step": 252700
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.6048059998469426e-05,
+      "loss": 1.087,
+      "step": 252710
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.604691206856967e-05,
+      "loss": 1.0685,
+      "step": 252720
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.6045764138669926e-05,
+      "loss": 1.0141,
+      "step": 252730
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.604461620877018e-05,
+      "loss": 1.0298,
+      "step": 252740
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.604346827887043e-05,
+      "loss": 1.0243,
+      "step": 252750
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.604232034897068e-05,
+      "loss": 1.0624,
+      "step": 252760
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.604117241907094e-05,
+      "loss": 1.092,
+      "step": 252770
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.604002448917119e-05,
+      "loss": 0.9377,
+      "step": 252780
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.6038876559271445e-05,
+      "loss": 0.9562,
+      "step": 252790
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.6037728629371695e-05,
+      "loss": 1.0583,
+      "step": 252800
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.6036580699471945e-05,
+      "loss": 1.0806,
+      "step": 252810
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.60354327695722e-05,
+      "loss": 0.9958,
+      "step": 252820
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.603428483967246e-05,
+      "loss": 1.1127,
+      "step": 252830
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.60331369097727e-05,
+      "loss": 1.1029,
+      "step": 252840
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.603198897987296e-05,
+      "loss": 1.0018,
+      "step": 252850
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.6030841049973215e-05,
+      "loss": 1.1318,
+      "step": 252860
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.602969312007346e-05,
+      "loss": 1.0603,
+      "step": 252870
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.6028545190173714e-05,
+      "loss": 1.105,
+      "step": 252880
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.602739726027397e-05,
+      "loss": 1.0206,
+      "step": 252890
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.602624933037422e-05,
+      "loss": 0.9206,
+      "step": 252900
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.602510140047447e-05,
+      "loss": 1.0868,
+      "step": 252910
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.602395347057473e-05,
+      "loss": 0.8814,
+      "step": 252920
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.602280554067498e-05,
+      "loss": 1.0195,
+      "step": 252930
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.6021657610775234e-05,
+      "loss": 1.1134,
+      "step": 252940
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.6020509680875484e-05,
+      "loss": 1.1051,
+      "step": 252950
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.6019361750975733e-05,
+      "loss": 1.0323,
+      "step": 252960
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.601821382107599e-05,
+      "loss": 1.0062,
+      "step": 252970
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.601706589117625e-05,
+      "loss": 1.0561,
+      "step": 252980
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.601591796127649e-05,
+      "loss": 1.1455,
+      "step": 252990
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.6014770031376746e-05,
+      "loss": 1.0347,
+      "step": 253000
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.6013622101477e-05,
+      "loss": 1.0706,
+      "step": 253010
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.6012474171577246e-05,
+      "loss": 0.9531,
+      "step": 253020
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.60113262416775e-05,
+      "loss": 1.079,
+      "step": 253030
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.601017831177776e-05,
+      "loss": 1.0916,
+      "step": 253040
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.600903038187801e-05,
+      "loss": 0.893,
+      "step": 253050
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.600788245197826e-05,
+      "loss": 0.9958,
+      "step": 253060
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.6006734522078516e-05,
+      "loss": 1.0912,
+      "step": 253070
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.6005586592178766e-05,
+      "loss": 1.0448,
+      "step": 253080
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.600443866227902e-05,
+      "loss": 1.1112,
+      "step": 253090
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.600329073237927e-05,
+      "loss": 1.0508,
+      "step": 253100
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.600214280247952e-05,
+      "loss": 1.0116,
+      "step": 253110
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.600099487257978e-05,
+      "loss": 0.8848,
+      "step": 253120
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.5999846942680035e-05,
+      "loss": 1.1077,
+      "step": 253130
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.599869901278028e-05,
+      "loss": 0.9562,
+      "step": 253140
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.5997551082880535e-05,
+      "loss": 1.0509,
+      "step": 253150
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.599640315298079e-05,
+      "loss": 0.9381,
+      "step": 253160
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.5995255223081035e-05,
+      "loss": 1.1106,
+      "step": 253170
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.599410729318129e-05,
+      "loss": 1.0184,
+      "step": 253180
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.599295936328155e-05,
+      "loss": 1.0625,
+      "step": 253190
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.59918114333818e-05,
+      "loss": 1.0529,
+      "step": 253200
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.599066350348205e-05,
+      "loss": 0.946,
+      "step": 253210
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.5989515573582304e-05,
+      "loss": 0.9421,
+      "step": 253220
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.5988367643682554e-05,
+      "loss": 1.102,
+      "step": 253230
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.598721971378281e-05,
+      "loss": 1.0869,
+      "step": 253240
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.598607178388306e-05,
+      "loss": 1.0283,
+      "step": 253250
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.598492385398331e-05,
+      "loss": 1.0048,
+      "step": 253260
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.598377592408357e-05,
+      "loss": 1.036,
+      "step": 253270
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.5982627994183824e-05,
+      "loss": 0.9456,
+      "step": 253280
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.598148006428407e-05,
+      "loss": 1.0332,
+      "step": 253290
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.598033213438432e-05,
+      "loss": 0.9265,
+      "step": 253300
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.597918420448458e-05,
+      "loss": 0.9109,
+      "step": 253310
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.597803627458482e-05,
+      "loss": 1.0357,
+      "step": 253320
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.597688834468508e-05,
+      "loss": 0.9893,
+      "step": 253330
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.5975740414785336e-05,
+      "loss": 0.9687,
+      "step": 253340
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.5974592484885586e-05,
+      "loss": 0.9743,
+      "step": 253350
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.5973444554985836e-05,
+      "loss": 1.05,
+      "step": 253360
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.597229662508609e-05,
+      "loss": 1.0427,
+      "step": 253370
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.597114869518634e-05,
+      "loss": 1.0422,
+      "step": 253380
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.59700007652866e-05,
+      "loss": 1.1377,
+      "step": 253390
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.596885283538685e-05,
+      "loss": 0.9611,
+      "step": 253400
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.59677049054871e-05,
+      "loss": 0.9602,
+      "step": 253410
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.5966556975587355e-05,
+      "loss": 1.0451,
+      "step": 253420
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.596540904568761e-05,
+      "loss": 1.0589,
+      "step": 253430
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.5964261115787855e-05,
+      "loss": 1.1021,
+      "step": 253440
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.596311318588811e-05,
+      "loss": 1.0448,
+      "step": 253450
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.596196525598837e-05,
+      "loss": 0.9862,
+      "step": 253460
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.596081732608861e-05,
+      "loss": 1.0993,
+      "step": 253470
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.595966939618887e-05,
+      "loss": 1.0195,
+      "step": 253480
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.5958521466289125e-05,
+      "loss": 1.0741,
+      "step": 253490
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.5957373536389375e-05,
+      "loss": 1.0681,
+      "step": 253500
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.5956225606489624e-05,
+      "loss": 1.0397,
+      "step": 253510
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.595507767658988e-05,
+      "loss": 1.0739,
+      "step": 253520
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.595392974669013e-05,
+      "loss": 1.0642,
+      "step": 253530
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.595278181679039e-05,
+      "loss": 1.1338,
+      "step": 253540
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.595163388689064e-05,
+      "loss": 1.0062,
+      "step": 253550
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.595048595699089e-05,
+      "loss": 0.956,
+      "step": 253560
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.5949338027091144e-05,
+      "loss": 0.9754,
+      "step": 253570
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.59481900971914e-05,
+      "loss": 1.1698,
+      "step": 253580
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.5947042167291644e-05,
+      "loss": 0.9353,
+      "step": 253590
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.59458942373919e-05,
+      "loss": 1.1552,
+      "step": 253600
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.594474630749216e-05,
+      "loss": 1.0694,
+      "step": 253610
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.59435983775924e-05,
+      "loss": 1.0485,
+      "step": 253620
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.594245044769266e-05,
+      "loss": 0.9758,
+      "step": 253630
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.594130251779291e-05,
+      "loss": 1.0784,
+      "step": 253640
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.594015458789316e-05,
+      "loss": 1.0595,
+      "step": 253650
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.593900665799341e-05,
+      "loss": 1.0614,
+      "step": 253660
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.593785872809367e-05,
+      "loss": 1.075,
+      "step": 253670
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.593671079819392e-05,
+      "loss": 0.966,
+      "step": 253680
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.5935562868294176e-05,
+      "loss": 1.0013,
+      "step": 253690
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.5934414938394426e-05,
+      "loss": 0.9666,
+      "step": 253700
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.5933267008494676e-05,
+      "loss": 1.0113,
+      "step": 253710
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.593211907859493e-05,
+      "loss": 1.0959,
+      "step": 253720
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.593097114869519e-05,
+      "loss": 1.1736,
+      "step": 253730
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.592982321879543e-05,
+      "loss": 0.9497,
+      "step": 253740
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.592867528889569e-05,
+      "loss": 1.0482,
+      "step": 253750
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.592752735899594e-05,
+      "loss": 1.0016,
+      "step": 253760
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.592637942909619e-05,
+      "loss": 0.9868,
+      "step": 253770
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.5925231499196445e-05,
+      "loss": 1.1873,
+      "step": 253780
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.5924083569296695e-05,
+      "loss": 0.8841,
+      "step": 253790
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.592293563939695e-05,
+      "loss": 1.0369,
+      "step": 253800
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.59217877094972e-05,
+      "loss": 1.0534,
+      "step": 253810
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.592063977959745e-05,
+      "loss": 0.9816,
+      "step": 253820
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.591949184969771e-05,
+      "loss": 0.9667,
+      "step": 253830
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.5918343919797965e-05,
+      "loss": 1.0687,
+      "step": 253840
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.591719598989821e-05,
+      "loss": 1.1038,
+      "step": 253850
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.5916048059998464e-05,
+      "loss": 1.1411,
+      "step": 253860
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.591490013009872e-05,
+      "loss": 1.1752,
+      "step": 253870
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.5913752200198964e-05,
+      "loss": 1.0663,
+      "step": 253880
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.591260427029922e-05,
+      "loss": 1.0514,
+      "step": 253890
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.591145634039948e-05,
+      "loss": 1.0688,
+      "step": 253900
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.591030841049973e-05,
+      "loss": 1.0679,
+      "step": 253910
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.590916048059998e-05,
+      "loss": 1.0341,
+      "step": 253920
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.5908012550700234e-05,
+      "loss": 1.0047,
+      "step": 253930
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.5906864620800483e-05,
+      "loss": 1.0429,
+      "step": 253940
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.590571669090074e-05,
+      "loss": 1.0085,
+      "step": 253950
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.590456876100099e-05,
+      "loss": 0.9943,
+      "step": 253960
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.590342083110124e-05,
+      "loss": 1.0377,
+      "step": 253970
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.5902272901201496e-05,
+      "loss": 1.0324,
+      "step": 253980
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.590112497130175e-05,
+      "loss": 1.0138,
+      "step": 253990
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.5899977041401996e-05,
+      "loss": 1.0178,
+      "step": 254000
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.589882911150225e-05,
+      "loss": 1.019,
+      "step": 254010
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.589768118160251e-05,
+      "loss": 0.9381,
+      "step": 254020
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.589653325170275e-05,
+      "loss": 1.0726,
+      "step": 254030
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.589538532180301e-05,
+      "loss": 0.9692,
+      "step": 254040
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.5894237391903266e-05,
+      "loss": 1.1271,
+      "step": 254050
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.5893089462003516e-05,
+      "loss": 1.0479,
+      "step": 254060
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.5891941532103765e-05,
+      "loss": 1.1435,
+      "step": 254070
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.589079360220402e-05,
+      "loss": 1.0566,
+      "step": 254080
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.588964567230427e-05,
+      "loss": 0.9778,
+      "step": 254090
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.588849774240453e-05,
+      "loss": 0.9521,
+      "step": 254100
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.588734981250478e-05,
+      "loss": 1.1774,
+      "step": 254110
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.588620188260503e-05,
+      "loss": 1.2656,
+      "step": 254120
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.5885053952705285e-05,
+      "loss": 1.029,
+      "step": 254130
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.588390602280554e-05,
+      "loss": 1.0452,
+      "step": 254140
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.5882758092905785e-05,
+      "loss": 1.0703,
+      "step": 254150
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.588161016300604e-05,
+      "loss": 1.147,
+      "step": 254160
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.58804622331063e-05,
+      "loss": 0.9808,
+      "step": 254170
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.587931430320654e-05,
+      "loss": 1.0123,
+      "step": 254180
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.58781663733068e-05,
+      "loss": 0.9807,
+      "step": 254190
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.5877018443407054e-05,
+      "loss": 1.0381,
+      "step": 254200
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.5875870513507304e-05,
+      "loss": 0.9941,
+      "step": 254210
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.5874722583607554e-05,
+      "loss": 1.0507,
+      "step": 254220
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.587357465370781e-05,
+      "loss": 1.1051,
+      "step": 254230
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.587242672380806e-05,
+      "loss": 1.1647,
+      "step": 254240
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.587127879390832e-05,
+      "loss": 0.9451,
+      "step": 254250
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.587013086400857e-05,
+      "loss": 1.0411,
+      "step": 254260
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.586898293410882e-05,
+      "loss": 1.0361,
+      "step": 254270
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.586783500420907e-05,
+      "loss": 1.0324,
+      "step": 254280
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.586668707430933e-05,
+      "loss": 0.9583,
+      "step": 254290
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.586553914440957e-05,
+      "loss": 0.9191,
+      "step": 254300
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.586439121450983e-05,
+      "loss": 1.0094,
+      "step": 254310
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.5863243284610086e-05,
+      "loss": 1.0854,
+      "step": 254320
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.586209535471033e-05,
+      "loss": 1.0458,
+      "step": 254330
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.5860947424810586e-05,
+      "loss": 1.0119,
+      "step": 254340
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.585979949491084e-05,
+      "loss": 0.9774,
+      "step": 254350
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.585865156501109e-05,
+      "loss": 1.0294,
+      "step": 254360
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.585750363511134e-05,
+      "loss": 1.1184,
+      "step": 254370
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.58563557052116e-05,
+      "loss": 0.9864,
+      "step": 254380
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.585520777531185e-05,
+      "loss": 1.0006,
+      "step": 254390
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.5854059845412105e-05,
+      "loss": 1.1009,
+      "step": 254400
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.5852911915512355e-05,
+      "loss": 1.0841,
+      "step": 254410
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.5851763985612605e-05,
+      "loss": 1.1492,
+      "step": 254420
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.585061605571286e-05,
+      "loss": 1.0009,
+      "step": 254430
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.584946812581312e-05,
+      "loss": 0.907,
+      "step": 254440
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.584832019591336e-05,
+      "loss": 1.087,
+      "step": 254450
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.584717226601362e-05,
+      "loss": 1.112,
+      "step": 254460
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.5846024336113875e-05,
+      "loss": 0.9987,
+      "step": 254470
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.584487640621412e-05,
+      "loss": 1.1072,
+      "step": 254480
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.5843728476314374e-05,
+      "loss": 0.9816,
+      "step": 254490
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.584258054641463e-05,
+      "loss": 1.0914,
+      "step": 254500
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.584143261651488e-05,
+      "loss": 0.9983,
+      "step": 254510
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.584028468661513e-05,
+      "loss": 0.9751,
+      "step": 254520
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.583913675671539e-05,
+      "loss": 1.1473,
+      "step": 254530
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.583798882681564e-05,
+      "loss": 1.051,
+      "step": 254540
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.5836840896915894e-05,
+      "loss": 0.9871,
+      "step": 254550
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.5835692967016144e-05,
+      "loss": 0.9692,
+      "step": 254560
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.5834545037116394e-05,
+      "loss": 0.9764,
+      "step": 254570
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.583339710721665e-05,
+      "loss": 1.1057,
+      "step": 254580
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.583224917731691e-05,
+      "loss": 1.1978,
+      "step": 254590
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.583110124741715e-05,
+      "loss": 1.0305,
+      "step": 254600
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.582995331751741e-05,
+      "loss": 0.9834,
+      "step": 254610
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.582880538761766e-05,
+      "loss": 1.0284,
+      "step": 254620
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.5827657457717906e-05,
+      "loss": 1.0499,
+      "step": 254630
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.582650952781816e-05,
+      "loss": 1.1509,
+      "step": 254640
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.582536159791842e-05,
+      "loss": 1.0362,
+      "step": 254650
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.582421366801867e-05,
+      "loss": 0.9568,
+      "step": 254660
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.582306573811892e-05,
+      "loss": 1.0045,
+      "step": 254670
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.5821917808219176e-05,
+      "loss": 0.9948,
+      "step": 254680
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.5820769878319426e-05,
+      "loss": 0.9908,
+      "step": 254690
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.581962194841968e-05,
+      "loss": 1.1119,
+      "step": 254700
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.581847401851993e-05,
+      "loss": 1.0598,
+      "step": 254710
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.581732608862018e-05,
+      "loss": 0.9375,
+      "step": 254720
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.581617815872044e-05,
+      "loss": 1.1723,
+      "step": 254730
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.5815030228820695e-05,
+      "loss": 0.9604,
+      "step": 254740
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.581388229892094e-05,
+      "loss": 1.0253,
+      "step": 254750
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.5812734369021195e-05,
+      "loss": 1.0174,
+      "step": 254760
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.581158643912145e-05,
+      "loss": 1.1857,
+      "step": 254770
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.5810438509221695e-05,
+      "loss": 1.1202,
+      "step": 254780
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.580929057932195e-05,
+      "loss": 1.0034,
+      "step": 254790
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.580814264942221e-05,
+      "loss": 1.0955,
+      "step": 254800
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.580699471952246e-05,
+      "loss": 0.9928,
+      "step": 254810
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.580584678962271e-05,
+      "loss": 1.0323,
+      "step": 254820
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.5804698859722964e-05,
+      "loss": 1.2745,
+      "step": 254830
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.5803550929823214e-05,
+      "loss": 1.1103,
+      "step": 254840
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.580240299992347e-05,
+      "loss": 1.0949,
+      "step": 254850
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.580125507002372e-05,
+      "loss": 0.9949,
+      "step": 254860
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.580010714012397e-05,
+      "loss": 1.0746,
+      "step": 254870
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.579895921022423e-05,
+      "loss": 1.0748,
+      "step": 254880
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.5797811280324484e-05,
+      "loss": 1.0165,
+      "step": 254890
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.579666335042473e-05,
+      "loss": 0.983,
+      "step": 254900
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.5795515420524984e-05,
+      "loss": 0.9998,
+      "step": 254910
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.579436749062524e-05,
+      "loss": 1.0582,
+      "step": 254920
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.579321956072548e-05,
+      "loss": 1.0302,
+      "step": 254930
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.579207163082574e-05,
+      "loss": 1.02,
+      "step": 254940
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.5790923700925997e-05,
+      "loss": 0.9656,
+      "step": 254950
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.5789775771026246e-05,
+      "loss": 1.0449,
+      "step": 254960
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.5788627841126496e-05,
+      "loss": 1.0017,
+      "step": 254970
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.578747991122675e-05,
+      "loss": 0.9293,
+      "step": 254980
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.5786331981327e-05,
+      "loss": 0.9901,
+      "step": 254990
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.578518405142726e-05,
+      "loss": 0.9883,
+      "step": 255000
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.578403612152751e-05,
+      "loss": 1.2018,
+      "step": 255010
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.578288819162776e-05,
+      "loss": 1.0842,
+      "step": 255020
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.5781740261728016e-05,
+      "loss": 1.0964,
+      "step": 255030
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.578059233182827e-05,
+      "loss": 0.9989,
+      "step": 255040
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.5779444401928515e-05,
+      "loss": 1.1622,
+      "step": 255050
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.577829647202877e-05,
+      "loss": 1.0182,
+      "step": 255060
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.577714854212903e-05,
+      "loss": 1.0532,
+      "step": 255070
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.577600061222927e-05,
+      "loss": 1.0151,
+      "step": 255080
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.577485268232953e-05,
+      "loss": 1.1009,
+      "step": 255090
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.5773704752429785e-05,
+      "loss": 1.0084,
+      "step": 255100
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.5772556822530035e-05,
+      "loss": 0.9707,
+      "step": 255110
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.5771408892630285e-05,
+      "loss": 0.9169,
+      "step": 255120
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.5770260962730535e-05,
+      "loss": 1.2157,
+      "step": 255130
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.576911303283079e-05,
+      "loss": 1.1515,
+      "step": 255140
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.576796510293105e-05,
+      "loss": 1.0997,
+      "step": 255150
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.576681717303129e-05,
+      "loss": 0.9251,
+      "step": 255160
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.576566924313155e-05,
+      "loss": 1.1803,
+      "step": 255170
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.5764521313231804e-05,
+      "loss": 1.0774,
+      "step": 255180
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.576337338333205e-05,
+      "loss": 0.981,
+      "step": 255190
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.5762225453432304e-05,
+      "loss": 0.9913,
+      "step": 255200
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.576107752353256e-05,
+      "loss": 1.0272,
+      "step": 255210
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.575992959363281e-05,
+      "loss": 1.0728,
+      "step": 255220
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.575878166373306e-05,
+      "loss": 1.2074,
+      "step": 255230
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.575763373383332e-05,
+      "loss": 1.1519,
+      "step": 255240
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.575648580393357e-05,
+      "loss": 1.0627,
+      "step": 255250
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.575533787403382e-05,
+      "loss": 0.9331,
+      "step": 255260
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.575418994413407e-05,
+      "loss": 1.1542,
+      "step": 255270
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.575304201423432e-05,
+      "loss": 1.0741,
+      "step": 255280
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.575189408433458e-05,
+      "loss": 1.0109,
+      "step": 255290
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.5750746154434836e-05,
+      "loss": 1.0336,
+      "step": 255300
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.574959822453508e-05,
+      "loss": 1.0326,
+      "step": 255310
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.5748450294635336e-05,
+      "loss": 1.0193,
+      "step": 255320
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.574730236473559e-05,
+      "loss": 1.1488,
+      "step": 255330
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.5746154434835836e-05,
+      "loss": 1.0348,
+      "step": 255340
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.574500650493609e-05,
+      "loss": 1.0121,
+      "step": 255350
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.574385857503635e-05,
+      "loss": 0.9838,
+      "step": 255360
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.57427106451366e-05,
+      "loss": 1.0782,
+      "step": 255370
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.574156271523685e-05,
+      "loss": 1.0501,
+      "step": 255380
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.5740414785337105e-05,
+      "loss": 0.9524,
+      "step": 255390
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.5739266855437355e-05,
+      "loss": 0.9982,
+      "step": 255400
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.573811892553761e-05,
+      "loss": 1.0498,
+      "step": 255410
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.573697099563786e-05,
+      "loss": 1.0903,
+      "step": 255420
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.573582306573811e-05,
+      "loss": 1.0736,
+      "step": 255430
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.573467513583837e-05,
+      "loss": 1.0839,
+      "step": 255440
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.5733527205938625e-05,
+      "loss": 1.0933,
+      "step": 255450
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.573237927603887e-05,
+      "loss": 0.9801,
+      "step": 255460
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.5731231346139124e-05,
+      "loss": 1.2148,
+      "step": 255470
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.573008341623938e-05,
+      "loss": 0.9751,
+      "step": 255480
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.5728935486339624e-05,
+      "loss": 1.0573,
+      "step": 255490
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.572778755643988e-05,
+      "loss": 1.0923,
+      "step": 255500
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.572663962654014e-05,
+      "loss": 1.0293,
+      "step": 255510
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.572549169664039e-05,
+      "loss": 0.9959,
+      "step": 255520
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.572434376674064e-05,
+      "loss": 1.0267,
+      "step": 255530
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.5723195836840894e-05,
+      "loss": 1.0451,
+      "step": 255540
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.5722047906941144e-05,
+      "loss": 1.0003,
+      "step": 255550
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.57208999770414e-05,
+      "loss": 0.9738,
+      "step": 255560
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.571975204714165e-05,
+      "loss": 0.9986,
+      "step": 255570
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.57186041172419e-05,
+      "loss": 1.155,
+      "step": 255580
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.5717456187342157e-05,
+      "loss": 1.046,
+      "step": 255590
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.571630825744241e-05,
+      "loss": 1.1208,
+      "step": 255600
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.5715160327542656e-05,
+      "loss": 0.9799,
+      "step": 255610
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.571401239764291e-05,
+      "loss": 0.9625,
+      "step": 255620
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.571286446774317e-05,
+      "loss": 0.956,
+      "step": 255630
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.571171653784341e-05,
+      "loss": 1.1627,
+      "step": 255640
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.571056860794367e-05,
+      "loss": 1.0064,
+      "step": 255650
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.5709420678043926e-05,
+      "loss": 1.0291,
+      "step": 255660
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.5708272748144176e-05,
+      "loss": 0.984,
+      "step": 255670
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.5707124818244426e-05,
+      "loss": 1.0443,
+      "step": 255680
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.570597688834468e-05,
+      "loss": 1.0137,
+      "step": 255690
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.570482895844493e-05,
+      "loss": 1.0067,
+      "step": 255700
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.570368102854519e-05,
+      "loss": 1.1482,
+      "step": 255710
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.570253309864544e-05,
+      "loss": 1.166,
+      "step": 255720
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.570138516874569e-05,
+      "loss": 0.9976,
+      "step": 255730
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.5700237238845945e-05,
+      "loss": 1.1337,
+      "step": 255740
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.56990893089462e-05,
+      "loss": 0.9475,
+      "step": 255750
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.5697941379046445e-05,
+      "loss": 1.0098,
+      "step": 255760
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.56967934491467e-05,
+      "loss": 1.0882,
+      "step": 255770
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.569564551924696e-05,
+      "loss": 1.0921,
+      "step": 255780
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.56944975893472e-05,
+      "loss": 1.011,
+      "step": 255790
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.569334965944746e-05,
+      "loss": 1.0448,
+      "step": 255800
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.5692201729547714e-05,
+      "loss": 0.9783,
+      "step": 255810
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.5691053799647964e-05,
+      "loss": 1.0365,
+      "step": 255820
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.5689905869748214e-05,
+      "loss": 1.0041,
+      "step": 255830
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.568875793984847e-05,
+      "loss": 1.018,
+      "step": 255840
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.568761000994872e-05,
+      "loss": 1.0811,
+      "step": 255850
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.568646208004898e-05,
+      "loss": 1.0623,
+      "step": 255860
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.568531415014923e-05,
+      "loss": 1.0403,
+      "step": 255870
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.568416622024948e-05,
+      "loss": 1.0122,
+      "step": 255880
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.5683018290349734e-05,
+      "loss": 1.0325,
+      "step": 255890
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.568187036044999e-05,
+      "loss": 1.0705,
+      "step": 255900
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.568072243055023e-05,
+      "loss": 0.9657,
+      "step": 255910
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.567957450065049e-05,
+      "loss": 0.9981,
+      "step": 255920
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.5678426570750747e-05,
+      "loss": 1.0211,
+      "step": 255930
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.567727864085099e-05,
+      "loss": 1.0492,
+      "step": 255940
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.5676130710951246e-05,
+      "loss": 0.9406,
+      "step": 255950
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.56749827810515e-05,
+      "loss": 1.1258,
+      "step": 255960
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.567383485115175e-05,
+      "loss": 1.056,
+      "step": 255970
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.5672686921252e-05,
+      "loss": 0.9942,
+      "step": 255980
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.567153899135226e-05,
+      "loss": 0.9437,
+      "step": 255990
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.567039106145251e-05,
+      "loss": 1.0632,
+      "step": 256000
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.5669243131552766e-05,
+      "loss": 1.0209,
+      "step": 256010
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.5668095201653016e-05,
+      "loss": 1.1457,
+      "step": 256020
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.5666947271753265e-05,
+      "loss": 0.9988,
+      "step": 256030
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.566579934185352e-05,
+      "loss": 0.9945,
+      "step": 256040
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.566465141195378e-05,
+      "loss": 0.9859,
+      "step": 256050
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.566350348205402e-05,
+      "loss": 1.0692,
+      "step": 256060
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.566235555215428e-05,
+      "loss": 1.0228,
+      "step": 256070
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.5661207622254535e-05,
+      "loss": 1.1108,
+      "step": 256080
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.566005969235478e-05,
+      "loss": 0.832,
+      "step": 256090
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.5658911762455035e-05,
+      "loss": 0.9647,
+      "step": 256100
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.565776383255529e-05,
+      "loss": 1.0468,
+      "step": 256110
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.565661590265554e-05,
+      "loss": 1.0327,
+      "step": 256120
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.565546797275579e-05,
+      "loss": 1.0069,
+      "step": 256130
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.565432004285605e-05,
+      "loss": 1.0415,
+      "step": 256140
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.56531721129563e-05,
+      "loss": 0.9596,
+      "step": 256150
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.5652024183056554e-05,
+      "loss": 1.0271,
+      "step": 256160
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.5650876253156804e-05,
+      "loss": 1.0682,
+      "step": 256170
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.5649728323257054e-05,
+      "loss": 1.0606,
+      "step": 256180
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.564858039335731e-05,
+      "loss": 1.0839,
+      "step": 256190
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.564743246345757e-05,
+      "loss": 1.0385,
+      "step": 256200
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.564628453355781e-05,
+      "loss": 1.0207,
+      "step": 256210
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.564513660365807e-05,
+      "loss": 1.1056,
+      "step": 256220
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.5643988673758323e-05,
+      "loss": 1.0783,
+      "step": 256230
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.5642840743858567e-05,
+      "loss": 1.2255,
+      "step": 256240
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.564169281395882e-05,
+      "loss": 1.0023,
+      "step": 256250
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.564054488405908e-05,
+      "loss": 1.0321,
+      "step": 256260
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.563939695415933e-05,
+      "loss": 1.0961,
+      "step": 256270
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.563824902425958e-05,
+      "loss": 1.0846,
+      "step": 256280
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.5637101094359836e-05,
+      "loss": 1.0965,
+      "step": 256290
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.5635953164460086e-05,
+      "loss": 1.046,
+      "step": 256300
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.563480523456034e-05,
+      "loss": 1.0905,
+      "step": 256310
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.563365730466059e-05,
+      "loss": 1.0271,
+      "step": 256320
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.563250937476084e-05,
+      "loss": 1.0612,
+      "step": 256330
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.56313614448611e-05,
+      "loss": 0.9429,
+      "step": 256340
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.5630213514961356e-05,
+      "loss": 1.0271,
+      "step": 256350
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.56290655850616e-05,
+      "loss": 1.1741,
+      "step": 256360
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.5627917655161855e-05,
+      "loss": 0.9952,
+      "step": 256370
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.562676972526211e-05,
+      "loss": 1.0529,
+      "step": 256380
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.5625621795362355e-05,
+      "loss": 1.0777,
+      "step": 256390
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.562447386546261e-05,
+      "loss": 1.0221,
+      "step": 256400
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.562332593556287e-05,
+      "loss": 1.0355,
+      "step": 256410
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.562217800566312e-05,
+      "loss": 1.0932,
+      "step": 256420
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.562103007576337e-05,
+      "loss": 1.0378,
+      "step": 256430
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.5619882145863625e-05,
+      "loss": 1.0908,
+      "step": 256440
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.5618734215963874e-05,
+      "loss": 1.1119,
+      "step": 256450
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.561758628606413e-05,
+      "loss": 1.0001,
+      "step": 256460
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.561643835616438e-05,
+      "loss": 1.0268,
+      "step": 256470
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.561529042626463e-05,
+      "loss": 1.0581,
+      "step": 256480
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.561414249636489e-05,
+      "loss": 1.0437,
+      "step": 256490
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.561299456646513e-05,
+      "loss": 1.1241,
+      "step": 256500
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.561184663656539e-05,
+      "loss": 1.064,
+      "step": 256510
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.5610698706665644e-05,
+      "loss": 1.1284,
+      "step": 256520
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.5609550776765894e-05,
+      "loss": 1.0215,
+      "step": 256530
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.5608402846866143e-05,
+      "loss": 1.0446,
+      "step": 256540
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.56072549169664e-05,
+      "loss": 1.0323,
+      "step": 256550
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.560610698706665e-05,
+      "loss": 0.9837,
+      "step": 256560
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.5604959057166907e-05,
+      "loss": 0.9792,
+      "step": 256570
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.5603811127267156e-05,
+      "loss": 1.0662,
+      "step": 256580
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.5602663197367406e-05,
+      "loss": 1.0231,
+      "step": 256590
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.560151526746766e-05,
+      "loss": 0.9971,
+      "step": 256600
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.560036733756792e-05,
+      "loss": 1.0408,
+      "step": 256610
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.559921940766816e-05,
+      "loss": 1.0365,
+      "step": 256620
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.559807147776842e-05,
+      "loss": 1.144,
+      "step": 256630
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.5596923547868676e-05,
+      "loss": 1.0078,
+      "step": 256640
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.559577561796892e-05,
+      "loss": 1.1315,
+      "step": 256650
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.5594627688069176e-05,
+      "loss": 1.0009,
+      "step": 256660
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.559347975816943e-05,
+      "loss": 1.0459,
+      "step": 256670
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.559233182826968e-05,
+      "loss": 1.0395,
+      "step": 256680
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.559118389836993e-05,
+      "loss": 0.9857,
+      "step": 256690
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.559003596847019e-05,
+      "loss": 1.1211,
+      "step": 256700
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.558888803857044e-05,
+      "loss": 0.9844,
+      "step": 256710
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.5587740108670695e-05,
+      "loss": 1.1078,
+      "step": 256720
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.5586592178770945e-05,
+      "loss": 0.9771,
+      "step": 256730
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.5585444248871195e-05,
+      "loss": 1.0248,
+      "step": 256740
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.558429631897145e-05,
+      "loss": 1.0361,
+      "step": 256750
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.558314838907171e-05,
+      "loss": 1.1092,
+      "step": 256760
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.558200045917195e-05,
+      "loss": 1.0986,
+      "step": 256770
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.558085252927221e-05,
+      "loss": 1.0806,
+      "step": 256780
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.5579704599372464e-05,
+      "loss": 1.0519,
+      "step": 256790
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.557855666947271e-05,
+      "loss": 1.0203,
+      "step": 256800
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.5577408739572964e-05,
+      "loss": 1.1296,
+      "step": 256810
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.557626080967322e-05,
+      "loss": 1.0654,
+      "step": 256820
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.557511287977347e-05,
+      "loss": 1.1133,
+      "step": 256830
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.557396494987372e-05,
+      "loss": 1.0838,
+      "step": 256840
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.557281701997398e-05,
+      "loss": 1.0632,
+      "step": 256850
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.557166909007423e-05,
+      "loss": 1.1128,
+      "step": 256860
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.5570521160174484e-05,
+      "loss": 1.0176,
+      "step": 256870
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.5569373230274733e-05,
+      "loss": 0.994,
+      "step": 256880
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.556822530037498e-05,
+      "loss": 1.0577,
+      "step": 256890
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.556707737047524e-05,
+      "loss": 1.0375,
+      "step": 256900
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.5565929440575497e-05,
+      "loss": 1.0099,
+      "step": 256910
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.556478151067574e-05,
+      "loss": 1.0114,
+      "step": 256920
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.5563633580775996e-05,
+      "loss": 1.0793,
+      "step": 256930
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.556248565087625e-05,
+      "loss": 0.9801,
+      "step": 256940
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.5561337720976496e-05,
+      "loss": 1.0429,
+      "step": 256950
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.556018979107675e-05,
+      "loss": 1.0484,
+      "step": 256960
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.555904186117701e-05,
+      "loss": 0.9753,
+      "step": 256970
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.555789393127726e-05,
+      "loss": 1.0172,
+      "step": 256980
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.555674600137751e-05,
+      "loss": 1.1325,
+      "step": 256990
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.5555598071477766e-05,
+      "loss": 1.0963,
+      "step": 257000
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.5554450141578015e-05,
+      "loss": 1.0215,
+      "step": 257010
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.555330221167827e-05,
+      "loss": 1.1486,
+      "step": 257020
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.555215428177852e-05,
+      "loss": 1.0454,
+      "step": 257030
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.555100635187877e-05,
+      "loss": 0.9942,
+      "step": 257040
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.554985842197903e-05,
+      "loss": 1.1702,
+      "step": 257050
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.5548710492079285e-05,
+      "loss": 1.0446,
+      "step": 257060
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.554756256217953e-05,
+      "loss": 1.0661,
+      "step": 257070
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.5546414632279785e-05,
+      "loss": 1.0648,
+      "step": 257080
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.554526670238004e-05,
+      "loss": 0.9938,
+      "step": 257090
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.5544118772480284e-05,
+      "loss": 1.0351,
+      "step": 257100
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.554297084258054e-05,
+      "loss": 0.967,
+      "step": 257110
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.55418229126808e-05,
+      "loss": 1.0646,
+      "step": 257120
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.554067498278105e-05,
+      "loss": 1.1223,
+      "step": 257130
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.55395270528813e-05,
+      "loss": 1.0714,
+      "step": 257140
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.5538379122981554e-05,
+      "loss": 1.1136,
+      "step": 257150
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.5537231193081804e-05,
+      "loss": 1.0222,
+      "step": 257160
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.553608326318206e-05,
+      "loss": 0.9798,
+      "step": 257170
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.553493533328231e-05,
+      "loss": 1.0295,
+      "step": 257180
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.553378740338256e-05,
+      "loss": 1.1205,
+      "step": 257190
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.553263947348282e-05,
+      "loss": 0.9943,
+      "step": 257200
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.5531491543583073e-05,
+      "loss": 0.9994,
+      "step": 257210
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.5530343613683317e-05,
+      "loss": 1.1421,
+      "step": 257220
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.552919568378357e-05,
+      "loss": 1.0239,
+      "step": 257230
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.552804775388383e-05,
+      "loss": 0.9439,
+      "step": 257240
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.552689982398407e-05,
+      "loss": 0.973,
+      "step": 257250
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.552575189408433e-05,
+      "loss": 1.021,
+      "step": 257260
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.5524603964184586e-05,
+      "loss": 1.0979,
+      "step": 257270
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.5523456034284836e-05,
+      "loss": 1.0187,
+      "step": 257280
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.5522308104385086e-05,
+      "loss": 0.962,
+      "step": 257290
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.552116017448534e-05,
+      "loss": 1.07,
+      "step": 257300
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.552001224458559e-05,
+      "loss": 1.0229,
+      "step": 257310
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.551886431468585e-05,
+      "loss": 1.0979,
+      "step": 257320
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.55177163847861e-05,
+      "loss": 1.0534,
+      "step": 257330
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.551656845488635e-05,
+      "loss": 1.1169,
+      "step": 257340
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.5515420524986605e-05,
+      "loss": 0.9651,
+      "step": 257350
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.551427259508686e-05,
+      "loss": 0.9769,
+      "step": 257360
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.5513124665187105e-05,
+      "loss": 1.0335,
+      "step": 257370
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.551197673528736e-05,
+      "loss": 0.9916,
+      "step": 257380
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.551082880538762e-05,
+      "loss": 0.9184,
+      "step": 257390
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.550968087548786e-05,
+      "loss": 1.1313,
+      "step": 257400
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.550853294558812e-05,
+      "loss": 1.0875,
+      "step": 257410
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.5507385015688375e-05,
+      "loss": 0.9923,
+      "step": 257420
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.5506237085788624e-05,
+      "loss": 1.0707,
+      "step": 257430
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.5505089155888874e-05,
+      "loss": 1.0884,
+      "step": 257440
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.550394122598913e-05,
+      "loss": 0.9382,
+      "step": 257450
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.550279329608938e-05,
+      "loss": 1.0884,
+      "step": 257460
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.550164536618964e-05,
+      "loss": 1.0401,
+      "step": 257470
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.550049743628989e-05,
+      "loss": 0.9899,
+      "step": 257480
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.549934950639014e-05,
+      "loss": 1.0137,
+      "step": 257490
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.5498201576490394e-05,
+      "loss": 0.9429,
+      "step": 257500
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.549705364659065e-05,
+      "loss": 1.1333,
+      "step": 257510
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.5495905716690893e-05,
+      "loss": 0.9414,
+      "step": 257520
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.549475778679115e-05,
+      "loss": 1.1285,
+      "step": 257530
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.549360985689141e-05,
+      "loss": 0.9988,
+      "step": 257540
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.549246192699165e-05,
+      "loss": 0.9779,
+      "step": 257550
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.5491313997091906e-05,
+      "loss": 1.0817,
+      "step": 257560
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.549016606719216e-05,
+      "loss": 0.9456,
+      "step": 257570
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.548901813729241e-05,
+      "loss": 0.9963,
+      "step": 257580
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.548787020739266e-05,
+      "loss": 1.091,
+      "step": 257590
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.548672227749292e-05,
+      "loss": 1.0789,
+      "step": 257600
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.548557434759317e-05,
+      "loss": 1.024,
+      "step": 257610
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.5484426417693426e-05,
+      "loss": 1.0667,
+      "step": 257620
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.5483278487793676e-05,
+      "loss": 1.1304,
+      "step": 257630
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.5482130557893926e-05,
+      "loss": 1.1132,
+      "step": 257640
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.548098262799418e-05,
+      "loss": 1.105,
+      "step": 257650
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.547983469809444e-05,
+      "loss": 1.1383,
+      "step": 257660
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.547868676819468e-05,
+      "loss": 1.0109,
+      "step": 257670
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.547753883829494e-05,
+      "loss": 1.1096,
+      "step": 257680
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.5476390908395195e-05,
+      "loss": 1.159,
+      "step": 257690
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.547524297849544e-05,
+      "loss": 1.1791,
+      "step": 257700
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.5474095048595695e-05,
+      "loss": 1.0972,
+      "step": 257710
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.547294711869595e-05,
+      "loss": 1.1347,
+      "step": 257720
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.54717991887962e-05,
+      "loss": 0.9811,
+      "step": 257730
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.547065125889645e-05,
+      "loss": 1.0294,
+      "step": 257740
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.546950332899671e-05,
+      "loss": 1.0871,
+      "step": 257750
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.546835539909696e-05,
+      "loss": 1.2746,
+      "step": 257760
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.5467207469197214e-05,
+      "loss": 1.0872,
+      "step": 257770
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.5466059539297464e-05,
+      "loss": 1.1138,
+      "step": 257780
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.5464911609397714e-05,
+      "loss": 1.1537,
+      "step": 257790
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.546376367949797e-05,
+      "loss": 0.9993,
+      "step": 257800
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.546261574959823e-05,
+      "loss": 1.129,
+      "step": 257810
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.546146781969847e-05,
+      "loss": 1.1321,
+      "step": 257820
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.546031988979873e-05,
+      "loss": 1.0744,
+      "step": 257830
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.5459171959898984e-05,
+      "loss": 1.126,
+      "step": 257840
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.545802402999923e-05,
+      "loss": 1.0593,
+      "step": 257850
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.5456876100099483e-05,
+      "loss": 1.0283,
+      "step": 257860
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.545572817019973e-05,
+      "loss": 1.1042,
+      "step": 257870
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.545458024029999e-05,
+      "loss": 1.0285,
+      "step": 257880
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.545343231040024e-05,
+      "loss": 1.0012,
+      "step": 257890
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.545228438050049e-05,
+      "loss": 1.0665,
+      "step": 257900
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.5451136450600746e-05,
+      "loss": 1.0425,
+      "step": 257910
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.5449988520701e-05,
+      "loss": 1.017,
+      "step": 257920
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.5448840590801246e-05,
+      "loss": 1.0297,
+      "step": 257930
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.54476926609015e-05,
+      "loss": 1.1285,
+      "step": 257940
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.544654473100176e-05,
+      "loss": 1.1795,
+      "step": 257950
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.5445396801102e-05,
+      "loss": 0.9777,
+      "step": 257960
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.544424887120226e-05,
+      "loss": 1.1124,
+      "step": 257970
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.5443100941302516e-05,
+      "loss": 0.9926,
+      "step": 257980
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.5441953011402765e-05,
+      "loss": 1.0357,
+      "step": 257990
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.5440805081503015e-05,
+      "loss": 1.0826,
+      "step": 258000
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.543965715160327e-05,
+      "loss": 1.081,
+      "step": 258010
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.543850922170352e-05,
+      "loss": 1.0007,
+      "step": 258020
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.543736129180378e-05,
+      "loss": 0.9648,
+      "step": 258030
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.543621336190403e-05,
+      "loss": 1.0577,
+      "step": 258040
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.543506543200428e-05,
+      "loss": 0.9858,
+      "step": 258050
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.5433917502104535e-05,
+      "loss": 1.0258,
+      "step": 258060
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.543276957220479e-05,
+      "loss": 0.9319,
+      "step": 258070
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.5431621642305034e-05,
+      "loss": 1.0727,
+      "step": 258080
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.543047371240529e-05,
+      "loss": 1.0527,
+      "step": 258090
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.542932578250555e-05,
+      "loss": 1.0911,
+      "step": 258100
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.542817785260579e-05,
+      "loss": 1.066,
+      "step": 258110
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.542702992270605e-05,
+      "loss": 0.9683,
+      "step": 258120
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.5425881992806304e-05,
+      "loss": 1.0205,
+      "step": 258130
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.5424734062906554e-05,
+      "loss": 0.9712,
+      "step": 258140
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.5423586133006804e-05,
+      "loss": 1.02,
+      "step": 258150
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.542243820310706e-05,
+      "loss": 1.0753,
+      "step": 258160
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.542129027320731e-05,
+      "loss": 1.1022,
+      "step": 258170
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.542014234330757e-05,
+      "loss": 1.0711,
+      "step": 258180
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.541899441340782e-05,
+      "loss": 1.1244,
+      "step": 258190
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.5417846483508067e-05,
+      "loss": 1.0865,
+      "step": 258200
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.541669855360832e-05,
+      "loss": 1.1226,
+      "step": 258210
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.541555062370858e-05,
+      "loss": 1.0714,
+      "step": 258220
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.541440269380882e-05,
+      "loss": 1.0503,
+      "step": 258230
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.541325476390908e-05,
+      "loss": 1.0365,
+      "step": 258240
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.5412106834009336e-05,
+      "loss": 1.0116,
+      "step": 258250
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.541095890410958e-05,
+      "loss": 1.0188,
+      "step": 258260
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.5409810974209836e-05,
+      "loss": 1.1411,
+      "step": 258270
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.540866304431009e-05,
+      "loss": 0.9956,
+      "step": 258280
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.540751511441034e-05,
+      "loss": 1.078,
+      "step": 258290
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.540636718451059e-05,
+      "loss": 1.1081,
+      "step": 258300
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.540521925461085e-05,
+      "loss": 1.0493,
+      "step": 258310
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.54040713247111e-05,
+      "loss": 0.9937,
+      "step": 258320
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.5402923394811355e-05,
+      "loss": 1.0699,
+      "step": 258330
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.5401775464911605e-05,
+      "loss": 1.0532,
+      "step": 258340
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.5400627535011855e-05,
+      "loss": 0.9788,
+      "step": 258350
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.539947960511211e-05,
+      "loss": 1.0103,
+      "step": 258360
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.539833167521237e-05,
+      "loss": 1.0596,
+      "step": 258370
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.539718374531261e-05,
+      "loss": 1.2187,
+      "step": 258380
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.539603581541287e-05,
+      "loss": 1.1114,
+      "step": 258390
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.5394887885513125e-05,
+      "loss": 1.1037,
+      "step": 258400
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.539373995561337e-05,
+      "loss": 1.0953,
+      "step": 258410
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.5392592025713624e-05,
+      "loss": 1.0051,
+      "step": 258420
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.539144409581388e-05,
+      "loss": 1.0421,
+      "step": 258430
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.539029616591413e-05,
+      "loss": 0.996,
+      "step": 258440
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.538914823601438e-05,
+      "loss": 1.0733,
+      "step": 258450
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.538800030611464e-05,
+      "loss": 0.9992,
+      "step": 258460
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.538685237621489e-05,
+      "loss": 1.2851,
+      "step": 258470
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.5385704446315144e-05,
+      "loss": 1.042,
+      "step": 258480
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.5384556516415394e-05,
+      "loss": 1.0116,
+      "step": 258490
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.5383408586515643e-05,
+      "loss": 0.9794,
+      "step": 258500
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.53822606566159e-05,
+      "loss": 1.0542,
+      "step": 258510
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.538111272671616e-05,
+      "loss": 1.0695,
+      "step": 258520
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.53799647968164e-05,
+      "loss": 1.0513,
+      "step": 258530
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.5378816866916656e-05,
+      "loss": 1.0194,
+      "step": 258540
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.537766893701691e-05,
+      "loss": 0.8637,
+      "step": 258550
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.5376521007117156e-05,
+      "loss": 1.0441,
+      "step": 258560
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.537537307721741e-05,
+      "loss": 0.9992,
+      "step": 258570
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.537422514731767e-05,
+      "loss": 1.152,
+      "step": 258580
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.537307721741792e-05,
+      "loss": 1.2606,
+      "step": 258590
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.537192928751817e-05,
+      "loss": 0.9999,
+      "step": 258600
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.5370781357618426e-05,
+      "loss": 0.9847,
+      "step": 258610
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.5369633427718676e-05,
+      "loss": 1.1085,
+      "step": 258620
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.536848549781893e-05,
+      "loss": 1.0006,
+      "step": 258630
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.536733756791918e-05,
+      "loss": 0.9944,
+      "step": 258640
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.536618963801943e-05,
+      "loss": 1.0677,
+      "step": 258650
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.536504170811969e-05,
+      "loss": 1.0913,
+      "step": 258660
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.5363893778219945e-05,
+      "loss": 0.909,
+      "step": 258670
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.536274584832019e-05,
+      "loss": 1.0613,
+      "step": 258680
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.5361597918420445e-05,
+      "loss": 1.0188,
+      "step": 258690
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.53604499885207e-05,
+      "loss": 0.9707,
+      "step": 258700
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.5359302058620945e-05,
+      "loss": 1.0831,
+      "step": 258710
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.53581541287212e-05,
+      "loss": 1.1941,
+      "step": 258720
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.535700619882146e-05,
+      "loss": 0.9133,
+      "step": 258730
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.535585826892171e-05,
+      "loss": 1.1034,
+      "step": 258740
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.535471033902196e-05,
+      "loss": 1.0003,
+      "step": 258750
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.5353562409122214e-05,
+      "loss": 1.0137,
+      "step": 258760
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.5352414479222464e-05,
+      "loss": 1.008,
+      "step": 258770
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.535126654932272e-05,
+      "loss": 1.0911,
+      "step": 258780
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.535011861942297e-05,
+      "loss": 1.1188,
+      "step": 258790
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.534897068952322e-05,
+      "loss": 1.1209,
+      "step": 258800
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.534782275962348e-05,
+      "loss": 1.1077,
+      "step": 258810
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.5346674829723734e-05,
+      "loss": 0.9583,
+      "step": 258820
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.534552689982398e-05,
+      "loss": 1.0841,
+      "step": 258830
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.5344378969924233e-05,
+      "loss": 1.0061,
+      "step": 258840
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.534323104002449e-05,
+      "loss": 1.2001,
+      "step": 258850
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.534208311012473e-05,
+      "loss": 1.0226,
+      "step": 258860
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.534093518022499e-05,
+      "loss": 0.9765,
+      "step": 258870
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.5339787250325246e-05,
+      "loss": 1.0402,
+      "step": 258880
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.5338639320425496e-05,
+      "loss": 1.0671,
+      "step": 258890
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.5337491390525746e-05,
+      "loss": 1.0377,
+      "step": 258900
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.5336343460626e-05,
+      "loss": 1.0411,
+      "step": 258910
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.533519553072625e-05,
+      "loss": 1.1209,
+      "step": 258920
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.533404760082651e-05,
+      "loss": 1.0677,
+      "step": 258930
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.533289967092676e-05,
+      "loss": 1.1141,
+      "step": 258940
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.533175174102701e-05,
+      "loss": 0.9955,
+      "step": 258950
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.5330603811127266e-05,
+      "loss": 0.9779,
+      "step": 258960
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.532945588122752e-05,
+      "loss": 1.0091,
+      "step": 258970
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.5328307951327765e-05,
+      "loss": 1.043,
+      "step": 258980
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.532716002142802e-05,
+      "loss": 1.0586,
+      "step": 258990
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.532601209152828e-05,
+      "loss": 1.0314,
+      "step": 259000
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.532486416162852e-05,
+      "loss": 1.0333,
+      "step": 259010
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.532371623172878e-05,
+      "loss": 0.9794,
+      "step": 259020
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.5322568301829035e-05,
+      "loss": 1.0175,
+      "step": 259030
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.5321420371929285e-05,
+      "loss": 1.0476,
+      "step": 259040
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.5320272442029535e-05,
+      "loss": 1.0444,
+      "step": 259050
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.531912451212979e-05,
+      "loss": 1.0784,
+      "step": 259060
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.531797658223004e-05,
+      "loss": 1.0531,
+      "step": 259070
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.53168286523303e-05,
+      "loss": 1.1385,
+      "step": 259080
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.531568072243055e-05,
+      "loss": 0.9896,
+      "step": 259090
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.53145327925308e-05,
+      "loss": 0.9849,
+      "step": 259100
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.5313384862631054e-05,
+      "loss": 1.0712,
+      "step": 259110
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.531223693273131e-05,
+      "loss": 1.0144,
+      "step": 259120
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.5311089002831554e-05,
+      "loss": 1.0822,
+      "step": 259130
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.530994107293181e-05,
+      "loss": 1.0765,
+      "step": 259140
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.530879314303207e-05,
+      "loss": 0.9486,
+      "step": 259150
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.530764521313231e-05,
+      "loss": 1.0499,
+      "step": 259160
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.530649728323257e-05,
+      "loss": 1.085,
+      "step": 259170
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.530534935333282e-05,
+      "loss": 1.105,
+      "step": 259180
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.530420142343307e-05,
+      "loss": 1.0176,
+      "step": 259190
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.530305349353332e-05,
+      "loss": 0.9672,
+      "step": 259200
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.530190556363358e-05,
+      "loss": 1.0736,
+      "step": 259210
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.530075763373383e-05,
+      "loss": 1.0576,
+      "step": 259220
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.5299609703834086e-05,
+      "loss": 1.0938,
+      "step": 259230
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.529846177393433e-05,
+      "loss": 0.998,
+      "step": 259240
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.5297313844034586e-05,
+      "loss": 0.9673,
+      "step": 259250
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.529616591413484e-05,
+      "loss": 0.9772,
+      "step": 259260
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.5295017984235086e-05,
+      "loss": 1.0168,
+      "step": 259270
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.529387005433534e-05,
+      "loss": 1.0688,
+      "step": 259280
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.52927221244356e-05,
+      "loss": 1.1492,
+      "step": 259290
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.529157419453585e-05,
+      "loss": 0.9606,
+      "step": 259300
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.52904262646361e-05,
+      "loss": 1.1677,
+      "step": 259310
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.5289278334736355e-05,
+      "loss": 1.1749,
+      "step": 259320
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.5288130404836605e-05,
+      "loss": 1.0072,
+      "step": 259330
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.528698247493686e-05,
+      "loss": 1.0571,
+      "step": 259340
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.528583454503711e-05,
+      "loss": 0.9492,
+      "step": 259350
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.528468661513736e-05,
+      "loss": 1.1159,
+      "step": 259360
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.528353868523762e-05,
+      "loss": 1.1984,
+      "step": 259370
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.5282390755337875e-05,
+      "loss": 1.0224,
+      "step": 259380
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.528124282543812e-05,
+      "loss": 1.0285,
+      "step": 259390
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.5280094895538374e-05,
+      "loss": 1.1143,
+      "step": 259400
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.527894696563863e-05,
+      "loss": 1.0755,
+      "step": 259410
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.5277799035738874e-05,
+      "loss": 0.941,
+      "step": 259420
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.527665110583913e-05,
+      "loss": 1.0101,
+      "step": 259430
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.527550317593939e-05,
+      "loss": 1.034,
+      "step": 259440
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.527435524603964e-05,
+      "loss": 1.0467,
+      "step": 259450
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.527320731613989e-05,
+      "loss": 1.0266,
+      "step": 259460
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.5272059386240144e-05,
+      "loss": 1.0672,
+      "step": 259470
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.5270911456340393e-05,
+      "loss": 1.1003,
+      "step": 259480
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.526976352644065e-05,
+      "loss": 1.0156,
+      "step": 259490
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.52686155965409e-05,
+      "loss": 1.0983,
+      "step": 259500
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.526746766664115e-05,
+      "loss": 1.1214,
+      "step": 259510
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.5266319736741406e-05,
+      "loss": 1.0694,
+      "step": 259520
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.526517180684166e-05,
+      "loss": 1.0364,
+      "step": 259530
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.5264023876941906e-05,
+      "loss": 1.0232,
+      "step": 259540
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.526287594704216e-05,
+      "loss": 0.9869,
+      "step": 259550
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.526172801714242e-05,
+      "loss": 1.0525,
+      "step": 259560
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.526058008724266e-05,
+      "loss": 1.0645,
+      "step": 259570
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.525943215734292e-05,
+      "loss": 1.1105,
+      "step": 259580
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.5258284227443176e-05,
+      "loss": 0.9472,
+      "step": 259590
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.5257136297543426e-05,
+      "loss": 0.9818,
+      "step": 259600
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.5255988367643675e-05,
+      "loss": 1.0654,
+      "step": 259610
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.525484043774393e-05,
+      "loss": 1.0333,
+      "step": 259620
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.525369250784418e-05,
+      "loss": 1.0048,
+      "step": 259630
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.525254457794444e-05,
+      "loss": 1.0309,
+      "step": 259640
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.525139664804469e-05,
+      "loss": 1.0582,
+      "step": 259650
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.525024871814494e-05,
+      "loss": 1.1014,
+      "step": 259660
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.5249100788245195e-05,
+      "loss": 1.1139,
+      "step": 259670
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.524795285834545e-05,
+      "loss": 1.0882,
+      "step": 259680
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.5246804928445695e-05,
+      "loss": 1.1266,
+      "step": 259690
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.524565699854595e-05,
+      "loss": 1.1352,
+      "step": 259700
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.524450906864621e-05,
+      "loss": 0.9565,
+      "step": 259710
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.524336113874645e-05,
+      "loss": 0.8979,
+      "step": 259720
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.524221320884671e-05,
+      "loss": 0.9648,
+      "step": 259730
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.5241065278946964e-05,
+      "loss": 1.143,
+      "step": 259740
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.5239917349047214e-05,
+      "loss": 0.9069,
+      "step": 259750
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.5238769419147464e-05,
+      "loss": 1.0312,
+      "step": 259760
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.523762148924772e-05,
+      "loss": 1.1075,
+      "step": 259770
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.523647355934797e-05,
+      "loss": 1.0764,
+      "step": 259780
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.523532562944823e-05,
+      "loss": 0.9742,
+      "step": 259790
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.523417769954848e-05,
+      "loss": 0.9575,
+      "step": 259800
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.523302976964873e-05,
+      "loss": 1.0752,
+      "step": 259810
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.5231881839748983e-05,
+      "loss": 1.0161,
+      "step": 259820
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.523073390984924e-05,
+      "loss": 0.989,
+      "step": 259830
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.522958597994948e-05,
+      "loss": 1.0402,
+      "step": 259840
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.522843805004974e-05,
+      "loss": 1.0752,
+      "step": 259850
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.5227290120149996e-05,
+      "loss": 1.0592,
+      "step": 259860
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.522614219025024e-05,
+      "loss": 0.9446,
+      "step": 259870
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.5224994260350496e-05,
+      "loss": 0.9439,
+      "step": 259880
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.522384633045075e-05,
+      "loss": 0.9488,
+      "step": 259890
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.5222698400551e-05,
+      "loss": 1.0857,
+      "step": 259900
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.522155047065125e-05,
+      "loss": 1.0506,
+      "step": 259910
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.522040254075151e-05,
+      "loss": 0.9467,
+      "step": 259920
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.521925461085176e-05,
+      "loss": 1.0321,
+      "step": 259930
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.5218106680952016e-05,
+      "loss": 0.9998,
+      "step": 259940
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.5216958751052265e-05,
+      "loss": 1.1087,
+      "step": 259950
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.5215810821152515e-05,
+      "loss": 1.0745,
+      "step": 259960
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.521466289125277e-05,
+      "loss": 0.9686,
+      "step": 259970
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.521351496135303e-05,
+      "loss": 1.0507,
+      "step": 259980
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.521236703145327e-05,
+      "loss": 1.1183,
+      "step": 259990
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.521121910155353e-05,
+      "loss": 1.0978,
+      "step": 260000
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.5210071171653785e-05,
+      "loss": 1.0684,
+      "step": 260010
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.520892324175403e-05,
+      "loss": 1.0565,
+      "step": 260020
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.5207775311854285e-05,
+      "loss": 1.0206,
+      "step": 260030
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.520662738195454e-05,
+      "loss": 1.0967,
+      "step": 260040
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.520547945205479e-05,
+      "loss": 0.892,
+      "step": 260050
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.520433152215504e-05,
+      "loss": 1.0361,
+      "step": 260060
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.52031835922553e-05,
+      "loss": 1.1639,
+      "step": 260070
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.520203566235555e-05,
+      "loss": 1.2459,
+      "step": 260080
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.5200887732455804e-05,
+      "loss": 1.0366,
+      "step": 260090
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.5199739802556054e-05,
+      "loss": 0.9948,
+      "step": 260100
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.5198591872656304e-05,
+      "loss": 1.0671,
+      "step": 260110
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.519744394275656e-05,
+      "loss": 0.9824,
+      "step": 260120
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.519629601285682e-05,
+      "loss": 0.9835,
+      "step": 260130
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.519514808295706e-05,
+      "loss": 1.0547,
+      "step": 260140
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.519400015305732e-05,
+      "loss": 0.9961,
+      "step": 260150
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.519285222315757e-05,
+      "loss": 1.0266,
+      "step": 260160
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.5191704293257816e-05,
+      "loss": 1.0453,
+      "step": 260170
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.519055636335807e-05,
+      "loss": 1.0977,
+      "step": 260180
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.518940843345833e-05,
+      "loss": 1.0124,
+      "step": 260190
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.518826050355858e-05,
+      "loss": 1.1032,
+      "step": 260200
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.518711257365883e-05,
+      "loss": 0.9733,
+      "step": 260210
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.5185964643759086e-05,
+      "loss": 1.0649,
+      "step": 260220
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.5184816713859336e-05,
+      "loss": 1.1123,
+      "step": 260230
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.518366878395959e-05,
+      "loss": 1.0276,
+      "step": 260240
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.518252085405984e-05,
+      "loss": 1.1064,
+      "step": 260250
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.518137292416009e-05,
+      "loss": 1.1727,
+      "step": 260260
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.518022499426035e-05,
+      "loss": 0.9615,
+      "step": 260270
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.5179077064360605e-05,
+      "loss": 1.0698,
+      "step": 260280
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.517792913446085e-05,
+      "loss": 1.0796,
+      "step": 260290
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.5176781204561105e-05,
+      "loss": 1.0876,
+      "step": 260300
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.517563327466136e-05,
+      "loss": 0.9754,
+      "step": 260310
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.5174485344761605e-05,
+      "loss": 1.0722,
+      "step": 260320
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.517333741486186e-05,
+      "loss": 1.0222,
+      "step": 260330
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.517218948496212e-05,
+      "loss": 0.9711,
+      "step": 260340
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.517104155506237e-05,
+      "loss": 0.9775,
+      "step": 260350
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.516989362516262e-05,
+      "loss": 0.9723,
+      "step": 260360
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.5168745695262874e-05,
+      "loss": 1.0342,
+      "step": 260370
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.5167597765363124e-05,
+      "loss": 1.0074,
+      "step": 260380
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.516644983546338e-05,
+      "loss": 1.0696,
+      "step": 260390
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.516530190556363e-05,
+      "loss": 1.026,
+      "step": 260400
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.516415397566388e-05,
+      "loss": 1.0432,
+      "step": 260410
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.516300604576414e-05,
+      "loss": 1.0574,
+      "step": 260420
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.5161858115864394e-05,
+      "loss": 0.9247,
+      "step": 260430
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.516071018596464e-05,
+      "loss": 0.9445,
+      "step": 260440
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.5159562256064894e-05,
+      "loss": 0.99,
+      "step": 260450
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.515841432616515e-05,
+      "loss": 1.1104,
+      "step": 260460
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.515726639626539e-05,
+      "loss": 0.9905,
+      "step": 260470
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.515611846636565e-05,
+      "loss": 1.0683,
+      "step": 260480
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.5154970536465907e-05,
+      "loss": 1.1346,
+      "step": 260490
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.5153822606566156e-05,
+      "loss": 1.0057,
+      "step": 260500
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.5152674676666406e-05,
+      "loss": 0.9542,
+      "step": 260510
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.515152674676666e-05,
+      "loss": 1.0287,
+      "step": 260520
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.515037881686691e-05,
+      "loss": 1.0725,
+      "step": 260530
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.514923088696717e-05,
+      "loss": 0.9901,
+      "step": 260540
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.514808295706742e-05,
+      "loss": 1.0616,
+      "step": 260550
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.514693502716767e-05,
+      "loss": 1.0711,
+      "step": 260560
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.5145787097267926e-05,
+      "loss": 1.0031,
+      "step": 260570
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.514463916736818e-05,
+      "loss": 1.0084,
+      "step": 260580
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.5143491237468425e-05,
+      "loss": 1.0985,
+      "step": 260590
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.514234330756868e-05,
+      "loss": 1.0688,
+      "step": 260600
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.514119537766893e-05,
+      "loss": 1.0301,
+      "step": 260610
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.514004744776918e-05,
+      "loss": 1.067,
+      "step": 260620
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.513889951786944e-05,
+      "loss": 1.1038,
+      "step": 260630
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.513775158796969e-05,
+      "loss": 0.8954,
+      "step": 260640
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.5136603658069945e-05,
+      "loss": 1.0399,
+      "step": 260650
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.5135455728170195e-05,
+      "loss": 1.1077,
+      "step": 260660
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.5134307798270445e-05,
+      "loss": 1.1862,
+      "step": 260670
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.51331598683707e-05,
+      "loss": 1.0164,
+      "step": 260680
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.513201193847096e-05,
+      "loss": 1.0896,
+      "step": 260690
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.51308640085712e-05,
+      "loss": 1.0648,
+      "step": 260700
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.512971607867146e-05,
+      "loss": 0.9804,
+      "step": 260710
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.5128568148771714e-05,
+      "loss": 0.9931,
+      "step": 260720
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.512742021887196e-05,
+      "loss": 0.9134,
+      "step": 260730
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.5126272288972214e-05,
+      "loss": 1.1035,
+      "step": 260740
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.512512435907247e-05,
+      "loss": 0.9811,
+      "step": 260750
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.512397642917272e-05,
+      "loss": 1.026,
+      "step": 260760
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.512282849927297e-05,
+      "loss": 1.0635,
+      "step": 260770
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.512168056937323e-05,
+      "loss": 1.0403,
+      "step": 260780
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.512053263947348e-05,
+      "loss": 1.0909,
+      "step": 260790
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.511938470957373e-05,
+      "loss": 1.0287,
+      "step": 260800
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.511823677967398e-05,
+      "loss": 0.9706,
+      "step": 260810
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.511708884977423e-05,
+      "loss": 0.9682,
+      "step": 260820
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.511594091987449e-05,
+      "loss": 1.081,
+      "step": 260830
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.5114792989974746e-05,
+      "loss": 1.1498,
+      "step": 260840
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.511364506007499e-05,
+      "loss": 1.0711,
+      "step": 260850
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.5112497130175246e-05,
+      "loss": 1.1383,
+      "step": 260860
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.51113492002755e-05,
+      "loss": 1.1209,
+      "step": 260870
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.5110201270375746e-05,
+      "loss": 1.1869,
+      "step": 260880
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.5109053340476e-05,
+      "loss": 1.064,
+      "step": 260890
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.510790541057626e-05,
+      "loss": 1.07,
+      "step": 260900
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.510675748067651e-05,
+      "loss": 1.0579,
+      "step": 260910
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.510560955077676e-05,
+      "loss": 1.0759,
+      "step": 260920
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.5104461620877015e-05,
+      "loss": 1.0645,
+      "step": 260930
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.5103313690977265e-05,
+      "loss": 1.0946,
+      "step": 260940
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.510216576107752e-05,
+      "loss": 1.0707,
+      "step": 260950
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.510101783117777e-05,
+      "loss": 0.9543,
+      "step": 260960
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.509986990127802e-05,
+      "loss": 1.0701,
+      "step": 260970
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.509872197137828e-05,
+      "loss": 0.9955,
+      "step": 260980
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.5097574041478535e-05,
+      "loss": 1.0205,
+      "step": 260990
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.509642611157878e-05,
+      "loss": 1.1412,
+      "step": 261000
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.5095278181679035e-05,
+      "loss": 1.0734,
+      "step": 261010
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.509413025177929e-05,
+      "loss": 1.1492,
+      "step": 261020
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.5092982321879534e-05,
+      "loss": 1.0497,
+      "step": 261030
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.509183439197979e-05,
+      "loss": 1.0317,
+      "step": 261040
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.509068646208005e-05,
+      "loss": 1.0381,
+      "step": 261050
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.50895385321803e-05,
+      "loss": 0.9139,
+      "step": 261060
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.508839060228055e-05,
+      "loss": 1.0752,
+      "step": 261070
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.5087242672380804e-05,
+      "loss": 1.0088,
+      "step": 261080
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.5086094742481054e-05,
+      "loss": 1.1397,
+      "step": 261090
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.508494681258131e-05,
+      "loss": 1.0106,
+      "step": 261100
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.508379888268156e-05,
+      "loss": 1.0517,
+      "step": 261110
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.508265095278181e-05,
+      "loss": 0.9678,
+      "step": 261120
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.508150302288207e-05,
+      "loss": 0.9558,
+      "step": 261130
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.508035509298232e-05,
+      "loss": 1.0734,
+      "step": 261140
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.5079207163082566e-05,
+      "loss": 1.078,
+      "step": 261150
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.507805923318282e-05,
+      "loss": 1.0743,
+      "step": 261160
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.507691130328308e-05,
+      "loss": 1.0407,
+      "step": 261170
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.507576337338332e-05,
+      "loss": 1.1141,
+      "step": 261180
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.507461544348358e-05,
+      "loss": 1.0726,
+      "step": 261190
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.5073467513583836e-05,
+      "loss": 0.9992,
+      "step": 261200
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.5072319583684086e-05,
+      "loss": 1.0761,
+      "step": 261210
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.5071171653784336e-05,
+      "loss": 1.0696,
+      "step": 261220
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.507002372388459e-05,
+      "loss": 1.0475,
+      "step": 261230
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.506887579398484e-05,
+      "loss": 1.0583,
+      "step": 261240
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.50677278640851e-05,
+      "loss": 1.1456,
+      "step": 261250
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.506657993418535e-05,
+      "loss": 1.0127,
+      "step": 261260
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.50654320042856e-05,
+      "loss": 1.0456,
+      "step": 261270
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.5064284074385855e-05,
+      "loss": 1.1745,
+      "step": 261280
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.506313614448611e-05,
+      "loss": 1.0652,
+      "step": 261290
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.5061988214586355e-05,
+      "loss": 1.0039,
+      "step": 261300
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.506084028468661e-05,
+      "loss": 0.997,
+      "step": 261310
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.505969235478687e-05,
+      "loss": 1.0981,
+      "step": 261320
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.505854442488711e-05,
+      "loss": 1.1135,
+      "step": 261330
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.505739649498737e-05,
+      "loss": 0.9374,
+      "step": 261340
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.5056248565087624e-05,
+      "loss": 1.1301,
+      "step": 261350
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.5055100635187874e-05,
+      "loss": 0.9846,
+      "step": 261360
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.5053952705288124e-05,
+      "loss": 1.0025,
+      "step": 261370
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.505280477538838e-05,
+      "loss": 0.9985,
+      "step": 261380
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.505165684548863e-05,
+      "loss": 1.0785,
+      "step": 261390
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.505050891558889e-05,
+      "loss": 0.8766,
+      "step": 261400
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.504936098568914e-05,
+      "loss": 1.0012,
+      "step": 261410
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.504821305578939e-05,
+      "loss": 1.2325,
+      "step": 261420
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.5047065125889644e-05,
+      "loss": 1.0606,
+      "step": 261430
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.50459171959899e-05,
+      "loss": 1.1373,
+      "step": 261440
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.504476926609014e-05,
+      "loss": 1.073,
+      "step": 261450
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.50436213361904e-05,
+      "loss": 1.1178,
+      "step": 261460
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.5042473406290657e-05,
+      "loss": 1.0722,
+      "step": 261470
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.50413254763909e-05,
+      "loss": 1.0722,
+      "step": 261480
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.5040177546491156e-05,
+      "loss": 0.9506,
+      "step": 261490
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.503902961659141e-05,
+      "loss": 1.0332,
+      "step": 261500
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.503788168669166e-05,
+      "loss": 1.0687,
+      "step": 261510
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.503673375679191e-05,
+      "loss": 1.1493,
+      "step": 261520
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.503558582689217e-05,
+      "loss": 0.9922,
+      "step": 261530
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.503443789699242e-05,
+      "loss": 1.0703,
+      "step": 261540
+    },
+    {
+      "epoch": 12.0,
+      "eval_cer": 0.6686114899770257,
+      "eval_em": 0.007357212003872217,
+      "eval_f1": 0.007357212003872217,
+      "eval_loss": 1.2141674757003784,
+      "eval_runtime": 1411.1659,
+      "eval_samples_per_second": 7.32,
+      "eval_steps_per_second": 0.916,
+      "eval_wer": 0.9926427879961278,
+      "step": 261540
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.5033289967092676e-05,
+      "loss": 1.1592,
+      "step": 261550
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.5032142037192926e-05,
+      "loss": 0.976,
+      "step": 261560
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.5030994107293175e-05,
+      "loss": 0.978,
+      "step": 261570
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.502984617739343e-05,
+      "loss": 0.9533,
+      "step": 261580
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.502869824749369e-05,
+      "loss": 0.8612,
+      "step": 261590
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.502755031759393e-05,
+      "loss": 0.8546,
+      "step": 261600
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.502640238769419e-05,
+      "loss": 0.9513,
+      "step": 261610
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.5025254457794445e-05,
+      "loss": 0.9746,
+      "step": 261620
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.502410652789469e-05,
+      "loss": 0.9027,
+      "step": 261630
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.5022958597994945e-05,
+      "loss": 1.0567,
+      "step": 261640
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.50218106680952e-05,
+      "loss": 0.9288,
+      "step": 261650
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.502066273819545e-05,
+      "loss": 0.9278,
+      "step": 261660
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.50195148082957e-05,
+      "loss": 0.9878,
+      "step": 261670
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.501836687839596e-05,
+      "loss": 0.9062,
+      "step": 261680
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.501721894849621e-05,
+      "loss": 0.9664,
+      "step": 261690
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.5016071018596464e-05,
+      "loss": 0.8667,
+      "step": 261700
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.5014923088696714e-05,
+      "loss": 0.9639,
+      "step": 261710
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.5013775158796964e-05,
+      "loss": 1.0402,
+      "step": 261720
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.501262722889722e-05,
+      "loss": 1.042,
+      "step": 261730
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.501147929899748e-05,
+      "loss": 0.9047,
+      "step": 261740
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.501033136909772e-05,
+      "loss": 1.0077,
+      "step": 261750
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.500918343919798e-05,
+      "loss": 0.9597,
+      "step": 261760
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.5008035509298234e-05,
+      "loss": 0.9962,
+      "step": 261770
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.5006887579398477e-05,
+      "loss": 0.9548,
+      "step": 261780
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.500573964949873e-05,
+      "loss": 1.0019,
+      "step": 261790
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.500459171959899e-05,
+      "loss": 1.035,
+      "step": 261800
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.500344378969924e-05,
+      "loss": 0.9097,
+      "step": 261810
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.500229585979949e-05,
+      "loss": 0.9789,
+      "step": 261820
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.5001147929899746e-05,
+      "loss": 1.0673,
+      "step": 261830
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 0.94,
+      "step": 261840
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.499885207010025e-05,
+      "loss": 1.0157,
+      "step": 261850
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.49977041402005e-05,
+      "loss": 0.9595,
+      "step": 261860
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.499655621030075e-05,
+      "loss": 0.9283,
+      "step": 261870
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.499540828040101e-05,
+      "loss": 1.0607,
+      "step": 261880
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.4994260350501266e-05,
+      "loss": 1.0439,
+      "step": 261890
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.499311242060151e-05,
+      "loss": 1.0436,
+      "step": 261900
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.4991964490701765e-05,
+      "loss": 0.9655,
+      "step": 261910
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.499081656080202e-05,
+      "loss": 1.0434,
+      "step": 261920
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.4989668630902265e-05,
+      "loss": 1.03,
+      "step": 261930
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.498852070100252e-05,
+      "loss": 1.1716,
+      "step": 261940
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.498737277110278e-05,
+      "loss": 1.0193,
+      "step": 261950
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.498622484120303e-05,
+      "loss": 0.926,
+      "step": 261960
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.498507691130328e-05,
+      "loss": 1.0671,
+      "step": 261970
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.498392898140353e-05,
+      "loss": 1.0333,
+      "step": 261980
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.4982781051503785e-05,
+      "loss": 1.1289,
+      "step": 261990
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.498163312160404e-05,
+      "loss": 0.8909,
+      "step": 262000
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.4980485191704284e-05,
+      "loss": 1.0152,
+      "step": 262010
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.497933726180454e-05,
+      "loss": 0.9775,
+      "step": 262020
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.49781893319048e-05,
+      "loss": 0.9999,
+      "step": 262030
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.497704140200504e-05,
+      "loss": 0.925,
+      "step": 262040
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.49758934721053e-05,
+      "loss": 1.1451,
+      "step": 262050
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.4974745542205554e-05,
+      "loss": 0.8857,
+      "step": 262060
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.4973597612305804e-05,
+      "loss": 1.0905,
+      "step": 262070
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.4972449682406054e-05,
+      "loss": 1.0212,
+      "step": 262080
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.497130175250631e-05,
+      "loss": 1.0266,
+      "step": 262090
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.497015382260656e-05,
+      "loss": 0.9237,
+      "step": 262100
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.496900589270682e-05,
+      "loss": 1.0839,
+      "step": 262110
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.4967857962807067e-05,
+      "loss": 1.0317,
+      "step": 262120
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.4966710032907316e-05,
+      "loss": 0.9979,
+      "step": 262130
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.496556210300757e-05,
+      "loss": 1.0735,
+      "step": 262140
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.496441417310783e-05,
+      "loss": 0.9397,
+      "step": 262150
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.496326624320807e-05,
+      "loss": 1.0006,
+      "step": 262160
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.496211831330833e-05,
+      "loss": 1.0205,
+      "step": 262170
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.4960970383408586e-05,
+      "loss": 0.9271,
+      "step": 262180
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.495982245350883e-05,
+      "loss": 0.9056,
+      "step": 262190
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.4958674523609086e-05,
+      "loss": 1.062,
+      "step": 262200
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.495752659370934e-05,
+      "loss": 0.9957,
+      "step": 262210
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.495637866380959e-05,
+      "loss": 1.0835,
+      "step": 262220
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.495523073390984e-05,
+      "loss": 1.0189,
+      "step": 262230
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.49540828040101e-05,
+      "loss": 1.0526,
+      "step": 262240
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.495293487411035e-05,
+      "loss": 1.0918,
+      "step": 262250
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.4951786944210605e-05,
+      "loss": 0.9418,
+      "step": 262260
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.4950639014310855e-05,
+      "loss": 0.9622,
+      "step": 262270
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.4949491084411105e-05,
+      "loss": 1.0755,
+      "step": 262280
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.494834315451136e-05,
+      "loss": 0.9615,
+      "step": 262290
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.494719522461162e-05,
+      "loss": 1.0069,
+      "step": 262300
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.494604729471186e-05,
+      "loss": 0.9584,
+      "step": 262310
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.494489936481212e-05,
+      "loss": 0.9007,
+      "step": 262320
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.4943751434912374e-05,
+      "loss": 0.9822,
+      "step": 262330
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.494260350501262e-05,
+      "loss": 0.9672,
+      "step": 262340
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.4941455575112874e-05,
+      "loss": 0.9486,
+      "step": 262350
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.494030764521313e-05,
+      "loss": 0.9735,
+      "step": 262360
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.493915971531338e-05,
+      "loss": 1.0182,
+      "step": 262370
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.493801178541363e-05,
+      "loss": 0.9913,
+      "step": 262380
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.493686385551389e-05,
+      "loss": 1.1054,
+      "step": 262390
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.493571592561414e-05,
+      "loss": 0.9128,
+      "step": 262400
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.4934567995714394e-05,
+      "loss": 1.0247,
+      "step": 262410
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.4933420065814643e-05,
+      "loss": 0.983,
+      "step": 262420
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.493227213591489e-05,
+      "loss": 0.9404,
+      "step": 262430
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.493112420601515e-05,
+      "loss": 0.9889,
+      "step": 262440
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.4929976276115407e-05,
+      "loss": 0.9797,
+      "step": 262450
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.492882834621565e-05,
+      "loss": 0.9857,
+      "step": 262460
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.4927680416315906e-05,
+      "loss": 0.9535,
+      "step": 262470
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.492653248641616e-05,
+      "loss": 0.9998,
+      "step": 262480
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.4925384556516406e-05,
+      "loss": 0.9345,
+      "step": 262490
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.492423662661666e-05,
+      "loss": 1.0869,
+      "step": 262500
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.492308869671692e-05,
+      "loss": 0.9801,
+      "step": 262510
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.492194076681717e-05,
+      "loss": 0.9766,
+      "step": 262520
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.492079283691742e-05,
+      "loss": 0.8942,
+      "step": 262530
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.4919644907017676e-05,
+      "loss": 1.0361,
+      "step": 262540
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.4918496977117925e-05,
+      "loss": 0.9919,
+      "step": 262550
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.491734904721818e-05,
+      "loss": 0.9753,
+      "step": 262560
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.491620111731843e-05,
+      "loss": 0.9403,
+      "step": 262570
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.491505318741868e-05,
+      "loss": 1.0488,
+      "step": 262580
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.491390525751894e-05,
+      "loss": 0.9645,
+      "step": 262590
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.4912757327619195e-05,
+      "loss": 0.9654,
+      "step": 262600
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.491160939771944e-05,
+      "loss": 0.9967,
+      "step": 262610
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.4910461467819695e-05,
+      "loss": 0.8949,
+      "step": 262620
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.490931353791995e-05,
+      "loss": 0.8804,
+      "step": 262630
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.4908165608020194e-05,
+      "loss": 0.9423,
+      "step": 262640
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.490701767812045e-05,
+      "loss": 0.9287,
+      "step": 262650
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.490586974822071e-05,
+      "loss": 0.9585,
+      "step": 262660
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.490472181832096e-05,
+      "loss": 0.9289,
+      "step": 262670
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.490357388842121e-05,
+      "loss": 1.0302,
+      "step": 262680
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.4902425958521464e-05,
+      "loss": 1.1012,
+      "step": 262690
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.4901278028621714e-05,
+      "loss": 1.0222,
+      "step": 262700
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.490013009872197e-05,
+      "loss": 0.9837,
+      "step": 262710
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.489898216882222e-05,
+      "loss": 1.0326,
+      "step": 262720
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.489783423892247e-05,
+      "loss": 1.0657,
+      "step": 262730
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.489668630902273e-05,
+      "loss": 1.0035,
+      "step": 262740
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.4895538379122984e-05,
+      "loss": 1.0354,
+      "step": 262750
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.4894390449223227e-05,
+      "loss": 1.0125,
+      "step": 262760
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.489324251932348e-05,
+      "loss": 0.9363,
+      "step": 262770
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.489209458942374e-05,
+      "loss": 0.9493,
+      "step": 262780
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.489094665952398e-05,
+      "loss": 0.9481,
+      "step": 262790
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.488979872962424e-05,
+      "loss": 1.0125,
+      "step": 262800
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.4888650799724496e-05,
+      "loss": 0.9752,
+      "step": 262810
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.4887502869824746e-05,
+      "loss": 1.0,
+      "step": 262820
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.4886354939924996e-05,
+      "loss": 1.056,
+      "step": 262830
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.488520701002525e-05,
+      "loss": 0.9819,
+      "step": 262840
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.48840590801255e-05,
+      "loss": 0.9689,
+      "step": 262850
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.488291115022576e-05,
+      "loss": 1.0147,
+      "step": 262860
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.488176322032601e-05,
+      "loss": 1.039,
+      "step": 262870
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.488061529042626e-05,
+      "loss": 1.0351,
+      "step": 262880
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.4879467360526515e-05,
+      "loss": 0.9916,
+      "step": 262890
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.487831943062677e-05,
+      "loss": 1.002,
+      "step": 262900
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.4877171500727015e-05,
+      "loss": 0.9495,
+      "step": 262910
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.487602357082727e-05,
+      "loss": 0.9352,
+      "step": 262920
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.487487564092753e-05,
+      "loss": 1.0453,
+      "step": 262930
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.487372771102777e-05,
+      "loss": 1.0175,
+      "step": 262940
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.487257978112803e-05,
+      "loss": 1.0089,
+      "step": 262950
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.4871431851228285e-05,
+      "loss": 0.9809,
+      "step": 262960
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.4870283921328535e-05,
+      "loss": 1.1123,
+      "step": 262970
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.4869135991428784e-05,
+      "loss": 1.1055,
+      "step": 262980
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.486798806152904e-05,
+      "loss": 0.9374,
+      "step": 262990
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.486684013162929e-05,
+      "loss": 0.9969,
+      "step": 263000
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.486569220172955e-05,
+      "loss": 0.878,
+      "step": 263010
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.48645442718298e-05,
+      "loss": 1.0347,
+      "step": 263020
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.486339634193005e-05,
+      "loss": 0.994,
+      "step": 263030
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.4862248412030304e-05,
+      "loss": 1.0609,
+      "step": 263040
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.486110048213056e-05,
+      "loss": 0.9798,
+      "step": 263050
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.4859952552230804e-05,
+      "loss": 1.0862,
+      "step": 263060
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.485880462233106e-05,
+      "loss": 1.02,
+      "step": 263070
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.485765669243132e-05,
+      "loss": 0.9605,
+      "step": 263080
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.485650876253156e-05,
+      "loss": 0.8907,
+      "step": 263090
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.4855360832631817e-05,
+      "loss": 0.9757,
+      "step": 263100
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.485421290273207e-05,
+      "loss": 0.978,
+      "step": 263110
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.485306497283232e-05,
+      "loss": 0.9625,
+      "step": 263120
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.485191704293257e-05,
+      "loss": 1.0785,
+      "step": 263130
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.485076911303283e-05,
+      "loss": 1.007,
+      "step": 263140
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.484962118313308e-05,
+      "loss": 1.0185,
+      "step": 263150
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.4848473253233336e-05,
+      "loss": 0.9883,
+      "step": 263160
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.4847325323333586e-05,
+      "loss": 1.1199,
+      "step": 263170
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.4846177393433836e-05,
+      "loss": 0.9302,
+      "step": 263180
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.484502946353409e-05,
+      "loss": 0.9875,
+      "step": 263190
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.484388153363435e-05,
+      "loss": 0.9205,
+      "step": 263200
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.484273360373459e-05,
+      "loss": 0.979,
+      "step": 263210
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.484158567383485e-05,
+      "loss": 0.9597,
+      "step": 263220
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.4840437743935105e-05,
+      "loss": 1.0478,
+      "step": 263230
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.483928981403535e-05,
+      "loss": 1.0141,
+      "step": 263240
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.4838141884135605e-05,
+      "loss": 0.9219,
+      "step": 263250
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.483699395423586e-05,
+      "loss": 1.0647,
+      "step": 263260
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.483584602433611e-05,
+      "loss": 1.0434,
+      "step": 263270
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.483469809443636e-05,
+      "loss": 1.0595,
+      "step": 263280
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.483355016453662e-05,
+      "loss": 0.9528,
+      "step": 263290
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.483240223463687e-05,
+      "loss": 1.1303,
+      "step": 263300
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.4831254304737124e-05,
+      "loss": 1.0722,
+      "step": 263310
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.4830106374837374e-05,
+      "loss": 0.9251,
+      "step": 263320
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.4828958444937624e-05,
+      "loss": 1.0396,
+      "step": 263330
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.482781051503788e-05,
+      "loss": 0.9341,
+      "step": 263340
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.4826662585138124e-05,
+      "loss": 0.9152,
+      "step": 263350
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.482551465523838e-05,
+      "loss": 0.9996,
+      "step": 263360
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.482436672533864e-05,
+      "loss": 1.0446,
+      "step": 263370
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.482321879543889e-05,
+      "loss": 1.04,
+      "step": 263380
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.482207086553914e-05,
+      "loss": 1.0653,
+      "step": 263390
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.4820922935639393e-05,
+      "loss": 0.9656,
+      "step": 263400
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.481977500573964e-05,
+      "loss": 0.9641,
+      "step": 263410
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.48186270758399e-05,
+      "loss": 1.0053,
+      "step": 263420
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.481747914594015e-05,
+      "loss": 0.9892,
+      "step": 263430
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.48163312160404e-05,
+      "loss": 1.0368,
+      "step": 263440
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.4815183286140656e-05,
+      "loss": 0.9941,
+      "step": 263450
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.481403535624091e-05,
+      "loss": 1.114,
+      "step": 263460
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.4812887426341156e-05,
+      "loss": 0.9635,
+      "step": 263470
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.481173949644141e-05,
+      "loss": 1.0246,
+      "step": 263480
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.481059156654167e-05,
+      "loss": 1.0483,
+      "step": 263490
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.480944363664191e-05,
+      "loss": 1.0255,
+      "step": 263500
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.480829570674217e-05,
+      "loss": 0.9798,
+      "step": 263510
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.4807147776842426e-05,
+      "loss": 0.9398,
+      "step": 263520
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.4805999846942675e-05,
+      "loss": 0.9533,
+      "step": 263530
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.4804851917042925e-05,
+      "loss": 0.9227,
+      "step": 263540
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.480370398714318e-05,
+      "loss": 0.9934,
+      "step": 263550
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.480255605724343e-05,
+      "loss": 1.0187,
+      "step": 263560
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.480140812734369e-05,
+      "loss": 1.006,
+      "step": 263570
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.480026019744394e-05,
+      "loss": 1.074,
+      "step": 263580
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.479911226754419e-05,
+      "loss": 0.9056,
+      "step": 263590
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.4797964337644445e-05,
+      "loss": 0.9497,
+      "step": 263600
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.47968164077447e-05,
+      "loss": 0.9174,
+      "step": 263610
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.4795668477844944e-05,
+      "loss": 1.0502,
+      "step": 263620
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.47945205479452e-05,
+      "loss": 0.9953,
+      "step": 263630
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.479337261804546e-05,
+      "loss": 1.069,
+      "step": 263640
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.47922246881457e-05,
+      "loss": 0.9822,
+      "step": 263650
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.479107675824596e-05,
+      "loss": 1.0313,
+      "step": 263660
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.4789928828346214e-05,
+      "loss": 0.9934,
+      "step": 263670
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.4788780898446464e-05,
+      "loss": 0.9287,
+      "step": 263680
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.4787632968546714e-05,
+      "loss": 0.9721,
+      "step": 263690
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.478648503864697e-05,
+      "loss": 0.9911,
+      "step": 263700
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.478533710874722e-05,
+      "loss": 0.9463,
+      "step": 263710
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.478418917884748e-05,
+      "loss": 0.9719,
+      "step": 263720
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.478304124894773e-05,
+      "loss": 1.0003,
+      "step": 263730
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.4781893319047977e-05,
+      "loss": 0.9706,
+      "step": 263740
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.478074538914823e-05,
+      "loss": 0.9921,
+      "step": 263750
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.477959745924849e-05,
+      "loss": 1.0493,
+      "step": 263760
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.477844952934873e-05,
+      "loss": 0.9999,
+      "step": 263770
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.477730159944899e-05,
+      "loss": 1.0068,
+      "step": 263780
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.4776153669549246e-05,
+      "loss": 0.899,
+      "step": 263790
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.477500573964949e-05,
+      "loss": 1.0082,
+      "step": 263800
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.4773857809749746e-05,
+      "loss": 0.983,
+      "step": 263810
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.477270987985e-05,
+      "loss": 0.9025,
+      "step": 263820
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.477156194995025e-05,
+      "loss": 0.9907,
+      "step": 263830
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.47704140200505e-05,
+      "loss": 0.9465,
+      "step": 263840
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.476926609015076e-05,
+      "loss": 1.1222,
+      "step": 263850
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.476811816025101e-05,
+      "loss": 1.0437,
+      "step": 263860
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.4766970230351265e-05,
+      "loss": 1.0305,
+      "step": 263870
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.4765822300451515e-05,
+      "loss": 0.9731,
+      "step": 263880
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.4764674370551765e-05,
+      "loss": 1.017,
+      "step": 263890
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.476352644065202e-05,
+      "loss": 1.0178,
+      "step": 263900
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.476237851075228e-05,
+      "loss": 0.9936,
+      "step": 263910
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.476123058085252e-05,
+      "loss": 1.0689,
+      "step": 263920
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.476008265095278e-05,
+      "loss": 0.8793,
+      "step": 263930
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.4758934721053035e-05,
+      "loss": 0.9916,
+      "step": 263940
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.475778679115328e-05,
+      "loss": 0.9599,
+      "step": 263950
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.4756638861253534e-05,
+      "loss": 1.0253,
+      "step": 263960
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.475549093135379e-05,
+      "loss": 1.0031,
+      "step": 263970
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.475434300145404e-05,
+      "loss": 1.0316,
+      "step": 263980
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.475319507155429e-05,
+      "loss": 1.0054,
+      "step": 263990
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.475204714165455e-05,
+      "loss": 1.0027,
+      "step": 264000
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.47508992117548e-05,
+      "loss": 1.0297,
+      "step": 264010
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.4749751281855054e-05,
+      "loss": 1.007,
+      "step": 264020
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.4748603351955304e-05,
+      "loss": 0.9549,
+      "step": 264030
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.4747455422055554e-05,
+      "loss": 1.0654,
+      "step": 264040
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.474630749215581e-05,
+      "loss": 0.9939,
+      "step": 264050
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.474515956225607e-05,
+      "loss": 1.0531,
+      "step": 264060
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.474401163235631e-05,
+      "loss": 1.0538,
+      "step": 264070
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.4742863702456567e-05,
+      "loss": 1.0027,
+      "step": 264080
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.474171577255682e-05,
+      "loss": 0.9482,
+      "step": 264090
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.4740567842657066e-05,
+      "loss": 0.9768,
+      "step": 264100
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.473941991275732e-05,
+      "loss": 0.9688,
+      "step": 264110
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.473827198285758e-05,
+      "loss": 1.0707,
+      "step": 264120
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.473712405295783e-05,
+      "loss": 0.9772,
+      "step": 264130
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.473597612305808e-05,
+      "loss": 0.9596,
+      "step": 264140
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.4734828193158336e-05,
+      "loss": 0.9707,
+      "step": 264150
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.4733680263258586e-05,
+      "loss": 0.8478,
+      "step": 264160
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.473253233335884e-05,
+      "loss": 1.0131,
+      "step": 264170
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.473138440345909e-05,
+      "loss": 0.9727,
+      "step": 264180
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.473023647355934e-05,
+      "loss": 1.0068,
+      "step": 264190
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.47290885436596e-05,
+      "loss": 0.9645,
+      "step": 264200
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.4727940613759855e-05,
+      "loss": 1.1182,
+      "step": 264210
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.47267926838601e-05,
+      "loss": 1.0074,
+      "step": 264220
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.4725644753960355e-05,
+      "loss": 1.0209,
+      "step": 264230
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.472449682406061e-05,
+      "loss": 0.955,
+      "step": 264240
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.4723348894160855e-05,
+      "loss": 0.9776,
+      "step": 264250
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.472220096426111e-05,
+      "loss": 1.0197,
+      "step": 264260
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.472105303436137e-05,
+      "loss": 1.0148,
+      "step": 264270
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.471990510446162e-05,
+      "loss": 1.0178,
+      "step": 264280
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.471875717456187e-05,
+      "loss": 0.9999,
+      "step": 264290
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.4717609244662124e-05,
+      "loss": 1.0014,
+      "step": 264300
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.4716461314762374e-05,
+      "loss": 0.967,
+      "step": 264310
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.471531338486263e-05,
+      "loss": 0.981,
+      "step": 264320
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.471416545496288e-05,
+      "loss": 1.0023,
+      "step": 264330
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.471301752506313e-05,
+      "loss": 0.9421,
+      "step": 264340
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.471186959516339e-05,
+      "loss": 1.0723,
+      "step": 264350
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.4710721665263644e-05,
+      "loss": 1.0303,
+      "step": 264360
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.470957373536389e-05,
+      "loss": 0.9697,
+      "step": 264370
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.4708425805464143e-05,
+      "loss": 0.9523,
+      "step": 264380
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.47072778755644e-05,
+      "loss": 1.142,
+      "step": 264390
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.470612994566464e-05,
+      "loss": 0.9067,
+      "step": 264400
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.47049820157649e-05,
+      "loss": 0.939,
+      "step": 264410
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.4703834085865156e-05,
+      "loss": 1.0041,
+      "step": 264420
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.4702686155965406e-05,
+      "loss": 1.0038,
+      "step": 264430
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.4701538226065656e-05,
+      "loss": 1.0091,
+      "step": 264440
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.470039029616591e-05,
+      "loss": 0.9398,
+      "step": 264450
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.469924236626616e-05,
+      "loss": 0.919,
+      "step": 264460
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.469809443636642e-05,
+      "loss": 0.9741,
+      "step": 264470
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.469694650646667e-05,
+      "loss": 0.9734,
+      "step": 264480
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.469579857656692e-05,
+      "loss": 0.9693,
+      "step": 264490
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.4694650646667176e-05,
+      "loss": 0.9584,
+      "step": 264500
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.469350271676743e-05,
+      "loss": 0.9428,
+      "step": 264510
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.4692354786867675e-05,
+      "loss": 1.1255,
+      "step": 264520
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.469120685696793e-05,
+      "loss": 1.0433,
+      "step": 264530
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.469005892706819e-05,
+      "loss": 0.9031,
+      "step": 264540
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.468891099716843e-05,
+      "loss": 1.0323,
+      "step": 264550
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.468776306726869e-05,
+      "loss": 0.9799,
+      "step": 264560
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.4686615137368945e-05,
+      "loss": 0.997,
+      "step": 264570
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.4685467207469195e-05,
+      "loss": 0.9601,
+      "step": 264580
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.4684319277569445e-05,
+      "loss": 1.025,
+      "step": 264590
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.46831713476697e-05,
+      "loss": 0.9007,
+      "step": 264600
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.468202341776995e-05,
+      "loss": 0.9369,
+      "step": 264610
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.468087548787021e-05,
+      "loss": 1.0202,
+      "step": 264620
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.467972755797046e-05,
+      "loss": 1.0308,
+      "step": 264630
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.467857962807071e-05,
+      "loss": 0.9931,
+      "step": 264640
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.4677431698170964e-05,
+      "loss": 0.9136,
+      "step": 264650
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.467628376827122e-05,
+      "loss": 1.0478,
+      "step": 264660
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.4675135838371464e-05,
+      "loss": 0.9592,
+      "step": 264670
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.467398790847172e-05,
+      "loss": 0.9308,
+      "step": 264680
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.467283997857198e-05,
+      "loss": 1.0446,
+      "step": 264690
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.467169204867222e-05,
+      "loss": 1.0465,
+      "step": 264700
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.467054411877248e-05,
+      "loss": 0.9396,
+      "step": 264710
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.466939618887273e-05,
+      "loss": 0.9228,
+      "step": 264720
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.466824825897298e-05,
+      "loss": 0.9925,
+      "step": 264730
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.466710032907323e-05,
+      "loss": 1.0221,
+      "step": 264740
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.466595239917348e-05,
+      "loss": 0.9608,
+      "step": 264750
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.466480446927374e-05,
+      "loss": 0.9556,
+      "step": 264760
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.4663656539373996e-05,
+      "loss": 1.0481,
+      "step": 264770
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.466250860947424e-05,
+      "loss": 0.9699,
+      "step": 264780
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.4661360679574496e-05,
+      "loss": 1.0588,
+      "step": 264790
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.466021274967475e-05,
+      "loss": 0.8854,
+      "step": 264800
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.4659064819774996e-05,
+      "loss": 0.9701,
+      "step": 264810
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.465791688987525e-05,
+      "loss": 0.9517,
+      "step": 264820
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.465676895997551e-05,
+      "loss": 1.0128,
+      "step": 264830
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.465562103007576e-05,
+      "loss": 1.0162,
+      "step": 264840
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.465447310017601e-05,
+      "loss": 1.0918,
+      "step": 264850
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.4653325170276265e-05,
+      "loss": 0.9863,
+      "step": 264860
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.4652177240376515e-05,
+      "loss": 1.0242,
+      "step": 264870
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.465102931047677e-05,
+      "loss": 0.8544,
+      "step": 264880
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.464988138057702e-05,
+      "loss": 0.9541,
+      "step": 264890
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.464873345067727e-05,
+      "loss": 0.9534,
+      "step": 264900
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.464758552077753e-05,
+      "loss": 0.9811,
+      "step": 264910
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.4646437590877785e-05,
+      "loss": 0.938,
+      "step": 264920
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.464528966097803e-05,
+      "loss": 0.954,
+      "step": 264930
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.4644141731078284e-05,
+      "loss": 1.0591,
+      "step": 264940
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.464299380117854e-05,
+      "loss": 1.0609,
+      "step": 264950
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.4641845871278784e-05,
+      "loss": 1.1119,
+      "step": 264960
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.464069794137904e-05,
+      "loss": 0.9953,
+      "step": 264970
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.46395500114793e-05,
+      "loss": 1.0356,
+      "step": 264980
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.463840208157955e-05,
+      "loss": 0.9948,
+      "step": 264990
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.46372541516798e-05,
+      "loss": 1.0417,
+      "step": 265000
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.4636106221780054e-05,
+      "loss": 0.9411,
+      "step": 265010
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.4634958291880304e-05,
+      "loss": 1.0315,
+      "step": 265020
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.463381036198056e-05,
+      "loss": 1.0111,
+      "step": 265030
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.463266243208081e-05,
+      "loss": 1.0218,
+      "step": 265040
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.463151450218106e-05,
+      "loss": 0.889,
+      "step": 265050
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.4630366572281317e-05,
+      "loss": 0.9846,
+      "step": 265060
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.462921864238157e-05,
+      "loss": 0.9164,
+      "step": 265070
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.4628070712481816e-05,
+      "loss": 1.016,
+      "step": 265080
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.462692278258207e-05,
+      "loss": 0.9307,
+      "step": 265090
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.462577485268233e-05,
+      "loss": 1.0117,
+      "step": 265100
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.462462692278257e-05,
+      "loss": 0.9582,
+      "step": 265110
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.462347899288283e-05,
+      "loss": 0.9293,
+      "step": 265120
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.4622331062983086e-05,
+      "loss": 1.0507,
+      "step": 265130
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.4621183133083336e-05,
+      "loss": 1.0622,
+      "step": 265140
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.4620035203183586e-05,
+      "loss": 1.0564,
+      "step": 265150
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.461888727328384e-05,
+      "loss": 0.9862,
+      "step": 265160
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.461773934338409e-05,
+      "loss": 0.9394,
+      "step": 265170
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.461659141348435e-05,
+      "loss": 0.9749,
+      "step": 265180
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.46154434835846e-05,
+      "loss": 1.0474,
+      "step": 265190
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.461429555368485e-05,
+      "loss": 0.9625,
+      "step": 265200
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.4613147623785105e-05,
+      "loss": 0.9905,
+      "step": 265210
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.461199969388536e-05,
+      "loss": 1.0769,
+      "step": 265220
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.4610851763985605e-05,
+      "loss": 0.9431,
+      "step": 265230
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.460970383408586e-05,
+      "loss": 1.0933,
+      "step": 265240
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.460855590418612e-05,
+      "loss": 1.053,
+      "step": 265250
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.460740797428636e-05,
+      "loss": 0.9216,
+      "step": 265260
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.460626004438662e-05,
+      "loss": 1.0085,
+      "step": 265270
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.4605112114486874e-05,
+      "loss": 1.0106,
+      "step": 265280
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.4603964184587124e-05,
+      "loss": 0.9564,
+      "step": 265290
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.4602816254687374e-05,
+      "loss": 0.9098,
+      "step": 265300
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.460166832478763e-05,
+      "loss": 1.0005,
+      "step": 265310
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.460052039488788e-05,
+      "loss": 0.9722,
+      "step": 265320
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.459937246498814e-05,
+      "loss": 1.0861,
+      "step": 265330
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.459822453508839e-05,
+      "loss": 0.9555,
+      "step": 265340
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.459707660518864e-05,
+      "loss": 1.0757,
+      "step": 265350
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.4595928675288893e-05,
+      "loss": 1.0705,
+      "step": 265360
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.459478074538915e-05,
+      "loss": 1.02,
+      "step": 265370
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.459363281548939e-05,
+      "loss": 0.9936,
+      "step": 265380
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.459248488558965e-05,
+      "loss": 0.9617,
+      "step": 265390
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.4591336955689906e-05,
+      "loss": 1.0143,
+      "step": 265400
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.459018902579015e-05,
+      "loss": 0.9715,
+      "step": 265410
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.4589041095890406e-05,
+      "loss": 0.9218,
+      "step": 265420
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.458789316599066e-05,
+      "loss": 0.9994,
+      "step": 265430
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.458674523609091e-05,
+      "loss": 1.0127,
+      "step": 265440
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.458559730619116e-05,
+      "loss": 0.9667,
+      "step": 265450
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.458444937629142e-05,
+      "loss": 1.0142,
+      "step": 265460
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.458330144639167e-05,
+      "loss": 0.9155,
+      "step": 265470
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.4582153516491926e-05,
+      "loss": 1.095,
+      "step": 265480
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.4581005586592175e-05,
+      "loss": 1.0355,
+      "step": 265490
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.4579857656692425e-05,
+      "loss": 0.8822,
+      "step": 265500
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.457870972679268e-05,
+      "loss": 1.0704,
+      "step": 265510
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.457756179689294e-05,
+      "loss": 1.0004,
+      "step": 265520
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.457641386699318e-05,
+      "loss": 0.9609,
+      "step": 265530
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.457526593709344e-05,
+      "loss": 1.0515,
+      "step": 265540
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.4574118007193695e-05,
+      "loss": 1.0422,
+      "step": 265550
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.457297007729394e-05,
+      "loss": 0.9598,
+      "step": 265560
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.4571822147394195e-05,
+      "loss": 0.9641,
+      "step": 265570
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.457067421749445e-05,
+      "loss": 1.0905,
+      "step": 265580
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.45695262875947e-05,
+      "loss": 0.945,
+      "step": 265590
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.456837835769495e-05,
+      "loss": 1.1196,
+      "step": 265600
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.456723042779521e-05,
+      "loss": 0.9388,
+      "step": 265610
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.456608249789546e-05,
+      "loss": 0.8538,
+      "step": 265620
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.4564934567995714e-05,
+      "loss": 0.9307,
+      "step": 265630
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.4563786638095964e-05,
+      "loss": 1.0634,
+      "step": 265640
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.4562638708196214e-05,
+      "loss": 1.0175,
+      "step": 265650
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.456149077829647e-05,
+      "loss": 1.0154,
+      "step": 265660
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.456034284839673e-05,
+      "loss": 0.9863,
+      "step": 265670
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.455919491849697e-05,
+      "loss": 0.9137,
+      "step": 265680
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.455804698859723e-05,
+      "loss": 0.9833,
+      "step": 265690
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.455689905869748e-05,
+      "loss": 0.8828,
+      "step": 265700
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.4555751128797726e-05,
+      "loss": 1.0642,
+      "step": 265710
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.455460319889798e-05,
+      "loss": 1.1117,
+      "step": 265720
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.455345526899824e-05,
+      "loss": 1.04,
+      "step": 265730
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.455230733909849e-05,
+      "loss": 0.9568,
+      "step": 265740
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.455115940919874e-05,
+      "loss": 0.9123,
+      "step": 265750
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.4550011479298996e-05,
+      "loss": 0.9694,
+      "step": 265760
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.4548863549399246e-05,
+      "loss": 0.9131,
+      "step": 265770
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.45477156194995e-05,
+      "loss": 0.9398,
+      "step": 265780
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.454656768959975e-05,
+      "loss": 0.9564,
+      "step": 265790
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.45454197597e-05,
+      "loss": 0.9135,
+      "step": 265800
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.454427182980026e-05,
+      "loss": 0.9854,
+      "step": 265810
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.4543123899900515e-05,
+      "loss": 0.9431,
+      "step": 265820
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.454197597000076e-05,
+      "loss": 1.0102,
+      "step": 265830
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.4540828040101015e-05,
+      "loss": 0.9594,
+      "step": 265840
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.453968011020127e-05,
+      "loss": 0.9133,
+      "step": 265850
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.4538532180301515e-05,
+      "loss": 1.0095,
+      "step": 265860
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.453738425040177e-05,
+      "loss": 1.1588,
+      "step": 265870
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.453623632050203e-05,
+      "loss": 0.9949,
+      "step": 265880
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.453508839060228e-05,
+      "loss": 1.1217,
+      "step": 265890
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.453394046070253e-05,
+      "loss": 0.9584,
+      "step": 265900
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.4532792530802785e-05,
+      "loss": 0.8718,
+      "step": 265910
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.4531644600903034e-05,
+      "loss": 1.0863,
+      "step": 265920
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.453049667100329e-05,
+      "loss": 0.901,
+      "step": 265930
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.452934874110354e-05,
+      "loss": 0.9567,
+      "step": 265940
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.452820081120379e-05,
+      "loss": 1.0118,
+      "step": 265950
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.452705288130405e-05,
+      "loss": 1.0996,
+      "step": 265960
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.4525904951404304e-05,
+      "loss": 0.8789,
+      "step": 265970
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.452475702150455e-05,
+      "loss": 0.9484,
+      "step": 265980
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.4523609091604804e-05,
+      "loss": 1.0063,
+      "step": 265990
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.452246116170506e-05,
+      "loss": 0.913,
+      "step": 266000
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.45213132318053e-05,
+      "loss": 0.9301,
+      "step": 266010
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.452016530190556e-05,
+      "loss": 1.0297,
+      "step": 266020
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.451901737200582e-05,
+      "loss": 0.9545,
+      "step": 266030
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.4517869442106066e-05,
+      "loss": 0.9934,
+      "step": 266040
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.4516721512206316e-05,
+      "loss": 1.0108,
+      "step": 266050
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.451557358230657e-05,
+      "loss": 1.0675,
+      "step": 266060
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.451442565240682e-05,
+      "loss": 1.0706,
+      "step": 266070
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.451327772250708e-05,
+      "loss": 1.0052,
+      "step": 266080
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.451212979260733e-05,
+      "loss": 0.946,
+      "step": 266090
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.451098186270758e-05,
+      "loss": 1.0851,
+      "step": 266100
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.4509833932807836e-05,
+      "loss": 0.9831,
+      "step": 266110
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.450868600290808e-05,
+      "loss": 0.9742,
+      "step": 266120
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.4507538073008336e-05,
+      "loss": 1.0657,
+      "step": 266130
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.450639014310859e-05,
+      "loss": 0.9989,
+      "step": 266140
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.450524221320884e-05,
+      "loss": 0.9056,
+      "step": 266150
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.450409428330909e-05,
+      "loss": 1.0027,
+      "step": 266160
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.450294635340935e-05,
+      "loss": 0.8831,
+      "step": 266170
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.45017984235096e-05,
+      "loss": 1.0385,
+      "step": 266180
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.4500650493609855e-05,
+      "loss": 1.0734,
+      "step": 266190
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.4499502563710105e-05,
+      "loss": 0.9914,
+      "step": 266200
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.4498354633810355e-05,
+      "loss": 1.0427,
+      "step": 266210
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.449720670391061e-05,
+      "loss": 1.0993,
+      "step": 266220
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.449605877401087e-05,
+      "loss": 0.9657,
+      "step": 266230
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.449491084411111e-05,
+      "loss": 0.9983,
+      "step": 266240
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.449376291421137e-05,
+      "loss": 1.1027,
+      "step": 266250
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.4492614984311624e-05,
+      "loss": 1.0872,
+      "step": 266260
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.449146705441187e-05,
+      "loss": 1.0134,
+      "step": 266270
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.4490319124512124e-05,
+      "loss": 0.9731,
+      "step": 266280
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.448917119461238e-05,
+      "loss": 0.903,
+      "step": 266290
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.448802326471263e-05,
+      "loss": 1.0136,
+      "step": 266300
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.448687533481288e-05,
+      "loss": 1.0039,
+      "step": 266310
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.448572740491314e-05,
+      "loss": 0.9724,
+      "step": 266320
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.448457947501339e-05,
+      "loss": 0.8643,
+      "step": 266330
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.4483431545113643e-05,
+      "loss": 0.9812,
+      "step": 266340
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.448228361521389e-05,
+      "loss": 1.0284,
+      "step": 266350
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.448113568531414e-05,
+      "loss": 1.0222,
+      "step": 266360
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.44799877554144e-05,
+      "loss": 1.0992,
+      "step": 266370
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.4478839825514656e-05,
+      "loss": 1.0434,
+      "step": 266380
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.44776918956149e-05,
+      "loss": 1.0622,
+      "step": 266390
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.4476543965715156e-05,
+      "loss": 0.8817,
+      "step": 266400
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.447539603581541e-05,
+      "loss": 0.977,
+      "step": 266410
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.4474248105915656e-05,
+      "loss": 0.9381,
+      "step": 266420
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.447310017601591e-05,
+      "loss": 0.9476,
+      "step": 266430
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.447195224611617e-05,
+      "loss": 1.0163,
+      "step": 266440
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.447080431621642e-05,
+      "loss": 0.9886,
+      "step": 266450
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.446965638631667e-05,
+      "loss": 1.0863,
+      "step": 266460
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.4468508456416925e-05,
+      "loss": 0.9863,
+      "step": 266470
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.4467360526517175e-05,
+      "loss": 1.0003,
+      "step": 266480
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.446621259661743e-05,
+      "loss": 0.9487,
+      "step": 266490
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.446506466671768e-05,
+      "loss": 0.9755,
+      "step": 266500
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.446391673681793e-05,
+      "loss": 0.9815,
+      "step": 266510
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.446276880691819e-05,
+      "loss": 1.0816,
+      "step": 266520
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.4461620877018445e-05,
+      "loss": 1.03,
+      "step": 266530
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.446047294711869e-05,
+      "loss": 0.9545,
+      "step": 266540
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.4459325017218945e-05,
+      "loss": 0.8844,
+      "step": 266550
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.44581770873192e-05,
+      "loss": 0.9281,
+      "step": 266560
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.4457029157419444e-05,
+      "loss": 1.0579,
+      "step": 266570
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.44558812275197e-05,
+      "loss": 0.9586,
+      "step": 266580
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.445473329761996e-05,
+      "loss": 0.9252,
+      "step": 266590
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.445358536772021e-05,
+      "loss": 0.9153,
+      "step": 266600
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.445243743782046e-05,
+      "loss": 1.0528,
+      "step": 266610
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.4451289507920714e-05,
+      "loss": 0.9373,
+      "step": 266620
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.4450141578020964e-05,
+      "loss": 0.9551,
+      "step": 266630
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.444899364812122e-05,
+      "loss": 0.9565,
+      "step": 266640
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.444784571822147e-05,
+      "loss": 0.9844,
+      "step": 266650
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.444669778832172e-05,
+      "loss": 1.077,
+      "step": 266660
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.444554985842198e-05,
+      "loss": 0.9892,
+      "step": 266670
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.444440192852223e-05,
+      "loss": 1.0296,
+      "step": 266680
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.4443253998622476e-05,
+      "loss": 1.0733,
+      "step": 266690
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.444210606872273e-05,
+      "loss": 1.0494,
+      "step": 266700
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.444095813882299e-05,
+      "loss": 0.9299,
+      "step": 266710
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.443981020892323e-05,
+      "loss": 0.9012,
+      "step": 266720
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.443866227902349e-05,
+      "loss": 0.9689,
+      "step": 266730
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.4437514349123746e-05,
+      "loss": 1.0687,
+      "step": 266740
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.4436366419223996e-05,
+      "loss": 1.0303,
+      "step": 266750
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.4435218489324246e-05,
+      "loss": 1.0277,
+      "step": 266760
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.44340705594245e-05,
+      "loss": 0.9844,
+      "step": 266770
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.443292262952475e-05,
+      "loss": 1.1229,
+      "step": 266780
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.443177469962501e-05,
+      "loss": 1.0434,
+      "step": 266790
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.443062676972526e-05,
+      "loss": 1.0104,
+      "step": 266800
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.442947883982551e-05,
+      "loss": 1.0984,
+      "step": 266810
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.4428330909925765e-05,
+      "loss": 1.0092,
+      "step": 266820
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.442718298002602e-05,
+      "loss": 1.0501,
+      "step": 266830
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.4426035050126265e-05,
+      "loss": 1.0936,
+      "step": 266840
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.442488712022652e-05,
+      "loss": 1.0831,
+      "step": 266850
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.442373919032678e-05,
+      "loss": 0.9382,
+      "step": 266860
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.442259126042702e-05,
+      "loss": 0.9766,
+      "step": 266870
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.442144333052728e-05,
+      "loss": 1.0466,
+      "step": 266880
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.4420295400627535e-05,
+      "loss": 0.9706,
+      "step": 266890
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.4419147470727784e-05,
+      "loss": 1.1121,
+      "step": 266900
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.4417999540828034e-05,
+      "loss": 1.0282,
+      "step": 266910
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.441685161092829e-05,
+      "loss": 1.0722,
+      "step": 266920
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.441570368102854e-05,
+      "loss": 0.887,
+      "step": 266930
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.44145557511288e-05,
+      "loss": 1.0123,
+      "step": 266940
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.441340782122905e-05,
+      "loss": 0.9678,
+      "step": 266950
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.44122598913293e-05,
+      "loss": 0.9416,
+      "step": 266960
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.4411111961429554e-05,
+      "loss": 0.9594,
+      "step": 266970
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.440996403152981e-05,
+      "loss": 1.1117,
+      "step": 266980
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.440881610163005e-05,
+      "loss": 1.0338,
+      "step": 266990
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.440766817173031e-05,
+      "loss": 1.1128,
+      "step": 267000
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.440652024183057e-05,
+      "loss": 1.0182,
+      "step": 267010
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.440537231193081e-05,
+      "loss": 1.0586,
+      "step": 267020
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.4404224382031066e-05,
+      "loss": 1.0026,
+      "step": 267030
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.440307645213132e-05,
+      "loss": 1.0169,
+      "step": 267040
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.440192852223157e-05,
+      "loss": 1.005,
+      "step": 267050
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.440078059233182e-05,
+      "loss": 0.945,
+      "step": 267060
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.439963266243208e-05,
+      "loss": 1.0665,
+      "step": 267070
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.439848473253233e-05,
+      "loss": 0.9521,
+      "step": 267080
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.4397336802632586e-05,
+      "loss": 1.1457,
+      "step": 267090
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.4396188872732836e-05,
+      "loss": 1.0168,
+      "step": 267100
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.4395040942833086e-05,
+      "loss": 1.0745,
+      "step": 267110
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.439389301293334e-05,
+      "loss": 0.9926,
+      "step": 267120
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.43927450830336e-05,
+      "loss": 1.0581,
+      "step": 267130
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.439159715313384e-05,
+      "loss": 1.073,
+      "step": 267140
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.43904492232341e-05,
+      "loss": 0.9436,
+      "step": 267150
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.4389301293334355e-05,
+      "loss": 0.9957,
+      "step": 267160
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.43881533634346e-05,
+      "loss": 1.0489,
+      "step": 267170
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.4387005433534855e-05,
+      "loss": 1.0506,
+      "step": 267180
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.438585750363511e-05,
+      "loss": 1.0121,
+      "step": 267190
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.438470957373536e-05,
+      "loss": 1.0089,
+      "step": 267200
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.438356164383561e-05,
+      "loss": 1.0254,
+      "step": 267210
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.438241371393587e-05,
+      "loss": 0.8943,
+      "step": 267220
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.438126578403612e-05,
+      "loss": 1.0823,
+      "step": 267230
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.4380117854136374e-05,
+      "loss": 0.9471,
+      "step": 267240
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.4378969924236624e-05,
+      "loss": 0.9829,
+      "step": 267250
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.4377821994336874e-05,
+      "loss": 0.9689,
+      "step": 267260
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.437667406443713e-05,
+      "loss": 1.0129,
+      "step": 267270
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.437552613453739e-05,
+      "loss": 0.9712,
+      "step": 267280
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.437437820463763e-05,
+      "loss": 0.8423,
+      "step": 267290
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.437323027473789e-05,
+      "loss": 1.0111,
+      "step": 267300
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.4372082344838144e-05,
+      "loss": 1.0898,
+      "step": 267310
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.437093441493839e-05,
+      "loss": 1.0031,
+      "step": 267320
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.436978648503864e-05,
+      "loss": 0.9882,
+      "step": 267330
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.43686385551389e-05,
+      "loss": 0.9735,
+      "step": 267340
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.436749062523915e-05,
+      "loss": 1.0364,
+      "step": 267350
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.43663426953394e-05,
+      "loss": 0.9529,
+      "step": 267360
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.4365194765439656e-05,
+      "loss": 1.0856,
+      "step": 267370
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.4364046835539906e-05,
+      "loss": 0.9981,
+      "step": 267380
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.436289890564016e-05,
+      "loss": 0.8872,
+      "step": 267390
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.436175097574041e-05,
+      "loss": 1.0378,
+      "step": 267400
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.436060304584066e-05,
+      "loss": 1.0378,
+      "step": 267410
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.435945511594092e-05,
+      "loss": 1.1534,
+      "step": 267420
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.4358307186041176e-05,
+      "loss": 0.9119,
+      "step": 267430
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.435715925614142e-05,
+      "loss": 0.9477,
+      "step": 267440
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.4356011326241675e-05,
+      "loss": 1.131,
+      "step": 267450
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.435486339634193e-05,
+      "loss": 1.0051,
+      "step": 267460
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.4353715466442175e-05,
+      "loss": 0.9762,
+      "step": 267470
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.435256753654243e-05,
+      "loss": 1.0143,
+      "step": 267480
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.435141960664268e-05,
+      "loss": 1.0617,
+      "step": 267490
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.435027167674294e-05,
+      "loss": 0.935,
+      "step": 267500
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.434912374684319e-05,
+      "loss": 0.8864,
+      "step": 267510
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.434797581694344e-05,
+      "loss": 1.1325,
+      "step": 267520
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.4346827887043695e-05,
+      "loss": 1.0514,
+      "step": 267530
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.434567995714395e-05,
+      "loss": 0.9499,
+      "step": 267540
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.4344532027244194e-05,
+      "loss": 1.0869,
+      "step": 267550
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.434338409734445e-05,
+      "loss": 1.0168,
+      "step": 267560
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.434223616744471e-05,
+      "loss": 1.0861,
+      "step": 267570
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.434108823754495e-05,
+      "loss": 0.9368,
+      "step": 267580
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.433994030764521e-05,
+      "loss": 1.007,
+      "step": 267590
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.4338792377745464e-05,
+      "loss": 0.9248,
+      "step": 267600
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.4337644447845714e-05,
+      "loss": 0.9429,
+      "step": 267610
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.4336496517945964e-05,
+      "loss": 1.0232,
+      "step": 267620
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.433534858804622e-05,
+      "loss": 0.9913,
+      "step": 267630
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.433420065814647e-05,
+      "loss": 1.0702,
+      "step": 267640
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.433305272824673e-05,
+      "loss": 0.9768,
+      "step": 267650
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.4331904798346977e-05,
+      "loss": 0.9622,
+      "step": 267660
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.4330756868447226e-05,
+      "loss": 0.942,
+      "step": 267670
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.432960893854748e-05,
+      "loss": 0.9764,
+      "step": 267680
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.432846100864774e-05,
+      "loss": 1.0092,
+      "step": 267690
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.432731307874798e-05,
+      "loss": 0.8752,
+      "step": 267700
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.432616514884824e-05,
+      "loss": 1.0606,
+      "step": 267710
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.4325017218948496e-05,
+      "loss": 0.9851,
+      "step": 267720
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.432386928904874e-05,
+      "loss": 0.9681,
+      "step": 267730
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.4322721359148996e-05,
+      "loss": 1.0941,
+      "step": 267740
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.432157342924925e-05,
+      "loss": 1.0822,
+      "step": 267750
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.43204254993495e-05,
+      "loss": 0.9706,
+      "step": 267760
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.431927756944975e-05,
+      "loss": 0.8771,
+      "step": 267770
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.431812963955001e-05,
+      "loss": 1.0377,
+      "step": 267780
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.431698170965026e-05,
+      "loss": 1.0191,
+      "step": 267790
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.4315833779750515e-05,
+      "loss": 1.0952,
+      "step": 267800
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.4314685849850765e-05,
+      "loss": 0.9959,
+      "step": 267810
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.4313537919951015e-05,
+      "loss": 1.0684,
+      "step": 267820
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.431238999005127e-05,
+      "loss": 1.0208,
+      "step": 267830
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.431124206015153e-05,
+      "loss": 1.0223,
+      "step": 267840
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.431009413025177e-05,
+      "loss": 0.9984,
+      "step": 267850
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.430894620035203e-05,
+      "loss": 0.9623,
+      "step": 267860
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.4307798270452285e-05,
+      "loss": 1.0432,
+      "step": 267870
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.430665034055253e-05,
+      "loss": 0.9947,
+      "step": 267880
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.4305502410652784e-05,
+      "loss": 1.0647,
+      "step": 267890
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.430435448075304e-05,
+      "loss": 1.0269,
+      "step": 267900
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.430320655085329e-05,
+      "loss": 0.9559,
+      "step": 267910
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.430205862095354e-05,
+      "loss": 1.1059,
+      "step": 267920
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.43009106910538e-05,
+      "loss": 0.9691,
+      "step": 267930
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.429976276115405e-05,
+      "loss": 0.9207,
+      "step": 267940
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.4298614831254304e-05,
+      "loss": 1.0315,
+      "step": 267950
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.4297466901354554e-05,
+      "loss": 1.0116,
+      "step": 267960
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.42963189714548e-05,
+      "loss": 1.0204,
+      "step": 267970
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.429517104155506e-05,
+      "loss": 0.9851,
+      "step": 267980
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.429402311165532e-05,
+      "loss": 1.085,
+      "step": 267990
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.429287518175556e-05,
+      "loss": 1.0034,
+      "step": 268000
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.4291727251855816e-05,
+      "loss": 1.0465,
+      "step": 268010
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.429057932195607e-05,
+      "loss": 1.0097,
+      "step": 268020
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.4289431392056316e-05,
+      "loss": 1.0772,
+      "step": 268030
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.428828346215657e-05,
+      "loss": 1.0133,
+      "step": 268040
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.428713553225683e-05,
+      "loss": 0.979,
+      "step": 268050
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.428598760235708e-05,
+      "loss": 0.8976,
+      "step": 268060
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.428483967245733e-05,
+      "loss": 1.0197,
+      "step": 268070
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.4283691742557586e-05,
+      "loss": 0.9883,
+      "step": 268080
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.4282543812657836e-05,
+      "loss": 1.0191,
+      "step": 268090
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.428139588275809e-05,
+      "loss": 0.9025,
+      "step": 268100
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.428024795285834e-05,
+      "loss": 1.0526,
+      "step": 268110
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.427910002295859e-05,
+      "loss": 1.0993,
+      "step": 268120
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.427795209305885e-05,
+      "loss": 1.0314,
+      "step": 268130
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.4276804163159105e-05,
+      "loss": 0.9914,
+      "step": 268140
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.427565623325935e-05,
+      "loss": 1.085,
+      "step": 268150
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.4274508303359605e-05,
+      "loss": 0.964,
+      "step": 268160
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.427336037345986e-05,
+      "loss": 0.9761,
+      "step": 268170
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.4272212443560105e-05,
+      "loss": 0.9891,
+      "step": 268180
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.427106451366036e-05,
+      "loss": 1.0297,
+      "step": 268190
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.426991658376062e-05,
+      "loss": 0.9437,
+      "step": 268200
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.426876865386087e-05,
+      "loss": 0.9497,
+      "step": 268210
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.426762072396112e-05,
+      "loss": 0.9987,
+      "step": 268220
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.4266472794061374e-05,
+      "loss": 0.9697,
+      "step": 268230
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.4265324864161624e-05,
+      "loss": 1.0126,
+      "step": 268240
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.426417693426188e-05,
+      "loss": 0.9501,
+      "step": 268250
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.426302900436213e-05,
+      "loss": 0.9534,
+      "step": 268260
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.426188107446238e-05,
+      "loss": 1.0168,
+      "step": 268270
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.426073314456264e-05,
+      "loss": 1.0222,
+      "step": 268280
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.4259585214662894e-05,
+      "loss": 1.0421,
+      "step": 268290
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.425843728476314e-05,
+      "loss": 0.9215,
+      "step": 268300
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.425728935486339e-05,
+      "loss": 0.947,
+      "step": 268310
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.425614142496365e-05,
+      "loss": 1.0146,
+      "step": 268320
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.425499349506389e-05,
+      "loss": 1.0411,
+      "step": 268330
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.425384556516415e-05,
+      "loss": 1.0867,
+      "step": 268340
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.4252697635264406e-05,
+      "loss": 1.0718,
+      "step": 268350
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.4251549705364656e-05,
+      "loss": 1.1346,
+      "step": 268360
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.4250401775464906e-05,
+      "loss": 0.9839,
+      "step": 268370
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.424925384556516e-05,
+      "loss": 1.0096,
+      "step": 268380
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.424810591566541e-05,
+      "loss": 1.0497,
+      "step": 268390
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.424695798576567e-05,
+      "loss": 0.9626,
+      "step": 268400
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.424581005586592e-05,
+      "loss": 1.009,
+      "step": 268410
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.424466212596617e-05,
+      "loss": 1.0945,
+      "step": 268420
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.4243514196066425e-05,
+      "loss": 1.0022,
+      "step": 268430
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.424236626616668e-05,
+      "loss": 0.9672,
+      "step": 268440
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.4241218336266925e-05,
+      "loss": 1.0532,
+      "step": 268450
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.424007040636718e-05,
+      "loss": 0.97,
+      "step": 268460
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.423892247646744e-05,
+      "loss": 1.0156,
+      "step": 268470
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.423777454656768e-05,
+      "loss": 1.1434,
+      "step": 268480
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.423662661666794e-05,
+      "loss": 0.9865,
+      "step": 268490
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.4235478686768195e-05,
+      "loss": 1.0791,
+      "step": 268500
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.4234330756868445e-05,
+      "loss": 0.9282,
+      "step": 268510
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.4233182826968694e-05,
+      "loss": 0.9176,
+      "step": 268520
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.423203489706895e-05,
+      "loss": 1.0057,
+      "step": 268530
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.42308869671692e-05,
+      "loss": 0.9458,
+      "step": 268540
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.422973903726946e-05,
+      "loss": 0.9329,
+      "step": 268550
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.422859110736971e-05,
+      "loss": 0.9495,
+      "step": 268560
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.422744317746996e-05,
+      "loss": 0.997,
+      "step": 268570
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.4226295247570214e-05,
+      "loss": 0.9733,
+      "step": 268580
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.422514731767047e-05,
+      "loss": 1.0951,
+      "step": 268590
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.4223999387770714e-05,
+      "loss": 0.9615,
+      "step": 268600
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.422285145787097e-05,
+      "loss": 0.7745,
+      "step": 268610
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.422170352797123e-05,
+      "loss": 1.0843,
+      "step": 268620
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.422055559807147e-05,
+      "loss": 1.0533,
+      "step": 268630
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.4219407668171727e-05,
+      "loss": 1.0397,
+      "step": 268640
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.421825973827198e-05,
+      "loss": 1.1647,
+      "step": 268650
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.421711180837223e-05,
+      "loss": 1.0196,
+      "step": 268660
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.421596387847248e-05,
+      "loss": 0.9445,
+      "step": 268670
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.421481594857274e-05,
+      "loss": 1.1113,
+      "step": 268680
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.421366801867299e-05,
+      "loss": 1.027,
+      "step": 268690
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.4212520088773246e-05,
+      "loss": 0.9024,
+      "step": 268700
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.4211372158873496e-05,
+      "loss": 1.1236,
+      "step": 268710
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.4210224228973746e-05,
+      "loss": 1.0775,
+      "step": 268720
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.4209076299074e-05,
+      "loss": 1.0024,
+      "step": 268730
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.420792836917426e-05,
+      "loss": 1.0036,
+      "step": 268740
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.42067804392745e-05,
+      "loss": 0.9565,
+      "step": 268750
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.420563250937476e-05,
+      "loss": 0.9669,
+      "step": 268760
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.4204484579475015e-05,
+      "loss": 1.032,
+      "step": 268770
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.420333664957526e-05,
+      "loss": 1.1243,
+      "step": 268780
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.4202188719675515e-05,
+      "loss": 1.0695,
+      "step": 268790
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.420104078977577e-05,
+      "loss": 0.9397,
+      "step": 268800
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.419989285987602e-05,
+      "loss": 1.0973,
+      "step": 268810
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.419874492997627e-05,
+      "loss": 0.8679,
+      "step": 268820
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.419759700007653e-05,
+      "loss": 1.0376,
+      "step": 268830
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.419644907017678e-05,
+      "loss": 0.9809,
+      "step": 268840
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.4195301140277034e-05,
+      "loss": 1.113,
+      "step": 268850
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.419415321037728e-05,
+      "loss": 1.1527,
+      "step": 268860
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.4193005280477534e-05,
+      "loss": 1.0895,
+      "step": 268870
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.419185735057779e-05,
+      "loss": 1.0486,
+      "step": 268880
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.4190709420678034e-05,
+      "loss": 0.9933,
+      "step": 268890
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.418956149077829e-05,
+      "loss": 0.9535,
+      "step": 268900
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.418841356087855e-05,
+      "loss": 0.9292,
+      "step": 268910
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.41872656309788e-05,
+      "loss": 1.0421,
+      "step": 268920
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.418611770107905e-05,
+      "loss": 0.9938,
+      "step": 268930
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.4184969771179304e-05,
+      "loss": 1.0992,
+      "step": 268940
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.418382184127955e-05,
+      "loss": 1.1491,
+      "step": 268950
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.418267391137981e-05,
+      "loss": 1.1068,
+      "step": 268960
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.418152598148006e-05,
+      "loss": 1.0971,
+      "step": 268970
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.418037805158031e-05,
+      "loss": 1.0384,
+      "step": 268980
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.4179230121680566e-05,
+      "loss": 1.0877,
+      "step": 268990
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.417808219178082e-05,
+      "loss": 0.8881,
+      "step": 269000
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.4176934261881066e-05,
+      "loss": 1.0768,
+      "step": 269010
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.417578633198132e-05,
+      "loss": 1.0124,
+      "step": 269020
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.417463840208158e-05,
+      "loss": 0.9094,
+      "step": 269030
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.417349047218182e-05,
+      "loss": 1.0217,
+      "step": 269040
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.417234254228208e-05,
+      "loss": 1.0865,
+      "step": 269050
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.4171194612382336e-05,
+      "loss": 1.0012,
+      "step": 269060
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.4170046682482585e-05,
+      "loss": 0.9903,
+      "step": 269070
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.4168898752582835e-05,
+      "loss": 1.1509,
+      "step": 269080
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.416775082268309e-05,
+      "loss": 1.0264,
+      "step": 269090
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.416660289278334e-05,
+      "loss": 1.0812,
+      "step": 269100
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.41654549628836e-05,
+      "loss": 1.0336,
+      "step": 269110
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.416430703298385e-05,
+      "loss": 0.9683,
+      "step": 269120
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.41631591030841e-05,
+      "loss": 1.0434,
+      "step": 269130
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.4162011173184355e-05,
+      "loss": 0.9309,
+      "step": 269140
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.416086324328461e-05,
+      "loss": 1.0034,
+      "step": 269150
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.4159715313384855e-05,
+      "loss": 1.0433,
+      "step": 269160
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.415856738348511e-05,
+      "loss": 0.9566,
+      "step": 269170
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.415741945358537e-05,
+      "loss": 0.9532,
+      "step": 269180
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.415627152368561e-05,
+      "loss": 1.0254,
+      "step": 269190
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.415512359378587e-05,
+      "loss": 1.0093,
+      "step": 269200
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.4153975663886124e-05,
+      "loss": 1.031,
+      "step": 269210
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.4152827733986374e-05,
+      "loss": 0.9848,
+      "step": 269220
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.4151679804086624e-05,
+      "loss": 1.1639,
+      "step": 269230
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.415053187418688e-05,
+      "loss": 1.0415,
+      "step": 269240
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.414938394428713e-05,
+      "loss": 0.929,
+      "step": 269250
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.414823601438739e-05,
+      "loss": 0.9701,
+      "step": 269260
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.414708808448764e-05,
+      "loss": 1.0696,
+      "step": 269270
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.414594015458789e-05,
+      "loss": 0.978,
+      "step": 269280
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.414479222468814e-05,
+      "loss": 0.9704,
+      "step": 269290
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.41436442947884e-05,
+      "loss": 1.0239,
+      "step": 269300
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.414249636488864e-05,
+      "loss": 1.0266,
+      "step": 269310
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.41413484349889e-05,
+      "loss": 0.9249,
+      "step": 269320
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.4140200505089156e-05,
+      "loss": 0.9695,
+      "step": 269330
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.41390525751894e-05,
+      "loss": 0.9574,
+      "step": 269340
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.4137904645289656e-05,
+      "loss": 0.9755,
+      "step": 269350
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.413675671538991e-05,
+      "loss": 1.0364,
+      "step": 269360
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.413560878549016e-05,
+      "loss": 0.9562,
+      "step": 269370
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.413446085559041e-05,
+      "loss": 0.9705,
+      "step": 269380
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.413331292569067e-05,
+      "loss": 1.0076,
+      "step": 269390
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.413216499579092e-05,
+      "loss": 0.9654,
+      "step": 269400
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.4131017065891175e-05,
+      "loss": 0.9675,
+      "step": 269410
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.4129869135991425e-05,
+      "loss": 1.0181,
+      "step": 269420
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.4128721206091675e-05,
+      "loss": 1.0438,
+      "step": 269430
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.412757327619193e-05,
+      "loss": 1.017,
+      "step": 269440
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.412642534629219e-05,
+      "loss": 1.0534,
+      "step": 269450
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.412527741639243e-05,
+      "loss": 0.9476,
+      "step": 269460
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.412412948649269e-05,
+      "loss": 1.0982,
+      "step": 269470
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.4122981556592945e-05,
+      "loss": 1.0239,
+      "step": 269480
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.412183362669319e-05,
+      "loss": 0.9397,
+      "step": 269490
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.4120685696793444e-05,
+      "loss": 1.0226,
+      "step": 269500
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.41195377668937e-05,
+      "loss": 0.9321,
+      "step": 269510
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.411838983699395e-05,
+      "loss": 0.9851,
+      "step": 269520
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.41172419070942e-05,
+      "loss": 0.9325,
+      "step": 269530
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.411609397719446e-05,
+      "loss": 0.9407,
+      "step": 269540
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.411494604729471e-05,
+      "loss": 1.0043,
+      "step": 269550
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.4113798117394964e-05,
+      "loss": 0.9927,
+      "step": 269560
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.4112650187495214e-05,
+      "loss": 1.0636,
+      "step": 269570
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.4111502257595464e-05,
+      "loss": 0.9367,
+      "step": 269580
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.411035432769572e-05,
+      "loss": 1.1333,
+      "step": 269590
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.410920639779598e-05,
+      "loss": 0.9913,
+      "step": 269600
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.410805846789622e-05,
+      "loss": 0.9311,
+      "step": 269610
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.4106910537996477e-05,
+      "loss": 0.9577,
+      "step": 269620
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.410576260809673e-05,
+      "loss": 1.0785,
+      "step": 269630
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.4104614678196976e-05,
+      "loss": 0.9866,
+      "step": 269640
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.410346674829723e-05,
+      "loss": 0.89,
+      "step": 269650
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.410231881839749e-05,
+      "loss": 1.0169,
+      "step": 269660
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.410117088849774e-05,
+      "loss": 1.03,
+      "step": 269670
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.410002295859799e-05,
+      "loss": 1.0907,
+      "step": 269680
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.4098875028698246e-05,
+      "loss": 1.0295,
+      "step": 269690
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.4097727098798496e-05,
+      "loss": 1.0152,
+      "step": 269700
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.409657916889875e-05,
+      "loss": 1.1065,
+      "step": 269710
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.4095431238999e-05,
+      "loss": 0.9672,
+      "step": 269720
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.409428330909925e-05,
+      "loss": 1.0736,
+      "step": 269730
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.409313537919951e-05,
+      "loss": 0.9748,
+      "step": 269740
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.4091987449299765e-05,
+      "loss": 1.0061,
+      "step": 269750
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.409083951940001e-05,
+      "loss": 0.9718,
+      "step": 269760
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.4089691589500265e-05,
+      "loss": 0.9972,
+      "step": 269770
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.408854365960052e-05,
+      "loss": 1.0826,
+      "step": 269780
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.4087395729700765e-05,
+      "loss": 0.9934,
+      "step": 269790
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.408624779980102e-05,
+      "loss": 0.9656,
+      "step": 269800
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.408509986990128e-05,
+      "loss": 0.9932,
+      "step": 269810
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.408395194000153e-05,
+      "loss": 1.0023,
+      "step": 269820
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.408280401010178e-05,
+      "loss": 1.0544,
+      "step": 269830
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.4081656080202034e-05,
+      "loss": 1.0752,
+      "step": 269840
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.4080508150302284e-05,
+      "loss": 1.1014,
+      "step": 269850
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.407936022040254e-05,
+      "loss": 1.0899,
+      "step": 269860
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.407821229050279e-05,
+      "loss": 0.9599,
+      "step": 269870
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.407706436060304e-05,
+      "loss": 0.9887,
+      "step": 269880
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.40759164307033e-05,
+      "loss": 1.0968,
+      "step": 269890
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.4074768500803554e-05,
+      "loss": 0.9692,
+      "step": 269900
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.40736205709038e-05,
+      "loss": 0.9512,
+      "step": 269910
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.4072472641004054e-05,
+      "loss": 0.9348,
+      "step": 269920
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.407132471110431e-05,
+      "loss": 1.0207,
+      "step": 269930
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.407017678120455e-05,
+      "loss": 1.0616,
+      "step": 269940
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.406902885130481e-05,
+      "loss": 1.0482,
+      "step": 269950
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.4067880921405066e-05,
+      "loss": 0.9905,
+      "step": 269960
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.4066732991505316e-05,
+      "loss": 0.9729,
+      "step": 269970
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.4065585061605566e-05,
+      "loss": 0.8923,
+      "step": 269980
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.406443713170582e-05,
+      "loss": 1.0351,
+      "step": 269990
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.406328920180607e-05,
+      "loss": 1.0589,
+      "step": 270000
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.406214127190633e-05,
+      "loss": 1.0599,
+      "step": 270010
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.406099334200658e-05,
+      "loss": 1.1202,
+      "step": 270020
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.405984541210683e-05,
+      "loss": 0.9887,
+      "step": 270030
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.4058697482207086e-05,
+      "loss": 0.9294,
+      "step": 270040
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.405754955230734e-05,
+      "loss": 0.8961,
+      "step": 270050
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.4056401622407585e-05,
+      "loss": 1.0116,
+      "step": 270060
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.405525369250784e-05,
+      "loss": 1.079,
+      "step": 270070
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.40541057626081e-05,
+      "loss": 0.8753,
+      "step": 270080
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.405295783270834e-05,
+      "loss": 1.0055,
+      "step": 270090
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.40518099028086e-05,
+      "loss": 0.9802,
+      "step": 270100
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.4050661972908855e-05,
+      "loss": 1.0045,
+      "step": 270110
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.4049514043009105e-05,
+      "loss": 1.0019,
+      "step": 270120
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.4048366113109355e-05,
+      "loss": 1.0556,
+      "step": 270130
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.404721818320961e-05,
+      "loss": 1.16,
+      "step": 270140
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.404607025330986e-05,
+      "loss": 0.9881,
+      "step": 270150
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.404492232341012e-05,
+      "loss": 0.9821,
+      "step": 270160
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.404377439351037e-05,
+      "loss": 1.1339,
+      "step": 270170
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.404262646361062e-05,
+      "loss": 0.8898,
+      "step": 270180
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.4041478533710874e-05,
+      "loss": 1.0652,
+      "step": 270190
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.404033060381113e-05,
+      "loss": 1.0452,
+      "step": 270200
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.4039182673911374e-05,
+      "loss": 1.0047,
+      "step": 270210
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.403803474401163e-05,
+      "loss": 1.0176,
+      "step": 270220
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.4036886814111874e-05,
+      "loss": 0.9789,
+      "step": 270230
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.403573888421213e-05,
+      "loss": 1.085,
+      "step": 270240
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.403459095431239e-05,
+      "loss": 0.9182,
+      "step": 270250
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.403344302441264e-05,
+      "loss": 1.0221,
+      "step": 270260
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.403229509451289e-05,
+      "loss": 0.9732,
+      "step": 270270
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.403114716461314e-05,
+      "loss": 1.0566,
+      "step": 270280
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.402999923471339e-05,
+      "loss": 0.9479,
+      "step": 270290
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.402885130481365e-05,
+      "loss": 1.0077,
+      "step": 270300
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.4027703374913906e-05,
+      "loss": 1.0748,
+      "step": 270310
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.402655544501415e-05,
+      "loss": 1.0763,
+      "step": 270320
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.4025407515114406e-05,
+      "loss": 1.0344,
+      "step": 270330
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.402425958521466e-05,
+      "loss": 0.9011,
+      "step": 270340
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.4023111655314906e-05,
+      "loss": 1.0526,
+      "step": 270350
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.402196372541516e-05,
+      "loss": 1.0572,
+      "step": 270360
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.402081579551542e-05,
+      "loss": 0.9868,
+      "step": 270370
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.401966786561566e-05,
+      "loss": 0.9888,
+      "step": 270380
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.401851993571592e-05,
+      "loss": 0.918,
+      "step": 270390
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.4017372005816175e-05,
+      "loss": 0.9949,
+      "step": 270400
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.4016224075916425e-05,
+      "loss": 1.0162,
+      "step": 270410
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.401507614601668e-05,
+      "loss": 1.0567,
+      "step": 270420
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.401392821611693e-05,
+      "loss": 1.0383,
+      "step": 270430
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.401278028621718e-05,
+      "loss": 1.0785,
+      "step": 270440
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.401163235631744e-05,
+      "loss": 1.0558,
+      "step": 270450
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.4010484426417695e-05,
+      "loss": 0.9553,
+      "step": 270460
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.400933649651794e-05,
+      "loss": 1.03,
+      "step": 270470
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.4008188566618194e-05,
+      "loss": 0.9641,
+      "step": 270480
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.400704063671845e-05,
+      "loss": 0.9798,
+      "step": 270490
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.4005892706818694e-05,
+      "loss": 0.9317,
+      "step": 270500
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.400474477691895e-05,
+      "loss": 0.9872,
+      "step": 270510
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.400359684701921e-05,
+      "loss": 1.0006,
+      "step": 270520
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.400244891711945e-05,
+      "loss": 0.9565,
+      "step": 270530
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.400130098721971e-05,
+      "loss": 1.12,
+      "step": 270540
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.4000153057319964e-05,
+      "loss": 0.9225,
+      "step": 270550
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.3999005127420214e-05,
+      "loss": 1.0411,
+      "step": 270560
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.399785719752047e-05,
+      "loss": 1.0065,
+      "step": 270570
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.399670926762072e-05,
+      "loss": 1.0522,
+      "step": 270580
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.399556133772097e-05,
+      "loss": 0.9676,
+      "step": 270590
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.3994413407821227e-05,
+      "loss": 0.9755,
+      "step": 270600
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.399326547792148e-05,
+      "loss": 0.9745,
+      "step": 270610
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.3992117548021726e-05,
+      "loss": 0.9335,
+      "step": 270620
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.399096961812198e-05,
+      "loss": 0.9138,
+      "step": 270630
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.398982168822224e-05,
+      "loss": 0.9897,
+      "step": 270640
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.398867375832248e-05,
+      "loss": 1.0077,
+      "step": 270650
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.398752582842274e-05,
+      "loss": 1.0338,
+      "step": 270660
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.3986377898522996e-05,
+      "loss": 0.9828,
+      "step": 270670
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.3985229968623246e-05,
+      "loss": 1.0461,
+      "step": 270680
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.3984082038723496e-05,
+      "loss": 1.0097,
+      "step": 270690
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.398293410882375e-05,
+      "loss": 0.9421,
+      "step": 270700
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.3981786178924e-05,
+      "loss": 0.9816,
+      "step": 270710
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.398063824902426e-05,
+      "loss": 1.0093,
+      "step": 270720
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.397949031912451e-05,
+      "loss": 0.9596,
+      "step": 270730
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.397834238922476e-05,
+      "loss": 0.9716,
+      "step": 270740
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.3977194459325015e-05,
+      "loss": 0.9662,
+      "step": 270750
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.397604652942527e-05,
+      "loss": 1.0319,
+      "step": 270760
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.3974898599525515e-05,
+      "loss": 1.0843,
+      "step": 270770
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.397375066962577e-05,
+      "loss": 1.081,
+      "step": 270780
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.397260273972603e-05,
+      "loss": 0.9453,
+      "step": 270790
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.397145480982627e-05,
+      "loss": 0.9608,
+      "step": 270800
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.397030687992653e-05,
+      "loss": 1.0532,
+      "step": 270810
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.3969158950026784e-05,
+      "loss": 0.9623,
+      "step": 270820
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.3968011020127034e-05,
+      "loss": 1.0505,
+      "step": 270830
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.3966863090227284e-05,
+      "loss": 1.0219,
+      "step": 270840
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.396571516032754e-05,
+      "loss": 0.9291,
+      "step": 270850
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.396456723042779e-05,
+      "loss": 0.9266,
+      "step": 270860
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.396341930052805e-05,
+      "loss": 0.9527,
+      "step": 270870
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.39622713706283e-05,
+      "loss": 0.9679,
+      "step": 270880
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.396112344072855e-05,
+      "loss": 1.0022,
+      "step": 270890
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.3959975510828804e-05,
+      "loss": 0.9067,
+      "step": 270900
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.395882758092906e-05,
+      "loss": 1.0835,
+      "step": 270910
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.39576796510293e-05,
+      "loss": 1.0679,
+      "step": 270920
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.395653172112956e-05,
+      "loss": 1.036,
+      "step": 270930
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.3955383791229816e-05,
+      "loss": 1.0843,
+      "step": 270940
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.395423586133006e-05,
+      "loss": 0.8979,
+      "step": 270950
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.3953087931430316e-05,
+      "loss": 1.0342,
+      "step": 270960
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.395194000153057e-05,
+      "loss": 1.0626,
+      "step": 270970
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.395079207163082e-05,
+      "loss": 0.8844,
+      "step": 270980
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.394964414173107e-05,
+      "loss": 0.949,
+      "step": 270990
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.394849621183133e-05,
+      "loss": 1.0749,
+      "step": 271000
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.394734828193158e-05,
+      "loss": 0.9384,
+      "step": 271010
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.3946200352031836e-05,
+      "loss": 0.9638,
+      "step": 271020
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.3945052422132085e-05,
+      "loss": 1.0915,
+      "step": 271030
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.3943904492232335e-05,
+      "loss": 0.9266,
+      "step": 271040
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.394275656233259e-05,
+      "loss": 1.033,
+      "step": 271050
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.394160863243285e-05,
+      "loss": 0.9507,
+      "step": 271060
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.394046070253309e-05,
+      "loss": 1.0732,
+      "step": 271070
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.393931277263335e-05,
+      "loss": 1.05,
+      "step": 271080
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.3938164842733605e-05,
+      "loss": 0.9797,
+      "step": 271090
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.393701691283385e-05,
+      "loss": 1.0877,
+      "step": 271100
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.3935868982934105e-05,
+      "loss": 1.0789,
+      "step": 271110
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.393472105303436e-05,
+      "loss": 1.0003,
+      "step": 271120
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.393357312313461e-05,
+      "loss": 1.0941,
+      "step": 271130
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.393242519323486e-05,
+      "loss": 0.988,
+      "step": 271140
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.393127726333512e-05,
+      "loss": 0.9333,
+      "step": 271150
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.393012933343537e-05,
+      "loss": 0.9935,
+      "step": 271160
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.3928981403535624e-05,
+      "loss": 1.0021,
+      "step": 271170
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.3927833473635874e-05,
+      "loss": 0.8105,
+      "step": 271180
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.3926685543736124e-05,
+      "loss": 0.977,
+      "step": 271190
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.392553761383638e-05,
+      "loss": 1.017,
+      "step": 271200
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.392438968393664e-05,
+      "loss": 1.0028,
+      "step": 271210
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.392324175403688e-05,
+      "loss": 1.0365,
+      "step": 271220
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.392209382413714e-05,
+      "loss": 1.123,
+      "step": 271230
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.3920945894237393e-05,
+      "loss": 0.8572,
+      "step": 271240
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.3919797964337636e-05,
+      "loss": 0.978,
+      "step": 271250
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.391865003443789e-05,
+      "loss": 1.028,
+      "step": 271260
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.391750210453815e-05,
+      "loss": 0.9167,
+      "step": 271270
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.39163541746384e-05,
+      "loss": 0.9669,
+      "step": 271280
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.391520624473865e-05,
+      "loss": 0.9121,
+      "step": 271290
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.3914058314838906e-05,
+      "loss": 0.9506,
+      "step": 271300
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.3912910384939156e-05,
+      "loss": 1.0249,
+      "step": 271310
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.391176245503941e-05,
+      "loss": 0.9027,
+      "step": 271320
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.391061452513966e-05,
+      "loss": 1.1618,
+      "step": 271330
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.390946659523991e-05,
+      "loss": 0.9435,
+      "step": 271340
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.390831866534017e-05,
+      "loss": 1.0636,
+      "step": 271350
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.3907170735440426e-05,
+      "loss": 0.981,
+      "step": 271360
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.390602280554067e-05,
+      "loss": 0.9187,
+      "step": 271370
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.3904874875640925e-05,
+      "loss": 0.9805,
+      "step": 271380
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.390372694574118e-05,
+      "loss": 0.9815,
+      "step": 271390
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.3902579015841425e-05,
+      "loss": 0.9314,
+      "step": 271400
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.390143108594168e-05,
+      "loss": 1.0256,
+      "step": 271410
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.390028315604194e-05,
+      "loss": 1.1147,
+      "step": 271420
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.389913522614219e-05,
+      "loss": 0.99,
+      "step": 271430
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.389798729624244e-05,
+      "loss": 0.9515,
+      "step": 271440
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.3896839366342695e-05,
+      "loss": 0.9609,
+      "step": 271450
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.3895691436442944e-05,
+      "loss": 0.9264,
+      "step": 271460
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.38945435065432e-05,
+      "loss": 0.9831,
+      "step": 271470
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.389339557664345e-05,
+      "loss": 0.96,
+      "step": 271480
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.38922476467437e-05,
+      "loss": 0.9994,
+      "step": 271490
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.389109971684396e-05,
+      "loss": 0.9833,
+      "step": 271500
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.3889951786944214e-05,
+      "loss": 1.0066,
+      "step": 271510
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.388880385704446e-05,
+      "loss": 0.9741,
+      "step": 271520
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.3887655927144714e-05,
+      "loss": 0.9594,
+      "step": 271530
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.388650799724497e-05,
+      "loss": 1.0245,
+      "step": 271540
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.3885360067345213e-05,
+      "loss": 1.0741,
+      "step": 271550
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.388421213744547e-05,
+      "loss": 0.9321,
+      "step": 271560
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.388306420754573e-05,
+      "loss": 0.9805,
+      "step": 271570
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.3881916277645977e-05,
+      "loss": 0.9816,
+      "step": 271580
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.3880768347746226e-05,
+      "loss": 1.0736,
+      "step": 271590
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.3879620417846476e-05,
+      "loss": 1.0676,
+      "step": 271600
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.387847248794673e-05,
+      "loss": 0.8932,
+      "step": 271610
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.387732455804699e-05,
+      "loss": 1.0304,
+      "step": 271620
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.387617662814723e-05,
+      "loss": 0.9859,
+      "step": 271630
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.387502869824749e-05,
+      "loss": 0.9923,
+      "step": 271640
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.3873880768347746e-05,
+      "loss": 0.9665,
+      "step": 271650
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.387273283844799e-05,
+      "loss": 1.1325,
+      "step": 271660
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.3871584908548246e-05,
+      "loss": 0.9669,
+      "step": 271670
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.38704369786485e-05,
+      "loss": 0.9983,
+      "step": 271680
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.3869289048748745e-05,
+      "loss": 1.0297,
+      "step": 271690
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.3868141118849e-05,
+      "loss": 1.0081,
+      "step": 271700
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.386699318894926e-05,
+      "loss": 1.0493,
+      "step": 271710
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.386584525904951e-05,
+      "loss": 1.0401,
+      "step": 271720
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.3864697329149765e-05,
+      "loss": 1.0413,
+      "step": 271730
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.3863549399250015e-05,
+      "loss": 1.0577,
+      "step": 271740
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.3862401469350265e-05,
+      "loss": 1.0056,
+      "step": 271750
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.386125353945052e-05,
+      "loss": 0.997,
+      "step": 271760
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.386010560955078e-05,
+      "loss": 1.0132,
+      "step": 271770
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.385895767965102e-05,
+      "loss": 1.0945,
+      "step": 271780
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.385780974975128e-05,
+      "loss": 1.0224,
+      "step": 271790
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.3856661819851534e-05,
+      "loss": 0.9921,
+      "step": 271800
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.385551388995178e-05,
+      "loss": 1.0144,
+      "step": 271810
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.3854365960052034e-05,
+      "loss": 0.9838,
+      "step": 271820
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.385321803015229e-05,
+      "loss": 0.8714,
+      "step": 271830
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.3852070100252534e-05,
+      "loss": 0.9001,
+      "step": 271840
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.385092217035279e-05,
+      "loss": 1.0569,
+      "step": 271850
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.384977424045305e-05,
+      "loss": 0.9712,
+      "step": 271860
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.38486263105533e-05,
+      "loss": 0.9744,
+      "step": 271870
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.3847478380653553e-05,
+      "loss": 1.0398,
+      "step": 271880
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.38463304507538e-05,
+      "loss": 1.0222,
+      "step": 271890
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.384518252085405e-05,
+      "loss": 1.0334,
+      "step": 271900
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.384403459095431e-05,
+      "loss": 0.94,
+      "step": 271910
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.3842886661054566e-05,
+      "loss": 0.9809,
+      "step": 271920
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.384173873115481e-05,
+      "loss": 1.0952,
+      "step": 271930
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.3840590801255066e-05,
+      "loss": 0.991,
+      "step": 271940
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.383944287135532e-05,
+      "loss": 1.0096,
+      "step": 271950
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.3838294941455566e-05,
+      "loss": 1.099,
+      "step": 271960
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.383714701155582e-05,
+      "loss": 1.0929,
+      "step": 271970
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.383599908165608e-05,
+      "loss": 0.9572,
+      "step": 271980
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.383485115175632e-05,
+      "loss": 1.0664,
+      "step": 271990
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.383370322185658e-05,
+      "loss": 1.0182,
+      "step": 272000
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.3832555291956835e-05,
+      "loss": 1.0054,
+      "step": 272010
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.3831407362057085e-05,
+      "loss": 1.0794,
+      "step": 272020
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.383025943215734e-05,
+      "loss": 1.0183,
+      "step": 272030
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.382911150225759e-05,
+      "loss": 0.9615,
+      "step": 272040
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.382796357235784e-05,
+      "loss": 0.9608,
+      "step": 272050
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.38268156424581e-05,
+      "loss": 0.9808,
+      "step": 272060
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.3825667712558355e-05,
+      "loss": 0.9699,
+      "step": 272070
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.38245197826586e-05,
+      "loss": 1.0277,
+      "step": 272080
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.3823371852758855e-05,
+      "loss": 0.9999,
+      "step": 272090
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.382222392285911e-05,
+      "loss": 0.9492,
+      "step": 272100
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.3821075992959354e-05,
+      "loss": 1.0228,
+      "step": 272110
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.381992806305961e-05,
+      "loss": 1.0093,
+      "step": 272120
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.381878013315987e-05,
+      "loss": 1.0355,
+      "step": 272130
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.381763220326011e-05,
+      "loss": 0.9622,
+      "step": 272140
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.381648427336037e-05,
+      "loss": 0.9684,
+      "step": 272150
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.3815336343460624e-05,
+      "loss": 1.0573,
+      "step": 272160
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.3814188413560874e-05,
+      "loss": 1.0907,
+      "step": 272170
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.381304048366113e-05,
+      "loss": 0.9658,
+      "step": 272180
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.381189255376138e-05,
+      "loss": 0.9466,
+      "step": 272190
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.381074462386163e-05,
+      "loss": 1.1419,
+      "step": 272200
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.380959669396189e-05,
+      "loss": 1.0366,
+      "step": 272210
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.3808448764062143e-05,
+      "loss": 0.9963,
+      "step": 272220
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.3807300834162386e-05,
+      "loss": 0.9535,
+      "step": 272230
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.380615290426264e-05,
+      "loss": 1.0099,
+      "step": 272240
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.38050049743629e-05,
+      "loss": 1.0359,
+      "step": 272250
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.380385704446314e-05,
+      "loss": 1.0949,
+      "step": 272260
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.38027091145634e-05,
+      "loss": 1.051,
+      "step": 272270
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.3801561184663656e-05,
+      "loss": 0.8922,
+      "step": 272280
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.38004132547639e-05,
+      "loss": 1.0596,
+      "step": 272290
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.3799265324864156e-05,
+      "loss": 1.0834,
+      "step": 272300
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.379811739496441e-05,
+      "loss": 1.0909,
+      "step": 272310
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.379696946506466e-05,
+      "loss": 0.9714,
+      "step": 272320
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.379582153516492e-05,
+      "loss": 0.9734,
+      "step": 272330
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.379467360526517e-05,
+      "loss": 1.0297,
+      "step": 272340
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.379352567536542e-05,
+      "loss": 0.9144,
+      "step": 272350
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.3792377745465675e-05,
+      "loss": 1.0494,
+      "step": 272360
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.379122981556593e-05,
+      "loss": 1.1301,
+      "step": 272370
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.3790081885666175e-05,
+      "loss": 0.9559,
+      "step": 272380
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.378893395576643e-05,
+      "loss": 1.0076,
+      "step": 272390
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.378778602586669e-05,
+      "loss": 1.0713,
+      "step": 272400
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.378663809596693e-05,
+      "loss": 0.983,
+      "step": 272410
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.378549016606719e-05,
+      "loss": 1.0165,
+      "step": 272420
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.3784342236167445e-05,
+      "loss": 0.9463,
+      "step": 272430
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.378319430626769e-05,
+      "loss": 1.025,
+      "step": 272440
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.3782046376367944e-05,
+      "loss": 1.0229,
+      "step": 272450
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.37808984464682e-05,
+      "loss": 0.9836,
+      "step": 272460
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.377975051656845e-05,
+      "loss": 0.9579,
+      "step": 272470
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.377860258666871e-05,
+      "loss": 0.927,
+      "step": 272480
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.377745465676896e-05,
+      "loss": 0.9951,
+      "step": 272490
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.377630672686921e-05,
+      "loss": 1.0079,
+      "step": 272500
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.3775158796969464e-05,
+      "loss": 0.9394,
+      "step": 272510
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.377401086706972e-05,
+      "loss": 1.024,
+      "step": 272520
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.3772862937169963e-05,
+      "loss": 0.9548,
+      "step": 272530
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.377171500727022e-05,
+      "loss": 1.0084,
+      "step": 272540
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.377056707737048e-05,
+      "loss": 1.0744,
+      "step": 272550
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.376941914747072e-05,
+      "loss": 1.0648,
+      "step": 272560
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.3768271217570976e-05,
+      "loss": 1.003,
+      "step": 272570
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.376712328767123e-05,
+      "loss": 0.9579,
+      "step": 272580
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.3765975357771476e-05,
+      "loss": 0.991,
+      "step": 272590
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.376482742787173e-05,
+      "loss": 0.9177,
+      "step": 272600
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.376367949797199e-05,
+      "loss": 0.9307,
+      "step": 272610
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.376253156807224e-05,
+      "loss": 0.9516,
+      "step": 272620
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.3761383638172496e-05,
+      "loss": 1.0356,
+      "step": 272630
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.3760235708272746e-05,
+      "loss": 0.9637,
+      "step": 272640
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.3759087778372996e-05,
+      "loss": 1.0467,
+      "step": 272650
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.375793984847325e-05,
+      "loss": 1.0882,
+      "step": 272660
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.375679191857351e-05,
+      "loss": 0.9512,
+      "step": 272670
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.375564398867375e-05,
+      "loss": 0.9824,
+      "step": 272680
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.375449605877401e-05,
+      "loss": 1.0368,
+      "step": 272690
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.3753348128874265e-05,
+      "loss": 0.9779,
+      "step": 272700
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.375220019897451e-05,
+      "loss": 1.0474,
+      "step": 272710
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.3751052269074765e-05,
+      "loss": 1.0585,
+      "step": 272720
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.374990433917502e-05,
+      "loss": 1.082,
+      "step": 272730
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.3748756409275265e-05,
+      "loss": 1.0494,
+      "step": 272740
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.374760847937552e-05,
+      "loss": 0.9966,
+      "step": 272750
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.374646054947578e-05,
+      "loss": 0.9994,
+      "step": 272760
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.374531261957603e-05,
+      "loss": 0.9323,
+      "step": 272770
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.3744164689676284e-05,
+      "loss": 1.0219,
+      "step": 272780
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.3743016759776534e-05,
+      "loss": 1.016,
+      "step": 272790
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.3741868829876784e-05,
+      "loss": 0.9781,
+      "step": 272800
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.374072089997704e-05,
+      "loss": 0.9841,
+      "step": 272810
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.37395729700773e-05,
+      "loss": 0.9723,
+      "step": 272820
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.373842504017754e-05,
+      "loss": 0.9615,
+      "step": 272830
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.37372771102778e-05,
+      "loss": 1.1287,
+      "step": 272840
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.3736129180378054e-05,
+      "loss": 0.9588,
+      "step": 272850
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.37349812504783e-05,
+      "loss": 1.0955,
+      "step": 272860
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.373383332057855e-05,
+      "loss": 1.0189,
+      "step": 272870
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.373268539067881e-05,
+      "loss": 0.9472,
+      "step": 272880
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.373153746077905e-05,
+      "loss": 1.0259,
+      "step": 272890
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.373038953087931e-05,
+      "loss": 0.9086,
+      "step": 272900
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.3729241600979566e-05,
+      "loss": 1.1282,
+      "step": 272910
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.3728093671079816e-05,
+      "loss": 0.9751,
+      "step": 272920
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.372694574118007e-05,
+      "loss": 1.0435,
+      "step": 272930
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.372579781128032e-05,
+      "loss": 0.9466,
+      "step": 272940
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.372464988138057e-05,
+      "loss": 0.9592,
+      "step": 272950
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.372350195148083e-05,
+      "loss": 0.9365,
+      "step": 272960
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.372235402158107e-05,
+      "loss": 0.9955,
+      "step": 272970
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.372120609168133e-05,
+      "loss": 1.0549,
+      "step": 272980
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.3720058161781585e-05,
+      "loss": 1.0438,
+      "step": 272990
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.371891023188183e-05,
+      "loss": 1.0055,
+      "step": 273000
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.3717762301982085e-05,
+      "loss": 0.9321,
+      "step": 273010
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.371661437208234e-05,
+      "loss": 1.021,
+      "step": 273020
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.371546644218259e-05,
+      "loss": 1.0816,
+      "step": 273030
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.371431851228284e-05,
+      "loss": 0.9879,
+      "step": 273040
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.37131705823831e-05,
+      "loss": 1.1437,
+      "step": 273050
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.371202265248335e-05,
+      "loss": 0.9993,
+      "step": 273060
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.3710874722583605e-05,
+      "loss": 0.9791,
+      "step": 273070
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.370972679268386e-05,
+      "loss": 1.0395,
+      "step": 273080
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.3708578862784104e-05,
+      "loss": 0.9751,
+      "step": 273090
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.370743093288436e-05,
+      "loss": 0.9359,
+      "step": 273100
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.370628300298462e-05,
+      "loss": 1.0919,
+      "step": 273110
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.370513507308486e-05,
+      "loss": 1.1433,
+      "step": 273120
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.370398714318512e-05,
+      "loss": 1.1262,
+      "step": 273130
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.3702839213285374e-05,
+      "loss": 0.9252,
+      "step": 273140
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.370169128338562e-05,
+      "loss": 1.0199,
+      "step": 273150
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.3700543353485874e-05,
+      "loss": 0.8997,
+      "step": 273160
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.369939542358613e-05,
+      "loss": 0.9689,
+      "step": 273170
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.369824749368638e-05,
+      "loss": 1.0632,
+      "step": 273180
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.369709956378663e-05,
+      "loss": 1.0393,
+      "step": 273190
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.369595163388689e-05,
+      "loss": 1.0177,
+      "step": 273200
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.3694803703987136e-05,
+      "loss": 0.9979,
+      "step": 273210
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.369365577408739e-05,
+      "loss": 1.0835,
+      "step": 273220
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.369250784418765e-05,
+      "loss": 0.9501,
+      "step": 273230
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.369135991428789e-05,
+      "loss": 1.1049,
+      "step": 273240
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.369021198438815e-05,
+      "loss": 0.9322,
+      "step": 273250
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.3689064054488406e-05,
+      "loss": 0.9956,
+      "step": 273260
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.368791612458865e-05,
+      "loss": 1.1248,
+      "step": 273270
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.3686768194688906e-05,
+      "loss": 1.0698,
+      "step": 273280
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.368562026478916e-05,
+      "loss": 1.011,
+      "step": 273290
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.3684472334889406e-05,
+      "loss": 0.9831,
+      "step": 273300
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.368332440498966e-05,
+      "loss": 0.903,
+      "step": 273310
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.368217647508992e-05,
+      "loss": 0.8945,
+      "step": 273320
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.368102854519017e-05,
+      "loss": 0.9901,
+      "step": 273330
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.367988061529042e-05,
+      "loss": 1.0764,
+      "step": 273340
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.3678732685390675e-05,
+      "loss": 1.0943,
+      "step": 273350
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.3677584755490925e-05,
+      "loss": 0.9974,
+      "step": 273360
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.367643682559118e-05,
+      "loss": 0.9788,
+      "step": 273370
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.367528889569144e-05,
+      "loss": 0.9782,
+      "step": 273380
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.367414096579168e-05,
+      "loss": 0.9796,
+      "step": 273390
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.367299303589194e-05,
+      "loss": 1.0425,
+      "step": 273400
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.3671845105992195e-05,
+      "loss": 0.9652,
+      "step": 273410
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.367069717609244e-05,
+      "loss": 1.0117,
+      "step": 273420
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.3669549246192694e-05,
+      "loss": 1.0255,
+      "step": 273430
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.366840131629295e-05,
+      "loss": 0.8851,
+      "step": 273440
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.3667253386393194e-05,
+      "loss": 1.0899,
+      "step": 273450
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.366610545649345e-05,
+      "loss": 0.9833,
+      "step": 273460
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.366495752659371e-05,
+      "loss": 1.0085,
+      "step": 273470
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.366380959669396e-05,
+      "loss": 1.0623,
+      "step": 273480
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.366266166679421e-05,
+      "loss": 1.0679,
+      "step": 273490
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.3661513736894464e-05,
+      "loss": 0.9938,
+      "step": 273500
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.3660365806994713e-05,
+      "loss": 1.082,
+      "step": 273510
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.365921787709497e-05,
+      "loss": 0.9866,
+      "step": 273520
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.365806994719523e-05,
+      "loss": 1.005,
+      "step": 273530
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.365692201729547e-05,
+      "loss": 0.9433,
+      "step": 273540
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.3655774087395726e-05,
+      "loss": 1.0811,
+      "step": 273550
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.365462615749598e-05,
+      "loss": 1.0744,
+      "step": 273560
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.3653478227596226e-05,
+      "loss": 0.9934,
+      "step": 273570
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.365233029769648e-05,
+      "loss": 0.9364,
+      "step": 273580
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.365118236779674e-05,
+      "loss": 0.9169,
+      "step": 273590
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.365003443789698e-05,
+      "loss": 0.9399,
+      "step": 273600
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.364888650799724e-05,
+      "loss": 0.9778,
+      "step": 273610
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.3647738578097496e-05,
+      "loss": 1.0443,
+      "step": 273620
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.3646590648197746e-05,
+      "loss": 0.9577,
+      "step": 273630
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.3645442718297995e-05,
+      "loss": 1.0548,
+      "step": 273640
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.364429478839825e-05,
+      "loss": 1.058,
+      "step": 273650
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.36431468584985e-05,
+      "loss": 1.1809,
+      "step": 273660
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.364199892859876e-05,
+      "loss": 0.9756,
+      "step": 273670
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.3640850998699015e-05,
+      "loss": 0.9551,
+      "step": 273680
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.363970306879926e-05,
+      "loss": 1.0694,
+      "step": 273690
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.3638555138899515e-05,
+      "loss": 1.0432,
+      "step": 273700
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.363740720899977e-05,
+      "loss": 0.9424,
+      "step": 273710
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.3636259279100015e-05,
+      "loss": 1.0823,
+      "step": 273720
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.363511134920027e-05,
+      "loss": 1.0278,
+      "step": 273730
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.363396341930053e-05,
+      "loss": 0.9398,
+      "step": 273740
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.363281548940077e-05,
+      "loss": 1.1029,
+      "step": 273750
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.363166755950103e-05,
+      "loss": 0.9508,
+      "step": 273760
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.3630519629601284e-05,
+      "loss": 1.105,
+      "step": 273770
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.3629371699701534e-05,
+      "loss": 1.0016,
+      "step": 273780
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.3628223769801784e-05,
+      "loss": 1.0938,
+      "step": 273790
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.362707583990204e-05,
+      "loss": 0.9858,
+      "step": 273800
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.362592791000229e-05,
+      "loss": 1.0311,
+      "step": 273810
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.362477998010255e-05,
+      "loss": 0.9942,
+      "step": 273820
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.3623632050202804e-05,
+      "loss": 0.9304,
+      "step": 273830
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.362248412030305e-05,
+      "loss": 1.0336,
+      "step": 273840
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.36213361904033e-05,
+      "loss": 1.0291,
+      "step": 273850
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.362018826050356e-05,
+      "loss": 1.0827,
+      "step": 273860
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.36190403306038e-05,
+      "loss": 0.9832,
+      "step": 273870
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.361789240070406e-05,
+      "loss": 1.04,
+      "step": 273880
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.3616744470804316e-05,
+      "loss": 0.9448,
+      "step": 273890
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.361559654090456e-05,
+      "loss": 1.047,
+      "step": 273900
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.3614448611004816e-05,
+      "loss": 1.045,
+      "step": 273910
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.361330068110507e-05,
+      "loss": 1.0201,
+      "step": 273920
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.361215275120532e-05,
+      "loss": 1.019,
+      "step": 273930
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.361100482130557e-05,
+      "loss": 1.0592,
+      "step": 273940
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.360985689140583e-05,
+      "loss": 1.0693,
+      "step": 273950
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.360870896150608e-05,
+      "loss": 1.0275,
+      "step": 273960
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.3607561031606335e-05,
+      "loss": 0.9693,
+      "step": 273970
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.360641310170659e-05,
+      "loss": 1.009,
+      "step": 273980
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.3605265171806835e-05,
+      "loss": 0.9988,
+      "step": 273990
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.360411724190709e-05,
+      "loss": 1.0271,
+      "step": 274000
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.360296931200735e-05,
+      "loss": 1.0224,
+      "step": 274010
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.360182138210759e-05,
+      "loss": 1.0897,
+      "step": 274020
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.360067345220785e-05,
+      "loss": 1.0713,
+      "step": 274030
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.3599525522308105e-05,
+      "loss": 1.0174,
+      "step": 274040
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.359837759240835e-05,
+      "loss": 0.958,
+      "step": 274050
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.3597229662508604e-05,
+      "loss": 0.9268,
+      "step": 274060
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.359608173260886e-05,
+      "loss": 0.9438,
+      "step": 274070
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.359493380270911e-05,
+      "loss": 1.0988,
+      "step": 274080
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.359378587280936e-05,
+      "loss": 1.1065,
+      "step": 274090
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.359263794290962e-05,
+      "loss": 0.972,
+      "step": 274100
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.359149001300987e-05,
+      "loss": 1.0131,
+      "step": 274110
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.3590342083110124e-05,
+      "loss": 0.9896,
+      "step": 274120
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.358919415321038e-05,
+      "loss": 0.9801,
+      "step": 274130
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.3588046223310624e-05,
+      "loss": 1.0917,
+      "step": 274140
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.358689829341088e-05,
+      "loss": 1.0234,
+      "step": 274150
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.358575036351114e-05,
+      "loss": 1.1481,
+      "step": 274160
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.358460243361138e-05,
+      "loss": 1.0024,
+      "step": 274170
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.358345450371164e-05,
+      "loss": 1.0327,
+      "step": 274180
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.358230657381189e-05,
+      "loss": 0.9891,
+      "step": 274190
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.3581158643912136e-05,
+      "loss": 0.9873,
+      "step": 274200
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.358001071401239e-05,
+      "loss": 1.0651,
+      "step": 274210
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.357886278411265e-05,
+      "loss": 0.9675,
+      "step": 274220
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.35777148542129e-05,
+      "loss": 1.0766,
+      "step": 274230
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.357656692431315e-05,
+      "loss": 0.9315,
+      "step": 274240
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.3575418994413406e-05,
+      "loss": 1.0123,
+      "step": 274250
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.3574271064513656e-05,
+      "loss": 0.9485,
+      "step": 274260
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.357312313461391e-05,
+      "loss": 0.938,
+      "step": 274270
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.357197520471417e-05,
+      "loss": 0.9859,
+      "step": 274280
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.357082727481441e-05,
+      "loss": 0.8801,
+      "step": 274290
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.356967934491467e-05,
+      "loss": 1.0035,
+      "step": 274300
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.3568531415014925e-05,
+      "loss": 1.0129,
+      "step": 274310
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.356738348511517e-05,
+      "loss": 1.1043,
+      "step": 274320
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.3566235555215425e-05,
+      "loss": 1.0139,
+      "step": 274330
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.3565087625315675e-05,
+      "loss": 0.976,
+      "step": 274340
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.3563939695415925e-05,
+      "loss": 1.0733,
+      "step": 274350
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.356279176551618e-05,
+      "loss": 1.0121,
+      "step": 274360
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.356164383561643e-05,
+      "loss": 0.9958,
+      "step": 274370
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.356049590571669e-05,
+      "loss": 1.0212,
+      "step": 274380
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.355934797581694e-05,
+      "loss": 1.0466,
+      "step": 274390
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.355820004591719e-05,
+      "loss": 1.0411,
+      "step": 274400
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.3557052116017444e-05,
+      "loss": 1.0247,
+      "step": 274410
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.35559041861177e-05,
+      "loss": 1.0329,
+      "step": 274420
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.3554756256217944e-05,
+      "loss": 0.9868,
+      "step": 274430
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.35536083263182e-05,
+      "loss": 0.9979,
+      "step": 274440
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.355246039641846e-05,
+      "loss": 0.9502,
+      "step": 274450
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.35513124665187e-05,
+      "loss": 1.0226,
+      "step": 274460
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.355016453661896e-05,
+      "loss": 1.0456,
+      "step": 274470
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.3549016606719214e-05,
+      "loss": 0.9399,
+      "step": 274480
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.3547868676819463e-05,
+      "loss": 0.8656,
+      "step": 274490
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.354672074691971e-05,
+      "loss": 1.0551,
+      "step": 274500
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.354557281701997e-05,
+      "loss": 1.093,
+      "step": 274510
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.354442488712022e-05,
+      "loss": 0.9623,
+      "step": 274520
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.3543276957220476e-05,
+      "loss": 0.9431,
+      "step": 274530
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.3542129027320726e-05,
+      "loss": 0.9459,
+      "step": 274540
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.3540981097420976e-05,
+      "loss": 1.0832,
+      "step": 274550
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.353983316752123e-05,
+      "loss": 1.0746,
+      "step": 274560
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.353868523762149e-05,
+      "loss": 1.0108,
+      "step": 274570
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.353753730772173e-05,
+      "loss": 1.1004,
+      "step": 274580
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.353638937782199e-05,
+      "loss": 0.9417,
+      "step": 274590
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.3535241447922246e-05,
+      "loss": 0.9638,
+      "step": 274600
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.353409351802249e-05,
+      "loss": 1.0196,
+      "step": 274610
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.3532945588122745e-05,
+      "loss": 1.0105,
+      "step": 274620
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.3531797658223e-05,
+      "loss": 1.0383,
+      "step": 274630
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.353064972832325e-05,
+      "loss": 1.0186,
+      "step": 274640
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.35295017984235e-05,
+      "loss": 1.0021,
+      "step": 274650
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.352835386852376e-05,
+      "loss": 0.991,
+      "step": 274660
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.352720593862401e-05,
+      "loss": 1.0806,
+      "step": 274670
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.3526058008724265e-05,
+      "loss": 1.1366,
+      "step": 274680
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.3524910078824515e-05,
+      "loss": 0.9955,
+      "step": 274690
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.3523762148924765e-05,
+      "loss": 1.0632,
+      "step": 274700
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.352261421902502e-05,
+      "loss": 1.1289,
+      "step": 274710
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.352146628912528e-05,
+      "loss": 1.0889,
+      "step": 274720
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.352031835922552e-05,
+      "loss": 1.0272,
+      "step": 274730
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.351917042932578e-05,
+      "loss": 0.8721,
+      "step": 274740
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.3518022499426034e-05,
+      "loss": 0.961,
+      "step": 274750
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.351687456952628e-05,
+      "loss": 1.0497,
+      "step": 274760
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.3515726639626534e-05,
+      "loss": 1.0538,
+      "step": 274770
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.351457870972679e-05,
+      "loss": 1.0427,
+      "step": 274780
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.351343077982704e-05,
+      "loss": 1.0394,
+      "step": 274790
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.351228284992729e-05,
+      "loss": 0.9791,
+      "step": 274800
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.351113492002755e-05,
+      "loss": 0.8932,
+      "step": 274810
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.35099869901278e-05,
+      "loss": 0.9673,
+      "step": 274820
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.350883906022805e-05,
+      "loss": 1.0592,
+      "step": 274830
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.35076911303283e-05,
+      "loss": 0.9746,
+      "step": 274840
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.350654320042855e-05,
+      "loss": 1.0877,
+      "step": 274850
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.350539527052881e-05,
+      "loss": 0.9818,
+      "step": 274860
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.3504247340629066e-05,
+      "loss": 1.0322,
+      "step": 274870
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.350309941072931e-05,
+      "loss": 1.0613,
+      "step": 274880
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.3501951480829566e-05,
+      "loss": 1.0184,
+      "step": 274890
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.350080355092982e-05,
+      "loss": 0.9491,
+      "step": 274900
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.3499655621030066e-05,
+      "loss": 1.0779,
+      "step": 274910
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.349850769113032e-05,
+      "loss": 0.9319,
+      "step": 274920
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.349735976123058e-05,
+      "loss": 0.9791,
+      "step": 274930
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.349621183133083e-05,
+      "loss": 0.9626,
+      "step": 274940
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.349506390143108e-05,
+      "loss": 1.193,
+      "step": 274950
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.3493915971531335e-05,
+      "loss": 0.9468,
+      "step": 274960
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.3492768041631585e-05,
+      "loss": 1.0375,
+      "step": 274970
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.349162011173184e-05,
+      "loss": 1.0142,
+      "step": 274980
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.349047218183209e-05,
+      "loss": 1.0509,
+      "step": 274990
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.348932425193234e-05,
+      "loss": 0.9717,
+      "step": 275000
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.34881763220326e-05,
+      "loss": 1.0246,
+      "step": 275010
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.3487028392132855e-05,
+      "loss": 0.9546,
+      "step": 275020
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.34858804622331e-05,
+      "loss": 1.003,
+      "step": 275030
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.3484732532333354e-05,
+      "loss": 0.9916,
+      "step": 275040
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.348358460243361e-05,
+      "loss": 1.0683,
+      "step": 275050
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.3482436672533854e-05,
+      "loss": 0.9516,
+      "step": 275060
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.348128874263411e-05,
+      "loss": 1.0001,
+      "step": 275070
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.348014081273437e-05,
+      "loss": 0.981,
+      "step": 275080
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.347899288283462e-05,
+      "loss": 0.941,
+      "step": 275090
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.347784495293487e-05,
+      "loss": 1.0767,
+      "step": 275100
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.3476697023035124e-05,
+      "loss": 0.9675,
+      "step": 275110
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.3475549093135374e-05,
+      "loss": 0.9139,
+      "step": 275120
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.347440116323563e-05,
+      "loss": 0.9301,
+      "step": 275130
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.347325323333588e-05,
+      "loss": 0.9777,
+      "step": 275140
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.347210530343613e-05,
+      "loss": 1.0373,
+      "step": 275150
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.347095737353639e-05,
+      "loss": 1.0488,
+      "step": 275160
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.346980944363664e-05,
+      "loss": 0.8962,
+      "step": 275170
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.3468661513736886e-05,
+      "loss": 1.06,
+      "step": 275180
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.346751358383714e-05,
+      "loss": 1.0483,
+      "step": 275190
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.34663656539374e-05,
+      "loss": 0.9277,
+      "step": 275200
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.346521772403764e-05,
+      "loss": 0.9889,
+      "step": 275210
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.34640697941379e-05,
+      "loss": 0.9466,
+      "step": 275220
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.3462921864238156e-05,
+      "loss": 0.9042,
+      "step": 275230
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.3461773934338406e-05,
+      "loss": 1.0054,
+      "step": 275240
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.3460626004438656e-05,
+      "loss": 0.9579,
+      "step": 275250
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.345947807453891e-05,
+      "loss": 0.9992,
+      "step": 275260
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.345833014463916e-05,
+      "loss": 1.026,
+      "step": 275270
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.345718221473942e-05,
+      "loss": 1.0507,
+      "step": 275280
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.345603428483967e-05,
+      "loss": 1.1143,
+      "step": 275290
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.345488635493992e-05,
+      "loss": 1.0722,
+      "step": 275300
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.3453738425040175e-05,
+      "loss": 1.08,
+      "step": 275310
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.345259049514043e-05,
+      "loss": 1.0046,
+      "step": 275320
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.3451442565240675e-05,
+      "loss": 1.0225,
+      "step": 275330
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.345029463534093e-05,
+      "loss": 1.0151,
+      "step": 275340
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.344914670544119e-05,
+      "loss": 0.8715,
+      "step": 275350
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.344799877554143e-05,
+      "loss": 0.9822,
+      "step": 275360
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.344685084564169e-05,
+      "loss": 1.0332,
+      "step": 275370
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.3445702915741944e-05,
+      "loss": 0.9288,
+      "step": 275380
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.3444554985842194e-05,
+      "loss": 0.9872,
+      "step": 275390
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.3443407055942444e-05,
+      "loss": 1.0022,
+      "step": 275400
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.34422591260427e-05,
+      "loss": 1.067,
+      "step": 275410
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.344111119614295e-05,
+      "loss": 1.0368,
+      "step": 275420
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.343996326624321e-05,
+      "loss": 1.0541,
+      "step": 275430
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.343881533634346e-05,
+      "loss": 0.9455,
+      "step": 275440
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.343766740644371e-05,
+      "loss": 0.9483,
+      "step": 275450
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.3436519476543964e-05,
+      "loss": 1.04,
+      "step": 275460
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.343537154664422e-05,
+      "loss": 1.1112,
+      "step": 275470
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.343422361674446e-05,
+      "loss": 1.0586,
+      "step": 275480
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.343307568684472e-05,
+      "loss": 1.03,
+      "step": 275490
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.3431927756944977e-05,
+      "loss": 0.9799,
+      "step": 275500
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.343077982704522e-05,
+      "loss": 1.0991,
+      "step": 275510
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.3429631897145476e-05,
+      "loss": 0.9544,
+      "step": 275520
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.342848396724573e-05,
+      "loss": 1.0321,
+      "step": 275530
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.342733603734598e-05,
+      "loss": 0.9871,
+      "step": 275540
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.342618810744623e-05,
+      "loss": 0.9742,
+      "step": 275550
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.342504017754649e-05,
+      "loss": 1.0085,
+      "step": 275560
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.342389224764674e-05,
+      "loss": 1.0383,
+      "step": 275570
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.3422744317746996e-05,
+      "loss": 0.9537,
+      "step": 275580
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.3421596387847246e-05,
+      "loss": 1.0158,
+      "step": 275590
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.3420448457947495e-05,
+      "loss": 1.0655,
+      "step": 275600
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.341930052804775e-05,
+      "loss": 0.9415,
+      "step": 275610
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.341815259814801e-05,
+      "loss": 1.0544,
+      "step": 275620
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.341700466824825e-05,
+      "loss": 1.0816,
+      "step": 275630
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.341585673834851e-05,
+      "loss": 0.9478,
+      "step": 275640
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.3414708808448765e-05,
+      "loss": 1.019,
+      "step": 275650
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.341356087854901e-05,
+      "loss": 1.0169,
+      "step": 275660
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.3412412948649265e-05,
+      "loss": 1.122,
+      "step": 275670
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.341126501874952e-05,
+      "loss": 1.0621,
+      "step": 275680
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.341011708884977e-05,
+      "loss": 1.0022,
+      "step": 275690
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.340896915895002e-05,
+      "loss": 1.0763,
+      "step": 275700
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.340782122905027e-05,
+      "loss": 0.9881,
+      "step": 275710
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.340667329915053e-05,
+      "loss": 0.9365,
+      "step": 275720
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.3405525369250784e-05,
+      "loss": 0.9584,
+      "step": 275730
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.340437743935103e-05,
+      "loss": 0.9314,
+      "step": 275740
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.3403229509451284e-05,
+      "loss": 0.9432,
+      "step": 275750
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.340208157955154e-05,
+      "loss": 1.0344,
+      "step": 275760
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.3400933649651784e-05,
+      "loss": 0.9813,
+      "step": 275770
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.339978571975204e-05,
+      "loss": 1.0452,
+      "step": 275780
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.33986377898523e-05,
+      "loss": 0.9714,
+      "step": 275790
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.339748985995255e-05,
+      "loss": 0.9523,
+      "step": 275800
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.3396341930052797e-05,
+      "loss": 0.9503,
+      "step": 275810
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.339519400015305e-05,
+      "loss": 1.0275,
+      "step": 275820
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.33940460702533e-05,
+      "loss": 0.9208,
+      "step": 275830
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.339289814035356e-05,
+      "loss": 0.9728,
+      "step": 275840
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.339175021045381e-05,
+      "loss": 1.1249,
+      "step": 275850
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.339060228055406e-05,
+      "loss": 0.9918,
+      "step": 275860
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.3389454350654316e-05,
+      "loss": 1.0656,
+      "step": 275870
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.338830642075457e-05,
+      "loss": 1.0139,
+      "step": 275880
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.3387158490854816e-05,
+      "loss": 0.9938,
+      "step": 275890
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.338601056095507e-05,
+      "loss": 1.1026,
+      "step": 275900
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.338486263105533e-05,
+      "loss": 0.9612,
+      "step": 275910
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.338371470115557e-05,
+      "loss": 0.9013,
+      "step": 275920
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.338256677125583e-05,
+      "loss": 1.0175,
+      "step": 275930
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.3381418841356085e-05,
+      "loss": 0.9525,
+      "step": 275940
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.3380270911456335e-05,
+      "loss": 1.0679,
+      "step": 275950
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.3379122981556585e-05,
+      "loss": 0.9797,
+      "step": 275960
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.337797505165684e-05,
+      "loss": 1.0371,
+      "step": 275970
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.337682712175709e-05,
+      "loss": 0.9216,
+      "step": 275980
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.337567919185735e-05,
+      "loss": 1.0494,
+      "step": 275990
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.33745312619576e-05,
+      "loss": 0.985,
+      "step": 276000
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.337338333205785e-05,
+      "loss": 0.8936,
+      "step": 276010
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.3372235402158104e-05,
+      "loss": 1.0147,
+      "step": 276020
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.337108747225836e-05,
+      "loss": 0.8851,
+      "step": 276030
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.3369939542358604e-05,
+      "loss": 1.0579,
+      "step": 276040
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.336879161245886e-05,
+      "loss": 1.002,
+      "step": 276050
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.336764368255912e-05,
+      "loss": 1.0247,
+      "step": 276060
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.336649575265936e-05,
+      "loss": 0.9731,
+      "step": 276070
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.336534782275962e-05,
+      "loss": 0.9236,
+      "step": 276080
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.3364199892859874e-05,
+      "loss": 1.0349,
+      "step": 276090
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.3363051962960124e-05,
+      "loss": 1.0426,
+      "step": 276100
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.3361904033060374e-05,
+      "loss": 0.9697,
+      "step": 276110
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.336075610316063e-05,
+      "loss": 0.9628,
+      "step": 276120
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.335960817326088e-05,
+      "loss": 1.0613,
+      "step": 276130
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.335846024336114e-05,
+      "loss": 1.0122,
+      "step": 276140
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.3357312313461386e-05,
+      "loss": 1.0477,
+      "step": 276150
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.3356164383561636e-05,
+      "loss": 1.1107,
+      "step": 276160
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.335501645366189e-05,
+      "loss": 1.0608,
+      "step": 276170
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.335386852376215e-05,
+      "loss": 1.0072,
+      "step": 276180
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.335272059386239e-05,
+      "loss": 1.04,
+      "step": 276190
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.335157266396265e-05,
+      "loss": 0.942,
+      "step": 276200
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.3350424734062906e-05,
+      "loss": 1.0266,
+      "step": 276210
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.334927680416315e-05,
+      "loss": 1.0141,
+      "step": 276220
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.3348128874263406e-05,
+      "loss": 0.9939,
+      "step": 276230
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.334698094436366e-05,
+      "loss": 0.9572,
+      "step": 276240
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.334583301446391e-05,
+      "loss": 0.9555,
+      "step": 276250
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.334468508456416e-05,
+      "loss": 1.0096,
+      "step": 276260
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.334353715466442e-05,
+      "loss": 1.0155,
+      "step": 276270
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.334238922476467e-05,
+      "loss": 1.038,
+      "step": 276280
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.3341241294864925e-05,
+      "loss": 1.0266,
+      "step": 276290
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.3340093364965175e-05,
+      "loss": 0.9218,
+      "step": 276300
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.3338945435065425e-05,
+      "loss": 1.0979,
+      "step": 276310
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.333779750516568e-05,
+      "loss": 1.0051,
+      "step": 276320
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.333664957526594e-05,
+      "loss": 1.0532,
+      "step": 276330
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.333550164536618e-05,
+      "loss": 0.9675,
+      "step": 276340
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.333435371546644e-05,
+      "loss": 0.9774,
+      "step": 276350
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.3333205785566694e-05,
+      "loss": 1.0602,
+      "step": 276360
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.333205785566694e-05,
+      "loss": 1.022,
+      "step": 276370
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.3330909925767194e-05,
+      "loss": 1.0863,
+      "step": 276380
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.332976199586745e-05,
+      "loss": 0.9316,
+      "step": 276390
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.33286140659677e-05,
+      "loss": 1.1629,
+      "step": 276400
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.332746613606795e-05,
+      "loss": 0.9958,
+      "step": 276410
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.332631820616821e-05,
+      "loss": 0.9982,
+      "step": 276420
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.332517027626846e-05,
+      "loss": 0.973,
+      "step": 276430
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.3324022346368714e-05,
+      "loss": 0.9013,
+      "step": 276440
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.3322874416468963e-05,
+      "loss": 0.9845,
+      "step": 276450
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.332172648656921e-05,
+      "loss": 0.9362,
+      "step": 276460
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.332057855666947e-05,
+      "loss": 1.0706,
+      "step": 276470
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.3319430626769727e-05,
+      "loss": 0.9974,
+      "step": 276480
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.331828269686997e-05,
+      "loss": 1.0452,
+      "step": 276490
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.3317134766970226e-05,
+      "loss": 0.9947,
+      "step": 276500
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.331598683707048e-05,
+      "loss": 1.0396,
+      "step": 276510
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.3314838907170726e-05,
+      "loss": 1.0514,
+      "step": 276520
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.331369097727098e-05,
+      "loss": 1.0176,
+      "step": 276530
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.331254304737124e-05,
+      "loss": 0.8988,
+      "step": 276540
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.331139511747149e-05,
+      "loss": 0.967,
+      "step": 276550
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.331024718757174e-05,
+      "loss": 1.0339,
+      "step": 276560
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.3309099257671996e-05,
+      "loss": 0.9122,
+      "step": 276570
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.3307951327772245e-05,
+      "loss": 0.8946,
+      "step": 276580
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.33068033978725e-05,
+      "loss": 1.0247,
+      "step": 276590
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.330565546797275e-05,
+      "loss": 1.084,
+      "step": 276600
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.3304507538073e-05,
+      "loss": 1.0273,
+      "step": 276610
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.330335960817326e-05,
+      "loss": 1.0269,
+      "step": 276620
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.3302211678273515e-05,
+      "loss": 1.0665,
+      "step": 276630
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.330106374837376e-05,
+      "loss": 1.0556,
+      "step": 276640
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.3299915818474015e-05,
+      "loss": 1.0497,
+      "step": 276650
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.329876788857427e-05,
+      "loss": 1.0257,
+      "step": 276660
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.3297619958674514e-05,
+      "loss": 1.0131,
+      "step": 276670
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.329647202877477e-05,
+      "loss": 0.8655,
+      "step": 276680
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.329532409887503e-05,
+      "loss": 1.0144,
+      "step": 276690
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.329417616897528e-05,
+      "loss": 1.0896,
+      "step": 276700
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.329302823907553e-05,
+      "loss": 1.0513,
+      "step": 276710
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.3291880309175784e-05,
+      "loss": 1.0987,
+      "step": 276720
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.3290732379276034e-05,
+      "loss": 1.0903,
+      "step": 276730
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.328958444937629e-05,
+      "loss": 1.057,
+      "step": 276740
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.328843651947654e-05,
+      "loss": 0.9461,
+      "step": 276750
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.328728858957679e-05,
+      "loss": 0.9196,
+      "step": 276760
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.328614065967705e-05,
+      "loss": 0.9739,
+      "step": 276770
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.3284992729777303e-05,
+      "loss": 1.011,
+      "step": 276780
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.3283844799877547e-05,
+      "loss": 1.0611,
+      "step": 276790
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.32826968699778e-05,
+      "loss": 1.0252,
+      "step": 276800
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.328154894007806e-05,
+      "loss": 1.0421,
+      "step": 276810
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.32804010101783e-05,
+      "loss": 0.9846,
+      "step": 276820
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.327925308027856e-05,
+      "loss": 1.0669,
+      "step": 276830
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.3278105150378816e-05,
+      "loss": 1.0233,
+      "step": 276840
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.3276957220479066e-05,
+      "loss": 1.0051,
+      "step": 276850
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.3275809290579316e-05,
+      "loss": 0.9737,
+      "step": 276860
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.327466136067957e-05,
+      "loss": 1.1335,
+      "step": 276870
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.327351343077982e-05,
+      "loss": 1.0193,
+      "step": 276880
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.327236550088008e-05,
+      "loss": 1.0487,
+      "step": 276890
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.327121757098033e-05,
+      "loss": 1.074,
+      "step": 276900
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.327006964108058e-05,
+      "loss": 1.0871,
+      "step": 276910
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.3268921711180835e-05,
+      "loss": 1.0707,
+      "step": 276920
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.326777378128109e-05,
+      "loss": 0.94,
+      "step": 276930
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.3266625851381335e-05,
+      "loss": 0.9824,
+      "step": 276940
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.326547792148159e-05,
+      "loss": 1.0468,
+      "step": 276950
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.326432999158185e-05,
+      "loss": 1.0976,
+      "step": 276960
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.326318206168209e-05,
+      "loss": 1.1087,
+      "step": 276970
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.326203413178235e-05,
+      "loss": 1.0082,
+      "step": 276980
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.3260886201882605e-05,
+      "loss": 0.975,
+      "step": 276990
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.3259738271982854e-05,
+      "loss": 1.1621,
+      "step": 277000
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.3258590342083104e-05,
+      "loss": 1.1513,
+      "step": 277010
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.325744241218336e-05,
+      "loss": 0.9411,
+      "step": 277020
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.325629448228361e-05,
+      "loss": 1.0232,
+      "step": 277030
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.325514655238387e-05,
+      "loss": 0.934,
+      "step": 277040
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.325399862248412e-05,
+      "loss": 1.0318,
+      "step": 277050
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.325285069258437e-05,
+      "loss": 0.9626,
+      "step": 277060
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.3251702762684624e-05,
+      "loss": 1.0023,
+      "step": 277070
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.325055483278487e-05,
+      "loss": 1.0707,
+      "step": 277080
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.3249406902885123e-05,
+      "loss": 0.9378,
+      "step": 277090
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.324825897298538e-05,
+      "loss": 0.9664,
+      "step": 277100
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.324711104308563e-05,
+      "loss": 1.0157,
+      "step": 277110
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.324596311318588e-05,
+      "loss": 0.9852,
+      "step": 277120
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.3244815183286136e-05,
+      "loss": 1.0305,
+      "step": 277130
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.3243667253386386e-05,
+      "loss": 1.0691,
+      "step": 277140
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.324251932348664e-05,
+      "loss": 1.0064,
+      "step": 277150
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.324137139358689e-05,
+      "loss": 1.0786,
+      "step": 277160
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.324022346368714e-05,
+      "loss": 1.0428,
+      "step": 277170
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.32390755337874e-05,
+      "loss": 1.1192,
+      "step": 277180
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.3237927603887656e-05,
+      "loss": 1.007,
+      "step": 277190
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.32367796739879e-05,
+      "loss": 0.9767,
+      "step": 277200
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.3235631744088156e-05,
+      "loss": 1.0332,
+      "step": 277210
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.323448381418841e-05,
+      "loss": 0.9844,
+      "step": 277220
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.3233335884288655e-05,
+      "loss": 1.0247,
+      "step": 277230
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.323218795438891e-05,
+      "loss": 0.9664,
+      "step": 277240
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.323104002448917e-05,
+      "loss": 1.1591,
+      "step": 277250
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.322989209458942e-05,
+      "loss": 0.9946,
+      "step": 277260
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.322874416468967e-05,
+      "loss": 1.0359,
+      "step": 277270
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.3227596234789925e-05,
+      "loss": 1.1448,
+      "step": 277280
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.3226448304890175e-05,
+      "loss": 1.0531,
+      "step": 277290
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.322530037499043e-05,
+      "loss": 0.9553,
+      "step": 277300
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.322415244509068e-05,
+      "loss": 0.9641,
+      "step": 277310
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.322300451519093e-05,
+      "loss": 1.14,
+      "step": 277320
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.322185658529119e-05,
+      "loss": 1.0238,
+      "step": 277330
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.3220708655391444e-05,
+      "loss": 1.0201,
+      "step": 277340
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.321956072549169e-05,
+      "loss": 1.0198,
+      "step": 277350
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.3218412795591944e-05,
+      "loss": 1.0633,
+      "step": 277360
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.32172648656922e-05,
+      "loss": 0.9769,
+      "step": 277370
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.3216116935792444e-05,
+      "loss": 1.0246,
+      "step": 277380
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.32149690058927e-05,
+      "loss": 0.9097,
+      "step": 277390
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.321382107599296e-05,
+      "loss": 0.9013,
+      "step": 277400
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.321267314609321e-05,
+      "loss": 1.053,
+      "step": 277410
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.321152521619346e-05,
+      "loss": 1.0296,
+      "step": 277420
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.3210377286293713e-05,
+      "loss": 0.9272,
+      "step": 277430
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.320922935639396e-05,
+      "loss": 1.0516,
+      "step": 277440
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.320808142649422e-05,
+      "loss": 0.9496,
+      "step": 277450
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.320693349659447e-05,
+      "loss": 0.9592,
+      "step": 277460
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.320578556669472e-05,
+      "loss": 0.9939,
+      "step": 277470
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.3204637636794976e-05,
+      "loss": 1.0528,
+      "step": 277480
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.320348970689523e-05,
+      "loss": 0.9374,
+      "step": 277490
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.3202341776995476e-05,
+      "loss": 1.078,
+      "step": 277500
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.320119384709573e-05,
+      "loss": 1.0548,
+      "step": 277510
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.320004591719599e-05,
+      "loss": 0.9939,
+      "step": 277520
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.319889798729623e-05,
+      "loss": 0.9484,
+      "step": 277530
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.319775005739649e-05,
+      "loss": 1.0447,
+      "step": 277540
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.3196602127496746e-05,
+      "loss": 0.9788,
+      "step": 277550
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.3195454197596995e-05,
+      "loss": 0.99,
+      "step": 277560
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.3194306267697245e-05,
+      "loss": 1.0069,
+      "step": 277570
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.31931583377975e-05,
+      "loss": 1.0667,
+      "step": 277580
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.319201040789775e-05,
+      "loss": 0.9905,
+      "step": 277590
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.319086247799801e-05,
+      "loss": 1.0786,
+      "step": 277600
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.318971454809826e-05,
+      "loss": 1.1561,
+      "step": 277610
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.318856661819851e-05,
+      "loss": 0.9981,
+      "step": 277620
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.3187418688298765e-05,
+      "loss": 0.974,
+      "step": 277630
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.318627075839902e-05,
+      "loss": 0.9882,
+      "step": 277640
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.3185122828499264e-05,
+      "loss": 1.0359,
+      "step": 277650
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.318397489859952e-05,
+      "loss": 1.0053,
+      "step": 277660
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.318282696869978e-05,
+      "loss": 0.9263,
+      "step": 277670
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.318167903880002e-05,
+      "loss": 0.9653,
+      "step": 277680
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.318053110890028e-05,
+      "loss": 1.0044,
+      "step": 277690
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.3179383179000534e-05,
+      "loss": 1.0793,
+      "step": 277700
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.3178235249100784e-05,
+      "loss": 0.9433,
+      "step": 277710
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.3177087319201034e-05,
+      "loss": 1.0502,
+      "step": 277720
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.317593938930129e-05,
+      "loss": 1.0585,
+      "step": 277730
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.317479145940154e-05,
+      "loss": 0.9362,
+      "step": 277740
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.31736435295018e-05,
+      "loss": 1.0199,
+      "step": 277750
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.317249559960205e-05,
+      "loss": 0.9356,
+      "step": 277760
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.3171347669702297e-05,
+      "loss": 0.9924,
+      "step": 277770
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.317019973980255e-05,
+      "loss": 0.9038,
+      "step": 277780
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.316905180990281e-05,
+      "loss": 1.0989,
+      "step": 277790
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.316790388000305e-05,
+      "loss": 0.9879,
+      "step": 277800
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.316675595010331e-05,
+      "loss": 1.0863,
+      "step": 277810
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.3165608020203566e-05,
+      "loss": 0.9844,
+      "step": 277820
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.316446009030381e-05,
+      "loss": 0.9865,
+      "step": 277830
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.3163312160404066e-05,
+      "loss": 1.0615,
+      "step": 277840
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.316216423050432e-05,
+      "loss": 1.0859,
+      "step": 277850
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.316101630060457e-05,
+      "loss": 1.0929,
+      "step": 277860
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.315986837070482e-05,
+      "loss": 1.1166,
+      "step": 277870
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.315872044080508e-05,
+      "loss": 0.9751,
+      "step": 277880
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.315757251090533e-05,
+      "loss": 0.9704,
+      "step": 277890
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.3156424581005585e-05,
+      "loss": 1.0596,
+      "step": 277900
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.3155276651105835e-05,
+      "loss": 0.9532,
+      "step": 277910
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.3154128721206085e-05,
+      "loss": 1.0832,
+      "step": 277920
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.315298079130634e-05,
+      "loss": 1.0139,
+      "step": 277930
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.31518328614066e-05,
+      "loss": 0.9632,
+      "step": 277940
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.315068493150684e-05,
+      "loss": 1.0396,
+      "step": 277950
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.31495370016071e-05,
+      "loss": 1.1358,
+      "step": 277960
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.3148389071707355e-05,
+      "loss": 1.0021,
+      "step": 277970
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.31472411418076e-05,
+      "loss": 0.998,
+      "step": 277980
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.3146093211907854e-05,
+      "loss": 1.1298,
+      "step": 277990
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.314494528200811e-05,
+      "loss": 0.9407,
+      "step": 278000
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.314379735210836e-05,
+      "loss": 0.9912,
+      "step": 278010
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.314264942220861e-05,
+      "loss": 1.032,
+      "step": 278020
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.314150149230887e-05,
+      "loss": 0.9985,
+      "step": 278030
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.314035356240912e-05,
+      "loss": 1.0814,
+      "step": 278040
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.3139205632509374e-05,
+      "loss": 1.0103,
+      "step": 278050
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.3138057702609624e-05,
+      "loss": 1.131,
+      "step": 278060
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.3136909772709873e-05,
+      "loss": 1.024,
+      "step": 278070
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.313576184281013e-05,
+      "loss": 0.9637,
+      "step": 278080
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.313461391291039e-05,
+      "loss": 1.0693,
+      "step": 278090
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.313346598301063e-05,
+      "loss": 0.9126,
+      "step": 278100
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.3132318053110886e-05,
+      "loss": 1.0425,
+      "step": 278110
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.313117012321114e-05,
+      "loss": 0.926,
+      "step": 278120
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.3130022193311386e-05,
+      "loss": 0.9956,
+      "step": 278130
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.312887426341164e-05,
+      "loss": 1.0621,
+      "step": 278140
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.31277263335119e-05,
+      "loss": 0.9999,
+      "step": 278150
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.312657840361215e-05,
+      "loss": 1.0116,
+      "step": 278160
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.31254304737124e-05,
+      "loss": 1.0358,
+      "step": 278170
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.3124282543812656e-05,
+      "loss": 1.0956,
+      "step": 278180
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.3123134613912906e-05,
+      "loss": 1.0066,
+      "step": 278190
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.312198668401316e-05,
+      "loss": 0.9958,
+      "step": 278200
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.312083875411341e-05,
+      "loss": 0.9513,
+      "step": 278210
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.311969082421366e-05,
+      "loss": 1.0792,
+      "step": 278220
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.311854289431392e-05,
+      "loss": 1.0929,
+      "step": 278230
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.3117394964414175e-05,
+      "loss": 0.9498,
+      "step": 278240
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.311624703451442e-05,
+      "loss": 0.9668,
+      "step": 278250
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.3115099104614675e-05,
+      "loss": 1.0044,
+      "step": 278260
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.311395117471493e-05,
+      "loss": 1.028,
+      "step": 278270
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.3112803244815175e-05,
+      "loss": 1.0144,
+      "step": 278280
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.311165531491543e-05,
+      "loss": 0.9985,
+      "step": 278290
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.311050738501569e-05,
+      "loss": 1.0055,
+      "step": 278300
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.310935945511594e-05,
+      "loss": 1.0151,
+      "step": 278310
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.310821152521619e-05,
+      "loss": 1.0331,
+      "step": 278320
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.3107063595316444e-05,
+      "loss": 0.957,
+      "step": 278330
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.3105915665416694e-05,
+      "loss": 1.1133,
+      "step": 278340
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.310476773551695e-05,
+      "loss": 1.0087,
+      "step": 278350
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.31036198056172e-05,
+      "loss": 0.963,
+      "step": 278360
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.310247187571745e-05,
+      "loss": 0.9745,
+      "step": 278370
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.310132394581771e-05,
+      "loss": 1.1312,
+      "step": 278380
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.3100176015917964e-05,
+      "loss": 1.0074,
+      "step": 278390
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.309902808601821e-05,
+      "loss": 1.1623,
+      "step": 278400
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.3097880156118463e-05,
+      "loss": 0.918,
+      "step": 278410
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.309673222621872e-05,
+      "loss": 0.9723,
+      "step": 278420
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.309558429631896e-05,
+      "loss": 0.9318,
+      "step": 278430
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.309443636641922e-05,
+      "loss": 0.9468,
+      "step": 278440
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.309328843651947e-05,
+      "loss": 1.1046,
+      "step": 278450
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.3092140506619726e-05,
+      "loss": 1.0807,
+      "step": 278460
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.3090992576719976e-05,
+      "loss": 0.946,
+      "step": 278470
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.3089844646820226e-05,
+      "loss": 1.0147,
+      "step": 278480
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.308869671692048e-05,
+      "loss": 0.9362,
+      "step": 278490
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.308754878702074e-05,
+      "loss": 0.9803,
+      "step": 278500
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.308640085712098e-05,
+      "loss": 0.9625,
+      "step": 278510
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.308525292722124e-05,
+      "loss": 0.9919,
+      "step": 278520
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.3084104997321496e-05,
+      "loss": 0.9811,
+      "step": 278530
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.308295706742174e-05,
+      "loss": 1.126,
+      "step": 278540
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.3081809137521995e-05,
+      "loss": 0.9628,
+      "step": 278550
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.308066120762225e-05,
+      "loss": 1.034,
+      "step": 278560
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.30795132777225e-05,
+      "loss": 0.9819,
+      "step": 278570
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.307836534782275e-05,
+      "loss": 0.9798,
+      "step": 278580
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.307721741792301e-05,
+      "loss": 0.9226,
+      "step": 278590
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.307606948802326e-05,
+      "loss": 0.9499,
+      "step": 278600
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.3074921558123515e-05,
+      "loss": 0.9285,
+      "step": 278610
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.3073773628223765e-05,
+      "loss": 1.1506,
+      "step": 278620
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.3072625698324014e-05,
+      "loss": 1.0184,
+      "step": 278630
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.307147776842427e-05,
+      "loss": 1.0469,
+      "step": 278640
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.307032983852453e-05,
+      "loss": 1.0338,
+      "step": 278650
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.306918190862477e-05,
+      "loss": 0.962,
+      "step": 278660
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.306803397872503e-05,
+      "loss": 1.0417,
+      "step": 278670
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.3066886048825284e-05,
+      "loss": 1.0489,
+      "step": 278680
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.306573811892553e-05,
+      "loss": 1.0734,
+      "step": 278690
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.3064590189025784e-05,
+      "loss": 1.0676,
+      "step": 278700
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.306344225912604e-05,
+      "loss": 0.9344,
+      "step": 278710
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.306229432922629e-05,
+      "loss": 0.9433,
+      "step": 278720
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.306114639932654e-05,
+      "loss": 1.0663,
+      "step": 278730
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.30599984694268e-05,
+      "loss": 0.9876,
+      "step": 278740
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.3058850539527047e-05,
+      "loss": 1.0138,
+      "step": 278750
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.30577026096273e-05,
+      "loss": 1.0155,
+      "step": 278760
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.305655467972755e-05,
+      "loss": 1.141,
+      "step": 278770
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.30554067498278e-05,
+      "loss": 1.1578,
+      "step": 278780
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.305425881992806e-05,
+      "loss": 1.0622,
+      "step": 278790
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.3053110890028316e-05,
+      "loss": 1.1124,
+      "step": 278800
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.305196296012856e-05,
+      "loss": 1.1348,
+      "step": 278810
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.3050815030228816e-05,
+      "loss": 0.9823,
+      "step": 278820
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.304966710032907e-05,
+      "loss": 0.9873,
+      "step": 278830
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.3048519170429316e-05,
+      "loss": 1.0515,
+      "step": 278840
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.304737124052957e-05,
+      "loss": 1.0648,
+      "step": 278850
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.304622331062983e-05,
+      "loss": 1.0704,
+      "step": 278860
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.304507538073008e-05,
+      "loss": 1.0522,
+      "step": 278870
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.304392745083033e-05,
+      "loss": 1.0375,
+      "step": 278880
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.3042779520930585e-05,
+      "loss": 0.9827,
+      "step": 278890
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.3041631591030835e-05,
+      "loss": 1.0684,
+      "step": 278900
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.304048366113109e-05,
+      "loss": 0.8899,
+      "step": 278910
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.303933573123134e-05,
+      "loss": 1.1301,
+      "step": 278920
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.303818780133159e-05,
+      "loss": 0.9595,
+      "step": 278930
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.303703987143185e-05,
+      "loss": 0.8915,
+      "step": 278940
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.3035891941532105e-05,
+      "loss": 1.0044,
+      "step": 278950
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.303474401163235e-05,
+      "loss": 1.0445,
+      "step": 278960
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.3033596081732604e-05,
+      "loss": 1.0432,
+      "step": 278970
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.303244815183286e-05,
+      "loss": 0.9716,
+      "step": 278980
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.3031300221933104e-05,
+      "loss": 1.0751,
+      "step": 278990
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.303015229203336e-05,
+      "loss": 0.9855,
+      "step": 279000
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.302900436213362e-05,
+      "loss": 0.9037,
+      "step": 279010
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.302785643223387e-05,
+      "loss": 0.9659,
+      "step": 279020
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.302670850233412e-05,
+      "loss": 1.0539,
+      "step": 279030
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.3025560572434374e-05,
+      "loss": 0.9531,
+      "step": 279040
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.3024412642534623e-05,
+      "loss": 0.9707,
+      "step": 279050
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.302326471263488e-05,
+      "loss": 1.0831,
+      "step": 279060
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.302211678273513e-05,
+      "loss": 0.9106,
+      "step": 279070
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.302096885283538e-05,
+      "loss": 0.9805,
+      "step": 279080
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.3019820922935636e-05,
+      "loss": 1.0426,
+      "step": 279090
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.301867299303589e-05,
+      "loss": 0.9654,
+      "step": 279100
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.3017525063136136e-05,
+      "loss": 1.0196,
+      "step": 279110
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.301637713323639e-05,
+      "loss": 0.9747,
+      "step": 279120
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.301522920333665e-05,
+      "loss": 1.0666,
+      "step": 279130
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.301408127343689e-05,
+      "loss": 1.0625,
+      "step": 279140
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.301293334353715e-05,
+      "loss": 0.8516,
+      "step": 279150
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.3011785413637406e-05,
+      "loss": 0.9162,
+      "step": 279160
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.3010637483737656e-05,
+      "loss": 1.0469,
+      "step": 279170
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.3009489553837905e-05,
+      "loss": 1.0179,
+      "step": 279180
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.300834162393816e-05,
+      "loss": 1.0181,
+      "step": 279190
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.300719369403841e-05,
+      "loss": 0.9682,
+      "step": 279200
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.300604576413867e-05,
+      "loss": 1.022,
+      "step": 279210
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.300489783423892e-05,
+      "loss": 0.9607,
+      "step": 279220
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.300374990433917e-05,
+      "loss": 1.0026,
+      "step": 279230
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.3002601974439425e-05,
+      "loss": 0.8946,
+      "step": 279240
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.300145404453968e-05,
+      "loss": 0.9451,
+      "step": 279250
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.3000306114639925e-05,
+      "loss": 0.9885,
+      "step": 279260
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.299915818474018e-05,
+      "loss": 1.0508,
+      "step": 279270
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.299801025484044e-05,
+      "loss": 0.9439,
+      "step": 279280
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.299686232494068e-05,
+      "loss": 0.9186,
+      "step": 279290
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.299571439504094e-05,
+      "loss": 1.0033,
+      "step": 279300
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.2994566465141194e-05,
+      "loss": 0.9856,
+      "step": 279310
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.2993418535241444e-05,
+      "loss": 1.1056,
+      "step": 279320
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.2992270605341694e-05,
+      "loss": 1.0566,
+      "step": 279330
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.299112267544195e-05,
+      "loss": 0.9324,
+      "step": 279340
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.29899747455422e-05,
+      "loss": 0.9256,
+      "step": 279350
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.298882681564246e-05,
+      "loss": 0.9949,
+      "step": 279360
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.298767888574271e-05,
+      "loss": 0.9744,
+      "step": 279370
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.298653095584296e-05,
+      "loss": 0.9689,
+      "step": 279380
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.2985383025943213e-05,
+      "loss": 0.9637,
+      "step": 279390
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.298423509604347e-05,
+      "loss": 0.9758,
+      "step": 279400
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.298308716614371e-05,
+      "loss": 1.003,
+      "step": 279410
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.298193923624397e-05,
+      "loss": 1.0676,
+      "step": 279420
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.2980791306344226e-05,
+      "loss": 1.0139,
+      "step": 279430
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.297964337644447e-05,
+      "loss": 0.9695,
+      "step": 279440
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.2978495446544726e-05,
+      "loss": 0.9862,
+      "step": 279450
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.297734751664498e-05,
+      "loss": 0.9916,
+      "step": 279460
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.297619958674523e-05,
+      "loss": 0.9656,
+      "step": 279470
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.297505165684548e-05,
+      "loss": 0.9499,
+      "step": 279480
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.297390372694574e-05,
+      "loss": 1.0736,
+      "step": 279490
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.297275579704599e-05,
+      "loss": 0.9608,
+      "step": 279500
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.2971607867146246e-05,
+      "loss": 0.9073,
+      "step": 279510
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.2970459937246495e-05,
+      "loss": 1.1086,
+      "step": 279520
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.2969312007346745e-05,
+      "loss": 0.8959,
+      "step": 279530
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.2968164077447e-05,
+      "loss": 1.0293,
+      "step": 279540
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.296701614754726e-05,
+      "loss": 0.9457,
+      "step": 279550
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.29658682176475e-05,
+      "loss": 1.0118,
+      "step": 279560
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.296472028774776e-05,
+      "loss": 1.1093,
+      "step": 279570
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.2963572357848015e-05,
+      "loss": 1.0151,
+      "step": 279580
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.296242442794826e-05,
+      "loss": 1.02,
+      "step": 279590
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.2961276498048515e-05,
+      "loss": 1.0106,
+      "step": 279600
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.296012856814877e-05,
+      "loss": 1.0065,
+      "step": 279610
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.295898063824902e-05,
+      "loss": 1.068,
+      "step": 279620
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.295783270834927e-05,
+      "loss": 1.1354,
+      "step": 279630
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.295668477844953e-05,
+      "loss": 0.9818,
+      "step": 279640
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.295553684854978e-05,
+      "loss": 0.9443,
+      "step": 279650
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.2954388918650034e-05,
+      "loss": 1.1095,
+      "step": 279660
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.2953240988750284e-05,
+      "loss": 0.944,
+      "step": 279670
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.2952093058850534e-05,
+      "loss": 1.004,
+      "step": 279680
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.295094512895079e-05,
+      "loss": 1.1298,
+      "step": 279690
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.294979719905105e-05,
+      "loss": 1.0188,
+      "step": 279700
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.294864926915129e-05,
+      "loss": 1.0123,
+      "step": 279710
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.294750133925155e-05,
+      "loss": 1.0948,
+      "step": 279720
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.29463534093518e-05,
+      "loss": 1.2051,
+      "step": 279730
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.2945205479452046e-05,
+      "loss": 0.996,
+      "step": 279740
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.29440575495523e-05,
+      "loss": 1.0555,
+      "step": 279750
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.294290961965256e-05,
+      "loss": 0.9737,
+      "step": 279760
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.294176168975281e-05,
+      "loss": 0.9428,
+      "step": 279770
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.294061375985306e-05,
+      "loss": 1.0987,
+      "step": 279780
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.2939465829953316e-05,
+      "loss": 0.9287,
+      "step": 279790
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.2938317900053566e-05,
+      "loss": 1.035,
+      "step": 279800
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.293716997015382e-05,
+      "loss": 1.0104,
+      "step": 279810
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.293602204025407e-05,
+      "loss": 1.0444,
+      "step": 279820
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.293487411035432e-05,
+      "loss": 0.9716,
+      "step": 279830
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.293372618045458e-05,
+      "loss": 0.914,
+      "step": 279840
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.293257825055482e-05,
+      "loss": 0.9991,
+      "step": 279850
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.293143032065508e-05,
+      "loss": 0.928,
+      "step": 279860
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.2930282390755335e-05,
+      "loss": 1.0189,
+      "step": 279870
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.2929134460855585e-05,
+      "loss": 0.9775,
+      "step": 279880
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.2927986530955835e-05,
+      "loss": 0.9529,
+      "step": 279890
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.292683860105609e-05,
+      "loss": 1.1781,
+      "step": 279900
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.292569067115634e-05,
+      "loss": 1.0485,
+      "step": 279910
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.29245427412566e-05,
+      "loss": 1.0356,
+      "step": 279920
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.292339481135685e-05,
+      "loss": 0.9539,
+      "step": 279930
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.29222468814571e-05,
+      "loss": 1.0339,
+      "step": 279940
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.2921098951557354e-05,
+      "loss": 0.9697,
+      "step": 279950
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.291995102165761e-05,
+      "loss": 0.9129,
+      "step": 279960
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.2918803091757854e-05,
+      "loss": 0.9524,
+      "step": 279970
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.291765516185811e-05,
+      "loss": 1.1042,
+      "step": 279980
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.291650723195837e-05,
+      "loss": 1.0888,
+      "step": 279990
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.291535930205861e-05,
+      "loss": 1.1528,
+      "step": 280000
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.291421137215887e-05,
+      "loss": 0.9595,
+      "step": 280010
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.2913063442259124e-05,
+      "loss": 0.9706,
+      "step": 280020
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.2911915512359373e-05,
+      "loss": 1.0174,
+      "step": 280030
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.291076758245962e-05,
+      "loss": 0.9538,
+      "step": 280040
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.290961965255988e-05,
+      "loss": 1.0004,
+      "step": 280050
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.290847172266013e-05,
+      "loss": 0.9427,
+      "step": 280060
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.2907323792760386e-05,
+      "loss": 1.0134,
+      "step": 280070
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.2906175862860636e-05,
+      "loss": 0.9931,
+      "step": 280080
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.2905027932960886e-05,
+      "loss": 1.0179,
+      "step": 280090
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.290388000306114e-05,
+      "loss": 1.0287,
+      "step": 280100
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.29027320731614e-05,
+      "loss": 0.9846,
+      "step": 280110
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.290158414326164e-05,
+      "loss": 0.9826,
+      "step": 280120
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.29004362133619e-05,
+      "loss": 0.958,
+      "step": 280130
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.2899288283462156e-05,
+      "loss": 0.9744,
+      "step": 280140
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.28981403535624e-05,
+      "loss": 0.9618,
+      "step": 280150
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.2896992423662655e-05,
+      "loss": 1.1116,
+      "step": 280160
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.289584449376291e-05,
+      "loss": 0.9788,
+      "step": 280170
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.289469656386316e-05,
+      "loss": 0.9489,
+      "step": 280180
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.289354863396341e-05,
+      "loss": 1.0739,
+      "step": 280190
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.289240070406367e-05,
+      "loss": 0.9664,
+      "step": 280200
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.289125277416392e-05,
+      "loss": 1.112,
+      "step": 280210
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.2890104844264175e-05,
+      "loss": 0.949,
+      "step": 280220
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.2888956914364425e-05,
+      "loss": 1.0653,
+      "step": 280230
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.2887808984464675e-05,
+      "loss": 0.9724,
+      "step": 280240
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.288666105456493e-05,
+      "loss": 0.9987,
+      "step": 280250
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.288551312466519e-05,
+      "loss": 1.055,
+      "step": 280260
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.288436519476543e-05,
+      "loss": 1.0952,
+      "step": 280270
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.288321726486569e-05,
+      "loss": 1.0454,
+      "step": 280280
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.2882069334965944e-05,
+      "loss": 0.98,
+      "step": 280290
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.288092140506619e-05,
+      "loss": 1.0639,
+      "step": 280300
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.2879773475166444e-05,
+      "loss": 1.0097,
+      "step": 280310
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.28786255452667e-05,
+      "loss": 1.0701,
+      "step": 280320
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.287747761536695e-05,
+      "loss": 1.0132,
+      "step": 280330
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.28763296854672e-05,
+      "loss": 1.0285,
+      "step": 280340
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.287518175556746e-05,
+      "loss": 1.0237,
+      "step": 280350
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.287403382566771e-05,
+      "loss": 1.0305,
+      "step": 280360
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.2872885895767963e-05,
+      "loss": 1.0863,
+      "step": 280370
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.287173796586821e-05,
+      "loss": 1.0046,
+      "step": 280380
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.287059003596846e-05,
+      "loss": 0.9195,
+      "step": 280390
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.286944210606872e-05,
+      "loss": 0.9091,
+      "step": 280400
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.2868294176168976e-05,
+      "loss": 0.9695,
+      "step": 280410
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.286714624626922e-05,
+      "loss": 1.0844,
+      "step": 280420
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.2865998316369476e-05,
+      "loss": 1.0009,
+      "step": 280430
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.286485038646973e-05,
+      "loss": 1.0688,
+      "step": 280440
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.2863702456569976e-05,
+      "loss": 1.0073,
+      "step": 280450
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.286255452667023e-05,
+      "loss": 0.9487,
+      "step": 280460
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.286140659677049e-05,
+      "loss": 0.9611,
+      "step": 280470
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.286025866687074e-05,
+      "loss": 0.9831,
+      "step": 280480
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.285911073697099e-05,
+      "loss": 1.1057,
+      "step": 280490
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.2857962807071245e-05,
+      "loss": 1.1084,
+      "step": 280500
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.2856814877171495e-05,
+      "loss": 1.1127,
+      "step": 280510
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.285566694727175e-05,
+      "loss": 1.0631,
+      "step": 280520
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.2854519017372e-05,
+      "loss": 0.9289,
+      "step": 280530
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.285337108747225e-05,
+      "loss": 0.962,
+      "step": 280540
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.285222315757251e-05,
+      "loss": 0.9754,
+      "step": 280550
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.2851075227672765e-05,
+      "loss": 0.9751,
+      "step": 280560
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.284992729777301e-05,
+      "loss": 0.9509,
+      "step": 280570
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.2848779367873265e-05,
+      "loss": 0.9663,
+      "step": 280580
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.284763143797352e-05,
+      "loss": 1.05,
+      "step": 280590
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.2846483508073764e-05,
+      "loss": 0.9755,
+      "step": 280600
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.284533557817402e-05,
+      "loss": 1.0064,
+      "step": 280610
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.284418764827428e-05,
+      "loss": 1.0912,
+      "step": 280620
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.284303971837453e-05,
+      "loss": 0.9987,
+      "step": 280630
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.284189178847478e-05,
+      "loss": 1.0871,
+      "step": 280640
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.2840743858575034e-05,
+      "loss": 0.9654,
+      "step": 280650
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.2839595928675284e-05,
+      "loss": 0.962,
+      "step": 280660
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.283844799877554e-05,
+      "loss": 1.0377,
+      "step": 280670
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.283730006887579e-05,
+      "loss": 0.9747,
+      "step": 280680
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.283615213897604e-05,
+      "loss": 1.0293,
+      "step": 280690
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.28350042090763e-05,
+      "loss": 1.088,
+      "step": 280700
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.283385627917655e-05,
+      "loss": 0.9746,
+      "step": 280710
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.2832708349276796e-05,
+      "loss": 0.9967,
+      "step": 280720
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.283156041937705e-05,
+      "loss": 1.1178,
+      "step": 280730
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.283041248947731e-05,
+      "loss": 0.9913,
+      "step": 280740
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.282926455957755e-05,
+      "loss": 0.9524,
+      "step": 280750
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.282811662967781e-05,
+      "loss": 0.9556,
+      "step": 280760
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.2826968699778066e-05,
+      "loss": 1.0328,
+      "step": 280770
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.2825820769878316e-05,
+      "loss": 1.006,
+      "step": 280780
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.2824672839978566e-05,
+      "loss": 1.0498,
+      "step": 280790
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.282352491007882e-05,
+      "loss": 0.9634,
+      "step": 280800
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.282237698017907e-05,
+      "loss": 0.9893,
+      "step": 280810
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.282122905027933e-05,
+      "loss": 1.0669,
+      "step": 280820
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.282008112037958e-05,
+      "loss": 0.9475,
+      "step": 280830
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.281893319047983e-05,
+      "loss": 1.0612,
+      "step": 280840
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.2817785260580085e-05,
+      "loss": 1.0211,
+      "step": 280850
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.281663733068034e-05,
+      "loss": 1.1143,
+      "step": 280860
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.2815489400780585e-05,
+      "loss": 0.9374,
+      "step": 280870
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.281434147088084e-05,
+      "loss": 1.0668,
+      "step": 280880
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.28131935409811e-05,
+      "loss": 1.0294,
+      "step": 280890
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.281204561108134e-05,
+      "loss": 1.019,
+      "step": 280900
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.28108976811816e-05,
+      "loss": 1.0673,
+      "step": 280910
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.2809749751281854e-05,
+      "loss": 1.089,
+      "step": 280920
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.2808601821382104e-05,
+      "loss": 1.0249,
+      "step": 280930
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.2807453891482354e-05,
+      "loss": 1.0842,
+      "step": 280940
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.280630596158261e-05,
+      "loss": 1.1403,
+      "step": 280950
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.280515803168286e-05,
+      "loss": 1.0212,
+      "step": 280960
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.280401010178312e-05,
+      "loss": 0.9572,
+      "step": 280970
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.280286217188337e-05,
+      "loss": 0.9562,
+      "step": 280980
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.280171424198362e-05,
+      "loss": 1.1248,
+      "step": 280990
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.2800566312083874e-05,
+      "loss": 1.1296,
+      "step": 281000
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.279941838218413e-05,
+      "loss": 1.065,
+      "step": 281010
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.279827045228437e-05,
+      "loss": 1.1304,
+      "step": 281020
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.279712252238463e-05,
+      "loss": 1.0507,
+      "step": 281030
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.2795974592484887e-05,
+      "loss": 1.1223,
+      "step": 281040
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.279482666258513e-05,
+      "loss": 1.1373,
+      "step": 281050
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.2793678732685386e-05,
+      "loss": 1.0651,
+      "step": 281060
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.279253080278564e-05,
+      "loss": 0.9829,
+      "step": 281070
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.279138287288589e-05,
+      "loss": 1.0923,
+      "step": 281080
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.279023494298614e-05,
+      "loss": 0.9863,
+      "step": 281090
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.27890870130864e-05,
+      "loss": 1.019,
+      "step": 281100
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.278793908318665e-05,
+      "loss": 1.0697,
+      "step": 281110
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.2786791153286906e-05,
+      "loss": 0.9858,
+      "step": 281120
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.2785643223387156e-05,
+      "loss": 1.0333,
+      "step": 281130
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.2784495293487405e-05,
+      "loss": 1.0337,
+      "step": 281140
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.278334736358766e-05,
+      "loss": 1.0236,
+      "step": 281150
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.278219943368792e-05,
+      "loss": 0.9994,
+      "step": 281160
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.278105150378816e-05,
+      "loss": 1.1395,
+      "step": 281170
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.277990357388842e-05,
+      "loss": 0.979,
+      "step": 281180
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.2778755643988675e-05,
+      "loss": 1.0475,
+      "step": 281190
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.277760771408892e-05,
+      "loss": 1.0413,
+      "step": 281200
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.2776459784189175e-05,
+      "loss": 0.9913,
+      "step": 281210
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.2775311854289425e-05,
+      "loss": 1.0231,
+      "step": 281220
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.277416392438968e-05,
+      "loss": 0.9825,
+      "step": 281230
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.277301599448993e-05,
+      "loss": 1.1039,
+      "step": 281240
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.277186806459018e-05,
+      "loss": 0.9103,
+      "step": 281250
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.277072013469044e-05,
+      "loss": 0.951,
+      "step": 281260
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.2769572204790694e-05,
+      "loss": 1.1232,
+      "step": 281270
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.276842427489094e-05,
+      "loss": 0.9462,
+      "step": 281280
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.2767276344991194e-05,
+      "loss": 1.0692,
+      "step": 281290
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.276612841509145e-05,
+      "loss": 1.0152,
+      "step": 281300
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.2764980485191694e-05,
+      "loss": 1.0419,
+      "step": 281310
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.276383255529195e-05,
+      "loss": 1.0463,
+      "step": 281320
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.276268462539221e-05,
+      "loss": 1.0552,
+      "step": 281330
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.276153669549246e-05,
+      "loss": 0.9968,
+      "step": 281340
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.276038876559271e-05,
+      "loss": 0.9645,
+      "step": 281350
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.275924083569296e-05,
+      "loss": 1.1069,
+      "step": 281360
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.275809290579321e-05,
+      "loss": 1.1869,
+      "step": 281370
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.275694497589347e-05,
+      "loss": 0.9785,
+      "step": 281380
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.275579704599372e-05,
+      "loss": 1.0965,
+      "step": 281390
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.275464911609397e-05,
+      "loss": 0.9599,
+      "step": 281400
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.2753501186194226e-05,
+      "loss": 0.9434,
+      "step": 281410
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.275235325629448e-05,
+      "loss": 0.9786,
+      "step": 281420
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.2751205326394726e-05,
+      "loss": 0.9844,
+      "step": 281430
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.275005739649498e-05,
+      "loss": 1.1006,
+      "step": 281440
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.274890946659524e-05,
+      "loss": 1.0623,
+      "step": 281450
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.274776153669548e-05,
+      "loss": 0.9915,
+      "step": 281460
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.274661360679574e-05,
+      "loss": 1.0056,
+      "step": 281470
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.2745465676895995e-05,
+      "loss": 1.0319,
+      "step": 281480
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.2744317746996245e-05,
+      "loss": 1.1544,
+      "step": 281490
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.2743169817096495e-05,
+      "loss": 1.0101,
+      "step": 281500
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.274202188719675e-05,
+      "loss": 1.0604,
+      "step": 281510
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.2740873957297e-05,
+      "loss": 0.9533,
+      "step": 281520
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.273972602739726e-05,
+      "loss": 0.9808,
+      "step": 281530
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.273857809749751e-05,
+      "loss": 0.996,
+      "step": 281540
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.273743016759776e-05,
+      "loss": 1.0495,
+      "step": 281550
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.2736282237698015e-05,
+      "loss": 0.9649,
+      "step": 281560
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.273513430779827e-05,
+      "loss": 1.135,
+      "step": 281570
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.2733986377898514e-05,
+      "loss": 0.9641,
+      "step": 281580
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.273283844799877e-05,
+      "loss": 0.9321,
+      "step": 281590
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.273169051809903e-05,
+      "loss": 0.949,
+      "step": 281600
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.273054258819927e-05,
+      "loss": 1.0753,
+      "step": 281610
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.272939465829953e-05,
+      "loss": 1.0082,
+      "step": 281620
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.2728246728399784e-05,
+      "loss": 1.0024,
+      "step": 281630
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.2727098798500034e-05,
+      "loss": 0.9795,
+      "step": 281640
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.2725950868600284e-05,
+      "loss": 0.9971,
+      "step": 281650
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.272480293870054e-05,
+      "loss": 0.9959,
+      "step": 281660
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.272365500880079e-05,
+      "loss": 1.0414,
+      "step": 281670
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.272250707890105e-05,
+      "loss": 1.0114,
+      "step": 281680
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.2721359149001297e-05,
+      "loss": 1.0439,
+      "step": 281690
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.2720211219101546e-05,
+      "loss": 1.0913,
+      "step": 281700
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.27190632892018e-05,
+      "loss": 1.0051,
+      "step": 281710
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.271791535930206e-05,
+      "loss": 0.9148,
+      "step": 281720
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.27167674294023e-05,
+      "loss": 1.0331,
+      "step": 281730
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.271561949950256e-05,
+      "loss": 1.0081,
+      "step": 281740
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.2714471569602816e-05,
+      "loss": 0.8994,
+      "step": 281750
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.271332363970306e-05,
+      "loss": 1.0441,
+      "step": 281760
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.2712175709803316e-05,
+      "loss": 1.0084,
+      "step": 281770
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.271102777990357e-05,
+      "loss": 0.9812,
+      "step": 281780
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.270987985000382e-05,
+      "loss": 1.0189,
+      "step": 281790
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.270873192010407e-05,
+      "loss": 0.9774,
+      "step": 281800
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.270758399020433e-05,
+      "loss": 1.109,
+      "step": 281810
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.270643606030458e-05,
+      "loss": 1.0509,
+      "step": 281820
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.2705288130404835e-05,
+      "loss": 1.1079,
+      "step": 281830
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.2704140200505085e-05,
+      "loss": 1.0191,
+      "step": 281840
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.2702992270605335e-05,
+      "loss": 1.2386,
+      "step": 281850
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.270184434070559e-05,
+      "loss": 1.0433,
+      "step": 281860
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.270069641080585e-05,
+      "loss": 1.0273,
+      "step": 281870
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.269954848090609e-05,
+      "loss": 1.0784,
+      "step": 281880
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.269840055100635e-05,
+      "loss": 0.9748,
+      "step": 281890
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.2697252621106604e-05,
+      "loss": 1.0042,
+      "step": 281900
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.269610469120685e-05,
+      "loss": 0.9916,
+      "step": 281910
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.2694956761307104e-05,
+      "loss": 0.9793,
+      "step": 281920
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.269380883140736e-05,
+      "loss": 0.932,
+      "step": 281930
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.269266090150761e-05,
+      "loss": 1.0889,
+      "step": 281940
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.269151297160786e-05,
+      "loss": 0.9176,
+      "step": 281950
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.269036504170812e-05,
+      "loss": 1.0445,
+      "step": 281960
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.268921711180837e-05,
+      "loss": 0.9285,
+      "step": 281970
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.2688069181908624e-05,
+      "loss": 1.0587,
+      "step": 281980
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.2686921252008873e-05,
+      "loss": 0.9871,
+      "step": 281990
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.268577332210912e-05,
+      "loss": 1.0036,
+      "step": 282000
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.268462539220938e-05,
+      "loss": 1.0719,
+      "step": 282010
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.2683477462309637e-05,
+      "loss": 1.0292,
+      "step": 282020
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.268232953240988e-05,
+      "loss": 0.9409,
+      "step": 282030
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.2681181602510136e-05,
+      "loss": 1.041,
+      "step": 282040
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.268003367261039e-05,
+      "loss": 0.9542,
+      "step": 282050
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.2678885742710636e-05,
+      "loss": 1.0061,
+      "step": 282060
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.267773781281089e-05,
+      "loss": 0.9916,
+      "step": 282070
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.267658988291115e-05,
+      "loss": 0.9432,
+      "step": 282080
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.26754419530114e-05,
+      "loss": 1.1288,
+      "step": 282090
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.267429402311165e-05,
+      "loss": 1.0683,
+      "step": 282100
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.2673146093211906e-05,
+      "loss": 1.0432,
+      "step": 282110
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.2671998163312155e-05,
+      "loss": 1.014,
+      "step": 282120
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.267085023341241e-05,
+      "loss": 1.035,
+      "step": 282130
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.266970230351266e-05,
+      "loss": 1.1195,
+      "step": 282140
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.266855437361291e-05,
+      "loss": 0.9997,
+      "step": 282150
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.266740644371317e-05,
+      "loss": 1.0327,
+      "step": 282160
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.2666258513813425e-05,
+      "loss": 1.0179,
+      "step": 282170
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.266511058391367e-05,
+      "loss": 0.9372,
+      "step": 282180
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.2663962654013925e-05,
+      "loss": 1.1005,
+      "step": 282190
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.266281472411418e-05,
+      "loss": 1.1183,
+      "step": 282200
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.2661666794214424e-05,
+      "loss": 0.9677,
+      "step": 282210
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.266051886431468e-05,
+      "loss": 0.9721,
+      "step": 282220
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.265937093441494e-05,
+      "loss": 1.0131,
+      "step": 282230
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.265822300451519e-05,
+      "loss": 1.0411,
+      "step": 282240
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.265707507461544e-05,
+      "loss": 1.0271,
+      "step": 282250
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.2655927144715694e-05,
+      "loss": 1.0327,
+      "step": 282260
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.2654779214815944e-05,
+      "loss": 1.0347,
+      "step": 282270
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.26536312849162e-05,
+      "loss": 1.0873,
+      "step": 282280
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.265248335501645e-05,
+      "loss": 1.0897,
+      "step": 282290
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.26513354251167e-05,
+      "loss": 1.1116,
+      "step": 282300
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.265018749521696e-05,
+      "loss": 1.041,
+      "step": 282310
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.2649039565317214e-05,
+      "loss": 0.9553,
+      "step": 282320
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.2647891635417457e-05,
+      "loss": 0.9335,
+      "step": 282330
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.264674370551771e-05,
+      "loss": 1.0166,
+      "step": 282340
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.264559577561797e-05,
+      "loss": 0.9077,
+      "step": 282350
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.264444784571821e-05,
+      "loss": 0.9892,
+      "step": 282360
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.264329991581847e-05,
+      "loss": 1.1265,
+      "step": 282370
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.2642151985918726e-05,
+      "loss": 1.1364,
+      "step": 282380
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.2641004056018976e-05,
+      "loss": 1.0861,
+      "step": 282390
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.2639856126119226e-05,
+      "loss": 0.9755,
+      "step": 282400
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.263870819621948e-05,
+      "loss": 1.0825,
+      "step": 282410
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.263756026631973e-05,
+      "loss": 0.9921,
+      "step": 282420
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.263641233641999e-05,
+      "loss": 1.0829,
+      "step": 282430
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.263526440652024e-05,
+      "loss": 0.9873,
+      "step": 282440
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.263411647662049e-05,
+      "loss": 1.0148,
+      "step": 282450
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.2632968546720745e-05,
+      "loss": 1.0414,
+      "step": 282460
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.2631820616821e-05,
+      "loss": 1.0142,
+      "step": 282470
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.2630672686921245e-05,
+      "loss": 1.0701,
+      "step": 282480
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.26295247570215e-05,
+      "loss": 1.0213,
+      "step": 282490
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.262837682712176e-05,
+      "loss": 1.0247,
+      "step": 282500
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.2627228897222e-05,
+      "loss": 0.9685,
+      "step": 282510
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.262608096732226e-05,
+      "loss": 0.996,
+      "step": 282520
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.2624933037422515e-05,
+      "loss": 0.9694,
+      "step": 282530
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.2623785107522765e-05,
+      "loss": 1.1006,
+      "step": 282540
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.2622637177623014e-05,
+      "loss": 0.9642,
+      "step": 282550
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.262148924772327e-05,
+      "loss": 0.9517,
+      "step": 282560
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.262034131782352e-05,
+      "loss": 1.0234,
+      "step": 282570
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.261919338792378e-05,
+      "loss": 1.0932,
+      "step": 282580
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.261804545802402e-05,
+      "loss": 0.9537,
+      "step": 282590
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.261689752812428e-05,
+      "loss": 0.9919,
+      "step": 282600
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.2615749598224534e-05,
+      "loss": 1.0218,
+      "step": 282610
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.261460166832478e-05,
+      "loss": 1.115,
+      "step": 282620
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.2613453738425034e-05,
+      "loss": 0.9993,
+      "step": 282630
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.261230580852529e-05,
+      "loss": 1.0254,
+      "step": 282640
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.261115787862554e-05,
+      "loss": 1.0902,
+      "step": 282650
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.261000994872579e-05,
+      "loss": 0.9922,
+      "step": 282660
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.2608862018826047e-05,
+      "loss": 0.9834,
+      "step": 282670
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.2607714088926296e-05,
+      "loss": 0.9354,
+      "step": 282680
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.260656615902655e-05,
+      "loss": 1.0659,
+      "step": 282690
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.26054182291268e-05,
+      "loss": 0.9697,
+      "step": 282700
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.260427029922705e-05,
+      "loss": 1.0186,
+      "step": 282710
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.260312236932731e-05,
+      "loss": 1.0171,
+      "step": 282720
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.2601974439427566e-05,
+      "loss": 1.031,
+      "step": 282730
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.260082650952781e-05,
+      "loss": 1.012,
+      "step": 282740
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.2599678579628066e-05,
+      "loss": 1.0501,
+      "step": 282750
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.259853064972832e-05,
+      "loss": 0.9715,
+      "step": 282760
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.2597382719828565e-05,
+      "loss": 1.0819,
+      "step": 282770
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.259623478992882e-05,
+      "loss": 1.0451,
+      "step": 282780
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.259508686002908e-05,
+      "loss": 1.011,
+      "step": 282790
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.259393893012933e-05,
+      "loss": 1.16,
+      "step": 282800
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.259279100022958e-05,
+      "loss": 1.0045,
+      "step": 282810
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.2591643070329835e-05,
+      "loss": 0.9694,
+      "step": 282820
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.2590495140430085e-05,
+      "loss": 1.0546,
+      "step": 282830
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.258934721053034e-05,
+      "loss": 0.8889,
+      "step": 282840
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.258819928063059e-05,
+      "loss": 1.0511,
+      "step": 282850
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.258705135073084e-05,
+      "loss": 1.0155,
+      "step": 282860
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.25859034208311e-05,
+      "loss": 1.1101,
+      "step": 282870
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.2584755490931354e-05,
+      "loss": 1.0158,
+      "step": 282880
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.25836075610316e-05,
+      "loss": 0.9819,
+      "step": 282890
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.2582459631131854e-05,
+      "loss": 1.0099,
+      "step": 282900
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.258131170123211e-05,
+      "loss": 0.9403,
+      "step": 282910
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.2580163771332354e-05,
+      "loss": 1.0988,
+      "step": 282920
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.257901584143261e-05,
+      "loss": 0.8941,
+      "step": 282930
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.257786791153287e-05,
+      "loss": 0.9831,
+      "step": 282940
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.257671998163312e-05,
+      "loss": 0.9447,
+      "step": 282950
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.257557205173337e-05,
+      "loss": 1.1066,
+      "step": 282960
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.2574424121833623e-05,
+      "loss": 1.0422,
+      "step": 282970
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.257327619193387e-05,
+      "loss": 1.0888,
+      "step": 282980
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.257212826203413e-05,
+      "loss": 0.9975,
+      "step": 282990
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.257098033213438e-05,
+      "loss": 1.0778,
+      "step": 283000
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.256983240223463e-05,
+      "loss": 0.9902,
+      "step": 283010
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.2568684472334886e-05,
+      "loss": 1.0206,
+      "step": 283020
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.256753654243514e-05,
+      "loss": 0.9847,
+      "step": 283030
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.2566388612535386e-05,
+      "loss": 1.0263,
+      "step": 283040
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.256524068263564e-05,
+      "loss": 1.0454,
+      "step": 283050
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.25640927527359e-05,
+      "loss": 1.0249,
+      "step": 283060
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.256294482283614e-05,
+      "loss": 0.9791,
+      "step": 283070
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.25617968929364e-05,
+      "loss": 1.0047,
+      "step": 283080
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.2560648963036656e-05,
+      "loss": 1.009,
+      "step": 283090
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.2559501033136905e-05,
+      "loss": 0.9517,
+      "step": 283100
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.2558353103237155e-05,
+      "loss": 1.1314,
+      "step": 283110
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.255720517333741e-05,
+      "loss": 1.0325,
+      "step": 283120
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.255605724343766e-05,
+      "loss": 1.0651,
+      "step": 283130
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.255490931353792e-05,
+      "loss": 1.0791,
+      "step": 283140
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.255376138363817e-05,
+      "loss": 1.0902,
+      "step": 283150
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.255261345373842e-05,
+      "loss": 1.0021,
+      "step": 283160
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.2551465523838675e-05,
+      "loss": 0.9991,
+      "step": 283170
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.255031759393893e-05,
+      "loss": 0.986,
+      "step": 283180
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.2549169664039174e-05,
+      "loss": 1.007,
+      "step": 283190
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.254802173413943e-05,
+      "loss": 1.1028,
+      "step": 283200
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.254687380423969e-05,
+      "loss": 1.0051,
+      "step": 283210
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.254572587433993e-05,
+      "loss": 1.102,
+      "step": 283220
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.254457794444019e-05,
+      "loss": 1.0028,
+      "step": 283230
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.2543430014540444e-05,
+      "loss": 1.0615,
+      "step": 283240
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.2542282084640694e-05,
+      "loss": 0.9039,
+      "step": 283250
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.2541134154740944e-05,
+      "loss": 1.0717,
+      "step": 283260
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.25399862248412e-05,
+      "loss": 1.0671,
+      "step": 283270
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.253883829494145e-05,
+      "loss": 1.0125,
+      "step": 283280
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.253769036504171e-05,
+      "loss": 1.0661,
+      "step": 283290
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.253654243514196e-05,
+      "loss": 1.0608,
+      "step": 283300
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.2535394505242207e-05,
+      "loss": 0.9681,
+      "step": 283310
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.253424657534246e-05,
+      "loss": 1.0184,
+      "step": 283320
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.253309864544272e-05,
+      "loss": 0.9258,
+      "step": 283330
+    },
+    {
+      "epoch": 13.0,
+      "eval_cer": 0.6690984190052434,
+      "eval_em": 0.007454017424975799,
+      "eval_f1": 0.007454017424975799,
+      "eval_loss": 1.2230417728424072,
+      "eval_runtime": 1677.6949,
+      "eval_samples_per_second": 6.157,
+      "eval_steps_per_second": 0.77,
+      "eval_wer": 0.9925459825750242,
+      "step": 283336
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.253195071554296e-05,
+      "loss": 0.9765,
+      "step": 283340
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.253080278564322e-05,
+      "loss": 0.9696,
+      "step": 283350
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.2529654855743476e-05,
+      "loss": 0.887,
+      "step": 283360
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.252850692584372e-05,
+      "loss": 0.9081,
+      "step": 283370
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.2527358995943976e-05,
+      "loss": 0.9766,
+      "step": 283380
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.252621106604423e-05,
+      "loss": 0.9563,
+      "step": 283390
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.252506313614448e-05,
+      "loss": 0.8664,
+      "step": 283400
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.252391520624473e-05,
+      "loss": 1.0544,
+      "step": 283410
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.252276727634499e-05,
+      "loss": 1.006,
+      "step": 283420
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.252161934644524e-05,
+      "loss": 1.0658,
+      "step": 283430
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.2520471416545495e-05,
+      "loss": 1.012,
+      "step": 283440
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.2519323486645745e-05,
+      "loss": 0.9323,
+      "step": 283450
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.2518175556745995e-05,
+      "loss": 0.9613,
+      "step": 283460
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.251702762684625e-05,
+      "loss": 0.9076,
+      "step": 283470
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.251587969694651e-05,
+      "loss": 0.9951,
+      "step": 283480
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.251473176704675e-05,
+      "loss": 0.9607,
+      "step": 283490
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.251358383714701e-05,
+      "loss": 1.0649,
+      "step": 283500
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.2512435907247265e-05,
+      "loss": 0.9791,
+      "step": 283510
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.251128797734751e-05,
+      "loss": 1.0303,
+      "step": 283520
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.2510140047447764e-05,
+      "loss": 1.0555,
+      "step": 283530
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.250899211754802e-05,
+      "loss": 1.0123,
+      "step": 283540
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.250784418764827e-05,
+      "loss": 1.037,
+      "step": 283550
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.250669625774852e-05,
+      "loss": 0.959,
+      "step": 283560
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.250554832784878e-05,
+      "loss": 1.0374,
+      "step": 283570
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.250440039794903e-05,
+      "loss": 0.9678,
+      "step": 283580
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.2503252468049284e-05,
+      "loss": 0.951,
+      "step": 283590
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.2502104538149534e-05,
+      "loss": 0.9162,
+      "step": 283600
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.2500956608249784e-05,
+      "loss": 0.9103,
+      "step": 283610
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.249980867835004e-05,
+      "loss": 1.0018,
+      "step": 283620
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.24986607484503e-05,
+      "loss": 0.9945,
+      "step": 283630
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.249751281855054e-05,
+      "loss": 0.9564,
+      "step": 283640
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.2496364888650797e-05,
+      "loss": 0.934,
+      "step": 283650
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.249521695875105e-05,
+      "loss": 0.9305,
+      "step": 283660
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.2494069028851296e-05,
+      "loss": 0.9957,
+      "step": 283670
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.249292109895155e-05,
+      "loss": 0.9586,
+      "step": 283680
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.249177316905181e-05,
+      "loss": 0.8374,
+      "step": 283690
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.249062523915206e-05,
+      "loss": 1.0258,
+      "step": 283700
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.248947730925231e-05,
+      "loss": 0.9731,
+      "step": 283710
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.2488329379352566e-05,
+      "loss": 1.0258,
+      "step": 283720
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.2487181449452816e-05,
+      "loss": 0.9875,
+      "step": 283730
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.248603351955307e-05,
+      "loss": 0.9752,
+      "step": 283740
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.248488558965332e-05,
+      "loss": 1.0116,
+      "step": 283750
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.248373765975357e-05,
+      "loss": 0.8882,
+      "step": 283760
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.248258972985383e-05,
+      "loss": 0.9834,
+      "step": 283770
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.2481441799954085e-05,
+      "loss": 1.0038,
+      "step": 283780
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.248029387005433e-05,
+      "loss": 0.9724,
+      "step": 283790
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.2479145940154585e-05,
+      "loss": 0.9339,
+      "step": 283800
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.247799801025484e-05,
+      "loss": 0.9398,
+      "step": 283810
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.2476850080355085e-05,
+      "loss": 0.9542,
+      "step": 283820
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.247570215045534e-05,
+      "loss": 0.9174,
+      "step": 283830
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.24745542205556e-05,
+      "loss": 0.9259,
+      "step": 283840
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.247340629065585e-05,
+      "loss": 0.9382,
+      "step": 283850
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.24722583607561e-05,
+      "loss": 0.9829,
+      "step": 283860
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.2471110430856354e-05,
+      "loss": 1.0077,
+      "step": 283870
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.2469962500956604e-05,
+      "loss": 1.0953,
+      "step": 283880
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.246881457105686e-05,
+      "loss": 0.9914,
+      "step": 283890
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.246766664115711e-05,
+      "loss": 0.9504,
+      "step": 283900
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.246651871125736e-05,
+      "loss": 0.919,
+      "step": 283910
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.246537078135762e-05,
+      "loss": 0.964,
+      "step": 283920
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.2464222851457874e-05,
+      "loss": 1.0226,
+      "step": 283930
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.246307492155812e-05,
+      "loss": 1.032,
+      "step": 283940
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.2461926991658373e-05,
+      "loss": 0.9077,
+      "step": 283950
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.246077906175862e-05,
+      "loss": 0.8555,
+      "step": 283960
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.245963113185887e-05,
+      "loss": 0.9408,
+      "step": 283970
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.245848320195913e-05,
+      "loss": 0.9174,
+      "step": 283980
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.245733527205938e-05,
+      "loss": 1.0054,
+      "step": 283990
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.2456187342159636e-05,
+      "loss": 1.026,
+      "step": 284000
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.2455039412259886e-05,
+      "loss": 0.9382,
+      "step": 284010
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.2453891482360136e-05,
+      "loss": 1.001,
+      "step": 284020
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.245274355246039e-05,
+      "loss": 0.9593,
+      "step": 284030
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.245159562256065e-05,
+      "loss": 1.0112,
+      "step": 284040
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.245044769266089e-05,
+      "loss": 0.9921,
+      "step": 284050
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.244929976276115e-05,
+      "loss": 1.0064,
+      "step": 284060
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.2448151832861406e-05,
+      "loss": 0.9774,
+      "step": 284070
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.244700390296165e-05,
+      "loss": 0.9656,
+      "step": 284080
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.2445855973061905e-05,
+      "loss": 0.9175,
+      "step": 284090
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.244470804316216e-05,
+      "loss": 0.9491,
+      "step": 284100
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.244356011326241e-05,
+      "loss": 0.939,
+      "step": 284110
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.244241218336266e-05,
+      "loss": 0.9742,
+      "step": 284120
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.244126425346292e-05,
+      "loss": 0.9537,
+      "step": 284130
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.244011632356317e-05,
+      "loss": 1.0075,
+      "step": 284140
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.2438968393663425e-05,
+      "loss": 0.9324,
+      "step": 284150
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.2437820463763675e-05,
+      "loss": 0.9622,
+      "step": 284160
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.2436672533863924e-05,
+      "loss": 1.0169,
+      "step": 284170
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.243552460396418e-05,
+      "loss": 1.0205,
+      "step": 284180
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.243437667406444e-05,
+      "loss": 0.9186,
+      "step": 284190
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.243322874416468e-05,
+      "loss": 1.0011,
+      "step": 284200
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.243208081426494e-05,
+      "loss": 0.8847,
+      "step": 284210
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.2430932884365194e-05,
+      "loss": 1.112,
+      "step": 284220
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.242978495446544e-05,
+      "loss": 1.0008,
+      "step": 284230
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.2428637024565694e-05,
+      "loss": 0.969,
+      "step": 284240
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.242748909466595e-05,
+      "loss": 0.9072,
+      "step": 284250
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.24263411647662e-05,
+      "loss": 1.0727,
+      "step": 284260
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.242519323486645e-05,
+      "loss": 0.8964,
+      "step": 284270
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.242404530496671e-05,
+      "loss": 0.9659,
+      "step": 284280
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.2422897375066957e-05,
+      "loss": 0.9685,
+      "step": 284290
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.242174944516721e-05,
+      "loss": 1.0068,
+      "step": 284300
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.242060151526746e-05,
+      "loss": 1.0029,
+      "step": 284310
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.241945358536771e-05,
+      "loss": 1.0654,
+      "step": 284320
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.241830565546797e-05,
+      "loss": 0.9023,
+      "step": 284330
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.2417157725568226e-05,
+      "loss": 1.0423,
+      "step": 284340
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.241600979566847e-05,
+      "loss": 0.8784,
+      "step": 284350
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.2414861865768726e-05,
+      "loss": 0.8846,
+      "step": 284360
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.241371393586898e-05,
+      "loss": 0.9985,
+      "step": 284370
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.2412566005969226e-05,
+      "loss": 0.9953,
+      "step": 284380
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.241141807606948e-05,
+      "loss": 0.9764,
+      "step": 284390
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.241027014616974e-05,
+      "loss": 0.8392,
+      "step": 284400
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.240912221626999e-05,
+      "loss": 0.9669,
+      "step": 284410
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.240797428637024e-05,
+      "loss": 0.9161,
+      "step": 284420
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.2406826356470495e-05,
+      "loss": 0.9132,
+      "step": 284430
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.2405678426570745e-05,
+      "loss": 0.8992,
+      "step": 284440
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.2404530496671e-05,
+      "loss": 0.9551,
+      "step": 284450
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.240338256677125e-05,
+      "loss": 0.9877,
+      "step": 284460
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.24022346368715e-05,
+      "loss": 0.9445,
+      "step": 284470
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.240108670697176e-05,
+      "loss": 0.9306,
+      "step": 284480
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.2399938777072015e-05,
+      "loss": 0.9942,
+      "step": 284490
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.239879084717226e-05,
+      "loss": 0.916,
+      "step": 284500
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.2397642917272514e-05,
+      "loss": 0.9613,
+      "step": 284510
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.239649498737277e-05,
+      "loss": 0.9876,
+      "step": 284520
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.2395347057473014e-05,
+      "loss": 0.906,
+      "step": 284530
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.239419912757327e-05,
+      "loss": 0.913,
+      "step": 284540
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.239305119767353e-05,
+      "loss": 1.0424,
+      "step": 284550
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.239190326777378e-05,
+      "loss": 0.938,
+      "step": 284560
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.239075533787403e-05,
+      "loss": 0.9951,
+      "step": 284570
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.2389607407974284e-05,
+      "loss": 0.9148,
+      "step": 284580
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.2388459478074534e-05,
+      "loss": 0.9965,
+      "step": 284590
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.238731154817479e-05,
+      "loss": 1.0505,
+      "step": 284600
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.238616361827504e-05,
+      "loss": 0.8872,
+      "step": 284610
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.238501568837529e-05,
+      "loss": 1.0215,
+      "step": 284620
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.2383867758475547e-05,
+      "loss": 0.961,
+      "step": 284630
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.23827198285758e-05,
+      "loss": 0.8442,
+      "step": 284640
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.2381571898676046e-05,
+      "loss": 0.9815,
+      "step": 284650
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.23804239687763e-05,
+      "loss": 0.9673,
+      "step": 284660
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.237927603887656e-05,
+      "loss": 0.9408,
+      "step": 284670
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.23781281089768e-05,
+      "loss": 0.914,
+      "step": 284680
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.237698017907706e-05,
+      "loss": 1.0337,
+      "step": 284690
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.2375832249177316e-05,
+      "loss": 1.0174,
+      "step": 284700
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.2374684319277566e-05,
+      "loss": 1.0195,
+      "step": 284710
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.2373536389377816e-05,
+      "loss": 0.9583,
+      "step": 284720
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.237238845947807e-05,
+      "loss": 1.0,
+      "step": 284730
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.237124052957832e-05,
+      "loss": 0.864,
+      "step": 284740
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.237009259967858e-05,
+      "loss": 0.9585,
+      "step": 284750
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.236894466977883e-05,
+      "loss": 0.9633,
+      "step": 284760
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.236779673987908e-05,
+      "loss": 0.885,
+      "step": 284770
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.2366648809979335e-05,
+      "loss": 0.9188,
+      "step": 284780
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.236550088007959e-05,
+      "loss": 1.0763,
+      "step": 284790
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.2364352950179835e-05,
+      "loss": 0.8967,
+      "step": 284800
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.236320502028009e-05,
+      "loss": 0.8717,
+      "step": 284810
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.236205709038035e-05,
+      "loss": 0.9793,
+      "step": 284820
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.236090916048059e-05,
+      "loss": 0.9867,
+      "step": 284830
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.235976123058085e-05,
+      "loss": 0.9403,
+      "step": 284840
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.2358613300681104e-05,
+      "loss": 0.9755,
+      "step": 284850
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.2357465370781354e-05,
+      "loss": 0.9295,
+      "step": 284860
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.2356317440881604e-05,
+      "loss": 0.9857,
+      "step": 284870
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.235516951098186e-05,
+      "loss": 0.9986,
+      "step": 284880
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.235402158108211e-05,
+      "loss": 0.8881,
+      "step": 284890
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.235287365118237e-05,
+      "loss": 1.0101,
+      "step": 284900
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.235172572128262e-05,
+      "loss": 0.9546,
+      "step": 284910
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.235057779138287e-05,
+      "loss": 1.0496,
+      "step": 284920
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.2349429861483123e-05,
+      "loss": 0.9816,
+      "step": 284930
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.234828193158338e-05,
+      "loss": 0.9993,
+      "step": 284940
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.234713400168362e-05,
+      "loss": 1.0331,
+      "step": 284950
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.234598607178388e-05,
+      "loss": 0.9917,
+      "step": 284960
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.2344838141884136e-05,
+      "loss": 1.0382,
+      "step": 284970
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.234369021198438e-05,
+      "loss": 1.0107,
+      "step": 284980
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.2342542282084636e-05,
+      "loss": 1.0489,
+      "step": 284990
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.234139435218489e-05,
+      "loss": 0.9823,
+      "step": 285000
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.234024642228514e-05,
+      "loss": 0.9639,
+      "step": 285010
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.233909849238539e-05,
+      "loss": 0.9761,
+      "step": 285020
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.233795056248565e-05,
+      "loss": 0.9774,
+      "step": 285030
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.23368026325859e-05,
+      "loss": 0.9643,
+      "step": 285040
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.2335654702686156e-05,
+      "loss": 1.0253,
+      "step": 285050
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.2334506772786405e-05,
+      "loss": 0.9832,
+      "step": 285060
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.2333358842886655e-05,
+      "loss": 1.0212,
+      "step": 285070
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.233221091298691e-05,
+      "loss": 1.01,
+      "step": 285080
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.233106298308717e-05,
+      "loss": 0.9129,
+      "step": 285090
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.232991505318741e-05,
+      "loss": 1.0425,
+      "step": 285100
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.232876712328767e-05,
+      "loss": 0.9292,
+      "step": 285110
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.2327619193387925e-05,
+      "loss": 0.9357,
+      "step": 285120
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.232647126348817e-05,
+      "loss": 0.9327,
+      "step": 285130
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.2325323333588425e-05,
+      "loss": 0.9968,
+      "step": 285140
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.232417540368868e-05,
+      "loss": 1.0732,
+      "step": 285150
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.232302747378893e-05,
+      "loss": 0.9191,
+      "step": 285160
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.232187954388918e-05,
+      "loss": 1.0198,
+      "step": 285170
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.232073161398944e-05,
+      "loss": 0.9733,
+      "step": 285180
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.231958368408969e-05,
+      "loss": 0.9174,
+      "step": 285190
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.2318435754189944e-05,
+      "loss": 0.9563,
+      "step": 285200
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.2317287824290194e-05,
+      "loss": 0.9333,
+      "step": 285210
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.2316139894390444e-05,
+      "loss": 0.9039,
+      "step": 285220
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.23149919644907e-05,
+      "loss": 0.9424,
+      "step": 285230
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.231384403459096e-05,
+      "loss": 0.9452,
+      "step": 285240
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.23126961046912e-05,
+      "loss": 0.9195,
+      "step": 285250
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.231154817479146e-05,
+      "loss": 1.0112,
+      "step": 285260
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.231040024489171e-05,
+      "loss": 0.9817,
+      "step": 285270
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.2309252314991956e-05,
+      "loss": 0.9275,
+      "step": 285280
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.230810438509221e-05,
+      "loss": 0.9089,
+      "step": 285290
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.230695645519247e-05,
+      "loss": 0.9364,
+      "step": 285300
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.230580852529272e-05,
+      "loss": 0.9662,
+      "step": 285310
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.230466059539297e-05,
+      "loss": 0.9544,
+      "step": 285320
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.230351266549322e-05,
+      "loss": 0.9561,
+      "step": 285330
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.2302364735593476e-05,
+      "loss": 0.9464,
+      "step": 285340
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.230121680569373e-05,
+      "loss": 0.9617,
+      "step": 285350
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.2300068875793976e-05,
+      "loss": 0.9487,
+      "step": 285360
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.229892094589423e-05,
+      "loss": 0.9336,
+      "step": 285370
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.229777301599449e-05,
+      "loss": 0.9659,
+      "step": 285380
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.229662508609473e-05,
+      "loss": 0.9862,
+      "step": 285390
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.229547715619499e-05,
+      "loss": 0.9796,
+      "step": 285400
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.2294329226295245e-05,
+      "loss": 1.0251,
+      "step": 285410
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.2293181296395495e-05,
+      "loss": 1.0025,
+      "step": 285420
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.2292033366495745e-05,
+      "loss": 0.8948,
+      "step": 285430
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.2290885436596e-05,
+      "loss": 0.9713,
+      "step": 285440
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.228973750669625e-05,
+      "loss": 0.9083,
+      "step": 285450
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.228858957679651e-05,
+      "loss": 0.9525,
+      "step": 285460
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.228744164689676e-05,
+      "loss": 0.9578,
+      "step": 285470
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.228629371699701e-05,
+      "loss": 0.9363,
+      "step": 285480
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.2285145787097264e-05,
+      "loss": 1.0466,
+      "step": 285490
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.228399785719752e-05,
+      "loss": 1.0044,
+      "step": 285500
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.2282849927297764e-05,
+      "loss": 1.0196,
+      "step": 285510
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.228170199739802e-05,
+      "loss": 0.918,
+      "step": 285520
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.228055406749828e-05,
+      "loss": 0.9443,
+      "step": 285530
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.227940613759852e-05,
+      "loss": 0.9302,
+      "step": 285540
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.227825820769878e-05,
+      "loss": 1.0209,
+      "step": 285550
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.2277110277799034e-05,
+      "loss": 1.0375,
+      "step": 285560
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.2275962347899284e-05,
+      "loss": 0.9041,
+      "step": 285570
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.2274814417999533e-05,
+      "loss": 0.9695,
+      "step": 285580
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.227366648809979e-05,
+      "loss": 1.0032,
+      "step": 285590
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.227251855820004e-05,
+      "loss": 0.985,
+      "step": 285600
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.2271370628300297e-05,
+      "loss": 0.9576,
+      "step": 285610
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.2270222698400546e-05,
+      "loss": 1.0724,
+      "step": 285620
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.2269074768500796e-05,
+      "loss": 0.9389,
+      "step": 285630
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.226792683860105e-05,
+      "loss": 0.9391,
+      "step": 285640
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.226677890870131e-05,
+      "loss": 0.9309,
+      "step": 285650
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.226563097880155e-05,
+      "loss": 1.0093,
+      "step": 285660
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.226448304890181e-05,
+      "loss": 0.9304,
+      "step": 285670
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.2263335119002066e-05,
+      "loss": 0.9965,
+      "step": 285680
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.226218718910231e-05,
+      "loss": 0.9742,
+      "step": 285690
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.2261039259202566e-05,
+      "loss": 1.0413,
+      "step": 285700
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.225989132930282e-05,
+      "loss": 0.9435,
+      "step": 285710
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.225874339940307e-05,
+      "loss": 0.9583,
+      "step": 285720
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.225759546950332e-05,
+      "loss": 0.9355,
+      "step": 285730
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.225644753960358e-05,
+      "loss": 0.9432,
+      "step": 285740
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.225529960970383e-05,
+      "loss": 0.8824,
+      "step": 285750
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.2254151679804085e-05,
+      "loss": 1.0267,
+      "step": 285760
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.2253003749904335e-05,
+      "loss": 0.9644,
+      "step": 285770
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.2251855820004585e-05,
+      "loss": 0.9339,
+      "step": 285780
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.225070789010484e-05,
+      "loss": 0.9642,
+      "step": 285790
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.22495599602051e-05,
+      "loss": 1.0155,
+      "step": 285800
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.224841203030534e-05,
+      "loss": 0.9992,
+      "step": 285810
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.22472641004056e-05,
+      "loss": 0.9761,
+      "step": 285820
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.2246116170505854e-05,
+      "loss": 0.9712,
+      "step": 285830
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.22449682406061e-05,
+      "loss": 0.9485,
+      "step": 285840
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.2243820310706354e-05,
+      "loss": 0.9099,
+      "step": 285850
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.224267238080661e-05,
+      "loss": 0.9829,
+      "step": 285860
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.224152445090686e-05,
+      "loss": 1.0674,
+      "step": 285870
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.224037652100711e-05,
+      "loss": 0.9209,
+      "step": 285880
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.223922859110737e-05,
+      "loss": 0.9628,
+      "step": 285890
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.223808066120762e-05,
+      "loss": 1.0478,
+      "step": 285900
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.2236932731307873e-05,
+      "loss": 0.9222,
+      "step": 285910
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.223578480140812e-05,
+      "loss": 0.9994,
+      "step": 285920
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.223463687150837e-05,
+      "loss": 0.9113,
+      "step": 285930
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.223348894160863e-05,
+      "loss": 1.0565,
+      "step": 285940
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.2232341011708886e-05,
+      "loss": 0.9205,
+      "step": 285950
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.223119308180913e-05,
+      "loss": 1.1421,
+      "step": 285960
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.2230045151909386e-05,
+      "loss": 1.0458,
+      "step": 285970
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.222889722200964e-05,
+      "loss": 0.9594,
+      "step": 285980
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.2227749292109886e-05,
+      "loss": 1.0963,
+      "step": 285990
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.222660136221014e-05,
+      "loss": 0.8971,
+      "step": 286000
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.22254534323104e-05,
+      "loss": 0.9415,
+      "step": 286010
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.222430550241065e-05,
+      "loss": 1.0672,
+      "step": 286020
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.22231575725109e-05,
+      "loss": 0.9571,
+      "step": 286030
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.2222009642611155e-05,
+      "loss": 0.9965,
+      "step": 286040
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.2220861712711405e-05,
+      "loss": 1.018,
+      "step": 286050
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.221971378281166e-05,
+      "loss": 0.9591,
+      "step": 286060
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.221856585291191e-05,
+      "loss": 1.0488,
+      "step": 286070
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.221741792301216e-05,
+      "loss": 0.9664,
+      "step": 286080
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.221626999311242e-05,
+      "loss": 0.915,
+      "step": 286090
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.2215122063212675e-05,
+      "loss": 0.9959,
+      "step": 286100
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.221397413331292e-05,
+      "loss": 0.9353,
+      "step": 286110
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.2212826203413175e-05,
+      "loss": 0.9844,
+      "step": 286120
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.221167827351343e-05,
+      "loss": 0.9178,
+      "step": 286130
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.2210530343613674e-05,
+      "loss": 0.9927,
+      "step": 286140
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.220938241371393e-05,
+      "loss": 1.0241,
+      "step": 286150
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.220823448381419e-05,
+      "loss": 0.9426,
+      "step": 286160
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.220708655391444e-05,
+      "loss": 0.9414,
+      "step": 286170
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.220593862401469e-05,
+      "loss": 0.9481,
+      "step": 286180
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.2204790694114944e-05,
+      "loss": 0.9809,
+      "step": 286190
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.2203642764215194e-05,
+      "loss": 1.0156,
+      "step": 286200
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.220249483431545e-05,
+      "loss": 0.9256,
+      "step": 286210
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.22013469044157e-05,
+      "loss": 0.9717,
+      "step": 286220
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.220019897451595e-05,
+      "loss": 1.0615,
+      "step": 286230
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.219905104461621e-05,
+      "loss": 0.8888,
+      "step": 286240
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.219790311471646e-05,
+      "loss": 0.8998,
+      "step": 286250
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.2196755184816706e-05,
+      "loss": 0.9355,
+      "step": 286260
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.219560725491696e-05,
+      "loss": 1.0241,
+      "step": 286270
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.219445932501722e-05,
+      "loss": 0.989,
+      "step": 286280
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.219331139511746e-05,
+      "loss": 1.0101,
+      "step": 286290
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.219216346521772e-05,
+      "loss": 1.1223,
+      "step": 286300
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.2191015535317976e-05,
+      "loss": 0.8878,
+      "step": 286310
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.2189867605418226e-05,
+      "loss": 1.0456,
+      "step": 286320
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.2188719675518476e-05,
+      "loss": 0.8949,
+      "step": 286330
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.218757174561873e-05,
+      "loss": 0.9361,
+      "step": 286340
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.218642381571898e-05,
+      "loss": 0.9852,
+      "step": 286350
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.218527588581924e-05,
+      "loss": 0.9127,
+      "step": 286360
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.218412795591949e-05,
+      "loss": 0.9988,
+      "step": 286370
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.218298002601974e-05,
+      "loss": 0.9393,
+      "step": 286380
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.2181832096119995e-05,
+      "loss": 0.999,
+      "step": 286390
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.218068416622025e-05,
+      "loss": 1.0524,
+      "step": 286400
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.2179536236320495e-05,
+      "loss": 0.9307,
+      "step": 286410
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.217838830642075e-05,
+      "loss": 0.8667,
+      "step": 286420
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.217724037652101e-05,
+      "loss": 0.9721,
+      "step": 286430
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.217609244662125e-05,
+      "loss": 0.9733,
+      "step": 286440
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.217494451672151e-05,
+      "loss": 1.0772,
+      "step": 286450
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.2173796586821765e-05,
+      "loss": 1.0059,
+      "step": 286460
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.2172648656922014e-05,
+      "loss": 1.1297,
+      "step": 286470
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.2171500727022264e-05,
+      "loss": 0.9241,
+      "step": 286480
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.217035279712252e-05,
+      "loss": 0.9615,
+      "step": 286490
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.216920486722277e-05,
+      "loss": 0.9426,
+      "step": 286500
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.216805693732303e-05,
+      "loss": 0.9543,
+      "step": 286510
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.216690900742328e-05,
+      "loss": 1.0719,
+      "step": 286520
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.216576107752353e-05,
+      "loss": 1.012,
+      "step": 286530
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.2164613147623784e-05,
+      "loss": 0.9643,
+      "step": 286540
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.216346521772404e-05,
+      "loss": 0.9475,
+      "step": 286550
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.216231728782428e-05,
+      "loss": 0.9592,
+      "step": 286560
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.216116935792454e-05,
+      "loss": 0.9398,
+      "step": 286570
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.21600214280248e-05,
+      "loss": 0.9574,
+      "step": 286580
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.215887349812504e-05,
+      "loss": 0.9269,
+      "step": 286590
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.2157725568225296e-05,
+      "loss": 0.9367,
+      "step": 286600
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.215657763832555e-05,
+      "loss": 0.8891,
+      "step": 286610
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.21554297084258e-05,
+      "loss": 0.9692,
+      "step": 286620
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.215428177852605e-05,
+      "loss": 1.0133,
+      "step": 286630
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.215313384862631e-05,
+      "loss": 0.916,
+      "step": 286640
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.215198591872656e-05,
+      "loss": 1.005,
+      "step": 286650
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.2150837988826816e-05,
+      "loss": 1.0148,
+      "step": 286660
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.2149690058927066e-05,
+      "loss": 1.0341,
+      "step": 286670
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.2148542129027316e-05,
+      "loss": 1.0546,
+      "step": 286680
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.214739419912757e-05,
+      "loss": 0.9374,
+      "step": 286690
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.2146246269227815e-05,
+      "loss": 0.9941,
+      "step": 286700
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.214509833932807e-05,
+      "loss": 0.9389,
+      "step": 286710
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.214395040942833e-05,
+      "loss": 0.9883,
+      "step": 286720
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.214280247952858e-05,
+      "loss": 1.0292,
+      "step": 286730
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.214165454962883e-05,
+      "loss": 1.0017,
+      "step": 286740
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.2140506619729085e-05,
+      "loss": 1.047,
+      "step": 286750
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.2139358689829335e-05,
+      "loss": 1.0606,
+      "step": 286760
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.213821075992959e-05,
+      "loss": 0.9785,
+      "step": 286770
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.213706283002984e-05,
+      "loss": 1.0002,
+      "step": 286780
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.213591490013009e-05,
+      "loss": 0.8975,
+      "step": 286790
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.213476697023035e-05,
+      "loss": 0.9304,
+      "step": 286800
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.2133619040330604e-05,
+      "loss": 0.9261,
+      "step": 286810
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.213247111043085e-05,
+      "loss": 1.0021,
+      "step": 286820
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.2131323180531104e-05,
+      "loss": 0.9953,
+      "step": 286830
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.213017525063136e-05,
+      "loss": 0.9812,
+      "step": 286840
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.2129027320731604e-05,
+      "loss": 0.9942,
+      "step": 286850
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.212787939083186e-05,
+      "loss": 1.0365,
+      "step": 286860
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.212673146093212e-05,
+      "loss": 1.0782,
+      "step": 286870
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.212558353103237e-05,
+      "loss": 0.8739,
+      "step": 286880
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.212443560113262e-05,
+      "loss": 0.9223,
+      "step": 286890
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.212328767123287e-05,
+      "loss": 1.0909,
+      "step": 286900
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.212213974133312e-05,
+      "loss": 0.926,
+      "step": 286910
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.212099181143338e-05,
+      "loss": 0.9409,
+      "step": 286920
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.211984388153363e-05,
+      "loss": 0.8441,
+      "step": 286930
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.211869595163388e-05,
+      "loss": 1.1488,
+      "step": 286940
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.2117548021734136e-05,
+      "loss": 0.9706,
+      "step": 286950
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.211640009183439e-05,
+      "loss": 0.9947,
+      "step": 286960
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.2115252161934636e-05,
+      "loss": 0.911,
+      "step": 286970
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.211410423203489e-05,
+      "loss": 0.9154,
+      "step": 286980
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.211295630213515e-05,
+      "loss": 0.9144,
+      "step": 286990
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.211180837223539e-05,
+      "loss": 0.8889,
+      "step": 287000
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.211066044233565e-05,
+      "loss": 0.9547,
+      "step": 287010
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.2109512512435905e-05,
+      "loss": 0.8621,
+      "step": 287020
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.2108364582536155e-05,
+      "loss": 0.977,
+      "step": 287030
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.2107216652636405e-05,
+      "loss": 0.9488,
+      "step": 287040
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.210606872273666e-05,
+      "loss": 0.9659,
+      "step": 287050
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.210492079283691e-05,
+      "loss": 1.0243,
+      "step": 287060
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.210377286293717e-05,
+      "loss": 1.0107,
+      "step": 287070
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.210262493303742e-05,
+      "loss": 0.9999,
+      "step": 287080
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.210147700313767e-05,
+      "loss": 0.9727,
+      "step": 287090
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.2100329073237925e-05,
+      "loss": 0.9207,
+      "step": 287100
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.209918114333818e-05,
+      "loss": 0.9543,
+      "step": 287110
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.2098033213438424e-05,
+      "loss": 0.8923,
+      "step": 287120
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.209688528353868e-05,
+      "loss": 1.0706,
+      "step": 287130
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.209573735363894e-05,
+      "loss": 0.9651,
+      "step": 287140
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.209458942373918e-05,
+      "loss": 0.983,
+      "step": 287150
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.209344149383944e-05,
+      "loss": 0.9317,
+      "step": 287160
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.2092293563939694e-05,
+      "loss": 0.9022,
+      "step": 287170
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.2091145634039944e-05,
+      "loss": 1.084,
+      "step": 287180
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.2089997704140194e-05,
+      "loss": 0.9192,
+      "step": 287190
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.208884977424045e-05,
+      "loss": 0.9457,
+      "step": 287200
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.20877018443407e-05,
+      "loss": 0.9548,
+      "step": 287210
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.208655391444096e-05,
+      "loss": 0.9977,
+      "step": 287220
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.2085405984541207e-05,
+      "loss": 0.8787,
+      "step": 287230
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.2084258054641456e-05,
+      "loss": 0.9206,
+      "step": 287240
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.208311012474171e-05,
+      "loss": 0.8577,
+      "step": 287250
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.208196219484197e-05,
+      "loss": 0.9563,
+      "step": 287260
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.208081426494221e-05,
+      "loss": 0.9432,
+      "step": 287270
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.207966633504247e-05,
+      "loss": 1.0284,
+      "step": 287280
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.2078518405142726e-05,
+      "loss": 0.9912,
+      "step": 287290
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.207737047524297e-05,
+      "loss": 0.9371,
+      "step": 287300
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.2076222545343226e-05,
+      "loss": 0.962,
+      "step": 287310
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.207507461544348e-05,
+      "loss": 1.0054,
+      "step": 287320
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.207392668554373e-05,
+      "loss": 0.9279,
+      "step": 287330
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.207277875564398e-05,
+      "loss": 0.9303,
+      "step": 287340
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.207163082574424e-05,
+      "loss": 0.8847,
+      "step": 287350
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.207048289584449e-05,
+      "loss": 0.9846,
+      "step": 287360
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.2069334965944745e-05,
+      "loss": 0.9873,
+      "step": 287370
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.2068187036044995e-05,
+      "loss": 0.9415,
+      "step": 287380
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.2067039106145245e-05,
+      "loss": 1.1061,
+      "step": 287390
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.20658911762455e-05,
+      "loss": 0.901,
+      "step": 287400
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.206474324634576e-05,
+      "loss": 0.9323,
+      "step": 287410
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.2063595316446e-05,
+      "loss": 0.9236,
+      "step": 287420
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.206244738654626e-05,
+      "loss": 0.9809,
+      "step": 287430
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.2061299456646515e-05,
+      "loss": 0.9855,
+      "step": 287440
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.206015152674676e-05,
+      "loss": 1.1318,
+      "step": 287450
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.2059003596847014e-05,
+      "loss": 0.9803,
+      "step": 287460
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.205785566694727e-05,
+      "loss": 0.8756,
+      "step": 287470
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.205670773704752e-05,
+      "loss": 0.9312,
+      "step": 287480
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.205555980714777e-05,
+      "loss": 1.0398,
+      "step": 287490
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.205441187724803e-05,
+      "loss": 0.9707,
+      "step": 287500
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.205326394734828e-05,
+      "loss": 0.9386,
+      "step": 287510
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.2052116017448534e-05,
+      "loss": 1.0627,
+      "step": 287520
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.2050968087548784e-05,
+      "loss": 1.0037,
+      "step": 287530
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.204982015764903e-05,
+      "loss": 1.0162,
+      "step": 287540
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.204867222774929e-05,
+      "loss": 1.0141,
+      "step": 287550
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.204752429784955e-05,
+      "loss": 0.9813,
+      "step": 287560
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.204637636794979e-05,
+      "loss": 0.8434,
+      "step": 287570
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.2045228438050046e-05,
+      "loss": 1.0074,
+      "step": 287580
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.20440805081503e-05,
+      "loss": 0.946,
+      "step": 287590
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.2042932578250546e-05,
+      "loss": 0.9843,
+      "step": 287600
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.20417846483508e-05,
+      "loss": 0.9611,
+      "step": 287610
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.204063671845106e-05,
+      "loss": 0.9959,
+      "step": 287620
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.203948878855131e-05,
+      "loss": 0.9831,
+      "step": 287630
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.203834085865156e-05,
+      "loss": 1.0088,
+      "step": 287640
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.2037192928751816e-05,
+      "loss": 1.0049,
+      "step": 287650
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.2036044998852066e-05,
+      "loss": 0.8993,
+      "step": 287660
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.203489706895232e-05,
+      "loss": 0.9306,
+      "step": 287670
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.203374913905257e-05,
+      "loss": 0.8938,
+      "step": 287680
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.203260120915282e-05,
+      "loss": 0.963,
+      "step": 287690
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.203145327925308e-05,
+      "loss": 0.9325,
+      "step": 287700
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.2030305349353335e-05,
+      "loss": 0.9127,
+      "step": 287710
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.202915741945358e-05,
+      "loss": 1.0277,
+      "step": 287720
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.2028009489553835e-05,
+      "loss": 1.0196,
+      "step": 287730
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.202686155965409e-05,
+      "loss": 0.9087,
+      "step": 287740
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.2025713629754335e-05,
+      "loss": 0.8828,
+      "step": 287750
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.202456569985459e-05,
+      "loss": 1.0061,
+      "step": 287760
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.202341776995485e-05,
+      "loss": 0.9151,
+      "step": 287770
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.20222698400551e-05,
+      "loss": 0.9404,
+      "step": 287780
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.202112191015535e-05,
+      "loss": 0.9563,
+      "step": 287790
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.2019973980255604e-05,
+      "loss": 0.9581,
+      "step": 287800
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.2018826050355854e-05,
+      "loss": 0.9305,
+      "step": 287810
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.201767812045611e-05,
+      "loss": 0.864,
+      "step": 287820
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.201653019055636e-05,
+      "loss": 0.9147,
+      "step": 287830
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.201538226065661e-05,
+      "loss": 0.985,
+      "step": 287840
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.201423433075687e-05,
+      "loss": 0.9568,
+      "step": 287850
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.2013086400857124e-05,
+      "loss": 1.0584,
+      "step": 287860
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.201193847095737e-05,
+      "loss": 1.0259,
+      "step": 287870
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.201079054105762e-05,
+      "loss": 0.8968,
+      "step": 287880
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.200964261115788e-05,
+      "loss": 1.0098,
+      "step": 287890
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.200849468125812e-05,
+      "loss": 0.9671,
+      "step": 287900
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.200734675135838e-05,
+      "loss": 0.9473,
+      "step": 287910
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.2006198821458636e-05,
+      "loss": 0.9754,
+      "step": 287920
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.2005050891558886e-05,
+      "loss": 0.9105,
+      "step": 287930
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.2003902961659136e-05,
+      "loss": 0.9493,
+      "step": 287940
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.200275503175939e-05,
+      "loss": 0.9734,
+      "step": 287950
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.200160710185964e-05,
+      "loss": 0.8935,
+      "step": 287960
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.20004591719599e-05,
+      "loss": 0.9974,
+      "step": 287970
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.199931124206015e-05,
+      "loss": 0.924,
+      "step": 287980
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.19981633121604e-05,
+      "loss": 0.8553,
+      "step": 287990
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.1997015382260655e-05,
+      "loss": 0.9627,
+      "step": 288000
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.199586745236091e-05,
+      "loss": 0.9721,
+      "step": 288010
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.1994719522461155e-05,
+      "loss": 1.0164,
+      "step": 288020
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.199357159256141e-05,
+      "loss": 1.1159,
+      "step": 288030
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.199242366266167e-05,
+      "loss": 0.9956,
+      "step": 288040
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.199127573276191e-05,
+      "loss": 1.0098,
+      "step": 288050
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.199012780286217e-05,
+      "loss": 0.9145,
+      "step": 288060
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.198897987296242e-05,
+      "loss": 0.9764,
+      "step": 288070
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.1987831943062675e-05,
+      "loss": 1.0291,
+      "step": 288080
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.1986684013162924e-05,
+      "loss": 0.9411,
+      "step": 288090
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.1985536083263174e-05,
+      "loss": 1.0153,
+      "step": 288100
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.198438815336343e-05,
+      "loss": 1.0072,
+      "step": 288110
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.198324022346369e-05,
+      "loss": 0.8989,
+      "step": 288120
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.198209229356393e-05,
+      "loss": 0.9464,
+      "step": 288130
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.198094436366419e-05,
+      "loss": 0.9158,
+      "step": 288140
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.1979796433764444e-05,
+      "loss": 1.0403,
+      "step": 288150
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.197864850386469e-05,
+      "loss": 0.9673,
+      "step": 288160
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.1977500573964944e-05,
+      "loss": 1.0797,
+      "step": 288170
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.19763526440652e-05,
+      "loss": 1.0134,
+      "step": 288180
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.197520471416545e-05,
+      "loss": 1.0019,
+      "step": 288190
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.19740567842657e-05,
+      "loss": 0.9879,
+      "step": 288200
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.1972908854365957e-05,
+      "loss": 0.9439,
+      "step": 288210
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.1971760924466206e-05,
+      "loss": 0.9725,
+      "step": 288220
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.197061299456646e-05,
+      "loss": 0.9633,
+      "step": 288230
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.196946506466671e-05,
+      "loss": 1.0602,
+      "step": 288240
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.196831713476696e-05,
+      "loss": 0.9736,
+      "step": 288250
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.196716920486722e-05,
+      "loss": 1.0662,
+      "step": 288260
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.1966021274967476e-05,
+      "loss": 0.9198,
+      "step": 288270
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.196487334506772e-05,
+      "loss": 1.0457,
+      "step": 288280
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.1963725415167976e-05,
+      "loss": 0.922,
+      "step": 288290
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.196257748526823e-05,
+      "loss": 0.9593,
+      "step": 288300
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.1961429555368475e-05,
+      "loss": 0.9141,
+      "step": 288310
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.196028162546873e-05,
+      "loss": 0.984,
+      "step": 288320
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.195913369556899e-05,
+      "loss": 0.9318,
+      "step": 288330
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.195798576566924e-05,
+      "loss": 1.0225,
+      "step": 288340
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.195683783576949e-05,
+      "loss": 1.0753,
+      "step": 288350
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.1955689905869745e-05,
+      "loss": 0.9659,
+      "step": 288360
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.1954541975969995e-05,
+      "loss": 0.9739,
+      "step": 288370
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.195339404607025e-05,
+      "loss": 0.8879,
+      "step": 288380
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.19522461161705e-05,
+      "loss": 0.9982,
+      "step": 288390
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.195109818627075e-05,
+      "loss": 1.0421,
+      "step": 288400
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.194995025637101e-05,
+      "loss": 0.9823,
+      "step": 288410
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.1948802326471265e-05,
+      "loss": 0.981,
+      "step": 288420
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.194765439657151e-05,
+      "loss": 0.9614,
+      "step": 288430
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.1946506466671764e-05,
+      "loss": 1.0455,
+      "step": 288440
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.194535853677202e-05,
+      "loss": 0.9598,
+      "step": 288450
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.1944210606872264e-05,
+      "loss": 1.0401,
+      "step": 288460
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.194306267697252e-05,
+      "loss": 0.996,
+      "step": 288470
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.194191474707278e-05,
+      "loss": 1.0874,
+      "step": 288480
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.194076681717303e-05,
+      "loss": 0.9655,
+      "step": 288490
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.193961888727328e-05,
+      "loss": 0.9493,
+      "step": 288500
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.1938470957373534e-05,
+      "loss": 0.919,
+      "step": 288510
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.193732302747378e-05,
+      "loss": 1.0458,
+      "step": 288520
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.193617509757404e-05,
+      "loss": 0.8932,
+      "step": 288530
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.193502716767429e-05,
+      "loss": 1.0675,
+      "step": 288540
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.193387923777454e-05,
+      "loss": 0.9428,
+      "step": 288550
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.1932731307874796e-05,
+      "loss": 0.8606,
+      "step": 288560
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.193158337797505e-05,
+      "loss": 0.9837,
+      "step": 288570
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.1930435448075296e-05,
+      "loss": 0.9824,
+      "step": 288580
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.192928751817555e-05,
+      "loss": 1.029,
+      "step": 288590
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.192813958827581e-05,
+      "loss": 1.0403,
+      "step": 288600
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.192699165837605e-05,
+      "loss": 1.0356,
+      "step": 288610
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.192584372847631e-05,
+      "loss": 1.0216,
+      "step": 288620
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.1924695798576566e-05,
+      "loss": 0.9538,
+      "step": 288630
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.1923547868676816e-05,
+      "loss": 0.939,
+      "step": 288640
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.1922399938777065e-05,
+      "loss": 0.9475,
+      "step": 288650
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.192125200887732e-05,
+      "loss": 0.9752,
+      "step": 288660
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.192010407897757e-05,
+      "loss": 0.9352,
+      "step": 288670
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.191895614907783e-05,
+      "loss": 1.0044,
+      "step": 288680
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.191780821917808e-05,
+      "loss": 0.9074,
+      "step": 288690
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.191666028927833e-05,
+      "loss": 0.9351,
+      "step": 288700
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.1915512359378585e-05,
+      "loss": 0.9869,
+      "step": 288710
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.191436442947884e-05,
+      "loss": 0.9729,
+      "step": 288720
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.1913216499579085e-05,
+      "loss": 1.052,
+      "step": 288730
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.191206856967934e-05,
+      "loss": 0.9945,
+      "step": 288740
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.19109206397796e-05,
+      "loss": 0.9511,
+      "step": 288750
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.190977270987984e-05,
+      "loss": 0.8692,
+      "step": 288760
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.19086247799801e-05,
+      "loss": 0.9334,
+      "step": 288770
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.1907476850080354e-05,
+      "loss": 0.8633,
+      "step": 288780
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.1906328920180604e-05,
+      "loss": 0.9105,
+      "step": 288790
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.1905180990280854e-05,
+      "loss": 1.0137,
+      "step": 288800
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.190403306038111e-05,
+      "loss": 0.9664,
+      "step": 288810
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.190288513048136e-05,
+      "loss": 0.9405,
+      "step": 288820
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.190173720058162e-05,
+      "loss": 0.9655,
+      "step": 288830
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.190058927068187e-05,
+      "loss": 1.1087,
+      "step": 288840
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.189944134078212e-05,
+      "loss": 1.1017,
+      "step": 288850
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.189829341088237e-05,
+      "loss": 1.0034,
+      "step": 288860
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.189714548098263e-05,
+      "loss": 0.937,
+      "step": 288870
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.189599755108287e-05,
+      "loss": 0.8704,
+      "step": 288880
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.189484962118313e-05,
+      "loss": 0.9445,
+      "step": 288890
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.1893701691283386e-05,
+      "loss": 0.9928,
+      "step": 288900
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.189255376138363e-05,
+      "loss": 0.9817,
+      "step": 288910
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.1891405831483886e-05,
+      "loss": 1.0028,
+      "step": 288920
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.189025790158414e-05,
+      "loss": 0.9399,
+      "step": 288930
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.188910997168439e-05,
+      "loss": 1.0268,
+      "step": 288940
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.188796204178464e-05,
+      "loss": 0.9578,
+      "step": 288950
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.18868141118849e-05,
+      "loss": 0.9742,
+      "step": 288960
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.188566618198515e-05,
+      "loss": 0.9665,
+      "step": 288970
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.1884518252085405e-05,
+      "loss": 0.9883,
+      "step": 288980
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.1883370322185655e-05,
+      "loss": 1.0647,
+      "step": 288990
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.1882222392285905e-05,
+      "loss": 0.9359,
+      "step": 289000
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.188107446238616e-05,
+      "loss": 0.9371,
+      "step": 289010
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.187992653248642e-05,
+      "loss": 0.9781,
+      "step": 289020
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.187877860258666e-05,
+      "loss": 1.0189,
+      "step": 289030
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.187763067268692e-05,
+      "loss": 0.8846,
+      "step": 289040
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.1876482742787175e-05,
+      "loss": 1.0242,
+      "step": 289050
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.187533481288742e-05,
+      "loss": 0.9159,
+      "step": 289060
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.1874186882987674e-05,
+      "loss": 0.9953,
+      "step": 289070
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.187303895308793e-05,
+      "loss": 0.9541,
+      "step": 289080
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.187189102318818e-05,
+      "loss": 0.9616,
+      "step": 289090
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.187074309328843e-05,
+      "loss": 0.9667,
+      "step": 289100
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.186959516338869e-05,
+      "loss": 1.1006,
+      "step": 289110
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.186844723348894e-05,
+      "loss": 0.9239,
+      "step": 289120
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.1867299303589194e-05,
+      "loss": 1.093,
+      "step": 289130
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.1866151373689444e-05,
+      "loss": 0.9345,
+      "step": 289140
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.1865003443789694e-05,
+      "loss": 0.9288,
+      "step": 289150
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.186385551388995e-05,
+      "loss": 1.0126,
+      "step": 289160
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.186270758399021e-05,
+      "loss": 0.9816,
+      "step": 289170
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.186155965409045e-05,
+      "loss": 1.0076,
+      "step": 289180
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.1860411724190707e-05,
+      "loss": 0.9517,
+      "step": 289190
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.185926379429096e-05,
+      "loss": 0.9325,
+      "step": 289200
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.1858115864391206e-05,
+      "loss": 0.9388,
+      "step": 289210
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.185696793449146e-05,
+      "loss": 0.9425,
+      "step": 289220
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.185582000459172e-05,
+      "loss": 0.9219,
+      "step": 289230
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.185467207469197e-05,
+      "loss": 1.0449,
+      "step": 289240
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.185352414479222e-05,
+      "loss": 0.9004,
+      "step": 289250
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.1852376214892476e-05,
+      "loss": 0.8548,
+      "step": 289260
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.1851228284992726e-05,
+      "loss": 0.9773,
+      "step": 289270
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.185008035509298e-05,
+      "loss": 1.033,
+      "step": 289280
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.184893242519323e-05,
+      "loss": 1.0891,
+      "step": 289290
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.184778449529348e-05,
+      "loss": 1.0046,
+      "step": 289300
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.184663656539374e-05,
+      "loss": 0.9063,
+      "step": 289310
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.1845488635493995e-05,
+      "loss": 0.9704,
+      "step": 289320
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.184434070559424e-05,
+      "loss": 1.0091,
+      "step": 289330
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.1843192775694495e-05,
+      "loss": 0.9746,
+      "step": 289340
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.184204484579475e-05,
+      "loss": 0.9048,
+      "step": 289350
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.1840896915894995e-05,
+      "loss": 0.9452,
+      "step": 289360
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.183974898599525e-05,
+      "loss": 0.9828,
+      "step": 289370
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.183860105609551e-05,
+      "loss": 1.0003,
+      "step": 289380
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.183745312619576e-05,
+      "loss": 1.0574,
+      "step": 289390
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.183630519629601e-05,
+      "loss": 0.9657,
+      "step": 289400
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.1835157266396264e-05,
+      "loss": 1.2069,
+      "step": 289410
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.1834009336496514e-05,
+      "loss": 0.9846,
+      "step": 289420
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.183286140659677e-05,
+      "loss": 0.8743,
+      "step": 289430
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.1831713476697014e-05,
+      "loss": 0.9988,
+      "step": 289440
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.183056554679727e-05,
+      "loss": 1.0366,
+      "step": 289450
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.182941761689753e-05,
+      "loss": 1.0453,
+      "step": 289460
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.182826968699777e-05,
+      "loss": 0.9412,
+      "step": 289470
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.182712175709803e-05,
+      "loss": 0.9569,
+      "step": 289480
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.1825973827198284e-05,
+      "loss": 1.0308,
+      "step": 289490
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.182482589729853e-05,
+      "loss": 0.9309,
+      "step": 289500
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.182367796739878e-05,
+      "loss": 0.9374,
+      "step": 289510
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.182253003749904e-05,
+      "loss": 0.9587,
+      "step": 289520
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.182138210759929e-05,
+      "loss": 1.0494,
+      "step": 289530
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.1820234177699546e-05,
+      "loss": 1.0397,
+      "step": 289540
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.1819086247799796e-05,
+      "loss": 0.9925,
+      "step": 289550
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.1817938317900046e-05,
+      "loss": 1.0479,
+      "step": 289560
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.18167903880003e-05,
+      "loss": 0.9628,
+      "step": 289570
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.181564245810056e-05,
+      "loss": 0.928,
+      "step": 289580
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.18144945282008e-05,
+      "loss": 1.0127,
+      "step": 289590
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.181334659830106e-05,
+      "loss": 0.9948,
+      "step": 289600
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.1812198668401316e-05,
+      "loss": 1.025,
+      "step": 289610
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.181105073850156e-05,
+      "loss": 0.9719,
+      "step": 289620
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.1809902808601815e-05,
+      "loss": 0.9032,
+      "step": 289630
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.180875487870207e-05,
+      "loss": 1.0262,
+      "step": 289640
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.180760694880232e-05,
+      "loss": 0.9786,
+      "step": 289650
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.180645901890257e-05,
+      "loss": 0.9489,
+      "step": 289660
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.180531108900283e-05,
+      "loss": 0.9284,
+      "step": 289670
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.180416315910308e-05,
+      "loss": 0.9347,
+      "step": 289680
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.1803015229203335e-05,
+      "loss": 0.9517,
+      "step": 289690
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.1801867299303585e-05,
+      "loss": 0.9808,
+      "step": 289700
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.1800719369403835e-05,
+      "loss": 1.0427,
+      "step": 289710
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.179957143950409e-05,
+      "loss": 0.8206,
+      "step": 289720
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.179842350960435e-05,
+      "loss": 0.95,
+      "step": 289730
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.179727557970459e-05,
+      "loss": 1.0051,
+      "step": 289740
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.179612764980485e-05,
+      "loss": 0.9364,
+      "step": 289750
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.1794979719905104e-05,
+      "loss": 0.9556,
+      "step": 289760
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.179383179000535e-05,
+      "loss": 0.9114,
+      "step": 289770
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.1792683860105604e-05,
+      "loss": 0.8658,
+      "step": 289780
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.179153593020586e-05,
+      "loss": 0.8275,
+      "step": 289790
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.179038800030611e-05,
+      "loss": 0.9519,
+      "step": 289800
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.178924007040636e-05,
+      "loss": 1.0088,
+      "step": 289810
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.178809214050662e-05,
+      "loss": 0.8848,
+      "step": 289820
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.178694421060687e-05,
+      "loss": 0.999,
+      "step": 289830
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.178579628070712e-05,
+      "loss": 0.9501,
+      "step": 289840
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.178464835080737e-05,
+      "loss": 0.8459,
+      "step": 289850
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.178350042090762e-05,
+      "loss": 0.9734,
+      "step": 289860
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.178235249100788e-05,
+      "loss": 0.8566,
+      "step": 289870
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.1781204561108136e-05,
+      "loss": 1.0377,
+      "step": 289880
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.178005663120838e-05,
+      "loss": 0.9665,
+      "step": 289890
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.1778908701308636e-05,
+      "loss": 1.0351,
+      "step": 289900
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.177776077140889e-05,
+      "loss": 0.9157,
+      "step": 289910
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.1776612841509136e-05,
+      "loss": 0.9864,
+      "step": 289920
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.177546491160939e-05,
+      "loss": 0.9516,
+      "step": 289930
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.177431698170965e-05,
+      "loss": 0.9019,
+      "step": 289940
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.17731690518099e-05,
+      "loss": 0.9552,
+      "step": 289950
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.177202112191015e-05,
+      "loss": 1.0008,
+      "step": 289960
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.1770873192010405e-05,
+      "loss": 0.8876,
+      "step": 289970
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.1769725262110655e-05,
+      "loss": 1.1422,
+      "step": 289980
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.176857733221091e-05,
+      "loss": 0.9317,
+      "step": 289990
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.176742940231116e-05,
+      "loss": 0.8998,
+      "step": 290000
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.176628147241141e-05,
+      "loss": 0.9986,
+      "step": 290010
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.176513354251167e-05,
+      "loss": 0.988,
+      "step": 290020
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.1763985612611925e-05,
+      "loss": 1.0681,
+      "step": 290030
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.176283768271217e-05,
+      "loss": 0.8848,
+      "step": 290040
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.1761689752812424e-05,
+      "loss": 0.9672,
+      "step": 290050
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.176054182291268e-05,
+      "loss": 0.973,
+      "step": 290060
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.1759393893012924e-05,
+      "loss": 0.8874,
+      "step": 290070
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.175824596311318e-05,
+      "loss": 1.0268,
+      "step": 290080
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.175709803321344e-05,
+      "loss": 1.0005,
+      "step": 290090
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.175595010331369e-05,
+      "loss": 1.0289,
+      "step": 290100
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.175480217341394e-05,
+      "loss": 1.0291,
+      "step": 290110
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.1753654243514194e-05,
+      "loss": 0.8989,
+      "step": 290120
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.1752506313614444e-05,
+      "loss": 0.9647,
+      "step": 290130
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.17513583837147e-05,
+      "loss": 0.9579,
+      "step": 290140
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.175021045381495e-05,
+      "loss": 0.9822,
+      "step": 290150
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.17490625239152e-05,
+      "loss": 0.9421,
+      "step": 290160
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.1747914594015457e-05,
+      "loss": 1.0329,
+      "step": 290170
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.174676666411571e-05,
+      "loss": 0.8711,
+      "step": 290180
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.1745618734215956e-05,
+      "loss": 0.9275,
+      "step": 290190
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.174447080431621e-05,
+      "loss": 0.9717,
+      "step": 290200
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.174332287441647e-05,
+      "loss": 0.903,
+      "step": 290210
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.174217494451671e-05,
+      "loss": 1.0894,
+      "step": 290220
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.174102701461697e-05,
+      "loss": 0.8974,
+      "step": 290230
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.1739879084717226e-05,
+      "loss": 1.0399,
+      "step": 290240
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.1738731154817476e-05,
+      "loss": 0.9419,
+      "step": 290250
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.1737583224917726e-05,
+      "loss": 0.9153,
+      "step": 290260
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.173643529501798e-05,
+      "loss": 0.9198,
+      "step": 290270
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.173528736511823e-05,
+      "loss": 1.0332,
+      "step": 290280
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.173413943521849e-05,
+      "loss": 1.0505,
+      "step": 290290
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.173299150531874e-05,
+      "loss": 0.9263,
+      "step": 290300
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.173184357541899e-05,
+      "loss": 0.9781,
+      "step": 290310
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.1730695645519245e-05,
+      "loss": 1.0356,
+      "step": 290320
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.17295477156195e-05,
+      "loss": 0.9941,
+      "step": 290330
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.1728399785719745e-05,
+      "loss": 0.9456,
+      "step": 290340
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.172725185582e-05,
+      "loss": 0.9488,
+      "step": 290350
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.172610392592026e-05,
+      "loss": 0.9393,
+      "step": 290360
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.17249559960205e-05,
+      "loss": 0.9718,
+      "step": 290370
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.172380806612076e-05,
+      "loss": 1.0109,
+      "step": 290380
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.1722660136221014e-05,
+      "loss": 0.9834,
+      "step": 290390
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.1721512206321264e-05,
+      "loss": 1.0448,
+      "step": 290400
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.1720364276421514e-05,
+      "loss": 1.0008,
+      "step": 290410
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.171921634652177e-05,
+      "loss": 0.9555,
+      "step": 290420
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.171806841662202e-05,
+      "loss": 0.9994,
+      "step": 290430
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.171692048672228e-05,
+      "loss": 1.0027,
+      "step": 290440
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.171577255682253e-05,
+      "loss": 0.9927,
+      "step": 290450
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.171462462692278e-05,
+      "loss": 0.996,
+      "step": 290460
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.1713476697023034e-05,
+      "loss": 1.0221,
+      "step": 290470
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.171232876712329e-05,
+      "loss": 0.9147,
+      "step": 290480
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.171118083722353e-05,
+      "loss": 0.9847,
+      "step": 290490
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.171003290732379e-05,
+      "loss": 0.9272,
+      "step": 290500
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.1708884977424046e-05,
+      "loss": 0.9154,
+      "step": 290510
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.170773704752429e-05,
+      "loss": 0.9407,
+      "step": 290520
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.1706589117624546e-05,
+      "loss": 0.9092,
+      "step": 290530
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.17054411877248e-05,
+      "loss": 0.9373,
+      "step": 290540
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.170429325782505e-05,
+      "loss": 1.0312,
+      "step": 290550
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.17031453279253e-05,
+      "loss": 0.9282,
+      "step": 290560
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.170199739802556e-05,
+      "loss": 1.0443,
+      "step": 290570
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.170084946812581e-05,
+      "loss": 0.9616,
+      "step": 290580
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.1699701538226066e-05,
+      "loss": 0.964,
+      "step": 290590
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.1698553608326316e-05,
+      "loss": 0.9756,
+      "step": 290600
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.1697405678426565e-05,
+      "loss": 0.9631,
+      "step": 290610
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.169625774852682e-05,
+      "loss": 1.016,
+      "step": 290620
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.169510981862708e-05,
+      "loss": 0.9307,
+      "step": 290630
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.169396188872732e-05,
+      "loss": 1.0107,
+      "step": 290640
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.169281395882758e-05,
+      "loss": 1.0242,
+      "step": 290650
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.1691666028927835e-05,
+      "loss": 1.0019,
+      "step": 290660
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.169051809902808e-05,
+      "loss": 0.977,
+      "step": 290670
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.1689370169128335e-05,
+      "loss": 0.9792,
+      "step": 290680
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.168822223922859e-05,
+      "loss": 0.9506,
+      "step": 290690
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.168707430932884e-05,
+      "loss": 1.0352,
+      "step": 290700
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.168592637942909e-05,
+      "loss": 0.9622,
+      "step": 290710
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.168477844952935e-05,
+      "loss": 1.0161,
+      "step": 290720
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.16836305196296e-05,
+      "loss": 0.9539,
+      "step": 290730
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.1682482589729854e-05,
+      "loss": 0.9541,
+      "step": 290740
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.1681334659830104e-05,
+      "loss": 0.9256,
+      "step": 290750
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.1680186729930354e-05,
+      "loss": 0.9256,
+      "step": 290760
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.167903880003061e-05,
+      "loss": 0.9711,
+      "step": 290770
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.167789087013087e-05,
+      "loss": 1.0471,
+      "step": 290780
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.167674294023111e-05,
+      "loss": 1.0139,
+      "step": 290790
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.167559501033137e-05,
+      "loss": 0.8216,
+      "step": 290800
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.167444708043162e-05,
+      "loss": 0.9739,
+      "step": 290810
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.1673299150531867e-05,
+      "loss": 1.0677,
+      "step": 290820
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.167215122063212e-05,
+      "loss": 0.9649,
+      "step": 290830
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.167100329073237e-05,
+      "loss": 1.0137,
+      "step": 290840
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.166985536083263e-05,
+      "loss": 0.9912,
+      "step": 290850
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.166870743093288e-05,
+      "loss": 0.9439,
+      "step": 290860
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.166755950103313e-05,
+      "loss": 0.9682,
+      "step": 290870
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.1666411571133386e-05,
+      "loss": 0.9245,
+      "step": 290880
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.166526364123364e-05,
+      "loss": 0.9248,
+      "step": 290890
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.1664115711333886e-05,
+      "loss": 1.0504,
+      "step": 290900
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.166296778143414e-05,
+      "loss": 1.0086,
+      "step": 290910
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.16618198515344e-05,
+      "loss": 1.0073,
+      "step": 290920
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.166067192163464e-05,
+      "loss": 0.9433,
+      "step": 290930
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.16595239917349e-05,
+      "loss": 0.9449,
+      "step": 290940
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.1658376061835155e-05,
+      "loss": 0.8626,
+      "step": 290950
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.1657228131935405e-05,
+      "loss": 1.0795,
+      "step": 290960
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.1656080202035655e-05,
+      "loss": 0.9338,
+      "step": 290970
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.165493227213591e-05,
+      "loss": 0.9168,
+      "step": 290980
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.165378434223616e-05,
+      "loss": 0.9888,
+      "step": 290990
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.165263641233642e-05,
+      "loss": 1.0636,
+      "step": 291000
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.165148848243667e-05,
+      "loss": 0.9682,
+      "step": 291010
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.165034055253692e-05,
+      "loss": 0.9396,
+      "step": 291020
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.1649192622637174e-05,
+      "loss": 1.0242,
+      "step": 291030
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.164804469273743e-05,
+      "loss": 0.9167,
+      "step": 291040
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.1646896762837674e-05,
+      "loss": 0.924,
+      "step": 291050
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.164574883293793e-05,
+      "loss": 0.9993,
+      "step": 291060
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.164460090303819e-05,
+      "loss": 0.8999,
+      "step": 291070
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.164345297313843e-05,
+      "loss": 0.892,
+      "step": 291080
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.164230504323869e-05,
+      "loss": 0.9293,
+      "step": 291090
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.1641157113338944e-05,
+      "loss": 1.0586,
+      "step": 291100
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.1640009183439194e-05,
+      "loss": 0.9616,
+      "step": 291110
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.1638861253539443e-05,
+      "loss": 1.0409,
+      "step": 291120
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.16377133236397e-05,
+      "loss": 1.0199,
+      "step": 291130
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.163656539373995e-05,
+      "loss": 0.8753,
+      "step": 291140
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.1635417463840207e-05,
+      "loss": 0.8985,
+      "step": 291150
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.1634269533940456e-05,
+      "loss": 1.1178,
+      "step": 291160
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.1633121604040706e-05,
+      "loss": 1.0428,
+      "step": 291170
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.163197367414096e-05,
+      "loss": 1.0183,
+      "step": 291180
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.163082574424122e-05,
+      "loss": 0.9646,
+      "step": 291190
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.162967781434146e-05,
+      "loss": 0.9018,
+      "step": 291200
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.162852988444172e-05,
+      "loss": 0.922,
+      "step": 291210
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.1627381954541976e-05,
+      "loss": 1.0446,
+      "step": 291220
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.162623402464222e-05,
+      "loss": 1.0942,
+      "step": 291230
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.1625086094742476e-05,
+      "loss": 0.9986,
+      "step": 291240
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.162393816484273e-05,
+      "loss": 1.0422,
+      "step": 291250
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.162279023494298e-05,
+      "loss": 0.9515,
+      "step": 291260
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.162164230504323e-05,
+      "loss": 1.008,
+      "step": 291270
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.162049437514349e-05,
+      "loss": 1.0189,
+      "step": 291280
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.161934644524374e-05,
+      "loss": 0.9582,
+      "step": 291290
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.1618198515343995e-05,
+      "loss": 0.9325,
+      "step": 291300
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.1617050585444245e-05,
+      "loss": 0.9752,
+      "step": 291310
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.1615902655544495e-05,
+      "loss": 0.9896,
+      "step": 291320
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.161475472564475e-05,
+      "loss": 0.9043,
+      "step": 291330
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.161360679574501e-05,
+      "loss": 1.0603,
+      "step": 291340
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.161245886584525e-05,
+      "loss": 1.0079,
+      "step": 291350
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.161131093594551e-05,
+      "loss": 0.9171,
+      "step": 291360
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.1610163006045764e-05,
+      "loss": 0.9219,
+      "step": 291370
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.160901507614601e-05,
+      "loss": 1.0141,
+      "step": 291380
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.1607867146246264e-05,
+      "loss": 0.9138,
+      "step": 291390
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.160671921634652e-05,
+      "loss": 0.9562,
+      "step": 291400
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.160557128644677e-05,
+      "loss": 0.9366,
+      "step": 291410
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.160442335654702e-05,
+      "loss": 1.0511,
+      "step": 291420
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.160327542664728e-05,
+      "loss": 0.974,
+      "step": 291430
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.160212749674753e-05,
+      "loss": 0.9959,
+      "step": 291440
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.1600979566847784e-05,
+      "loss": 0.9085,
+      "step": 291450
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.159983163694803e-05,
+      "loss": 0.9662,
+      "step": 291460
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.159868370704828e-05,
+      "loss": 0.993,
+      "step": 291470
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.159753577714854e-05,
+      "loss": 1.028,
+      "step": 291480
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.1596387847248796e-05,
+      "loss": 0.9848,
+      "step": 291490
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.159523991734904e-05,
+      "loss": 0.9189,
+      "step": 291500
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.1594091987449296e-05,
+      "loss": 1.0286,
+      "step": 291510
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.159294405754955e-05,
+      "loss": 0.9778,
+      "step": 291520
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.1591796127649796e-05,
+      "loss": 1.0354,
+      "step": 291530
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.159064819775005e-05,
+      "loss": 0.8758,
+      "step": 291540
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.158950026785031e-05,
+      "loss": 1.0519,
+      "step": 291550
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.158835233795056e-05,
+      "loss": 0.9147,
+      "step": 291560
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.158720440805081e-05,
+      "loss": 1.0591,
+      "step": 291570
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.1586056478151065e-05,
+      "loss": 0.939,
+      "step": 291580
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.1584908548251315e-05,
+      "loss": 0.9445,
+      "step": 291590
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.158376061835157e-05,
+      "loss": 0.9433,
+      "step": 291600
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.158261268845182e-05,
+      "loss": 1.0297,
+      "step": 291610
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.158146475855207e-05,
+      "loss": 0.9923,
+      "step": 291620
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.158031682865233e-05,
+      "loss": 0.9178,
+      "step": 291630
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.1579168898752585e-05,
+      "loss": 1.0098,
+      "step": 291640
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.157802096885283e-05,
+      "loss": 0.9315,
+      "step": 291650
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.1576873038953085e-05,
+      "loss": 0.9534,
+      "step": 291660
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.157572510905334e-05,
+      "loss": 1.0263,
+      "step": 291670
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.1574577179153584e-05,
+      "loss": 0.9423,
+      "step": 291680
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.157342924925384e-05,
+      "loss": 0.9888,
+      "step": 291690
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.15722813193541e-05,
+      "loss": 1.0442,
+      "step": 291700
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.157113338945435e-05,
+      "loss": 1.0138,
+      "step": 291710
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.15699854595546e-05,
+      "loss": 1.0372,
+      "step": 291720
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.1568837529654854e-05,
+      "loss": 0.9562,
+      "step": 291730
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.1567689599755104e-05,
+      "loss": 1.1176,
+      "step": 291740
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.156654166985536e-05,
+      "loss": 0.9521,
+      "step": 291750
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.156539373995561e-05,
+      "loss": 0.9949,
+      "step": 291760
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.156424581005586e-05,
+      "loss": 0.9435,
+      "step": 291770
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.156309788015612e-05,
+      "loss": 1.0139,
+      "step": 291780
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.1561949950256373e-05,
+      "loss": 0.9885,
+      "step": 291790
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.1560802020356616e-05,
+      "loss": 1.0605,
+      "step": 291800
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.155965409045687e-05,
+      "loss": 0.9107,
+      "step": 291810
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.155850616055713e-05,
+      "loss": 0.9832,
+      "step": 291820
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.155735823065737e-05,
+      "loss": 0.9865,
+      "step": 291830
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.155621030075763e-05,
+      "loss": 0.8924,
+      "step": 291840
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.1555062370857886e-05,
+      "loss": 0.912,
+      "step": 291850
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.1553914440958136e-05,
+      "loss": 1.0561,
+      "step": 291860
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.1552766511058386e-05,
+      "loss": 0.9261,
+      "step": 291870
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.155161858115864e-05,
+      "loss": 0.9621,
+      "step": 291880
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.155047065125889e-05,
+      "loss": 0.9261,
+      "step": 291890
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.154932272135915e-05,
+      "loss": 0.9555,
+      "step": 291900
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.15481747914594e-05,
+      "loss": 1.0594,
+      "step": 291910
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.154702686155965e-05,
+      "loss": 1.0604,
+      "step": 291920
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.1545878931659905e-05,
+      "loss": 0.939,
+      "step": 291930
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.154473100176016e-05,
+      "loss": 1.0927,
+      "step": 291940
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.1543583071860405e-05,
+      "loss": 0.9455,
+      "step": 291950
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.154243514196066e-05,
+      "loss": 0.9814,
+      "step": 291960
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.154128721206092e-05,
+      "loss": 0.8631,
+      "step": 291970
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.154013928216116e-05,
+      "loss": 0.9668,
+      "step": 291980
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.153899135226142e-05,
+      "loss": 0.9618,
+      "step": 291990
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.1537843422361675e-05,
+      "loss": 0.9893,
+      "step": 292000
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.1536695492461924e-05,
+      "loss": 0.9272,
+      "step": 292010
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.1535547562562174e-05,
+      "loss": 1.0808,
+      "step": 292020
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.153439963266243e-05,
+      "loss": 0.947,
+      "step": 292030
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.153325170276268e-05,
+      "loss": 0.8582,
+      "step": 292040
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.153210377286294e-05,
+      "loss": 1.006,
+      "step": 292050
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.153095584296319e-05,
+      "loss": 0.9893,
+      "step": 292060
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.152980791306344e-05,
+      "loss": 0.9876,
+      "step": 292070
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.1528659983163694e-05,
+      "loss": 1.0541,
+      "step": 292080
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.152751205326395e-05,
+      "loss": 0.9685,
+      "step": 292090
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.1526364123364193e-05,
+      "loss": 1.04,
+      "step": 292100
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.152521619346445e-05,
+      "loss": 0.9927,
+      "step": 292110
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.152406826356471e-05,
+      "loss": 1.0042,
+      "step": 292120
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.152292033366495e-05,
+      "loss": 0.9422,
+      "step": 292130
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.1521772403765206e-05,
+      "loss": 1.0527,
+      "step": 292140
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.152062447386546e-05,
+      "loss": 0.982,
+      "step": 292150
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.151947654396571e-05,
+      "loss": 0.8752,
+      "step": 292160
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.151832861406596e-05,
+      "loss": 1.0251,
+      "step": 292170
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.151718068416621e-05,
+      "loss": 0.8844,
+      "step": 292180
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.151603275426647e-05,
+      "loss": 0.9948,
+      "step": 292190
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.1514884824366726e-05,
+      "loss": 0.9331,
+      "step": 292200
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.151373689446697e-05,
+      "loss": 1.0635,
+      "step": 292210
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.1512588964567226e-05,
+      "loss": 1.0181,
+      "step": 292220
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.151144103466748e-05,
+      "loss": 0.9263,
+      "step": 292230
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.1510293104767725e-05,
+      "loss": 0.9856,
+      "step": 292240
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.150914517486798e-05,
+      "loss": 0.9962,
+      "step": 292250
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.150799724496824e-05,
+      "loss": 1.1033,
+      "step": 292260
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.150684931506849e-05,
+      "loss": 0.8958,
+      "step": 292270
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.150570138516874e-05,
+      "loss": 1.0599,
+      "step": 292280
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.1504553455268995e-05,
+      "loss": 1.1289,
+      "step": 292290
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.1503405525369245e-05,
+      "loss": 1.0463,
+      "step": 292300
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.15022575954695e-05,
+      "loss": 0.9694,
+      "step": 292310
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.150110966556975e-05,
+      "loss": 0.9902,
+      "step": 292320
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.149996173567e-05,
+      "loss": 0.9306,
+      "step": 292330
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.149881380577026e-05,
+      "loss": 0.9848,
+      "step": 292340
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.1497665875870514e-05,
+      "loss": 0.877,
+      "step": 292350
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.149651794597076e-05,
+      "loss": 0.9273,
+      "step": 292360
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.1495370016071014e-05,
+      "loss": 1.0015,
+      "step": 292370
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.149422208617127e-05,
+      "loss": 1.0196,
+      "step": 292380
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.1493074156271514e-05,
+      "loss": 0.9667,
+      "step": 292390
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.149192622637177e-05,
+      "loss": 0.9421,
+      "step": 292400
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.149077829647203e-05,
+      "loss": 1.0046,
+      "step": 292410
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.148963036657228e-05,
+      "loss": 0.953,
+      "step": 292420
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.148848243667253e-05,
+      "loss": 0.9728,
+      "step": 292430
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.148733450677278e-05,
+      "loss": 0.9656,
+      "step": 292440
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.148618657687303e-05,
+      "loss": 1.0027,
+      "step": 292450
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.148503864697329e-05,
+      "loss": 1.0689,
+      "step": 292460
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.148389071707354e-05,
+      "loss": 0.9462,
+      "step": 292470
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.148274278717379e-05,
+      "loss": 1.0341,
+      "step": 292480
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.1481594857274046e-05,
+      "loss": 1.0939,
+      "step": 292490
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.14804469273743e-05,
+      "loss": 0.9935,
+      "step": 292500
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.1479298997474546e-05,
+      "loss": 0.9399,
+      "step": 292510
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.14781510675748e-05,
+      "loss": 0.924,
+      "step": 292520
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.147700313767506e-05,
+      "loss": 0.9459,
+      "step": 292530
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.14758552077753e-05,
+      "loss": 0.9687,
+      "step": 292540
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.147470727787556e-05,
+      "loss": 0.9993,
+      "step": 292550
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.1473559347975815e-05,
+      "loss": 1.0169,
+      "step": 292560
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.1472411418076065e-05,
+      "loss": 0.9911,
+      "step": 292570
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.1471263488176315e-05,
+      "loss": 0.985,
+      "step": 292580
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.147011555827657e-05,
+      "loss": 0.9614,
+      "step": 292590
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.146896762837682e-05,
+      "loss": 0.9308,
+      "step": 292600
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.146781969847708e-05,
+      "loss": 0.9718,
+      "step": 292610
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.146667176857733e-05,
+      "loss": 0.939,
+      "step": 292620
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.146552383867758e-05,
+      "loss": 0.9488,
+      "step": 292630
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.1464375908777835e-05,
+      "loss": 0.9268,
+      "step": 292640
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.146322797887809e-05,
+      "loss": 1.0452,
+      "step": 292650
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.1462080048978334e-05,
+      "loss": 1.0167,
+      "step": 292660
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.146093211907859e-05,
+      "loss": 0.9789,
+      "step": 292670
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.145978418917885e-05,
+      "loss": 0.9507,
+      "step": 292680
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.145863625927909e-05,
+      "loss": 0.8944,
+      "step": 292690
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.145748832937935e-05,
+      "loss": 0.9786,
+      "step": 292700
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.1456340399479604e-05,
+      "loss": 0.9956,
+      "step": 292710
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.1455192469579854e-05,
+      "loss": 1.0065,
+      "step": 292720
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.1454044539680104e-05,
+      "loss": 0.87,
+      "step": 292730
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.145289660978036e-05,
+      "loss": 0.9482,
+      "step": 292740
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.145174867988061e-05,
+      "loss": 0.892,
+      "step": 292750
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.145060074998087e-05,
+      "loss": 0.966,
+      "step": 292760
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.144945282008112e-05,
+      "loss": 0.9997,
+      "step": 292770
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.1448304890181366e-05,
+      "loss": 1.0175,
+      "step": 292780
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.144715696028162e-05,
+      "loss": 0.9734,
+      "step": 292790
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.144600903038188e-05,
+      "loss": 0.9918,
+      "step": 292800
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.144486110048212e-05,
+      "loss": 1.1125,
+      "step": 292810
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.144371317058238e-05,
+      "loss": 1.0714,
+      "step": 292820
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.1442565240682636e-05,
+      "loss": 1.0149,
+      "step": 292830
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.144141731078288e-05,
+      "loss": 0.8763,
+      "step": 292840
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.1440269380883136e-05,
+      "loss": 0.9632,
+      "step": 292850
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.143912145098339e-05,
+      "loss": 1.0657,
+      "step": 292860
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.143797352108364e-05,
+      "loss": 0.9571,
+      "step": 292870
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.143682559118389e-05,
+      "loss": 0.9684,
+      "step": 292880
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.143567766128415e-05,
+      "loss": 0.9631,
+      "step": 292890
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.14345297313844e-05,
+      "loss": 1.0254,
+      "step": 292900
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.1433381801484655e-05,
+      "loss": 1.0074,
+      "step": 292910
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.1432233871584905e-05,
+      "loss": 0.9133,
+      "step": 292920
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.1431085941685155e-05,
+      "loss": 0.9264,
+      "step": 292930
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.142993801178541e-05,
+      "loss": 0.9869,
+      "step": 292940
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.142879008188567e-05,
+      "loss": 0.9714,
+      "step": 292950
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.142764215198591e-05,
+      "loss": 0.9574,
+      "step": 292960
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.142649422208617e-05,
+      "loss": 0.9801,
+      "step": 292970
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.1425346292186425e-05,
+      "loss": 1.0924,
+      "step": 292980
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.142419836228667e-05,
+      "loss": 0.9944,
+      "step": 292990
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.1423050432386924e-05,
+      "loss": 1.0397,
+      "step": 293000
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.142190250248718e-05,
+      "loss": 1.0126,
+      "step": 293010
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.142075457258743e-05,
+      "loss": 1.0778,
+      "step": 293020
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.141960664268768e-05,
+      "loss": 0.925,
+      "step": 293030
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.141845871278794e-05,
+      "loss": 1.0704,
+      "step": 293040
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.141731078288819e-05,
+      "loss": 0.8988,
+      "step": 293050
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.1416162852988444e-05,
+      "loss": 1.0626,
+      "step": 293060
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.1415014923088694e-05,
+      "loss": 0.9243,
+      "step": 293070
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.1413866993188943e-05,
+      "loss": 0.9773,
+      "step": 293080
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.14127190632892e-05,
+      "loss": 0.9311,
+      "step": 293090
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.141157113338946e-05,
+      "loss": 0.8464,
+      "step": 293100
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.14104232034897e-05,
+      "loss": 0.8691,
+      "step": 293110
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.1409275273589956e-05,
+      "loss": 1.019,
+      "step": 293120
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.140812734369021e-05,
+      "loss": 0.9542,
+      "step": 293130
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.1406979413790456e-05,
+      "loss": 0.9311,
+      "step": 293140
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.140583148389071e-05,
+      "loss": 1.0081,
+      "step": 293150
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.140468355399097e-05,
+      "loss": 0.9399,
+      "step": 293160
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.140353562409122e-05,
+      "loss": 1.0111,
+      "step": 293170
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.140238769419147e-05,
+      "loss": 0.9993,
+      "step": 293180
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.1401239764291726e-05,
+      "loss": 1.1188,
+      "step": 293190
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.1400091834391976e-05,
+      "loss": 0.921,
+      "step": 293200
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.139894390449223e-05,
+      "loss": 1.0343,
+      "step": 293210
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.139779597459248e-05,
+      "loss": 0.8885,
+      "step": 293220
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.139664804469273e-05,
+      "loss": 0.993,
+      "step": 293230
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.139550011479299e-05,
+      "loss": 0.9189,
+      "step": 293240
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.1394352184893245e-05,
+      "loss": 0.9532,
+      "step": 293250
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.139320425499349e-05,
+      "loss": 1.0066,
+      "step": 293260
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.1392056325093745e-05,
+      "loss": 1.1001,
+      "step": 293270
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.1390908395194e-05,
+      "loss": 1.0185,
+      "step": 293280
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.1389760465294245e-05,
+      "loss": 0.8857,
+      "step": 293290
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.13886125353945e-05,
+      "loss": 1.05,
+      "step": 293300
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.138746460549476e-05,
+      "loss": 0.9514,
+      "step": 293310
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.138631667559501e-05,
+      "loss": 0.9051,
+      "step": 293320
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.138516874569526e-05,
+      "loss": 0.9543,
+      "step": 293330
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.1384020815795514e-05,
+      "loss": 0.9451,
+      "step": 293340
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.1382872885895764e-05,
+      "loss": 0.9214,
+      "step": 293350
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.138172495599602e-05,
+      "loss": 1.0027,
+      "step": 293360
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.138057702609627e-05,
+      "loss": 1.0348,
+      "step": 293370
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.137942909619652e-05,
+      "loss": 0.9337,
+      "step": 293380
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.137828116629678e-05,
+      "loss": 0.9971,
+      "step": 293390
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.1377133236397034e-05,
+      "loss": 0.9266,
+      "step": 293400
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.137598530649728e-05,
+      "loss": 0.9757,
+      "step": 293410
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.137483737659753e-05,
+      "loss": 0.9066,
+      "step": 293420
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.137368944669779e-05,
+      "loss": 1.0379,
+      "step": 293430
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.137254151679803e-05,
+      "loss": 0.9528,
+      "step": 293440
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.137139358689829e-05,
+      "loss": 0.9683,
+      "step": 293450
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.1370245656998546e-05,
+      "loss": 0.9954,
+      "step": 293460
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.1369097727098796e-05,
+      "loss": 1.0217,
+      "step": 293470
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.1367949797199046e-05,
+      "loss": 0.922,
+      "step": 293480
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.13668018672993e-05,
+      "loss": 1.041,
+      "step": 293490
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.136565393739955e-05,
+      "loss": 1.0432,
+      "step": 293500
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.136450600749981e-05,
+      "loss": 0.938,
+      "step": 293510
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.136335807760006e-05,
+      "loss": 0.92,
+      "step": 293520
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.136221014770031e-05,
+      "loss": 1.0623,
+      "step": 293530
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.1361062217800565e-05,
+      "loss": 1.0203,
+      "step": 293540
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.135991428790081e-05,
+      "loss": 1.1792,
+      "step": 293550
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.1358766358001065e-05,
+      "loss": 0.9359,
+      "step": 293560
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.135761842810132e-05,
+      "loss": 1.0628,
+      "step": 293570
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.135647049820157e-05,
+      "loss": 1.0366,
+      "step": 293580
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.135532256830182e-05,
+      "loss": 0.9463,
+      "step": 293590
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.135417463840208e-05,
+      "loss": 0.9035,
+      "step": 293600
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.135302670850233e-05,
+      "loss": 0.9379,
+      "step": 293610
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.1351878778602585e-05,
+      "loss": 0.9748,
+      "step": 293620
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.1350730848702835e-05,
+      "loss": 0.9592,
+      "step": 293630
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.1349582918803084e-05,
+      "loss": 0.9446,
+      "step": 293640
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.134843498890334e-05,
+      "loss": 1.0666,
+      "step": 293650
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.13472870590036e-05,
+      "loss": 0.9938,
+      "step": 293660
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.134613912910384e-05,
+      "loss": 0.9072,
+      "step": 293670
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.13449911992041e-05,
+      "loss": 0.9725,
+      "step": 293680
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.1343843269304354e-05,
+      "loss": 1.0693,
+      "step": 293690
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.13426953394046e-05,
+      "loss": 1.0274,
+      "step": 293700
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.1341547409504854e-05,
+      "loss": 1.0664,
+      "step": 293710
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.134039947960511e-05,
+      "loss": 1.0321,
+      "step": 293720
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.133925154970536e-05,
+      "loss": 0.9359,
+      "step": 293730
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.133810361980561e-05,
+      "loss": 0.9676,
+      "step": 293740
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.133695568990587e-05,
+      "loss": 0.8651,
+      "step": 293750
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.1335807760006116e-05,
+      "loss": 1.0374,
+      "step": 293760
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.133465983010637e-05,
+      "loss": 1.0392,
+      "step": 293770
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.133351190020662e-05,
+      "loss": 0.9331,
+      "step": 293780
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.133236397030687e-05,
+      "loss": 1.0549,
+      "step": 293790
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.133121604040713e-05,
+      "loss": 0.964,
+      "step": 293800
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.1330068110507386e-05,
+      "loss": 0.9916,
+      "step": 293810
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.132892018060763e-05,
+      "loss": 0.9662,
+      "step": 293820
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.1327772250707886e-05,
+      "loss": 0.9999,
+      "step": 293830
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.132662432080814e-05,
+      "loss": 1.0558,
+      "step": 293840
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.1325476390908386e-05,
+      "loss": 0.9372,
+      "step": 293850
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.132432846100864e-05,
+      "loss": 1.0043,
+      "step": 293860
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.13231805311089e-05,
+      "loss": 0.9661,
+      "step": 293870
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.132203260120915e-05,
+      "loss": 0.9905,
+      "step": 293880
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.13208846713094e-05,
+      "loss": 1.0206,
+      "step": 293890
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.1319736741409655e-05,
+      "loss": 1.0602,
+      "step": 293900
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.1318588811509905e-05,
+      "loss": 0.9846,
+      "step": 293910
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.131744088161016e-05,
+      "loss": 0.9108,
+      "step": 293920
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.131629295171041e-05,
+      "loss": 0.9645,
+      "step": 293930
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.131514502181066e-05,
+      "loss": 0.9392,
+      "step": 293940
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.131399709191092e-05,
+      "loss": 1.0012,
+      "step": 293950
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.1312849162011175e-05,
+      "loss": 1.0046,
+      "step": 293960
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.131170123211142e-05,
+      "loss": 1.0903,
+      "step": 293970
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.1310553302211674e-05,
+      "loss": 1.0282,
+      "step": 293980
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.130940537231193e-05,
+      "loss": 0.8982,
+      "step": 293990
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.1308257442412174e-05,
+      "loss": 0.9421,
+      "step": 294000
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.130710951251243e-05,
+      "loss": 0.9372,
+      "step": 294010
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.130596158261269e-05,
+      "loss": 0.9517,
+      "step": 294020
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.130481365271294e-05,
+      "loss": 0.9091,
+      "step": 294030
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.130366572281319e-05,
+      "loss": 0.9737,
+      "step": 294040
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.1302517792913444e-05,
+      "loss": 1.0774,
+      "step": 294050
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.1301369863013693e-05,
+      "loss": 0.9637,
+      "step": 294060
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.130022193311395e-05,
+      "loss": 0.9181,
+      "step": 294070
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.12990740032142e-05,
+      "loss": 0.9064,
+      "step": 294080
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.129792607331445e-05,
+      "loss": 0.9733,
+      "step": 294090
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.1296778143414706e-05,
+      "loss": 0.912,
+      "step": 294100
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.129563021351496e-05,
+      "loss": 0.9159,
+      "step": 294110
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.1294482283615206e-05,
+      "loss": 0.9423,
+      "step": 294120
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.129333435371546e-05,
+      "loss": 0.998,
+      "step": 294130
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.129218642381572e-05,
+      "loss": 1.0723,
+      "step": 294140
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.129103849391596e-05,
+      "loss": 1.024,
+      "step": 294150
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.128989056401622e-05,
+      "loss": 0.8401,
+      "step": 294160
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.1288742634116476e-05,
+      "loss": 0.9829,
+      "step": 294170
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.1287594704216726e-05,
+      "loss": 0.9356,
+      "step": 294180
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.1286446774316975e-05,
+      "loss": 0.9098,
+      "step": 294190
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.128529884441723e-05,
+      "loss": 0.9156,
+      "step": 294200
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.128415091451748e-05,
+      "loss": 0.9493,
+      "step": 294210
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.128300298461774e-05,
+      "loss": 0.9212,
+      "step": 294220
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.128185505471799e-05,
+      "loss": 0.9496,
+      "step": 294230
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.128070712481824e-05,
+      "loss": 0.8834,
+      "step": 294240
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.1279559194918495e-05,
+      "loss": 1.073,
+      "step": 294250
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.127841126501875e-05,
+      "loss": 0.9629,
+      "step": 294260
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.1277263335118995e-05,
+      "loss": 0.9487,
+      "step": 294270
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.127611540521925e-05,
+      "loss": 1.0423,
+      "step": 294280
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.127496747531951e-05,
+      "loss": 1.0189,
+      "step": 294290
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.127381954541975e-05,
+      "loss": 1.0139,
+      "step": 294300
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.127267161552001e-05,
+      "loss": 0.9494,
+      "step": 294310
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.1271523685620264e-05,
+      "loss": 0.9497,
+      "step": 294320
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.1270375755720514e-05,
+      "loss": 0.9722,
+      "step": 294330
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.1269227825820764e-05,
+      "loss": 1.1025,
+      "step": 294340
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.126807989592102e-05,
+      "loss": 1.0283,
+      "step": 294350
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.126693196602127e-05,
+      "loss": 1.0052,
+      "step": 294360
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.126578403612153e-05,
+      "loss": 0.9954,
+      "step": 294370
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.126463610622178e-05,
+      "loss": 1.0076,
+      "step": 294380
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.126348817632203e-05,
+      "loss": 1.0196,
+      "step": 294390
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.126234024642228e-05,
+      "loss": 0.8965,
+      "step": 294400
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.126119231652254e-05,
+      "loss": 1.0046,
+      "step": 294410
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.126004438662278e-05,
+      "loss": 0.977,
+      "step": 294420
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.125889645672304e-05,
+      "loss": 1.0291,
+      "step": 294430
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.1257748526823296e-05,
+      "loss": 1.0548,
+      "step": 294440
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.125660059692354e-05,
+      "loss": 1.0063,
+      "step": 294450
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.1255452667023796e-05,
+      "loss": 0.9901,
+      "step": 294460
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.125430473712405e-05,
+      "loss": 0.9566,
+      "step": 294470
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.12531568072243e-05,
+      "loss": 0.9699,
+      "step": 294480
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.125200887732455e-05,
+      "loss": 1.0402,
+      "step": 294490
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.125086094742481e-05,
+      "loss": 1.1498,
+      "step": 294500
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.124971301752506e-05,
+      "loss": 0.9838,
+      "step": 294510
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.1248565087625315e-05,
+      "loss": 0.9752,
+      "step": 294520
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.1247417157725565e-05,
+      "loss": 0.9824,
+      "step": 294530
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.1246269227825815e-05,
+      "loss": 1.0372,
+      "step": 294540
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.124512129792607e-05,
+      "loss": 0.8983,
+      "step": 294550
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.124397336802633e-05,
+      "loss": 0.97,
+      "step": 294560
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.124282543812657e-05,
+      "loss": 0.9859,
+      "step": 294570
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.124167750822683e-05,
+      "loss": 0.9934,
+      "step": 294580
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.1240529578327085e-05,
+      "loss": 0.9275,
+      "step": 294590
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.123938164842733e-05,
+      "loss": 1.0166,
+      "step": 294600
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.1238233718527584e-05,
+      "loss": 0.9393,
+      "step": 294610
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.123708578862784e-05,
+      "loss": 0.969,
+      "step": 294620
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.123593785872809e-05,
+      "loss": 0.9132,
+      "step": 294630
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.123478992882834e-05,
+      "loss": 0.9224,
+      "step": 294640
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.12336419989286e-05,
+      "loss": 1.0295,
+      "step": 294650
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.123249406902885e-05,
+      "loss": 0.9839,
+      "step": 294660
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.1231346139129104e-05,
+      "loss": 0.9954,
+      "step": 294670
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.1230198209229354e-05,
+      "loss": 0.9421,
+      "step": 294680
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.1229050279329604e-05,
+      "loss": 0.9854,
+      "step": 294690
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.122790234942986e-05,
+      "loss": 1.0153,
+      "step": 294700
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.122675441953012e-05,
+      "loss": 0.9484,
+      "step": 294710
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.122560648963036e-05,
+      "loss": 0.9766,
+      "step": 294720
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.122445855973062e-05,
+      "loss": 0.9189,
+      "step": 294730
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.122331062983087e-05,
+      "loss": 0.8996,
+      "step": 294740
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.1222162699931116e-05,
+      "loss": 1.0292,
+      "step": 294750
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.122101477003137e-05,
+      "loss": 0.8964,
+      "step": 294760
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.121986684013163e-05,
+      "loss": 0.8886,
+      "step": 294770
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.121871891023188e-05,
+      "loss": 1.0467,
+      "step": 294780
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.121757098033213e-05,
+      "loss": 1.0554,
+      "step": 294790
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.1216423050432386e-05,
+      "loss": 0.8145,
+      "step": 294800
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.1215275120532636e-05,
+      "loss": 0.982,
+      "step": 294810
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.121412719063289e-05,
+      "loss": 0.9345,
+      "step": 294820
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.121297926073314e-05,
+      "loss": 0.9457,
+      "step": 294830
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.121183133083339e-05,
+      "loss": 0.909,
+      "step": 294840
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.121068340093365e-05,
+      "loss": 0.9865,
+      "step": 294850
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.1209535471033905e-05,
+      "loss": 0.9726,
+      "step": 294860
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.120838754113415e-05,
+      "loss": 0.9117,
+      "step": 294870
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.1207239611234405e-05,
+      "loss": 1.0592,
+      "step": 294880
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.120609168133466e-05,
+      "loss": 0.9676,
+      "step": 294890
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.1204943751434905e-05,
+      "loss": 1.0827,
+      "step": 294900
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.120379582153516e-05,
+      "loss": 1.0002,
+      "step": 294910
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.120264789163541e-05,
+      "loss": 0.9591,
+      "step": 294920
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.120149996173567e-05,
+      "loss": 0.9543,
+      "step": 294930
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.120035203183592e-05,
+      "loss": 1.0092,
+      "step": 294940
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.119920410193617e-05,
+      "loss": 1.036,
+      "step": 294950
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.1198056172036424e-05,
+      "loss": 1.0103,
+      "step": 294960
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.119690824213668e-05,
+      "loss": 1.0343,
+      "step": 294970
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.1195760312236924e-05,
+      "loss": 0.9786,
+      "step": 294980
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.119461238233718e-05,
+      "loss": 0.9626,
+      "step": 294990
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.119346445243744e-05,
+      "loss": 1.1173,
+      "step": 295000
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.119231652253768e-05,
+      "loss": 1.0336,
+      "step": 295010
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.119116859263794e-05,
+      "loss": 0.9289,
+      "step": 295020
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.1190020662738194e-05,
+      "loss": 0.9586,
+      "step": 295030
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.1188872732838443e-05,
+      "loss": 1.0687,
+      "step": 295040
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.118772480293869e-05,
+      "loss": 0.9167,
+      "step": 295050
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.118657687303895e-05,
+      "loss": 0.9231,
+      "step": 295060
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.11854289431392e-05,
+      "loss": 1.0381,
+      "step": 295070
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.1184281013239456e-05,
+      "loss": 0.9158,
+      "step": 295080
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.1183133083339706e-05,
+      "loss": 0.987,
+      "step": 295090
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.1181985153439956e-05,
+      "loss": 1.0861,
+      "step": 295100
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.118083722354021e-05,
+      "loss": 1.0894,
+      "step": 295110
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.117968929364047e-05,
+      "loss": 1.0034,
+      "step": 295120
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.117854136374071e-05,
+      "loss": 0.9963,
+      "step": 295130
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.117739343384097e-05,
+      "loss": 0.9711,
+      "step": 295140
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.1176245503941226e-05,
+      "loss": 0.8425,
+      "step": 295150
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.117509757404147e-05,
+      "loss": 0.9199,
+      "step": 295160
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.1173949644141725e-05,
+      "loss": 1.0013,
+      "step": 295170
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.117280171424198e-05,
+      "loss": 1.0476,
+      "step": 295180
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.117165378434223e-05,
+      "loss": 1.0162,
+      "step": 295190
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.117050585444248e-05,
+      "loss": 1.0783,
+      "step": 295200
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.116935792454274e-05,
+      "loss": 1.1003,
+      "step": 295210
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.116820999464299e-05,
+      "loss": 1.0644,
+      "step": 295220
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.1167062064743245e-05,
+      "loss": 0.9832,
+      "step": 295230
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.1165914134843495e-05,
+      "loss": 0.972,
+      "step": 295240
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.1164766204943745e-05,
+      "loss": 0.9816,
+      "step": 295250
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.1163618275044e-05,
+      "loss": 0.9485,
+      "step": 295260
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.116247034514426e-05,
+      "loss": 0.9225,
+      "step": 295270
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.11613224152445e-05,
+      "loss": 1.0509,
+      "step": 295280
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.116017448534476e-05,
+      "loss": 1.0381,
+      "step": 295290
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.1159026555445014e-05,
+      "loss": 0.9348,
+      "step": 295300
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.115787862554526e-05,
+      "loss": 1.0053,
+      "step": 295310
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.1156730695645514e-05,
+      "loss": 0.9691,
+      "step": 295320
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.115558276574577e-05,
+      "loss": 0.9519,
+      "step": 295330
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.115443483584602e-05,
+      "loss": 1.008,
+      "step": 295340
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.115328690594627e-05,
+      "loss": 0.9285,
+      "step": 295350
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.115213897604653e-05,
+      "loss": 0.9841,
+      "step": 295360
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.115099104614678e-05,
+      "loss": 0.8978,
+      "step": 295370
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.114984311624703e-05,
+      "loss": 1.0017,
+      "step": 295380
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.114869518634728e-05,
+      "loss": 1.0541,
+      "step": 295390
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.114754725644753e-05,
+      "loss": 1.0273,
+      "step": 295400
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.114639932654779e-05,
+      "loss": 0.8988,
+      "step": 295410
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.1145251396648046e-05,
+      "loss": 0.9338,
+      "step": 295420
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.114410346674829e-05,
+      "loss": 0.9569,
+      "step": 295430
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.1142955536848546e-05,
+      "loss": 0.9418,
+      "step": 295440
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.11418076069488e-05,
+      "loss": 1.0565,
+      "step": 295450
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.1140659677049046e-05,
+      "loss": 0.9584,
+      "step": 295460
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.11395117471493e-05,
+      "loss": 0.8883,
+      "step": 295470
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.113836381724956e-05,
+      "loss": 0.9585,
+      "step": 295480
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.113721588734981e-05,
+      "loss": 0.9319,
+      "step": 295490
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.113606795745006e-05,
+      "loss": 1.028,
+      "step": 295500
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.1134920027550315e-05,
+      "loss": 0.995,
+      "step": 295510
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.1133772097650565e-05,
+      "loss": 0.9731,
+      "step": 295520
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.113262416775082e-05,
+      "loss": 0.9041,
+      "step": 295530
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.113147623785107e-05,
+      "loss": 0.9637,
+      "step": 295540
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.113032830795132e-05,
+      "loss": 0.9461,
+      "step": 295550
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.112918037805158e-05,
+      "loss": 1.0534,
+      "step": 295560
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.1128032448151835e-05,
+      "loss": 0.8858,
+      "step": 295570
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.112688451825208e-05,
+      "loss": 0.9957,
+      "step": 295580
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.1125736588352334e-05,
+      "loss": 0.9769,
+      "step": 295590
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.112458865845259e-05,
+      "loss": 1.0222,
+      "step": 295600
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.1123440728552834e-05,
+      "loss": 1.0356,
+      "step": 295610
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.112229279865309e-05,
+      "loss": 0.9531,
+      "step": 295620
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.112114486875335e-05,
+      "loss": 0.8948,
+      "step": 295630
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.11199969388536e-05,
+      "loss": 0.9961,
+      "step": 295640
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.111884900895385e-05,
+      "loss": 1.0257,
+      "step": 295650
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.1117701079054104e-05,
+      "loss": 1.0843,
+      "step": 295660
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.1116553149154354e-05,
+      "loss": 0.9259,
+      "step": 295670
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.111540521925461e-05,
+      "loss": 0.9421,
+      "step": 295680
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.111425728935486e-05,
+      "loss": 0.9875,
+      "step": 295690
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.111310935945511e-05,
+      "loss": 0.9827,
+      "step": 295700
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.111196142955537e-05,
+      "loss": 1.0157,
+      "step": 295710
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.111081349965562e-05,
+      "loss": 0.9423,
+      "step": 295720
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.1109665569755866e-05,
+      "loss": 0.956,
+      "step": 295730
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.110851763985612e-05,
+      "loss": 0.9286,
+      "step": 295740
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.110736970995638e-05,
+      "loss": 0.9596,
+      "step": 295750
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.110622178005662e-05,
+      "loss": 1.1181,
+      "step": 295760
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.110507385015688e-05,
+      "loss": 0.929,
+      "step": 295770
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.1103925920257136e-05,
+      "loss": 0.9121,
+      "step": 295780
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.1102777990357386e-05,
+      "loss": 0.9749,
+      "step": 295790
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.1101630060457636e-05,
+      "loss": 0.9504,
+      "step": 295800
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.110048213055789e-05,
+      "loss": 0.9849,
+      "step": 295810
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.109933420065814e-05,
+      "loss": 0.976,
+      "step": 295820
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.10981862707584e-05,
+      "loss": 0.9749,
+      "step": 295830
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.109703834085865e-05,
+      "loss": 1.0617,
+      "step": 295840
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.10958904109589e-05,
+      "loss": 1.0479,
+      "step": 295850
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.1094742481059155e-05,
+      "loss": 0.9482,
+      "step": 295860
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.109359455115941e-05,
+      "loss": 1.0203,
+      "step": 295870
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.1092446621259655e-05,
+      "loss": 0.9383,
+      "step": 295880
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.109129869135991e-05,
+      "loss": 1.017,
+      "step": 295890
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.109015076146017e-05,
+      "loss": 0.9741,
+      "step": 295900
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.108900283156041e-05,
+      "loss": 0.9196,
+      "step": 295910
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.108785490166067e-05,
+      "loss": 0.9339,
+      "step": 295920
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.1086706971760924e-05,
+      "loss": 1.0494,
+      "step": 295930
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.1085559041861174e-05,
+      "loss": 1.0279,
+      "step": 295940
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.1084411111961424e-05,
+      "loss": 0.9301,
+      "step": 295950
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.108326318206168e-05,
+      "loss": 1.0483,
+      "step": 295960
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.108211525216193e-05,
+      "loss": 0.9305,
+      "step": 295970
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.108096732226219e-05,
+      "loss": 0.9802,
+      "step": 295980
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.107981939236244e-05,
+      "loss": 0.9903,
+      "step": 295990
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.107867146246269e-05,
+      "loss": 1.014,
+      "step": 296000
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.1077523532562944e-05,
+      "loss": 1.0862,
+      "step": 296010
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.10763756026632e-05,
+      "loss": 1.0345,
+      "step": 296020
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.107522767276344e-05,
+      "loss": 0.9482,
+      "step": 296030
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.10740797428637e-05,
+      "loss": 1.0046,
+      "step": 296040
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.1072931812963957e-05,
+      "loss": 1.0162,
+      "step": 296050
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.10717838830642e-05,
+      "loss": 0.9156,
+      "step": 296060
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.1070635953164456e-05,
+      "loss": 1.0467,
+      "step": 296070
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.106948802326471e-05,
+      "loss": 0.9432,
+      "step": 296080
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.106834009336496e-05,
+      "loss": 0.9306,
+      "step": 296090
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.106719216346521e-05,
+      "loss": 0.9333,
+      "step": 296100
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.106604423356547e-05,
+      "loss": 1.0087,
+      "step": 296110
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.106489630366572e-05,
+      "loss": 0.9776,
+      "step": 296120
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.1063748373765976e-05,
+      "loss": 1.0124,
+      "step": 296130
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.1062600443866226e-05,
+      "loss": 0.9733,
+      "step": 296140
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.1061452513966475e-05,
+      "loss": 1.0388,
+      "step": 296150
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.106030458406673e-05,
+      "loss": 1.0236,
+      "step": 296160
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.105915665416699e-05,
+      "loss": 0.8941,
+      "step": 296170
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.105800872426723e-05,
+      "loss": 1.058,
+      "step": 296180
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.105686079436749e-05,
+      "loss": 1.0204,
+      "step": 296190
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.1055712864467745e-05,
+      "loss": 1.038,
+      "step": 296200
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.105456493456799e-05,
+      "loss": 1.0265,
+      "step": 296210
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.1053417004668245e-05,
+      "loss": 0.9171,
+      "step": 296220
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.10522690747685e-05,
+      "loss": 1.0427,
+      "step": 296230
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.105112114486875e-05,
+      "loss": 0.9224,
+      "step": 296240
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.1049973214969e-05,
+      "loss": 0.9647,
+      "step": 296250
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.104882528506926e-05,
+      "loss": 1.1182,
+      "step": 296260
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.104767735516951e-05,
+      "loss": 1.0721,
+      "step": 296270
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.1046529425269764e-05,
+      "loss": 1.0276,
+      "step": 296280
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.1045381495370014e-05,
+      "loss": 0.9187,
+      "step": 296290
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.1044233565470264e-05,
+      "loss": 1.0614,
+      "step": 296300
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.104308563557052e-05,
+      "loss": 0.9317,
+      "step": 296310
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.1041937705670764e-05,
+      "loss": 1.1115,
+      "step": 296320
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.104078977577102e-05,
+      "loss": 1.0003,
+      "step": 296330
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.103964184587128e-05,
+      "loss": 0.9423,
+      "step": 296340
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.103849391597153e-05,
+      "loss": 0.9743,
+      "step": 296350
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.1037345986071777e-05,
+      "loss": 0.8405,
+      "step": 296360
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.103619805617203e-05,
+      "loss": 1.0163,
+      "step": 296370
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.103505012627228e-05,
+      "loss": 1.022,
+      "step": 296380
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.103390219637254e-05,
+      "loss": 0.9312,
+      "step": 296390
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.103275426647279e-05,
+      "loss": 0.9473,
+      "step": 296400
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.103160633657304e-05,
+      "loss": 0.9296,
+      "step": 296410
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.1030458406673296e-05,
+      "loss": 0.933,
+      "step": 296420
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.102931047677355e-05,
+      "loss": 0.9653,
+      "step": 296430
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.1028162546873796e-05,
+      "loss": 0.9513,
+      "step": 296440
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.102701461697405e-05,
+      "loss": 0.8571,
+      "step": 296450
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.102586668707431e-05,
+      "loss": 0.9815,
+      "step": 296460
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.102471875717455e-05,
+      "loss": 0.9651,
+      "step": 296470
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.102357082727481e-05,
+      "loss": 0.9084,
+      "step": 296480
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.1022422897375065e-05,
+      "loss": 0.9093,
+      "step": 296490
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.1021274967475315e-05,
+      "loss": 0.9601,
+      "step": 296500
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.1020127037575565e-05,
+      "loss": 1.0173,
+      "step": 296510
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.101897910767582e-05,
+      "loss": 0.9308,
+      "step": 296520
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.101783117777607e-05,
+      "loss": 1.017,
+      "step": 296530
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.101668324787633e-05,
+      "loss": 0.9485,
+      "step": 296540
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.101553531797658e-05,
+      "loss": 1.0362,
+      "step": 296550
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.101438738807683e-05,
+      "loss": 1.0526,
+      "step": 296560
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.1013239458177084e-05,
+      "loss": 0.9765,
+      "step": 296570
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.101209152827734e-05,
+      "loss": 0.9513,
+      "step": 296580
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.1010943598377584e-05,
+      "loss": 1.1814,
+      "step": 296590
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.100979566847784e-05,
+      "loss": 1.0964,
+      "step": 296600
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.10086477385781e-05,
+      "loss": 1.1562,
+      "step": 296610
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.100749980867834e-05,
+      "loss": 0.9633,
+      "step": 296620
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.10063518787786e-05,
+      "loss": 1.0266,
+      "step": 296630
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.1005203948878854e-05,
+      "loss": 0.9772,
+      "step": 296640
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.1004056018979104e-05,
+      "loss": 0.9011,
+      "step": 296650
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.1002908089079354e-05,
+      "loss": 1.1281,
+      "step": 296660
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.100176015917961e-05,
+      "loss": 0.9199,
+      "step": 296670
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.100061222927986e-05,
+      "loss": 0.8646,
+      "step": 296680
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.099946429938012e-05,
+      "loss": 0.9405,
+      "step": 296690
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.0998316369480366e-05,
+      "loss": 1.0474,
+      "step": 296700
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.0997168439580616e-05,
+      "loss": 0.9613,
+      "step": 296710
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.099602050968087e-05,
+      "loss": 0.957,
+      "step": 296720
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.099487257978113e-05,
+      "loss": 0.9439,
+      "step": 296730
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.099372464988137e-05,
+      "loss": 1.041,
+      "step": 296740
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.099257671998163e-05,
+      "loss": 0.9814,
+      "step": 296750
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.0991428790081886e-05,
+      "loss": 0.9655,
+      "step": 296760
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.099028086018213e-05,
+      "loss": 0.9986,
+      "step": 296770
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.0989132930282386e-05,
+      "loss": 1.0854,
+      "step": 296780
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.098798500038264e-05,
+      "loss": 0.907,
+      "step": 296790
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.098683707048289e-05,
+      "loss": 0.9963,
+      "step": 296800
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.098568914058314e-05,
+      "loss": 1.1462,
+      "step": 296810
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.09845412106834e-05,
+      "loss": 0.9604,
+      "step": 296820
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.098339328078365e-05,
+      "loss": 0.9485,
+      "step": 296830
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.0982245350883905e-05,
+      "loss": 1.0137,
+      "step": 296840
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.0981097420984155e-05,
+      "loss": 0.9818,
+      "step": 296850
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.0979949491084405e-05,
+      "loss": 0.9606,
+      "step": 296860
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.097880156118466e-05,
+      "loss": 0.9662,
+      "step": 296870
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.097765363128492e-05,
+      "loss": 1.0053,
+      "step": 296880
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.097650570138516e-05,
+      "loss": 1.0232,
+      "step": 296890
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.097535777148542e-05,
+      "loss": 0.9797,
+      "step": 296900
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.0974209841585674e-05,
+      "loss": 1.0187,
+      "step": 296910
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.097306191168592e-05,
+      "loss": 1.0104,
+      "step": 296920
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.0971913981786174e-05,
+      "loss": 1.0766,
+      "step": 296930
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.097076605188643e-05,
+      "loss": 1.0498,
+      "step": 296940
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.096961812198668e-05,
+      "loss": 0.945,
+      "step": 296950
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.096847019208693e-05,
+      "loss": 0.8997,
+      "step": 296960
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.096732226218719e-05,
+      "loss": 0.9408,
+      "step": 296970
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.096617433228744e-05,
+      "loss": 0.9963,
+      "step": 296980
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.0965026402387694e-05,
+      "loss": 0.9124,
+      "step": 296990
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.0963878472487943e-05,
+      "loss": 0.9715,
+      "step": 297000
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.096273054258819e-05,
+      "loss": 1.024,
+      "step": 297010
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.096158261268845e-05,
+      "loss": 1.0771,
+      "step": 297020
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.0960434682788707e-05,
+      "loss": 1.0294,
+      "step": 297030
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.095928675288895e-05,
+      "loss": 0.9813,
+      "step": 297040
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.0958138822989206e-05,
+      "loss": 0.9388,
+      "step": 297050
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.095699089308946e-05,
+      "loss": 0.9815,
+      "step": 297060
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.0955842963189706e-05,
+      "loss": 1.1005,
+      "step": 297070
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.095469503328996e-05,
+      "loss": 0.9475,
+      "step": 297080
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.095354710339022e-05,
+      "loss": 0.9198,
+      "step": 297090
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.095239917349047e-05,
+      "loss": 0.9664,
+      "step": 297100
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.095125124359072e-05,
+      "loss": 0.9805,
+      "step": 297110
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.0950103313690976e-05,
+      "loss": 1.0514,
+      "step": 297120
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.0948955383791225e-05,
+      "loss": 0.8952,
+      "step": 297130
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.094780745389148e-05,
+      "loss": 1.0617,
+      "step": 297140
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.094665952399173e-05,
+      "loss": 1.0091,
+      "step": 297150
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.094551159409198e-05,
+      "loss": 1.074,
+      "step": 297160
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.094436366419224e-05,
+      "loss": 0.9575,
+      "step": 297170
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.0943215734292495e-05,
+      "loss": 0.9289,
+      "step": 297180
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.094206780439274e-05,
+      "loss": 0.9097,
+      "step": 297190
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.0940919874492995e-05,
+      "loss": 0.9194,
+      "step": 297200
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.093977194459325e-05,
+      "loss": 0.9531,
+      "step": 297210
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.0938624014693494e-05,
+      "loss": 0.9633,
+      "step": 297220
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.093747608479375e-05,
+      "loss": 0.9789,
+      "step": 297230
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.093632815489401e-05,
+      "loss": 0.9688,
+      "step": 297240
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.093518022499426e-05,
+      "loss": 1.048,
+      "step": 297250
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.093403229509451e-05,
+      "loss": 0.9561,
+      "step": 297260
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.0932884365194764e-05,
+      "loss": 0.9821,
+      "step": 297270
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.0931736435295014e-05,
+      "loss": 0.9854,
+      "step": 297280
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.093058850539527e-05,
+      "loss": 0.9321,
+      "step": 297290
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.092944057549552e-05,
+      "loss": 0.9606,
+      "step": 297300
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.092829264559577e-05,
+      "loss": 0.9661,
+      "step": 297310
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.092714471569603e-05,
+      "loss": 0.9818,
+      "step": 297320
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.0925996785796283e-05,
+      "loss": 1.0481,
+      "step": 297330
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.0924848855896527e-05,
+      "loss": 0.9386,
+      "step": 297340
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.092370092599678e-05,
+      "loss": 0.9859,
+      "step": 297350
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.092255299609704e-05,
+      "loss": 0.9709,
+      "step": 297360
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.092140506619728e-05,
+      "loss": 0.9708,
+      "step": 297370
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.092025713629754e-05,
+      "loss": 0.9736,
+      "step": 297380
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.0919109206397796e-05,
+      "loss": 0.9725,
+      "step": 297390
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.0917961276498046e-05,
+      "loss": 1.1315,
+      "step": 297400
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.0916813346598296e-05,
+      "loss": 0.9913,
+      "step": 297410
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.091566541669855e-05,
+      "loss": 0.9571,
+      "step": 297420
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.09145174867988e-05,
+      "loss": 1.0172,
+      "step": 297430
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.091336955689906e-05,
+      "loss": 1.007,
+      "step": 297440
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.091222162699931e-05,
+      "loss": 0.8961,
+      "step": 297450
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.091107369709956e-05,
+      "loss": 0.92,
+      "step": 297460
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.0909925767199815e-05,
+      "loss": 1.0017,
+      "step": 297470
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.090877783730007e-05,
+      "loss": 0.9312,
+      "step": 297480
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.0907629907400315e-05,
+      "loss": 0.8587,
+      "step": 297490
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.090648197750057e-05,
+      "loss": 0.9687,
+      "step": 297500
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.090533404760083e-05,
+      "loss": 0.9628,
+      "step": 297510
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.090418611770107e-05,
+      "loss": 0.9906,
+      "step": 297520
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.090303818780133e-05,
+      "loss": 1.0589,
+      "step": 297530
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.0901890257901585e-05,
+      "loss": 0.9954,
+      "step": 297540
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.0900742328001834e-05,
+      "loss": 1.0003,
+      "step": 297550
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.0899594398102084e-05,
+      "loss": 0.9993,
+      "step": 297560
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.089844646820234e-05,
+      "loss": 0.9757,
+      "step": 297570
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.089729853830259e-05,
+      "loss": 0.9903,
+      "step": 297580
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.089615060840285e-05,
+      "loss": 0.8875,
+      "step": 297590
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.08950026785031e-05,
+      "loss": 0.9975,
+      "step": 297600
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.089385474860335e-05,
+      "loss": 0.9192,
+      "step": 297610
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.0892706818703604e-05,
+      "loss": 0.9998,
+      "step": 297620
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.089155888880386e-05,
+      "loss": 0.938,
+      "step": 297630
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.0890410958904104e-05,
+      "loss": 0.9513,
+      "step": 297640
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.088926302900436e-05,
+      "loss": 0.9849,
+      "step": 297650
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.088811509910462e-05,
+      "loss": 0.9896,
+      "step": 297660
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.088696716920486e-05,
+      "loss": 1.0004,
+      "step": 297670
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.0885819239305116e-05,
+      "loss": 0.9045,
+      "step": 297680
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.0884671309405366e-05,
+      "loss": 0.9628,
+      "step": 297690
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.088352337950562e-05,
+      "loss": 0.9863,
+      "step": 297700
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.088237544960587e-05,
+      "loss": 0.9353,
+      "step": 297710
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.088122751970612e-05,
+      "loss": 0.8862,
+      "step": 297720
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.088007958980638e-05,
+      "loss": 0.9513,
+      "step": 297730
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.0878931659906636e-05,
+      "loss": 0.9875,
+      "step": 297740
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.087778373000688e-05,
+      "loss": 1.0439,
+      "step": 297750
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.0876635800107136e-05,
+      "loss": 1.0507,
+      "step": 297760
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.087548787020739e-05,
+      "loss": 0.9843,
+      "step": 297770
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.0874339940307635e-05,
+      "loss": 0.9245,
+      "step": 297780
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.087319201040789e-05,
+      "loss": 0.9979,
+      "step": 297790
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.087204408050815e-05,
+      "loss": 0.9297,
+      "step": 297800
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.08708961506084e-05,
+      "loss": 0.8964,
+      "step": 297810
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.086974822070865e-05,
+      "loss": 0.9155,
+      "step": 297820
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.0868600290808905e-05,
+      "loss": 1.0298,
+      "step": 297830
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.0867452360909155e-05,
+      "loss": 0.9275,
+      "step": 297840
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.086630443100941e-05,
+      "loss": 0.9552,
+      "step": 297850
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.086515650110966e-05,
+      "loss": 0.8937,
+      "step": 297860
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.086400857120991e-05,
+      "loss": 0.9638,
+      "step": 297870
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.086286064131017e-05,
+      "loss": 0.9527,
+      "step": 297880
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.0861712711410424e-05,
+      "loss": 0.8835,
+      "step": 297890
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.086056478151067e-05,
+      "loss": 1.0192,
+      "step": 297900
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.0859416851610924e-05,
+      "loss": 0.964,
+      "step": 297910
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.085826892171118e-05,
+      "loss": 0.9492,
+      "step": 297920
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.0857120991811424e-05,
+      "loss": 0.9722,
+      "step": 297930
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.085597306191168e-05,
+      "loss": 1.0164,
+      "step": 297940
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.085482513201194e-05,
+      "loss": 0.9538,
+      "step": 297950
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.085367720211219e-05,
+      "loss": 0.9653,
+      "step": 297960
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.085252927221244e-05,
+      "loss": 0.9572,
+      "step": 297970
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.0851381342312693e-05,
+      "loss": 0.8899,
+      "step": 297980
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.085023341241294e-05,
+      "loss": 0.9541,
+      "step": 297990
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.08490854825132e-05,
+      "loss": 1.0762,
+      "step": 298000
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.084793755261345e-05,
+      "loss": 1.0996,
+      "step": 298010
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.08467896227137e-05,
+      "loss": 0.9698,
+      "step": 298020
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.0845641692813956e-05,
+      "loss": 1.0086,
+      "step": 298030
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.084449376291421e-05,
+      "loss": 0.9816,
+      "step": 298040
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.0843345833014456e-05,
+      "loss": 0.8945,
+      "step": 298050
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.084219790311471e-05,
+      "loss": 1.0517,
+      "step": 298060
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.084104997321497e-05,
+      "loss": 0.9769,
+      "step": 298070
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.083990204331521e-05,
+      "loss": 0.9408,
+      "step": 298080
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.083875411341547e-05,
+      "loss": 0.8963,
+      "step": 298090
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.0837606183515726e-05,
+      "loss": 0.9635,
+      "step": 298100
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.0836458253615975e-05,
+      "loss": 0.9749,
+      "step": 298110
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.0835310323716225e-05,
+      "loss": 0.9999,
+      "step": 298120
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.083416239381648e-05,
+      "loss": 1.0392,
+      "step": 298130
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.083301446391673e-05,
+      "loss": 1.0127,
+      "step": 298140
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.083186653401699e-05,
+      "loss": 1.0522,
+      "step": 298150
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.083071860411724e-05,
+      "loss": 1.0267,
+      "step": 298160
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.082957067421749e-05,
+      "loss": 0.9583,
+      "step": 298170
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.0828422744317745e-05,
+      "loss": 0.9769,
+      "step": 298180
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.0827274814418e-05,
+      "loss": 0.9476,
+      "step": 298190
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.0826126884518244e-05,
+      "loss": 1.0516,
+      "step": 298200
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.08249789546185e-05,
+      "loss": 0.9508,
+      "step": 298210
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.082383102471876e-05,
+      "loss": 0.9552,
+      "step": 298220
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.0822683094819e-05,
+      "loss": 1.0254,
+      "step": 298230
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.082153516491926e-05,
+      "loss": 1.058,
+      "step": 298240
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.0820387235019514e-05,
+      "loss": 0.9453,
+      "step": 298250
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.0819239305119764e-05,
+      "loss": 1.0662,
+      "step": 298260
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.0818091375220014e-05,
+      "loss": 0.9319,
+      "step": 298270
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.081694344532027e-05,
+      "loss": 0.9513,
+      "step": 298280
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.081579551542052e-05,
+      "loss": 0.9441,
+      "step": 298290
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.081464758552078e-05,
+      "loss": 0.9303,
+      "step": 298300
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.081349965562103e-05,
+      "loss": 0.9456,
+      "step": 298310
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.0812351725721277e-05,
+      "loss": 1.0804,
+      "step": 298320
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.081120379582153e-05,
+      "loss": 1.0776,
+      "step": 298330
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.081005586592179e-05,
+      "loss": 0.9637,
+      "step": 298340
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.080890793602203e-05,
+      "loss": 1.1191,
+      "step": 298350
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.080776000612229e-05,
+      "loss": 1.071,
+      "step": 298360
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.0806612076222546e-05,
+      "loss": 0.9441,
+      "step": 298370
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.080546414632279e-05,
+      "loss": 0.8964,
+      "step": 298380
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.0804316216423046e-05,
+      "loss": 0.9563,
+      "step": 298390
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.08031682865233e-05,
+      "loss": 1.0041,
+      "step": 298400
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.080202035662355e-05,
+      "loss": 1.0296,
+      "step": 298410
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.08008724267238e-05,
+      "loss": 1.0237,
+      "step": 298420
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.079972449682406e-05,
+      "loss": 1.0745,
+      "step": 298430
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.079857656692431e-05,
+      "loss": 0.9049,
+      "step": 298440
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.0797428637024565e-05,
+      "loss": 0.8953,
+      "step": 298450
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.0796280707124815e-05,
+      "loss": 1.0078,
+      "step": 298460
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.0795132777225065e-05,
+      "loss": 1.0222,
+      "step": 298470
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.079398484732532e-05,
+      "loss": 0.9997,
+      "step": 298480
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.079283691742558e-05,
+      "loss": 0.9656,
+      "step": 298490
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.079168898752582e-05,
+      "loss": 1.1159,
+      "step": 298500
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.079054105762608e-05,
+      "loss": 0.9829,
+      "step": 298510
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.0789393127726335e-05,
+      "loss": 0.9334,
+      "step": 298520
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.078824519782658e-05,
+      "loss": 0.9188,
+      "step": 298530
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.0787097267926834e-05,
+      "loss": 0.898,
+      "step": 298540
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.078594933802709e-05,
+      "loss": 0.9147,
+      "step": 298550
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.078480140812734e-05,
+      "loss": 0.9767,
+      "step": 298560
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.078365347822759e-05,
+      "loss": 1.0414,
+      "step": 298570
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.078250554832785e-05,
+      "loss": 0.954,
+      "step": 298580
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.07813576184281e-05,
+      "loss": 0.9429,
+      "step": 298590
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.0780209688528354e-05,
+      "loss": 0.8948,
+      "step": 298600
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.0779061758628604e-05,
+      "loss": 0.9584,
+      "step": 298610
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.0777913828728853e-05,
+      "loss": 0.899,
+      "step": 298620
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.077676589882911e-05,
+      "loss": 1.0514,
+      "step": 298630
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.077561796892937e-05,
+      "loss": 0.978,
+      "step": 298640
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.077447003902961e-05,
+      "loss": 1.0335,
+      "step": 298650
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.0773322109129866e-05,
+      "loss": 0.9951,
+      "step": 298660
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.077217417923012e-05,
+      "loss": 1.0008,
+      "step": 298670
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.0771026249330366e-05,
+      "loss": 0.9862,
+      "step": 298680
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.076987831943062e-05,
+      "loss": 0.9298,
+      "step": 298690
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.076873038953088e-05,
+      "loss": 1.0584,
+      "step": 298700
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.076758245963113e-05,
+      "loss": 0.9292,
+      "step": 298710
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.076643452973138e-05,
+      "loss": 1.0856,
+      "step": 298720
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.0765286599831636e-05,
+      "loss": 1.0308,
+      "step": 298730
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.0764138669931886e-05,
+      "loss": 1.107,
+      "step": 298740
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.076299074003214e-05,
+      "loss": 0.8674,
+      "step": 298750
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.076184281013239e-05,
+      "loss": 1.0316,
+      "step": 298760
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.076069488023264e-05,
+      "loss": 1.1555,
+      "step": 298770
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.07595469503329e-05,
+      "loss": 0.9652,
+      "step": 298780
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.0758399020433155e-05,
+      "loss": 0.9991,
+      "step": 298790
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.07572510905334e-05,
+      "loss": 0.887,
+      "step": 298800
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.0756103160633655e-05,
+      "loss": 0.9707,
+      "step": 298810
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.075495523073391e-05,
+      "loss": 0.9791,
+      "step": 298820
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.0753807300834155e-05,
+      "loss": 0.9649,
+      "step": 298830
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.075265937093441e-05,
+      "loss": 0.9978,
+      "step": 298840
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.075151144103467e-05,
+      "loss": 0.9762,
+      "step": 298850
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.075036351113492e-05,
+      "loss": 0.995,
+      "step": 298860
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.074921558123517e-05,
+      "loss": 0.8852,
+      "step": 298870
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.0748067651335424e-05,
+      "loss": 0.9368,
+      "step": 298880
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.0746919721435674e-05,
+      "loss": 0.9158,
+      "step": 298890
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.074577179153593e-05,
+      "loss": 0.9866,
+      "step": 298900
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.074462386163618e-05,
+      "loss": 0.8931,
+      "step": 298910
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.074347593173643e-05,
+      "loss": 0.9669,
+      "step": 298920
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.074232800183669e-05,
+      "loss": 0.9947,
+      "step": 298930
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.0741180071936944e-05,
+      "loss": 0.8577,
+      "step": 298940
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.074003214203719e-05,
+      "loss": 0.9853,
+      "step": 298950
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.0738884212137443e-05,
+      "loss": 1.0684,
+      "step": 298960
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.07377362822377e-05,
+      "loss": 0.9482,
+      "step": 298970
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.073658835233794e-05,
+      "loss": 1.0266,
+      "step": 298980
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.07354404224382e-05,
+      "loss": 0.9007,
+      "step": 298990
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.0734292492538456e-05,
+      "loss": 0.9696,
+      "step": 299000
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.0733144562638706e-05,
+      "loss": 0.994,
+      "step": 299010
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.0731996632738956e-05,
+      "loss": 0.968,
+      "step": 299020
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.073084870283921e-05,
+      "loss": 1.0397,
+      "step": 299030
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.072970077293946e-05,
+      "loss": 0.9574,
+      "step": 299040
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.072855284303972e-05,
+      "loss": 1.036,
+      "step": 299050
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.072740491313996e-05,
+      "loss": 1.118,
+      "step": 299060
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.072625698324022e-05,
+      "loss": 1.0841,
+      "step": 299070
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.0725109053340476e-05,
+      "loss": 0.9184,
+      "step": 299080
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.072396112344072e-05,
+      "loss": 0.9343,
+      "step": 299090
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.0722813193540975e-05,
+      "loss": 0.9946,
+      "step": 299100
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.072166526364123e-05,
+      "loss": 0.8992,
+      "step": 299110
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.072051733374148e-05,
+      "loss": 0.9315,
+      "step": 299120
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.071936940384173e-05,
+      "loss": 1.048,
+      "step": 299130
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.071822147394199e-05,
+      "loss": 1.0153,
+      "step": 299140
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.071707354404224e-05,
+      "loss": 1.0664,
+      "step": 299150
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.0715925614142495e-05,
+      "loss": 0.9594,
+      "step": 299160
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.0714777684242745e-05,
+      "loss": 0.9717,
+      "step": 299170
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.0713629754342994e-05,
+      "loss": 1.0062,
+      "step": 299180
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.071248182444325e-05,
+      "loss": 0.9566,
+      "step": 299190
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.071133389454351e-05,
+      "loss": 0.984,
+      "step": 299200
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.071018596464375e-05,
+      "loss": 1.0637,
+      "step": 299210
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.070903803474401e-05,
+      "loss": 1.0221,
+      "step": 299220
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.0707890104844264e-05,
+      "loss": 0.9769,
+      "step": 299230
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.070674217494451e-05,
+      "loss": 0.9323,
+      "step": 299240
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.0705594245044764e-05,
+      "loss": 0.9583,
+      "step": 299250
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.070444631514502e-05,
+      "loss": 0.9901,
+      "step": 299260
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.070329838524527e-05,
+      "loss": 0.9531,
+      "step": 299270
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.070215045534552e-05,
+      "loss": 0.9977,
+      "step": 299280
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.070100252544578e-05,
+      "loss": 1.0774,
+      "step": 299290
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.0699854595546027e-05,
+      "loss": 0.9817,
+      "step": 299300
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.069870666564628e-05,
+      "loss": 1.0497,
+      "step": 299310
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.069755873574653e-05,
+      "loss": 0.9647,
+      "step": 299320
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.069641080584678e-05,
+      "loss": 0.8666,
+      "step": 299330
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.069526287594704e-05,
+      "loss": 0.9926,
+      "step": 299340
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.0694114946047296e-05,
+      "loss": 1.0275,
+      "step": 299350
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.069296701614754e-05,
+      "loss": 0.9836,
+      "step": 299360
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.0691819086247796e-05,
+      "loss": 0.9865,
+      "step": 299370
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.069067115634805e-05,
+      "loss": 0.9212,
+      "step": 299380
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.0689523226448296e-05,
+      "loss": 0.9678,
+      "step": 299390
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.068837529654855e-05,
+      "loss": 0.9191,
+      "step": 299400
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.068722736664881e-05,
+      "loss": 0.9968,
+      "step": 299410
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.068607943674906e-05,
+      "loss": 0.9481,
+      "step": 299420
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.068493150684931e-05,
+      "loss": 0.9857,
+      "step": 299430
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.0683783576949565e-05,
+      "loss": 1.0175,
+      "step": 299440
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.0682635647049815e-05,
+      "loss": 0.9238,
+      "step": 299450
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.068148771715007e-05,
+      "loss": 0.946,
+      "step": 299460
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.068033978725032e-05,
+      "loss": 1.0962,
+      "step": 299470
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.067919185735057e-05,
+      "loss": 1.0577,
+      "step": 299480
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.067804392745083e-05,
+      "loss": 0.9592,
+      "step": 299490
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.0676895997551085e-05,
+      "loss": 1.0179,
+      "step": 299500
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.067574806765133e-05,
+      "loss": 0.952,
+      "step": 299510
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.0674600137751584e-05,
+      "loss": 0.9738,
+      "step": 299520
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.067345220785184e-05,
+      "loss": 0.9725,
+      "step": 299530
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.0672304277952084e-05,
+      "loss": 0.9381,
+      "step": 299540
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.067115634805234e-05,
+      "loss": 0.9639,
+      "step": 299550
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.06700084181526e-05,
+      "loss": 0.9013,
+      "step": 299560
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.066886048825285e-05,
+      "loss": 0.9588,
+      "step": 299570
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.06677125583531e-05,
+      "loss": 0.956,
+      "step": 299580
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.0666564628453354e-05,
+      "loss": 0.9374,
+      "step": 299590
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.0665416698553603e-05,
+      "loss": 1.0963,
+      "step": 299600
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.066426876865386e-05,
+      "loss": 0.9625,
+      "step": 299610
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.066312083875411e-05,
+      "loss": 1.0245,
+      "step": 299620
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.066197290885436e-05,
+      "loss": 0.9322,
+      "step": 299630
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.0660824978954616e-05,
+      "loss": 0.9759,
+      "step": 299640
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.065967704905487e-05,
+      "loss": 0.9636,
+      "step": 299650
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.0658529119155116e-05,
+      "loss": 1.0595,
+      "step": 299660
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.065738118925537e-05,
+      "loss": 0.9618,
+      "step": 299670
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.065623325935563e-05,
+      "loss": 0.9481,
+      "step": 299680
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.065508532945587e-05,
+      "loss": 0.8398,
+      "step": 299690
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.065393739955613e-05,
+      "loss": 1.1134,
+      "step": 299700
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.0652789469656386e-05,
+      "loss": 0.931,
+      "step": 299710
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.0651641539756636e-05,
+      "loss": 0.9782,
+      "step": 299720
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.0650493609856885e-05,
+      "loss": 0.8721,
+      "step": 299730
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.064934567995714e-05,
+      "loss": 0.9683,
+      "step": 299740
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.064819775005739e-05,
+      "loss": 0.9901,
+      "step": 299750
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.064704982015765e-05,
+      "loss": 1.0269,
+      "step": 299760
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.06459018902579e-05,
+      "loss": 0.9453,
+      "step": 299770
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.064475396035815e-05,
+      "loss": 0.9479,
+      "step": 299780
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.0643606030458405e-05,
+      "loss": 1.0163,
+      "step": 299790
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.064245810055866e-05,
+      "loss": 0.9936,
+      "step": 299800
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.0641310170658905e-05,
+      "loss": 0.9615,
+      "step": 299810
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.064016224075916e-05,
+      "loss": 0.9046,
+      "step": 299820
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.063901431085942e-05,
+      "loss": 1.047,
+      "step": 299830
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.063786638095966e-05,
+      "loss": 0.9595,
+      "step": 299840
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.063671845105992e-05,
+      "loss": 0.928,
+      "step": 299850
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.0635570521160174e-05,
+      "loss": 0.9939,
+      "step": 299860
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.0634422591260424e-05,
+      "loss": 0.9593,
+      "step": 299870
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.0633274661360674e-05,
+      "loss": 0.9892,
+      "step": 299880
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.063212673146093e-05,
+      "loss": 0.9598,
+      "step": 299890
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.063097880156118e-05,
+      "loss": 0.9728,
+      "step": 299900
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.062983087166144e-05,
+      "loss": 0.9487,
+      "step": 299910
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.062868294176169e-05,
+      "loss": 1.0035,
+      "step": 299920
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.062753501186194e-05,
+      "loss": 1.0077,
+      "step": 299930
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.0626387081962193e-05,
+      "loss": 0.9558,
+      "step": 299940
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.062523915206245e-05,
+      "loss": 0.9529,
+      "step": 299950
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.062409122216269e-05,
+      "loss": 0.9657,
+      "step": 299960
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.062294329226295e-05,
+      "loss": 0.9936,
+      "step": 299970
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.0621795362363206e-05,
+      "loss": 1.0271,
+      "step": 299980
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.062064743246345e-05,
+      "loss": 0.9569,
+      "step": 299990
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.0619499502563706e-05,
+      "loss": 1.0359,
+      "step": 300000
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.061835157266396e-05,
+      "loss": 1.0199,
+      "step": 300010
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.061720364276421e-05,
+      "loss": 0.9709,
+      "step": 300020
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.061605571286446e-05,
+      "loss": 0.9872,
+      "step": 300030
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.061490778296472e-05,
+      "loss": 0.9475,
+      "step": 300040
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.061375985306497e-05,
+      "loss": 1.0157,
+      "step": 300050
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.0612611923165226e-05,
+      "loss": 0.9378,
+      "step": 300060
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.0611463993265475e-05,
+      "loss": 1.0351,
+      "step": 300070
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.0610316063365725e-05,
+      "loss": 0.9237,
+      "step": 300080
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.060916813346598e-05,
+      "loss": 0.9706,
+      "step": 300090
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.060802020356624e-05,
+      "loss": 1.0041,
+      "step": 300100
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.060687227366648e-05,
+      "loss": 0.9659,
+      "step": 300110
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.060572434376674e-05,
+      "loss": 1.0102,
+      "step": 300120
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.0604576413866995e-05,
+      "loss": 0.9056,
+      "step": 300130
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.060342848396724e-05,
+      "loss": 1.0209,
+      "step": 300140
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.0602280554067495e-05,
+      "loss": 0.9769,
+      "step": 300150
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.060113262416775e-05,
+      "loss": 1.0818,
+      "step": 300160
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.0599984694268e-05,
+      "loss": 0.9541,
+      "step": 300170
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.059883676436825e-05,
+      "loss": 1.0269,
+      "step": 300180
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.059768883446851e-05,
+      "loss": 0.9423,
+      "step": 300190
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.059654090456876e-05,
+      "loss": 0.998,
+      "step": 300200
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.0595392974669014e-05,
+      "loss": 1.092,
+      "step": 300210
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.0594245044769264e-05,
+      "loss": 0.9223,
+      "step": 300220
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.0593097114869514e-05,
+      "loss": 0.9396,
+      "step": 300230
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.059194918496977e-05,
+      "loss": 0.9014,
+      "step": 300240
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.059080125507003e-05,
+      "loss": 0.9842,
+      "step": 300250
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.058965332517027e-05,
+      "loss": 0.9588,
+      "step": 300260
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.058850539527053e-05,
+      "loss": 1.0456,
+      "step": 300270
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.058735746537078e-05,
+      "loss": 0.9871,
+      "step": 300280
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.0586209535471026e-05,
+      "loss": 0.9974,
+      "step": 300290
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.058506160557128e-05,
+      "loss": 0.9652,
+      "step": 300300
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.058391367567154e-05,
+      "loss": 0.9992,
+      "step": 300310
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.058276574577179e-05,
+      "loss": 0.99,
+      "step": 300320
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.058161781587204e-05,
+      "loss": 0.9705,
+      "step": 300330
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.0580469885972296e-05,
+      "loss": 0.984,
+      "step": 300340
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.0579321956072546e-05,
+      "loss": 1.0049,
+      "step": 300350
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.05781740261728e-05,
+      "loss": 0.9794,
+      "step": 300360
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.057702609627305e-05,
+      "loss": 0.9979,
+      "step": 300370
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.05758781663733e-05,
+      "loss": 0.8606,
+      "step": 300380
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.057473023647356e-05,
+      "loss": 0.9957,
+      "step": 300390
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.0573582306573815e-05,
+      "loss": 0.9755,
+      "step": 300400
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.057243437667406e-05,
+      "loss": 0.9983,
+      "step": 300410
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.0571286446774315e-05,
+      "loss": 0.9,
+      "step": 300420
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.0570138516874565e-05,
+      "loss": 0.9028,
+      "step": 300430
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.0568990586974815e-05,
+      "loss": 0.971,
+      "step": 300440
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.056784265707507e-05,
+      "loss": 1.0127,
+      "step": 300450
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.056669472717532e-05,
+      "loss": 0.9374,
+      "step": 300460
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.056554679727558e-05,
+      "loss": 0.9986,
+      "step": 300470
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.056439886737583e-05,
+      "loss": 0.9883,
+      "step": 300480
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.056325093747608e-05,
+      "loss": 0.9946,
+      "step": 300490
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.0562103007576334e-05,
+      "loss": 0.935,
+      "step": 300500
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.056095507767659e-05,
+      "loss": 0.9846,
+      "step": 300510
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.0559807147776834e-05,
+      "loss": 0.9334,
+      "step": 300520
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.055865921787709e-05,
+      "loss": 1.0369,
+      "step": 300530
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.055751128797735e-05,
+      "loss": 0.8959,
+      "step": 300540
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.055636335807759e-05,
+      "loss": 0.987,
+      "step": 300550
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.055521542817785e-05,
+      "loss": 1.012,
+      "step": 300560
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.0554067498278104e-05,
+      "loss": 1.0322,
+      "step": 300570
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.0552919568378353e-05,
+      "loss": 0.9249,
+      "step": 300580
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.05517716384786e-05,
+      "loss": 1.0746,
+      "step": 300590
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.055062370857886e-05,
+      "loss": 1.001,
+      "step": 300600
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.054947577867911e-05,
+      "loss": 0.9663,
+      "step": 300610
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.0548327848779366e-05,
+      "loss": 1.0054,
+      "step": 300620
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.0547179918879616e-05,
+      "loss": 1.0242,
+      "step": 300630
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.0546031988979866e-05,
+      "loss": 1.0088,
+      "step": 300640
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.054488405908012e-05,
+      "loss": 1.0028,
+      "step": 300650
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.054373612918038e-05,
+      "loss": 0.9789,
+      "step": 300660
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.054258819928062e-05,
+      "loss": 1.0727,
+      "step": 300670
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.054144026938088e-05,
+      "loss": 0.9915,
+      "step": 300680
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.0540292339481136e-05,
+      "loss": 0.8783,
+      "step": 300690
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.053914440958138e-05,
+      "loss": 1.0515,
+      "step": 300700
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.0537996479681635e-05,
+      "loss": 0.9663,
+      "step": 300710
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.053684854978189e-05,
+      "loss": 1.0315,
+      "step": 300720
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.053570061988214e-05,
+      "loss": 1.0118,
+      "step": 300730
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.053455268998239e-05,
+      "loss": 0.9734,
+      "step": 300740
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.053340476008265e-05,
+      "loss": 0.9667,
+      "step": 300750
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.05322568301829e-05,
+      "loss": 0.9381,
+      "step": 300760
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.0531108900283155e-05,
+      "loss": 1.0089,
+      "step": 300770
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.0529960970383405e-05,
+      "loss": 0.9993,
+      "step": 300780
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.0528813040483655e-05,
+      "loss": 0.9609,
+      "step": 300790
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.052766511058391e-05,
+      "loss": 0.9391,
+      "step": 300800
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.052651718068417e-05,
+      "loss": 0.844,
+      "step": 300810
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.052536925078441e-05,
+      "loss": 1.0165,
+      "step": 300820
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.052422132088467e-05,
+      "loss": 0.9919,
+      "step": 300830
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.0523073390984924e-05,
+      "loss": 0.9666,
+      "step": 300840
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.052192546108517e-05,
+      "loss": 0.9513,
+      "step": 300850
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.0520777531185424e-05,
+      "loss": 1.0268,
+      "step": 300860
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.051962960128568e-05,
+      "loss": 1.0043,
+      "step": 300870
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.051848167138593e-05,
+      "loss": 0.9003,
+      "step": 300880
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.051733374148618e-05,
+      "loss": 1.0116,
+      "step": 300890
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.051618581158644e-05,
+      "loss": 1.0029,
+      "step": 300900
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.051503788168669e-05,
+      "loss": 1.0829,
+      "step": 300910
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.0513889951786943e-05,
+      "loss": 1.0857,
+      "step": 300920
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.051274202188719e-05,
+      "loss": 0.852,
+      "step": 300930
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.051159409198744e-05,
+      "loss": 0.961,
+      "step": 300940
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.05104461620877e-05,
+      "loss": 0.8994,
+      "step": 300950
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.0509298232187956e-05,
+      "loss": 0.9077,
+      "step": 300960
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.05081503022882e-05,
+      "loss": 0.8578,
+      "step": 300970
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.0507002372388456e-05,
+      "loss": 1.025,
+      "step": 300980
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.050585444248871e-05,
+      "loss": 0.9876,
+      "step": 300990
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.0504706512588956e-05,
+      "loss": 0.9786,
+      "step": 301000
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.050355858268921e-05,
+      "loss": 0.8635,
+      "step": 301010
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.050241065278947e-05,
+      "loss": 0.9347,
+      "step": 301020
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.050126272288972e-05,
+      "loss": 1.002,
+      "step": 301030
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.050011479298997e-05,
+      "loss": 0.941,
+      "step": 301040
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.0498966863090225e-05,
+      "loss": 0.9905,
+      "step": 301050
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.0497818933190475e-05,
+      "loss": 0.9662,
+      "step": 301060
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.049667100329073e-05,
+      "loss": 1.0788,
+      "step": 301070
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.049552307339098e-05,
+      "loss": 0.9959,
+      "step": 301080
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.049437514349123e-05,
+      "loss": 0.9585,
+      "step": 301090
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.049322721359149e-05,
+      "loss": 0.926,
+      "step": 301100
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.0492079283691745e-05,
+      "loss": 0.9914,
+      "step": 301110
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.049093135379199e-05,
+      "loss": 0.963,
+      "step": 301120
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.0489783423892245e-05,
+      "loss": 0.9569,
+      "step": 301130
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.04886354939925e-05,
+      "loss": 0.9733,
+      "step": 301140
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.0487487564092744e-05,
+      "loss": 1.0889,
+      "step": 301150
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.0486339634193e-05,
+      "loss": 1.0254,
+      "step": 301160
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.048519170429326e-05,
+      "loss": 1.0162,
+      "step": 301170
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.048404377439351e-05,
+      "loss": 0.9139,
+      "step": 301180
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.048289584449376e-05,
+      "loss": 0.9647,
+      "step": 301190
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.0481747914594014e-05,
+      "loss": 0.951,
+      "step": 301200
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.0480599984694264e-05,
+      "loss": 0.9948,
+      "step": 301210
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.047945205479452e-05,
+      "loss": 1.0126,
+      "step": 301220
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.047830412489477e-05,
+      "loss": 0.9169,
+      "step": 301230
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.047715619499502e-05,
+      "loss": 0.9663,
+      "step": 301240
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.047600826509528e-05,
+      "loss": 0.9775,
+      "step": 301250
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.047486033519553e-05,
+      "loss": 0.8536,
+      "step": 301260
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.0473712405295776e-05,
+      "loss": 1.0965,
+      "step": 301270
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.047256447539603e-05,
+      "loss": 0.9362,
+      "step": 301280
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.047141654549629e-05,
+      "loss": 0.9652,
+      "step": 301290
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.047026861559653e-05,
+      "loss": 1.0292,
+      "step": 301300
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.046912068569679e-05,
+      "loss": 0.996,
+      "step": 301310
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.0467972755797046e-05,
+      "loss": 1.0765,
+      "step": 301320
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.0466824825897296e-05,
+      "loss": 0.9946,
+      "step": 301330
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.0465676895997546e-05,
+      "loss": 1.0207,
+      "step": 301340
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.04645289660978e-05,
+      "loss": 1.0143,
+      "step": 301350
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.046338103619805e-05,
+      "loss": 1.0871,
+      "step": 301360
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.046223310629831e-05,
+      "loss": 1.0671,
+      "step": 301370
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.046108517639856e-05,
+      "loss": 0.977,
+      "step": 301380
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.045993724649881e-05,
+      "loss": 1.0157,
+      "step": 301390
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.0458789316599065e-05,
+      "loss": 0.8966,
+      "step": 301400
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.045764138669932e-05,
+      "loss": 1.0033,
+      "step": 301410
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.0456493456799565e-05,
+      "loss": 1.0288,
+      "step": 301420
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.045534552689982e-05,
+      "loss": 1.0032,
+      "step": 301430
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.045419759700008e-05,
+      "loss": 0.9098,
+      "step": 301440
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.045304966710032e-05,
+      "loss": 0.9558,
+      "step": 301450
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.045190173720058e-05,
+      "loss": 1.017,
+      "step": 301460
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.0450753807300834e-05,
+      "loss": 0.9982,
+      "step": 301470
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.0449605877401084e-05,
+      "loss": 1.0474,
+      "step": 301480
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.0448457947501334e-05,
+      "loss": 1.0122,
+      "step": 301490
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.044731001760159e-05,
+      "loss": 1.0191,
+      "step": 301500
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.044616208770184e-05,
+      "loss": 1.0388,
+      "step": 301510
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.04450141578021e-05,
+      "loss": 0.92,
+      "step": 301520
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.044386622790235e-05,
+      "loss": 0.9788,
+      "step": 301530
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.04427182980026e-05,
+      "loss": 0.9729,
+      "step": 301540
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.0441570368102854e-05,
+      "loss": 0.9733,
+      "step": 301550
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.044042243820311e-05,
+      "loss": 1.0307,
+      "step": 301560
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.043927450830335e-05,
+      "loss": 1.0373,
+      "step": 301570
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.043812657840361e-05,
+      "loss": 0.9193,
+      "step": 301580
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.0436978648503867e-05,
+      "loss": 0.8627,
+      "step": 301590
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.043583071860411e-05,
+      "loss": 0.9087,
+      "step": 301600
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.0434682788704366e-05,
+      "loss": 0.9341,
+      "step": 301610
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.043353485880462e-05,
+      "loss": 1.1424,
+      "step": 301620
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.043238692890487e-05,
+      "loss": 1.0614,
+      "step": 301630
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.043123899900512e-05,
+      "loss": 0.9871,
+      "step": 301640
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.043009106910538e-05,
+      "loss": 0.8816,
+      "step": 301650
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.042894313920563e-05,
+      "loss": 1.079,
+      "step": 301660
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.0427795209305886e-05,
+      "loss": 0.9755,
+      "step": 301670
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.0426647279406136e-05,
+      "loss": 1.0177,
+      "step": 301680
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.0425499349506385e-05,
+      "loss": 0.9125,
+      "step": 301690
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.042435141960664e-05,
+      "loss": 1.0482,
+      "step": 301700
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.04232034897069e-05,
+      "loss": 0.963,
+      "step": 301710
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.042205555980714e-05,
+      "loss": 1.0339,
+      "step": 301720
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.04209076299074e-05,
+      "loss": 1.046,
+      "step": 301730
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.0419759700007655e-05,
+      "loss": 1.0314,
+      "step": 301740
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.04186117701079e-05,
+      "loss": 1.0089,
+      "step": 301750
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.0417463840208155e-05,
+      "loss": 0.9544,
+      "step": 301760
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.041631591030841e-05,
+      "loss": 1.0492,
+      "step": 301770
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.041516798040866e-05,
+      "loss": 0.9903,
+      "step": 301780
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.041402005050891e-05,
+      "loss": 0.9645,
+      "step": 301790
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.041287212060916e-05,
+      "loss": 1.0977,
+      "step": 301800
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.041172419070942e-05,
+      "loss": 0.9284,
+      "step": 301810
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.0410576260809674e-05,
+      "loss": 0.9592,
+      "step": 301820
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.040942833090992e-05,
+      "loss": 0.9311,
+      "step": 301830
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.0408280401010174e-05,
+      "loss": 0.9657,
+      "step": 301840
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.040713247111043e-05,
+      "loss": 0.9238,
+      "step": 301850
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.0405984541210674e-05,
+      "loss": 0.9286,
+      "step": 301860
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.040483661131093e-05,
+      "loss": 0.9812,
+      "step": 301870
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.040368868141119e-05,
+      "loss": 0.9914,
+      "step": 301880
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.040254075151144e-05,
+      "loss": 0.9515,
+      "step": 301890
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.040139282161169e-05,
+      "loss": 1.0571,
+      "step": 301900
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.040024489171194e-05,
+      "loss": 0.9744,
+      "step": 301910
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.039909696181219e-05,
+      "loss": 0.9854,
+      "step": 301920
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.039794903191245e-05,
+      "loss": 1.0056,
+      "step": 301930
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.03968011020127e-05,
+      "loss": 1.0049,
+      "step": 301940
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.039565317211295e-05,
+      "loss": 0.9938,
+      "step": 301950
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.0394505242213206e-05,
+      "loss": 0.9801,
+      "step": 301960
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.039335731231346e-05,
+      "loss": 0.9717,
+      "step": 301970
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.0392209382413706e-05,
+      "loss": 1.0162,
+      "step": 301980
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.039106145251396e-05,
+      "loss": 0.8988,
+      "step": 301990
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.038991352261422e-05,
+      "loss": 0.9967,
+      "step": 302000
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.038876559271446e-05,
+      "loss": 0.9983,
+      "step": 302010
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.038761766281472e-05,
+      "loss": 1.0467,
+      "step": 302020
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.0386469732914975e-05,
+      "loss": 1.0124,
+      "step": 302030
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.0385321803015225e-05,
+      "loss": 0.9011,
+      "step": 302040
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.0384173873115475e-05,
+      "loss": 0.9868,
+      "step": 302050
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.038302594321573e-05,
+      "loss": 1.0319,
+      "step": 302060
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.038187801331598e-05,
+      "loss": 1.1278,
+      "step": 302070
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.038073008341624e-05,
+      "loss": 1.0698,
+      "step": 302080
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.037958215351649e-05,
+      "loss": 0.9203,
+      "step": 302090
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.037843422361674e-05,
+      "loss": 0.9322,
+      "step": 302100
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.0377286293716995e-05,
+      "loss": 0.9101,
+      "step": 302110
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.037613836381725e-05,
+      "loss": 0.9774,
+      "step": 302120
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.0374990433917494e-05,
+      "loss": 0.9481,
+      "step": 302130
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.037384250401775e-05,
+      "loss": 0.9863,
+      "step": 302140
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.037269457411801e-05,
+      "loss": 1.0327,
+      "step": 302150
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.037154664421825e-05,
+      "loss": 0.9925,
+      "step": 302160
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.037039871431851e-05,
+      "loss": 0.953,
+      "step": 302170
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.0369250784418764e-05,
+      "loss": 1.0402,
+      "step": 302180
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.0368102854519014e-05,
+      "loss": 0.9273,
+      "step": 302190
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.0366954924619264e-05,
+      "loss": 0.9399,
+      "step": 302200
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.036580699471952e-05,
+      "loss": 0.916,
+      "step": 302210
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.036465906481977e-05,
+      "loss": 1.0246,
+      "step": 302220
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.036351113492003e-05,
+      "loss": 1.0264,
+      "step": 302230
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.0362363205020277e-05,
+      "loss": 0.9439,
+      "step": 302240
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.0361215275120526e-05,
+      "loss": 0.9497,
+      "step": 302250
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.036006734522078e-05,
+      "loss": 1.0126,
+      "step": 302260
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.035891941532104e-05,
+      "loss": 1.0011,
+      "step": 302270
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.035777148542128e-05,
+      "loss": 0.984,
+      "step": 302280
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.035662355552154e-05,
+      "loss": 0.9651,
+      "step": 302290
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.0355475625621796e-05,
+      "loss": 0.991,
+      "step": 302300
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.035432769572204e-05,
+      "loss": 1.0542,
+      "step": 302310
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.0353179765822296e-05,
+      "loss": 1.1081,
+      "step": 302320
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.035203183592255e-05,
+      "loss": 1.0064,
+      "step": 302330
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.03508839060228e-05,
+      "loss": 1.0017,
+      "step": 302340
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.034973597612305e-05,
+      "loss": 0.9333,
+      "step": 302350
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.034858804622331e-05,
+      "loss": 0.9645,
+      "step": 302360
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.034744011632356e-05,
+      "loss": 0.9038,
+      "step": 302370
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.0346292186423815e-05,
+      "loss": 0.9348,
+      "step": 302380
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.0345144256524065e-05,
+      "loss": 1.0074,
+      "step": 302390
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.0343996326624315e-05,
+      "loss": 0.9419,
+      "step": 302400
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.034284839672457e-05,
+      "loss": 1.0001,
+      "step": 302410
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.034170046682483e-05,
+      "loss": 0.9762,
+      "step": 302420
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.034055253692507e-05,
+      "loss": 1.012,
+      "step": 302430
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.033940460702533e-05,
+      "loss": 1.0463,
+      "step": 302440
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.0338256677125584e-05,
+      "loss": 0.985,
+      "step": 302450
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.033710874722583e-05,
+      "loss": 1.0335,
+      "step": 302460
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.0335960817326084e-05,
+      "loss": 1.0885,
+      "step": 302470
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.033481288742634e-05,
+      "loss": 0.9991,
+      "step": 302480
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.033366495752659e-05,
+      "loss": 0.9961,
+      "step": 302490
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.033251702762684e-05,
+      "loss": 0.9539,
+      "step": 302500
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.03313690977271e-05,
+      "loss": 1.0252,
+      "step": 302510
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.033022116782735e-05,
+      "loss": 1.0765,
+      "step": 302520
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.0329073237927604e-05,
+      "loss": 1.0883,
+      "step": 302530
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.0327925308027853e-05,
+      "loss": 1.0131,
+      "step": 302540
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.03267773781281e-05,
+      "loss": 1.1048,
+      "step": 302550
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.032562944822836e-05,
+      "loss": 0.9588,
+      "step": 302560
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.0324481518328617e-05,
+      "loss": 1.1233,
+      "step": 302570
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.032333358842886e-05,
+      "loss": 0.9769,
+      "step": 302580
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.0322185658529116e-05,
+      "loss": 0.9739,
+      "step": 302590
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.032103772862937e-05,
+      "loss": 1.0252,
+      "step": 302600
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.0319889798729616e-05,
+      "loss": 1.0206,
+      "step": 302610
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.031874186882987e-05,
+      "loss": 0.9404,
+      "step": 302620
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.031759393893013e-05,
+      "loss": 0.9519,
+      "step": 302630
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.031644600903038e-05,
+      "loss": 0.9697,
+      "step": 302640
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.031529807913063e-05,
+      "loss": 0.9888,
+      "step": 302650
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.0314150149230886e-05,
+      "loss": 1.0107,
+      "step": 302660
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.0313002219331135e-05,
+      "loss": 1.0274,
+      "step": 302670
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.031185428943139e-05,
+      "loss": 0.9573,
+      "step": 302680
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.031070635953164e-05,
+      "loss": 1.0839,
+      "step": 302690
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.030955842963189e-05,
+      "loss": 0.9663,
+      "step": 302700
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.030841049973215e-05,
+      "loss": 0.9275,
+      "step": 302710
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.0307262569832405e-05,
+      "loss": 0.9592,
+      "step": 302720
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.030611463993265e-05,
+      "loss": 0.9224,
+      "step": 302730
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.0304966710032905e-05,
+      "loss": 1.0857,
+      "step": 302740
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.030381878013316e-05,
+      "loss": 1.0466,
+      "step": 302750
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.0302670850233404e-05,
+      "loss": 1.0339,
+      "step": 302760
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.030152292033366e-05,
+      "loss": 0.9927,
+      "step": 302770
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.030037499043392e-05,
+      "loss": 0.8763,
+      "step": 302780
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.029922706053417e-05,
+      "loss": 1.0321,
+      "step": 302790
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.029807913063442e-05,
+      "loss": 0.9678,
+      "step": 302800
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.0296931200734674e-05,
+      "loss": 1.1053,
+      "step": 302810
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.0295783270834924e-05,
+      "loss": 0.9284,
+      "step": 302820
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.029463534093518e-05,
+      "loss": 1.044,
+      "step": 302830
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.029348741103543e-05,
+      "loss": 1.0376,
+      "step": 302840
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.029233948113568e-05,
+      "loss": 1.0222,
+      "step": 302850
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.029119155123594e-05,
+      "loss": 0.9785,
+      "step": 302860
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.0290043621336194e-05,
+      "loss": 1.0305,
+      "step": 302870
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.028889569143644e-05,
+      "loss": 1.0153,
+      "step": 302880
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.028774776153669e-05,
+      "loss": 0.9441,
+      "step": 302890
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.028659983163695e-05,
+      "loss": 0.9935,
+      "step": 302900
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.028545190173719e-05,
+      "loss": 0.9782,
+      "step": 302910
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.028430397183745e-05,
+      "loss": 0.927,
+      "step": 302920
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.0283156041937706e-05,
+      "loss": 0.8859,
+      "step": 302930
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.0282008112037956e-05,
+      "loss": 0.9738,
+      "step": 302940
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.0280860182138206e-05,
+      "loss": 0.9581,
+      "step": 302950
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.027971225223846e-05,
+      "loss": 1.0765,
+      "step": 302960
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.027856432233871e-05,
+      "loss": 1.026,
+      "step": 302970
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.027741639243897e-05,
+      "loss": 1.0798,
+      "step": 302980
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.027626846253922e-05,
+      "loss": 1.025,
+      "step": 302990
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.027512053263947e-05,
+      "loss": 1.0204,
+      "step": 303000
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.0273972602739725e-05,
+      "loss": 1.0106,
+      "step": 303010
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.027282467283998e-05,
+      "loss": 1.018,
+      "step": 303020
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.0271676742940225e-05,
+      "loss": 1.0034,
+      "step": 303030
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.027052881304048e-05,
+      "loss": 0.9382,
+      "step": 303040
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.026938088314074e-05,
+      "loss": 0.8825,
+      "step": 303050
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.026823295324098e-05,
+      "loss": 1.0446,
+      "step": 303060
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.026708502334124e-05,
+      "loss": 0.9501,
+      "step": 303070
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.0265937093441495e-05,
+      "loss": 0.9876,
+      "step": 303080
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.0264789163541745e-05,
+      "loss": 0.9938,
+      "step": 303090
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.0263641233641994e-05,
+      "loss": 0.9453,
+      "step": 303100
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.026249330374225e-05,
+      "loss": 0.9731,
+      "step": 303110
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.02613453738425e-05,
+      "loss": 1.0557,
+      "step": 303120
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.026019744394276e-05,
+      "loss": 0.9434,
+      "step": 303130
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.025904951404301e-05,
+      "loss": 0.9872,
+      "step": 303140
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.025790158414326e-05,
+      "loss": 0.9968,
+      "step": 303150
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.0256753654243514e-05,
+      "loss": 1.0183,
+      "step": 303160
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.025560572434376e-05,
+      "loss": 0.9954,
+      "step": 303170
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.0254457794444014e-05,
+      "loss": 1.0006,
+      "step": 303180
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.025330986454427e-05,
+      "loss": 1.0382,
+      "step": 303190
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.025216193464452e-05,
+      "loss": 1.0001,
+      "step": 303200
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.025101400474477e-05,
+      "loss": 0.968,
+      "step": 303210
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.0249866074845027e-05,
+      "loss": 0.9939,
+      "step": 303220
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.0248718144945276e-05,
+      "loss": 0.9784,
+      "step": 303230
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.024757021504553e-05,
+      "loss": 0.9899,
+      "step": 303240
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.024642228514578e-05,
+      "loss": 0.9694,
+      "step": 303250
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.024527435524603e-05,
+      "loss": 1.0613,
+      "step": 303260
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.024412642534629e-05,
+      "loss": 0.9488,
+      "step": 303270
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.0242978495446546e-05,
+      "loss": 0.9421,
+      "step": 303280
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.024183056554679e-05,
+      "loss": 1.0055,
+      "step": 303290
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.0240682635647046e-05,
+      "loss": 1.0052,
+      "step": 303300
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.02395347057473e-05,
+      "loss": 0.9703,
+      "step": 303310
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.0238386775847545e-05,
+      "loss": 0.9983,
+      "step": 303320
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.02372388459478e-05,
+      "loss": 0.991,
+      "step": 303330
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.023609091604806e-05,
+      "loss": 1.0593,
+      "step": 303340
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.023494298614831e-05,
+      "loss": 1.0442,
+      "step": 303350
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.023379505624856e-05,
+      "loss": 0.9855,
+      "step": 303360
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.0232647126348815e-05,
+      "loss": 1.059,
+      "step": 303370
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.0231499196449065e-05,
+      "loss": 1.0435,
+      "step": 303380
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.023035126654932e-05,
+      "loss": 0.9107,
+      "step": 303390
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.022920333664957e-05,
+      "loss": 0.9971,
+      "step": 303400
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.022805540674982e-05,
+      "loss": 0.8963,
+      "step": 303410
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.022690747685008e-05,
+      "loss": 0.9448,
+      "step": 303420
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.0225759546950334e-05,
+      "loss": 0.9085,
+      "step": 303430
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.022461161705058e-05,
+      "loss": 0.9224,
+      "step": 303440
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.0223463687150834e-05,
+      "loss": 1.0006,
+      "step": 303450
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.022231575725109e-05,
+      "loss": 1.0474,
+      "step": 303460
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.0221167827351334e-05,
+      "loss": 1.1634,
+      "step": 303470
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.022001989745159e-05,
+      "loss": 0.9882,
+      "step": 303480
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.021887196755185e-05,
+      "loss": 0.9553,
+      "step": 303490
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.02177240376521e-05,
+      "loss": 1.0724,
+      "step": 303500
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.021657610775235e-05,
+      "loss": 0.8638,
+      "step": 303510
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.0215428177852603e-05,
+      "loss": 0.9484,
+      "step": 303520
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.021428024795285e-05,
+      "loss": 1.0688,
+      "step": 303530
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.021313231805311e-05,
+      "loss": 1.0228,
+      "step": 303540
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.021198438815336e-05,
+      "loss": 0.9787,
+      "step": 303550
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.021083645825361e-05,
+      "loss": 0.9783,
+      "step": 303560
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.0209688528353866e-05,
+      "loss": 1.0308,
+      "step": 303570
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.020854059845412e-05,
+      "loss": 0.9835,
+      "step": 303580
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.0207392668554366e-05,
+      "loss": 0.9753,
+      "step": 303590
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.020624473865462e-05,
+      "loss": 1.1031,
+      "step": 303600
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.020509680875488e-05,
+      "loss": 1.0793,
+      "step": 303610
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.020394887885512e-05,
+      "loss": 1.0132,
+      "step": 303620
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.020280094895538e-05,
+      "loss": 1.0012,
+      "step": 303630
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.0201653019055636e-05,
+      "loss": 1.0233,
+      "step": 303640
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.0200505089155885e-05,
+      "loss": 0.9981,
+      "step": 303650
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.0199357159256135e-05,
+      "loss": 1.0581,
+      "step": 303660
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.019820922935639e-05,
+      "loss": 0.9596,
+      "step": 303670
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.019706129945664e-05,
+      "loss": 1.1044,
+      "step": 303680
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.01959133695569e-05,
+      "loss": 1.0201,
+      "step": 303690
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.019476543965715e-05,
+      "loss": 0.8911,
+      "step": 303700
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.01936175097574e-05,
+      "loss": 0.9275,
+      "step": 303710
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.0192469579857655e-05,
+      "loss": 1.0234,
+      "step": 303720
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.019132164995791e-05,
+      "loss": 0.8826,
+      "step": 303730
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.0190173720058154e-05,
+      "loss": 1.0245,
+      "step": 303740
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.018902579015841e-05,
+      "loss": 1.0557,
+      "step": 303750
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.018787786025867e-05,
+      "loss": 0.988,
+      "step": 303760
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.018672993035891e-05,
+      "loss": 1.0601,
+      "step": 303770
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.018558200045917e-05,
+      "loss": 1.0164,
+      "step": 303780
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.0184434070559424e-05,
+      "loss": 0.9268,
+      "step": 303790
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.0183286140659674e-05,
+      "loss": 0.8354,
+      "step": 303800
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.0182138210759924e-05,
+      "loss": 1.0246,
+      "step": 303810
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.018099028086018e-05,
+      "loss": 0.9642,
+      "step": 303820
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.017984235096043e-05,
+      "loss": 1.003,
+      "step": 303830
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.017869442106069e-05,
+      "loss": 0.9528,
+      "step": 303840
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.017754649116094e-05,
+      "loss": 0.9676,
+      "step": 303850
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.0176398561261187e-05,
+      "loss": 1.0045,
+      "step": 303860
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.017525063136144e-05,
+      "loss": 0.9734,
+      "step": 303870
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.01741027014617e-05,
+      "loss": 0.9268,
+      "step": 303880
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.017295477156194e-05,
+      "loss": 1.0637,
+      "step": 303890
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.01718068416622e-05,
+      "loss": 1.0493,
+      "step": 303900
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.0170658911762456e-05,
+      "loss": 1.0688,
+      "step": 303910
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.01695109818627e-05,
+      "loss": 0.979,
+      "step": 303920
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.0168363051962956e-05,
+      "loss": 1.0455,
+      "step": 303930
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.016721512206321e-05,
+      "loss": 0.8932,
+      "step": 303940
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.016606719216346e-05,
+      "loss": 1.0337,
+      "step": 303950
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.016491926226371e-05,
+      "loss": 1.0387,
+      "step": 303960
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.016377133236397e-05,
+      "loss": 0.9754,
+      "step": 303970
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.016262340246422e-05,
+      "loss": 1.0028,
+      "step": 303980
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.0161475472564475e-05,
+      "loss": 1.0726,
+      "step": 303990
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.0160327542664725e-05,
+      "loss": 0.952,
+      "step": 304000
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.0159179612764975e-05,
+      "loss": 0.9283,
+      "step": 304010
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.015803168286523e-05,
+      "loss": 1.0341,
+      "step": 304020
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.015688375296549e-05,
+      "loss": 0.9434,
+      "step": 304030
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.015573582306573e-05,
+      "loss": 1.0149,
+      "step": 304040
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.015458789316599e-05,
+      "loss": 0.9161,
+      "step": 304050
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.0153439963266245e-05,
+      "loss": 1.0311,
+      "step": 304060
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.015229203336649e-05,
+      "loss": 1.05,
+      "step": 304070
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.0151144103466744e-05,
+      "loss": 0.9845,
+      "step": 304080
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.0149996173567e-05,
+      "loss": 0.994,
+      "step": 304090
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.014884824366725e-05,
+      "loss": 0.9632,
+      "step": 304100
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.01477003137675e-05,
+      "loss": 0.9321,
+      "step": 304110
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.014655238386776e-05,
+      "loss": 1.0077,
+      "step": 304120
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.014540445396801e-05,
+      "loss": 1.0466,
+      "step": 304130
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.0144256524068264e-05,
+      "loss": 1.0301,
+      "step": 304140
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.0143108594168514e-05,
+      "loss": 1.0139,
+      "step": 304150
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.0141960664268764e-05,
+      "loss": 0.9894,
+      "step": 304160
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.014081273436902e-05,
+      "loss": 0.8869,
+      "step": 304170
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.013966480446928e-05,
+      "loss": 0.9023,
+      "step": 304180
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.013851687456952e-05,
+      "loss": 1.0299,
+      "step": 304190
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.0137368944669777e-05,
+      "loss": 0.9226,
+      "step": 304200
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.013622101477003e-05,
+      "loss": 0.9617,
+      "step": 304210
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.0135073084870276e-05,
+      "loss": 1.0267,
+      "step": 304220
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.013392515497053e-05,
+      "loss": 0.9833,
+      "step": 304230
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.013277722507079e-05,
+      "loss": 1.0296,
+      "step": 304240
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.013162929517104e-05,
+      "loss": 0.8743,
+      "step": 304250
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.013048136527129e-05,
+      "loss": 0.8366,
+      "step": 304260
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.0129333435371546e-05,
+      "loss": 0.9687,
+      "step": 304270
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.0128185505471796e-05,
+      "loss": 0.9012,
+      "step": 304280
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.012703757557205e-05,
+      "loss": 0.9431,
+      "step": 304290
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.01258896456723e-05,
+      "loss": 1.0242,
+      "step": 304300
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.012474171577255e-05,
+      "loss": 0.9949,
+      "step": 304310
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.012359378587281e-05,
+      "loss": 0.9474,
+      "step": 304320
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.0122445855973065e-05,
+      "loss": 0.9328,
+      "step": 304330
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.012129792607331e-05,
+      "loss": 0.9392,
+      "step": 304340
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.0120149996173565e-05,
+      "loss": 0.9462,
+      "step": 304350
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.011900206627382e-05,
+      "loss": 1.0961,
+      "step": 304360
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.0117854136374065e-05,
+      "loss": 0.882,
+      "step": 304370
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.011670620647432e-05,
+      "loss": 0.9253,
+      "step": 304380
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.011555827657458e-05,
+      "loss": 0.9643,
+      "step": 304390
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.011441034667483e-05,
+      "loss": 0.9632,
+      "step": 304400
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.011326241677508e-05,
+      "loss": 0.9132,
+      "step": 304410
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.0112114486875334e-05,
+      "loss": 1.0679,
+      "step": 304420
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.0110966556975584e-05,
+      "loss": 0.9957,
+      "step": 304430
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.010981862707584e-05,
+      "loss": 0.9762,
+      "step": 304440
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.010867069717609e-05,
+      "loss": 1.0502,
+      "step": 304450
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.010752276727634e-05,
+      "loss": 0.9318,
+      "step": 304460
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.01063748373766e-05,
+      "loss": 0.9011,
+      "step": 304470
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.0105226907476854e-05,
+      "loss": 1.0226,
+      "step": 304480
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.01040789775771e-05,
+      "loss": 0.9629,
+      "step": 304490
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.0102931047677353e-05,
+      "loss": 1.011,
+      "step": 304500
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.010178311777761e-05,
+      "loss": 1.0419,
+      "step": 304510
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.010063518787785e-05,
+      "loss": 0.9359,
+      "step": 304520
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.009948725797811e-05,
+      "loss": 1.0008,
+      "step": 304530
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.009833932807836e-05,
+      "loss": 1.0339,
+      "step": 304540
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.0097191398178616e-05,
+      "loss": 1.111,
+      "step": 304550
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.0096043468278866e-05,
+      "loss": 1.0043,
+      "step": 304560
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.0094895538379116e-05,
+      "loss": 1.03,
+      "step": 304570
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.009374760847937e-05,
+      "loss": 0.9872,
+      "step": 304580
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.009259967857963e-05,
+      "loss": 0.9948,
+      "step": 304590
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.009145174867987e-05,
+      "loss": 0.9754,
+      "step": 304600
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.009030381878013e-05,
+      "loss": 1.0451,
+      "step": 304610
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.0089155888880386e-05,
+      "loss": 0.9581,
+      "step": 304620
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.008800795898063e-05,
+      "loss": 1.0639,
+      "step": 304630
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.0086860029080885e-05,
+      "loss": 0.9942,
+      "step": 304640
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.008571209918114e-05,
+      "loss": 1.0239,
+      "step": 304650
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.008456416928139e-05,
+      "loss": 1.0777,
+      "step": 304660
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.008341623938164e-05,
+      "loss": 0.9492,
+      "step": 304670
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.00822683094819e-05,
+      "loss": 0.9821,
+      "step": 304680
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.008112037958215e-05,
+      "loss": 1.0173,
+      "step": 304690
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.0079972449682405e-05,
+      "loss": 1.0451,
+      "step": 304700
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.0078824519782655e-05,
+      "loss": 0.8948,
+      "step": 304710
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.0077676589882904e-05,
+      "loss": 1.0411,
+      "step": 304720
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.007652865998316e-05,
+      "loss": 1.101,
+      "step": 304730
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.007538073008342e-05,
+      "loss": 0.8266,
+      "step": 304740
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.007423280018366e-05,
+      "loss": 0.9783,
+      "step": 304750
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.007308487028392e-05,
+      "loss": 0.9903,
+      "step": 304760
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.0071936940384174e-05,
+      "loss": 1.0048,
+      "step": 304770
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.007078901048442e-05,
+      "loss": 0.9478,
+      "step": 304780
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.0069641080584674e-05,
+      "loss": 0.9644,
+      "step": 304790
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.006849315068493e-05,
+      "loss": 0.9992,
+      "step": 304800
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.006734522078518e-05,
+      "loss": 0.9125,
+      "step": 304810
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.006619729088543e-05,
+      "loss": 0.96,
+      "step": 304820
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.006504936098569e-05,
+      "loss": 0.9636,
+      "step": 304830
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.0063901431085937e-05,
+      "loss": 1.011,
+      "step": 304840
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.006275350118619e-05,
+      "loss": 0.9885,
+      "step": 304850
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.006160557128644e-05,
+      "loss": 0.9459,
+      "step": 304860
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.006045764138669e-05,
+      "loss": 0.9144,
+      "step": 304870
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.005930971148695e-05,
+      "loss": 0.9852,
+      "step": 304880
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.0058161781587206e-05,
+      "loss": 0.9298,
+      "step": 304890
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.005701385168745e-05,
+      "loss": 0.9658,
+      "step": 304900
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.0055865921787706e-05,
+      "loss": 0.9825,
+      "step": 304910
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.005471799188796e-05,
+      "loss": 1.0194,
+      "step": 304920
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.0053570061988206e-05,
+      "loss": 0.9857,
+      "step": 304930
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.005242213208846e-05,
+      "loss": 1.072,
+      "step": 304940
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.005127420218872e-05,
+      "loss": 0.8948,
+      "step": 304950
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.005012627228897e-05,
+      "loss": 0.9895,
+      "step": 304960
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.004897834238922e-05,
+      "loss": 1.0448,
+      "step": 304970
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.0047830412489475e-05,
+      "loss": 1.0449,
+      "step": 304980
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.0046682482589725e-05,
+      "loss": 0.9271,
+      "step": 304990
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.004553455268998e-05,
+      "loss": 0.9375,
+      "step": 305000
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.004438662279023e-05,
+      "loss": 0.9411,
+      "step": 305010
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.004323869289048e-05,
+      "loss": 1.0064,
+      "step": 305020
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.004209076299074e-05,
+      "loss": 1.0754,
+      "step": 305030
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.0040942833090995e-05,
+      "loss": 0.9279,
+      "step": 305040
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.003979490319124e-05,
+      "loss": 0.9279,
+      "step": 305050
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.0038646973291494e-05,
+      "loss": 0.9436,
+      "step": 305060
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.003749904339175e-05,
+      "loss": 0.9801,
+      "step": 305070
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.0036351113491994e-05,
+      "loss": 0.9783,
+      "step": 305080
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.003520318359225e-05,
+      "loss": 0.9897,
+      "step": 305090
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.003405525369251e-05,
+      "loss": 1.0229,
+      "step": 305100
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.003290732379276e-05,
+      "loss": 1.125,
+      "step": 305110
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.003175939389301e-05,
+      "loss": 1.0487,
+      "step": 305120
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.0030611463993264e-05,
+      "loss": 0.9393,
+      "step": 305130
+    },
+    {
+      "epoch": 14.0,
+      "eval_cer": 0.66873181492475,
+      "eval_em": 0.007454017424975799,
+      "eval_f1": 0.007454017424975799,
+      "eval_loss": 1.2341563701629639,
+      "eval_runtime": 1678.5851,
+      "eval_samples_per_second": 6.154,
+      "eval_steps_per_second": 0.77,
+      "eval_wer": 0.9925459825750242,
+      "step": 305131
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.0029463534093514e-05,
+      "loss": 0.8934,
+      "step": 305140
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.002831560419377e-05,
+      "loss": 0.9814,
+      "step": 305150
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.002716767429402e-05,
+      "loss": 0.8597,
+      "step": 305160
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.002601974439427e-05,
+      "loss": 0.9025,
+      "step": 305170
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.0024871814494527e-05,
+      "loss": 0.8966,
+      "step": 305180
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.002372388459478e-05,
+      "loss": 0.9948,
+      "step": 305190
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.0022575954695026e-05,
+      "loss": 0.933,
+      "step": 305200
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.002142802479528e-05,
+      "loss": 0.9322,
+      "step": 305210
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.002028009489554e-05,
+      "loss": 0.9005,
+      "step": 305220
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.001913216499578e-05,
+      "loss": 1.0416,
+      "step": 305230
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.001798423509604e-05,
+      "loss": 0.9586,
+      "step": 305240
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.0016836305196296e-05,
+      "loss": 0.9265,
+      "step": 305250
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.0015688375296546e-05,
+      "loss": 0.9101,
+      "step": 305260
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.0014540445396796e-05,
+      "loss": 0.923,
+      "step": 305270
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.001339251549705e-05,
+      "loss": 0.8702,
+      "step": 305280
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.00122445855973e-05,
+      "loss": 0.9367,
+      "step": 305290
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.001109665569756e-05,
+      "loss": 0.9538,
+      "step": 305300
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.000994872579781e-05,
+      "loss": 0.9345,
+      "step": 305310
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.000880079589806e-05,
+      "loss": 0.8805,
+      "step": 305320
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.0007652865998315e-05,
+      "loss": 0.9086,
+      "step": 305330
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.000650493609857e-05,
+      "loss": 0.8883,
+      "step": 305340
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.0005357006198815e-05,
+      "loss": 0.9809,
+      "step": 305350
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.000420907629907e-05,
+      "loss": 0.9604,
+      "step": 305360
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.000306114639933e-05,
+      "loss": 0.9368,
+      "step": 305370
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.000191321649957e-05,
+      "loss": 0.9309,
+      "step": 305380
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.000076528659983e-05,
+      "loss": 0.8255,
+      "step": 305390
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.9999617356700084e-05,
+      "loss": 0.9534,
+      "step": 305400
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.9998469426800334e-05,
+      "loss": 0.8667,
+      "step": 305410
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.9997321496900584e-05,
+      "loss": 0.9111,
+      "step": 305420
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.999617356700084e-05,
+      "loss": 0.9033,
+      "step": 305430
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.999502563710109e-05,
+      "loss": 0.9535,
+      "step": 305440
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.999387770720135e-05,
+      "loss": 0.9238,
+      "step": 305450
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.99927297773016e-05,
+      "loss": 0.9365,
+      "step": 305460
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.999158184740185e-05,
+      "loss": 0.9518,
+      "step": 305470
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.9990433917502103e-05,
+      "loss": 0.9805,
+      "step": 305480
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.998928598760236e-05,
+      "loss": 0.9299,
+      "step": 305490
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.99881380577026e-05,
+      "loss": 1.0272,
+      "step": 305500
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.998699012780286e-05,
+      "loss": 0.9445,
+      "step": 305510
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.9985842197903116e-05,
+      "loss": 0.9877,
+      "step": 305520
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.998469426800336e-05,
+      "loss": 0.9034,
+      "step": 305530
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.9983546338103616e-05,
+      "loss": 0.9231,
+      "step": 305540
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.998239840820387e-05,
+      "loss": 0.9818,
+      "step": 305550
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.998125047830412e-05,
+      "loss": 0.8731,
+      "step": 305560
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.998010254840437e-05,
+      "loss": 1.0062,
+      "step": 305570
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.997895461850463e-05,
+      "loss": 0.9724,
+      "step": 305580
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.997780668860488e-05,
+      "loss": 0.8554,
+      "step": 305590
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.9976658758705136e-05,
+      "loss": 0.9007,
+      "step": 305600
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.9975510828805385e-05,
+      "loss": 0.9098,
+      "step": 305610
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.9974362898905635e-05,
+      "loss": 0.932,
+      "step": 305620
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.997321496900589e-05,
+      "loss": 0.8964,
+      "step": 305630
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.997206703910615e-05,
+      "loss": 0.8866,
+      "step": 305640
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.997091910920639e-05,
+      "loss": 1.0224,
+      "step": 305650
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.996977117930665e-05,
+      "loss": 0.836,
+      "step": 305660
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.9968623249406905e-05,
+      "loss": 0.9104,
+      "step": 305670
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.996747531950715e-05,
+      "loss": 1.0174,
+      "step": 305680
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.9966327389607405e-05,
+      "loss": 0.885,
+      "step": 305690
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.996517945970766e-05,
+      "loss": 0.9028,
+      "step": 305700
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.996403152980791e-05,
+      "loss": 0.9915,
+      "step": 305710
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.996288359990816e-05,
+      "loss": 0.9803,
+      "step": 305720
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.996173567000842e-05,
+      "loss": 0.9473,
+      "step": 305730
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.996058774010867e-05,
+      "loss": 0.8725,
+      "step": 305740
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.9959439810208924e-05,
+      "loss": 0.847,
+      "step": 305750
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.9958291880309174e-05,
+      "loss": 0.9676,
+      "step": 305760
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.9957143950409424e-05,
+      "loss": 0.8943,
+      "step": 305770
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.995599602050968e-05,
+      "loss": 1.0837,
+      "step": 305780
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.995484809060994e-05,
+      "loss": 0.9207,
+      "step": 305790
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.995370016071018e-05,
+      "loss": 0.8874,
+      "step": 305800
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.995255223081044e-05,
+      "loss": 0.9582,
+      "step": 305810
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.995140430091069e-05,
+      "loss": 0.8956,
+      "step": 305820
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.9950256371010936e-05,
+      "loss": 0.9592,
+      "step": 305830
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.994910844111119e-05,
+      "loss": 0.9467,
+      "step": 305840
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.994796051121145e-05,
+      "loss": 0.8871,
+      "step": 305850
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.99468125813117e-05,
+      "loss": 0.9045,
+      "step": 305860
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.994566465141195e-05,
+      "loss": 0.9234,
+      "step": 305870
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.9944516721512206e-05,
+      "loss": 0.9201,
+      "step": 305880
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.9943368791612456e-05,
+      "loss": 0.8424,
+      "step": 305890
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.994222086171271e-05,
+      "loss": 0.9306,
+      "step": 305900
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.9941072931812956e-05,
+      "loss": 0.8414,
+      "step": 305910
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.993992500191321e-05,
+      "loss": 1.0364,
+      "step": 305920
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.993877707201347e-05,
+      "loss": 0.9226,
+      "step": 305930
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.993762914211371e-05,
+      "loss": 0.9978,
+      "step": 305940
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.993648121221397e-05,
+      "loss": 0.9718,
+      "step": 305950
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.9935333282314225e-05,
+      "loss": 0.9124,
+      "step": 305960
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.9934185352414475e-05,
+      "loss": 0.9367,
+      "step": 305970
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.9933037422514725e-05,
+      "loss": 0.9735,
+      "step": 305980
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.993188949261498e-05,
+      "loss": 0.9285,
+      "step": 305990
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.993074156271523e-05,
+      "loss": 0.9193,
+      "step": 306000
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.992959363281549e-05,
+      "loss": 0.9413,
+      "step": 306010
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.992844570291574e-05,
+      "loss": 0.8984,
+      "step": 306020
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.992729777301599e-05,
+      "loss": 1.0068,
+      "step": 306030
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.9926149843116244e-05,
+      "loss": 0.8978,
+      "step": 306040
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.99250019132165e-05,
+      "loss": 0.9178,
+      "step": 306050
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.9923853983316744e-05,
+      "loss": 0.8916,
+      "step": 306060
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.9922706053417e-05,
+      "loss": 0.9507,
+      "step": 306070
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.992155812351726e-05,
+      "loss": 0.935,
+      "step": 306080
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.99204101936175e-05,
+      "loss": 0.8405,
+      "step": 306090
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.991926226371776e-05,
+      "loss": 1.0017,
+      "step": 306100
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 3.9918114333818014e-05,
+      "loss": 0.9855,
+      "step": 306110
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.9916966403918264e-05,
+      "loss": 0.8979,
+      "step": 306120
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.9915818474018513e-05,
+      "loss": 1.0175,
+      "step": 306130
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.991467054411877e-05,
+      "loss": 0.9084,
+      "step": 306140
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.991352261421902e-05,
+      "loss": 0.8288,
+      "step": 306150
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.9912374684319277e-05,
+      "loss": 0.9184,
+      "step": 306160
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.9911226754419526e-05,
+      "loss": 0.9642,
+      "step": 306170
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.9910078824519776e-05,
+      "loss": 0.9157,
+      "step": 306180
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.990893089462003e-05,
+      "loss": 0.904,
+      "step": 306190
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.990778296472029e-05,
+      "loss": 0.9017,
+      "step": 306200
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.990663503482053e-05,
+      "loss": 1.0034,
+      "step": 306210
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.990548710492079e-05,
+      "loss": 0.9323,
+      "step": 306220
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.9904339175021046e-05,
+      "loss": 0.91,
+      "step": 306230
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.990319124512129e-05,
+      "loss": 0.8608,
+      "step": 306240
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.9902043315221546e-05,
+      "loss": 0.9542,
+      "step": 306250
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.99008953853218e-05,
+      "loss": 0.9692,
+      "step": 306260
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.989974745542205e-05,
+      "loss": 0.8874,
+      "step": 306270
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.98985995255223e-05,
+      "loss": 0.8896,
+      "step": 306280
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.989745159562256e-05,
+      "loss": 0.8611,
+      "step": 306290
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.989630366572281e-05,
+      "loss": 0.9953,
+      "step": 306300
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.9895155735823065e-05,
+      "loss": 0.9241,
+      "step": 306310
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.9894007805923315e-05,
+      "loss": 0.8884,
+      "step": 306320
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 3.9892859876023565e-05,
+      "loss": 0.8245,
+      "step": 306330
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.989171194612382e-05,
+      "loss": 0.9573,
+      "step": 306340
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.989056401622408e-05,
+      "loss": 0.9772,
+      "step": 306350
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.988941608632432e-05,
+      "loss": 0.9948,
+      "step": 306360
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.988826815642458e-05,
+      "loss": 0.7927,
+      "step": 306370
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.9887120226524834e-05,
+      "loss": 0.8539,
+      "step": 306380
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.988597229662508e-05,
+      "loss": 0.9942,
+      "step": 306390
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.9884824366725334e-05,
+      "loss": 0.8858,
+      "step": 306400
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.988367643682559e-05,
+      "loss": 1.0498,
+      "step": 306410
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.988252850692584e-05,
+      "loss": 0.8081,
+      "step": 306420
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.988138057702609e-05,
+      "loss": 0.9124,
+      "step": 306430
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.988023264712635e-05,
+      "loss": 0.8807,
+      "step": 306440
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.98790847172266e-05,
+      "loss": 0.8907,
+      "step": 306450
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.9877936787326853e-05,
+      "loss": 0.9155,
+      "step": 306460
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.98767888574271e-05,
+      "loss": 0.918,
+      "step": 306470
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.987564092752735e-05,
+      "loss": 0.8853,
+      "step": 306480
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.987449299762761e-05,
+      "loss": 0.8156,
+      "step": 306490
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.9873345067727866e-05,
+      "loss": 0.8783,
+      "step": 306500
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.987219713782811e-05,
+      "loss": 0.9555,
+      "step": 306510
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.9871049207928366e-05,
+      "loss": 0.9063,
+      "step": 306520
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.986990127802862e-05,
+      "loss": 0.9205,
+      "step": 306530
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.9868753348128866e-05,
+      "loss": 1.0906,
+      "step": 306540
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.986760541822912e-05,
+      "loss": 0.8792,
+      "step": 306550
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.986645748832938e-05,
+      "loss": 0.976,
+      "step": 306560
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.986530955842963e-05,
+      "loss": 0.9422,
+      "step": 306570
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.986416162852988e-05,
+      "loss": 0.905,
+      "step": 306580
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.9863013698630135e-05,
+      "loss": 0.9752,
+      "step": 306590
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.9861865768730385e-05,
+      "loss": 1.0408,
+      "step": 306600
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.986071783883064e-05,
+      "loss": 0.8533,
+      "step": 306610
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.985956990893089e-05,
+      "loss": 1.0977,
+      "step": 306620
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.985842197903114e-05,
+      "loss": 0.9554,
+      "step": 306630
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.98572740491314e-05,
+      "loss": 0.9471,
+      "step": 306640
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.9856126119231655e-05,
+      "loss": 0.9392,
+      "step": 306650
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.98549781893319e-05,
+      "loss": 0.8471,
+      "step": 306660
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.9853830259432155e-05,
+      "loss": 0.921,
+      "step": 306670
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.985268232953241e-05,
+      "loss": 0.9103,
+      "step": 306680
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.9851534399632654e-05,
+      "loss": 0.8671,
+      "step": 306690
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.985038646973291e-05,
+      "loss": 0.9088,
+      "step": 306700
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.984923853983317e-05,
+      "loss": 0.985,
+      "step": 306710
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.984809060993342e-05,
+      "loss": 0.9148,
+      "step": 306720
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.984694268003367e-05,
+      "loss": 0.8971,
+      "step": 306730
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.9845794750133924e-05,
+      "loss": 0.935,
+      "step": 306740
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.9844646820234174e-05,
+      "loss": 0.834,
+      "step": 306750
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.984349889033443e-05,
+      "loss": 0.9888,
+      "step": 306760
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.984235096043468e-05,
+      "loss": 1.1,
+      "step": 306770
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.984120303053493e-05,
+      "loss": 1.0319,
+      "step": 306780
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.984005510063519e-05,
+      "loss": 0.8499,
+      "step": 306790
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.983890717073544e-05,
+      "loss": 0.9852,
+      "step": 306800
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.9837759240835686e-05,
+      "loss": 0.9497,
+      "step": 306810
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.983661131093594e-05,
+      "loss": 0.8905,
+      "step": 306820
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.98354633810362e-05,
+      "loss": 0.9236,
+      "step": 306830
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.983431545113644e-05,
+      "loss": 0.9292,
+      "step": 306840
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.98331675212367e-05,
+      "loss": 0.967,
+      "step": 306850
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.9832019591336956e-05,
+      "loss": 0.8263,
+      "step": 306860
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.9830871661437206e-05,
+      "loss": 0.8758,
+      "step": 306870
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.9829723731537456e-05,
+      "loss": 0.9428,
+      "step": 306880
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.982857580163771e-05,
+      "loss": 1.0529,
+      "step": 306890
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.982742787173796e-05,
+      "loss": 0.9375,
+      "step": 306900
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.982627994183822e-05,
+      "loss": 0.934,
+      "step": 306910
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.982513201193847e-05,
+      "loss": 0.9991,
+      "step": 306920
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.982398408203872e-05,
+      "loss": 1.0149,
+      "step": 306930
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.9822836152138975e-05,
+      "loss": 0.9278,
+      "step": 306940
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.982168822223923e-05,
+      "loss": 0.9766,
+      "step": 306950
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.9820540292339475e-05,
+      "loss": 0.918,
+      "step": 306960
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.981939236243973e-05,
+      "loss": 0.9222,
+      "step": 306970
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 3.981824443253999e-05,
+      "loss": 0.8766,
+      "step": 306980
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.981709650264023e-05,
+      "loss": 0.9537,
+      "step": 306990
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.981594857274049e-05,
+      "loss": 0.9094,
+      "step": 307000
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.9814800642840745e-05,
+      "loss": 1.0274,
+      "step": 307010
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.9813652712940994e-05,
+      "loss": 1.0426,
+      "step": 307020
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.9812504783041244e-05,
+      "loss": 0.8494,
+      "step": 307030
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.98113568531415e-05,
+      "loss": 0.8399,
+      "step": 307040
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.981020892324175e-05,
+      "loss": 0.9372,
+      "step": 307050
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.980906099334201e-05,
+      "loss": 0.897,
+      "step": 307060
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.980791306344226e-05,
+      "loss": 0.9998,
+      "step": 307070
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.980676513354251e-05,
+      "loss": 0.8814,
+      "step": 307080
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.9805617203642764e-05,
+      "loss": 0.8803,
+      "step": 307090
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.980446927374302e-05,
+      "loss": 0.9125,
+      "step": 307100
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.9803321343843263e-05,
+      "loss": 0.914,
+      "step": 307110
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.980217341394352e-05,
+      "loss": 0.9744,
+      "step": 307120
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.980102548404378e-05,
+      "loss": 1.0,
+      "step": 307130
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.979987755414402e-05,
+      "loss": 0.946,
+      "step": 307140
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.9798729624244276e-05,
+      "loss": 0.9441,
+      "step": 307150
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.979758169434453e-05,
+      "loss": 0.9533,
+      "step": 307160
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.979643376444478e-05,
+      "loss": 0.9297,
+      "step": 307170
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.979528583454503e-05,
+      "loss": 0.8443,
+      "step": 307180
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.979413790464529e-05,
+      "loss": 0.8642,
+      "step": 307190
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 3.979298997474554e-05,
+      "loss": 0.9347,
+      "step": 307200
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.9791842044845796e-05,
+      "loss": 0.8883,
+      "step": 307210
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.9790694114946046e-05,
+      "loss": 0.9893,
+      "step": 307220
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.9789546185046296e-05,
+      "loss": 0.9063,
+      "step": 307230
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.978839825514655e-05,
+      "loss": 1.0505,
+      "step": 307240
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.978725032524681e-05,
+      "loss": 0.9851,
+      "step": 307250
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.978610239534705e-05,
+      "loss": 0.935,
+      "step": 307260
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.978495446544731e-05,
+      "loss": 0.8806,
+      "step": 307270
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.978380653554755e-05,
+      "loss": 0.8956,
+      "step": 307280
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.978265860564781e-05,
+      "loss": 0.9151,
+      "step": 307290
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.9781510675748065e-05,
+      "loss": 0.9893,
+      "step": 307300
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.9780362745848315e-05,
+      "loss": 0.9006,
+      "step": 307310
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.977921481594857e-05,
+      "loss": 0.9277,
+      "step": 307320
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.977806688604882e-05,
+      "loss": 1.0147,
+      "step": 307330
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.977691895614907e-05,
+      "loss": 0.8196,
+      "step": 307340
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.977577102624933e-05,
+      "loss": 0.8922,
+      "step": 307350
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.9774623096349584e-05,
+      "loss": 0.9902,
+      "step": 307360
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.977347516644983e-05,
+      "loss": 0.9235,
+      "step": 307370
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.9772327236550084e-05,
+      "loss": 0.8659,
+      "step": 307380
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.977117930665034e-05,
+      "loss": 0.9013,
+      "step": 307390
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.9770031376750584e-05,
+      "loss": 0.9091,
+      "step": 307400
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.976888344685084e-05,
+      "loss": 0.9196,
+      "step": 307410
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 3.97677355169511e-05,
+      "loss": 0.8825,
+      "step": 307420
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.976658758705134e-05,
+      "loss": 0.9563,
+      "step": 307430
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.97654396571516e-05,
+      "loss": 0.9003,
+      "step": 307440
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.976429172725185e-05,
+      "loss": 0.992,
+      "step": 307450
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.97631437973521e-05,
+      "loss": 0.959,
+      "step": 307460
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.976199586745236e-05,
+      "loss": 0.9403,
+      "step": 307470
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.976084793755261e-05,
+      "loss": 0.9191,
+      "step": 307480
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.975970000765286e-05,
+      "loss": 1.0014,
+      "step": 307490
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.9758552077753116e-05,
+      "loss": 0.8932,
+      "step": 307500
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.975740414785337e-05,
+      "loss": 0.9319,
+      "step": 307510
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.9756256217953616e-05,
+      "loss": 0.9472,
+      "step": 307520
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.975510828805387e-05,
+      "loss": 0.9034,
+      "step": 307530
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.975396035815413e-05,
+      "loss": 0.9331,
+      "step": 307540
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.975281242825437e-05,
+      "loss": 0.9286,
+      "step": 307550
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.975166449835463e-05,
+      "loss": 0.8751,
+      "step": 307560
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.9750516568454885e-05,
+      "loss": 0.8561,
+      "step": 307570
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.974936863855513e-05,
+      "loss": 0.9219,
+      "step": 307580
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.9748220708655385e-05,
+      "loss": 0.9252,
+      "step": 307590
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.974707277875564e-05,
+      "loss": 0.9274,
+      "step": 307600
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.974592484885589e-05,
+      "loss": 0.9587,
+      "step": 307610
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.974477691895615e-05,
+      "loss": 1.0254,
+      "step": 307620
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 3.97436289890564e-05,
+      "loss": 0.8974,
+      "step": 307630
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.974248105915665e-05,
+      "loss": 0.9204,
+      "step": 307640
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.9741333129256905e-05,
+      "loss": 0.9533,
+      "step": 307650
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.974018519935716e-05,
+      "loss": 0.955,
+      "step": 307660
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.9739037269457404e-05,
+      "loss": 0.8829,
+      "step": 307670
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.973788933955766e-05,
+      "loss": 0.9186,
+      "step": 307680
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.973674140965792e-05,
+      "loss": 0.9093,
+      "step": 307690
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.973559347975816e-05,
+      "loss": 0.9822,
+      "step": 307700
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.973444554985842e-05,
+      "loss": 0.8901,
+      "step": 307710
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.9733297619958674e-05,
+      "loss": 0.95,
+      "step": 307720
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.973214969005892e-05,
+      "loss": 0.9067,
+      "step": 307730
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.9731001760159174e-05,
+      "loss": 0.9622,
+      "step": 307740
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.972985383025943e-05,
+      "loss": 0.9908,
+      "step": 307750
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.972870590035968e-05,
+      "loss": 0.9278,
+      "step": 307760
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.972755797045994e-05,
+      "loss": 1.069,
+      "step": 307770
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.9726410040560187e-05,
+      "loss": 0.9623,
+      "step": 307780
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.9725262110660436e-05,
+      "loss": 0.8873,
+      "step": 307790
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.972411418076069e-05,
+      "loss": 0.9942,
+      "step": 307800
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.972296625086095e-05,
+      "loss": 0.8652,
+      "step": 307810
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.972181832096119e-05,
+      "loss": 0.8888,
+      "step": 307820
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.972067039106145e-05,
+      "loss": 0.8829,
+      "step": 307830
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.9719522461161706e-05,
+      "loss": 0.909,
+      "step": 307840
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 3.971837453126195e-05,
+      "loss": 1.0064,
+      "step": 307850
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.9717226601362206e-05,
+      "loss": 1.0432,
+      "step": 307860
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.971607867146246e-05,
+      "loss": 0.9328,
+      "step": 307870
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.9714930741562705e-05,
+      "loss": 0.8842,
+      "step": 307880
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.971378281166296e-05,
+      "loss": 0.9024,
+      "step": 307890
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.971263488176322e-05,
+      "loss": 0.8781,
+      "step": 307900
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.971148695186347e-05,
+      "loss": 0.9486,
+      "step": 307910
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.9710339021963725e-05,
+      "loss": 1.0219,
+      "step": 307920
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.9709191092063975e-05,
+      "loss": 0.9442,
+      "step": 307930
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.9708043162164225e-05,
+      "loss": 0.8212,
+      "step": 307940
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.970689523226448e-05,
+      "loss": 1.0089,
+      "step": 307950
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.970574730236474e-05,
+      "loss": 0.8891,
+      "step": 307960
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.970459937246498e-05,
+      "loss": 0.812,
+      "step": 307970
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.970345144256524e-05,
+      "loss": 0.8923,
+      "step": 307980
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.9702303512665495e-05,
+      "loss": 0.954,
+      "step": 307990
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.970115558276574e-05,
+      "loss": 0.9813,
+      "step": 308000
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.9700007652865994e-05,
+      "loss": 0.8776,
+      "step": 308010
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.969885972296625e-05,
+      "loss": 0.9661,
+      "step": 308020
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.9697711793066494e-05,
+      "loss": 1.0291,
+      "step": 308030
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.969656386316675e-05,
+      "loss": 0.8694,
+      "step": 308040
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.969541593326701e-05,
+      "loss": 0.9424,
+      "step": 308050
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.969426800336726e-05,
+      "loss": 0.8539,
+      "step": 308060
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 3.9693120073467514e-05,
+      "loss": 0.9625,
+      "step": 308070
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.9691972143567764e-05,
+      "loss": 0.9423,
+      "step": 308080
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.969082421366801e-05,
+      "loss": 0.8921,
+      "step": 308090
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.968967628376827e-05,
+      "loss": 0.9648,
+      "step": 308100
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.968852835386853e-05,
+      "loss": 0.8812,
+      "step": 308110
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.968738042396877e-05,
+      "loss": 0.9607,
+      "step": 308120
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.9686232494069026e-05,
+      "loss": 0.812,
+      "step": 308130
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.968508456416928e-05,
+      "loss": 0.9313,
+      "step": 308140
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.9683936634269526e-05,
+      "loss": 0.9483,
+      "step": 308150
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.968278870436978e-05,
+      "loss": 0.9874,
+      "step": 308160
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.968164077447004e-05,
+      "loss": 0.9293,
+      "step": 308170
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.968049284457028e-05,
+      "loss": 0.8932,
+      "step": 308180
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.967934491467054e-05,
+      "loss": 0.8864,
+      "step": 308190
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.9678196984770796e-05,
+      "loss": 0.9536,
+      "step": 308200
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.9677049054871046e-05,
+      "loss": 0.9805,
+      "step": 308210
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.96759011249713e-05,
+      "loss": 0.8602,
+      "step": 308220
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.967475319507155e-05,
+      "loss": 0.8304,
+      "step": 308230
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.96736052651718e-05,
+      "loss": 0.8987,
+      "step": 308240
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.967245733527206e-05,
+      "loss": 0.9513,
+      "step": 308250
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.9671309405372315e-05,
+      "loss": 0.8744,
+      "step": 308260
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.967016147547256e-05,
+      "loss": 0.9327,
+      "step": 308270
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.9669013545572815e-05,
+      "loss": 0.8499,
+      "step": 308280
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 3.966786561567307e-05,
+      "loss": 1.0068,
+      "step": 308290
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.9666717685773315e-05,
+      "loss": 1.0341,
+      "step": 308300
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.966556975587357e-05,
+      "loss": 0.8581,
+      "step": 308310
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.966442182597383e-05,
+      "loss": 1.0475,
+      "step": 308320
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.966327389607407e-05,
+      "loss": 0.8672,
+      "step": 308330
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.966212596617433e-05,
+      "loss": 1.0699,
+      "step": 308340
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.9660978036274584e-05,
+      "loss": 1.0262,
+      "step": 308350
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.9659830106374834e-05,
+      "loss": 0.9449,
+      "step": 308360
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.965868217647509e-05,
+      "loss": 0.9547,
+      "step": 308370
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.965753424657534e-05,
+      "loss": 0.9911,
+      "step": 308380
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.965638631667559e-05,
+      "loss": 1.0315,
+      "step": 308390
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.965523838677585e-05,
+      "loss": 1.0383,
+      "step": 308400
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.9654090456876104e-05,
+      "loss": 0.8578,
+      "step": 308410
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.965294252697635e-05,
+      "loss": 0.9267,
+      "step": 308420
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.96517945970766e-05,
+      "loss": 0.9615,
+      "step": 308430
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.965064666717686e-05,
+      "loss": 0.8658,
+      "step": 308440
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.96494987372771e-05,
+      "loss": 0.9024,
+      "step": 308450
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.964835080737736e-05,
+      "loss": 0.945,
+      "step": 308460
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.9647202877477616e-05,
+      "loss": 0.9159,
+      "step": 308470
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.964605494757786e-05,
+      "loss": 0.9787,
+      "step": 308480
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.9644907017678116e-05,
+      "loss": 0.9469,
+      "step": 308490
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 3.964375908777837e-05,
+      "loss": 1.0133,
+      "step": 308500
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.964261115787862e-05,
+      "loss": 0.9229,
+      "step": 308510
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.964146322797888e-05,
+      "loss": 0.9914,
+      "step": 308520
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.964031529807913e-05,
+      "loss": 0.9749,
+      "step": 308530
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.963916736817938e-05,
+      "loss": 0.9185,
+      "step": 308540
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.9638019438279635e-05,
+      "loss": 0.9288,
+      "step": 308550
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.963687150837989e-05,
+      "loss": 0.9055,
+      "step": 308560
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.9635723578480135e-05,
+      "loss": 0.9654,
+      "step": 308570
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.963457564858039e-05,
+      "loss": 0.9958,
+      "step": 308580
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.963342771868065e-05,
+      "loss": 0.9386,
+      "step": 308590
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.963227978878089e-05,
+      "loss": 0.9204,
+      "step": 308600
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.963113185888115e-05,
+      "loss": 0.9224,
+      "step": 308610
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.9629983928981405e-05,
+      "loss": 0.9334,
+      "step": 308620
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.962883599908165e-05,
+      "loss": 0.8925,
+      "step": 308630
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.9627688069181904e-05,
+      "loss": 0.9418,
+      "step": 308640
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.9626540139282154e-05,
+      "loss": 0.9855,
+      "step": 308650
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.962539220938241e-05,
+      "loss": 1.0076,
+      "step": 308660
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.962424427948267e-05,
+      "loss": 0.9398,
+      "step": 308670
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.962309634958291e-05,
+      "loss": 0.9811,
+      "step": 308680
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.962194841968317e-05,
+      "loss": 0.8965,
+      "step": 308690
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.9620800489783424e-05,
+      "loss": 0.9689,
+      "step": 308700
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.961965255988367e-05,
+      "loss": 0.9474,
+      "step": 308710
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 3.9618504629983924e-05,
+      "loss": 0.9168,
+      "step": 308720
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.961735670008418e-05,
+      "loss": 0.9949,
+      "step": 308730
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.961620877018442e-05,
+      "loss": 0.9112,
+      "step": 308740
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.961506084028468e-05,
+      "loss": 0.9274,
+      "step": 308750
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.9613912910384937e-05,
+      "loss": 0.9849,
+      "step": 308760
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.9612764980485186e-05,
+      "loss": 0.8901,
+      "step": 308770
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.9611617050585436e-05,
+      "loss": 0.8763,
+      "step": 308780
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.961046912068569e-05,
+      "loss": 0.8872,
+      "step": 308790
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.960932119078594e-05,
+      "loss": 0.9213,
+      "step": 308800
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.96081732608862e-05,
+      "loss": 0.9003,
+      "step": 308810
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.9607025330986456e-05,
+      "loss": 0.836,
+      "step": 308820
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.96058774010867e-05,
+      "loss": 0.8452,
+      "step": 308830
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.9604729471186956e-05,
+      "loss": 0.9092,
+      "step": 308840
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.960358154128721e-05,
+      "loss": 0.8789,
+      "step": 308850
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.9602433611387455e-05,
+      "loss": 0.9916,
+      "step": 308860
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.960128568148771e-05,
+      "loss": 1.0129,
+      "step": 308870
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.960013775158797e-05,
+      "loss": 0.9773,
+      "step": 308880
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.959898982168821e-05,
+      "loss": 0.7711,
+      "step": 308890
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.959784189178847e-05,
+      "loss": 0.9045,
+      "step": 308900
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.9596693961888725e-05,
+      "loss": 0.9446,
+      "step": 308910
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.9595546031988975e-05,
+      "loss": 0.8659,
+      "step": 308920
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.9594398102089225e-05,
+      "loss": 0.8531,
+      "step": 308930
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 3.959325017218948e-05,
+      "loss": 0.9184,
+      "step": 308940
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.959210224228973e-05,
+      "loss": 0.9181,
+      "step": 308950
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.959095431238999e-05,
+      "loss": 0.9663,
+      "step": 308960
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.9589806382490245e-05,
+      "loss": 0.9499,
+      "step": 308970
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.958865845259049e-05,
+      "loss": 0.9491,
+      "step": 308980
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.9587510522690744e-05,
+      "loss": 0.9474,
+      "step": 308990
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.9586362592791e-05,
+      "loss": 0.9605,
+      "step": 309000
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.9585214662891244e-05,
+      "loss": 0.9008,
+      "step": 309010
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.95840667329915e-05,
+      "loss": 1.0122,
+      "step": 309020
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.958291880309176e-05,
+      "loss": 1.0066,
+      "step": 309030
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.9581770873192e-05,
+      "loss": 0.9721,
+      "step": 309040
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.958062294329226e-05,
+      "loss": 0.9535,
+      "step": 309050
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.9579475013392514e-05,
+      "loss": 0.9533,
+      "step": 309060
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.957832708349276e-05,
+      "loss": 1.0918,
+      "step": 309070
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.957717915359301e-05,
+      "loss": 0.9516,
+      "step": 309080
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.957603122369327e-05,
+      "loss": 1.0097,
+      "step": 309090
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.957488329379352e-05,
+      "loss": 0.959,
+      "step": 309100
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.9573735363893776e-05,
+      "loss": 0.9755,
+      "step": 309110
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.957258743399403e-05,
+      "loss": 0.9035,
+      "step": 309120
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.9571439504094276e-05,
+      "loss": 0.9226,
+      "step": 309130
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.957029157419453e-05,
+      "loss": 0.8913,
+      "step": 309140
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.956914364429479e-05,
+      "loss": 0.8972,
+      "step": 309150
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 3.956799571439503e-05,
+      "loss": 0.8919,
+      "step": 309160
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.956684778449529e-05,
+      "loss": 1.0382,
+      "step": 309170
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.9565699854595546e-05,
+      "loss": 1.0257,
+      "step": 309180
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.956455192469579e-05,
+      "loss": 0.9441,
+      "step": 309190
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.9563403994796045e-05,
+      "loss": 1.049,
+      "step": 309200
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.95622560648963e-05,
+      "loss": 1.0366,
+      "step": 309210
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.956110813499655e-05,
+      "loss": 1.0018,
+      "step": 309220
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.95599602050968e-05,
+      "loss": 1.0248,
+      "step": 309230
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.955881227519706e-05,
+      "loss": 0.9246,
+      "step": 309240
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.955766434529731e-05,
+      "loss": 0.8953,
+      "step": 309250
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.9556516415397565e-05,
+      "loss": 0.9306,
+      "step": 309260
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.955536848549782e-05,
+      "loss": 0.8688,
+      "step": 309270
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.9554220555598065e-05,
+      "loss": 1.0418,
+      "step": 309280
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.955307262569832e-05,
+      "loss": 0.9533,
+      "step": 309290
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.955192469579858e-05,
+      "loss": 0.9492,
+      "step": 309300
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.955077676589882e-05,
+      "loss": 0.9871,
+      "step": 309310
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.954962883599908e-05,
+      "loss": 0.8782,
+      "step": 309320
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.9548480906099334e-05,
+      "loss": 0.9373,
+      "step": 309330
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.954733297619958e-05,
+      "loss": 0.912,
+      "step": 309340
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.9546185046299834e-05,
+      "loss": 0.9803,
+      "step": 309350
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.954503711640009e-05,
+      "loss": 0.8711,
+      "step": 309360
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.954388918650034e-05,
+      "loss": 0.9599,
+      "step": 309370
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 3.954274125660059e-05,
+      "loss": 0.8791,
+      "step": 309380
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.954159332670085e-05,
+      "loss": 1.0382,
+      "step": 309390
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.95404453968011e-05,
+      "loss": 0.9392,
+      "step": 309400
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.953929746690135e-05,
+      "loss": 0.929,
+      "step": 309410
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.953814953700161e-05,
+      "loss": 0.9479,
+      "step": 309420
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.953700160710185e-05,
+      "loss": 0.916,
+      "step": 309430
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.953585367720211e-05,
+      "loss": 0.8968,
+      "step": 309440
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.9534705747302366e-05,
+      "loss": 0.8578,
+      "step": 309450
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.953355781740261e-05,
+      "loss": 1.0185,
+      "step": 309460
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.9532409887502866e-05,
+      "loss": 0.9026,
+      "step": 309470
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.953126195760312e-05,
+      "loss": 0.8608,
+      "step": 309480
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.9530114027703366e-05,
+      "loss": 0.9167,
+      "step": 309490
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.952896609780362e-05,
+      "loss": 0.9084,
+      "step": 309500
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.952781816790388e-05,
+      "loss": 0.9431,
+      "step": 309510
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.952667023800413e-05,
+      "loss": 0.9298,
+      "step": 309520
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.952552230810438e-05,
+      "loss": 0.9482,
+      "step": 309530
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.9524374378204635e-05,
+      "loss": 0.9604,
+      "step": 309540
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.9523226448304885e-05,
+      "loss": 0.9326,
+      "step": 309550
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.952207851840514e-05,
+      "loss": 0.9456,
+      "step": 309560
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.95209305885054e-05,
+      "loss": 1.0248,
+      "step": 309570
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.951978265860564e-05,
+      "loss": 0.9559,
+      "step": 309580
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 3.95186347287059e-05,
+      "loss": 0.9943,
+      "step": 309590
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.9517486798806155e-05,
+      "loss": 0.9075,
+      "step": 309600
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.95163388689064e-05,
+      "loss": 1.0129,
+      "step": 309610
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.9515190939006654e-05,
+      "loss": 0.9386,
+      "step": 309620
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.951404300910691e-05,
+      "loss": 0.9005,
+      "step": 309630
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.9512895079207154e-05,
+      "loss": 0.9801,
+      "step": 309640
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.951174714930741e-05,
+      "loss": 1.0187,
+      "step": 309650
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.951059921940767e-05,
+      "loss": 0.8857,
+      "step": 309660
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.950945128950792e-05,
+      "loss": 0.9658,
+      "step": 309670
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.950830335960817e-05,
+      "loss": 0.892,
+      "step": 309680
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.9507155429708424e-05,
+      "loss": 0.9441,
+      "step": 309690
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.9506007499808674e-05,
+      "loss": 0.8944,
+      "step": 309700
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.950485956990893e-05,
+      "loss": 0.9009,
+      "step": 309710
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.950371164000919e-05,
+      "loss": 1.0536,
+      "step": 309720
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.950256371010943e-05,
+      "loss": 0.8825,
+      "step": 309730
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.9501415780209687e-05,
+      "loss": 0.9907,
+      "step": 309740
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.950026785030994e-05,
+      "loss": 0.9662,
+      "step": 309750
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.9499119920410186e-05,
+      "loss": 0.8885,
+      "step": 309760
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.949797199051044e-05,
+      "loss": 0.9,
+      "step": 309770
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.94968240606107e-05,
+      "loss": 0.9933,
+      "step": 309780
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.949567613071094e-05,
+      "loss": 0.9245,
+      "step": 309790
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.94945282008112e-05,
+      "loss": 0.9802,
+      "step": 309800
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 3.9493380270911456e-05,
+      "loss": 1.0196,
+      "step": 309810
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.9492232341011706e-05,
+      "loss": 0.9997,
+      "step": 309820
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.9491084411111956e-05,
+      "loss": 0.8815,
+      "step": 309830
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.948993648121221e-05,
+      "loss": 0.8747,
+      "step": 309840
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.948878855131246e-05,
+      "loss": 1.0291,
+      "step": 309850
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.948764062141272e-05,
+      "loss": 0.9213,
+      "step": 309860
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.9486492691512975e-05,
+      "loss": 1.0024,
+      "step": 309870
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.948534476161322e-05,
+      "loss": 1.0063,
+      "step": 309880
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.9484196831713475e-05,
+      "loss": 0.9156,
+      "step": 309890
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.948304890181373e-05,
+      "loss": 1.0057,
+      "step": 309900
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.9481900971913975e-05,
+      "loss": 1.0446,
+      "step": 309910
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.948075304201423e-05,
+      "loss": 0.8708,
+      "step": 309920
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.947960511211449e-05,
+      "loss": 0.9739,
+      "step": 309930
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.947845718221473e-05,
+      "loss": 1.0099,
+      "step": 309940
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.947730925231499e-05,
+      "loss": 1.0976,
+      "step": 309950
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.9476161322415244e-05,
+      "loss": 0.9471,
+      "step": 309960
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.9475013392515494e-05,
+      "loss": 0.9719,
+      "step": 309970
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.9473865462615744e-05,
+      "loss": 1.0466,
+      "step": 309980
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.9472717532716e-05,
+      "loss": 0.9618,
+      "step": 309990
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.947156960281625e-05,
+      "loss": 1.0245,
+      "step": 310000
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.947042167291651e-05,
+      "loss": 0.9962,
+      "step": 310010
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.946927374301675e-05,
+      "loss": 0.9486,
+      "step": 310020
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 3.946812581311701e-05,
+      "loss": 0.9279,
+      "step": 310030
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.9466977883217264e-05,
+      "loss": 0.8807,
+      "step": 310040
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.9465829953317507e-05,
+      "loss": 0.9413,
+      "step": 310050
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.946468202341776e-05,
+      "loss": 0.997,
+      "step": 310060
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.946353409351802e-05,
+      "loss": 0.983,
+      "step": 310070
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.946238616361827e-05,
+      "loss": 0.8515,
+      "step": 310080
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.946123823371852e-05,
+      "loss": 0.9088,
+      "step": 310090
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.9460090303818776e-05,
+      "loss": 0.9442,
+      "step": 310100
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.9458942373919026e-05,
+      "loss": 0.8961,
+      "step": 310110
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.945779444401928e-05,
+      "loss": 0.9447,
+      "step": 310120
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.945664651411953e-05,
+      "loss": 0.9728,
+      "step": 310130
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.945549858421978e-05,
+      "loss": 0.9672,
+      "step": 310140
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.945435065432004e-05,
+      "loss": 0.9199,
+      "step": 310150
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.9453202724420296e-05,
+      "loss": 0.912,
+      "step": 310160
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.945205479452054e-05,
+      "loss": 0.9505,
+      "step": 310170
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.9450906864620795e-05,
+      "loss": 1.0206,
+      "step": 310180
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.944975893472105e-05,
+      "loss": 0.9678,
+      "step": 310190
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.9448611004821295e-05,
+      "loss": 0.9243,
+      "step": 310200
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.944746307492155e-05,
+      "loss": 1.0229,
+      "step": 310210
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.944631514502181e-05,
+      "loss": 0.9436,
+      "step": 310220
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.944516721512206e-05,
+      "loss": 0.933,
+      "step": 310230
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.944401928522231e-05,
+      "loss": 0.902,
+      "step": 310240
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 3.9442871355322565e-05,
+      "loss": 0.94,
+      "step": 310250
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.9441723425422815e-05,
+      "loss": 1.0611,
+      "step": 310260
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.944057549552307e-05,
+      "loss": 0.9066,
+      "step": 310270
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.943942756562332e-05,
+      "loss": 0.9048,
+      "step": 310280
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.943827963572357e-05,
+      "loss": 0.8964,
+      "step": 310290
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.943713170582383e-05,
+      "loss": 1.1096,
+      "step": 310300
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.9435983775924084e-05,
+      "loss": 0.9391,
+      "step": 310310
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.943483584602433e-05,
+      "loss": 0.9426,
+      "step": 310320
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.9433687916124584e-05,
+      "loss": 0.8863,
+      "step": 310330
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.943253998622484e-05,
+      "loss": 0.8878,
+      "step": 310340
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.9431392056325084e-05,
+      "loss": 0.9236,
+      "step": 310350
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.943024412642534e-05,
+      "loss": 0.9475,
+      "step": 310360
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.94290961965256e-05,
+      "loss": 0.9202,
+      "step": 310370
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.942794826662585e-05,
+      "loss": 0.913,
+      "step": 310380
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.9426800336726097e-05,
+      "loss": 0.9051,
+      "step": 310390
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.942565240682635e-05,
+      "loss": 1.0378,
+      "step": 310400
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.94245044769266e-05,
+      "loss": 0.8899,
+      "step": 310410
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.942335654702686e-05,
+      "loss": 0.9788,
+      "step": 310420
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.942220861712711e-05,
+      "loss": 0.8826,
+      "step": 310430
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.942106068722736e-05,
+      "loss": 0.9464,
+      "step": 310440
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.9419912757327616e-05,
+      "loss": 1.0507,
+      "step": 310450
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.941876482742787e-05,
+      "loss": 0.9694,
+      "step": 310460
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 3.9417616897528116e-05,
+      "loss": 0.9065,
+      "step": 310470
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.941646896762837e-05,
+      "loss": 0.8886,
+      "step": 310480
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.941532103772863e-05,
+      "loss": 0.8885,
+      "step": 310490
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.941417310782887e-05,
+      "loss": 0.9932,
+      "step": 310500
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.941302517792913e-05,
+      "loss": 0.968,
+      "step": 310510
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.9411877248029385e-05,
+      "loss": 1.0185,
+      "step": 310520
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.9410729318129635e-05,
+      "loss": 0.9266,
+      "step": 310530
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.9409581388229885e-05,
+      "loss": 0.9114,
+      "step": 310540
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.940843345833014e-05,
+      "loss": 0.9713,
+      "step": 310550
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.940728552843039e-05,
+      "loss": 0.9699,
+      "step": 310560
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.940613759853065e-05,
+      "loss": 0.8913,
+      "step": 310570
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.9404989668630905e-05,
+      "loss": 0.9457,
+      "step": 310580
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.940384173873115e-05,
+      "loss": 0.9144,
+      "step": 310590
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.9402693808831404e-05,
+      "loss": 1.0348,
+      "step": 310600
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.940154587893166e-05,
+      "loss": 0.872,
+      "step": 310610
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.9400397949031904e-05,
+      "loss": 0.8301,
+      "step": 310620
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.939925001913216e-05,
+      "loss": 0.9741,
+      "step": 310630
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.939810208923242e-05,
+      "loss": 0.9059,
+      "step": 310640
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.939695415933266e-05,
+      "loss": 0.9233,
+      "step": 310650
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.939580622943292e-05,
+      "loss": 0.9295,
+      "step": 310660
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.9394658299533174e-05,
+      "loss": 0.9756,
+      "step": 310670
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 3.9393510369633424e-05,
+      "loss": 0.8817,
+      "step": 310680
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.9392362439733673e-05,
+      "loss": 1.066,
+      "step": 310690
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.939121450983393e-05,
+      "loss": 0.9008,
+      "step": 310700
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.939006657993418e-05,
+      "loss": 0.9415,
+      "step": 310710
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.9388918650034437e-05,
+      "loss": 0.9903,
+      "step": 310720
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.938777072013469e-05,
+      "loss": 0.9729,
+      "step": 310730
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.9386622790234936e-05,
+      "loss": 0.9171,
+      "step": 310740
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.938547486033519e-05,
+      "loss": 0.9623,
+      "step": 310750
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.938432693043545e-05,
+      "loss": 0.9373,
+      "step": 310760
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.938317900053569e-05,
+      "loss": 0.9451,
+      "step": 310770
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.938203107063595e-05,
+      "loss": 0.9488,
+      "step": 310780
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.9380883140736206e-05,
+      "loss": 0.9074,
+      "step": 310790
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.937973521083645e-05,
+      "loss": 0.9446,
+      "step": 310800
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.9378587280936706e-05,
+      "loss": 0.8743,
+      "step": 310810
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.937743935103696e-05,
+      "loss": 0.9925,
+      "step": 310820
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.937629142113721e-05,
+      "loss": 0.9847,
+      "step": 310830
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.937514349123746e-05,
+      "loss": 0.9644,
+      "step": 310840
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.937399556133772e-05,
+      "loss": 1.0474,
+      "step": 310850
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.937284763143797e-05,
+      "loss": 0.83,
+      "step": 310860
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.9371699701538225e-05,
+      "loss": 0.8742,
+      "step": 310870
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.937055177163848e-05,
+      "loss": 0.952,
+      "step": 310880
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.9369403841738725e-05,
+      "loss": 0.9817,
+      "step": 310890
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 3.936825591183898e-05,
+      "loss": 0.8515,
+      "step": 310900
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.936710798193924e-05,
+      "loss": 1.0587,
+      "step": 310910
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.936596005203948e-05,
+      "loss": 0.971,
+      "step": 310920
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.936481212213974e-05,
+      "loss": 0.8614,
+      "step": 310930
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.9363664192239994e-05,
+      "loss": 1.0183,
+      "step": 310940
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.936251626234024e-05,
+      "loss": 0.9505,
+      "step": 310950
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.9361368332440494e-05,
+      "loss": 0.9236,
+      "step": 310960
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.936022040254075e-05,
+      "loss": 1.046,
+      "step": 310970
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.9359072472641e-05,
+      "loss": 0.8977,
+      "step": 310980
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.935792454274125e-05,
+      "loss": 0.9275,
+      "step": 310990
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.935677661284151e-05,
+      "loss": 0.8413,
+      "step": 311000
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.935562868294176e-05,
+      "loss": 0.8756,
+      "step": 311010
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.9354480753042014e-05,
+      "loss": 0.9856,
+      "step": 311020
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.935333282314227e-05,
+      "loss": 1.007,
+      "step": 311030
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.935218489324251e-05,
+      "loss": 0.9287,
+      "step": 311040
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.935103696334277e-05,
+      "loss": 0.9291,
+      "step": 311050
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.9349889033443026e-05,
+      "loss": 0.8737,
+      "step": 311060
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.934874110354327e-05,
+      "loss": 0.99,
+      "step": 311070
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.9347593173643526e-05,
+      "loss": 0.9629,
+      "step": 311080
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.934644524374378e-05,
+      "loss": 1.0117,
+      "step": 311090
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.9345297313844026e-05,
+      "loss": 0.8648,
+      "step": 311100
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.934414938394428e-05,
+      "loss": 1.0025,
+      "step": 311110
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 3.934300145404454e-05,
+      "loss": 0.9096,
+      "step": 311120
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.934185352414479e-05,
+      "loss": 0.8747,
+      "step": 311130
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.934070559424504e-05,
+      "loss": 1.007,
+      "step": 311140
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.9339557664345296e-05,
+      "loss": 0.9174,
+      "step": 311150
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.9338409734445545e-05,
+      "loss": 0.9418,
+      "step": 311160
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.93372618045458e-05,
+      "loss": 0.8788,
+      "step": 311170
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.933611387464606e-05,
+      "loss": 1.0018,
+      "step": 311180
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.93349659447463e-05,
+      "loss": 0.9266,
+      "step": 311190
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.933381801484656e-05,
+      "loss": 0.9836,
+      "step": 311200
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.9332670084946815e-05,
+      "loss": 0.9875,
+      "step": 311210
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.933152215504706e-05,
+      "loss": 0.8535,
+      "step": 311220
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.9330374225147315e-05,
+      "loss": 0.9252,
+      "step": 311230
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.932922629524757e-05,
+      "loss": 0.9292,
+      "step": 311240
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.9328078365347814e-05,
+      "loss": 0.9976,
+      "step": 311250
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.932693043544807e-05,
+      "loss": 0.9661,
+      "step": 311260
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.932578250554833e-05,
+      "loss": 1.039,
+      "step": 311270
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.932463457564858e-05,
+      "loss": 0.9443,
+      "step": 311280
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.932348664574883e-05,
+      "loss": 0.9035,
+      "step": 311290
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.9322338715849084e-05,
+      "loss": 0.9373,
+      "step": 311300
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.9321190785949334e-05,
+      "loss": 0.9399,
+      "step": 311310
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.932004285604959e-05,
+      "loss": 0.9644,
+      "step": 311320
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.931889492614985e-05,
+      "loss": 0.8495,
+      "step": 311330
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 3.931774699625009e-05,
+      "loss": 1.0222,
+      "step": 311340
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.931659906635035e-05,
+      "loss": 0.9711,
+      "step": 311350
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.9315451136450603e-05,
+      "loss": 0.9267,
+      "step": 311360
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.9314303206550847e-05,
+      "loss": 0.8381,
+      "step": 311370
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.93131552766511e-05,
+      "loss": 0.879,
+      "step": 311380
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.931200734675136e-05,
+      "loss": 0.9886,
+      "step": 311390
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.93108594168516e-05,
+      "loss": 0.9749,
+      "step": 311400
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.930971148695186e-05,
+      "loss": 0.8665,
+      "step": 311410
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.930856355705211e-05,
+      "loss": 0.9685,
+      "step": 311420
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.9307415627152366e-05,
+      "loss": 0.8913,
+      "step": 311430
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.9306267697252616e-05,
+      "loss": 1.0335,
+      "step": 311440
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.9305119767352866e-05,
+      "loss": 1.0468,
+      "step": 311450
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.930397183745312e-05,
+      "loss": 1.0459,
+      "step": 311460
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.930282390755338e-05,
+      "loss": 0.8587,
+      "step": 311470
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.930167597765362e-05,
+      "loss": 0.9298,
+      "step": 311480
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.930052804775388e-05,
+      "loss": 0.9596,
+      "step": 311490
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.9299380117854135e-05,
+      "loss": 0.9071,
+      "step": 311500
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.929823218795438e-05,
+      "loss": 0.9973,
+      "step": 311510
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.9297084258054635e-05,
+      "loss": 0.9515,
+      "step": 311520
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.929593632815489e-05,
+      "loss": 0.8989,
+      "step": 311530
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.929478839825514e-05,
+      "loss": 0.8973,
+      "step": 311540
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.929364046835539e-05,
+      "loss": 0.9234,
+      "step": 311550
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 3.929249253845565e-05,
+      "loss": 0.9737,
+      "step": 311560
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.92913446085559e-05,
+      "loss": 0.9192,
+      "step": 311570
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.9290196678656154e-05,
+      "loss": 0.9236,
+      "step": 311580
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.9289048748756404e-05,
+      "loss": 0.8957,
+      "step": 311590
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.9287900818856654e-05,
+      "loss": 0.9301,
+      "step": 311600
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.928675288895691e-05,
+      "loss": 0.9394,
+      "step": 311610
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.928560495905717e-05,
+      "loss": 0.9456,
+      "step": 311620
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.928445702915741e-05,
+      "loss": 0.8955,
+      "step": 311630
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.928330909925767e-05,
+      "loss": 0.8937,
+      "step": 311640
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.9282161169357924e-05,
+      "loss": 0.9539,
+      "step": 311650
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.928101323945817e-05,
+      "loss": 1.0455,
+      "step": 311660
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.9279865309558423e-05,
+      "loss": 0.9312,
+      "step": 311670
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.927871737965868e-05,
+      "loss": 0.8934,
+      "step": 311680
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.927756944975893e-05,
+      "loss": 0.9271,
+      "step": 311690
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.927642151985918e-05,
+      "loss": 0.9837,
+      "step": 311700
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.9275273589959436e-05,
+      "loss": 0.8898,
+      "step": 311710
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.9274125660059686e-05,
+      "loss": 1.0657,
+      "step": 311720
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.927297773015994e-05,
+      "loss": 0.9442,
+      "step": 311730
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.927182980026019e-05,
+      "loss": 0.9523,
+      "step": 311740
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.927068187036044e-05,
+      "loss": 1.0008,
+      "step": 311750
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.92695339404607e-05,
+      "loss": 0.8768,
+      "step": 311760
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 3.9268386010560956e-05,
+      "loss": 1.0482,
+      "step": 311770
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.92672380806612e-05,
+      "loss": 0.9266,
+      "step": 311780
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.9266090150761456e-05,
+      "loss": 0.9191,
+      "step": 311790
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.926494222086171e-05,
+      "loss": 0.8775,
+      "step": 311800
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.9263794290961955e-05,
+      "loss": 1.0244,
+      "step": 311810
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.926264636106221e-05,
+      "loss": 0.9034,
+      "step": 311820
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.926149843116247e-05,
+      "loss": 0.9202,
+      "step": 311830
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.926035050126272e-05,
+      "loss": 0.9303,
+      "step": 311840
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.925920257136297e-05,
+      "loss": 0.8989,
+      "step": 311850
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.9258054641463225e-05,
+      "loss": 1.0017,
+      "step": 311860
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.9256906711563475e-05,
+      "loss": 0.9186,
+      "step": 311870
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.925575878166373e-05,
+      "loss": 1.0199,
+      "step": 311880
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.925461085176398e-05,
+      "loss": 1.0669,
+      "step": 311890
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.925346292186423e-05,
+      "loss": 0.8962,
+      "step": 311900
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.925231499196449e-05,
+      "loss": 0.8855,
+      "step": 311910
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.9251167062064744e-05,
+      "loss": 0.9208,
+      "step": 311920
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.925001913216499e-05,
+      "loss": 1.003,
+      "step": 311930
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.9248871202265244e-05,
+      "loss": 0.8991,
+      "step": 311940
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.92477232723655e-05,
+      "loss": 0.9736,
+      "step": 311950
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.9246575342465744e-05,
+      "loss": 0.876,
+      "step": 311960
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.9245427412566e-05,
+      "loss": 0.9187,
+      "step": 311970
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.924427948266626e-05,
+      "loss": 0.9261,
+      "step": 311980
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 3.924313155276651e-05,
+      "loss": 0.9775,
+      "step": 311990
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.924198362286676e-05,
+      "loss": 0.9427,
+      "step": 312000
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.924083569296701e-05,
+      "loss": 0.9906,
+      "step": 312010
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.923968776306726e-05,
+      "loss": 0.9596,
+      "step": 312020
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.923853983316752e-05,
+      "loss": 0.8688,
+      "step": 312030
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.923739190326777e-05,
+      "loss": 0.8375,
+      "step": 312040
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.923624397336802e-05,
+      "loss": 1.0424,
+      "step": 312050
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.9235096043468276e-05,
+      "loss": 0.9265,
+      "step": 312060
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.923394811356853e-05,
+      "loss": 0.8945,
+      "step": 312070
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.9232800183668776e-05,
+      "loss": 0.9555,
+      "step": 312080
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.923165225376903e-05,
+      "loss": 0.9468,
+      "step": 312090
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.923050432386929e-05,
+      "loss": 0.9403,
+      "step": 312100
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.922935639396953e-05,
+      "loss": 0.9435,
+      "step": 312110
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.922820846406979e-05,
+      "loss": 1.0202,
+      "step": 312120
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.9227060534170046e-05,
+      "loss": 0.9938,
+      "step": 312130
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.9225912604270295e-05,
+      "loss": 0.8364,
+      "step": 312140
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.9224764674370545e-05,
+      "loss": 0.9178,
+      "step": 312150
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.92236167444708e-05,
+      "loss": 0.9513,
+      "step": 312160
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.922246881457105e-05,
+      "loss": 0.9665,
+      "step": 312170
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.922132088467131e-05,
+      "loss": 0.8977,
+      "step": 312180
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.922017295477156e-05,
+      "loss": 0.9924,
+      "step": 312190
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.921902502487181e-05,
+      "loss": 0.923,
+      "step": 312200
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 3.9217877094972065e-05,
+      "loss": 0.8946,
+      "step": 312210
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.921672916507232e-05,
+      "loss": 1.0108,
+      "step": 312220
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.9215581235172564e-05,
+      "loss": 0.9429,
+      "step": 312230
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.921443330527282e-05,
+      "loss": 0.9283,
+      "step": 312240
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.921328537537308e-05,
+      "loss": 0.9127,
+      "step": 312250
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.921213744547332e-05,
+      "loss": 0.927,
+      "step": 312260
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.921098951557358e-05,
+      "loss": 0.9506,
+      "step": 312270
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.9209841585673834e-05,
+      "loss": 1.0498,
+      "step": 312280
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.9208693655774084e-05,
+      "loss": 0.973,
+      "step": 312290
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.9207545725874334e-05,
+      "loss": 0.9384,
+      "step": 312300
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.920639779597459e-05,
+      "loss": 1.0018,
+      "step": 312310
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.920524986607484e-05,
+      "loss": 0.955,
+      "step": 312320
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.92041019361751e-05,
+      "loss": 1.0065,
+      "step": 312330
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.920295400627535e-05,
+      "loss": 0.9004,
+      "step": 312340
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.9201806076375597e-05,
+      "loss": 1.006,
+      "step": 312350
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.920065814647585e-05,
+      "loss": 0.9945,
+      "step": 312360
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.919951021657611e-05,
+      "loss": 0.9351,
+      "step": 312370
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.919836228667635e-05,
+      "loss": 0.9987,
+      "step": 312380
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.919721435677661e-05,
+      "loss": 0.8315,
+      "step": 312390
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.9196066426876866e-05,
+      "loss": 0.9247,
+      "step": 312400
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.919491849697711e-05,
+      "loss": 0.9119,
+      "step": 312410
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.9193770567077366e-05,
+      "loss": 0.7813,
+      "step": 312420
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 3.919262263717762e-05,
+      "loss": 0.8879,
+      "step": 312430
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.919147470727787e-05,
+      "loss": 0.9298,
+      "step": 312440
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.919032677737812e-05,
+      "loss": 0.9716,
+      "step": 312450
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.918917884747838e-05,
+      "loss": 0.9715,
+      "step": 312460
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.918803091757863e-05,
+      "loss": 0.9676,
+      "step": 312470
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.9186882987678885e-05,
+      "loss": 0.9837,
+      "step": 312480
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.9185735057779135e-05,
+      "loss": 0.9415,
+      "step": 312490
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.9184587127879385e-05,
+      "loss": 0.958,
+      "step": 312500
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.918343919797964e-05,
+      "loss": 0.9031,
+      "step": 312510
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.91822912680799e-05,
+      "loss": 0.9075,
+      "step": 312520
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.918114333818014e-05,
+      "loss": 0.9801,
+      "step": 312530
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.91799954082804e-05,
+      "loss": 0.8847,
+      "step": 312540
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.9178847478380655e-05,
+      "loss": 0.8556,
+      "step": 312550
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.91776995484809e-05,
+      "loss": 0.8329,
+      "step": 312560
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.9176551618581154e-05,
+      "loss": 0.9022,
+      "step": 312570
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.917540368868141e-05,
+      "loss": 0.9066,
+      "step": 312580
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.917425575878166e-05,
+      "loss": 0.9636,
+      "step": 312590
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.917310782888191e-05,
+      "loss": 0.9124,
+      "step": 312600
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.917195989898217e-05,
+      "loss": 1.0003,
+      "step": 312610
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.917081196908242e-05,
+      "loss": 0.9443,
+      "step": 312620
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.9169664039182674e-05,
+      "loss": 0.9411,
+      "step": 312630
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.9168516109282924e-05,
+      "loss": 0.9237,
+      "step": 312640
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 3.9167368179383173e-05,
+      "loss": 0.8878,
+      "step": 312650
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.916622024948343e-05,
+      "loss": 0.9707,
+      "step": 312660
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.916507231958369e-05,
+      "loss": 0.9196,
+      "step": 312670
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.916392438968393e-05,
+      "loss": 0.9812,
+      "step": 312680
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.9162776459784186e-05,
+      "loss": 0.898,
+      "step": 312690
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.916162852988444e-05,
+      "loss": 0.8734,
+      "step": 312700
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.9160480599984686e-05,
+      "loss": 0.9405,
+      "step": 312710
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.915933267008494e-05,
+      "loss": 0.9662,
+      "step": 312720
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.91581847401852e-05,
+      "loss": 0.9999,
+      "step": 312730
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.915703681028545e-05,
+      "loss": 0.9204,
+      "step": 312740
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.91558888803857e-05,
+      "loss": 0.9382,
+      "step": 312750
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.9154740950485956e-05,
+      "loss": 0.8876,
+      "step": 312760
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.9153593020586206e-05,
+      "loss": 0.9676,
+      "step": 312770
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.915244509068646e-05,
+      "loss": 0.9371,
+      "step": 312780
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.9151297160786705e-05,
+      "loss": 0.9578,
+      "step": 312790
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.915014923088696e-05,
+      "loss": 0.9531,
+      "step": 312800
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.914900130098722e-05,
+      "loss": 0.9004,
+      "step": 312810
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.914785337108746e-05,
+      "loss": 0.8878,
+      "step": 312820
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.914670544118772e-05,
+      "loss": 0.9714,
+      "step": 312830
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.9145557511287975e-05,
+      "loss": 1.0549,
+      "step": 312840
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.9144409581388225e-05,
+      "loss": 0.9918,
+      "step": 312850
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 3.9143261651488475e-05,
+      "loss": 0.9836,
+      "step": 312860
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.914211372158873e-05,
+      "loss": 0.9962,
+      "step": 312870
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.914096579168898e-05,
+      "loss": 1.085,
+      "step": 312880
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.913981786178924e-05,
+      "loss": 0.9835,
+      "step": 312890
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.913866993188949e-05,
+      "loss": 0.975,
+      "step": 312900
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.913752200198974e-05,
+      "loss": 0.966,
+      "step": 312910
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.9136374072089994e-05,
+      "loss": 0.939,
+      "step": 312920
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.913522614219025e-05,
+      "loss": 0.9321,
+      "step": 312930
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.9134078212290494e-05,
+      "loss": 0.995,
+      "step": 312940
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.913293028239075e-05,
+      "loss": 0.9311,
+      "step": 312950
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.913178235249101e-05,
+      "loss": 1.0056,
+      "step": 312960
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.913063442259125e-05,
+      "loss": 0.9188,
+      "step": 312970
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.912948649269151e-05,
+      "loss": 1.0024,
+      "step": 312980
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.912833856279176e-05,
+      "loss": 0.9474,
+      "step": 312990
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.912719063289201e-05,
+      "loss": 1.0574,
+      "step": 313000
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.912604270299226e-05,
+      "loss": 0.9832,
+      "step": 313010
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.912489477309252e-05,
+      "loss": 0.9646,
+      "step": 313020
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.912374684319277e-05,
+      "loss": 0.8621,
+      "step": 313030
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.9122598913293026e-05,
+      "loss": 0.9047,
+      "step": 313040
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.9121450983393276e-05,
+      "loss": 0.9906,
+      "step": 313050
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.9120303053493526e-05,
+      "loss": 0.9838,
+      "step": 313060
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.911915512359378e-05,
+      "loss": 0.8498,
+      "step": 313070
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 3.911800719369404e-05,
+      "loss": 0.942,
+      "step": 313080
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.911685926379428e-05,
+      "loss": 1.0,
+      "step": 313090
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.911571133389454e-05,
+      "loss": 0.9361,
+      "step": 313100
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.9114563403994795e-05,
+      "loss": 1.0197,
+      "step": 313110
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.911341547409504e-05,
+      "loss": 0.9286,
+      "step": 313120
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.9112267544195295e-05,
+      "loss": 0.9981,
+      "step": 313130
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.911111961429555e-05,
+      "loss": 0.8647,
+      "step": 313140
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.91099716843958e-05,
+      "loss": 0.9588,
+      "step": 313150
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.910882375449605e-05,
+      "loss": 0.8629,
+      "step": 313160
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.910767582459631e-05,
+      "loss": 0.8617,
+      "step": 313170
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.910652789469656e-05,
+      "loss": 0.8948,
+      "step": 313180
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.9105379964796815e-05,
+      "loss": 0.8724,
+      "step": 313190
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.9104232034897065e-05,
+      "loss": 0.9975,
+      "step": 313200
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.9103084104997314e-05,
+      "loss": 0.8809,
+      "step": 313210
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.910193617509757e-05,
+      "loss": 0.914,
+      "step": 313220
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.910078824519783e-05,
+      "loss": 0.9367,
+      "step": 313230
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.909964031529807e-05,
+      "loss": 0.9091,
+      "step": 313240
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.909849238539833e-05,
+      "loss": 0.9405,
+      "step": 313250
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.9097344455498584e-05,
+      "loss": 0.9651,
+      "step": 313260
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.909619652559883e-05,
+      "loss": 0.9058,
+      "step": 313270
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.9095048595699084e-05,
+      "loss": 0.9657,
+      "step": 313280
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.909390066579934e-05,
+      "loss": 0.9167,
+      "step": 313290
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 3.909275273589959e-05,
+      "loss": 0.9706,
+      "step": 313300
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.909160480599984e-05,
+      "loss": 0.8823,
+      "step": 313310
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.90904568761001e-05,
+      "loss": 1.0137,
+      "step": 313320
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.9089308946200346e-05,
+      "loss": 0.9176,
+      "step": 313330
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.90881610163006e-05,
+      "loss": 0.9729,
+      "step": 313340
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.908701308640085e-05,
+      "loss": 0.9055,
+      "step": 313350
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.90858651565011e-05,
+      "loss": 0.8389,
+      "step": 313360
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.908471722660136e-05,
+      "loss": 1.0317,
+      "step": 313370
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.9083569296701616e-05,
+      "loss": 0.9032,
+      "step": 313380
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.908242136680186e-05,
+      "loss": 0.9074,
+      "step": 313390
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.9081273436902116e-05,
+      "loss": 0.9483,
+      "step": 313400
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.908012550700237e-05,
+      "loss": 1.0721,
+      "step": 313410
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.9078977577102616e-05,
+      "loss": 0.8753,
+      "step": 313420
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.907782964720287e-05,
+      "loss": 0.9707,
+      "step": 313430
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.907668171730313e-05,
+      "loss": 0.9529,
+      "step": 313440
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.907553378740338e-05,
+      "loss": 1.0782,
+      "step": 313450
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.907438585750363e-05,
+      "loss": 1.0013,
+      "step": 313460
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.9073237927603885e-05,
+      "loss": 0.9106,
+      "step": 313470
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.9072089997704135e-05,
+      "loss": 1.0333,
+      "step": 313480
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.907094206780439e-05,
+      "loss": 1.0218,
+      "step": 313490
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.906979413790464e-05,
+      "loss": 0.8686,
+      "step": 313500
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.906864620800489e-05,
+      "loss": 0.8564,
+      "step": 313510
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 3.906749827810515e-05,
+      "loss": 0.9453,
+      "step": 313520
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.9066350348205405e-05,
+      "loss": 0.8842,
+      "step": 313530
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.906520241830565e-05,
+      "loss": 1.0643,
+      "step": 313540
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.9064054488405904e-05,
+      "loss": 0.9076,
+      "step": 313550
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.906290655850616e-05,
+      "loss": 0.9646,
+      "step": 313560
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.9061758628606404e-05,
+      "loss": 0.9685,
+      "step": 313570
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.906061069870666e-05,
+      "loss": 0.9198,
+      "step": 313580
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.905946276880692e-05,
+      "loss": 0.8739,
+      "step": 313590
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.905831483890717e-05,
+      "loss": 0.9381,
+      "step": 313600
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.905716690900742e-05,
+      "loss": 0.9506,
+      "step": 313610
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.9056018979107674e-05,
+      "loss": 0.9215,
+      "step": 313620
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.9054871049207923e-05,
+      "loss": 0.8961,
+      "step": 313630
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.905372311930818e-05,
+      "loss": 0.8634,
+      "step": 313640
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.905257518940843e-05,
+      "loss": 0.9795,
+      "step": 313650
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.905142725950868e-05,
+      "loss": 1.0236,
+      "step": 313660
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.9050279329608936e-05,
+      "loss": 0.9655,
+      "step": 313670
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.904913139970919e-05,
+      "loss": 0.9937,
+      "step": 313680
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.9047983469809436e-05,
+      "loss": 0.903,
+      "step": 313690
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.904683553990969e-05,
+      "loss": 0.9029,
+      "step": 313700
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.904568761000995e-05,
+      "loss": 1.0682,
+      "step": 313710
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.904453968011019e-05,
+      "loss": 1.0582,
+      "step": 313720
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.904339175021045e-05,
+      "loss": 0.9794,
+      "step": 313730
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 3.9042243820310706e-05,
+      "loss": 0.9357,
+      "step": 313740
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.9041095890410956e-05,
+      "loss": 0.874,
+      "step": 313750
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.9039947960511205e-05,
+      "loss": 0.9474,
+      "step": 313760
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.903880003061146e-05,
+      "loss": 0.971,
+      "step": 313770
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.903765210071171e-05,
+      "loss": 1.0065,
+      "step": 313780
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.903650417081197e-05,
+      "loss": 0.9059,
+      "step": 313790
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.903535624091222e-05,
+      "loss": 0.8657,
+      "step": 313800
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.903420831101247e-05,
+      "loss": 0.9413,
+      "step": 313810
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.9033060381112725e-05,
+      "loss": 0.9607,
+      "step": 313820
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.903191245121298e-05,
+      "loss": 0.9445,
+      "step": 313830
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.9030764521313225e-05,
+      "loss": 0.9284,
+      "step": 313840
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.902961659141348e-05,
+      "loss": 0.9041,
+      "step": 313850
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.902846866151374e-05,
+      "loss": 0.9178,
+      "step": 313860
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.902732073161398e-05,
+      "loss": 0.9748,
+      "step": 313870
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.902617280171424e-05,
+      "loss": 0.8896,
+      "step": 313880
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.9025024871814494e-05,
+      "loss": 1.0749,
+      "step": 313890
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.9023876941914744e-05,
+      "loss": 0.9777,
+      "step": 313900
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.9022729012014994e-05,
+      "loss": 0.9302,
+      "step": 313910
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.902158108211525e-05,
+      "loss": 0.9841,
+      "step": 313920
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.90204331522155e-05,
+      "loss": 0.9509,
+      "step": 313930
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.901928522231576e-05,
+      "loss": 1.0611,
+      "step": 313940
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 3.901813729241601e-05,
+      "loss": 0.8962,
+      "step": 313950
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.901698936251626e-05,
+      "loss": 0.9898,
+      "step": 313960
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.901584143261651e-05,
+      "loss": 0.9178,
+      "step": 313970
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.901469350271677e-05,
+      "loss": 0.8875,
+      "step": 313980
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.901354557281701e-05,
+      "loss": 0.9577,
+      "step": 313990
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.901239764291727e-05,
+      "loss": 0.9186,
+      "step": 314000
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.9011249713017526e-05,
+      "loss": 0.9997,
+      "step": 314010
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.901010178311777e-05,
+      "loss": 0.9984,
+      "step": 314020
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.9008953853218026e-05,
+      "loss": 0.9171,
+      "step": 314030
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.900780592331828e-05,
+      "loss": 0.8847,
+      "step": 314040
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.900665799341853e-05,
+      "loss": 0.9801,
+      "step": 314050
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.900551006351878e-05,
+      "loss": 1.0008,
+      "step": 314060
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.900436213361904e-05,
+      "loss": 0.9978,
+      "step": 314070
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.900321420371929e-05,
+      "loss": 0.9629,
+      "step": 314080
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.9002066273819545e-05,
+      "loss": 0.9811,
+      "step": 314090
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.9000918343919795e-05,
+      "loss": 0.9974,
+      "step": 314100
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.8999770414020045e-05,
+      "loss": 0.8631,
+      "step": 314110
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.89986224841203e-05,
+      "loss": 1.0287,
+      "step": 314120
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.899747455422056e-05,
+      "loss": 0.9148,
+      "step": 314130
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.89963266243208e-05,
+      "loss": 1.003,
+      "step": 314140
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.899517869442106e-05,
+      "loss": 0.9117,
+      "step": 314150
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.899403076452131e-05,
+      "loss": 0.9804,
+      "step": 314160
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 3.899288283462156e-05,
+      "loss": 0.9065,
+      "step": 314170
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.8991734904721815e-05,
+      "loss": 0.8336,
+      "step": 314180
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.8990586974822064e-05,
+      "loss": 0.943,
+      "step": 314190
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.898943904492232e-05,
+      "loss": 0.9406,
+      "step": 314200
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.898829111502257e-05,
+      "loss": 0.8957,
+      "step": 314210
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.898714318512282e-05,
+      "loss": 0.9471,
+      "step": 314220
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.898599525522308e-05,
+      "loss": 0.9862,
+      "step": 314230
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.8984847325323334e-05,
+      "loss": 0.9621,
+      "step": 314240
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.898369939542358e-05,
+      "loss": 0.9595,
+      "step": 314250
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.8982551465523834e-05,
+      "loss": 0.806,
+      "step": 314260
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.898140353562409e-05,
+      "loss": 0.9455,
+      "step": 314270
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.898025560572433e-05,
+      "loss": 0.9814,
+      "step": 314280
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.897910767582459e-05,
+      "loss": 0.9733,
+      "step": 314290
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.897795974592485e-05,
+      "loss": 0.929,
+      "step": 314300
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.8976811816025096e-05,
+      "loss": 0.9038,
+      "step": 314310
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.8975663886125346e-05,
+      "loss": 0.9549,
+      "step": 314320
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.89745159562256e-05,
+      "loss": 0.9506,
+      "step": 314330
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.897336802632585e-05,
+      "loss": 0.9191,
+      "step": 314340
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.897222009642611e-05,
+      "loss": 0.9673,
+      "step": 314350
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.897107216652636e-05,
+      "loss": 1.0063,
+      "step": 314360
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.896992423662661e-05,
+      "loss": 0.9015,
+      "step": 314370
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.8968776306726866e-05,
+      "loss": 0.9156,
+      "step": 314380
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 3.896762837682712e-05,
+      "loss": 0.9271,
+      "step": 314390
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.8966480446927366e-05,
+      "loss": 0.9773,
+      "step": 314400
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.896533251702762e-05,
+      "loss": 0.9705,
+      "step": 314410
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.896418458712788e-05,
+      "loss": 0.9436,
+      "step": 314420
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.896303665722812e-05,
+      "loss": 1.0404,
+      "step": 314430
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.896188872732838e-05,
+      "loss": 0.9528,
+      "step": 314440
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.8960740797428635e-05,
+      "loss": 0.9486,
+      "step": 314450
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.8959592867528885e-05,
+      "loss": 1.0422,
+      "step": 314460
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.8958444937629135e-05,
+      "loss": 0.9913,
+      "step": 314470
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.895729700772939e-05,
+      "loss": 0.9717,
+      "step": 314480
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.895614907782964e-05,
+      "loss": 1.0336,
+      "step": 314490
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.89550011479299e-05,
+      "loss": 0.8838,
+      "step": 314500
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.895385321803015e-05,
+      "loss": 0.9178,
+      "step": 314510
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.89527052881304e-05,
+      "loss": 0.9856,
+      "step": 314520
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.8951557358230654e-05,
+      "loss": 0.9247,
+      "step": 314530
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.895040942833091e-05,
+      "loss": 0.9191,
+      "step": 314540
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.8949261498431154e-05,
+      "loss": 0.9208,
+      "step": 314550
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.894811356853141e-05,
+      "loss": 0.9818,
+      "step": 314560
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.894696563863167e-05,
+      "loss": 0.9903,
+      "step": 314570
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.894581770873191e-05,
+      "loss": 0.8605,
+      "step": 314580
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.894466977883217e-05,
+      "loss": 0.9648,
+      "step": 314590
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.8943521848932424e-05,
+      "loss": 0.9646,
+      "step": 314600
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 3.8942373919032673e-05,
+      "loss": 1.0231,
+      "step": 314610
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.894122598913292e-05,
+      "loss": 0.988,
+      "step": 314620
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.894007805923318e-05,
+      "loss": 0.8836,
+      "step": 314630
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.893893012933343e-05,
+      "loss": 0.918,
+      "step": 314640
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.8937782199433686e-05,
+      "loss": 0.8717,
+      "step": 314650
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.8936634269533936e-05,
+      "loss": 1.06,
+      "step": 314660
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.8935486339634186e-05,
+      "loss": 0.9982,
+      "step": 314670
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.893433840973444e-05,
+      "loss": 0.8665,
+      "step": 314680
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.89331904798347e-05,
+      "loss": 1.0223,
+      "step": 314690
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.893204254993494e-05,
+      "loss": 0.9682,
+      "step": 314700
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.89308946200352e-05,
+      "loss": 1.0241,
+      "step": 314710
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.8929746690135456e-05,
+      "loss": 0.9093,
+      "step": 314720
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.89285987602357e-05,
+      "loss": 0.9814,
+      "step": 314730
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.8927450830335955e-05,
+      "loss": 0.9569,
+      "step": 314740
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.892630290043621e-05,
+      "loss": 0.9403,
+      "step": 314750
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.892515497053646e-05,
+      "loss": 0.8535,
+      "step": 314760
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.892400704063671e-05,
+      "loss": 1.023,
+      "step": 314770
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.892285911073697e-05,
+      "loss": 0.9169,
+      "step": 314780
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.892171118083722e-05,
+      "loss": 0.9143,
+      "step": 314790
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.8920563250937475e-05,
+      "loss": 0.9877,
+      "step": 314800
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.8919415321037725e-05,
+      "loss": 1.0047,
+      "step": 314810
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.8918267391137975e-05,
+      "loss": 0.88,
+      "step": 314820
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 3.891711946123823e-05,
+      "loss": 0.9854,
+      "step": 314830
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.891597153133849e-05,
+      "loss": 0.991,
+      "step": 314840
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.891482360143873e-05,
+      "loss": 1.0206,
+      "step": 314850
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.891367567153899e-05,
+      "loss": 0.9874,
+      "step": 314860
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.8912527741639244e-05,
+      "loss": 0.9789,
+      "step": 314870
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.891137981173949e-05,
+      "loss": 0.955,
+      "step": 314880
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.8910231881839744e-05,
+      "loss": 0.9818,
+      "step": 314890
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.890908395194e-05,
+      "loss": 0.9626,
+      "step": 314900
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.890793602204025e-05,
+      "loss": 0.9715,
+      "step": 314910
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.89067880921405e-05,
+      "loss": 0.7818,
+      "step": 314920
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.890564016224076e-05,
+      "loss": 0.8647,
+      "step": 314930
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.890449223234101e-05,
+      "loss": 1.0336,
+      "step": 314940
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.890334430244126e-05,
+      "loss": 0.8997,
+      "step": 314950
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.890219637254151e-05,
+      "loss": 0.9899,
+      "step": 314960
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.890104844264176e-05,
+      "loss": 0.9863,
+      "step": 314970
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.889990051274202e-05,
+      "loss": 1.071,
+      "step": 314980
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.8898752582842276e-05,
+      "loss": 0.954,
+      "step": 314990
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.889760465294252e-05,
+      "loss": 1.0626,
+      "step": 315000
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.8896456723042776e-05,
+      "loss": 0.8503,
+      "step": 315010
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.889530879314303e-05,
+      "loss": 0.9812,
+      "step": 315020
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.8894160863243276e-05,
+      "loss": 0.9778,
+      "step": 315030
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 3.889301293334353e-05,
+      "loss": 0.938,
+      "step": 315040
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.889186500344379e-05,
+      "loss": 1.0126,
+      "step": 315050
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.889071707354404e-05,
+      "loss": 0.9372,
+      "step": 315060
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.888956914364429e-05,
+      "loss": 0.9664,
+      "step": 315070
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.8888421213744545e-05,
+      "loss": 0.9637,
+      "step": 315080
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.8887273283844795e-05,
+      "loss": 0.9552,
+      "step": 315090
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.888612535394505e-05,
+      "loss": 0.9316,
+      "step": 315100
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.88849774240453e-05,
+      "loss": 0.9035,
+      "step": 315110
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.888382949414555e-05,
+      "loss": 1.012,
+      "step": 315120
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.888268156424581e-05,
+      "loss": 1.1302,
+      "step": 315130
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.8881533634346065e-05,
+      "loss": 0.9663,
+      "step": 315140
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.888038570444631e-05,
+      "loss": 0.8847,
+      "step": 315150
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.8879237774546565e-05,
+      "loss": 0.9734,
+      "step": 315160
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.887808984464682e-05,
+      "loss": 1.0126,
+      "step": 315170
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.8876941914747064e-05,
+      "loss": 0.9405,
+      "step": 315180
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.887579398484732e-05,
+      "loss": 0.9129,
+      "step": 315190
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.887464605494758e-05,
+      "loss": 0.9646,
+      "step": 315200
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.887349812504783e-05,
+      "loss": 0.9781,
+      "step": 315210
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.887235019514808e-05,
+      "loss": 1.0066,
+      "step": 315220
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.8871202265248334e-05,
+      "loss": 0.9516,
+      "step": 315230
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.8870054335348584e-05,
+      "loss": 0.8838,
+      "step": 315240
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.886890640544884e-05,
+      "loss": 0.9371,
+      "step": 315250
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 3.886775847554909e-05,
+      "loss": 0.9373,
+      "step": 315260
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.886661054564934e-05,
+      "loss": 0.9317,
+      "step": 315270
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.88654626157496e-05,
+      "loss": 0.9781,
+      "step": 315280
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.886431468584985e-05,
+      "loss": 0.9524,
+      "step": 315290
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.8863166755950096e-05,
+      "loss": 0.9554,
+      "step": 315300
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.886201882605035e-05,
+      "loss": 0.9741,
+      "step": 315310
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.886087089615061e-05,
+      "loss": 1.0376,
+      "step": 315320
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.885972296625085e-05,
+      "loss": 0.9628,
+      "step": 315330
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.885857503635111e-05,
+      "loss": 1.0545,
+      "step": 315340
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.8857427106451366e-05,
+      "loss": 0.9916,
+      "step": 315350
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.8856279176551616e-05,
+      "loss": 0.9489,
+      "step": 315360
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.8855131246651866e-05,
+      "loss": 0.8725,
+      "step": 315370
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.885398331675212e-05,
+      "loss": 1.0179,
+      "step": 315380
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.885283538685237e-05,
+      "loss": 1.0008,
+      "step": 315390
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.885168745695263e-05,
+      "loss": 0.9436,
+      "step": 315400
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.885053952705288e-05,
+      "loss": 0.9951,
+      "step": 315410
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.884939159715313e-05,
+      "loss": 0.9506,
+      "step": 315420
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.8848243667253385e-05,
+      "loss": 1.0375,
+      "step": 315430
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.884709573735364e-05,
+      "loss": 0.9996,
+      "step": 315440
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.8845947807453885e-05,
+      "loss": 0.8749,
+      "step": 315450
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.884479987755414e-05,
+      "loss": 0.8512,
+      "step": 315460
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.88436519476544e-05,
+      "loss": 0.98,
+      "step": 315470
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 3.884250401775464e-05,
+      "loss": 0.946,
+      "step": 315480
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.88413560878549e-05,
+      "loss": 0.9449,
+      "step": 315490
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.8840208157955154e-05,
+      "loss": 0.9409,
+      "step": 315500
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.8839060228055404e-05,
+      "loss": 0.9388,
+      "step": 315510
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.8837912298155654e-05,
+      "loss": 0.9475,
+      "step": 315520
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.8836764368255904e-05,
+      "loss": 0.9513,
+      "step": 315530
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.883561643835616e-05,
+      "loss": 1.0151,
+      "step": 315540
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.883446850845642e-05,
+      "loss": 0.918,
+      "step": 315550
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.883332057855666e-05,
+      "loss": 0.9078,
+      "step": 315560
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.883217264865692e-05,
+      "loss": 0.9542,
+      "step": 315570
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.8831024718757174e-05,
+      "loss": 0.9528,
+      "step": 315580
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.882987678885742e-05,
+      "loss": 0.9558,
+      "step": 315590
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.882872885895767e-05,
+      "loss": 0.9413,
+      "step": 315600
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.882758092905793e-05,
+      "loss": 0.9613,
+      "step": 315610
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.882643299915818e-05,
+      "loss": 0.9074,
+      "step": 315620
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.882528506925843e-05,
+      "loss": 0.9233,
+      "step": 315630
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.8824137139358686e-05,
+      "loss": 1.0251,
+      "step": 315640
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.8822989209458936e-05,
+      "loss": 0.9732,
+      "step": 315650
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.882184127955919e-05,
+      "loss": 1.0699,
+      "step": 315660
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.882069334965944e-05,
+      "loss": 0.9954,
+      "step": 315670
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.881954541975969e-05,
+      "loss": 0.9131,
+      "step": 315680
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.881839748985995e-05,
+      "loss": 0.9701,
+      "step": 315690
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 3.8817249559960206e-05,
+      "loss": 0.9388,
+      "step": 315700
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.881610163006045e-05,
+      "loss": 0.9509,
+      "step": 315710
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.8814953700160705e-05,
+      "loss": 0.9502,
+      "step": 315720
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.881380577026096e-05,
+      "loss": 0.9764,
+      "step": 315730
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.8812657840361205e-05,
+      "loss": 1.0256,
+      "step": 315740
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.881150991046146e-05,
+      "loss": 0.909,
+      "step": 315750
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.881036198056172e-05,
+      "loss": 0.9772,
+      "step": 315760
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.880921405066197e-05,
+      "loss": 0.9736,
+      "step": 315770
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.880806612076222e-05,
+      "loss": 0.986,
+      "step": 315780
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.8806918190862475e-05,
+      "loss": 0.9702,
+      "step": 315790
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.8805770260962725e-05,
+      "loss": 1.0199,
+      "step": 315800
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.880462233106298e-05,
+      "loss": 0.9598,
+      "step": 315810
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.880347440116323e-05,
+      "loss": 1.0538,
+      "step": 315820
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.880232647126348e-05,
+      "loss": 0.9147,
+      "step": 315830
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.880117854136374e-05,
+      "loss": 0.9651,
+      "step": 315840
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.8800030611463994e-05,
+      "loss": 0.9414,
+      "step": 315850
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.879888268156424e-05,
+      "loss": 0.9659,
+      "step": 315860
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.8797734751664494e-05,
+      "loss": 0.9803,
+      "step": 315870
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.879658682176475e-05,
+      "loss": 0.8826,
+      "step": 315880
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.8795438891864994e-05,
+      "loss": 1.0157,
+      "step": 315890
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.879429096196525e-05,
+      "loss": 0.9747,
+      "step": 315900
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.879314303206551e-05,
+      "loss": 1.0743,
+      "step": 315910
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 3.879199510216576e-05,
+      "loss": 1.0308,
+      "step": 315920
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.8790847172266007e-05,
+      "loss": 0.9513,
+      "step": 315930
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.878969924236626e-05,
+      "loss": 0.9969,
+      "step": 315940
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.878855131246651e-05,
+      "loss": 0.9221,
+      "step": 315950
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.878740338256677e-05,
+      "loss": 0.9061,
+      "step": 315960
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.878625545266702e-05,
+      "loss": 1.0755,
+      "step": 315970
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.878510752276727e-05,
+      "loss": 0.8964,
+      "step": 315980
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.8783959592867526e-05,
+      "loss": 0.9111,
+      "step": 315990
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.878281166296778e-05,
+      "loss": 0.9976,
+      "step": 316000
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.8781663733068026e-05,
+      "loss": 0.9088,
+      "step": 316010
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.878051580316828e-05,
+      "loss": 1.0513,
+      "step": 316020
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.877936787326854e-05,
+      "loss": 0.8727,
+      "step": 316030
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.877821994336878e-05,
+      "loss": 0.9912,
+      "step": 316040
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.877707201346904e-05,
+      "loss": 0.9818,
+      "step": 316050
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.8775924083569295e-05,
+      "loss": 0.9163,
+      "step": 316060
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.8774776153669545e-05,
+      "loss": 0.9584,
+      "step": 316070
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.8773628223769795e-05,
+      "loss": 1.009,
+      "step": 316080
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.877248029387005e-05,
+      "loss": 0.9402,
+      "step": 316090
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.87713323639703e-05,
+      "loss": 1.0189,
+      "step": 316100
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.877018443407056e-05,
+      "loss": 1.0193,
+      "step": 316110
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.876903650417081e-05,
+      "loss": 0.9777,
+      "step": 316120
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 3.876788857427106e-05,
+      "loss": 0.841,
+      "step": 316130
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.8766740644371314e-05,
+      "loss": 0.93,
+      "step": 316140
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.876559271447157e-05,
+      "loss": 0.9907,
+      "step": 316150
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.8764444784571814e-05,
+      "loss": 1.0324,
+      "step": 316160
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.876329685467207e-05,
+      "loss": 1.0023,
+      "step": 316170
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.876214892477233e-05,
+      "loss": 0.9321,
+      "step": 316180
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.876100099487257e-05,
+      "loss": 1.021,
+      "step": 316190
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.875985306497283e-05,
+      "loss": 1.0096,
+      "step": 316200
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.8758705135073084e-05,
+      "loss": 1.0113,
+      "step": 316210
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.8757557205173334e-05,
+      "loss": 0.9798,
+      "step": 316220
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.8756409275273584e-05,
+      "loss": 0.938,
+      "step": 316230
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.875526134537384e-05,
+      "loss": 1.043,
+      "step": 316240
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.875411341547409e-05,
+      "loss": 0.9831,
+      "step": 316250
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.875296548557435e-05,
+      "loss": 1.0355,
+      "step": 316260
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.8751817555674596e-05,
+      "loss": 0.9999,
+      "step": 316270
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.8750669625774846e-05,
+      "loss": 0.9937,
+      "step": 316280
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.87495216958751e-05,
+      "loss": 1.0029,
+      "step": 316290
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.874837376597536e-05,
+      "loss": 0.884,
+      "step": 316300
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.87472258360756e-05,
+      "loss": 0.975,
+      "step": 316310
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.874607790617586e-05,
+      "loss": 1.0052,
+      "step": 316320
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.8744929976276116e-05,
+      "loss": 0.9151,
+      "step": 316330
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.874378204637636e-05,
+      "loss": 0.9298,
+      "step": 316340
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 3.8742634116476616e-05,
+      "loss": 0.9632,
+      "step": 316350
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.874148618657687e-05,
+      "loss": 0.9222,
+      "step": 316360
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.874033825667712e-05,
+      "loss": 0.9304,
+      "step": 316370
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.873919032677737e-05,
+      "loss": 0.8853,
+      "step": 316380
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.873804239687763e-05,
+      "loss": 0.9502,
+      "step": 316390
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.873689446697788e-05,
+      "loss": 0.8793,
+      "step": 316400
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.8735746537078135e-05,
+      "loss": 0.8514,
+      "step": 316410
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.8734598607178385e-05,
+      "loss": 0.9817,
+      "step": 316420
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.8733450677278635e-05,
+      "loss": 1.0187,
+      "step": 316430
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.873230274737889e-05,
+      "loss": 1.061,
+      "step": 316440
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.873115481747915e-05,
+      "loss": 1.0664,
+      "step": 316450
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.873000688757939e-05,
+      "loss": 0.9365,
+      "step": 316460
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.872885895767965e-05,
+      "loss": 1.0225,
+      "step": 316470
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.8727711027779904e-05,
+      "loss": 0.9659,
+      "step": 316480
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.872656309788015e-05,
+      "loss": 0.9387,
+      "step": 316490
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.8725415167980404e-05,
+      "loss": 0.8811,
+      "step": 316500
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.872426723808066e-05,
+      "loss": 0.8977,
+      "step": 316510
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.872311930818091e-05,
+      "loss": 0.9753,
+      "step": 316520
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.872197137828116e-05,
+      "loss": 0.9389,
+      "step": 316530
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.872082344838142e-05,
+      "loss": 0.9063,
+      "step": 316540
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.871967551848167e-05,
+      "loss": 0.9538,
+      "step": 316550
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.8718527588581924e-05,
+      "loss": 0.9634,
+      "step": 316560
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 3.8717379658682173e-05,
+      "loss": 0.8867,
+      "step": 316570
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.871623172878242e-05,
+      "loss": 0.975,
+      "step": 316580
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.871508379888268e-05,
+      "loss": 0.9449,
+      "step": 316590
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.8713935868982937e-05,
+      "loss": 0.9202,
+      "step": 316600
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.871278793908318e-05,
+      "loss": 0.9994,
+      "step": 316610
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.8711640009183436e-05,
+      "loss": 0.9384,
+      "step": 316620
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.871049207928369e-05,
+      "loss": 0.9634,
+      "step": 316630
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.8709344149383936e-05,
+      "loss": 0.8603,
+      "step": 316640
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.870819621948419e-05,
+      "loss": 0.9226,
+      "step": 316650
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.870704828958445e-05,
+      "loss": 0.9298,
+      "step": 316660
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.87059003596847e-05,
+      "loss": 0.9449,
+      "step": 316670
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.870475242978495e-05,
+      "loss": 0.9731,
+      "step": 316680
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.8703604499885206e-05,
+      "loss": 0.866,
+      "step": 316690
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.8702456569985455e-05,
+      "loss": 1.0033,
+      "step": 316700
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.870130864008571e-05,
+      "loss": 1.0013,
+      "step": 316710
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.870016071018596e-05,
+      "loss": 0.9926,
+      "step": 316720
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.869901278028621e-05,
+      "loss": 1.0091,
+      "step": 316730
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.869786485038647e-05,
+      "loss": 0.9344,
+      "step": 316740
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.8696716920486725e-05,
+      "loss": 0.912,
+      "step": 316750
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.869556899058697e-05,
+      "loss": 0.9418,
+      "step": 316760
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.8694421060687225e-05,
+      "loss": 0.9542,
+      "step": 316770
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.869327313078748e-05,
+      "loss": 0.9992,
+      "step": 316780
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 3.8692125200887724e-05,
+      "loss": 0.9118,
+      "step": 316790
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.869097727098798e-05,
+      "loss": 0.8708,
+      "step": 316800
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.868982934108824e-05,
+      "loss": 1.0656,
+      "step": 316810
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.868868141118849e-05,
+      "loss": 0.9068,
+      "step": 316820
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.868753348128874e-05,
+      "loss": 1.0522,
+      "step": 316830
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.8686385551388994e-05,
+      "loss": 0.9529,
+      "step": 316840
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.8685237621489244e-05,
+      "loss": 0.9657,
+      "step": 316850
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.86840896915895e-05,
+      "loss": 0.9876,
+      "step": 316860
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.868294176168975e-05,
+      "loss": 0.8874,
+      "step": 316870
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.868179383179e-05,
+      "loss": 1.0078,
+      "step": 316880
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.868064590189026e-05,
+      "loss": 0.9487,
+      "step": 316890
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.86794979719905e-05,
+      "loss": 0.9448,
+      "step": 316900
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.8678350042090757e-05,
+      "loss": 0.9816,
+      "step": 316910
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.867720211219101e-05,
+      "loss": 0.953,
+      "step": 316920
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.867605418229126e-05,
+      "loss": 0.8707,
+      "step": 316930
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.867490625239151e-05,
+      "loss": 0.9551,
+      "step": 316940
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.867375832249177e-05,
+      "loss": 0.9438,
+      "step": 316950
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.867261039259202e-05,
+      "loss": 1.1071,
+      "step": 316960
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.8671462462692276e-05,
+      "loss": 0.9848,
+      "step": 316970
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.8670314532792526e-05,
+      "loss": 1.0045,
+      "step": 316980
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.8669166602892776e-05,
+      "loss": 1.0786,
+      "step": 316990
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.866801867299303e-05,
+      "loss": 0.9338,
+      "step": 317000
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 3.866687074309329e-05,
+      "loss": 1.0057,
+      "step": 317010
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.866572281319353e-05,
+      "loss": 0.9378,
+      "step": 317020
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.866457488329379e-05,
+      "loss": 0.8401,
+      "step": 317030
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.8663426953394045e-05,
+      "loss": 0.9595,
+      "step": 317040
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.866227902349429e-05,
+      "loss": 0.9838,
+      "step": 317050
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.8661131093594545e-05,
+      "loss": 0.9984,
+      "step": 317060
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.86599831636948e-05,
+      "loss": 0.8545,
+      "step": 317070
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.865883523379505e-05,
+      "loss": 0.855,
+      "step": 317080
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.86576873038953e-05,
+      "loss": 0.9351,
+      "step": 317090
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.865653937399556e-05,
+      "loss": 0.9242,
+      "step": 317100
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.865539144409581e-05,
+      "loss": 0.9903,
+      "step": 317110
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.8654243514196064e-05,
+      "loss": 0.9541,
+      "step": 317120
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.8653095584296314e-05,
+      "loss": 0.8891,
+      "step": 317130
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.8651947654396564e-05,
+      "loss": 0.9262,
+      "step": 317140
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.865079972449682e-05,
+      "loss": 0.9385,
+      "step": 317150
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.864965179459708e-05,
+      "loss": 0.9739,
+      "step": 317160
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.864850386469732e-05,
+      "loss": 0.8797,
+      "step": 317170
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.864735593479758e-05,
+      "loss": 0.9274,
+      "step": 317180
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.8646208004897834e-05,
+      "loss": 1.0205,
+      "step": 317190
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.864506007499808e-05,
+      "loss": 0.8725,
+      "step": 317200
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.8643912145098334e-05,
+      "loss": 1.0162,
+      "step": 317210
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 3.864276421519859e-05,
+      "loss": 0.9986,
+      "step": 317220
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.864161628529884e-05,
+      "loss": 1.0373,
+      "step": 317230
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.864046835539909e-05,
+      "loss": 0.9427,
+      "step": 317240
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.8639320425499346e-05,
+      "loss": 0.9157,
+      "step": 317250
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.8638172495599596e-05,
+      "loss": 0.9755,
+      "step": 317260
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.863702456569985e-05,
+      "loss": 1.0154,
+      "step": 317270
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.86358766358001e-05,
+      "loss": 0.9566,
+      "step": 317280
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.863472870590035e-05,
+      "loss": 1.0126,
+      "step": 317290
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.863358077600061e-05,
+      "loss": 0.9236,
+      "step": 317300
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.8632432846100866e-05,
+      "loss": 1.0476,
+      "step": 317310
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.863128491620111e-05,
+      "loss": 0.9574,
+      "step": 317320
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.8630136986301366e-05,
+      "loss": 0.9734,
+      "step": 317330
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.862898905640162e-05,
+      "loss": 1.036,
+      "step": 317340
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.8627841126501865e-05,
+      "loss": 0.9458,
+      "step": 317350
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.862669319660212e-05,
+      "loss": 0.9629,
+      "step": 317360
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.862554526670238e-05,
+      "loss": 0.9402,
+      "step": 317370
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.862439733680263e-05,
+      "loss": 0.9923,
+      "step": 317380
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.862324940690288e-05,
+      "loss": 0.9593,
+      "step": 317390
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.8622101477003135e-05,
+      "loss": 1.03,
+      "step": 317400
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.8620953547103385e-05,
+      "loss": 0.9349,
+      "step": 317410
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.861980561720364e-05,
+      "loss": 0.9686,
+      "step": 317420
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.861865768730389e-05,
+      "loss": 0.9331,
+      "step": 317430
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 3.861750975740414e-05,
+      "loss": 0.9493,
+      "step": 317440
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.86163618275044e-05,
+      "loss": 0.8838,
+      "step": 317450
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.8615213897604654e-05,
+      "loss": 0.9729,
+      "step": 317460
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.86140659677049e-05,
+      "loss": 0.96,
+      "step": 317470
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.8612918037805154e-05,
+      "loss": 0.9378,
+      "step": 317480
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.861177010790541e-05,
+      "loss": 0.9741,
+      "step": 317490
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.8610622178005654e-05,
+      "loss": 1.0767,
+      "step": 317500
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.860947424810591e-05,
+      "loss": 0.8836,
+      "step": 317510
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.860832631820617e-05,
+      "loss": 0.949,
+      "step": 317520
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.860717838830642e-05,
+      "loss": 0.9007,
+      "step": 317530
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.860603045840667e-05,
+      "loss": 1.0093,
+      "step": 317540
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.8604882528506923e-05,
+      "loss": 0.8211,
+      "step": 317550
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.860373459860717e-05,
+      "loss": 0.8943,
+      "step": 317560
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.860258666870743e-05,
+      "loss": 0.8924,
+      "step": 317570
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.860143873880768e-05,
+      "loss": 0.97,
+      "step": 317580
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.860029080890793e-05,
+      "loss": 0.9344,
+      "step": 317590
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.8599142879008186e-05,
+      "loss": 0.938,
+      "step": 317600
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.859799494910844e-05,
+      "loss": 0.9519,
+      "step": 317610
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.8596847019208686e-05,
+      "loss": 0.8847,
+      "step": 317620
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.859569908930894e-05,
+      "loss": 0.9193,
+      "step": 317630
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.85945511594092e-05,
+      "loss": 0.9079,
+      "step": 317640
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.859340322950944e-05,
+      "loss": 0.903,
+      "step": 317650
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 3.85922552996097e-05,
+      "loss": 0.9686,
+      "step": 317660
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.8591107369709956e-05,
+      "loss": 0.9002,
+      "step": 317670
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.8589959439810205e-05,
+      "loss": 0.9596,
+      "step": 317680
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.8588811509910455e-05,
+      "loss": 0.9668,
+      "step": 317690
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.858766358001071e-05,
+      "loss": 0.9086,
+      "step": 317700
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.858651565011096e-05,
+      "loss": 0.9903,
+      "step": 317710
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.858536772021122e-05,
+      "loss": 0.9879,
+      "step": 317720
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.858421979031147e-05,
+      "loss": 1.1028,
+      "step": 317730
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.858307186041172e-05,
+      "loss": 1.0385,
+      "step": 317740
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.8581923930511975e-05,
+      "loss": 0.9163,
+      "step": 317750
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.858077600061223e-05,
+      "loss": 0.9229,
+      "step": 317760
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.8579628070712474e-05,
+      "loss": 0.943,
+      "step": 317770
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.857848014081273e-05,
+      "loss": 0.9983,
+      "step": 317780
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.857733221091299e-05,
+      "loss": 0.9734,
+      "step": 317790
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.857618428101323e-05,
+      "loss": 1.003,
+      "step": 317800
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.857503635111349e-05,
+      "loss": 0.9106,
+      "step": 317810
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.8573888421213744e-05,
+      "loss": 0.8643,
+      "step": 317820
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.8572740491313994e-05,
+      "loss": 1.0193,
+      "step": 317830
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.8571592561414244e-05,
+      "loss": 0.9966,
+      "step": 317840
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.85704446315145e-05,
+      "loss": 0.9771,
+      "step": 317850
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.856929670161475e-05,
+      "loss": 0.9865,
+      "step": 317860
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.856814877171501e-05,
+      "loss": 0.9547,
+      "step": 317870
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 3.856700084181526e-05,
+      "loss": 0.9647,
+      "step": 317880
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.8565852911915507e-05,
+      "loss": 0.9926,
+      "step": 317890
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.856470498201576e-05,
+      "loss": 0.9925,
+      "step": 317900
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.856355705211602e-05,
+      "loss": 0.9188,
+      "step": 317910
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.856240912221626e-05,
+      "loss": 0.9864,
+      "step": 317920
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.856126119231652e-05,
+      "loss": 0.9315,
+      "step": 317930
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.8560113262416776e-05,
+      "loss": 0.887,
+      "step": 317940
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.855896533251702e-05,
+      "loss": 0.907,
+      "step": 317950
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.8557817402617276e-05,
+      "loss": 1.0607,
+      "step": 317960
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.855666947271753e-05,
+      "loss": 0.992,
+      "step": 317970
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.855552154281778e-05,
+      "loss": 0.912,
+      "step": 317980
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.855437361291803e-05,
+      "loss": 0.9906,
+      "step": 317990
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.855322568301829e-05,
+      "loss": 0.8719,
+      "step": 318000
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.855207775311854e-05,
+      "loss": 0.9259,
+      "step": 318010
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.8550929823218795e-05,
+      "loss": 0.9713,
+      "step": 318020
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.8549781893319045e-05,
+      "loss": 0.9242,
+      "step": 318030
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.8548633963419295e-05,
+      "loss": 0.983,
+      "step": 318040
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.854748603351955e-05,
+      "loss": 0.9285,
+      "step": 318050
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.854633810361981e-05,
+      "loss": 0.9818,
+      "step": 318060
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.854519017372005e-05,
+      "loss": 0.8606,
+      "step": 318070
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.854404224382031e-05,
+      "loss": 0.9655,
+      "step": 318080
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.8542894313920565e-05,
+      "loss": 0.9574,
+      "step": 318090
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.854174638402081e-05,
+      "loss": 0.9036,
+      "step": 318100
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.8540598454121064e-05,
+      "loss": 1.0138,
+      "step": 318110
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.853945052422132e-05,
+      "loss": 0.9943,
+      "step": 318120
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.853830259432157e-05,
+      "loss": 0.9222,
+      "step": 318130
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.853715466442182e-05,
+      "loss": 0.9468,
+      "step": 318140
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.853600673452208e-05,
+      "loss": 1.1193,
+      "step": 318150
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.853485880462233e-05,
+      "loss": 1.0001,
+      "step": 318160
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.8533710874722584e-05,
+      "loss": 0.977,
+      "step": 318170
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.8532562944822834e-05,
+      "loss": 1.025,
+      "step": 318180
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.8531415014923084e-05,
+      "loss": 0.8789,
+      "step": 318190
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.853026708502334e-05,
+      "loss": 1.1239,
+      "step": 318200
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.85291191551236e-05,
+      "loss": 0.9578,
+      "step": 318210
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.852797122522384e-05,
+      "loss": 0.9165,
+      "step": 318220
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.8526823295324096e-05,
+      "loss": 0.9338,
+      "step": 318230
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.852567536542435e-05,
+      "loss": 0.9385,
+      "step": 318240
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.8524527435524596e-05,
+      "loss": 0.9911,
+      "step": 318250
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.852337950562485e-05,
+      "loss": 1.0308,
+      "step": 318260
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.85222315757251e-05,
+      "loss": 0.9423,
+      "step": 318270
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.852108364582536e-05,
+      "loss": 1.0679,
+      "step": 318280
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.851993571592561e-05,
+      "loss": 1.0632,
+      "step": 318290
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.851878778602586e-05,
+      "loss": 0.9024,
+      "step": 318300
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.8517639856126116e-05,
+      "loss": 0.9076,
+      "step": 318310
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.851649192622637e-05,
+      "loss": 0.9183,
+      "step": 318320
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.8515343996326615e-05,
+      "loss": 0.8461,
+      "step": 318330
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.851419606642687e-05,
+      "loss": 0.9178,
+      "step": 318340
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.851304813652713e-05,
+      "loss": 0.8672,
+      "step": 318350
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.851190020662737e-05,
+      "loss": 0.8574,
+      "step": 318360
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.851075227672763e-05,
+      "loss": 0.9612,
+      "step": 318370
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.8509604346827885e-05,
+      "loss": 0.9755,
+      "step": 318380
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.8508456416928135e-05,
+      "loss": 1.0515,
+      "step": 318390
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.8507308487028385e-05,
+      "loss": 0.8877,
+      "step": 318400
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.850616055712864e-05,
+      "loss": 0.9158,
+      "step": 318410
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.850501262722889e-05,
+      "loss": 0.9691,
+      "step": 318420
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.850386469732915e-05,
+      "loss": 0.9644,
+      "step": 318430
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.85027167674294e-05,
+      "loss": 1.0435,
+      "step": 318440
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.850156883752965e-05,
+      "loss": 0.9409,
+      "step": 318450
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.8500420907629904e-05,
+      "loss": 0.9689,
+      "step": 318460
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.849927297773016e-05,
+      "loss": 0.9481,
+      "step": 318470
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.8498125047830404e-05,
+      "loss": 0.8561,
+      "step": 318480
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.849697711793066e-05,
+      "loss": 0.957,
+      "step": 318490
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.849582918803092e-05,
+      "loss": 1.0524,
+      "step": 318500
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.849468125813116e-05,
+      "loss": 0.9693,
+      "step": 318510
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.849353332823142e-05,
+      "loss": 0.9556,
+      "step": 318520
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.8492385398331673e-05,
+      "loss": 0.9331,
+      "step": 318530
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.849123746843192e-05,
+      "loss": 0.9862,
+      "step": 318540
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.849008953853217e-05,
+      "loss": 0.9417,
+      "step": 318550
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.848894160863243e-05,
+      "loss": 0.9832,
+      "step": 318560
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.848779367873268e-05,
+      "loss": 0.8968,
+      "step": 318570
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.8486645748832936e-05,
+      "loss": 1.002,
+      "step": 318580
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.8485497818933186e-05,
+      "loss": 0.9587,
+      "step": 318590
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.8484349889033436e-05,
+      "loss": 0.9851,
+      "step": 318600
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.848320195913369e-05,
+      "loss": 1.0438,
+      "step": 318610
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.848205402923395e-05,
+      "loss": 0.9688,
+      "step": 318620
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.848090609933419e-05,
+      "loss": 0.997,
+      "step": 318630
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.847975816943445e-05,
+      "loss": 0.905,
+      "step": 318640
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.8478610239534706e-05,
+      "loss": 1.0142,
+      "step": 318650
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.847746230963495e-05,
+      "loss": 1.0198,
+      "step": 318660
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.8476314379735205e-05,
+      "loss": 0.9811,
+      "step": 318670
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.847516644983546e-05,
+      "loss": 0.9843,
+      "step": 318680
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.847401851993571e-05,
+      "loss": 1.0165,
+      "step": 318690
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.847287059003596e-05,
+      "loss": 0.8482,
+      "step": 318700
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.847172266013622e-05,
+      "loss": 1.0476,
+      "step": 318710
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.847057473023647e-05,
+      "loss": 0.9704,
+      "step": 318720
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.8469426800336725e-05,
+      "loss": 0.9081,
+      "step": 318730
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.8468278870436975e-05,
+      "loss": 0.9729,
+      "step": 318740
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.8467130940537224e-05,
+      "loss": 0.9933,
+      "step": 318750
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.846598301063748e-05,
+      "loss": 0.9828,
+      "step": 318760
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.846483508073774e-05,
+      "loss": 0.8657,
+      "step": 318770
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.846368715083798e-05,
+      "loss": 1.028,
+      "step": 318780
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.846253922093824e-05,
+      "loss": 0.8161,
+      "step": 318790
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.8461391291038494e-05,
+      "loss": 1.0481,
+      "step": 318800
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.846024336113874e-05,
+      "loss": 0.9297,
+      "step": 318810
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.8459095431238994e-05,
+      "loss": 0.9509,
+      "step": 318820
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.845794750133925e-05,
+      "loss": 0.8942,
+      "step": 318830
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.84567995714395e-05,
+      "loss": 0.9277,
+      "step": 318840
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.845565164153975e-05,
+      "loss": 0.943,
+      "step": 318850
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.845450371164001e-05,
+      "loss": 0.8848,
+      "step": 318860
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.8453355781740257e-05,
+      "loss": 0.958,
+      "step": 318870
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.845220785184051e-05,
+      "loss": 1.0501,
+      "step": 318880
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.845105992194076e-05,
+      "loss": 1.0729,
+      "step": 318890
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.844991199204101e-05,
+      "loss": 0.8724,
+      "step": 318900
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.844876406214127e-05,
+      "loss": 0.9041,
+      "step": 318910
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.8447616132241526e-05,
+      "loss": 0.9656,
+      "step": 318920
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.844646820234177e-05,
+      "loss": 0.9928,
+      "step": 318930
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.8445320272442026e-05,
+      "loss": 0.9763,
+      "step": 318940
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.844417234254228e-05,
+      "loss": 0.9559,
+      "step": 318950
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.8443024412642526e-05,
+      "loss": 0.8868,
+      "step": 318960
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.844187648274278e-05,
+      "loss": 0.9912,
+      "step": 318970
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.844072855284304e-05,
+      "loss": 0.952,
+      "step": 318980
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.843958062294329e-05,
+      "loss": 1.03,
+      "step": 318990
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.843843269304354e-05,
+      "loss": 0.8965,
+      "step": 319000
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.8437284763143795e-05,
+      "loss": 0.9224,
+      "step": 319010
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.8436136833244045e-05,
+      "loss": 0.8903,
+      "step": 319020
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.84349889033443e-05,
+      "loss": 1.0026,
+      "step": 319030
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.843384097344455e-05,
+      "loss": 0.8941,
+      "step": 319040
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.84326930435448e-05,
+      "loss": 0.9169,
+      "step": 319050
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.843154511364506e-05,
+      "loss": 1.046,
+      "step": 319060
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.8430397183745315e-05,
+      "loss": 0.8595,
+      "step": 319070
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.842924925384556e-05,
+      "loss": 0.9261,
+      "step": 319080
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.8428101323945814e-05,
+      "loss": 1.0908,
+      "step": 319090
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.842695339404607e-05,
+      "loss": 1.0469,
+      "step": 319100
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.8425805464146314e-05,
+      "loss": 0.9972,
+      "step": 319110
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.842465753424657e-05,
+      "loss": 1.0497,
+      "step": 319120
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.842350960434683e-05,
+      "loss": 0.9497,
+      "step": 319130
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.842236167444708e-05,
+      "loss": 0.9076,
+      "step": 319140
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.842121374454733e-05,
+      "loss": 0.9936,
+      "step": 319150
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.8420065814647584e-05,
+      "loss": 0.9456,
+      "step": 319160
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.8418917884747834e-05,
+      "loss": 0.9838,
+      "step": 319170
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.841776995484809e-05,
+      "loss": 0.9622,
+      "step": 319180
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.841662202494834e-05,
+      "loss": 0.9215,
+      "step": 319190
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.841547409504859e-05,
+      "loss": 0.9739,
+      "step": 319200
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.8414326165148846e-05,
+      "loss": 0.9971,
+      "step": 319210
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.84131782352491e-05,
+      "loss": 0.9445,
+      "step": 319220
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.8412030305349346e-05,
+      "loss": 0.9268,
+      "step": 319230
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.84108823754496e-05,
+      "loss": 1.0113,
+      "step": 319240
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.840973444554986e-05,
+      "loss": 1.0049,
+      "step": 319250
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.84085865156501e-05,
+      "loss": 0.9076,
+      "step": 319260
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.840743858575036e-05,
+      "loss": 0.928,
+      "step": 319270
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.8406290655850616e-05,
+      "loss": 0.9992,
+      "step": 319280
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.8405142725950866e-05,
+      "loss": 0.8386,
+      "step": 319290
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.8403994796051115e-05,
+      "loss": 1.0171,
+      "step": 319300
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.840284686615137e-05,
+      "loss": 0.9287,
+      "step": 319310
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.840169893625162e-05,
+      "loss": 0.9491,
+      "step": 319320
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.840055100635188e-05,
+      "loss": 0.9175,
+      "step": 319330
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.839940307645213e-05,
+      "loss": 1.008,
+      "step": 319340
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.839825514655238e-05,
+      "loss": 0.9786,
+      "step": 319350
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.8397107216652635e-05,
+      "loss": 0.9384,
+      "step": 319360
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.839595928675289e-05,
+      "loss": 0.9289,
+      "step": 319370
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.8394811356853135e-05,
+      "loss": 1.0103,
+      "step": 319380
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.839366342695339e-05,
+      "loss": 0.9795,
+      "step": 319390
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 3.839251549705365e-05,
+      "loss": 0.9484,
+      "step": 319400
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.839136756715389e-05,
+      "loss": 1.0399,
+      "step": 319410
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.839021963725415e-05,
+      "loss": 0.8855,
+      "step": 319420
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.8389071707354404e-05,
+      "loss": 0.9227,
+      "step": 319430
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.8387923777454654e-05,
+      "loss": 0.9933,
+      "step": 319440
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.8386775847554904e-05,
+      "loss": 0.9462,
+      "step": 319450
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.838562791765516e-05,
+      "loss": 0.9755,
+      "step": 319460
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.838447998775541e-05,
+      "loss": 1.0617,
+      "step": 319470
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.838333205785567e-05,
+      "loss": 0.9423,
+      "step": 319480
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.838218412795592e-05,
+      "loss": 0.9327,
+      "step": 319490
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.838103619805617e-05,
+      "loss": 0.9146,
+      "step": 319500
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.8379888268156423e-05,
+      "loss": 0.9311,
+      "step": 319510
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.837874033825668e-05,
+      "loss": 0.873,
+      "step": 319520
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.837759240835692e-05,
+      "loss": 1.0556,
+      "step": 319530
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.837644447845718e-05,
+      "loss": 1.0502,
+      "step": 319540
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.8375296548557436e-05,
+      "loss": 0.8532,
+      "step": 319550
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.837414861865768e-05,
+      "loss": 0.9881,
+      "step": 319560
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.8373000688757936e-05,
+      "loss": 0.9207,
+      "step": 319570
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.837185275885819e-05,
+      "loss": 0.9935,
+      "step": 319580
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.837070482895844e-05,
+      "loss": 0.8892,
+      "step": 319590
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.836955689905869e-05,
+      "loss": 0.9589,
+      "step": 319600
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.836840896915895e-05,
+      "loss": 0.9891,
+      "step": 319610
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.83672610392592e-05,
+      "loss": 0.9585,
+      "step": 319620
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.8366113109359456e-05,
+      "loss": 0.9041,
+      "step": 319630
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.83649651794597e-05,
+      "loss": 0.9201,
+      "step": 319640
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.8363817249559955e-05,
+      "loss": 0.9936,
+      "step": 319650
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.836266931966021e-05,
+      "loss": 0.8478,
+      "step": 319660
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.8361521389760455e-05,
+      "loss": 0.9302,
+      "step": 319670
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.836037345986071e-05,
+      "loss": 1.0031,
+      "step": 319680
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.835922552996097e-05,
+      "loss": 0.8349,
+      "step": 319690
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.835807760006122e-05,
+      "loss": 1.0854,
+      "step": 319700
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.835692967016147e-05,
+      "loss": 0.9601,
+      "step": 319710
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.8355781740261725e-05,
+      "loss": 0.8821,
+      "step": 319720
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.8354633810361974e-05,
+      "loss": 0.9892,
+      "step": 319730
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.835348588046223e-05,
+      "loss": 0.9985,
+      "step": 319740
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.835233795056248e-05,
+      "loss": 1.0035,
+      "step": 319750
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.835119002066273e-05,
+      "loss": 0.9779,
+      "step": 319760
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.835004209076299e-05,
+      "loss": 0.9691,
+      "step": 319770
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.8348894160863244e-05,
+      "loss": 0.9866,
+      "step": 319780
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.834774623096349e-05,
+      "loss": 1.0249,
+      "step": 319790
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.8346598301063744e-05,
+      "loss": 0.9567,
+      "step": 319800
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.8345450371164e-05,
+      "loss": 0.7707,
+      "step": 319810
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.8344302441264243e-05,
+      "loss": 0.8905,
+      "step": 319820
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.83431545113645e-05,
+      "loss": 1.0256,
+      "step": 319830
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 3.834200658146476e-05,
+      "loss": 0.9922,
+      "step": 319840
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.8340858651565007e-05,
+      "loss": 0.8913,
+      "step": 319850
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.8339710721665256e-05,
+      "loss": 0.8658,
+      "step": 319860
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.833856279176551e-05,
+      "loss": 1.0677,
+      "step": 319870
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.833741486186576e-05,
+      "loss": 0.996,
+      "step": 319880
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.833626693196602e-05,
+      "loss": 0.9106,
+      "step": 319890
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.833511900206627e-05,
+      "loss": 0.9581,
+      "step": 319900
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.833397107216652e-05,
+      "loss": 0.95,
+      "step": 319910
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.8332823142266776e-05,
+      "loss": 0.8736,
+      "step": 319920
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.833167521236703e-05,
+      "loss": 0.965,
+      "step": 319930
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.8330527282467276e-05,
+      "loss": 0.9384,
+      "step": 319940
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.832937935256753e-05,
+      "loss": 0.8242,
+      "step": 319950
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.832823142266779e-05,
+      "loss": 1.0087,
+      "step": 319960
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.832708349276803e-05,
+      "loss": 0.8436,
+      "step": 319970
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.832593556286829e-05,
+      "loss": 1.0114,
+      "step": 319980
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.8324787632968545e-05,
+      "loss": 1.0112,
+      "step": 319990
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.8323639703068795e-05,
+      "loss": 1.004,
+      "step": 320000
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.8322491773169045e-05,
+      "loss": 1.0626,
+      "step": 320010
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.83213438432693e-05,
+      "loss": 0.9227,
+      "step": 320020
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.832019591336955e-05,
+      "loss": 0.9359,
+      "step": 320030
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.831904798346981e-05,
+      "loss": 0.9784,
+      "step": 320040
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.831790005357006e-05,
+      "loss": 0.9716,
+      "step": 320050
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 3.831675212367031e-05,
+      "loss": 1.0854,
+      "step": 320060
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.8315604193770564e-05,
+      "loss": 0.9277,
+      "step": 320070
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.831445626387082e-05,
+      "loss": 0.9519,
+      "step": 320080
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.8313308333971064e-05,
+      "loss": 0.8157,
+      "step": 320090
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.831216040407132e-05,
+      "loss": 0.8187,
+      "step": 320100
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.831101247417158e-05,
+      "loss": 0.8755,
+      "step": 320110
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.830986454427182e-05,
+      "loss": 0.9704,
+      "step": 320120
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.830871661437208e-05,
+      "loss": 1.0318,
+      "step": 320130
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.8307568684472334e-05,
+      "loss": 0.9231,
+      "step": 320140
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.8306420754572583e-05,
+      "loss": 0.9641,
+      "step": 320150
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.830527282467283e-05,
+      "loss": 1.045,
+      "step": 320160
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.830412489477309e-05,
+      "loss": 0.8968,
+      "step": 320170
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.830297696487334e-05,
+      "loss": 0.9556,
+      "step": 320180
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.8301829034973596e-05,
+      "loss": 1.0177,
+      "step": 320190
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.8300681105073846e-05,
+      "loss": 0.9363,
+      "step": 320200
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.8299533175174096e-05,
+      "loss": 1.0616,
+      "step": 320210
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.829838524527435e-05,
+      "loss": 1.046,
+      "step": 320220
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.829723731537461e-05,
+      "loss": 0.9238,
+      "step": 320230
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.829608938547485e-05,
+      "loss": 0.9536,
+      "step": 320240
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.829494145557511e-05,
+      "loss": 1.0023,
+      "step": 320250
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.8293793525675366e-05,
+      "loss": 0.9081,
+      "step": 320260
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 3.829264559577561e-05,
+      "loss": 1.0185,
+      "step": 320270
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.8291497665875865e-05,
+      "loss": 0.9126,
+      "step": 320280
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.829034973597612e-05,
+      "loss": 0.9061,
+      "step": 320290
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.828920180607637e-05,
+      "loss": 0.9623,
+      "step": 320300
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.828805387617662e-05,
+      "loss": 0.8613,
+      "step": 320310
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.828690594627688e-05,
+      "loss": 0.8661,
+      "step": 320320
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.828575801637713e-05,
+      "loss": 0.9446,
+      "step": 320330
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.8284610086477385e-05,
+      "loss": 0.8831,
+      "step": 320340
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.8283462156577635e-05,
+      "loss": 0.9145,
+      "step": 320350
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.8282314226677885e-05,
+      "loss": 1.0273,
+      "step": 320360
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.828116629677814e-05,
+      "loss": 0.9546,
+      "step": 320370
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.82800183668784e-05,
+      "loss": 0.924,
+      "step": 320380
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.827887043697864e-05,
+      "loss": 1.0013,
+      "step": 320390
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.82777225070789e-05,
+      "loss": 0.9521,
+      "step": 320400
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.8276574577179154e-05,
+      "loss": 1.0812,
+      "step": 320410
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.82754266472794e-05,
+      "loss": 0.951,
+      "step": 320420
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.8274278717379654e-05,
+      "loss": 1.0359,
+      "step": 320430
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.827313078747991e-05,
+      "loss": 0.9469,
+      "step": 320440
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.827198285758016e-05,
+      "loss": 0.9635,
+      "step": 320450
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.827083492768041e-05,
+      "loss": 0.8957,
+      "step": 320460
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.826968699778067e-05,
+      "loss": 1.0033,
+      "step": 320470
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.826853906788092e-05,
+      "loss": 0.9514,
+      "step": 320480
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 3.8267391137981173e-05,
+      "loss": 0.9543,
+      "step": 320490
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.826624320808142e-05,
+      "loss": 0.9895,
+      "step": 320500
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.826509527818167e-05,
+      "loss": 0.9333,
+      "step": 320510
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.826394734828193e-05,
+      "loss": 0.9603,
+      "step": 320520
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.8262799418382186e-05,
+      "loss": 0.9395,
+      "step": 320530
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.826165148848243e-05,
+      "loss": 0.9255,
+      "step": 320540
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.8260503558582686e-05,
+      "loss": 0.9659,
+      "step": 320550
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.825935562868294e-05,
+      "loss": 0.9995,
+      "step": 320560
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.8258207698783186e-05,
+      "loss": 1.0204,
+      "step": 320570
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.825705976888344e-05,
+      "loss": 1.0568,
+      "step": 320580
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.82559118389837e-05,
+      "loss": 0.9716,
+      "step": 320590
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.825476390908395e-05,
+      "loss": 0.9017,
+      "step": 320600
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.82536159791842e-05,
+      "loss": 1.0228,
+      "step": 320610
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.8252468049284455e-05,
+      "loss": 0.9539,
+      "step": 320620
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.8251320119384705e-05,
+      "loss": 0.9172,
+      "step": 320630
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.825017218948496e-05,
+      "loss": 0.9419,
+      "step": 320640
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.824902425958521e-05,
+      "loss": 0.9884,
+      "step": 320650
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.824787632968546e-05,
+      "loss": 0.9142,
+      "step": 320660
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.824672839978572e-05,
+      "loss": 0.9065,
+      "step": 320670
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.8245580469885975e-05,
+      "loss": 0.9451,
+      "step": 320680
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.824443253998622e-05,
+      "loss": 0.9026,
+      "step": 320690
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.8243284610086475e-05,
+      "loss": 0.9438,
+      "step": 320700
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 3.824213668018673e-05,
+      "loss": 0.9705,
+      "step": 320710
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.8240988750286974e-05,
+      "loss": 0.9237,
+      "step": 320720
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.823984082038723e-05,
+      "loss": 0.942,
+      "step": 320730
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.823869289048749e-05,
+      "loss": 0.9354,
+      "step": 320740
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.823754496058774e-05,
+      "loss": 0.8704,
+      "step": 320750
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.823639703068799e-05,
+      "loss": 0.9901,
+      "step": 320760
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.8235249100788244e-05,
+      "loss": 0.9989,
+      "step": 320770
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.8234101170888494e-05,
+      "loss": 0.9946,
+      "step": 320780
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.823295324098875e-05,
+      "loss": 0.9638,
+      "step": 320790
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.8231805311089e-05,
+      "loss": 0.965,
+      "step": 320800
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.823065738118925e-05,
+      "loss": 0.9246,
+      "step": 320810
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.822950945128951e-05,
+      "loss": 0.9612,
+      "step": 320820
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.822836152138976e-05,
+      "loss": 0.9853,
+      "step": 320830
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.8227213591490006e-05,
+      "loss": 0.975,
+      "step": 320840
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.822606566159026e-05,
+      "loss": 0.8604,
+      "step": 320850
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.822491773169052e-05,
+      "loss": 1.056,
+      "step": 320860
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.822376980179076e-05,
+      "loss": 1.0187,
+      "step": 320870
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.822262187189102e-05,
+      "loss": 0.991,
+      "step": 320880
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.8221473941991276e-05,
+      "loss": 0.9186,
+      "step": 320890
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.8220326012091526e-05,
+      "loss": 1.0136,
+      "step": 320900
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.8219178082191776e-05,
+      "loss": 0.9383,
+      "step": 320910
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.821803015229203e-05,
+      "loss": 0.9697,
+      "step": 320920
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 3.821688222239228e-05,
+      "loss": 0.9451,
+      "step": 320930
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.821573429249254e-05,
+      "loss": 0.9195,
+      "step": 320940
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.821458636259279e-05,
+      "loss": 0.9153,
+      "step": 320950
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.821343843269304e-05,
+      "loss": 1.0529,
+      "step": 320960
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.8212290502793295e-05,
+      "loss": 0.9638,
+      "step": 320970
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.821114257289355e-05,
+      "loss": 0.8996,
+      "step": 320980
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.8209994642993795e-05,
+      "loss": 1.0059,
+      "step": 320990
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.820884671309405e-05,
+      "loss": 0.9752,
+      "step": 321000
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.82076987831943e-05,
+      "loss": 0.9851,
+      "step": 321010
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.820655085329455e-05,
+      "loss": 0.992,
+      "step": 321020
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.820540292339481e-05,
+      "loss": 0.9866,
+      "step": 321030
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.820425499349506e-05,
+      "loss": 0.9587,
+      "step": 321040
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.8203107063595314e-05,
+      "loss": 0.8605,
+      "step": 321050
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.8201959133695564e-05,
+      "loss": 0.8984,
+      "step": 321060
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.8200811203795814e-05,
+      "loss": 0.9027,
+      "step": 321070
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.819966327389607e-05,
+      "loss": 0.9497,
+      "step": 321080
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.819851534399633e-05,
+      "loss": 0.985,
+      "step": 321090
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.819736741409657e-05,
+      "loss": 1.0461,
+      "step": 321100
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.819621948419683e-05,
+      "loss": 0.971,
+      "step": 321110
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.8195071554297084e-05,
+      "loss": 1.0335,
+      "step": 321120
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.819392362439733e-05,
+      "loss": 0.905,
+      "step": 321130
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.819277569449758e-05,
+      "loss": 0.8782,
+      "step": 321140
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 3.819162776459784e-05,
+      "loss": 0.9194,
+      "step": 321150
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.819047983469809e-05,
+      "loss": 0.9672,
+      "step": 321160
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.818933190479834e-05,
+      "loss": 0.9512,
+      "step": 321170
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.8188183974898596e-05,
+      "loss": 1.0713,
+      "step": 321180
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.8187036044998846e-05,
+      "loss": 0.9614,
+      "step": 321190
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.81858881150991e-05,
+      "loss": 0.9259,
+      "step": 321200
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.818474018519935e-05,
+      "loss": 0.8895,
+      "step": 321210
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.81835922552996e-05,
+      "loss": 0.9832,
+      "step": 321220
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.818244432539986e-05,
+      "loss": 1.0249,
+      "step": 321230
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.8181296395500116e-05,
+      "loss": 0.9847,
+      "step": 321240
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.818014846560036e-05,
+      "loss": 0.9901,
+      "step": 321250
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.8179000535700615e-05,
+      "loss": 1.018,
+      "step": 321260
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.817785260580087e-05,
+      "loss": 0.9488,
+      "step": 321270
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.8176704675901115e-05,
+      "loss": 0.951,
+      "step": 321280
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.817555674600137e-05,
+      "loss": 0.9549,
+      "step": 321290
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.817440881610163e-05,
+      "loss": 0.9075,
+      "step": 321300
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.817326088620188e-05,
+      "loss": 0.9067,
+      "step": 321310
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.817211295630213e-05,
+      "loss": 0.9916,
+      "step": 321320
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.8170965026402385e-05,
+      "loss": 0.9419,
+      "step": 321330
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.8169817096502635e-05,
+      "loss": 0.9605,
+      "step": 321340
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.816866916660289e-05,
+      "loss": 1.0472,
+      "step": 321350
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 3.816752123670314e-05,
+      "loss": 0.9029,
+      "step": 321360
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.816637330680339e-05,
+      "loss": 0.8823,
+      "step": 321370
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.816522537690365e-05,
+      "loss": 0.9614,
+      "step": 321380
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.8164077447003904e-05,
+      "loss": 1.1234,
+      "step": 321390
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.816292951710415e-05,
+      "loss": 0.9525,
+      "step": 321400
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.8161781587204404e-05,
+      "loss": 0.9042,
+      "step": 321410
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.816063365730466e-05,
+      "loss": 1.0352,
+      "step": 321420
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.8159485727404904e-05,
+      "loss": 0.9652,
+      "step": 321430
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.815833779750516e-05,
+      "loss": 0.9709,
+      "step": 321440
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.815718986760542e-05,
+      "loss": 0.8974,
+      "step": 321450
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.815604193770567e-05,
+      "loss": 0.9872,
+      "step": 321460
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.815489400780592e-05,
+      "loss": 0.921,
+      "step": 321470
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.815374607790617e-05,
+      "loss": 0.9928,
+      "step": 321480
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.815259814800642e-05,
+      "loss": 0.9983,
+      "step": 321490
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.815145021810668e-05,
+      "loss": 0.872,
+      "step": 321500
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.815030228820693e-05,
+      "loss": 0.9306,
+      "step": 321510
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.814915435830718e-05,
+      "loss": 0.9056,
+      "step": 321520
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.8148006428407436e-05,
+      "loss": 0.9986,
+      "step": 321530
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.814685849850769e-05,
+      "loss": 0.9885,
+      "step": 321540
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.8145710568607936e-05,
+      "loss": 0.9509,
+      "step": 321550
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.814456263870819e-05,
+      "loss": 0.9275,
+      "step": 321560
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.814341470880845e-05,
+      "loss": 0.9744,
+      "step": 321570
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 3.814226677890869e-05,
+      "loss": 0.8927,
+      "step": 321580
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.814111884900895e-05,
+      "loss": 0.9523,
+      "step": 321590
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.8139970919109205e-05,
+      "loss": 0.9647,
+      "step": 321600
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.8138822989209455e-05,
+      "loss": 0.8607,
+      "step": 321610
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.8137675059309705e-05,
+      "loss": 0.9413,
+      "step": 321620
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.813652712940996e-05,
+      "loss": 0.9625,
+      "step": 321630
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.813537919951021e-05,
+      "loss": 0.9182,
+      "step": 321640
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.813423126961047e-05,
+      "loss": 0.9292,
+      "step": 321650
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.813308333971072e-05,
+      "loss": 0.9573,
+      "step": 321660
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.813193540981097e-05,
+      "loss": 0.9621,
+      "step": 321670
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.8130787479911225e-05,
+      "loss": 0.924,
+      "step": 321680
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.812963955001148e-05,
+      "loss": 0.8888,
+      "step": 321690
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.8128491620111724e-05,
+      "loss": 0.9257,
+      "step": 321700
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.812734369021198e-05,
+      "loss": 0.8903,
+      "step": 321710
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.812619576031224e-05,
+      "loss": 0.8658,
+      "step": 321720
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.812504783041248e-05,
+      "loss": 1.0165,
+      "step": 321730
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.812389990051274e-05,
+      "loss": 0.8891,
+      "step": 321740
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.8122751970612994e-05,
+      "loss": 0.9041,
+      "step": 321750
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.8121604040713244e-05,
+      "loss": 1.0216,
+      "step": 321760
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.8120456110813494e-05,
+      "loss": 0.8405,
+      "step": 321770
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.811930818091375e-05,
+      "loss": 1.0628,
+      "step": 321780
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.8118160251014e-05,
+      "loss": 1.0163,
+      "step": 321790
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 3.811701232111426e-05,
+      "loss": 0.9199,
+      "step": 321800
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.8115864391214507e-05,
+      "loss": 0.9306,
+      "step": 321810
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.8114716461314756e-05,
+      "loss": 0.9972,
+      "step": 321820
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.811356853141501e-05,
+      "loss": 0.8811,
+      "step": 321830
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.811242060151527e-05,
+      "loss": 0.9305,
+      "step": 321840
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.811127267161551e-05,
+      "loss": 0.8586,
+      "step": 321850
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.811012474171577e-05,
+      "loss": 0.9007,
+      "step": 321860
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.8108976811816026e-05,
+      "loss": 0.9417,
+      "step": 321870
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.810782888191627e-05,
+      "loss": 0.9247,
+      "step": 321880
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.8106680952016526e-05,
+      "loss": 0.9515,
+      "step": 321890
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.810553302211678e-05,
+      "loss": 0.9425,
+      "step": 321900
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.810438509221703e-05,
+      "loss": 0.9292,
+      "step": 321910
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.810323716231728e-05,
+      "loss": 0.8446,
+      "step": 321920
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.810208923241754e-05,
+      "loss": 1.0408,
+      "step": 321930
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.810094130251779e-05,
+      "loss": 0.967,
+      "step": 321940
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.8099793372618045e-05,
+      "loss": 0.9332,
+      "step": 321950
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.8098645442718295e-05,
+      "loss": 1.0009,
+      "step": 321960
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.8097497512818545e-05,
+      "loss": 1.0008,
+      "step": 321970
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.80963495829188e-05,
+      "loss": 1.0059,
+      "step": 321980
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.809520165301906e-05,
+      "loss": 0.9622,
+      "step": 321990
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.80940537231193e-05,
+      "loss": 0.8711,
+      "step": 322000
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.809290579321956e-05,
+      "loss": 0.9635,
+      "step": 322010
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 3.8091757863319814e-05,
+      "loss": 1.009,
+      "step": 322020
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.809060993342006e-05,
+      "loss": 0.9649,
+      "step": 322030
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.8089462003520314e-05,
+      "loss": 0.9975,
+      "step": 322040
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.808831407362057e-05,
+      "loss": 0.9457,
+      "step": 322050
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.808716614372082e-05,
+      "loss": 1.007,
+      "step": 322060
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.808601821382107e-05,
+      "loss": 0.9097,
+      "step": 322070
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.808487028392133e-05,
+      "loss": 1.0395,
+      "step": 322080
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.808372235402158e-05,
+      "loss": 0.9775,
+      "step": 322090
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.8082574424121834e-05,
+      "loss": 0.9515,
+      "step": 322100
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.8081426494222083e-05,
+      "loss": 0.9855,
+      "step": 322110
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.808027856432233e-05,
+      "loss": 1.0299,
+      "step": 322120
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.807913063442259e-05,
+      "loss": 0.9646,
+      "step": 322130
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.8077982704522847e-05,
+      "loss": 0.9066,
+      "step": 322140
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.807683477462309e-05,
+      "loss": 1.0018,
+      "step": 322150
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.8075686844723346e-05,
+      "loss": 1.0222,
+      "step": 322160
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.80745389148236e-05,
+      "loss": 0.9872,
+      "step": 322170
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.8073390984923846e-05,
+      "loss": 1.0293,
+      "step": 322180
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.80722430550241e-05,
+      "loss": 0.9188,
+      "step": 322190
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.807109512512436e-05,
+      "loss": 0.9439,
+      "step": 322200
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.806994719522461e-05,
+      "loss": 0.9327,
+      "step": 322210
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.806879926532486e-05,
+      "loss": 1.0108,
+      "step": 322220
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.8067651335425116e-05,
+      "loss": 0.9224,
+      "step": 322230
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 3.8066503405525365e-05,
+      "loss": 0.9606,
+      "step": 322240
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.806535547562562e-05,
+      "loss": 0.859,
+      "step": 322250
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.806420754572587e-05,
+      "loss": 0.9415,
+      "step": 322260
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.806305961582612e-05,
+      "loss": 0.8076,
+      "step": 322270
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.806191168592638e-05,
+      "loss": 0.8899,
+      "step": 322280
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.8060763756026635e-05,
+      "loss": 0.927,
+      "step": 322290
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.805961582612688e-05,
+      "loss": 0.876,
+      "step": 322300
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.8058467896227135e-05,
+      "loss": 0.9022,
+      "step": 322310
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.805731996632739e-05,
+      "loss": 0.925,
+      "step": 322320
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.8056172036427634e-05,
+      "loss": 0.944,
+      "step": 322330
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.805502410652789e-05,
+      "loss": 0.8339,
+      "step": 322340
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.805387617662815e-05,
+      "loss": 0.97,
+      "step": 322350
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.80527282467284e-05,
+      "loss": 0.937,
+      "step": 322360
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.805158031682865e-05,
+      "loss": 1.0556,
+      "step": 322370
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.80504323869289e-05,
+      "loss": 0.9753,
+      "step": 322380
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.8049284457029154e-05,
+      "loss": 0.89,
+      "step": 322390
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.804813652712941e-05,
+      "loss": 1.0064,
+      "step": 322400
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.8046988597229654e-05,
+      "loss": 0.9786,
+      "step": 322410
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.804584066732991e-05,
+      "loss": 1.0051,
+      "step": 322420
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.804469273743017e-05,
+      "loss": 0.894,
+      "step": 322430
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.804354480753041e-05,
+      "loss": 0.9644,
+      "step": 322440
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 3.804239687763067e-05,
+      "loss": 0.9734,
+      "step": 322450
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.804124894773092e-05,
+      "loss": 0.9255,
+      "step": 322460
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.804010101783117e-05,
+      "loss": 0.8604,
+      "step": 322470
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.803895308793142e-05,
+      "loss": 0.9055,
+      "step": 322480
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.803780515803168e-05,
+      "loss": 0.9599,
+      "step": 322490
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.803665722813193e-05,
+      "loss": 0.9833,
+      "step": 322500
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.8035509298232186e-05,
+      "loss": 0.9275,
+      "step": 322510
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.8034361368332436e-05,
+      "loss": 0.953,
+      "step": 322520
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.8033213438432686e-05,
+      "loss": 1.0092,
+      "step": 322530
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.803206550853294e-05,
+      "loss": 0.9535,
+      "step": 322540
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.80309175786332e-05,
+      "loss": 1.0139,
+      "step": 322550
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.802976964873344e-05,
+      "loss": 0.8798,
+      "step": 322560
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.80286217188337e-05,
+      "loss": 1.0656,
+      "step": 322570
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.8027473788933955e-05,
+      "loss": 0.9696,
+      "step": 322580
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.80263258590342e-05,
+      "loss": 0.9452,
+      "step": 322590
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.8025177929134455e-05,
+      "loss": 0.8776,
+      "step": 322600
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.802402999923471e-05,
+      "loss": 0.9635,
+      "step": 322610
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.802288206933496e-05,
+      "loss": 0.9919,
+      "step": 322620
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.802173413943521e-05,
+      "loss": 0.9221,
+      "step": 322630
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.802058620953547e-05,
+      "loss": 0.9801,
+      "step": 322640
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.801943827963572e-05,
+      "loss": 0.9812,
+      "step": 322650
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.8018290349735975e-05,
+      "loss": 0.974,
+      "step": 322660
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 3.8017142419836224e-05,
+      "loss": 0.9497,
+      "step": 322670
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.8015994489936474e-05,
+      "loss": 1.0555,
+      "step": 322680
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.801484656003673e-05,
+      "loss": 0.955,
+      "step": 322690
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.801369863013699e-05,
+      "loss": 0.9502,
+      "step": 322700
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.801255070023723e-05,
+      "loss": 0.9742,
+      "step": 322710
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.801140277033749e-05,
+      "loss": 0.9622,
+      "step": 322720
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.8010254840437744e-05,
+      "loss": 0.8983,
+      "step": 322730
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.800910691053799e-05,
+      "loss": 0.9374,
+      "step": 322740
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.8007958980638244e-05,
+      "loss": 0.9479,
+      "step": 322750
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.80068110507385e-05,
+      "loss": 0.9451,
+      "step": 322760
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.800566312083875e-05,
+      "loss": 0.8852,
+      "step": 322770
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.8004515190939e-05,
+      "loss": 1.0731,
+      "step": 322780
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.8003367261039257e-05,
+      "loss": 0.9515,
+      "step": 322790
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.8002219331139506e-05,
+      "loss": 0.9654,
+      "step": 322800
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.800107140123976e-05,
+      "loss": 0.9186,
+      "step": 322810
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.799992347134001e-05,
+      "loss": 1.0154,
+      "step": 322820
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.799877554144026e-05,
+      "loss": 1.0231,
+      "step": 322830
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.799762761154052e-05,
+      "loss": 0.9086,
+      "step": 322840
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.7996479681640776e-05,
+      "loss": 0.9607,
+      "step": 322850
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.799533175174102e-05,
+      "loss": 1.0843,
+      "step": 322860
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.7994183821841276e-05,
+      "loss": 1.1097,
+      "step": 322870
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.799303589194153e-05,
+      "loss": 0.9409,
+      "step": 322880
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 3.7991887962041775e-05,
+      "loss": 0.9803,
+      "step": 322890
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.799074003214203e-05,
+      "loss": 0.9351,
+      "step": 322900
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.798959210224229e-05,
+      "loss": 0.9598,
+      "step": 322910
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.798844417234254e-05,
+      "loss": 1.0034,
+      "step": 322920
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.798729624244279e-05,
+      "loss": 0.8844,
+      "step": 322930
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.7986148312543045e-05,
+      "loss": 0.9643,
+      "step": 322940
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.7985000382643295e-05,
+      "loss": 1.0569,
+      "step": 322950
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.798385245274355e-05,
+      "loss": 1.0882,
+      "step": 322960
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.79827045228438e-05,
+      "loss": 0.8639,
+      "step": 322970
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.798155659294405e-05,
+      "loss": 0.9117,
+      "step": 322980
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.798040866304431e-05,
+      "loss": 0.9256,
+      "step": 322990
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.7979260733144564e-05,
+      "loss": 0.9287,
+      "step": 323000
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.797811280324481e-05,
+      "loss": 0.9531,
+      "step": 323010
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.7976964873345064e-05,
+      "loss": 0.9195,
+      "step": 323020
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.797581694344532e-05,
+      "loss": 1.0646,
+      "step": 323030
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.7974669013545564e-05,
+      "loss": 1.0238,
+      "step": 323040
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.797352108364582e-05,
+      "loss": 0.9601,
+      "step": 323050
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.797237315374608e-05,
+      "loss": 1.009,
+      "step": 323060
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.797122522384633e-05,
+      "loss": 1.1197,
+      "step": 323070
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.797007729394658e-05,
+      "loss": 1.0091,
+      "step": 323080
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.7968929364046833e-05,
+      "loss": 0.9732,
+      "step": 323090
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.796778143414708e-05,
+      "loss": 0.9055,
+      "step": 323100
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 3.796663350424734e-05,
+      "loss": 0.9954,
+      "step": 323110
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.796548557434759e-05,
+      "loss": 0.9689,
+      "step": 323120
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.796433764444784e-05,
+      "loss": 1.0376,
+      "step": 323130
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.7963189714548096e-05,
+      "loss": 0.9984,
+      "step": 323140
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.796204178464835e-05,
+      "loss": 1.0734,
+      "step": 323150
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.7960893854748596e-05,
+      "loss": 0.9387,
+      "step": 323160
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.795974592484885e-05,
+      "loss": 1.1127,
+      "step": 323170
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.795859799494911e-05,
+      "loss": 0.8715,
+      "step": 323180
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.795745006504935e-05,
+      "loss": 0.9157,
+      "step": 323190
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.795630213514961e-05,
+      "loss": 0.9561,
+      "step": 323200
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.7955154205249866e-05,
+      "loss": 0.9817,
+      "step": 323210
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.7954006275350115e-05,
+      "loss": 0.9553,
+      "step": 323220
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.7952858345450365e-05,
+      "loss": 0.8654,
+      "step": 323230
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.795171041555062e-05,
+      "loss": 0.9189,
+      "step": 323240
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.795056248565087e-05,
+      "loss": 1.0059,
+      "step": 323250
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.794941455575113e-05,
+      "loss": 0.9583,
+      "step": 323260
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.794826662585138e-05,
+      "loss": 1.0581,
+      "step": 323270
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.794711869595163e-05,
+      "loss": 0.9151,
+      "step": 323280
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.7945970766051885e-05,
+      "loss": 0.922,
+      "step": 323290
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.794482283615214e-05,
+      "loss": 0.8935,
+      "step": 323300
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.7943674906252384e-05,
+      "loss": 0.895,
+      "step": 323310
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.794252697635264e-05,
+      "loss": 0.8796,
+      "step": 323320
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 3.79413790464529e-05,
+      "loss": 0.9704,
+      "step": 323330
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.794023111655314e-05,
+      "loss": 0.9449,
+      "step": 323340
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.79390831866534e-05,
+      "loss": 0.9822,
+      "step": 323350
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.7937935256753654e-05,
+      "loss": 0.9982,
+      "step": 323360
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.7936787326853904e-05,
+      "loss": 0.9851,
+      "step": 323370
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.7935639396954154e-05,
+      "loss": 0.9842,
+      "step": 323380
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.793449146705441e-05,
+      "loss": 1.0384,
+      "step": 323390
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.793334353715466e-05,
+      "loss": 0.9615,
+      "step": 323400
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.793219560725492e-05,
+      "loss": 0.9287,
+      "step": 323410
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.793104767735517e-05,
+      "loss": 1.0329,
+      "step": 323420
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.792989974745542e-05,
+      "loss": 0.9746,
+      "step": 323430
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.792875181755567e-05,
+      "loss": 0.8973,
+      "step": 323440
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.792760388765593e-05,
+      "loss": 0.9702,
+      "step": 323450
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.792645595775617e-05,
+      "loss": 0.9825,
+      "step": 323460
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.792530802785643e-05,
+      "loss": 0.8567,
+      "step": 323470
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.7924160097956686e-05,
+      "loss": 0.8839,
+      "step": 323480
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.792301216805693e-05,
+      "loss": 0.983,
+      "step": 323490
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.7921864238157186e-05,
+      "loss": 0.914,
+      "step": 323500
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.792071630825744e-05,
+      "loss": 0.8956,
+      "step": 323510
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.791956837835769e-05,
+      "loss": 0.9325,
+      "step": 323520
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.791842044845794e-05,
+      "loss": 0.9298,
+      "step": 323530
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 3.79172725185582e-05,
+      "loss": 0.9464,
+      "step": 323540
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.791612458865845e-05,
+      "loss": 0.9996,
+      "step": 323550
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.7914976658758705e-05,
+      "loss": 0.8518,
+      "step": 323560
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.7913828728858955e-05,
+      "loss": 0.9357,
+      "step": 323570
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.7912680798959205e-05,
+      "loss": 1.0291,
+      "step": 323580
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.791153286905946e-05,
+      "loss": 0.9098,
+      "step": 323590
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.791038493915972e-05,
+      "loss": 0.9124,
+      "step": 323600
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.790923700925996e-05,
+      "loss": 0.9498,
+      "step": 323610
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.790808907936022e-05,
+      "loss": 0.9085,
+      "step": 323620
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.7906941149460475e-05,
+      "loss": 0.9481,
+      "step": 323630
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.790579321956072e-05,
+      "loss": 0.9638,
+      "step": 323640
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.7904645289660974e-05,
+      "loss": 0.9174,
+      "step": 323650
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.790349735976123e-05,
+      "loss": 0.8753,
+      "step": 323660
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.790234942986148e-05,
+      "loss": 0.9091,
+      "step": 323670
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.790120149996173e-05,
+      "loss": 0.9409,
+      "step": 323680
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.790005357006199e-05,
+      "loss": 1.0461,
+      "step": 323690
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.789890564016224e-05,
+      "loss": 0.9402,
+      "step": 323700
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.7897757710262494e-05,
+      "loss": 0.969,
+      "step": 323710
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.7896609780362744e-05,
+      "loss": 0.8954,
+      "step": 323720
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.7895461850462994e-05,
+      "loss": 1.0882,
+      "step": 323730
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.789431392056325e-05,
+      "loss": 0.9478,
+      "step": 323740
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.789316599066349e-05,
+      "loss": 0.9487,
+      "step": 323750
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 3.789201806076375e-05,
+      "loss": 0.9692,
+      "step": 323760
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.7890870130864007e-05,
+      "loss": 0.9085,
+      "step": 323770
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.7889722200964256e-05,
+      "loss": 0.9656,
+      "step": 323780
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.7888574271064506e-05,
+      "loss": 1.0755,
+      "step": 323790
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.788742634116476e-05,
+      "loss": 0.8996,
+      "step": 323800
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.788627841126501e-05,
+      "loss": 0.9186,
+      "step": 323810
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.788513048136527e-05,
+      "loss": 0.9387,
+      "step": 323820
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.788398255146552e-05,
+      "loss": 0.9554,
+      "step": 323830
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.788283462156577e-05,
+      "loss": 0.9602,
+      "step": 323840
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.7881686691666026e-05,
+      "loss": 1.0475,
+      "step": 323850
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.788053876176628e-05,
+      "loss": 1.0587,
+      "step": 323860
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.7879390831866525e-05,
+      "loss": 0.9872,
+      "step": 323870
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.787824290196678e-05,
+      "loss": 1.0325,
+      "step": 323880
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.787709497206704e-05,
+      "loss": 0.9718,
+      "step": 323890
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.787594704216728e-05,
+      "loss": 0.9853,
+      "step": 323900
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.787479911226754e-05,
+      "loss": 0.9383,
+      "step": 323910
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.7873651182367795e-05,
+      "loss": 0.9576,
+      "step": 323920
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.7872503252468045e-05,
+      "loss": 0.991,
+      "step": 323930
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.7871355322568295e-05,
+      "loss": 0.9147,
+      "step": 323940
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.787020739266855e-05,
+      "loss": 0.984,
+      "step": 323950
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.78690594627688e-05,
+      "loss": 0.9057,
+      "step": 323960
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.786791153286906e-05,
+      "loss": 0.9481,
+      "step": 323970
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 3.786676360296931e-05,
+      "loss": 0.9957,
+      "step": 323980
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.786561567306956e-05,
+      "loss": 0.9412,
+      "step": 323990
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.7864467743169814e-05,
+      "loss": 0.8394,
+      "step": 324000
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.786331981327007e-05,
+      "loss": 1.0162,
+      "step": 324010
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.7862171883370314e-05,
+      "loss": 0.9788,
+      "step": 324020
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.786102395347057e-05,
+      "loss": 1.0036,
+      "step": 324030
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.785987602357083e-05,
+      "loss": 0.964,
+      "step": 324040
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.785872809367107e-05,
+      "loss": 0.945,
+      "step": 324050
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.785758016377133e-05,
+      "loss": 0.9825,
+      "step": 324060
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.7856432233871583e-05,
+      "loss": 0.9724,
+      "step": 324070
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.785528430397183e-05,
+      "loss": 0.8507,
+      "step": 324080
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.785413637407208e-05,
+      "loss": 0.9638,
+      "step": 324090
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.785298844417234e-05,
+      "loss": 0.9159,
+      "step": 324100
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.785184051427259e-05,
+      "loss": 0.9557,
+      "step": 324110
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.7850692584372846e-05,
+      "loss": 0.9212,
+      "step": 324120
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.7849544654473096e-05,
+      "loss": 0.8427,
+      "step": 324130
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.7848396724573346e-05,
+      "loss": 1.0187,
+      "step": 324140
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.78472487946736e-05,
+      "loss": 0.8538,
+      "step": 324150
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.784610086477386e-05,
+      "loss": 0.9128,
+      "step": 324160
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.78449529348741e-05,
+      "loss": 0.9907,
+      "step": 324170
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.784380500497436e-05,
+      "loss": 0.9484,
+      "step": 324180
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.7842657075074616e-05,
+      "loss": 0.9617,
+      "step": 324190
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.784150914517486e-05,
+      "loss": 0.9378,
+      "step": 324200
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.7840361215275115e-05,
+      "loss": 1.0032,
+      "step": 324210
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.783921328537537e-05,
+      "loss": 0.8863,
+      "step": 324220
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.783806535547562e-05,
+      "loss": 0.9791,
+      "step": 324230
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.783691742557587e-05,
+      "loss": 0.9166,
+      "step": 324240
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.783576949567613e-05,
+      "loss": 0.7909,
+      "step": 324250
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.783462156577638e-05,
+      "loss": 0.9774,
+      "step": 324260
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.7833473635876635e-05,
+      "loss": 0.9189,
+      "step": 324270
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.7832325705976885e-05,
+      "loss": 0.9977,
+      "step": 324280
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.7831177776077134e-05,
+      "loss": 0.9221,
+      "step": 324290
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.783002984617739e-05,
+      "loss": 1.018,
+      "step": 324300
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.782888191627765e-05,
+      "loss": 0.9738,
+      "step": 324310
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.782773398637789e-05,
+      "loss": 1.0219,
+      "step": 324320
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.782658605647815e-05,
+      "loss": 1.0404,
+      "step": 324330
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.7825438126578404e-05,
+      "loss": 0.9412,
+      "step": 324340
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.782429019667865e-05,
+      "loss": 0.927,
+      "step": 324350
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.7823142266778904e-05,
+      "loss": 0.9013,
+      "step": 324360
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.782199433687916e-05,
+      "loss": 0.9476,
+      "step": 324370
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.782084640697941e-05,
+      "loss": 0.9434,
+      "step": 324380
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.781969847707966e-05,
+      "loss": 0.9993,
+      "step": 324390
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.781855054717992e-05,
+      "loss": 0.9309,
+      "step": 324400
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.781740261728017e-05,
+      "loss": 0.9774,
+      "step": 324410
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.781625468738042e-05,
+      "loss": 0.9511,
+      "step": 324420
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.781510675748067e-05,
+      "loss": 0.9659,
+      "step": 324430
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.781395882758092e-05,
+      "loss": 0.9823,
+      "step": 324440
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.781281089768118e-05,
+      "loss": 1.0144,
+      "step": 324450
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.7811662967781436e-05,
+      "loss": 0.9576,
+      "step": 324460
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.781051503788168e-05,
+      "loss": 1.0223,
+      "step": 324470
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.7809367107981936e-05,
+      "loss": 0.9862,
+      "step": 324480
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.780821917808219e-05,
+      "loss": 0.9237,
+      "step": 324490
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.7807071248182436e-05,
+      "loss": 0.9211,
+      "step": 324500
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.780592331828269e-05,
+      "loss": 0.9888,
+      "step": 324510
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.780477538838295e-05,
+      "loss": 0.873,
+      "step": 324520
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.78036274584832e-05,
+      "loss": 0.9343,
+      "step": 324530
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.780247952858345e-05,
+      "loss": 0.9662,
+      "step": 324540
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.7801331598683705e-05,
+      "loss": 0.9181,
+      "step": 324550
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.7800183668783955e-05,
+      "loss": 0.9099,
+      "step": 324560
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.779903573888421e-05,
+      "loss": 1.0256,
+      "step": 324570
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.779788780898446e-05,
+      "loss": 1.0159,
+      "step": 324580
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.779673987908471e-05,
+      "loss": 1.0004,
+      "step": 324590
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.779559194918497e-05,
+      "loss": 0.9795,
+      "step": 324600
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.7794444019285225e-05,
+      "loss": 1.0679,
+      "step": 324610
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.779329608938547e-05,
+      "loss": 0.903,
+      "step": 324620
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 3.7792148159485724e-05,
+      "loss": 1.0277,
+      "step": 324630
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.779100022958598e-05,
+      "loss": 0.946,
+      "step": 324640
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.7789852299686224e-05,
+      "loss": 1.0176,
+      "step": 324650
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.778870436978648e-05,
+      "loss": 1.0179,
+      "step": 324660
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.778755643988674e-05,
+      "loss": 0.8788,
+      "step": 324670
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.778640850998699e-05,
+      "loss": 0.9593,
+      "step": 324680
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.778526058008724e-05,
+      "loss": 1.0573,
+      "step": 324690
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.7784112650187494e-05,
+      "loss": 0.9772,
+      "step": 324700
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.7782964720287744e-05,
+      "loss": 1.0416,
+      "step": 324710
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.7781816790388e-05,
+      "loss": 0.9537,
+      "step": 324720
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.778066886048825e-05,
+      "loss": 0.8319,
+      "step": 324730
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.77795209305885e-05,
+      "loss": 0.9299,
+      "step": 324740
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.7778373000688757e-05,
+      "loss": 0.9402,
+      "step": 324750
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.777722507078901e-05,
+      "loss": 0.8912,
+      "step": 324760
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.7776077140889256e-05,
+      "loss": 0.9822,
+      "step": 324770
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.777492921098951e-05,
+      "loss": 0.8461,
+      "step": 324780
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.777378128108977e-05,
+      "loss": 0.9359,
+      "step": 324790
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.777263335119001e-05,
+      "loss": 0.9447,
+      "step": 324800
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.777148542129027e-05,
+      "loss": 0.8371,
+      "step": 324810
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.7770337491390526e-05,
+      "loss": 0.9699,
+      "step": 324820
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.7769189561490776e-05,
+      "loss": 1.0796,
+      "step": 324830
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.7768041631591026e-05,
+      "loss": 0.9334,
+      "step": 324840
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 3.776689370169128e-05,
+      "loss": 0.952,
+      "step": 324850
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.776574577179153e-05,
+      "loss": 0.962,
+      "step": 324860
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.776459784189179e-05,
+      "loss": 1.0329,
+      "step": 324870
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.776344991199204e-05,
+      "loss": 1.0341,
+      "step": 324880
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.776230198209229e-05,
+      "loss": 1.031,
+      "step": 324890
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.7761154052192545e-05,
+      "loss": 1.0581,
+      "step": 324900
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.77600061222928e-05,
+      "loss": 0.9092,
+      "step": 324910
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.7758858192393045e-05,
+      "loss": 0.9159,
+      "step": 324920
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.77577102624933e-05,
+      "loss": 0.8467,
+      "step": 324930
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.775656233259356e-05,
+      "loss": 0.9093,
+      "step": 324940
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.77554144026938e-05,
+      "loss": 0.9515,
+      "step": 324950
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.775426647279406e-05,
+      "loss": 1.047,
+      "step": 324960
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.7753118542894314e-05,
+      "loss": 0.899,
+      "step": 324970
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.7751970612994564e-05,
+      "loss": 0.9609,
+      "step": 324980
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.7750822683094814e-05,
+      "loss": 0.9376,
+      "step": 324990
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.774967475319507e-05,
+      "loss": 0.8717,
+      "step": 325000
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.774852682329532e-05,
+      "loss": 1.0225,
+      "step": 325010
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.774737889339558e-05,
+      "loss": 1.0407,
+      "step": 325020
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.774623096349583e-05,
+      "loss": 0.9493,
+      "step": 325030
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.774508303359608e-05,
+      "loss": 0.8582,
+      "step": 325040
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.7743935103696333e-05,
+      "loss": 0.9208,
+      "step": 325050
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.774278717379659e-05,
+      "loss": 0.9512,
+      "step": 325060
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 3.774163924389683e-05,
+      "loss": 0.9225,
+      "step": 325070
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.774049131399709e-05,
+      "loss": 0.9616,
+      "step": 325080
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.7739343384097346e-05,
+      "loss": 1.0279,
+      "step": 325090
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.773819545419759e-05,
+      "loss": 0.9033,
+      "step": 325100
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.7737047524297846e-05,
+      "loss": 0.8683,
+      "step": 325110
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.7735899594398096e-05,
+      "loss": 0.9751,
+      "step": 325120
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.773475166449835e-05,
+      "loss": 1.0171,
+      "step": 325130
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.77336037345986e-05,
+      "loss": 0.9566,
+      "step": 325140
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.773245580469885e-05,
+      "loss": 0.9873,
+      "step": 325150
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.773130787479911e-05,
+      "loss": 0.9246,
+      "step": 325160
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.7730159944899366e-05,
+      "loss": 0.8771,
+      "step": 325170
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.772901201499961e-05,
+      "loss": 0.9739,
+      "step": 325180
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.7727864085099865e-05,
+      "loss": 0.9187,
+      "step": 325190
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.772671615520012e-05,
+      "loss": 1.0055,
+      "step": 325200
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.7725568225300365e-05,
+      "loss": 0.9432,
+      "step": 325210
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.772442029540062e-05,
+      "loss": 1.0485,
+      "step": 325220
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.772327236550088e-05,
+      "loss": 1.0038,
+      "step": 325230
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.772212443560113e-05,
+      "loss": 1.0498,
+      "step": 325240
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.772097650570138e-05,
+      "loss": 0.8585,
+      "step": 325250
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.7719828575801635e-05,
+      "loss": 1.0138,
+      "step": 325260
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.7718680645901884e-05,
+      "loss": 0.9902,
+      "step": 325270
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.771753271600214e-05,
+      "loss": 1.0096,
+      "step": 325280
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 3.771638478610239e-05,
+      "loss": 0.9311,
+      "step": 325290
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.771523685620264e-05,
+      "loss": 0.9462,
+      "step": 325300
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.77140889263029e-05,
+      "loss": 0.8607,
+      "step": 325310
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.7712940996403154e-05,
+      "loss": 0.8539,
+      "step": 325320
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.77117930665034e-05,
+      "loss": 0.9259,
+      "step": 325330
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.7710645136603654e-05,
+      "loss": 0.9313,
+      "step": 325340
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.770949720670391e-05,
+      "loss": 1.0147,
+      "step": 325350
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.7708349276804153e-05,
+      "loss": 0.8871,
+      "step": 325360
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.770720134690441e-05,
+      "loss": 0.9934,
+      "step": 325370
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.770605341700467e-05,
+      "loss": 0.9176,
+      "step": 325380
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.7704905487104917e-05,
+      "loss": 0.8941,
+      "step": 325390
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.7703757557205166e-05,
+      "loss": 0.9418,
+      "step": 325400
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.770260962730542e-05,
+      "loss": 0.9228,
+      "step": 325410
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.770146169740567e-05,
+      "loss": 0.9639,
+      "step": 325420
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.770031376750593e-05,
+      "loss": 0.8908,
+      "step": 325430
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.769916583760618e-05,
+      "loss": 0.9571,
+      "step": 325440
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.769801790770643e-05,
+      "loss": 1.0333,
+      "step": 325450
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.7696869977806686e-05,
+      "loss": 1.0948,
+      "step": 325460
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.769572204790694e-05,
+      "loss": 0.9019,
+      "step": 325470
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.7694574118007186e-05,
+      "loss": 0.9811,
+      "step": 325480
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.769342618810744e-05,
+      "loss": 0.9213,
+      "step": 325490
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.76922782582077e-05,
+      "loss": 0.8681,
+      "step": 325500
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 3.769113032830794e-05,
+      "loss": 0.8716,
+      "step": 325510
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.76899823984082e-05,
+      "loss": 0.8608,
+      "step": 325520
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.7688834468508455e-05,
+      "loss": 0.9668,
+      "step": 325530
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.7687686538608705e-05,
+      "loss": 1.0737,
+      "step": 325540
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.7686538608708955e-05,
+      "loss": 0.8952,
+      "step": 325550
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.768539067880921e-05,
+      "loss": 0.9877,
+      "step": 325560
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.768424274890946e-05,
+      "loss": 0.9918,
+      "step": 325570
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.768309481900972e-05,
+      "loss": 1.0016,
+      "step": 325580
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.768194688910997e-05,
+      "loss": 1.0127,
+      "step": 325590
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.768079895921022e-05,
+      "loss": 0.9307,
+      "step": 325600
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.7679651029310474e-05,
+      "loss": 0.975,
+      "step": 325610
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.767850309941073e-05,
+      "loss": 0.9531,
+      "step": 325620
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.7677355169510974e-05,
+      "loss": 0.9081,
+      "step": 325630
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.767620723961123e-05,
+      "loss": 1.0191,
+      "step": 325640
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.767505930971149e-05,
+      "loss": 0.9623,
+      "step": 325650
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.767391137981173e-05,
+      "loss": 0.9339,
+      "step": 325660
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.767276344991199e-05,
+      "loss": 1.0012,
+      "step": 325670
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.7671615520012244e-05,
+      "loss": 0.943,
+      "step": 325680
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.7670467590112494e-05,
+      "loss": 0.9603,
+      "step": 325690
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.7669319660212743e-05,
+      "loss": 0.9182,
+      "step": 325700
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.7668171730313e-05,
+      "loss": 1.0525,
+      "step": 325710
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 3.766702380041325e-05,
+      "loss": 0.919,
+      "step": 325720
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.7665875870513507e-05,
+      "loss": 0.9543,
+      "step": 325730
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.7664727940613756e-05,
+      "loss": 0.9271,
+      "step": 325740
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.7663580010714006e-05,
+      "loss": 0.9345,
+      "step": 325750
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.766243208081426e-05,
+      "loss": 1.032,
+      "step": 325760
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.766128415091452e-05,
+      "loss": 1.0026,
+      "step": 325770
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.766013622101476e-05,
+      "loss": 0.9586,
+      "step": 325780
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.765898829111502e-05,
+      "loss": 0.9721,
+      "step": 325790
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.7657840361215276e-05,
+      "loss": 0.938,
+      "step": 325800
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.765669243131552e-05,
+      "loss": 0.9074,
+      "step": 325810
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.7655544501415776e-05,
+      "loss": 0.9424,
+      "step": 325820
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.765439657151603e-05,
+      "loss": 0.9579,
+      "step": 325830
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.765324864161628e-05,
+      "loss": 0.9251,
+      "step": 325840
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.765210071171653e-05,
+      "loss": 0.9405,
+      "step": 325850
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.765095278181679e-05,
+      "loss": 0.9917,
+      "step": 325860
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.764980485191704e-05,
+      "loss": 0.8962,
+      "step": 325870
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.7648656922017295e-05,
+      "loss": 1.0065,
+      "step": 325880
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.7647508992117545e-05,
+      "loss": 0.9429,
+      "step": 325890
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.7646361062217795e-05,
+      "loss": 0.9881,
+      "step": 325900
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.764521313231805e-05,
+      "loss": 0.9849,
+      "step": 325910
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.764406520241831e-05,
+      "loss": 1.0871,
+      "step": 325920
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.764291727251855e-05,
+      "loss": 1.0394,
+      "step": 325930
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 3.764176934261881e-05,
+      "loss": 0.9999,
+      "step": 325940
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.7640621412719064e-05,
+      "loss": 0.9853,
+      "step": 325950
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.763947348281931e-05,
+      "loss": 0.9798,
+      "step": 325960
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.7638325552919564e-05,
+      "loss": 0.9475,
+      "step": 325970
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.763717762301982e-05,
+      "loss": 0.9178,
+      "step": 325980
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.763602969312007e-05,
+      "loss": 0.9463,
+      "step": 325990
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.763488176322032e-05,
+      "loss": 0.9582,
+      "step": 326000
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.763373383332058e-05,
+      "loss": 0.8757,
+      "step": 326010
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.763258590342083e-05,
+      "loss": 0.932,
+      "step": 326020
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.7631437973521083e-05,
+      "loss": 0.9429,
+      "step": 326030
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.763029004362133e-05,
+      "loss": 1.0173,
+      "step": 326040
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.762914211372158e-05,
+      "loss": 0.95,
+      "step": 326050
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.762799418382184e-05,
+      "loss": 0.9748,
+      "step": 326060
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.7626846253922096e-05,
+      "loss": 0.9568,
+      "step": 326070
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.762569832402234e-05,
+      "loss": 1.0412,
+      "step": 326080
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.7624550394122596e-05,
+      "loss": 0.8992,
+      "step": 326090
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.762340246422285e-05,
+      "loss": 0.9582,
+      "step": 326100
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.7622254534323096e-05,
+      "loss": 0.9952,
+      "step": 326110
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.762110660442335e-05,
+      "loss": 0.9657,
+      "step": 326120
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.761995867452361e-05,
+      "loss": 0.9054,
+      "step": 326130
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.761881074462386e-05,
+      "loss": 1.0283,
+      "step": 326140
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.761766281472411e-05,
+      "loss": 0.9193,
+      "step": 326150
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.7616514884824365e-05,
+      "loss": 0.9774,
+      "step": 326160
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.7615366954924615e-05,
+      "loss": 0.9136,
+      "step": 326170
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.761421902502487e-05,
+      "loss": 0.9808,
+      "step": 326180
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.761307109512512e-05,
+      "loss": 1.0272,
+      "step": 326190
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.761192316522537e-05,
+      "loss": 0.9243,
+      "step": 326200
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.761077523532563e-05,
+      "loss": 1.0335,
+      "step": 326210
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.7609627305425885e-05,
+      "loss": 0.9832,
+      "step": 326220
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.760847937552613e-05,
+      "loss": 1.0036,
+      "step": 326230
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.7607331445626385e-05,
+      "loss": 0.9309,
+      "step": 326240
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.760618351572664e-05,
+      "loss": 0.9362,
+      "step": 326250
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.7605035585826884e-05,
+      "loss": 0.9189,
+      "step": 326260
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.760388765592714e-05,
+      "loss": 1.0324,
+      "step": 326270
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.76027397260274e-05,
+      "loss": 0.928,
+      "step": 326280
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.760159179612765e-05,
+      "loss": 0.9421,
+      "step": 326290
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.76004438662279e-05,
+      "loss": 0.9363,
+      "step": 326300
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.7599295936328154e-05,
+      "loss": 1.0696,
+      "step": 326310
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.7598148006428404e-05,
+      "loss": 0.9402,
+      "step": 326320
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.759700007652866e-05,
+      "loss": 0.9243,
+      "step": 326330
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.759585214662891e-05,
+      "loss": 0.9763,
+      "step": 326340
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.759470421672916e-05,
+      "loss": 0.9639,
+      "step": 326350
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.759355628682942e-05,
+      "loss": 1.0482,
+      "step": 326360
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.759240835692967e-05,
+      "loss": 0.9554,
+      "step": 326370
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 3.7591260427029916e-05,
+      "loss": 0.9451,
+      "step": 326380
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.759011249713017e-05,
+      "loss": 0.9553,
+      "step": 326390
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.758896456723043e-05,
+      "loss": 0.8976,
+      "step": 326400
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.758781663733067e-05,
+      "loss": 0.9939,
+      "step": 326410
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.758666870743093e-05,
+      "loss": 0.9692,
+      "step": 326420
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.7585520777531186e-05,
+      "loss": 0.899,
+      "step": 326430
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.7584372847631436e-05,
+      "loss": 1.0232,
+      "step": 326440
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.7583224917731686e-05,
+      "loss": 0.9514,
+      "step": 326450
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.758207698783194e-05,
+      "loss": 1.002,
+      "step": 326460
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.758092905793219e-05,
+      "loss": 0.9268,
+      "step": 326470
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.757978112803245e-05,
+      "loss": 0.8989,
+      "step": 326480
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.757863319813269e-05,
+      "loss": 0.8719,
+      "step": 326490
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.757748526823295e-05,
+      "loss": 0.9227,
+      "step": 326500
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.7576337338333205e-05,
+      "loss": 0.9399,
+      "step": 326510
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.757518940843345e-05,
+      "loss": 0.9296,
+      "step": 326520
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.7574041478533705e-05,
+      "loss": 1.0058,
+      "step": 326530
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.757289354863396e-05,
+      "loss": 1.1064,
+      "step": 326540
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.757174561873421e-05,
+      "loss": 0.9679,
+      "step": 326550
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.757059768883446e-05,
+      "loss": 0.9331,
+      "step": 326560
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.756944975893472e-05,
+      "loss": 0.9195,
+      "step": 326570
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.756830182903497e-05,
+      "loss": 0.9167,
+      "step": 326580
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.7567153899135224e-05,
+      "loss": 0.8957,
+      "step": 326590
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 3.7566005969235474e-05,
+      "loss": 0.8649,
+      "step": 326600
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.7564858039335724e-05,
+      "loss": 0.9497,
+      "step": 326610
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.756371010943598e-05,
+      "loss": 1.0153,
+      "step": 326620
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.756256217953624e-05,
+      "loss": 0.8696,
+      "step": 326630
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.756141424963648e-05,
+      "loss": 0.9891,
+      "step": 326640
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.756026631973674e-05,
+      "loss": 0.9648,
+      "step": 326650
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.7559118389836994e-05,
+      "loss": 0.9355,
+      "step": 326660
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.755797045993724e-05,
+      "loss": 1.0215,
+      "step": 326670
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.7556822530037493e-05,
+      "loss": 0.9693,
+      "step": 326680
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.755567460013775e-05,
+      "loss": 0.8519,
+      "step": 326690
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.7554526670238e-05,
+      "loss": 1.0242,
+      "step": 326700
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.755337874033825e-05,
+      "loss": 0.991,
+      "step": 326710
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.7552230810438506e-05,
+      "loss": 0.9684,
+      "step": 326720
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.7551082880538756e-05,
+      "loss": 0.9626,
+      "step": 326730
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.754993495063901e-05,
+      "loss": 0.9437,
+      "step": 326740
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.754878702073926e-05,
+      "loss": 0.9066,
+      "step": 326750
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.754763909083951e-05,
+      "loss": 0.8809,
+      "step": 326760
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.754649116093977e-05,
+      "loss": 0.9623,
+      "step": 326770
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.7545343231040026e-05,
+      "loss": 0.922,
+      "step": 326780
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.754419530114027e-05,
+      "loss": 0.9206,
+      "step": 326790
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.7543047371240526e-05,
+      "loss": 0.949,
+      "step": 326800
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 3.754189944134078e-05,
+      "loss": 0.9599,
+      "step": 326810
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.7540751511441025e-05,
+      "loss": 0.9729,
+      "step": 326820
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.753960358154128e-05,
+      "loss": 0.9873,
+      "step": 326830
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.753845565164154e-05,
+      "loss": 0.8913,
+      "step": 326840
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.753730772174179e-05,
+      "loss": 0.957,
+      "step": 326850
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.753615979184204e-05,
+      "loss": 0.9523,
+      "step": 326860
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.7535011861942295e-05,
+      "loss": 1.026,
+      "step": 326870
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.7533863932042545e-05,
+      "loss": 0.9723,
+      "step": 326880
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.75327160021428e-05,
+      "loss": 1.0144,
+      "step": 326890
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.753156807224305e-05,
+      "loss": 0.9098,
+      "step": 326900
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.75304201423433e-05,
+      "loss": 0.9593,
+      "step": 326910
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.752927221244356e-05,
+      "loss": 1.0262,
+      "step": 326920
+    },
+    {
+      "epoch": 15.0,
+      "eval_cer": 0.6683523105087306,
+      "eval_em": 0.007357212003872217,
+      "eval_f1": 0.007357212003872217,
+      "eval_loss": 1.2376947402954102,
+      "eval_runtime": 1617.3498,
+      "eval_samples_per_second": 6.387,
+      "eval_steps_per_second": 0.799,
+      "eval_wer": 0.9926427879961278,
+      "step": 326927
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.7528124282543814e-05,
+      "loss": 0.8851,
+      "step": 326930
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.752697635264406e-05,
+      "loss": 0.92,
+      "step": 326940
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.7525828422744314e-05,
+      "loss": 0.8069,
+      "step": 326950
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.752468049284457e-05,
+      "loss": 0.9229,
+      "step": 326960
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.7523532562944814e-05,
+      "loss": 0.9065,
+      "step": 326970
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.752238463304507e-05,
+      "loss": 0.8705,
+      "step": 326980
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.752123670314533e-05,
+      "loss": 0.8158,
+      "step": 326990
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.752008877324558e-05,
+      "loss": 0.9127,
+      "step": 327000
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.751894084334583e-05,
+      "loss": 0.9325,
+      "step": 327010
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.751779291344608e-05,
+      "loss": 0.8947,
+      "step": 327020
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 3.751664498354633e-05,
+      "loss": 0.8257,
+      "step": 327030
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.751549705364659e-05,
+      "loss": 0.9241,
+      "step": 327040
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.751434912374684e-05,
+      "loss": 0.8667,
+      "step": 327050
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.751320119384709e-05,
+      "loss": 0.8413,
+      "step": 327060
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.7512053263947346e-05,
+      "loss": 0.8506,
+      "step": 327070
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.75109053340476e-05,
+      "loss": 0.876,
+      "step": 327080
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.7509757404147846e-05,
+      "loss": 0.9178,
+      "step": 327090
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.75086094742481e-05,
+      "loss": 0.9056,
+      "step": 327100
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.750746154434836e-05,
+      "loss": 0.8217,
+      "step": 327110
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.75063136144486e-05,
+      "loss": 0.8336,
+      "step": 327120
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.750516568454886e-05,
+      "loss": 0.9361,
+      "step": 327130
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.7504017754649115e-05,
+      "loss": 0.8193,
+      "step": 327140
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.7502869824749365e-05,
+      "loss": 0.9113,
+      "step": 327150
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.7501721894849615e-05,
+      "loss": 0.8947,
+      "step": 327160
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.750057396494987e-05,
+      "loss": 0.9469,
+      "step": 327170
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.749942603505012e-05,
+      "loss": 0.8378,
+      "step": 327180
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.749827810515038e-05,
+      "loss": 1.0235,
+      "step": 327190
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.749713017525063e-05,
+      "loss": 0.9172,
+      "step": 327200
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.7495982245350885e-05,
+      "loss": 0.9907,
+      "step": 327210
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.7494834315451135e-05,
+      "loss": 0.8973,
+      "step": 327220
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.7493686385551384e-05,
+      "loss": 0.8291,
+      "step": 327230
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.749253845565164e-05,
+      "loss": 0.9543,
+      "step": 327240
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 3.749139052575189e-05,
+      "loss": 1.0126,
+      "step": 327250
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.749024259585214e-05,
+      "loss": 0.9076,
+      "step": 327260
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.74890946659524e-05,
+      "loss": 0.8833,
+      "step": 327270
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.748794673605265e-05,
+      "loss": 0.8723,
+      "step": 327280
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.74867988061529e-05,
+      "loss": 0.9816,
+      "step": 327290
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.7485650876253154e-05,
+      "loss": 0.852,
+      "step": 327300
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.7484502946353404e-05,
+      "loss": 0.9009,
+      "step": 327310
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.748335501645366e-05,
+      "loss": 0.8685,
+      "step": 327320
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.748220708655391e-05,
+      "loss": 0.8617,
+      "step": 327330
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.748105915665417e-05,
+      "loss": 0.8887,
+      "step": 327340
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.7479911226754417e-05,
+      "loss": 0.8145,
+      "step": 327350
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.747876329685467e-05,
+      "loss": 0.9693,
+      "step": 327360
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.747761536695492e-05,
+      "loss": 0.9385,
+      "step": 327370
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.747646743705517e-05,
+      "loss": 0.8507,
+      "step": 327380
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.747531950715543e-05,
+      "loss": 0.8088,
+      "step": 327390
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.747417157725568e-05,
+      "loss": 1.0042,
+      "step": 327400
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.747302364735593e-05,
+      "loss": 0.987,
+      "step": 327410
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.7471875717456186e-05,
+      "loss": 1.0438,
+      "step": 327420
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.7470727787556436e-05,
+      "loss": 0.8802,
+      "step": 327430
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.7469579857656686e-05,
+      "loss": 0.9102,
+      "step": 327440
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.746843192775694e-05,
+      "loss": 0.8482,
+      "step": 327450
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.746728399785719e-05,
+      "loss": 0.9749,
+      "step": 327460
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 3.746613606795745e-05,
+      "loss": 0.9176,
+      "step": 327470
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.74649881380577e-05,
+      "loss": 0.9813,
+      "step": 327480
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.7463840208157955e-05,
+      "loss": 0.8453,
+      "step": 327490
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.7462692278258205e-05,
+      "loss": 0.9612,
+      "step": 327500
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.746154434835846e-05,
+      "loss": 0.8692,
+      "step": 327510
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.746039641845871e-05,
+      "loss": 0.896,
+      "step": 327520
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.745924848855896e-05,
+      "loss": 0.9266,
+      "step": 327530
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.745810055865921e-05,
+      "loss": 0.8543,
+      "step": 327540
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.745695262875947e-05,
+      "loss": 0.8834,
+      "step": 327550
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.745580469885972e-05,
+      "loss": 0.84,
+      "step": 327560
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.745465676895997e-05,
+      "loss": 0.8854,
+      "step": 327570
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.7453508839060224e-05,
+      "loss": 0.8307,
+      "step": 327580
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.7452360909160474e-05,
+      "loss": 0.8786,
+      "step": 327590
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.745121297926073e-05,
+      "loss": 0.922,
+      "step": 327600
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.745006504936098e-05,
+      "loss": 0.8916,
+      "step": 327610
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.744891711946124e-05,
+      "loss": 0.8777,
+      "step": 327620
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.744776918956149e-05,
+      "loss": 0.8229,
+      "step": 327630
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.7446621259661744e-05,
+      "loss": 0.8549,
+      "step": 327640
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.7445473329761994e-05,
+      "loss": 0.8781,
+      "step": 327650
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.7444325399862243e-05,
+      "loss": 0.8913,
+      "step": 327660
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.74431774699625e-05,
+      "loss": 0.8499,
+      "step": 327670
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 3.744202954006275e-05,
+      "loss": 0.8901,
+      "step": 327680
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.7440881610163e-05,
+      "loss": 0.9223,
+      "step": 327690
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.7439733680263256e-05,
+      "loss": 0.9054,
+      "step": 327700
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.7438585750363506e-05,
+      "loss": 0.8195,
+      "step": 327710
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.7437437820463756e-05,
+      "loss": 0.9647,
+      "step": 327720
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.743628989056401e-05,
+      "loss": 0.9209,
+      "step": 327730
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.743514196066426e-05,
+      "loss": 0.8793,
+      "step": 327740
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.743399403076452e-05,
+      "loss": 0.8644,
+      "step": 327750
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.743284610086477e-05,
+      "loss": 0.874,
+      "step": 327760
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.7431698170965026e-05,
+      "loss": 0.8679,
+      "step": 327770
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.7430550241065276e-05,
+      "loss": 0.8496,
+      "step": 327780
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.742940231116553e-05,
+      "loss": 0.8425,
+      "step": 327790
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.742825438126578e-05,
+      "loss": 0.9116,
+      "step": 327800
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.742710645136603e-05,
+      "loss": 0.9256,
+      "step": 327810
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.742595852146629e-05,
+      "loss": 0.8786,
+      "step": 327820
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.742481059156654e-05,
+      "loss": 0.9113,
+      "step": 327830
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.742366266166679e-05,
+      "loss": 0.8744,
+      "step": 327840
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.7422514731767045e-05,
+      "loss": 0.9679,
+      "step": 327850
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.7421366801867295e-05,
+      "loss": 0.8083,
+      "step": 327860
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.7420218871967545e-05,
+      "loss": 0.8931,
+      "step": 327870
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.74190709420678e-05,
+      "loss": 0.9598,
+      "step": 327880
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.741792301216805e-05,
+      "loss": 0.9781,
+      "step": 327890
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 3.741677508226831e-05,
+      "loss": 0.9362,
+      "step": 327900
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.741562715236856e-05,
+      "loss": 0.9389,
+      "step": 327910
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.7414479222468814e-05,
+      "loss": 0.9559,
+      "step": 327920
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.7413331292569064e-05,
+      "loss": 0.8939,
+      "step": 327930
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.741218336266932e-05,
+      "loss": 0.9244,
+      "step": 327940
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.741103543276957e-05,
+      "loss": 0.9494,
+      "step": 327950
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.740988750286982e-05,
+      "loss": 0.8757,
+      "step": 327960
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.740873957297008e-05,
+      "loss": 0.9118,
+      "step": 327970
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.740759164307033e-05,
+      "loss": 0.9113,
+      "step": 327980
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.740644371317058e-05,
+      "loss": 0.9064,
+      "step": 327990
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.740529578327083e-05,
+      "loss": 1.0208,
+      "step": 328000
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.740414785337108e-05,
+      "loss": 0.8662,
+      "step": 328010
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.740299992347133e-05,
+      "loss": 0.9365,
+      "step": 328020
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.740185199357159e-05,
+      "loss": 0.8357,
+      "step": 328030
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.740070406367184e-05,
+      "loss": 0.8588,
+      "step": 328040
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.7399556133772096e-05,
+      "loss": 1.0122,
+      "step": 328050
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.7398408203872346e-05,
+      "loss": 0.8029,
+      "step": 328060
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.73972602739726e-05,
+      "loss": 0.8141,
+      "step": 328070
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.739611234407285e-05,
+      "loss": 0.8569,
+      "step": 328080
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.739496441417311e-05,
+      "loss": 0.985,
+      "step": 328090
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.739381648427336e-05,
+      "loss": 0.9024,
+      "step": 328100
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.739266855437361e-05,
+      "loss": 0.931,
+      "step": 328110
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 3.7391520624473865e-05,
+      "loss": 0.8403,
+      "step": 328120
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.7390372694574115e-05,
+      "loss": 0.9476,
+      "step": 328130
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.7389224764674365e-05,
+      "loss": 0.9258,
+      "step": 328140
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.738807683477462e-05,
+      "loss": 0.8412,
+      "step": 328150
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.738692890487487e-05,
+      "loss": 0.8693,
+      "step": 328160
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.738578097497512e-05,
+      "loss": 0.9036,
+      "step": 328170
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.738463304507538e-05,
+      "loss": 0.938,
+      "step": 328180
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.738348511517563e-05,
+      "loss": 0.7828,
+      "step": 328190
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.7382337185275885e-05,
+      "loss": 0.8728,
+      "step": 328200
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.7381189255376134e-05,
+      "loss": 0.9802,
+      "step": 328210
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.738004132547639e-05,
+      "loss": 0.8461,
+      "step": 328220
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.737889339557664e-05,
+      "loss": 0.9897,
+      "step": 328230
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.737774546567689e-05,
+      "loss": 0.923,
+      "step": 328240
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.737659753577715e-05,
+      "loss": 0.9518,
+      "step": 328250
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.73754496058774e-05,
+      "loss": 0.9182,
+      "step": 328260
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.737430167597765e-05,
+      "loss": 0.8876,
+      "step": 328270
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.7373153746077904e-05,
+      "loss": 0.8986,
+      "step": 328280
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.7372005816178154e-05,
+      "loss": 0.9596,
+      "step": 328290
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.7370857886278403e-05,
+      "loss": 0.8375,
+      "step": 328300
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.736970995637866e-05,
+      "loss": 0.927,
+      "step": 328310
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.736856202647891e-05,
+      "loss": 0.8979,
+      "step": 328320
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.7367414096579167e-05,
+      "loss": 0.9003,
+      "step": 328330
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 3.7366266166679416e-05,
+      "loss": 0.9784,
+      "step": 328340
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.736511823677967e-05,
+      "loss": 0.9518,
+      "step": 328350
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.736397030687992e-05,
+      "loss": 0.958,
+      "step": 328360
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.736282237698018e-05,
+      "loss": 0.9117,
+      "step": 328370
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.736167444708043e-05,
+      "loss": 0.9944,
+      "step": 328380
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.736052651718068e-05,
+      "loss": 0.8786,
+      "step": 328390
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.7359378587280936e-05,
+      "loss": 1.0295,
+      "step": 328400
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.7358230657381186e-05,
+      "loss": 0.9281,
+      "step": 328410
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.7357082727481436e-05,
+      "loss": 0.8466,
+      "step": 328420
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.735593479758169e-05,
+      "loss": 0.8792,
+      "step": 328430
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.735478686768194e-05,
+      "loss": 0.8449,
+      "step": 328440
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.735363893778219e-05,
+      "loss": 0.8016,
+      "step": 328450
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.735249100788245e-05,
+      "loss": 0.9113,
+      "step": 328460
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.73513430779827e-05,
+      "loss": 0.8391,
+      "step": 328470
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.7350195148082955e-05,
+      "loss": 0.843,
+      "step": 328480
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.7349047218183205e-05,
+      "loss": 1.0292,
+      "step": 328490
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.734789928828346e-05,
+      "loss": 1.0067,
+      "step": 328500
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.734675135838371e-05,
+      "loss": 0.8559,
+      "step": 328510
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.734560342848397e-05,
+      "loss": 0.9244,
+      "step": 328520
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.734445549858422e-05,
+      "loss": 0.9254,
+      "step": 328530
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.734330756868447e-05,
+      "loss": 0.905,
+      "step": 328540
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.7342159638784724e-05,
+      "loss": 0.9517,
+      "step": 328550
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 3.7341011708884974e-05,
+      "loss": 0.9432,
+      "step": 328560
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.7339863778985224e-05,
+      "loss": 0.9038,
+      "step": 328570
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.733871584908548e-05,
+      "loss": 1.0117,
+      "step": 328580
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.733756791918573e-05,
+      "loss": 0.8304,
+      "step": 328590
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.733641998928598e-05,
+      "loss": 0.9586,
+      "step": 328600
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.733527205938624e-05,
+      "loss": 0.8787,
+      "step": 328610
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.733412412948649e-05,
+      "loss": 0.9648,
+      "step": 328620
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.7332976199586744e-05,
+      "loss": 0.9265,
+      "step": 328630
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.7331828269686993e-05,
+      "loss": 0.9942,
+      "step": 328640
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.733068033978725e-05,
+      "loss": 0.9568,
+      "step": 328650
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.73295324098875e-05,
+      "loss": 0.8149,
+      "step": 328660
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.7328384479987757e-05,
+      "loss": 0.8814,
+      "step": 328670
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.7327236550088006e-05,
+      "loss": 0.978,
+      "step": 328680
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.7326088620188256e-05,
+      "loss": 0.9154,
+      "step": 328690
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.732494069028851e-05,
+      "loss": 0.9116,
+      "step": 328700
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.732379276038876e-05,
+      "loss": 0.9397,
+      "step": 328710
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.732264483048901e-05,
+      "loss": 1.0566,
+      "step": 328720
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.732149690058927e-05,
+      "loss": 0.9026,
+      "step": 328730
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.732034897068952e-05,
+      "loss": 0.9642,
+      "step": 328740
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.731920104078977e-05,
+      "loss": 0.9494,
+      "step": 328750
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.7318053110890026e-05,
+      "loss": 0.9595,
+      "step": 328760
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 3.7316905180990275e-05,
+      "loss": 1.0321,
+      "step": 328770
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.731575725109053e-05,
+      "loss": 0.9402,
+      "step": 328780
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.731460932119078e-05,
+      "loss": 0.9862,
+      "step": 328790
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.731346139129104e-05,
+      "loss": 1.0307,
+      "step": 328800
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.731231346139129e-05,
+      "loss": 0.8491,
+      "step": 328810
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.7311165531491545e-05,
+      "loss": 0.9033,
+      "step": 328820
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.7310017601591795e-05,
+      "loss": 0.9294,
+      "step": 328830
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.7308869671692045e-05,
+      "loss": 0.9003,
+      "step": 328840
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.73077217417923e-05,
+      "loss": 0.9625,
+      "step": 328850
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.730657381189255e-05,
+      "loss": 0.9081,
+      "step": 328860
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.73054258819928e-05,
+      "loss": 0.8876,
+      "step": 328870
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.730427795209306e-05,
+      "loss": 0.8837,
+      "step": 328880
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.730313002219331e-05,
+      "loss": 0.9174,
+      "step": 328890
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.730198209229356e-05,
+      "loss": 0.8974,
+      "step": 328900
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.7300834162393814e-05,
+      "loss": 0.8768,
+      "step": 328910
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.7299686232494064e-05,
+      "loss": 0.9961,
+      "step": 328920
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.729853830259432e-05,
+      "loss": 0.9523,
+      "step": 328930
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.729739037269457e-05,
+      "loss": 0.8133,
+      "step": 328940
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.729624244279483e-05,
+      "loss": 0.8717,
+      "step": 328950
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.729509451289508e-05,
+      "loss": 0.858,
+      "step": 328960
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.729394658299533e-05,
+      "loss": 0.8213,
+      "step": 328970
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.729279865309558e-05,
+      "loss": 0.9704,
+      "step": 328980
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 3.729165072319583e-05,
+      "loss": 0.9133,
+      "step": 328990
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.729050279329608e-05,
+      "loss": 0.9534,
+      "step": 329000
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.728935486339634e-05,
+      "loss": 0.8357,
+      "step": 329010
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.728820693349659e-05,
+      "loss": 0.891,
+      "step": 329020
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.728705900359684e-05,
+      "loss": 0.8564,
+      "step": 329030
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.7285911073697096e-05,
+      "loss": 0.9659,
+      "step": 329040
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.7284763143797346e-05,
+      "loss": 0.8439,
+      "step": 329050
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.72836152138976e-05,
+      "loss": 0.9397,
+      "step": 329060
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.728246728399785e-05,
+      "loss": 0.8443,
+      "step": 329070
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.728131935409811e-05,
+      "loss": 0.9888,
+      "step": 329080
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.728017142419836e-05,
+      "loss": 0.888,
+      "step": 329090
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.7279023494298615e-05,
+      "loss": 0.8736,
+      "step": 329100
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.7277875564398865e-05,
+      "loss": 0.8873,
+      "step": 329110
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.7276727634499115e-05,
+      "loss": 0.8042,
+      "step": 329120
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.727557970459937e-05,
+      "loss": 0.9789,
+      "step": 329130
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.727443177469962e-05,
+      "loss": 0.9088,
+      "step": 329140
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.727328384479987e-05,
+      "loss": 0.8961,
+      "step": 329150
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.727213591490013e-05,
+      "loss": 0.9472,
+      "step": 329160
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.727098798500038e-05,
+      "loss": 0.8889,
+      "step": 329170
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.726984005510063e-05,
+      "loss": 0.9331,
+      "step": 329180
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.7268692125200884e-05,
+      "loss": 0.8488,
+      "step": 329190
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.7267544195301134e-05,
+      "loss": 0.9256,
+      "step": 329200
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 3.726639626540139e-05,
+      "loss": 0.9673,
+      "step": 329210
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.726524833550164e-05,
+      "loss": 1.0297,
+      "step": 329220
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.72641004056019e-05,
+      "loss": 0.9686,
+      "step": 329230
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.726295247570215e-05,
+      "loss": 0.8455,
+      "step": 329240
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.7261804545802404e-05,
+      "loss": 0.8133,
+      "step": 329250
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.7260656615902654e-05,
+      "loss": 1.0042,
+      "step": 329260
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.7259508686002904e-05,
+      "loss": 0.9062,
+      "step": 329270
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.725836075610316e-05,
+      "loss": 0.962,
+      "step": 329280
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.725721282620341e-05,
+      "loss": 0.8636,
+      "step": 329290
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.725606489630366e-05,
+      "loss": 0.9406,
+      "step": 329300
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.7254916966403917e-05,
+      "loss": 0.8884,
+      "step": 329310
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.7253769036504166e-05,
+      "loss": 0.8411,
+      "step": 329320
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.7252621106604416e-05,
+      "loss": 0.8655,
+      "step": 329330
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.725147317670467e-05,
+      "loss": 0.9139,
+      "step": 329340
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.725032524680492e-05,
+      "loss": 0.9948,
+      "step": 329350
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.724917731690518e-05,
+      "loss": 0.8748,
+      "step": 329360
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.724802938700543e-05,
+      "loss": 0.8699,
+      "step": 329370
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.7246881457105686e-05,
+      "loss": 0.9517,
+      "step": 329380
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.7245733527205936e-05,
+      "loss": 0.9198,
+      "step": 329390
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.724458559730619e-05,
+      "loss": 0.8405,
+      "step": 329400
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.724343766740644e-05,
+      "loss": 0.922,
+      "step": 329410
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.724228973750669e-05,
+      "loss": 0.8486,
+      "step": 329420
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 3.724114180760695e-05,
+      "loss": 0.8758,
+      "step": 329430
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.72399938777072e-05,
+      "loss": 0.8826,
+      "step": 329440
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.723884594780745e-05,
+      "loss": 0.9444,
+      "step": 329450
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.7237698017907705e-05,
+      "loss": 0.925,
+      "step": 329460
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.7236550088007955e-05,
+      "loss": 0.8673,
+      "step": 329470
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.7235402158108205e-05,
+      "loss": 0.9002,
+      "step": 329480
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.723425422820846e-05,
+      "loss": 0.8791,
+      "step": 329490
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.723310629830871e-05,
+      "loss": 0.9396,
+      "step": 329500
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.723195836840897e-05,
+      "loss": 0.9553,
+      "step": 329510
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.723081043850922e-05,
+      "loss": 0.9997,
+      "step": 329520
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.7229662508609474e-05,
+      "loss": 0.9608,
+      "step": 329530
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.7228514578709724e-05,
+      "loss": 0.9115,
+      "step": 329540
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.722736664880998e-05,
+      "loss": 0.9162,
+      "step": 329550
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.722621871891023e-05,
+      "loss": 0.8363,
+      "step": 329560
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.722507078901048e-05,
+      "loss": 0.9929,
+      "step": 329570
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.722392285911073e-05,
+      "loss": 0.9741,
+      "step": 329580
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.722277492921099e-05,
+      "loss": 0.9209,
+      "step": 329590
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.722162699931124e-05,
+      "loss": 0.9661,
+      "step": 329600
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.722047906941149e-05,
+      "loss": 0.9001,
+      "step": 329610
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.721933113951174e-05,
+      "loss": 0.9365,
+      "step": 329620
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.721818320961199e-05,
+      "loss": 0.8502,
+      "step": 329630
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.721703527971225e-05,
+      "loss": 0.8753,
+      "step": 329640
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 3.72158873498125e-05,
+      "loss": 0.9549,
+      "step": 329650
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.7214739419912756e-05,
+      "loss": 0.9187,
+      "step": 329660
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.7213591490013006e-05,
+      "loss": 0.9165,
+      "step": 329670
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.721244356011326e-05,
+      "loss": 0.9442,
+      "step": 329680
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.721129563021351e-05,
+      "loss": 0.9233,
+      "step": 329690
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.721014770031376e-05,
+      "loss": 0.9813,
+      "step": 329700
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.720899977041402e-05,
+      "loss": 0.98,
+      "step": 329710
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.720785184051427e-05,
+      "loss": 0.8241,
+      "step": 329720
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.720670391061452e-05,
+      "loss": 0.9738,
+      "step": 329730
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.7205555980714776e-05,
+      "loss": 0.9347,
+      "step": 329740
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.7204408050815025e-05,
+      "loss": 0.8572,
+      "step": 329750
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.7203260120915275e-05,
+      "loss": 0.9104,
+      "step": 329760
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.720211219101553e-05,
+      "loss": 0.8463,
+      "step": 329770
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.720096426111578e-05,
+      "loss": 0.8761,
+      "step": 329780
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.719981633121604e-05,
+      "loss": 0.919,
+      "step": 329790
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.719866840131629e-05,
+      "loss": 0.9082,
+      "step": 329800
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.7197520471416545e-05,
+      "loss": 0.942,
+      "step": 329810
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.7196372541516795e-05,
+      "loss": 0.9559,
+      "step": 329820
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.719522461161705e-05,
+      "loss": 0.904,
+      "step": 329830
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.71940766817173e-05,
+      "loss": 0.9434,
+      "step": 329840
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.719292875181755e-05,
+      "loss": 0.961,
+      "step": 329850
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 3.719178082191781e-05,
+      "loss": 0.853,
+      "step": 329860
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.719063289201806e-05,
+      "loss": 0.9532,
+      "step": 329870
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.718948496211831e-05,
+      "loss": 0.932,
+      "step": 329880
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.7188337032218564e-05,
+      "loss": 0.9372,
+      "step": 329890
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.7187189102318814e-05,
+      "loss": 0.8443,
+      "step": 329900
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.7186041172419064e-05,
+      "loss": 0.898,
+      "step": 329910
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.718489324251932e-05,
+      "loss": 0.9685,
+      "step": 329920
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.718374531261957e-05,
+      "loss": 1.0109,
+      "step": 329930
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.718259738271983e-05,
+      "loss": 0.8614,
+      "step": 329940
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.718144945282008e-05,
+      "loss": 0.9066,
+      "step": 329950
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.718030152292033e-05,
+      "loss": 0.966,
+      "step": 329960
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.717915359302058e-05,
+      "loss": 0.8838,
+      "step": 329970
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.717800566312084e-05,
+      "loss": 0.9101,
+      "step": 329980
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.717685773322109e-05,
+      "loss": 0.996,
+      "step": 329990
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.717570980332134e-05,
+      "loss": 0.9359,
+      "step": 330000
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.7174561873421596e-05,
+      "loss": 0.8851,
+      "step": 330010
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.7173413943521846e-05,
+      "loss": 0.9262,
+      "step": 330020
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.7172266013622096e-05,
+      "loss": 0.7623,
+      "step": 330030
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.717111808372235e-05,
+      "loss": 0.8496,
+      "step": 330040
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.71699701538226e-05,
+      "loss": 0.8689,
+      "step": 330050
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.716882222392285e-05,
+      "loss": 0.8596,
+      "step": 330060
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.716767429402311e-05,
+      "loss": 0.9886,
+      "step": 330070
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 3.716652636412336e-05,
+      "loss": 0.8893,
+      "step": 330080
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.7165378434223615e-05,
+      "loss": 0.8653,
+      "step": 330090
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.7164230504323865e-05,
+      "loss": 0.7722,
+      "step": 330100
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.716308257442412e-05,
+      "loss": 0.8123,
+      "step": 330110
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.716193464452437e-05,
+      "loss": 1.0231,
+      "step": 330120
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.716078671462463e-05,
+      "loss": 0.9554,
+      "step": 330130
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.715963878472488e-05,
+      "loss": 0.9841,
+      "step": 330140
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.715849085482513e-05,
+      "loss": 0.9228,
+      "step": 330150
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.7157342924925385e-05,
+      "loss": 0.9718,
+      "step": 330160
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.7156194995025634e-05,
+      "loss": 0.8574,
+      "step": 330170
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.7155047065125884e-05,
+      "loss": 0.9777,
+      "step": 330180
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.715389913522614e-05,
+      "loss": 0.8796,
+      "step": 330190
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.715275120532639e-05,
+      "loss": 0.9325,
+      "step": 330200
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.715160327542664e-05,
+      "loss": 0.9702,
+      "step": 330210
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.71504553455269e-05,
+      "loss": 0.9907,
+      "step": 330220
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.714930741562715e-05,
+      "loss": 0.8433,
+      "step": 330230
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.7148159485727404e-05,
+      "loss": 0.8815,
+      "step": 330240
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.7147011555827654e-05,
+      "loss": 0.9188,
+      "step": 330250
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.714586362592791e-05,
+      "loss": 0.9086,
+      "step": 330260
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.714471569602816e-05,
+      "loss": 0.8328,
+      "step": 330270
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.714356776612841e-05,
+      "loss": 1.0253,
+      "step": 330280
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.7142419836228667e-05,
+      "loss": 0.852,
+      "step": 330290
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 3.7141271906328916e-05,
+      "loss": 0.9467,
+      "step": 330300
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.7140123976429166e-05,
+      "loss": 0.8785,
+      "step": 330310
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.713897604652942e-05,
+      "loss": 0.8997,
+      "step": 330320
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.713782811662967e-05,
+      "loss": 0.9232,
+      "step": 330330
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.713668018672992e-05,
+      "loss": 0.8227,
+      "step": 330340
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.713553225683018e-05,
+      "loss": 0.9012,
+      "step": 330350
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.713438432693043e-05,
+      "loss": 0.8411,
+      "step": 330360
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.7133236397030686e-05,
+      "loss": 0.8939,
+      "step": 330370
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.7132088467130936e-05,
+      "loss": 0.9264,
+      "step": 330380
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.713094053723119e-05,
+      "loss": 0.9133,
+      "step": 330390
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.712979260733144e-05,
+      "loss": 0.9187,
+      "step": 330400
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.71286446774317e-05,
+      "loss": 0.8945,
+      "step": 330410
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.712749674753195e-05,
+      "loss": 0.88,
+      "step": 330420
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.71263488176322e-05,
+      "loss": 0.9473,
+      "step": 330430
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.7125200887732455e-05,
+      "loss": 0.8697,
+      "step": 330440
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.7124052957832705e-05,
+      "loss": 0.8229,
+      "step": 330450
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.7122905027932955e-05,
+      "loss": 0.8889,
+      "step": 330460
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.712175709803321e-05,
+      "loss": 0.908,
+      "step": 330470
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.712060916813346e-05,
+      "loss": 0.9319,
+      "step": 330480
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.711946123823371e-05,
+      "loss": 0.9184,
+      "step": 330490
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.711831330833397e-05,
+      "loss": 0.9448,
+      "step": 330500
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.711716537843422e-05,
+      "loss": 0.8163,
+      "step": 330510
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 3.7116017448534474e-05,
+      "loss": 0.9355,
+      "step": 330520
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.7114869518634724e-05,
+      "loss": 0.922,
+      "step": 330530
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.711372158873498e-05,
+      "loss": 0.8893,
+      "step": 330540
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.711257365883523e-05,
+      "loss": 0.9504,
+      "step": 330550
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.711142572893549e-05,
+      "loss": 0.9483,
+      "step": 330560
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.711027779903574e-05,
+      "loss": 0.9179,
+      "step": 330570
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.710912986913599e-05,
+      "loss": 0.8949,
+      "step": 330580
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.7107981939236244e-05,
+      "loss": 0.9331,
+      "step": 330590
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.710683400933649e-05,
+      "loss": 0.8244,
+      "step": 330600
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.710568607943674e-05,
+      "loss": 0.8597,
+      "step": 330610
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.7104538149537e-05,
+      "loss": 0.8643,
+      "step": 330620
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.710339021963725e-05,
+      "loss": 0.9644,
+      "step": 330630
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.71022422897375e-05,
+      "loss": 0.8693,
+      "step": 330640
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.7101094359837756e-05,
+      "loss": 0.8681,
+      "step": 330650
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.7099946429938006e-05,
+      "loss": 0.9241,
+      "step": 330660
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.709879850003826e-05,
+      "loss": 0.9631,
+      "step": 330670
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.709765057013851e-05,
+      "loss": 0.8688,
+      "step": 330680
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.709650264023877e-05,
+      "loss": 0.957,
+      "step": 330690
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.709535471033902e-05,
+      "loss": 0.8927,
+      "step": 330700
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.7094206780439276e-05,
+      "loss": 0.9186,
+      "step": 330710
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.7093058850539526e-05,
+      "loss": 0.9302,
+      "step": 330720
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.7091910920639775e-05,
+      "loss": 0.9606,
+      "step": 330730
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 3.709076299074003e-05,
+      "loss": 0.9533,
+      "step": 330740
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.708961506084028e-05,
+      "loss": 0.9514,
+      "step": 330750
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.708846713094053e-05,
+      "loss": 0.9122,
+      "step": 330760
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.708731920104079e-05,
+      "loss": 0.9475,
+      "step": 330770
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.708617127114104e-05,
+      "loss": 0.9155,
+      "step": 330780
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.708502334124129e-05,
+      "loss": 0.863,
+      "step": 330790
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.7083875411341545e-05,
+      "loss": 1.0148,
+      "step": 330800
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.7082727481441795e-05,
+      "loss": 0.8393,
+      "step": 330810
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.708157955154205e-05,
+      "loss": 1.0565,
+      "step": 330820
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.70804316216423e-05,
+      "loss": 0.8505,
+      "step": 330830
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.707928369174256e-05,
+      "loss": 0.855,
+      "step": 330840
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.707813576184281e-05,
+      "loss": 0.9004,
+      "step": 330850
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.7076987831943064e-05,
+      "loss": 0.8996,
+      "step": 330860
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.7075839902043314e-05,
+      "loss": 0.8892,
+      "step": 330870
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.7074691972143564e-05,
+      "loss": 0.9451,
+      "step": 330880
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.707354404224382e-05,
+      "loss": 0.9653,
+      "step": 330890
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.707239611234407e-05,
+      "loss": 0.9947,
+      "step": 330900
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.707124818244432e-05,
+      "loss": 0.848,
+      "step": 330910
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.707010025254458e-05,
+      "loss": 0.9619,
+      "step": 330920
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.706895232264483e-05,
+      "loss": 0.9351,
+      "step": 330930
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.7067804392745077e-05,
+      "loss": 0.948,
+      "step": 330940
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 3.706665646284533e-05,
+      "loss": 0.8291,
+      "step": 330950
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.706550853294558e-05,
+      "loss": 0.9598,
+      "step": 330960
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.706436060304584e-05,
+      "loss": 0.8478,
+      "step": 330970
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.706321267314609e-05,
+      "loss": 0.8875,
+      "step": 330980
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.7062064743246346e-05,
+      "loss": 0.8595,
+      "step": 330990
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.7060916813346596e-05,
+      "loss": 0.9932,
+      "step": 331000
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.7059768883446846e-05,
+      "loss": 0.912,
+      "step": 331010
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.70586209535471e-05,
+      "loss": 0.9913,
+      "step": 331020
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.705747302364735e-05,
+      "loss": 0.9504,
+      "step": 331030
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.70563250937476e-05,
+      "loss": 0.8861,
+      "step": 331040
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.705517716384786e-05,
+      "loss": 0.8378,
+      "step": 331050
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.705402923394811e-05,
+      "loss": 0.9217,
+      "step": 331060
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.705288130404836e-05,
+      "loss": 0.8974,
+      "step": 331070
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.7051733374148615e-05,
+      "loss": 0.8835,
+      "step": 331080
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.7050585444248865e-05,
+      "loss": 0.9111,
+      "step": 331090
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.704943751434912e-05,
+      "loss": 0.9031,
+      "step": 331100
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.704828958444937e-05,
+      "loss": 0.9139,
+      "step": 331110
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.704714165454963e-05,
+      "loss": 0.9425,
+      "step": 331120
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.704599372464988e-05,
+      "loss": 0.9035,
+      "step": 331130
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.7044845794750135e-05,
+      "loss": 0.9226,
+      "step": 331140
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.7043697864850384e-05,
+      "loss": 0.9172,
+      "step": 331150
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.7042549934950634e-05,
+      "loss": 0.8529,
+      "step": 331160
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 3.704140200505089e-05,
+      "loss": 1.0088,
+      "step": 331170
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.704025407515114e-05,
+      "loss": 0.9217,
+      "step": 331180
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.703910614525139e-05,
+      "loss": 0.8694,
+      "step": 331190
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.703795821535165e-05,
+      "loss": 0.9905,
+      "step": 331200
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.70368102854519e-05,
+      "loss": 0.9165,
+      "step": 331210
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.703566235555215e-05,
+      "loss": 0.9561,
+      "step": 331220
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.7034514425652404e-05,
+      "loss": 0.9453,
+      "step": 331230
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.7033366495752653e-05,
+      "loss": 0.8291,
+      "step": 331240
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.703221856585291e-05,
+      "loss": 0.9377,
+      "step": 331250
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.703107063595316e-05,
+      "loss": 0.8956,
+      "step": 331260
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.7029922706053417e-05,
+      "loss": 0.8981,
+      "step": 331270
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.7028774776153666e-05,
+      "loss": 0.9215,
+      "step": 331280
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.702762684625392e-05,
+      "loss": 0.8441,
+      "step": 331290
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.702647891635417e-05,
+      "loss": 0.9386,
+      "step": 331300
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.702533098645442e-05,
+      "loss": 0.9248,
+      "step": 331310
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.702418305655468e-05,
+      "loss": 0.9277,
+      "step": 331320
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.702303512665493e-05,
+      "loss": 0.9075,
+      "step": 331330
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.702188719675518e-05,
+      "loss": 0.9106,
+      "step": 331340
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.7020739266855436e-05,
+      "loss": 0.9276,
+      "step": 331350
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.7019591336955686e-05,
+      "loss": 0.9227,
+      "step": 331360
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.7018443407055935e-05,
+      "loss": 0.9769,
+      "step": 331370
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.701729547715619e-05,
+      "loss": 0.8677,
+      "step": 331380
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 3.701614754725644e-05,
+      "loss": 1.0348,
+      "step": 331390
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.70149996173567e-05,
+      "loss": 0.8798,
+      "step": 331400
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.701385168745695e-05,
+      "loss": 1.0448,
+      "step": 331410
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.7012703757557205e-05,
+      "loss": 0.9485,
+      "step": 331420
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.7011555827657455e-05,
+      "loss": 1.0124,
+      "step": 331430
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.701040789775771e-05,
+      "loss": 0.8909,
+      "step": 331440
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.700925996785796e-05,
+      "loss": 0.8786,
+      "step": 331450
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.700811203795821e-05,
+      "loss": 0.8403,
+      "step": 331460
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.700696410805847e-05,
+      "loss": 0.85,
+      "step": 331470
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.700581617815872e-05,
+      "loss": 0.889,
+      "step": 331480
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.700466824825897e-05,
+      "loss": 0.8088,
+      "step": 331490
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.7003520318359224e-05,
+      "loss": 0.8342,
+      "step": 331500
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.7002372388459474e-05,
+      "loss": 0.9085,
+      "step": 331510
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.7001224458559724e-05,
+      "loss": 1.0228,
+      "step": 331520
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.700007652865998e-05,
+      "loss": 0.9392,
+      "step": 331530
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.699892859876023e-05,
+      "loss": 0.9234,
+      "step": 331540
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.699778066886049e-05,
+      "loss": 0.8818,
+      "step": 331550
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.699663273896074e-05,
+      "loss": 0.8537,
+      "step": 331560
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.6995484809060994e-05,
+      "loss": 0.8675,
+      "step": 331570
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.699433687916124e-05,
+      "loss": 0.9391,
+      "step": 331580
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.69931889492615e-05,
+      "loss": 0.9444,
+      "step": 331590
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.699204101936175e-05,
+      "loss": 0.9068,
+      "step": 331600
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 3.6990893089462e-05,
+      "loss": 0.8633,
+      "step": 331610
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.6989745159562256e-05,
+      "loss": 0.8551,
+      "step": 331620
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.6988597229662506e-05,
+      "loss": 0.9029,
+      "step": 331630
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.6987449299762756e-05,
+      "loss": 0.933,
+      "step": 331640
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.6986301369863006e-05,
+      "loss": 0.8887,
+      "step": 331650
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.698515343996326e-05,
+      "loss": 0.8952,
+      "step": 331660
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.698400551006351e-05,
+      "loss": 0.9492,
+      "step": 331670
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.698285758016377e-05,
+      "loss": 0.9344,
+      "step": 331680
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.698170965026402e-05,
+      "loss": 0.9703,
+      "step": 331690
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.6980561720364276e-05,
+      "loss": 0.8857,
+      "step": 331700
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.6979413790464525e-05,
+      "loss": 0.964,
+      "step": 331710
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.697826586056478e-05,
+      "loss": 0.9162,
+      "step": 331720
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.697711793066503e-05,
+      "loss": 0.9272,
+      "step": 331730
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.697597000076528e-05,
+      "loss": 0.9064,
+      "step": 331740
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.697482207086554e-05,
+      "loss": 0.9159,
+      "step": 331750
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.697367414096579e-05,
+      "loss": 0.9411,
+      "step": 331760
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.697252621106604e-05,
+      "loss": 0.932,
+      "step": 331770
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.6971378281166295e-05,
+      "loss": 0.9442,
+      "step": 331780
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.6970230351266545e-05,
+      "loss": 0.8809,
+      "step": 331790
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.6969082421366794e-05,
+      "loss": 0.9722,
+      "step": 331800
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.696793449146705e-05,
+      "loss": 0.9608,
+      "step": 331810
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.69667865615673e-05,
+      "loss": 0.9026,
+      "step": 331820
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 3.696563863166756e-05,
+      "loss": 0.8905,
+      "step": 331830
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.696449070176781e-05,
+      "loss": 0.9133,
+      "step": 331840
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.6963342771868064e-05,
+      "loss": 0.9049,
+      "step": 331850
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.6962194841968314e-05,
+      "loss": 0.9576,
+      "step": 331860
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.696104691206857e-05,
+      "loss": 0.957,
+      "step": 331870
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.695989898216882e-05,
+      "loss": 0.9024,
+      "step": 331880
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.695875105226907e-05,
+      "loss": 1.0072,
+      "step": 331890
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.695760312236933e-05,
+      "loss": 0.8178,
+      "step": 331900
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.695645519246958e-05,
+      "loss": 0.994,
+      "step": 331910
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.6955307262569827e-05,
+      "loss": 0.9623,
+      "step": 331920
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.695415933267008e-05,
+      "loss": 0.9584,
+      "step": 331930
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.695301140277033e-05,
+      "loss": 0.9571,
+      "step": 331940
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.695186347287058e-05,
+      "loss": 0.9031,
+      "step": 331950
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.695071554297084e-05,
+      "loss": 0.9516,
+      "step": 331960
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.694956761307109e-05,
+      "loss": 0.9942,
+      "step": 331970
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.6948419683171346e-05,
+      "loss": 0.8355,
+      "step": 331980
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.6947271753271596e-05,
+      "loss": 0.9197,
+      "step": 331990
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.694612382337185e-05,
+      "loss": 0.8672,
+      "step": 332000
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.69449758934721e-05,
+      "loss": 0.8036,
+      "step": 332010
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.694382796357236e-05,
+      "loss": 0.9622,
+      "step": 332020
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.694268003367261e-05,
+      "loss": 0.992,
+      "step": 332030
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 3.694153210377286e-05,
+      "loss": 0.907,
+      "step": 332040
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.6940384173873115e-05,
+      "loss": 0.8371,
+      "step": 332050
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.6939236243973365e-05,
+      "loss": 0.915,
+      "step": 332060
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.6938088314073615e-05,
+      "loss": 0.9769,
+      "step": 332070
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.693694038417387e-05,
+      "loss": 0.8721,
+      "step": 332080
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.693579245427412e-05,
+      "loss": 0.8496,
+      "step": 332090
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.693464452437437e-05,
+      "loss": 0.9224,
+      "step": 332100
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.693349659447463e-05,
+      "loss": 0.889,
+      "step": 332110
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.693234866457488e-05,
+      "loss": 0.9825,
+      "step": 332120
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.6931200734675134e-05,
+      "loss": 0.892,
+      "step": 332130
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.6930052804775384e-05,
+      "loss": 0.832,
+      "step": 332140
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.692890487487564e-05,
+      "loss": 0.8896,
+      "step": 332150
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.692775694497589e-05,
+      "loss": 0.9309,
+      "step": 332160
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.692660901507615e-05,
+      "loss": 0.9512,
+      "step": 332170
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.69254610851764e-05,
+      "loss": 0.8983,
+      "step": 332180
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.692431315527665e-05,
+      "loss": 0.946,
+      "step": 332190
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.6923165225376904e-05,
+      "loss": 0.919,
+      "step": 332200
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.6922017295477154e-05,
+      "loss": 0.999,
+      "step": 332210
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.6920869365577403e-05,
+      "loss": 0.9564,
+      "step": 332220
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.691972143567766e-05,
+      "loss": 0.8483,
+      "step": 332230
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.691857350577791e-05,
+      "loss": 0.9269,
+      "step": 332240
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.691742557587816e-05,
+      "loss": 0.874,
+      "step": 332250
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 3.6916277645978416e-05,
+      "loss": 0.9202,
+      "step": 332260
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.6915129716078666e-05,
+      "loss": 0.8687,
+      "step": 332270
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.691398178617892e-05,
+      "loss": 0.8845,
+      "step": 332280
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.691283385627917e-05,
+      "loss": 0.9653,
+      "step": 332290
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.691168592637943e-05,
+      "loss": 1.0212,
+      "step": 332300
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.691053799647968e-05,
+      "loss": 0.8761,
+      "step": 332310
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.6909390066579936e-05,
+      "loss": 0.9063,
+      "step": 332320
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.6908242136680186e-05,
+      "loss": 0.8698,
+      "step": 332330
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.6907094206780436e-05,
+      "loss": 0.9501,
+      "step": 332340
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.6905946276880685e-05,
+      "loss": 0.9498,
+      "step": 332350
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.690479834698094e-05,
+      "loss": 0.8834,
+      "step": 332360
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.690365041708119e-05,
+      "loss": 0.9246,
+      "step": 332370
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.690250248718144e-05,
+      "loss": 0.9041,
+      "step": 332380
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.69013545572817e-05,
+      "loss": 0.9766,
+      "step": 332390
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.690020662738195e-05,
+      "loss": 0.8945,
+      "step": 332400
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.6899058697482205e-05,
+      "loss": 1.007,
+      "step": 332410
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.6897910767582455e-05,
+      "loss": 0.9058,
+      "step": 332420
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.689676283768271e-05,
+      "loss": 0.889,
+      "step": 332430
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.689561490778296e-05,
+      "loss": 0.8549,
+      "step": 332440
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.689446697788322e-05,
+      "loss": 0.8652,
+      "step": 332450
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.689331904798347e-05,
+      "loss": 0.8878,
+      "step": 332460
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.689217111808372e-05,
+      "loss": 0.9489,
+      "step": 332470
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 3.6891023188183974e-05,
+      "loss": 0.918,
+      "step": 332480
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.6889875258284224e-05,
+      "loss": 0.8468,
+      "step": 332490
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.6888727328384474e-05,
+      "loss": 0.9021,
+      "step": 332500
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.688757939848473e-05,
+      "loss": 0.9093,
+      "step": 332510
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.688643146858498e-05,
+      "loss": 0.9466,
+      "step": 332520
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.688528353868523e-05,
+      "loss": 0.8172,
+      "step": 332530
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.688413560878549e-05,
+      "loss": 0.9547,
+      "step": 332540
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.688298767888574e-05,
+      "loss": 0.8878,
+      "step": 332550
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.688183974898599e-05,
+      "loss": 0.8864,
+      "step": 332560
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.688069181908624e-05,
+      "loss": 0.9189,
+      "step": 332570
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.68795438891865e-05,
+      "loss": 0.8162,
+      "step": 332580
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.687839595928675e-05,
+      "loss": 0.9416,
+      "step": 332590
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.6877248029387006e-05,
+      "loss": 0.9854,
+      "step": 332600
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.6876100099487256e-05,
+      "loss": 0.9729,
+      "step": 332610
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.6874952169587506e-05,
+      "loss": 0.8553,
+      "step": 332620
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.687380423968776e-05,
+      "loss": 1.0118,
+      "step": 332630
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.687265630978801e-05,
+      "loss": 0.964,
+      "step": 332640
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.687150837988826e-05,
+      "loss": 0.8705,
+      "step": 332650
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.687036044998852e-05,
+      "loss": 0.8239,
+      "step": 332660
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.686921252008877e-05,
+      "loss": 0.8813,
+      "step": 332670
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.686806459018902e-05,
+      "loss": 0.9025,
+      "step": 332680
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.6866916660289275e-05,
+      "loss": 0.9376,
+      "step": 332690
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 3.6865768730389525e-05,
+      "loss": 0.8237,
+      "step": 332700
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.686462080048978e-05,
+      "loss": 0.8917,
+      "step": 332710
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.686347287059003e-05,
+      "loss": 0.986,
+      "step": 332720
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.686232494069029e-05,
+      "loss": 0.8311,
+      "step": 332730
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.686117701079054e-05,
+      "loss": 0.8681,
+      "step": 332740
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.6860029080890795e-05,
+      "loss": 0.8761,
+      "step": 332750
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.6858881150991045e-05,
+      "loss": 0.9714,
+      "step": 332760
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.6857733221091295e-05,
+      "loss": 1.0195,
+      "step": 332770
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.685658529119155e-05,
+      "loss": 0.8429,
+      "step": 332780
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.68554373612918e-05,
+      "loss": 0.8848,
+      "step": 332790
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.685428943139205e-05,
+      "loss": 0.8981,
+      "step": 332800
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.685314150149231e-05,
+      "loss": 0.9301,
+      "step": 332810
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.685199357159256e-05,
+      "loss": 0.8654,
+      "step": 332820
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.685084564169281e-05,
+      "loss": 0.9339,
+      "step": 332830
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.6849697711793064e-05,
+      "loss": 0.842,
+      "step": 332840
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.6848549781893314e-05,
+      "loss": 0.9684,
+      "step": 332850
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.684740185199357e-05,
+      "loss": 0.8792,
+      "step": 332860
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.684625392209382e-05,
+      "loss": 0.8962,
+      "step": 332870
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.684510599219408e-05,
+      "loss": 0.9019,
+      "step": 332880
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.684395806229433e-05,
+      "loss": 0.9632,
+      "step": 332890
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.684281013239458e-05,
+      "loss": 0.9307,
+      "step": 332900
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.684166220249483e-05,
+      "loss": 0.9445,
+      "step": 332910
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 3.684051427259508e-05,
+      "loss": 0.898,
+      "step": 332920
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.683936634269534e-05,
+      "loss": 0.9111,
+      "step": 332930
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.683821841279559e-05,
+      "loss": 0.8687,
+      "step": 332940
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.683707048289584e-05,
+      "loss": 1.0007,
+      "step": 332950
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.6835922552996096e-05,
+      "loss": 0.929,
+      "step": 332960
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.6834774623096346e-05,
+      "loss": 0.9133,
+      "step": 332970
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.6833626693196596e-05,
+      "loss": 0.8857,
+      "step": 332980
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.683247876329685e-05,
+      "loss": 1.0406,
+      "step": 332990
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.68313308333971e-05,
+      "loss": 0.9075,
+      "step": 333000
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.683018290349736e-05,
+      "loss": 0.9364,
+      "step": 333010
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.682903497359761e-05,
+      "loss": 1.0105,
+      "step": 333020
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.6827887043697865e-05,
+      "loss": 0.9799,
+      "step": 333030
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.6826739113798115e-05,
+      "loss": 0.9933,
+      "step": 333040
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.6825591183898365e-05,
+      "loss": 0.8567,
+      "step": 333050
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.682444325399862e-05,
+      "loss": 0.8303,
+      "step": 333060
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.682329532409887e-05,
+      "loss": 0.9256,
+      "step": 333070
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.682214739419912e-05,
+      "loss": 0.815,
+      "step": 333080
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.682099946429938e-05,
+      "loss": 0.8805,
+      "step": 333090
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.681985153439963e-05,
+      "loss": 0.9498,
+      "step": 333100
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.681870360449988e-05,
+      "loss": 1.0288,
+      "step": 333110
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.6817555674600134e-05,
+      "loss": 0.8658,
+      "step": 333120
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 3.6816407744700384e-05,
+      "loss": 1.0135,
+      "step": 333130
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.681525981480064e-05,
+      "loss": 0.9681,
+      "step": 333140
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.681411188490089e-05,
+      "loss": 0.8707,
+      "step": 333150
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.681296395500115e-05,
+      "loss": 0.9664,
+      "step": 333160
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.68118160251014e-05,
+      "loss": 0.8942,
+      "step": 333170
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.6810668095201654e-05,
+      "loss": 0.9591,
+      "step": 333180
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.6809520165301904e-05,
+      "loss": 0.8556,
+      "step": 333190
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.6808372235402153e-05,
+      "loss": 0.995,
+      "step": 333200
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.680722430550241e-05,
+      "loss": 0.9838,
+      "step": 333210
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.680607637560266e-05,
+      "loss": 0.9622,
+      "step": 333220
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.680492844570291e-05,
+      "loss": 0.8122,
+      "step": 333230
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.6803780515803166e-05,
+      "loss": 0.917,
+      "step": 333240
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.6802632585903416e-05,
+      "loss": 0.8504,
+      "step": 333250
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.6801484656003666e-05,
+      "loss": 0.937,
+      "step": 333260
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.680033672610392e-05,
+      "loss": 0.8584,
+      "step": 333270
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.679918879620417e-05,
+      "loss": 0.9222,
+      "step": 333280
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.679804086630443e-05,
+      "loss": 0.92,
+      "step": 333290
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.679689293640468e-05,
+      "loss": 0.9931,
+      "step": 333300
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.6795745006504936e-05,
+      "loss": 0.8533,
+      "step": 333310
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.6794597076605186e-05,
+      "loss": 0.941,
+      "step": 333320
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.679344914670544e-05,
+      "loss": 0.9505,
+      "step": 333330
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.679230121680569e-05,
+      "loss": 0.9186,
+      "step": 333340
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 3.679115328690594e-05,
+      "loss": 0.9166,
+      "step": 333350
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.67900053570062e-05,
+      "loss": 0.9578,
+      "step": 333360
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.678885742710645e-05,
+      "loss": 0.923,
+      "step": 333370
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.67877094972067e-05,
+      "loss": 0.8765,
+      "step": 333380
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.6786561567306955e-05,
+      "loss": 0.8978,
+      "step": 333390
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.6785413637407205e-05,
+      "loss": 0.9553,
+      "step": 333400
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.6784265707507455e-05,
+      "loss": 0.8722,
+      "step": 333410
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.678311777760771e-05,
+      "loss": 0.8909,
+      "step": 333420
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.678196984770796e-05,
+      "loss": 0.9034,
+      "step": 333430
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.678082191780822e-05,
+      "loss": 0.9248,
+      "step": 333440
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.677967398790847e-05,
+      "loss": 0.9047,
+      "step": 333450
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.6778526058008724e-05,
+      "loss": 0.8225,
+      "step": 333460
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.6777378128108974e-05,
+      "loss": 0.8081,
+      "step": 333470
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.677623019820923e-05,
+      "loss": 0.7924,
+      "step": 333480
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.677508226830948e-05,
+      "loss": 0.932,
+      "step": 333490
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.677393433840973e-05,
+      "loss": 0.9721,
+      "step": 333500
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.677278640850999e-05,
+      "loss": 0.9812,
+      "step": 333510
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.677163847861024e-05,
+      "loss": 0.8713,
+      "step": 333520
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.677049054871049e-05,
+      "loss": 0.9826,
+      "step": 333530
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.676934261881074e-05,
+      "loss": 0.9811,
+      "step": 333540
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.676819468891099e-05,
+      "loss": 0.8398,
+      "step": 333550
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.676704675901124e-05,
+      "loss": 0.9267,
+      "step": 333560
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 3.67658988291115e-05,
+      "loss": 0.9112,
+      "step": 333570
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.676475089921175e-05,
+      "loss": 1.0182,
+      "step": 333580
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.6763602969312006e-05,
+      "loss": 0.9657,
+      "step": 333590
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.6762455039412256e-05,
+      "loss": 0.9081,
+      "step": 333600
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.676130710951251e-05,
+      "loss": 0.8668,
+      "step": 333610
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.676015917961276e-05,
+      "loss": 0.9893,
+      "step": 333620
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.675901124971302e-05,
+      "loss": 0.8786,
+      "step": 333630
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.675786331981327e-05,
+      "loss": 0.8267,
+      "step": 333640
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.675671538991352e-05,
+      "loss": 0.8732,
+      "step": 333650
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.6755567460013776e-05,
+      "loss": 0.9353,
+      "step": 333660
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.6754419530114025e-05,
+      "loss": 0.9634,
+      "step": 333670
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.6753271600214275e-05,
+      "loss": 0.9172,
+      "step": 333680
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.675212367031453e-05,
+      "loss": 0.9231,
+      "step": 333690
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.675097574041478e-05,
+      "loss": 1.0038,
+      "step": 333700
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.674982781051503e-05,
+      "loss": 0.8987,
+      "step": 333710
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.674867988061529e-05,
+      "loss": 0.8769,
+      "step": 333720
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.674753195071554e-05,
+      "loss": 0.8935,
+      "step": 333730
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.6746384020815795e-05,
+      "loss": 0.9341,
+      "step": 333740
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.6745236090916045e-05,
+      "loss": 0.9344,
+      "step": 333750
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.67440881610163e-05,
+      "loss": 0.8867,
+      "step": 333760
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.674294023111655e-05,
+      "loss": 0.8679,
+      "step": 333770
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.67417923012168e-05,
+      "loss": 0.9004,
+      "step": 333780
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 3.674064437131706e-05,
+      "loss": 0.8964,
+      "step": 333790
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.673949644141731e-05,
+      "loss": 1.0263,
+      "step": 333800
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.673834851151756e-05,
+      "loss": 1.0284,
+      "step": 333810
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.6737200581617814e-05,
+      "loss": 0.9164,
+      "step": 333820
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.6736052651718064e-05,
+      "loss": 0.921,
+      "step": 333830
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.6734904721818314e-05,
+      "loss": 0.9436,
+      "step": 333840
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.673375679191857e-05,
+      "loss": 0.9487,
+      "step": 333850
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.673260886201882e-05,
+      "loss": 1.0345,
+      "step": 333860
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.673146093211908e-05,
+      "loss": 0.9272,
+      "step": 333870
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.6730313002219327e-05,
+      "loss": 0.8937,
+      "step": 333880
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.672916507231958e-05,
+      "loss": 0.9033,
+      "step": 333890
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.672801714241983e-05,
+      "loss": 0.9287,
+      "step": 333900
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.672686921252009e-05,
+      "loss": 0.8751,
+      "step": 333910
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.672572128262034e-05,
+      "loss": 0.9408,
+      "step": 333920
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.672457335272059e-05,
+      "loss": 0.922,
+      "step": 333930
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.6723425422820846e-05,
+      "loss": 0.9154,
+      "step": 333940
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.6722277492921096e-05,
+      "loss": 0.9317,
+      "step": 333950
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.6721129563021346e-05,
+      "loss": 0.9347,
+      "step": 333960
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.67199816331216e-05,
+      "loss": 0.9322,
+      "step": 333970
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.671883370322185e-05,
+      "loss": 0.8846,
+      "step": 333980
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.67176857733221e-05,
+      "loss": 0.9841,
+      "step": 333990
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.671653784342236e-05,
+      "loss": 0.9332,
+      "step": 334000
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 3.671538991352261e-05,
+      "loss": 0.8541,
+      "step": 334010
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.6714241983622865e-05,
+      "loss": 0.8644,
+      "step": 334020
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.6713094053723115e-05,
+      "loss": 0.9011,
+      "step": 334030
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.671194612382337e-05,
+      "loss": 0.8894,
+      "step": 334040
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.671079819392362e-05,
+      "loss": 1.002,
+      "step": 334050
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.670965026402388e-05,
+      "loss": 0.8932,
+      "step": 334060
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.670850233412413e-05,
+      "loss": 0.8679,
+      "step": 334070
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.670735440422438e-05,
+      "loss": 0.9237,
+      "step": 334080
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.6706206474324634e-05,
+      "loss": 0.8946,
+      "step": 334090
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.6705058544424884e-05,
+      "loss": 0.8779,
+      "step": 334100
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.6703910614525134e-05,
+      "loss": 1.0003,
+      "step": 334110
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.670276268462539e-05,
+      "loss": 0.9888,
+      "step": 334120
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.670161475472564e-05,
+      "loss": 0.9589,
+      "step": 334130
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.670046682482589e-05,
+      "loss": 0.8367,
+      "step": 334140
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.669931889492615e-05,
+      "loss": 0.8867,
+      "step": 334150
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.66981709650264e-05,
+      "loss": 1.0033,
+      "step": 334160
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.6697023035126654e-05,
+      "loss": 0.9012,
+      "step": 334170
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.6695875105226903e-05,
+      "loss": 0.9512,
+      "step": 334180
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.669472717532716e-05,
+      "loss": 0.8786,
+      "step": 334190
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.669357924542741e-05,
+      "loss": 0.933,
+      "step": 334200
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.6692431315527667e-05,
+      "loss": 1.0124,
+      "step": 334210
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 3.6691283385627916e-05,
+      "loss": 0.9594,
+      "step": 334220
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.6690135455728166e-05,
+      "loss": 0.9546,
+      "step": 334230
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.668898752582842e-05,
+      "loss": 1.0122,
+      "step": 334240
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.668783959592867e-05,
+      "loss": 0.8859,
+      "step": 334250
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.668669166602892e-05,
+      "loss": 0.9445,
+      "step": 334260
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.668554373612918e-05,
+      "loss": 0.8487,
+      "step": 334270
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.668439580622943e-05,
+      "loss": 1.0061,
+      "step": 334280
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.668324787632968e-05,
+      "loss": 1.059,
+      "step": 334290
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.6682099946429936e-05,
+      "loss": 0.9392,
+      "step": 334300
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.6680952016530185e-05,
+      "loss": 0.8959,
+      "step": 334310
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.667980408663044e-05,
+      "loss": 0.876,
+      "step": 334320
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.667865615673069e-05,
+      "loss": 0.8484,
+      "step": 334330
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.667750822683095e-05,
+      "loss": 0.8933,
+      "step": 334340
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.66763602969312e-05,
+      "loss": 0.873,
+      "step": 334350
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.6675212367031455e-05,
+      "loss": 0.8803,
+      "step": 334360
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.6674064437131705e-05,
+      "loss": 0.9064,
+      "step": 334370
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.6672916507231955e-05,
+      "loss": 0.8939,
+      "step": 334380
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.6671768577332205e-05,
+      "loss": 0.9867,
+      "step": 334390
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.667062064743246e-05,
+      "loss": 0.9344,
+      "step": 334400
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.666947271753271e-05,
+      "loss": 0.9262,
+      "step": 334410
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.666832478763296e-05,
+      "loss": 0.8863,
+      "step": 334420
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.666717685773322e-05,
+      "loss": 0.8015,
+      "step": 334430
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.666602892783347e-05,
+      "loss": 0.9438,
+      "step": 334440
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.6664880997933724e-05,
+      "loss": 0.9722,
+      "step": 334450
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.6663733068033974e-05,
+      "loss": 0.9393,
+      "step": 334460
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.666258513813423e-05,
+      "loss": 0.9327,
+      "step": 334470
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.666143720823448e-05,
+      "loss": 0.9264,
+      "step": 334480
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.666028927833474e-05,
+      "loss": 0.9514,
+      "step": 334490
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.665914134843499e-05,
+      "loss": 0.8517,
+      "step": 334500
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.665799341853524e-05,
+      "loss": 0.8372,
+      "step": 334510
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.665684548863549e-05,
+      "loss": 0.9423,
+      "step": 334520
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.665569755873574e-05,
+      "loss": 0.949,
+      "step": 334530
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.665454962883599e-05,
+      "loss": 0.9255,
+      "step": 334540
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.665340169893625e-05,
+      "loss": 0.893,
+      "step": 334550
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.66522537690365e-05,
+      "loss": 0.8292,
+      "step": 334560
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.665110583913675e-05,
+      "loss": 0.9009,
+      "step": 334570
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.6649957909237006e-05,
+      "loss": 0.877,
+      "step": 334580
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.6648809979337256e-05,
+      "loss": 1.0036,
+      "step": 334590
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.664766204943751e-05,
+      "loss": 0.9952,
+      "step": 334600
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.664651411953776e-05,
+      "loss": 0.8586,
+      "step": 334610
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.664536618963802e-05,
+      "loss": 0.9056,
+      "step": 334620
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.664421825973827e-05,
+      "loss": 0.9031,
+      "step": 334630
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.6643070329838525e-05,
+      "loss": 0.9135,
+      "step": 334640
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.6641922399938775e-05,
+      "loss": 0.9465,
+      "step": 334650
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 3.6640774470039025e-05,
+      "loss": 0.9764,
+      "step": 334660
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.663962654013928e-05,
+      "loss": 0.9642,
+      "step": 334670
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.663847861023953e-05,
+      "loss": 0.9248,
+      "step": 334680
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.663733068033978e-05,
+      "loss": 0.9359,
+      "step": 334690
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.663618275044004e-05,
+      "loss": 0.9493,
+      "step": 334700
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.663503482054029e-05,
+      "loss": 0.8554,
+      "step": 334710
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.663388689064054e-05,
+      "loss": 0.8944,
+      "step": 334720
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.6632738960740795e-05,
+      "loss": 0.9034,
+      "step": 334730
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.6631591030841044e-05,
+      "loss": 0.9349,
+      "step": 334740
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.66304431009413e-05,
+      "loss": 0.9617,
+      "step": 334750
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.662929517104155e-05,
+      "loss": 0.9004,
+      "step": 334760
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.662814724114181e-05,
+      "loss": 0.8809,
+      "step": 334770
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.662699931124206e-05,
+      "loss": 0.9789,
+      "step": 334780
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.6625851381342314e-05,
+      "loss": 0.8743,
+      "step": 334790
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.6624703451442564e-05,
+      "loss": 1.0101,
+      "step": 334800
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.6623555521542814e-05,
+      "loss": 0.9007,
+      "step": 334810
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.662240759164307e-05,
+      "loss": 0.8619,
+      "step": 334820
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.662125966174332e-05,
+      "loss": 0.956,
+      "step": 334830
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.662011173184357e-05,
+      "loss": 0.9039,
+      "step": 334840
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.661896380194383e-05,
+      "loss": 0.9338,
+      "step": 334850
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.6617815872044077e-05,
+      "loss": 0.8082,
+      "step": 334860
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.6616667942144326e-05,
+      "loss": 0.8898,
+      "step": 334870
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 3.661552001224458e-05,
+      "loss": 0.9725,
+      "step": 334880
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.661437208234483e-05,
+      "loss": 0.9135,
+      "step": 334890
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.661322415244509e-05,
+      "loss": 0.9624,
+      "step": 334900
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.661207622254534e-05,
+      "loss": 0.955,
+      "step": 334910
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.6610928292645596e-05,
+      "loss": 0.8667,
+      "step": 334920
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.6609780362745846e-05,
+      "loss": 0.9079,
+      "step": 334930
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.66086324328461e-05,
+      "loss": 0.8792,
+      "step": 334940
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.660748450294635e-05,
+      "loss": 0.9245,
+      "step": 334950
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.66063365730466e-05,
+      "loss": 0.9714,
+      "step": 334960
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.660518864314686e-05,
+      "loss": 0.9252,
+      "step": 334970
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.660404071324711e-05,
+      "loss": 0.9924,
+      "step": 334980
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.660289278334736e-05,
+      "loss": 0.9863,
+      "step": 334990
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.6601744853447615e-05,
+      "loss": 0.9215,
+      "step": 335000
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.6600596923547865e-05,
+      "loss": 0.9754,
+      "step": 335010
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.6599448993648115e-05,
+      "loss": 0.9333,
+      "step": 335020
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.659830106374837e-05,
+      "loss": 0.9895,
+      "step": 335030
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.659715313384862e-05,
+      "loss": 0.8797,
+      "step": 335040
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.659600520394888e-05,
+      "loss": 0.8798,
+      "step": 335050
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.659485727404913e-05,
+      "loss": 0.8785,
+      "step": 335060
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.6593709344149384e-05,
+      "loss": 0.9563,
+      "step": 335070
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.6592561414249634e-05,
+      "loss": 0.8494,
+      "step": 335080
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.6591413484349884e-05,
+      "loss": 0.873,
+      "step": 335090
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 3.659026555445014e-05,
+      "loss": 0.858,
+      "step": 335100
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.658911762455039e-05,
+      "loss": 0.9574,
+      "step": 335110
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.658796969465064e-05,
+      "loss": 0.9123,
+      "step": 335120
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.65868217647509e-05,
+      "loss": 0.9609,
+      "step": 335130
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.658567383485115e-05,
+      "loss": 0.9159,
+      "step": 335140
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.65845259049514e-05,
+      "loss": 0.8956,
+      "step": 335150
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.6583377975051653e-05,
+      "loss": 0.9165,
+      "step": 335160
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.65822300451519e-05,
+      "loss": 0.8317,
+      "step": 335170
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.658108211525216e-05,
+      "loss": 0.9391,
+      "step": 335180
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.657993418535241e-05,
+      "loss": 0.839,
+      "step": 335190
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.6578786255452666e-05,
+      "loss": 0.8946,
+      "step": 335200
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.6577638325552916e-05,
+      "loss": 0.9141,
+      "step": 335210
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.657649039565317e-05,
+      "loss": 0.8346,
+      "step": 335220
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.657534246575342e-05,
+      "loss": 0.8881,
+      "step": 335230
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.657419453585367e-05,
+      "loss": 0.9299,
+      "step": 335240
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.657304660595393e-05,
+      "loss": 0.867,
+      "step": 335250
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.657189867605418e-05,
+      "loss": 1.0443,
+      "step": 335260
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.657075074615443e-05,
+      "loss": 0.9562,
+      "step": 335270
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.6569602816254686e-05,
+      "loss": 0.8839,
+      "step": 335280
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.6568454886354935e-05,
+      "loss": 0.9237,
+      "step": 335290
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.6567306956455185e-05,
+      "loss": 0.9053,
+      "step": 335300
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 3.656615902655544e-05,
+      "loss": 0.8503,
+      "step": 335310
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.656501109665569e-05,
+      "loss": 0.8856,
+      "step": 335320
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.656386316675595e-05,
+      "loss": 0.8063,
+      "step": 335330
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.65627152368562e-05,
+      "loss": 0.8095,
+      "step": 335340
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.6561567306956455e-05,
+      "loss": 0.9832,
+      "step": 335350
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.6560419377056705e-05,
+      "loss": 0.8753,
+      "step": 335360
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.655927144715696e-05,
+      "loss": 0.9617,
+      "step": 335370
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.655812351725721e-05,
+      "loss": 0.951,
+      "step": 335380
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.655697558735746e-05,
+      "loss": 0.8932,
+      "step": 335390
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.655582765745772e-05,
+      "loss": 0.9802,
+      "step": 335400
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.655467972755797e-05,
+      "loss": 0.8332,
+      "step": 335410
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.655353179765822e-05,
+      "loss": 0.9777,
+      "step": 335420
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.6552383867758474e-05,
+      "loss": 0.9577,
+      "step": 335430
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.6551235937858724e-05,
+      "loss": 0.8971,
+      "step": 335440
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.6550088007958974e-05,
+      "loss": 0.9776,
+      "step": 335450
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.654894007805923e-05,
+      "loss": 0.8925,
+      "step": 335460
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.654779214815948e-05,
+      "loss": 0.9467,
+      "step": 335470
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.654664421825974e-05,
+      "loss": 0.9269,
+      "step": 335480
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.654549628835999e-05,
+      "loss": 0.9388,
+      "step": 335490
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.654434835846024e-05,
+      "loss": 0.9865,
+      "step": 335500
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.654320042856049e-05,
+      "loss": 0.9868,
+      "step": 335510
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.654205249866075e-05,
+      "loss": 0.8706,
+      "step": 335520
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 3.6540904568761e-05,
+      "loss": 0.8316,
+      "step": 335530
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.653975663886125e-05,
+      "loss": 0.9607,
+      "step": 335540
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.6538608708961506e-05,
+      "loss": 0.8467,
+      "step": 335550
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.6537460779061756e-05,
+      "loss": 0.8992,
+      "step": 335560
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.6536312849162006e-05,
+      "loss": 0.8935,
+      "step": 335570
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.653516491926226e-05,
+      "loss": 0.85,
+      "step": 335580
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.653401698936251e-05,
+      "loss": 0.8449,
+      "step": 335590
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.653286905946276e-05,
+      "loss": 0.9373,
+      "step": 335600
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.653172112956302e-05,
+      "loss": 0.9586,
+      "step": 335610
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.653057319966327e-05,
+      "loss": 0.9715,
+      "step": 335620
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.6529425269763525e-05,
+      "loss": 0.9583,
+      "step": 335630
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.6528277339863775e-05,
+      "loss": 0.9293,
+      "step": 335640
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.652712940996403e-05,
+      "loss": 0.9539,
+      "step": 335650
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.652598148006428e-05,
+      "loss": 0.9616,
+      "step": 335660
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.652483355016454e-05,
+      "loss": 0.9679,
+      "step": 335670
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.652368562026479e-05,
+      "loss": 0.9643,
+      "step": 335680
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.652253769036504e-05,
+      "loss": 0.9089,
+      "step": 335690
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.6521389760465295e-05,
+      "loss": 0.8278,
+      "step": 335700
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.6520241830565545e-05,
+      "loss": 1.0496,
+      "step": 335710
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.6519093900665794e-05,
+      "loss": 0.9041,
+      "step": 335720
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.651794597076605e-05,
+      "loss": 0.8723,
+      "step": 335730
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.65167980408663e-05,
+      "loss": 0.8994,
+      "step": 335740
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 3.651565011096655e-05,
+      "loss": 0.8788,
+      "step": 335750
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.651450218106681e-05,
+      "loss": 1.0352,
+      "step": 335760
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.651335425116706e-05,
+      "loss": 0.9091,
+      "step": 335770
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.6512206321267314e-05,
+      "loss": 0.8862,
+      "step": 335780
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.6511058391367564e-05,
+      "loss": 0.931,
+      "step": 335790
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.650991046146782e-05,
+      "loss": 0.84,
+      "step": 335800
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.650876253156807e-05,
+      "loss": 1.0168,
+      "step": 335810
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.650761460166832e-05,
+      "loss": 0.9897,
+      "step": 335820
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.650646667176858e-05,
+      "loss": 0.8731,
+      "step": 335830
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.6505318741868826e-05,
+      "loss": 0.9436,
+      "step": 335840
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.6504170811969076e-05,
+      "loss": 0.9212,
+      "step": 335850
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.650302288206933e-05,
+      "loss": 0.9579,
+      "step": 335860
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.650187495216958e-05,
+      "loss": 0.9337,
+      "step": 335870
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.650072702226983e-05,
+      "loss": 0.9054,
+      "step": 335880
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.649957909237009e-05,
+      "loss": 0.9544,
+      "step": 335890
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.649843116247034e-05,
+      "loss": 0.9159,
+      "step": 335900
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.6497283232570596e-05,
+      "loss": 0.9825,
+      "step": 335910
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.6496135302670846e-05,
+      "loss": 0.922,
+      "step": 335920
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.64949873727711e-05,
+      "loss": 0.9758,
+      "step": 335930
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.649383944287135e-05,
+      "loss": 0.8536,
+      "step": 335940
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.649269151297161e-05,
+      "loss": 0.9713,
+      "step": 335950
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.649154358307186e-05,
+      "loss": 0.939,
+      "step": 335960
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 3.649039565317211e-05,
+      "loss": 0.962,
+      "step": 335970
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.6489247723272365e-05,
+      "loss": 0.9761,
+      "step": 335980
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.6488099793372615e-05,
+      "loss": 1.0232,
+      "step": 335990
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.6486951863472865e-05,
+      "loss": 0.922,
+      "step": 336000
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.648580393357312e-05,
+      "loss": 0.8756,
+      "step": 336010
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.648465600367337e-05,
+      "loss": 0.8756,
+      "step": 336020
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.648350807377362e-05,
+      "loss": 0.8753,
+      "step": 336030
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.648236014387388e-05,
+      "loss": 0.9897,
+      "step": 336040
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.648121221397413e-05,
+      "loss": 0.845,
+      "step": 336050
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.6480064284074384e-05,
+      "loss": 0.9158,
+      "step": 336060
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.6478916354174634e-05,
+      "loss": 0.8765,
+      "step": 336070
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.647776842427489e-05,
+      "loss": 0.9163,
+      "step": 336080
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.647662049437514e-05,
+      "loss": 0.9028,
+      "step": 336090
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.64754725644754e-05,
+      "loss": 0.8873,
+      "step": 336100
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.647432463457565e-05,
+      "loss": 0.9673,
+      "step": 336110
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.64731767046759e-05,
+      "loss": 0.8468,
+      "step": 336120
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.6472028774776154e-05,
+      "loss": 0.9641,
+      "step": 336130
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.6470880844876403e-05,
+      "loss": 0.8507,
+      "step": 336140
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.646973291497665e-05,
+      "loss": 0.8774,
+      "step": 336150
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.646858498507691e-05,
+      "loss": 0.9435,
+      "step": 336160
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.646743705517716e-05,
+      "loss": 0.9937,
+      "step": 336170
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.646628912527741e-05,
+      "loss": 0.9398,
+      "step": 336180
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 3.6465141195377666e-05,
+      "loss": 0.9269,
+      "step": 336190
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.6463993265477916e-05,
+      "loss": 0.8737,
+      "step": 336200
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.646284533557817e-05,
+      "loss": 0.8951,
+      "step": 336210
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.646169740567842e-05,
+      "loss": 0.8935,
+      "step": 336220
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.646054947577868e-05,
+      "loss": 1.005,
+      "step": 336230
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.645940154587893e-05,
+      "loss": 0.8988,
+      "step": 336240
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.6458253615979186e-05,
+      "loss": 0.9694,
+      "step": 336250
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.6457105686079436e-05,
+      "loss": 0.9091,
+      "step": 336260
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.6455957756179685e-05,
+      "loss": 0.8719,
+      "step": 336270
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.645480982627994e-05,
+      "loss": 0.8712,
+      "step": 336280
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.645366189638019e-05,
+      "loss": 0.8782,
+      "step": 336290
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.645251396648044e-05,
+      "loss": 0.9484,
+      "step": 336300
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.64513660365807e-05,
+      "loss": 0.8848,
+      "step": 336310
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.645021810668095e-05,
+      "loss": 0.9746,
+      "step": 336320
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.64490701767812e-05,
+      "loss": 0.9107,
+      "step": 336330
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.6447922246881455e-05,
+      "loss": 0.8887,
+      "step": 336340
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.6446774316981705e-05,
+      "loss": 0.8552,
+      "step": 336350
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.644562638708196e-05,
+      "loss": 0.8897,
+      "step": 336360
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.644447845718221e-05,
+      "loss": 0.9126,
+      "step": 336370
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.644333052728247e-05,
+      "loss": 0.8613,
+      "step": 336380
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.644218259738272e-05,
+      "loss": 0.8925,
+      "step": 336390
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 3.6441034667482974e-05,
+      "loss": 0.9493,
+      "step": 336400
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.6439886737583224e-05,
+      "loss": 0.9334,
+      "step": 336410
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.6438738807683474e-05,
+      "loss": 0.877,
+      "step": 336420
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.643759087778373e-05,
+      "loss": 0.9925,
+      "step": 336430
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.643644294788398e-05,
+      "loss": 0.854,
+      "step": 336440
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.643529501798423e-05,
+      "loss": 0.8695,
+      "step": 336450
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.643414708808448e-05,
+      "loss": 0.9689,
+      "step": 336460
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.643299915818474e-05,
+      "loss": 1.0168,
+      "step": 336470
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.6431851228284987e-05,
+      "loss": 0.9537,
+      "step": 336480
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.643070329838524e-05,
+      "loss": 0.9963,
+      "step": 336490
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.642955536848549e-05,
+      "loss": 0.9262,
+      "step": 336500
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.642840743858575e-05,
+      "loss": 0.8847,
+      "step": 336510
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.6427259508686e-05,
+      "loss": 0.9427,
+      "step": 336520
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.6426111578786256e-05,
+      "loss": 0.9342,
+      "step": 336530
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.6424963648886506e-05,
+      "loss": 0.9309,
+      "step": 336540
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.6423815718986756e-05,
+      "loss": 0.9283,
+      "step": 336550
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.642266778908701e-05,
+      "loss": 0.899,
+      "step": 336560
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.642151985918726e-05,
+      "loss": 0.8898,
+      "step": 336570
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.642037192928751e-05,
+      "loss": 0.8716,
+      "step": 336580
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.641922399938777e-05,
+      "loss": 1.0759,
+      "step": 336590
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.641807606948802e-05,
+      "loss": 0.9942,
+      "step": 336600
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.641692813958827e-05,
+      "loss": 0.909,
+      "step": 336610
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 3.6415780209688525e-05,
+      "loss": 0.9422,
+      "step": 336620
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.6414632279788775e-05,
+      "loss": 0.939,
+      "step": 336630
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.641348434988903e-05,
+      "loss": 0.9415,
+      "step": 336640
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.641233641998928e-05,
+      "loss": 0.9467,
+      "step": 336650
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.641118849008954e-05,
+      "loss": 0.9588,
+      "step": 336660
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.641004056018979e-05,
+      "loss": 0.8599,
+      "step": 336670
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.6408892630290045e-05,
+      "loss": 0.8863,
+      "step": 336680
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.6407744700390295e-05,
+      "loss": 0.8506,
+      "step": 336690
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.6406596770490544e-05,
+      "loss": 0.8488,
+      "step": 336700
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.64054488405908e-05,
+      "loss": 0.9418,
+      "step": 336710
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.640430091069105e-05,
+      "loss": 0.9387,
+      "step": 336720
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.64031529807913e-05,
+      "loss": 0.9322,
+      "step": 336730
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.640200505089156e-05,
+      "loss": 0.9661,
+      "step": 336740
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.640085712099181e-05,
+      "loss": 0.9522,
+      "step": 336750
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.639970919109206e-05,
+      "loss": 0.9206,
+      "step": 336760
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.6398561261192314e-05,
+      "loss": 0.818,
+      "step": 336770
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.6397413331292564e-05,
+      "loss": 0.8567,
+      "step": 336780
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.639626540139282e-05,
+      "loss": 0.95,
+      "step": 336790
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.639511747149307e-05,
+      "loss": 0.8658,
+      "step": 336800
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.639396954159333e-05,
+      "loss": 0.913,
+      "step": 336810
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.6392821611693576e-05,
+      "loss": 0.8989,
+      "step": 336820
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.639167368179383e-05,
+      "loss": 0.9456,
+      "step": 336830
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 3.639052575189408e-05,
+      "loss": 0.8862,
+      "step": 336840
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.638937782199433e-05,
+      "loss": 0.9904,
+      "step": 336850
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.638822989209459e-05,
+      "loss": 0.9071,
+      "step": 336860
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.638708196219484e-05,
+      "loss": 0.9494,
+      "step": 336870
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.638593403229509e-05,
+      "loss": 0.9515,
+      "step": 336880
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.6384786102395346e-05,
+      "loss": 0.8582,
+      "step": 336890
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.6383638172495596e-05,
+      "loss": 0.9789,
+      "step": 336900
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.6382490242595846e-05,
+      "loss": 0.9163,
+      "step": 336910
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.63813423126961e-05,
+      "loss": 0.9565,
+      "step": 336920
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.638019438279635e-05,
+      "loss": 1.0507,
+      "step": 336930
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.637904645289661e-05,
+      "loss": 0.9164,
+      "step": 336940
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.637789852299686e-05,
+      "loss": 0.8684,
+      "step": 336950
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.6376750593097115e-05,
+      "loss": 0.8648,
+      "step": 336960
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.6375602663197365e-05,
+      "loss": 0.9474,
+      "step": 336970
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.637445473329762e-05,
+      "loss": 0.9221,
+      "step": 336980
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.637330680339787e-05,
+      "loss": 0.9488,
+      "step": 336990
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.637215887349812e-05,
+      "loss": 0.9262,
+      "step": 337000
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.637101094359838e-05,
+      "loss": 1.0307,
+      "step": 337010
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.636986301369863e-05,
+      "loss": 0.9425,
+      "step": 337020
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.636871508379888e-05,
+      "loss": 1.0292,
+      "step": 337030
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.6367567153899134e-05,
+      "loss": 0.9633,
+      "step": 337040
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.6366419223999384e-05,
+      "loss": 0.9292,
+      "step": 337050
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 3.6365271294099634e-05,
+      "loss": 0.9801,
+      "step": 337060
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.636412336419989e-05,
+      "loss": 0.933,
+      "step": 337070
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.636297543430014e-05,
+      "loss": 0.9085,
+      "step": 337080
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.63618275044004e-05,
+      "loss": 1.0128,
+      "step": 337090
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.636067957450065e-05,
+      "loss": 0.9599,
+      "step": 337100
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.6359531644600904e-05,
+      "loss": 0.8511,
+      "step": 337110
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.6358383714701153e-05,
+      "loss": 0.8896,
+      "step": 337120
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.63572357848014e-05,
+      "loss": 0.8984,
+      "step": 337130
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.635608785490166e-05,
+      "loss": 0.8894,
+      "step": 337140
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.635493992500191e-05,
+      "loss": 0.9445,
+      "step": 337150
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.635379199510216e-05,
+      "loss": 0.9124,
+      "step": 337160
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.6352644065202416e-05,
+      "loss": 0.8838,
+      "step": 337170
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.6351496135302666e-05,
+      "loss": 0.9724,
+      "step": 337180
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.6350348205402916e-05,
+      "loss": 0.8396,
+      "step": 337190
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.634920027550317e-05,
+      "loss": 0.9339,
+      "step": 337200
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.634805234560342e-05,
+      "loss": 0.8821,
+      "step": 337210
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.634690441570368e-05,
+      "loss": 0.9494,
+      "step": 337220
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.634575648580393e-05,
+      "loss": 0.9317,
+      "step": 337230
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.6344608555904186e-05,
+      "loss": 0.8519,
+      "step": 337240
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.6343460626004435e-05,
+      "loss": 0.8984,
+      "step": 337250
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.634231269610469e-05,
+      "loss": 1.0389,
+      "step": 337260
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 3.634116476620494e-05,
+      "loss": 0.9501,
+      "step": 337270
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.634001683630519e-05,
+      "loss": 0.9927,
+      "step": 337280
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.633886890640545e-05,
+      "loss": 0.8736,
+      "step": 337290
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.63377209765057e-05,
+      "loss": 0.9872,
+      "step": 337300
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.633657304660595e-05,
+      "loss": 0.9724,
+      "step": 337310
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.6335425116706205e-05,
+      "loss": 0.8263,
+      "step": 337320
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.6334277186806455e-05,
+      "loss": 0.7659,
+      "step": 337330
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.6333129256906704e-05,
+      "loss": 0.9788,
+      "step": 337340
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.633198132700696e-05,
+      "loss": 0.9324,
+      "step": 337350
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.633083339710721e-05,
+      "loss": 0.9434,
+      "step": 337360
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.632968546720747e-05,
+      "loss": 0.895,
+      "step": 337370
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.632853753730772e-05,
+      "loss": 0.9662,
+      "step": 337380
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.6327389607407974e-05,
+      "loss": 0.9038,
+      "step": 337390
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.6326241677508224e-05,
+      "loss": 0.9335,
+      "step": 337400
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.632509374760848e-05,
+      "loss": 0.8921,
+      "step": 337410
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.632394581770873e-05,
+      "loss": 0.9277,
+      "step": 337420
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.632279788780898e-05,
+      "loss": 0.9145,
+      "step": 337430
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.632164995790924e-05,
+      "loss": 0.9356,
+      "step": 337440
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.632050202800949e-05,
+      "loss": 0.8874,
+      "step": 337450
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.6319354098109737e-05,
+      "loss": 0.9973,
+      "step": 337460
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.631820616820999e-05,
+      "loss": 0.8106,
+      "step": 337470
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.631705823831024e-05,
+      "loss": 0.9665,
+      "step": 337480
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 3.631591030841049e-05,
+      "loss": 0.9233,
+      "step": 337490
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.631476237851075e-05,
+      "loss": 1.0367,
+      "step": 337500
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.6313614448611e-05,
+      "loss": 1.031,
+      "step": 337510
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.6312466518711256e-05,
+      "loss": 0.9079,
+      "step": 337520
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.6311318588811506e-05,
+      "loss": 0.9115,
+      "step": 337530
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.631017065891176e-05,
+      "loss": 0.8928,
+      "step": 337540
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.630902272901201e-05,
+      "loss": 0.9173,
+      "step": 337550
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.630787479911227e-05,
+      "loss": 0.9374,
+      "step": 337560
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.630672686921252e-05,
+      "loss": 0.9069,
+      "step": 337570
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.630557893931277e-05,
+      "loss": 0.9553,
+      "step": 337580
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.6304431009413025e-05,
+      "loss": 1.0475,
+      "step": 337590
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.6303283079513275e-05,
+      "loss": 0.9415,
+      "step": 337600
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.6302135149613525e-05,
+      "loss": 0.8776,
+      "step": 337610
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.630098721971378e-05,
+      "loss": 0.8873,
+      "step": 337620
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.629983928981403e-05,
+      "loss": 0.9972,
+      "step": 337630
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.629869135991428e-05,
+      "loss": 0.9683,
+      "step": 337640
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.629754343001454e-05,
+      "loss": 0.9231,
+      "step": 337650
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.629639550011479e-05,
+      "loss": 0.9672,
+      "step": 337660
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.6295247570215044e-05,
+      "loss": 0.8799,
+      "step": 337670
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.6294099640315294e-05,
+      "loss": 0.9541,
+      "step": 337680
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.629295171041555e-05,
+      "loss": 0.9975,
+      "step": 337690
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.62918037805158e-05,
+      "loss": 1.0283,
+      "step": 337700
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 3.629065585061606e-05,
+      "loss": 0.8494,
+      "step": 337710
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.628950792071631e-05,
+      "loss": 0.8172,
+      "step": 337720
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.628835999081656e-05,
+      "loss": 0.982,
+      "step": 337730
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.6287212060916814e-05,
+      "loss": 0.8578,
+      "step": 337740
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.6286064131017064e-05,
+      "loss": 0.9785,
+      "step": 337750
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.6284916201117314e-05,
+      "loss": 0.8696,
+      "step": 337760
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.628376827121757e-05,
+      "loss": 0.9647,
+      "step": 337770
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.628262034131782e-05,
+      "loss": 0.9513,
+      "step": 337780
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.628147241141807e-05,
+      "loss": 0.9906,
+      "step": 337790
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.6280324481518326e-05,
+      "loss": 0.8684,
+      "step": 337800
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.6279176551618576e-05,
+      "loss": 0.903,
+      "step": 337810
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.627802862171883e-05,
+      "loss": 0.9812,
+      "step": 337820
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.627688069181908e-05,
+      "loss": 0.9364,
+      "step": 337830
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.627573276191934e-05,
+      "loss": 0.9555,
+      "step": 337840
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.627458483201959e-05,
+      "loss": 0.8589,
+      "step": 337850
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.627343690211984e-05,
+      "loss": 0.9904,
+      "step": 337860
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.6272288972220096e-05,
+      "loss": 0.9027,
+      "step": 337870
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.6271141042320346e-05,
+      "loss": 0.8671,
+      "step": 337880
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.6269993112420596e-05,
+      "loss": 0.9053,
+      "step": 337890
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.626884518252085e-05,
+      "loss": 0.9373,
+      "step": 337900
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.62676972526211e-05,
+      "loss": 0.9918,
+      "step": 337910
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.626654932272135e-05,
+      "loss": 0.8368,
+      "step": 337920
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 3.626540139282161e-05,
+      "loss": 0.9434,
+      "step": 337930
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.626425346292186e-05,
+      "loss": 0.8874,
+      "step": 337940
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.6263105533022115e-05,
+      "loss": 1.0214,
+      "step": 337950
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.6261957603122365e-05,
+      "loss": 0.8742,
+      "step": 337960
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.626080967322262e-05,
+      "loss": 0.917,
+      "step": 337970
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.625966174332287e-05,
+      "loss": 0.8689,
+      "step": 337980
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.625851381342313e-05,
+      "loss": 0.931,
+      "step": 337990
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.625736588352338e-05,
+      "loss": 0.8551,
+      "step": 338000
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.625621795362363e-05,
+      "loss": 0.9273,
+      "step": 338010
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.6255070023723884e-05,
+      "loss": 0.9585,
+      "step": 338020
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.6253922093824134e-05,
+      "loss": 0.8719,
+      "step": 338030
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.6252774163924384e-05,
+      "loss": 1.0035,
+      "step": 338040
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.625162623402464e-05,
+      "loss": 0.8843,
+      "step": 338050
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.625047830412489e-05,
+      "loss": 0.9206,
+      "step": 338060
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.624933037422514e-05,
+      "loss": 0.8932,
+      "step": 338070
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.62481824443254e-05,
+      "loss": 0.9134,
+      "step": 338080
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.624703451442565e-05,
+      "loss": 0.9395,
+      "step": 338090
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.6245886584525903e-05,
+      "loss": 0.9048,
+      "step": 338100
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.624473865462615e-05,
+      "loss": 0.8701,
+      "step": 338110
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.624359072472641e-05,
+      "loss": 0.8625,
+      "step": 338120
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.624244279482666e-05,
+      "loss": 0.9121,
+      "step": 338130
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.6241294864926916e-05,
+      "loss": 0.9663,
+      "step": 338140
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 3.6240146935027166e-05,
+      "loss": 0.9307,
+      "step": 338150
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.6238999005127416e-05,
+      "loss": 1.0348,
+      "step": 338160
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.623785107522767e-05,
+      "loss": 0.9255,
+      "step": 338170
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.623670314532792e-05,
+      "loss": 0.9196,
+      "step": 338180
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.623555521542817e-05,
+      "loss": 0.9135,
+      "step": 338190
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.623440728552843e-05,
+      "loss": 0.9606,
+      "step": 338200
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.623325935562868e-05,
+      "loss": 0.9334,
+      "step": 338210
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.623211142572893e-05,
+      "loss": 1.0307,
+      "step": 338220
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.6230963495829185e-05,
+      "loss": 0.9025,
+      "step": 338230
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.6229815565929435e-05,
+      "loss": 0.8928,
+      "step": 338240
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.622866763602969e-05,
+      "loss": 0.9862,
+      "step": 338250
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.622751970612994e-05,
+      "loss": 0.9252,
+      "step": 338260
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.62263717762302e-05,
+      "loss": 0.873,
+      "step": 338270
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.622522384633045e-05,
+      "loss": 0.8017,
+      "step": 338280
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.6224075916430705e-05,
+      "loss": 0.9205,
+      "step": 338290
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.6222927986530955e-05,
+      "loss": 0.8196,
+      "step": 338300
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.6221780056631205e-05,
+      "loss": 1.0943,
+      "step": 338310
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.622063212673146e-05,
+      "loss": 1.0151,
+      "step": 338320
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.621948419683171e-05,
+      "loss": 0.8294,
+      "step": 338330
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.621833626693196e-05,
+      "loss": 0.8818,
+      "step": 338340
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.621718833703222e-05,
+      "loss": 0.8977,
+      "step": 338350
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 3.621604040713247e-05,
+      "loss": 1.0804,
+      "step": 338360
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.621489247723272e-05,
+      "loss": 0.9956,
+      "step": 338370
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.6213744547332974e-05,
+      "loss": 0.8708,
+      "step": 338380
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.6212596617433224e-05,
+      "loss": 0.9017,
+      "step": 338390
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.621144868753348e-05,
+      "loss": 0.9378,
+      "step": 338400
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.621030075763373e-05,
+      "loss": 0.8996,
+      "step": 338410
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.620915282773399e-05,
+      "loss": 0.9894,
+      "step": 338420
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.620800489783424e-05,
+      "loss": 0.9832,
+      "step": 338430
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.620685696793449e-05,
+      "loss": 1.0745,
+      "step": 338440
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.620570903803474e-05,
+      "loss": 0.8796,
+      "step": 338450
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.620456110813499e-05,
+      "loss": 0.9402,
+      "step": 338460
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.620341317823525e-05,
+      "loss": 0.9016,
+      "step": 338470
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.62022652483355e-05,
+      "loss": 1.0265,
+      "step": 338480
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.620111731843575e-05,
+      "loss": 0.8707,
+      "step": 338490
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.6199969388536e-05,
+      "loss": 0.9006,
+      "step": 338500
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.6198821458636256e-05,
+      "loss": 0.9268,
+      "step": 338510
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.6197673528736506e-05,
+      "loss": 0.9587,
+      "step": 338520
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.619652559883676e-05,
+      "loss": 0.9585,
+      "step": 338530
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.619537766893701e-05,
+      "loss": 0.999,
+      "step": 338540
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.619422973903727e-05,
+      "loss": 0.8842,
+      "step": 338550
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.619308180913752e-05,
+      "loss": 0.922,
+      "step": 338560
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.6191933879237775e-05,
+      "loss": 0.8379,
+      "step": 338570
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 3.6190785949338025e-05,
+      "loss": 0.8722,
+      "step": 338580
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.6189638019438275e-05,
+      "loss": 0.9868,
+      "step": 338590
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.618849008953853e-05,
+      "loss": 0.9296,
+      "step": 338600
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.618734215963878e-05,
+      "loss": 0.8836,
+      "step": 338610
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.618619422973903e-05,
+      "loss": 1.0248,
+      "step": 338620
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.618504629983929e-05,
+      "loss": 0.9084,
+      "step": 338630
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.618389836993954e-05,
+      "loss": 1.0027,
+      "step": 338640
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.618275044003979e-05,
+      "loss": 0.9202,
+      "step": 338650
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.6181602510140044e-05,
+      "loss": 0.9229,
+      "step": 338660
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.6180454580240294e-05,
+      "loss": 1.0136,
+      "step": 338670
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.617930665034055e-05,
+      "loss": 0.8485,
+      "step": 338680
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.61781587204408e-05,
+      "loss": 0.8891,
+      "step": 338690
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.617701079054106e-05,
+      "loss": 0.8857,
+      "step": 338700
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.617586286064131e-05,
+      "loss": 0.8778,
+      "step": 338710
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.6174714930741564e-05,
+      "loss": 0.9987,
+      "step": 338720
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.6173567000841814e-05,
+      "loss": 0.9066,
+      "step": 338730
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.6172419070942064e-05,
+      "loss": 0.8695,
+      "step": 338740
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.617127114104232e-05,
+      "loss": 0.9872,
+      "step": 338750
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.617012321114257e-05,
+      "loss": 0.9072,
+      "step": 338760
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.616897528124282e-05,
+      "loss": 0.966,
+      "step": 338770
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.6167827351343076e-05,
+      "loss": 0.9164,
+      "step": 338780
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.6166679421443326e-05,
+      "loss": 0.8958,
+      "step": 338790
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 3.6165531491543576e-05,
+      "loss": 0.9217,
+      "step": 338800
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.616438356164383e-05,
+      "loss": 0.9586,
+      "step": 338810
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.616323563174408e-05,
+      "loss": 1.0155,
+      "step": 338820
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.616208770184434e-05,
+      "loss": 1.0123,
+      "step": 338830
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.616093977194459e-05,
+      "loss": 0.8866,
+      "step": 338840
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.6159791842044846e-05,
+      "loss": 0.9662,
+      "step": 338850
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.6158643912145096e-05,
+      "loss": 0.9171,
+      "step": 338860
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.615749598224535e-05,
+      "loss": 1.0157,
+      "step": 338870
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.61563480523456e-05,
+      "loss": 0.9902,
+      "step": 338880
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.615520012244585e-05,
+      "loss": 0.9025,
+      "step": 338890
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.615405219254611e-05,
+      "loss": 0.9613,
+      "step": 338900
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.615290426264636e-05,
+      "loss": 0.9308,
+      "step": 338910
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.615175633274661e-05,
+      "loss": 0.8412,
+      "step": 338920
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.6150608402846865e-05,
+      "loss": 0.917,
+      "step": 338930
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.6149460472947115e-05,
+      "loss": 0.8603,
+      "step": 338940
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.6148312543047365e-05,
+      "loss": 0.86,
+      "step": 338950
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.614716461314762e-05,
+      "loss": 0.8779,
+      "step": 338960
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.614601668324787e-05,
+      "loss": 0.9552,
+      "step": 338970
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.614486875334813e-05,
+      "loss": 0.8856,
+      "step": 338980
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.614372082344838e-05,
+      "loss": 0.9208,
+      "step": 338990
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.6142572893548634e-05,
+      "loss": 0.8909,
+      "step": 339000
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.6141424963648884e-05,
+      "loss": 1.0242,
+      "step": 339010
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 3.614027703374914e-05,
+      "loss": 0.9596,
+      "step": 339020
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.613912910384939e-05,
+      "loss": 0.9072,
+      "step": 339030
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.613798117394964e-05,
+      "loss": 0.9372,
+      "step": 339040
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.61368332440499e-05,
+      "loss": 0.9457,
+      "step": 339050
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.613568531415015e-05,
+      "loss": 0.8203,
+      "step": 339060
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.61345373842504e-05,
+      "loss": 0.9712,
+      "step": 339070
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.6133389454350653e-05,
+      "loss": 0.9209,
+      "step": 339080
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.61322415244509e-05,
+      "loss": 1.0071,
+      "step": 339090
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.613109359455115e-05,
+      "loss": 0.8873,
+      "step": 339100
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.612994566465141e-05,
+      "loss": 0.8912,
+      "step": 339110
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.612879773475166e-05,
+      "loss": 0.8551,
+      "step": 339120
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.6127649804851916e-05,
+      "loss": 1.0279,
+      "step": 339130
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.6126501874952166e-05,
+      "loss": 0.9128,
+      "step": 339140
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.612535394505242e-05,
+      "loss": 0.9584,
+      "step": 339150
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.612420601515267e-05,
+      "loss": 1.0817,
+      "step": 339160
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.612305808525293e-05,
+      "loss": 0.9846,
+      "step": 339170
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.612191015535318e-05,
+      "loss": 1.0432,
+      "step": 339180
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.612076222545343e-05,
+      "loss": 0.9296,
+      "step": 339190
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.611961429555368e-05,
+      "loss": 0.9218,
+      "step": 339200
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.6118466365653935e-05,
+      "loss": 1.0352,
+      "step": 339210
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.6117318435754185e-05,
+      "loss": 0.9051,
+      "step": 339220
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.6116170505854435e-05,
+      "loss": 0.878,
+      "step": 339230
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 3.611502257595469e-05,
+      "loss": 0.8908,
+      "step": 339240
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.611387464605494e-05,
+      "loss": 0.9286,
+      "step": 339250
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.61127267161552e-05,
+      "loss": 0.9434,
+      "step": 339260
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.611157878625545e-05,
+      "loss": 0.8914,
+      "step": 339270
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.6110430856355705e-05,
+      "loss": 0.7818,
+      "step": 339280
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.6109282926455955e-05,
+      "loss": 0.9299,
+      "step": 339290
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.610813499655621e-05,
+      "loss": 0.8912,
+      "step": 339300
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.610698706665646e-05,
+      "loss": 0.9394,
+      "step": 339310
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.610583913675671e-05,
+      "loss": 0.8392,
+      "step": 339320
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.610469120685697e-05,
+      "loss": 0.8486,
+      "step": 339330
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.610354327695722e-05,
+      "loss": 0.8387,
+      "step": 339340
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.610239534705747e-05,
+      "loss": 0.895,
+      "step": 339350
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.6101247417157724e-05,
+      "loss": 0.9187,
+      "step": 339360
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.6100099487257974e-05,
+      "loss": 0.8478,
+      "step": 339370
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.6098951557358224e-05,
+      "loss": 0.9518,
+      "step": 339380
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.609780362745848e-05,
+      "loss": 0.8668,
+      "step": 339390
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.609665569755873e-05,
+      "loss": 0.9963,
+      "step": 339400
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.609550776765899e-05,
+      "loss": 0.812,
+      "step": 339410
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.6094359837759237e-05,
+      "loss": 0.9026,
+      "step": 339420
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.609321190785949e-05,
+      "loss": 0.9651,
+      "step": 339430
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.609206397795974e-05,
+      "loss": 0.9427,
+      "step": 339440
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 3.609091604806e-05,
+      "loss": 0.9418,
+      "step": 339450
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.608976811816025e-05,
+      "loss": 0.9528,
+      "step": 339460
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.60886201882605e-05,
+      "loss": 0.8713,
+      "step": 339470
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.6087472258360756e-05,
+      "loss": 0.8217,
+      "step": 339480
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.6086324328461006e-05,
+      "loss": 0.8059,
+      "step": 339490
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.6085176398561256e-05,
+      "loss": 0.9092,
+      "step": 339500
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.608402846866151e-05,
+      "loss": 0.9898,
+      "step": 339510
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.608288053876176e-05,
+      "loss": 0.9282,
+      "step": 339520
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.608173260886201e-05,
+      "loss": 0.9184,
+      "step": 339530
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.608058467896227e-05,
+      "loss": 1.0103,
+      "step": 339540
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.607943674906252e-05,
+      "loss": 0.8167,
+      "step": 339550
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.6078288819162775e-05,
+      "loss": 0.9615,
+      "step": 339560
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.6077140889263025e-05,
+      "loss": 0.9231,
+      "step": 339570
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.607599295936328e-05,
+      "loss": 0.8675,
+      "step": 339580
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.607484502946353e-05,
+      "loss": 0.9009,
+      "step": 339590
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.607369709956379e-05,
+      "loss": 0.9081,
+      "step": 339600
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.607254916966404e-05,
+      "loss": 0.979,
+      "step": 339610
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.607140123976429e-05,
+      "loss": 0.9038,
+      "step": 339620
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.6070253309864544e-05,
+      "loss": 0.9689,
+      "step": 339630
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.6069105379964794e-05,
+      "loss": 0.9006,
+      "step": 339640
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.6067957450065044e-05,
+      "loss": 0.7952,
+      "step": 339650
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.60668095201653e-05,
+      "loss": 0.9691,
+      "step": 339660
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 3.606566159026555e-05,
+      "loss": 0.9413,
+      "step": 339670
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.60645136603658e-05,
+      "loss": 0.9025,
+      "step": 339680
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.606336573046606e-05,
+      "loss": 0.9019,
+      "step": 339690
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.606221780056631e-05,
+      "loss": 0.9316,
+      "step": 339700
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.6061069870666564e-05,
+      "loss": 0.9684,
+      "step": 339710
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.6059921940766814e-05,
+      "loss": 0.8917,
+      "step": 339720
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.605877401086707e-05,
+      "loss": 0.951,
+      "step": 339730
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.605762608096732e-05,
+      "loss": 0.9089,
+      "step": 339740
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.605647815106758e-05,
+      "loss": 0.93,
+      "step": 339750
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.6055330221167826e-05,
+      "loss": 0.9213,
+      "step": 339760
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.6054182291268076e-05,
+      "loss": 1.0036,
+      "step": 339770
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.605303436136833e-05,
+      "loss": 0.8417,
+      "step": 339780
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.605188643146858e-05,
+      "loss": 0.9507,
+      "step": 339790
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.605073850156883e-05,
+      "loss": 0.9769,
+      "step": 339800
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.604959057166909e-05,
+      "loss": 0.9331,
+      "step": 339810
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.604844264176934e-05,
+      "loss": 0.9519,
+      "step": 339820
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.604729471186959e-05,
+      "loss": 0.9612,
+      "step": 339830
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.6046146781969846e-05,
+      "loss": 0.8276,
+      "step": 339840
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.6044998852070095e-05,
+      "loss": 0.8711,
+      "step": 339850
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.604385092217035e-05,
+      "loss": 0.9405,
+      "step": 339860
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.60427029922706e-05,
+      "loss": 0.9324,
+      "step": 339870
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.604155506237086e-05,
+      "loss": 0.9467,
+      "step": 339880
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 3.604040713247111e-05,
+      "loss": 0.8891,
+      "step": 339890
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.603925920257136e-05,
+      "loss": 0.8926,
+      "step": 339900
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.6038111272671615e-05,
+      "loss": 0.8386,
+      "step": 339910
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.6036963342771865e-05,
+      "loss": 0.9535,
+      "step": 339920
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.6035815412872115e-05,
+      "loss": 1.0423,
+      "step": 339930
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.603466748297237e-05,
+      "loss": 0.9284,
+      "step": 339940
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.603351955307262e-05,
+      "loss": 1.0071,
+      "step": 339950
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.603237162317287e-05,
+      "loss": 0.9712,
+      "step": 339960
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.603122369327313e-05,
+      "loss": 0.9001,
+      "step": 339970
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.603007576337338e-05,
+      "loss": 0.9406,
+      "step": 339980
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.6028927833473634e-05,
+      "loss": 0.8821,
+      "step": 339990
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.6027779903573884e-05,
+      "loss": 0.9861,
+      "step": 340000
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.602663197367414e-05,
+      "loss": 0.8581,
+      "step": 340010
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.602548404377439e-05,
+      "loss": 0.8368,
+      "step": 340020
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.602433611387465e-05,
+      "loss": 0.9554,
+      "step": 340030
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.60231881839749e-05,
+      "loss": 0.8845,
+      "step": 340040
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.602204025407515e-05,
+      "loss": 0.8955,
+      "step": 340050
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.6020892324175403e-05,
+      "loss": 0.8604,
+      "step": 340060
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.601974439427565e-05,
+      "loss": 0.8503,
+      "step": 340070
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.60185964643759e-05,
+      "loss": 1.0093,
+      "step": 340080
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.601744853447616e-05,
+      "loss": 1.0041,
+      "step": 340090
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.601630060457641e-05,
+      "loss": 0.9062,
+      "step": 340100
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 3.601515267467666e-05,
+      "loss": 0.8635,
+      "step": 340110
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.6014004744776916e-05,
+      "loss": 0.9524,
+      "step": 340120
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.6012856814877166e-05,
+      "loss": 0.902,
+      "step": 340130
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.601170888497742e-05,
+      "loss": 0.8461,
+      "step": 340140
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.601056095507767e-05,
+      "loss": 0.8478,
+      "step": 340150
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.600941302517793e-05,
+      "loss": 0.8399,
+      "step": 340160
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.600826509527818e-05,
+      "loss": 0.8665,
+      "step": 340170
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.6007117165378436e-05,
+      "loss": 0.9217,
+      "step": 340180
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.6005969235478685e-05,
+      "loss": 0.8556,
+      "step": 340190
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.6004821305578935e-05,
+      "loss": 0.9235,
+      "step": 340200
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.600367337567919e-05,
+      "loss": 0.9584,
+      "step": 340210
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.600252544577944e-05,
+      "loss": 0.828,
+      "step": 340220
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.600137751587969e-05,
+      "loss": 0.9616,
+      "step": 340230
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.600022958597995e-05,
+      "loss": 0.9278,
+      "step": 340240
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.59990816560802e-05,
+      "loss": 0.9077,
+      "step": 340250
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.599793372618045e-05,
+      "loss": 0.8548,
+      "step": 340260
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.5996785796280705e-05,
+      "loss": 1.0368,
+      "step": 340270
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.5995637866380954e-05,
+      "loss": 0.9629,
+      "step": 340280
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.599448993648121e-05,
+      "loss": 0.9594,
+      "step": 340290
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.599334200658146e-05,
+      "loss": 0.9267,
+      "step": 340300
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.599219407668172e-05,
+      "loss": 0.8911,
+      "step": 340310
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.599104614678197e-05,
+      "loss": 0.9086,
+      "step": 340320
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 3.5989898216882224e-05,
+      "loss": 0.9458,
+      "step": 340330
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.5988750286982474e-05,
+      "loss": 0.9865,
+      "step": 340340
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.5987602357082724e-05,
+      "loss": 0.9066,
+      "step": 340350
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.598645442718298e-05,
+      "loss": 0.8843,
+      "step": 340360
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.598530649728323e-05,
+      "loss": 0.8744,
+      "step": 340370
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.598415856738348e-05,
+      "loss": 0.9913,
+      "step": 340380
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.598301063748374e-05,
+      "loss": 0.9857,
+      "step": 340390
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.5981862707583987e-05,
+      "loss": 0.8694,
+      "step": 340400
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.5980714777684236e-05,
+      "loss": 0.9643,
+      "step": 340410
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.597956684778449e-05,
+      "loss": 1.0141,
+      "step": 340420
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.597841891788474e-05,
+      "loss": 0.9622,
+      "step": 340430
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.5977270987985e-05,
+      "loss": 0.879,
+      "step": 340440
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.597612305808525e-05,
+      "loss": 0.9152,
+      "step": 340450
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.5974975128185506e-05,
+      "loss": 0.8237,
+      "step": 340460
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.5973827198285756e-05,
+      "loss": 1.1198,
+      "step": 340470
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.597267926838601e-05,
+      "loss": 0.9722,
+      "step": 340480
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.597153133848626e-05,
+      "loss": 0.9156,
+      "step": 340490
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.597038340858651e-05,
+      "loss": 0.9674,
+      "step": 340500
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.596923547868677e-05,
+      "loss": 0.9485,
+      "step": 340510
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.596808754878702e-05,
+      "loss": 0.958,
+      "step": 340520
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.596693961888727e-05,
+      "loss": 0.9503,
+      "step": 340530
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 3.5965791688987525e-05,
+      "loss": 0.9521,
+      "step": 340540
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.5964643759087775e-05,
+      "loss": 0.9088,
+      "step": 340550
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.5963495829188025e-05,
+      "loss": 0.8372,
+      "step": 340560
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.596234789928828e-05,
+      "loss": 0.8733,
+      "step": 340570
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.596119996938853e-05,
+      "loss": 0.8443,
+      "step": 340580
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.596005203948879e-05,
+      "loss": 0.9425,
+      "step": 340590
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.595890410958904e-05,
+      "loss": 0.9093,
+      "step": 340600
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.5957756179689294e-05,
+      "loss": 1.0135,
+      "step": 340610
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.5956608249789544e-05,
+      "loss": 0.927,
+      "step": 340620
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.5955460319889794e-05,
+      "loss": 0.865,
+      "step": 340630
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.595431238999005e-05,
+      "loss": 0.9228,
+      "step": 340640
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.59531644600903e-05,
+      "loss": 0.9015,
+      "step": 340650
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.595201653019055e-05,
+      "loss": 0.8936,
+      "step": 340660
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.595086860029081e-05,
+      "loss": 0.9839,
+      "step": 340670
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.594972067039106e-05,
+      "loss": 0.9363,
+      "step": 340680
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.594857274049131e-05,
+      "loss": 0.8615,
+      "step": 340690
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.5947424810591564e-05,
+      "loss": 0.9196,
+      "step": 340700
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.594627688069181e-05,
+      "loss": 0.9186,
+      "step": 340710
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.594512895079207e-05,
+      "loss": 0.9207,
+      "step": 340720
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.594398102089232e-05,
+      "loss": 0.9716,
+      "step": 340730
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.5942833090992576e-05,
+      "loss": 0.9178,
+      "step": 340740
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.5941685161092826e-05,
+      "loss": 0.89,
+      "step": 340750
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 3.594053723119308e-05,
+      "loss": 0.9159,
+      "step": 340760
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.593938930129333e-05,
+      "loss": 0.9517,
+      "step": 340770
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.593824137139358e-05,
+      "loss": 0.9273,
+      "step": 340780
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.593709344149384e-05,
+      "loss": 1.05,
+      "step": 340790
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.593594551159409e-05,
+      "loss": 0.9275,
+      "step": 340800
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.593479758169434e-05,
+      "loss": 0.9439,
+      "step": 340810
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.5933649651794596e-05,
+      "loss": 0.9535,
+      "step": 340820
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.5932501721894845e-05,
+      "loss": 0.8085,
+      "step": 340830
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.5931353791995095e-05,
+      "loss": 0.9871,
+      "step": 340840
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.593020586209535e-05,
+      "loss": 0.9271,
+      "step": 340850
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.59290579321956e-05,
+      "loss": 0.9719,
+      "step": 340860
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.592791000229586e-05,
+      "loss": 0.9236,
+      "step": 340870
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.592676207239611e-05,
+      "loss": 1.0714,
+      "step": 340880
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.5925614142496365e-05,
+      "loss": 0.9221,
+      "step": 340890
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.5924466212596615e-05,
+      "loss": 0.8972,
+      "step": 340900
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.592331828269687e-05,
+      "loss": 0.9221,
+      "step": 340910
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.592217035279712e-05,
+      "loss": 0.9477,
+      "step": 340920
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.592102242289737e-05,
+      "loss": 0.9955,
+      "step": 340930
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.591987449299763e-05,
+      "loss": 0.8478,
+      "step": 340940
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.591872656309788e-05,
+      "loss": 0.9145,
+      "step": 340950
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.591757863319813e-05,
+      "loss": 0.9411,
+      "step": 340960
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.5916430703298384e-05,
+      "loss": 0.9472,
+      "step": 340970
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 3.5915282773398634e-05,
+      "loss": 0.9505,
+      "step": 340980
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.5914134843498884e-05,
+      "loss": 0.982,
+      "step": 340990
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.591298691359914e-05,
+      "loss": 0.9089,
+      "step": 341000
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.591183898369939e-05,
+      "loss": 0.9569,
+      "step": 341010
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.591069105379965e-05,
+      "loss": 0.9509,
+      "step": 341020
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.59095431238999e-05,
+      "loss": 0.8902,
+      "step": 341030
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.5908395194000153e-05,
+      "loss": 0.9805,
+      "step": 341040
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.59072472641004e-05,
+      "loss": 0.9569,
+      "step": 341050
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.590609933420066e-05,
+      "loss": 0.9531,
+      "step": 341060
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.590495140430091e-05,
+      "loss": 0.9495,
+      "step": 341070
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.590380347440116e-05,
+      "loss": 0.9598,
+      "step": 341080
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.5902655544501416e-05,
+      "loss": 0.9607,
+      "step": 341090
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.5901507614601666e-05,
+      "loss": 0.9509,
+      "step": 341100
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.5900359684701916e-05,
+      "loss": 0.8937,
+      "step": 341110
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.589921175480217e-05,
+      "loss": 0.933,
+      "step": 341120
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.589806382490242e-05,
+      "loss": 0.9866,
+      "step": 341130
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.589691589500267e-05,
+      "loss": 0.8515,
+      "step": 341140
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.589576796510293e-05,
+      "loss": 0.8549,
+      "step": 341150
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.589462003520318e-05,
+      "loss": 0.9514,
+      "step": 341160
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.5893472105303435e-05,
+      "loss": 1.0017,
+      "step": 341170
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.5892324175403685e-05,
+      "loss": 0.931,
+      "step": 341180
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.589117624550394e-05,
+      "loss": 0.905,
+      "step": 341190
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 3.589002831560419e-05,
+      "loss": 0.8413,
+      "step": 341200
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.588888038570445e-05,
+      "loss": 0.9105,
+      "step": 341210
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.58877324558047e-05,
+      "loss": 0.9368,
+      "step": 341220
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.588658452590495e-05,
+      "loss": 0.9691,
+      "step": 341230
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.5885436596005205e-05,
+      "loss": 1.0439,
+      "step": 341240
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.5884288666105455e-05,
+      "loss": 0.9495,
+      "step": 341250
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.5883140736205704e-05,
+      "loss": 0.9537,
+      "step": 341260
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.5881992806305954e-05,
+      "loss": 0.9091,
+      "step": 341270
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.588084487640621e-05,
+      "loss": 0.9206,
+      "step": 341280
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.587969694650646e-05,
+      "loss": 0.8971,
+      "step": 341290
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.587854901660672e-05,
+      "loss": 0.8558,
+      "step": 341300
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.587740108670697e-05,
+      "loss": 0.8606,
+      "step": 341310
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.5876253156807224e-05,
+      "loss": 1.0078,
+      "step": 341320
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.5875105226907474e-05,
+      "loss": 1.0397,
+      "step": 341330
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.587395729700773e-05,
+      "loss": 0.825,
+      "step": 341340
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.587280936710798e-05,
+      "loss": 0.8881,
+      "step": 341350
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.587166143720823e-05,
+      "loss": 0.8557,
+      "step": 341360
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.587051350730849e-05,
+      "loss": 0.9093,
+      "step": 341370
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.5869365577408737e-05,
+      "loss": 0.9735,
+      "step": 341380
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.5868217647508986e-05,
+      "loss": 0.8773,
+      "step": 341390
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.586706971760924e-05,
+      "loss": 0.8762,
+      "step": 341400
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.586592178770949e-05,
+      "loss": 0.8285,
+      "step": 341410
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 3.586477385780974e-05,
+      "loss": 0.9245,
+      "step": 341420
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.586362592791e-05,
+      "loss": 0.9886,
+      "step": 341430
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.586247799801025e-05,
+      "loss": 0.8547,
+      "step": 341440
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.5861330068110506e-05,
+      "loss": 0.9184,
+      "step": 341450
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.5860182138210756e-05,
+      "loss": 0.8867,
+      "step": 341460
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.585903420831101e-05,
+      "loss": 0.8909,
+      "step": 341470
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.585788627841126e-05,
+      "loss": 1.056,
+      "step": 341480
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.585673834851152e-05,
+      "loss": 0.9246,
+      "step": 341490
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.585559041861177e-05,
+      "loss": 0.9451,
+      "step": 341500
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.585444248871202e-05,
+      "loss": 1.0122,
+      "step": 341510
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.5853294558812275e-05,
+      "loss": 0.9675,
+      "step": 341520
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.5852146628912525e-05,
+      "loss": 0.9329,
+      "step": 341530
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.5850998699012775e-05,
+      "loss": 0.9213,
+      "step": 341540
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.584985076911303e-05,
+      "loss": 0.8873,
+      "step": 341550
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.584870283921328e-05,
+      "loss": 0.9262,
+      "step": 341560
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.584755490931353e-05,
+      "loss": 0.9276,
+      "step": 341570
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.584640697941379e-05,
+      "loss": 0.9052,
+      "step": 341580
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.584525904951404e-05,
+      "loss": 1.0078,
+      "step": 341590
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.5844111119614294e-05,
+      "loss": 0.911,
+      "step": 341600
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.5842963189714544e-05,
+      "loss": 0.9926,
+      "step": 341610
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.58418152598148e-05,
+      "loss": 0.9141,
+      "step": 341620
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 3.584066732991505e-05,
+      "loss": 0.9519,
+      "step": 341630
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.583951940001531e-05,
+      "loss": 0.9607,
+      "step": 341640
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.583837147011556e-05,
+      "loss": 0.8835,
+      "step": 341650
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.583722354021581e-05,
+      "loss": 0.8818,
+      "step": 341660
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.5836075610316064e-05,
+      "loss": 0.8895,
+      "step": 341670
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.5834927680416313e-05,
+      "loss": 0.9545,
+      "step": 341680
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.583377975051656e-05,
+      "loss": 0.963,
+      "step": 341690
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.583263182061682e-05,
+      "loss": 0.9185,
+      "step": 341700
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.583148389071707e-05,
+      "loss": 1.0753,
+      "step": 341710
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.583033596081732e-05,
+      "loss": 0.9143,
+      "step": 341720
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.5829188030917576e-05,
+      "loss": 1.0461,
+      "step": 341730
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.5828040101017826e-05,
+      "loss": 0.9512,
+      "step": 341740
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.582689217111808e-05,
+      "loss": 0.959,
+      "step": 341750
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.582574424121833e-05,
+      "loss": 0.9239,
+      "step": 341760
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.582459631131859e-05,
+      "loss": 0.8858,
+      "step": 341770
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.582344838141884e-05,
+      "loss": 0.8641,
+      "step": 341780
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.5822300451519096e-05,
+      "loss": 0.9968,
+      "step": 341790
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.5821152521619346e-05,
+      "loss": 0.9094,
+      "step": 341800
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.5820004591719595e-05,
+      "loss": 1.0018,
+      "step": 341810
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.581885666181985e-05,
+      "loss": 0.9615,
+      "step": 341820
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.58177087319201e-05,
+      "loss": 0.9303,
+      "step": 341830
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.581656080202035e-05,
+      "loss": 0.8895,
+      "step": 341840
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 3.581541287212061e-05,
+      "loss": 1.0221,
+      "step": 341850
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.581426494222086e-05,
+      "loss": 0.8747,
+      "step": 341860
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.581311701232111e-05,
+      "loss": 0.881,
+      "step": 341870
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.5811969082421365e-05,
+      "loss": 0.9212,
+      "step": 341880
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.5810821152521615e-05,
+      "loss": 0.9689,
+      "step": 341890
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.580967322262187e-05,
+      "loss": 0.9959,
+      "step": 341900
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.580852529272212e-05,
+      "loss": 0.8481,
+      "step": 341910
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.580737736282238e-05,
+      "loss": 0.9591,
+      "step": 341920
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.580622943292263e-05,
+      "loss": 0.9407,
+      "step": 341930
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.580508150302288e-05,
+      "loss": 0.9518,
+      "step": 341940
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.5803933573123134e-05,
+      "loss": 1.022,
+      "step": 341950
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.5802785643223384e-05,
+      "loss": 0.8279,
+      "step": 341960
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.5801637713323634e-05,
+      "loss": 0.8727,
+      "step": 341970
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.580048978342389e-05,
+      "loss": 0.934,
+      "step": 341980
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.579934185352414e-05,
+      "loss": 0.958,
+      "step": 341990
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.579819392362439e-05,
+      "loss": 0.9507,
+      "step": 342000
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.579704599372465e-05,
+      "loss": 0.8317,
+      "step": 342010
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.57958980638249e-05,
+      "loss": 0.8571,
+      "step": 342020
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.579475013392515e-05,
+      "loss": 0.9882,
+      "step": 342030
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.57936022040254e-05,
+      "loss": 0.9832,
+      "step": 342040
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.579245427412566e-05,
+      "loss": 0.8537,
+      "step": 342050
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.579130634422591e-05,
+      "loss": 0.9481,
+      "step": 342060
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 3.5790158414326166e-05,
+      "loss": 0.9293,
+      "step": 342070
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.5789010484426416e-05,
+      "loss": 0.8353,
+      "step": 342080
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.5787862554526666e-05,
+      "loss": 0.993,
+      "step": 342090
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.578671462462692e-05,
+      "loss": 0.8617,
+      "step": 342100
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.578556669472717e-05,
+      "loss": 0.9413,
+      "step": 342110
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.578441876482742e-05,
+      "loss": 1.0417,
+      "step": 342120
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.578327083492768e-05,
+      "loss": 0.9838,
+      "step": 342130
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.578212290502793e-05,
+      "loss": 0.9064,
+      "step": 342140
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.578097497512818e-05,
+      "loss": 0.9457,
+      "step": 342150
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.5779827045228435e-05,
+      "loss": 0.8448,
+      "step": 342160
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.5778679115328685e-05,
+      "loss": 0.9517,
+      "step": 342170
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.577753118542894e-05,
+      "loss": 0.9432,
+      "step": 342180
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.577638325552919e-05,
+      "loss": 1.0229,
+      "step": 342190
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.577523532562945e-05,
+      "loss": 0.9241,
+      "step": 342200
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.57740873957297e-05,
+      "loss": 0.9817,
+      "step": 342210
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.5772939465829955e-05,
+      "loss": 0.9032,
+      "step": 342220
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.5771791535930205e-05,
+      "loss": 0.9058,
+      "step": 342230
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.5770643606030454e-05,
+      "loss": 1.0573,
+      "step": 342240
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.576949567613071e-05,
+      "loss": 0.8914,
+      "step": 342250
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.576834774623096e-05,
+      "loss": 0.8838,
+      "step": 342260
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.576719981633121e-05,
+      "loss": 0.959,
+      "step": 342270
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.576605188643147e-05,
+      "loss": 0.9964,
+      "step": 342280
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 3.576490395653172e-05,
+      "loss": 0.942,
+      "step": 342290
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.576375602663197e-05,
+      "loss": 0.9587,
+      "step": 342300
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.5762608096732224e-05,
+      "loss": 0.8357,
+      "step": 342310
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.5761460166832474e-05,
+      "loss": 0.9652,
+      "step": 342320
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.576031223693273e-05,
+      "loss": 0.9982,
+      "step": 342330
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.575916430703298e-05,
+      "loss": 0.9466,
+      "step": 342340
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.575801637713324e-05,
+      "loss": 0.9441,
+      "step": 342350
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.5756868447233487e-05,
+      "loss": 0.872,
+      "step": 342360
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.575572051733374e-05,
+      "loss": 0.9766,
+      "step": 342370
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.575457258743399e-05,
+      "loss": 1.0036,
+      "step": 342380
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.575342465753424e-05,
+      "loss": 0.9463,
+      "step": 342390
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.57522767276345e-05,
+      "loss": 0.8992,
+      "step": 342400
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.575112879773475e-05,
+      "loss": 0.9777,
+      "step": 342410
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.5749980867835e-05,
+      "loss": 0.9137,
+      "step": 342420
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.5748832937935256e-05,
+      "loss": 0.9407,
+      "step": 342430
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.5747685008035506e-05,
+      "loss": 0.989,
+      "step": 342440
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.5746537078135756e-05,
+      "loss": 1.002,
+      "step": 342450
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.574538914823601e-05,
+      "loss": 0.9475,
+      "step": 342460
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.574424121833626e-05,
+      "loss": 0.9797,
+      "step": 342470
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.574309328843652e-05,
+      "loss": 0.9127,
+      "step": 342480
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.574194535853677e-05,
+      "loss": 0.9809,
+      "step": 342490
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.5740797428637025e-05,
+      "loss": 0.9908,
+      "step": 342500
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 3.5739649498737275e-05,
+      "loss": 0.919,
+      "step": 342510
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.573850156883753e-05,
+      "loss": 1.0505,
+      "step": 342520
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.573735363893778e-05,
+      "loss": 0.838,
+      "step": 342530
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.573620570903803e-05,
+      "loss": 0.9363,
+      "step": 342540
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.573505777913829e-05,
+      "loss": 0.9032,
+      "step": 342550
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.573390984923854e-05,
+      "loss": 0.9262,
+      "step": 342560
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.573276191933879e-05,
+      "loss": 0.8864,
+      "step": 342570
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.5731613989439044e-05,
+      "loss": 0.9218,
+      "step": 342580
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.5730466059539294e-05,
+      "loss": 0.9129,
+      "step": 342590
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.5729318129639544e-05,
+      "loss": 0.9188,
+      "step": 342600
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.57281701997398e-05,
+      "loss": 0.9203,
+      "step": 342610
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.572702226984005e-05,
+      "loss": 0.9292,
+      "step": 342620
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.572587433994031e-05,
+      "loss": 0.9466,
+      "step": 342630
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.572472641004056e-05,
+      "loss": 0.9727,
+      "step": 342640
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.5723578480140814e-05,
+      "loss": 0.9003,
+      "step": 342650
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.5722430550241063e-05,
+      "loss": 1.0334,
+      "step": 342660
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.572128262034131e-05,
+      "loss": 0.9659,
+      "step": 342670
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.572013469044157e-05,
+      "loss": 0.9477,
+      "step": 342680
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.571898676054182e-05,
+      "loss": 0.9608,
+      "step": 342690
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.571783883064207e-05,
+      "loss": 0.8926,
+      "step": 342700
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.5716690900742326e-05,
+      "loss": 0.9817,
+      "step": 342710
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 3.5715542970842576e-05,
+      "loss": 0.9509,
+      "step": 342720
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.5714395040942826e-05,
+      "loss": 0.9426,
+      "step": 342730
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.571324711104308e-05,
+      "loss": 0.8686,
+      "step": 342740
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.571209918114333e-05,
+      "loss": 0.9741,
+      "step": 342750
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.571095125124359e-05,
+      "loss": 0.9555,
+      "step": 342760
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.570980332134384e-05,
+      "loss": 0.8713,
+      "step": 342770
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.5708655391444096e-05,
+      "loss": 0.8517,
+      "step": 342780
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.5707507461544345e-05,
+      "loss": 0.7994,
+      "step": 342790
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.57063595316446e-05,
+      "loss": 0.9323,
+      "step": 342800
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.570521160174485e-05,
+      "loss": 0.9543,
+      "step": 342810
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.57040636718451e-05,
+      "loss": 0.9363,
+      "step": 342820
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.570291574194536e-05,
+      "loss": 0.8388,
+      "step": 342830
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.570176781204561e-05,
+      "loss": 0.8985,
+      "step": 342840
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.570061988214586e-05,
+      "loss": 0.9387,
+      "step": 342850
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.5699471952246115e-05,
+      "loss": 0.9416,
+      "step": 342860
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.5698324022346365e-05,
+      "loss": 1.1201,
+      "step": 342870
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.5697176092446614e-05,
+      "loss": 0.986,
+      "step": 342880
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.569602816254687e-05,
+      "loss": 0.9776,
+      "step": 342890
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.569488023264712e-05,
+      "loss": 0.9201,
+      "step": 342900
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.569373230274738e-05,
+      "loss": 0.9031,
+      "step": 342910
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.569258437284763e-05,
+      "loss": 0.891,
+      "step": 342920
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.5691436442947884e-05,
+      "loss": 0.9319,
+      "step": 342930
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 3.5690288513048134e-05,
+      "loss": 0.9046,
+      "step": 342940
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.568914058314839e-05,
+      "loss": 0.953,
+      "step": 342950
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.568799265324864e-05,
+      "loss": 0.8546,
+      "step": 342960
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.568684472334889e-05,
+      "loss": 0.9285,
+      "step": 342970
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.568569679344915e-05,
+      "loss": 0.927,
+      "step": 342980
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.56845488635494e-05,
+      "loss": 0.8739,
+      "step": 342990
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.568340093364965e-05,
+      "loss": 0.834,
+      "step": 343000
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.56822530037499e-05,
+      "loss": 0.891,
+      "step": 343010
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.568110507385015e-05,
+      "loss": 0.8895,
+      "step": 343020
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.56799571439504e-05,
+      "loss": 0.9314,
+      "step": 343030
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.567880921405066e-05,
+      "loss": 0.9582,
+      "step": 343040
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.567766128415091e-05,
+      "loss": 0.9899,
+      "step": 343050
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.5676513354251166e-05,
+      "loss": 0.8227,
+      "step": 343060
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.5675365424351416e-05,
+      "loss": 0.9292,
+      "step": 343070
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.567421749445167e-05,
+      "loss": 0.924,
+      "step": 343080
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.567306956455192e-05,
+      "loss": 0.9926,
+      "step": 343090
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.567192163465218e-05,
+      "loss": 0.9863,
+      "step": 343100
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.567077370475243e-05,
+      "loss": 0.9107,
+      "step": 343110
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.566962577485268e-05,
+      "loss": 0.9289,
+      "step": 343120
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.5668477844952935e-05,
+      "loss": 0.9396,
+      "step": 343130
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.5667329915053185e-05,
+      "loss": 0.8381,
+      "step": 343140
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.5666181985153435e-05,
+      "loss": 1.0411,
+      "step": 343150
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 3.566503405525369e-05,
+      "loss": 1.0156,
+      "step": 343160
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.566388612535394e-05,
+      "loss": 1.0386,
+      "step": 343170
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.566273819545419e-05,
+      "loss": 0.8951,
+      "step": 343180
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.566159026555445e-05,
+      "loss": 0.8312,
+      "step": 343190
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.56604423356547e-05,
+      "loss": 0.9937,
+      "step": 343200
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.5659294405754955e-05,
+      "loss": 0.8052,
+      "step": 343210
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.5658146475855204e-05,
+      "loss": 0.8669,
+      "step": 343220
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.565699854595546e-05,
+      "loss": 0.893,
+      "step": 343230
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.565585061605571e-05,
+      "loss": 0.9477,
+      "step": 343240
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.565470268615597e-05,
+      "loss": 0.8948,
+      "step": 343250
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.565355475625622e-05,
+      "loss": 0.8728,
+      "step": 343260
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.565240682635647e-05,
+      "loss": 0.9561,
+      "step": 343270
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.5651258896456724e-05,
+      "loss": 0.935,
+      "step": 343280
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.5650110966556974e-05,
+      "loss": 1.0682,
+      "step": 343290
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.5648963036657224e-05,
+      "loss": 0.9515,
+      "step": 343300
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.5647815106757473e-05,
+      "loss": 0.9466,
+      "step": 343310
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.564666717685773e-05,
+      "loss": 0.9732,
+      "step": 343320
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.564551924695798e-05,
+      "loss": 0.8698,
+      "step": 343330
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.5644371317058237e-05,
+      "loss": 0.8536,
+      "step": 343340
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.5643223387158486e-05,
+      "loss": 0.9327,
+      "step": 343350
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.564207545725874e-05,
+      "loss": 1.0125,
+      "step": 343360
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.564092752735899e-05,
+      "loss": 1.0101,
+      "step": 343370
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 3.563977959745925e-05,
+      "loss": 0.8626,
+      "step": 343380
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.56386316675595e-05,
+      "loss": 0.8951,
+      "step": 343390
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.563748373765975e-05,
+      "loss": 0.9219,
+      "step": 343400
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.5636335807760006e-05,
+      "loss": 0.9847,
+      "step": 343410
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.5635187877860256e-05,
+      "loss": 0.937,
+      "step": 343420
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.5634039947960506e-05,
+      "loss": 0.996,
+      "step": 343430
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.563289201806076e-05,
+      "loss": 0.9528,
+      "step": 343440
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.563174408816101e-05,
+      "loss": 0.9276,
+      "step": 343450
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.563059615826126e-05,
+      "loss": 0.8979,
+      "step": 343460
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.562944822836152e-05,
+      "loss": 0.8911,
+      "step": 343470
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.562830029846177e-05,
+      "loss": 0.9219,
+      "step": 343480
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.5627152368562025e-05,
+      "loss": 0.8661,
+      "step": 343490
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.5626004438662275e-05,
+      "loss": 0.9935,
+      "step": 343500
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.562485650876253e-05,
+      "loss": 0.9126,
+      "step": 343510
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.562370857886278e-05,
+      "loss": 0.887,
+      "step": 343520
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.562256064896304e-05,
+      "loss": 0.9933,
+      "step": 343530
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.562141271906329e-05,
+      "loss": 1.0223,
+      "step": 343540
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.562026478916354e-05,
+      "loss": 0.9263,
+      "step": 343550
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.5619116859263794e-05,
+      "loss": 0.9208,
+      "step": 343560
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.5617968929364044e-05,
+      "loss": 0.9789,
+      "step": 343570
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.5616820999464294e-05,
+      "loss": 0.8018,
+      "step": 343580
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.561567306956455e-05,
+      "loss": 0.9304,
+      "step": 343590
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 3.56145251396648e-05,
+      "loss": 0.8811,
+      "step": 343600
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.561337720976505e-05,
+      "loss": 0.9532,
+      "step": 343610
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.561222927986531e-05,
+      "loss": 0.9563,
+      "step": 343620
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.561108134996556e-05,
+      "loss": 0.937,
+      "step": 343630
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.5609933420065813e-05,
+      "loss": 0.9625,
+      "step": 343640
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.560878549016606e-05,
+      "loss": 0.9741,
+      "step": 343650
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.560763756026632e-05,
+      "loss": 0.89,
+      "step": 343660
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.560648963036657e-05,
+      "loss": 0.8655,
+      "step": 343670
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.5605341700466826e-05,
+      "loss": 0.949,
+      "step": 343680
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.5604193770567076e-05,
+      "loss": 0.9513,
+      "step": 343690
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.5603045840667326e-05,
+      "loss": 0.942,
+      "step": 343700
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.560189791076758e-05,
+      "loss": 0.9507,
+      "step": 343710
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.560074998086783e-05,
+      "loss": 0.8677,
+      "step": 343720
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.559960205096808e-05,
+      "loss": 0.8598,
+      "step": 343730
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.559845412106834e-05,
+      "loss": 0.9832,
+      "step": 343740
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.559730619116859e-05,
+      "loss": 0.8857,
+      "step": 343750
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.559615826126884e-05,
+      "loss": 0.985,
+      "step": 343760
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.5595010331369095e-05,
+      "loss": 0.9987,
+      "step": 343770
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.5593862401469345e-05,
+      "loss": 0.924,
+      "step": 343780
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.55927144715696e-05,
+      "loss": 0.986,
+      "step": 343790
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.559156654166985e-05,
+      "loss": 0.9433,
+      "step": 343800
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 3.559041861177011e-05,
+      "loss": 1.0057,
+      "step": 343810
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.558927068187036e-05,
+      "loss": 0.8906,
+      "step": 343820
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.5588122751970615e-05,
+      "loss": 0.9355,
+      "step": 343830
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.5586974822070865e-05,
+      "loss": 0.9347,
+      "step": 343840
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.5585826892171115e-05,
+      "loss": 0.857,
+      "step": 343850
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.558467896227137e-05,
+      "loss": 0.9695,
+      "step": 343860
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.558353103237162e-05,
+      "loss": 0.9462,
+      "step": 343870
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.558238310247187e-05,
+      "loss": 0.9246,
+      "step": 343880
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.558123517257213e-05,
+      "loss": 0.9259,
+      "step": 343890
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.558008724267238e-05,
+      "loss": 0.9713,
+      "step": 343900
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.557893931277263e-05,
+      "loss": 0.9274,
+      "step": 343910
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.5577791382872884e-05,
+      "loss": 0.8497,
+      "step": 343920
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.5576643452973134e-05,
+      "loss": 0.9612,
+      "step": 343930
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.557549552307339e-05,
+      "loss": 0.9214,
+      "step": 343940
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.557434759317364e-05,
+      "loss": 0.9117,
+      "step": 343950
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.55731996632739e-05,
+      "loss": 0.9435,
+      "step": 343960
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.557205173337415e-05,
+      "loss": 0.9667,
+      "step": 343970
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.5570903803474403e-05,
+      "loss": 0.9313,
+      "step": 343980
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.556975587357465e-05,
+      "loss": 0.9127,
+      "step": 343990
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.55686079436749e-05,
+      "loss": 0.9576,
+      "step": 344000
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.556746001377515e-05,
+      "loss": 0.9137,
+      "step": 344010
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.556631208387541e-05,
+      "loss": 0.9303,
+      "step": 344020
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 3.556516415397566e-05,
+      "loss": 0.8617,
+      "step": 344030
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.556401622407591e-05,
+      "loss": 0.9724,
+      "step": 344040
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.5562868294176166e-05,
+      "loss": 0.8886,
+      "step": 344050
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.5561720364276416e-05,
+      "loss": 0.8851,
+      "step": 344060
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.556057243437667e-05,
+      "loss": 0.9071,
+      "step": 344070
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.555942450447692e-05,
+      "loss": 0.9224,
+      "step": 344080
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.555827657457718e-05,
+      "loss": 0.8909,
+      "step": 344090
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.555712864467743e-05,
+      "loss": 0.9871,
+      "step": 344100
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.5555980714777685e-05,
+      "loss": 0.94,
+      "step": 344110
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.5554832784877935e-05,
+      "loss": 0.9632,
+      "step": 344120
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.5553684854978185e-05,
+      "loss": 0.8711,
+      "step": 344130
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.555253692507844e-05,
+      "loss": 0.8673,
+      "step": 344140
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.555138899517869e-05,
+      "loss": 0.9527,
+      "step": 344150
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.555024106527894e-05,
+      "loss": 0.9526,
+      "step": 344160
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.55490931353792e-05,
+      "loss": 0.979,
+      "step": 344170
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.554794520547945e-05,
+      "loss": 0.9279,
+      "step": 344180
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.55467972755797e-05,
+      "loss": 0.9543,
+      "step": 344190
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.5545649345679954e-05,
+      "loss": 1.0391,
+      "step": 344200
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.5544501415780204e-05,
+      "loss": 0.9737,
+      "step": 344210
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.554335348588046e-05,
+      "loss": 0.8714,
+      "step": 344220
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.554220555598071e-05,
+      "loss": 0.9545,
+      "step": 344230
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.554105762608097e-05,
+      "loss": 0.8955,
+      "step": 344240
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 3.553990969618122e-05,
+      "loss": 0.9384,
+      "step": 344250
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.5538761766281474e-05,
+      "loss": 0.8587,
+      "step": 344260
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.5537613836381724e-05,
+      "loss": 0.8691,
+      "step": 344270
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.5536465906481974e-05,
+      "loss": 0.9637,
+      "step": 344280
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.553531797658223e-05,
+      "loss": 1.0028,
+      "step": 344290
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.553417004668248e-05,
+      "loss": 0.969,
+      "step": 344300
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.553302211678273e-05,
+      "loss": 0.8934,
+      "step": 344310
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.5531874186882987e-05,
+      "loss": 0.9095,
+      "step": 344320
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.5530726256983236e-05,
+      "loss": 0.8778,
+      "step": 344330
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.5529578327083486e-05,
+      "loss": 1.0246,
+      "step": 344340
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.552843039718374e-05,
+      "loss": 1.0025,
+      "step": 344350
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.552728246728399e-05,
+      "loss": 0.9122,
+      "step": 344360
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.552613453738425e-05,
+      "loss": 0.8855,
+      "step": 344370
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.55249866074845e-05,
+      "loss": 0.9503,
+      "step": 344380
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.5523838677584756e-05,
+      "loss": 0.9095,
+      "step": 344390
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.5522690747685006e-05,
+      "loss": 0.8751,
+      "step": 344400
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.552154281778526e-05,
+      "loss": 1.0414,
+      "step": 344410
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.552039488788551e-05,
+      "loss": 0.8598,
+      "step": 344420
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.551924695798576e-05,
+      "loss": 0.8466,
+      "step": 344430
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.551809902808602e-05,
+      "loss": 0.9841,
+      "step": 344440
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.551695109818627e-05,
+      "loss": 0.9143,
+      "step": 344450
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.551580316828652e-05,
+      "loss": 0.908,
+      "step": 344460
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 3.5514655238386775e-05,
+      "loss": 0.9509,
+      "step": 344470
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.5513507308487025e-05,
+      "loss": 0.8722,
+      "step": 344480
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.5512359378587275e-05,
+      "loss": 0.8386,
+      "step": 344490
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.551121144868753e-05,
+      "loss": 0.9903,
+      "step": 344500
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.551006351878778e-05,
+      "loss": 0.9622,
+      "step": 344510
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.550891558888804e-05,
+      "loss": 0.9233,
+      "step": 344520
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.550776765898829e-05,
+      "loss": 0.8944,
+      "step": 344530
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.5506619729088544e-05,
+      "loss": 0.9282,
+      "step": 344540
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.5505471799188794e-05,
+      "loss": 0.871,
+      "step": 344550
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.550432386928905e-05,
+      "loss": 0.9143,
+      "step": 344560
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.55031759393893e-05,
+      "loss": 0.9064,
+      "step": 344570
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.550202800948955e-05,
+      "loss": 0.9257,
+      "step": 344580
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.550088007958981e-05,
+      "loss": 1.0038,
+      "step": 344590
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.549973214969006e-05,
+      "loss": 0.8743,
+      "step": 344600
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.549858421979031e-05,
+      "loss": 0.9418,
+      "step": 344610
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.5497436289890563e-05,
+      "loss": 0.9339,
+      "step": 344620
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.549628835999081e-05,
+      "loss": 0.8946,
+      "step": 344630
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.549514043009106e-05,
+      "loss": 1.1083,
+      "step": 344640
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.549399250019132e-05,
+      "loss": 0.9912,
+      "step": 344650
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.549284457029157e-05,
+      "loss": 0.8593,
+      "step": 344660
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.5491696640391826e-05,
+      "loss": 0.9414,
+      "step": 344670
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.5490548710492076e-05,
+      "loss": 0.9549,
+      "step": 344680
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 3.548940078059233e-05,
+      "loss": 1.0622,
+      "step": 344690
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.548825285069258e-05,
+      "loss": 0.9335,
+      "step": 344700
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.548710492079283e-05,
+      "loss": 0.9595,
+      "step": 344710
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.548595699089309e-05,
+      "loss": 1.0164,
+      "step": 344720
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.548480906099334e-05,
+      "loss": 0.9286,
+      "step": 344730
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.548366113109359e-05,
+      "loss": 0.8837,
+      "step": 344740
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.5482513201193845e-05,
+      "loss": 0.9513,
+      "step": 344750
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.5481365271294095e-05,
+      "loss": 0.9549,
+      "step": 344760
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.5480217341394345e-05,
+      "loss": 0.8946,
+      "step": 344770
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.54790694114946e-05,
+      "loss": 0.9343,
+      "step": 344780
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.547792148159485e-05,
+      "loss": 0.8427,
+      "step": 344790
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.54767735516951e-05,
+      "loss": 0.9716,
+      "step": 344800
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.547562562179536e-05,
+      "loss": 0.9636,
+      "step": 344810
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.5474477691895615e-05,
+      "loss": 0.9763,
+      "step": 344820
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.5473329761995865e-05,
+      "loss": 0.9899,
+      "step": 344830
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.547218183209612e-05,
+      "loss": 0.8748,
+      "step": 344840
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.547103390219637e-05,
+      "loss": 0.853,
+      "step": 344850
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.546988597229662e-05,
+      "loss": 0.9667,
+      "step": 344860
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.546873804239688e-05,
+      "loss": 0.9068,
+      "step": 344870
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.546759011249713e-05,
+      "loss": 0.8652,
+      "step": 344880
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.546644218259738e-05,
+      "loss": 0.9231,
+      "step": 344890
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 3.5465294252697634e-05,
+      "loss": 0.8849,
+      "step": 344900
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.5464146322797884e-05,
+      "loss": 0.9079,
+      "step": 344910
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.5462998392898134e-05,
+      "loss": 0.8134,
+      "step": 344920
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.546185046299839e-05,
+      "loss": 0.9832,
+      "step": 344930
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.546070253309864e-05,
+      "loss": 0.9852,
+      "step": 344940
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.545955460319889e-05,
+      "loss": 0.9758,
+      "step": 344950
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.545840667329915e-05,
+      "loss": 1.0645,
+      "step": 344960
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.54572587433994e-05,
+      "loss": 0.9199,
+      "step": 344970
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.545611081349965e-05,
+      "loss": 0.9906,
+      "step": 344980
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.545496288359991e-05,
+      "loss": 0.9085,
+      "step": 344990
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.545381495370016e-05,
+      "loss": 0.8408,
+      "step": 345000
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.545266702380041e-05,
+      "loss": 1.0046,
+      "step": 345010
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.5451519093900666e-05,
+      "loss": 0.967,
+      "step": 345020
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.5450371164000916e-05,
+      "loss": 0.8768,
+      "step": 345030
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.5449223234101166e-05,
+      "loss": 0.925,
+      "step": 345040
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.544807530420142e-05,
+      "loss": 0.9251,
+      "step": 345050
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.544692737430167e-05,
+      "loss": 0.8954,
+      "step": 345060
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.544577944440192e-05,
+      "loss": 1.0031,
+      "step": 345070
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.544463151450218e-05,
+      "loss": 0.9278,
+      "step": 345080
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.544348358460243e-05,
+      "loss": 0.9986,
+      "step": 345090
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.544233565470268e-05,
+      "loss": 0.8424,
+      "step": 345100
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.5441187724802935e-05,
+      "loss": 0.9753,
+      "step": 345110
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 3.544003979490319e-05,
+      "loss": 0.8986,
+      "step": 345120
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.543889186500344e-05,
+      "loss": 1.0257,
+      "step": 345130
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.54377439351037e-05,
+      "loss": 0.9687,
+      "step": 345140
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.543659600520395e-05,
+      "loss": 0.8829,
+      "step": 345150
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.54354480753042e-05,
+      "loss": 0.9517,
+      "step": 345160
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.5434300145404455e-05,
+      "loss": 0.996,
+      "step": 345170
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.5433152215504704e-05,
+      "loss": 0.9278,
+      "step": 345180
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.5432004285604954e-05,
+      "loss": 0.9727,
+      "step": 345190
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.543085635570521e-05,
+      "loss": 0.88,
+      "step": 345200
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.542970842580546e-05,
+      "loss": 0.9532,
+      "step": 345210
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.542856049590571e-05,
+      "loss": 1.057,
+      "step": 345220
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.542741256600597e-05,
+      "loss": 0.8655,
+      "step": 345230
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.542626463610622e-05,
+      "loss": 0.896,
+      "step": 345240
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.542511670620647e-05,
+      "loss": 0.9551,
+      "step": 345250
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.5423968776306724e-05,
+      "loss": 0.9206,
+      "step": 345260
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.542282084640698e-05,
+      "loss": 0.9856,
+      "step": 345270
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.542167291650723e-05,
+      "loss": 0.8661,
+      "step": 345280
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.542052498660749e-05,
+      "loss": 0.8593,
+      "step": 345290
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.5419377056707737e-05,
+      "loss": 0.9105,
+      "step": 345300
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.5418229126807986e-05,
+      "loss": 0.8719,
+      "step": 345310
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.541708119690824e-05,
+      "loss": 0.8613,
+      "step": 345320
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.541593326700849e-05,
+      "loss": 0.9988,
+      "step": 345330
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 3.541478533710874e-05,
+      "loss": 0.9508,
+      "step": 345340
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.5413637407209e-05,
+      "loss": 0.9067,
+      "step": 345350
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.541248947730925e-05,
+      "loss": 1.036,
+      "step": 345360
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.54113415474095e-05,
+      "loss": 1.0886,
+      "step": 345370
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.541019361750975e-05,
+      "loss": 0.9131,
+      "step": 345380
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.5409045687610006e-05,
+      "loss": 0.9125,
+      "step": 345390
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.5407897757710255e-05,
+      "loss": 0.9769,
+      "step": 345400
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.540674982781051e-05,
+      "loss": 0.9795,
+      "step": 345410
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.540560189791077e-05,
+      "loss": 0.8552,
+      "step": 345420
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.540445396801102e-05,
+      "loss": 0.9512,
+      "step": 345430
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.540330603811127e-05,
+      "loss": 0.9505,
+      "step": 345440
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.5402158108211525e-05,
+      "loss": 0.9029,
+      "step": 345450
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.5401010178311775e-05,
+      "loss": 1.0415,
+      "step": 345460
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.5399862248412025e-05,
+      "loss": 0.8799,
+      "step": 345470
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.539871431851228e-05,
+      "loss": 0.8907,
+      "step": 345480
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.539756638861253e-05,
+      "loss": 1.0162,
+      "step": 345490
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.539641845871278e-05,
+      "loss": 0.9161,
+      "step": 345500
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.539527052881304e-05,
+      "loss": 0.9842,
+      "step": 345510
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.539412259891329e-05,
+      "loss": 0.9947,
+      "step": 345520
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.539297466901354e-05,
+      "loss": 0.8562,
+      "step": 345530
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.5391826739113794e-05,
+      "loss": 0.8915,
+      "step": 345540
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.5390678809214044e-05,
+      "loss": 0.8559,
+      "step": 345550
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 3.53895308793143e-05,
+      "loss": 0.9807,
+      "step": 345560
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.538838294941456e-05,
+      "loss": 1.0033,
+      "step": 345570
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.538723501951481e-05,
+      "loss": 0.897,
+      "step": 345580
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.538608708961506e-05,
+      "loss": 0.9364,
+      "step": 345590
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.5384939159715313e-05,
+      "loss": 0.8667,
+      "step": 345600
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.538379122981556e-05,
+      "loss": 0.9635,
+      "step": 345610
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.538264329991581e-05,
+      "loss": 0.9187,
+      "step": 345620
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.538149537001607e-05,
+      "loss": 0.8768,
+      "step": 345630
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.538034744011632e-05,
+      "loss": 0.9014,
+      "step": 345640
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.537919951021657e-05,
+      "loss": 0.9984,
+      "step": 345650
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.5378051580316826e-05,
+      "loss": 0.9634,
+      "step": 345660
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.5376903650417076e-05,
+      "loss": 0.8821,
+      "step": 345670
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.5375755720517326e-05,
+      "loss": 0.9665,
+      "step": 345680
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.537460779061758e-05,
+      "loss": 0.9161,
+      "step": 345690
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.537345986071783e-05,
+      "loss": 0.9258,
+      "step": 345700
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.537231193081809e-05,
+      "loss": 0.9798,
+      "step": 345710
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.5371164000918346e-05,
+      "loss": 0.8794,
+      "step": 345720
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.5370016071018595e-05,
+      "loss": 0.9314,
+      "step": 345730
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.5368868141118845e-05,
+      "loss": 0.984,
+      "step": 345740
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.53677202112191e-05,
+      "loss": 0.9979,
+      "step": 345750
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.536657228131935e-05,
+      "loss": 0.8957,
+      "step": 345760
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.53654243514196e-05,
+      "loss": 0.9198,
+      "step": 345770
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 3.536427642151986e-05,
+      "loss": 0.9205,
+      "step": 345780
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.536312849162011e-05,
+      "loss": 0.8831,
+      "step": 345790
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.536198056172036e-05,
+      "loss": 0.9176,
+      "step": 345800
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.5360832631820615e-05,
+      "loss": 0.976,
+      "step": 345810
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.5359684701920864e-05,
+      "loss": 0.9523,
+      "step": 345820
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.5358536772021114e-05,
+      "loss": 0.8478,
+      "step": 345830
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.535738884212137e-05,
+      "loss": 0.8732,
+      "step": 345840
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.535624091222162e-05,
+      "loss": 0.931,
+      "step": 345850
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.535509298232188e-05,
+      "loss": 1.0029,
+      "step": 345860
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.5353945052422134e-05,
+      "loss": 0.9919,
+      "step": 345870
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.5352797122522384e-05,
+      "loss": 0.9344,
+      "step": 345880
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.5351649192622634e-05,
+      "loss": 1.0415,
+      "step": 345890
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.535050126272289e-05,
+      "loss": 0.9154,
+      "step": 345900
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.534935333282314e-05,
+      "loss": 0.9713,
+      "step": 345910
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.534820540292339e-05,
+      "loss": 0.9604,
+      "step": 345920
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.534705747302365e-05,
+      "loss": 0.9138,
+      "step": 345930
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.53459095431239e-05,
+      "loss": 0.8868,
+      "step": 345940
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.5344761613224146e-05,
+      "loss": 1.0274,
+      "step": 345950
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.53436136833244e-05,
+      "loss": 0.8877,
+      "step": 345960
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.534246575342465e-05,
+      "loss": 0.8818,
+      "step": 345970
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.53413178235249e-05,
+      "loss": 0.8519,
+      "step": 345980
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 3.534016989362516e-05,
+      "loss": 1.0409,
+      "step": 345990
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.533902196372541e-05,
+      "loss": 0.8954,
+      "step": 346000
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.5337874033825666e-05,
+      "loss": 1.0084,
+      "step": 346010
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.533672610392592e-05,
+      "loss": 1.0527,
+      "step": 346020
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.533557817402617e-05,
+      "loss": 0.9669,
+      "step": 346030
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.533443024412642e-05,
+      "loss": 0.9645,
+      "step": 346040
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.533328231422667e-05,
+      "loss": 0.8987,
+      "step": 346050
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.533213438432693e-05,
+      "loss": 1.0006,
+      "step": 346060
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.533098645442718e-05,
+      "loss": 0.9037,
+      "step": 346070
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.532983852452743e-05,
+      "loss": 0.9874,
+      "step": 346080
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.5328690594627685e-05,
+      "loss": 0.8502,
+      "step": 346090
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.5327542664727935e-05,
+      "loss": 1.0571,
+      "step": 346100
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.5326394734828185e-05,
+      "loss": 0.9293,
+      "step": 346110
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.532524680492844e-05,
+      "loss": 0.8721,
+      "step": 346120
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.532409887502869e-05,
+      "loss": 0.9162,
+      "step": 346130
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.532295094512895e-05,
+      "loss": 0.8562,
+      "step": 346140
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.53218030152292e-05,
+      "loss": 0.9272,
+      "step": 346150
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.5320655085329454e-05,
+      "loss": 0.9847,
+      "step": 346160
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.5319507155429704e-05,
+      "loss": 0.9378,
+      "step": 346170
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.531835922552996e-05,
+      "loss": 0.9464,
+      "step": 346180
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.531721129563021e-05,
+      "loss": 0.9521,
+      "step": 346190
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.531606336573046e-05,
+      "loss": 0.8742,
+      "step": 346200
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 3.531491543583072e-05,
+      "loss": 0.9095,
+      "step": 346210
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.531376750593097e-05,
+      "loss": 0.9478,
+      "step": 346220
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.531261957603122e-05,
+      "loss": 0.8831,
+      "step": 346230
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.5311471646131474e-05,
+      "loss": 1.0144,
+      "step": 346240
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.5310323716231723e-05,
+      "loss": 0.8602,
+      "step": 346250
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.530917578633197e-05,
+      "loss": 1.0118,
+      "step": 346260
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.530802785643223e-05,
+      "loss": 0.9967,
+      "step": 346270
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.530687992653248e-05,
+      "loss": 0.9444,
+      "step": 346280
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.5305731996632736e-05,
+      "loss": 0.9932,
+      "step": 346290
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.5304584066732986e-05,
+      "loss": 1.0138,
+      "step": 346300
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.530343613683324e-05,
+      "loss": 0.9434,
+      "step": 346310
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.530228820693349e-05,
+      "loss": 0.8822,
+      "step": 346320
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.530114027703375e-05,
+      "loss": 0.889,
+      "step": 346330
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.5299992347134e-05,
+      "loss": 0.9473,
+      "step": 346340
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.529884441723425e-05,
+      "loss": 0.8198,
+      "step": 346350
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.5297696487334506e-05,
+      "loss": 0.9711,
+      "step": 346360
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.5296548557434756e-05,
+      "loss": 0.9239,
+      "step": 346370
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.5295400627535005e-05,
+      "loss": 0.9125,
+      "step": 346380
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.529425269763526e-05,
+      "loss": 0.9028,
+      "step": 346390
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.529310476773551e-05,
+      "loss": 0.89,
+      "step": 346400
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.529195683783576e-05,
+      "loss": 0.9295,
+      "step": 346410
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.529080890793602e-05,
+      "loss": 0.773,
+      "step": 346420
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 3.528966097803627e-05,
+      "loss": 0.8751,
+      "step": 346430
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.5288513048136525e-05,
+      "loss": 1.0528,
+      "step": 346440
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.5287365118236775e-05,
+      "loss": 0.9879,
+      "step": 346450
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.528621718833703e-05,
+      "loss": 0.8925,
+      "step": 346460
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.528506925843728e-05,
+      "loss": 0.8792,
+      "step": 346470
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.528392132853754e-05,
+      "loss": 0.8257,
+      "step": 346480
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.528277339863779e-05,
+      "loss": 0.9946,
+      "step": 346490
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.528162546873804e-05,
+      "loss": 0.9787,
+      "step": 346500
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.5280477538838294e-05,
+      "loss": 0.9342,
+      "step": 346510
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.5279329608938544e-05,
+      "loss": 0.987,
+      "step": 346520
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.5278181679038794e-05,
+      "loss": 0.9077,
+      "step": 346530
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.527703374913905e-05,
+      "loss": 0.9824,
+      "step": 346540
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.52758858192393e-05,
+      "loss": 0.9453,
+      "step": 346550
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.527473788933955e-05,
+      "loss": 0.9794,
+      "step": 346560
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.527358995943981e-05,
+      "loss": 0.9141,
+      "step": 346570
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.527244202954006e-05,
+      "loss": 0.9554,
+      "step": 346580
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.527129409964031e-05,
+      "loss": 0.974,
+      "step": 346590
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.527014616974056e-05,
+      "loss": 1.0003,
+      "step": 346600
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.526899823984082e-05,
+      "loss": 0.9578,
+      "step": 346610
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.526785030994107e-05,
+      "loss": 0.9146,
+      "step": 346620
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.5266702380041326e-05,
+      "loss": 0.977,
+      "step": 346630
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.5265554450141576e-05,
+      "loss": 0.8533,
+      "step": 346640
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 3.5264406520241826e-05,
+      "loss": 0.9353,
+      "step": 346650
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.526325859034208e-05,
+      "loss": 0.9728,
+      "step": 346660
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.526211066044233e-05,
+      "loss": 0.9143,
+      "step": 346670
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.526096273054258e-05,
+      "loss": 0.9417,
+      "step": 346680
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.525981480064284e-05,
+      "loss": 1.0981,
+      "step": 346690
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.525866687074309e-05,
+      "loss": 1.0648,
+      "step": 346700
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.525751894084334e-05,
+      "loss": 0.8604,
+      "step": 346710
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.5256371010943595e-05,
+      "loss": 0.9793,
+      "step": 346720
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.5255223081043845e-05,
+      "loss": 0.9043,
+      "step": 346730
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.52540751511441e-05,
+      "loss": 1.0278,
+      "step": 346740
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.525292722124435e-05,
+      "loss": 0.9117,
+      "step": 346750
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.525177929134461e-05,
+      "loss": 0.8752,
+      "step": 346760
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.525063136144486e-05,
+      "loss": 0.8454,
+      "step": 346770
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.524948343154511e-05,
+      "loss": 1.0231,
+      "step": 346780
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.5248335501645365e-05,
+      "loss": 0.95,
+      "step": 346790
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.5247187571745614e-05,
+      "loss": 1.0139,
+      "step": 346800
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.5246039641845864e-05,
+      "loss": 0.9401,
+      "step": 346810
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.524489171194612e-05,
+      "loss": 0.9134,
+      "step": 346820
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.524374378204637e-05,
+      "loss": 0.8818,
+      "step": 346830
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.524259585214662e-05,
+      "loss": 1.0368,
+      "step": 346840
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.524144792224688e-05,
+      "loss": 0.9284,
+      "step": 346850
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 3.524029999234713e-05,
+      "loss": 0.9672,
+      "step": 346860
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.5239152062447384e-05,
+      "loss": 1.0025,
+      "step": 346870
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.5238004132547634e-05,
+      "loss": 0.8922,
+      "step": 346880
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.523685620264789e-05,
+      "loss": 0.901,
+      "step": 346890
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.523570827274814e-05,
+      "loss": 0.9046,
+      "step": 346900
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.52345603428484e-05,
+      "loss": 1.0274,
+      "step": 346910
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.5233412412948647e-05,
+      "loss": 0.9946,
+      "step": 346920
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.5232264483048896e-05,
+      "loss": 0.9404,
+      "step": 346930
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.523111655314915e-05,
+      "loss": 0.9712,
+      "step": 346940
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.52299686232494e-05,
+      "loss": 0.9066,
+      "step": 346950
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.522882069334965e-05,
+      "loss": 0.9216,
+      "step": 346960
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.522767276344991e-05,
+      "loss": 0.9788,
+      "step": 346970
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.522652483355016e-05,
+      "loss": 0.8267,
+      "step": 346980
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.522537690365041e-05,
+      "loss": 0.9354,
+      "step": 346990
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.5224228973750666e-05,
+      "loss": 0.9638,
+      "step": 347000
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.5223081043850916e-05,
+      "loss": 0.9751,
+      "step": 347010
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.522193311395117e-05,
+      "loss": 0.9214,
+      "step": 347020
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.522078518405142e-05,
+      "loss": 0.8855,
+      "step": 347030
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.521963725415168e-05,
+      "loss": 1.008,
+      "step": 347040
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.521848932425193e-05,
+      "loss": 0.9052,
+      "step": 347050
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.5217341394352185e-05,
+      "loss": 0.8698,
+      "step": 347060
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.5216193464452435e-05,
+      "loss": 0.8831,
+      "step": 347070
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 3.5215045534552685e-05,
+      "loss": 0.8396,
+      "step": 347080
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.521389760465294e-05,
+      "loss": 0.9193,
+      "step": 347090
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.521274967475319e-05,
+      "loss": 0.9856,
+      "step": 347100
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.521160174485344e-05,
+      "loss": 0.9153,
+      "step": 347110
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.52104538149537e-05,
+      "loss": 0.9298,
+      "step": 347120
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.520930588505395e-05,
+      "loss": 1.0572,
+      "step": 347130
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.52081579551542e-05,
+      "loss": 0.9548,
+      "step": 347140
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.5207010025254454e-05,
+      "loss": 0.9274,
+      "step": 347150
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.5205862095354704e-05,
+      "loss": 0.8997,
+      "step": 347160
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.520471416545496e-05,
+      "loss": 0.8151,
+      "step": 347170
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.520356623555521e-05,
+      "loss": 0.8927,
+      "step": 347180
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.520241830565547e-05,
+      "loss": 0.9264,
+      "step": 347190
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.520127037575572e-05,
+      "loss": 0.9514,
+      "step": 347200
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.5200122445855974e-05,
+      "loss": 0.9332,
+      "step": 347210
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.5198974515956224e-05,
+      "loss": 0.9034,
+      "step": 347220
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.5197826586056473e-05,
+      "loss": 0.987,
+      "step": 347230
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.519667865615673e-05,
+      "loss": 0.9667,
+      "step": 347240
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.519553072625698e-05,
+      "loss": 0.8896,
+      "step": 347250
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.519438279635723e-05,
+      "loss": 0.9433,
+      "step": 347260
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.5193234866457486e-05,
+      "loss": 0.9228,
+      "step": 347270
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.5192086936557736e-05,
+      "loss": 0.8945,
+      "step": 347280
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.5190939006657986e-05,
+      "loss": 0.9928,
+      "step": 347290
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 3.518979107675824e-05,
+      "loss": 0.9792,
+      "step": 347300
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.518864314685849e-05,
+      "loss": 0.8965,
+      "step": 347310
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.518749521695875e-05,
+      "loss": 0.9283,
+      "step": 347320
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.5186347287059e-05,
+      "loss": 0.9253,
+      "step": 347330
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.5185199357159256e-05,
+      "loss": 0.9232,
+      "step": 347340
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.5184051427259506e-05,
+      "loss": 0.8881,
+      "step": 347350
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.518290349735976e-05,
+      "loss": 0.897,
+      "step": 347360
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.518175556746001e-05,
+      "loss": 0.9458,
+      "step": 347370
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.518060763756026e-05,
+      "loss": 0.8853,
+      "step": 347380
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.517945970766052e-05,
+      "loss": 0.9053,
+      "step": 347390
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.517831177776077e-05,
+      "loss": 0.92,
+      "step": 347400
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.517716384786102e-05,
+      "loss": 0.9087,
+      "step": 347410
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.5176015917961275e-05,
+      "loss": 1.0052,
+      "step": 347420
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.5174867988061525e-05,
+      "loss": 0.8697,
+      "step": 347430
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.5173720058161775e-05,
+      "loss": 0.8467,
+      "step": 347440
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.517257212826203e-05,
+      "loss": 0.9633,
+      "step": 347450
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.517142419836228e-05,
+      "loss": 0.9295,
+      "step": 347460
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.517027626846254e-05,
+      "loss": 0.9104,
+      "step": 347470
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.516912833856279e-05,
+      "loss": 0.9837,
+      "step": 347480
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.5167980408663044e-05,
+      "loss": 0.912,
+      "step": 347490
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.5166832478763294e-05,
+      "loss": 0.8971,
+      "step": 347500
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.5165684548863544e-05,
+      "loss": 0.9669,
+      "step": 347510
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 3.51645366189638e-05,
+      "loss": 0.7787,
+      "step": 347520
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.516338868906405e-05,
+      "loss": 0.9246,
+      "step": 347530
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.51622407591643e-05,
+      "loss": 0.8895,
+      "step": 347540
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.516109282926456e-05,
+      "loss": 0.9027,
+      "step": 347550
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.515994489936481e-05,
+      "loss": 0.9364,
+      "step": 347560
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.5158796969465057e-05,
+      "loss": 0.9051,
+      "step": 347570
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.515764903956531e-05,
+      "loss": 0.8912,
+      "step": 347580
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.515650110966556e-05,
+      "loss": 0.9018,
+      "step": 347590
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.515535317976582e-05,
+      "loss": 0.8897,
+      "step": 347600
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.515420524986607e-05,
+      "loss": 0.9251,
+      "step": 347610
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.5153057319966326e-05,
+      "loss": 0.8692,
+      "step": 347620
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.5151909390066576e-05,
+      "loss": 0.9797,
+      "step": 347630
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.515076146016683e-05,
+      "loss": 1.0208,
+      "step": 347640
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.514961353026708e-05,
+      "loss": 0.7564,
+      "step": 347650
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.514846560036733e-05,
+      "loss": 0.9892,
+      "step": 347660
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.514731767046759e-05,
+      "loss": 0.9401,
+      "step": 347670
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.514616974056784e-05,
+      "loss": 0.8556,
+      "step": 347680
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.514502181066809e-05,
+      "loss": 0.9433,
+      "step": 347690
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.5143873880768345e-05,
+      "loss": 0.9039,
+      "step": 347700
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.5142725950868595e-05,
+      "loss": 0.9049,
+      "step": 347710
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.5141578020968845e-05,
+      "loss": 0.949,
+      "step": 347720
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.51404300910691e-05,
+      "loss": 0.9548,
+      "step": 347730
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 3.513928216116935e-05,
+      "loss": 0.9113,
+      "step": 347740
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.513813423126961e-05,
+      "loss": 0.9251,
+      "step": 347750
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.513698630136986e-05,
+      "loss": 0.98,
+      "step": 347760
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.5135838371470115e-05,
+      "loss": 0.9413,
+      "step": 347770
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.5134690441570364e-05,
+      "loss": 0.9264,
+      "step": 347780
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.513354251167062e-05,
+      "loss": 0.9668,
+      "step": 347790
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.513239458177087e-05,
+      "loss": 0.8865,
+      "step": 347800
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.513124665187112e-05,
+      "loss": 0.8666,
+      "step": 347810
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.513009872197138e-05,
+      "loss": 0.9533,
+      "step": 347820
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.512895079207163e-05,
+      "loss": 0.912,
+      "step": 347830
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.512780286217188e-05,
+      "loss": 0.9773,
+      "step": 347840
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.5126654932272134e-05,
+      "loss": 0.9343,
+      "step": 347850
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.5125507002372384e-05,
+      "loss": 0.9504,
+      "step": 347860
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.5124359072472633e-05,
+      "loss": 1.0473,
+      "step": 347870
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.512321114257289e-05,
+      "loss": 0.8952,
+      "step": 347880
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.512206321267314e-05,
+      "loss": 1.0512,
+      "step": 347890
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.5120915282773397e-05,
+      "loss": 0.9034,
+      "step": 347900
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.5119767352873646e-05,
+      "loss": 0.868,
+      "step": 347910
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.51186194229739e-05,
+      "loss": 1.021,
+      "step": 347920
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.511747149307415e-05,
+      "loss": 0.9253,
+      "step": 347930
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.511632356317441e-05,
+      "loss": 0.9075,
+      "step": 347940
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 3.511517563327466e-05,
+      "loss": 1.0404,
+      "step": 347950
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.511402770337491e-05,
+      "loss": 0.8923,
+      "step": 347960
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.5112879773475166e-05,
+      "loss": 0.9295,
+      "step": 347970
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.5111731843575416e-05,
+      "loss": 0.9682,
+      "step": 347980
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.5110583913675666e-05,
+      "loss": 0.8964,
+      "step": 347990
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.510943598377592e-05,
+      "loss": 0.9677,
+      "step": 348000
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.510828805387617e-05,
+      "loss": 0.9217,
+      "step": 348010
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.510714012397642e-05,
+      "loss": 0.9247,
+      "step": 348020
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.510599219407668e-05,
+      "loss": 0.9582,
+      "step": 348030
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.510484426417693e-05,
+      "loss": 0.8683,
+      "step": 348040
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.5103696334277185e-05,
+      "loss": 0.8434,
+      "step": 348050
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.5102548404377435e-05,
+      "loss": 0.9546,
+      "step": 348060
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.510140047447769e-05,
+      "loss": 0.9802,
+      "step": 348070
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.510025254457794e-05,
+      "loss": 0.9913,
+      "step": 348080
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.50991046146782e-05,
+      "loss": 0.9579,
+      "step": 348090
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.509795668477845e-05,
+      "loss": 0.9468,
+      "step": 348100
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.50968087548787e-05,
+      "loss": 0.9409,
+      "step": 348110
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.509566082497895e-05,
+      "loss": 1.0067,
+      "step": 348120
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.5094512895079204e-05,
+      "loss": 1.0162,
+      "step": 348130
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.5093364965179454e-05,
+      "loss": 0.9128,
+      "step": 348140
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.5092217035279704e-05,
+      "loss": 0.8676,
+      "step": 348150
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.509106910537996e-05,
+      "loss": 0.9429,
+      "step": 348160
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 3.508992117548021e-05,
+      "loss": 0.9203,
+      "step": 348170
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.508877324558047e-05,
+      "loss": 0.9947,
+      "step": 348180
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.508762531568072e-05,
+      "loss": 0.8863,
+      "step": 348190
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.5086477385780974e-05,
+      "loss": 0.9485,
+      "step": 348200
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.5085329455881223e-05,
+      "loss": 0.9179,
+      "step": 348210
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.508418152598148e-05,
+      "loss": 0.932,
+      "step": 348220
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.508303359608173e-05,
+      "loss": 0.9618,
+      "step": 348230
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.508188566618198e-05,
+      "loss": 0.9535,
+      "step": 348240
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.5080737736282236e-05,
+      "loss": 0.9354,
+      "step": 348250
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.5079589806382486e-05,
+      "loss": 0.9391,
+      "step": 348260
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.5078441876482736e-05,
+      "loss": 0.9102,
+      "step": 348270
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.507729394658299e-05,
+      "loss": 0.9096,
+      "step": 348280
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.507614601668324e-05,
+      "loss": 0.9454,
+      "step": 348290
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.507499808678349e-05,
+      "loss": 0.9816,
+      "step": 348300
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.507385015688375e-05,
+      "loss": 0.9701,
+      "step": 348310
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.5072702226984e-05,
+      "loss": 0.8996,
+      "step": 348320
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.5071554297084256e-05,
+      "loss": 0.9719,
+      "step": 348330
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.5070406367184505e-05,
+      "loss": 1.0219,
+      "step": 348340
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.506925843728476e-05,
+      "loss": 0.9883,
+      "step": 348350
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.506811050738501e-05,
+      "loss": 0.8817,
+      "step": 348360
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.506696257748527e-05,
+      "loss": 0.9878,
+      "step": 348370
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.506581464758552e-05,
+      "loss": 0.9259,
+      "step": 348380
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 3.506466671768577e-05,
+      "loss": 0.8839,
+      "step": 348390
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.5063518787786025e-05,
+      "loss": 0.9529,
+      "step": 348400
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.5062370857886275e-05,
+      "loss": 0.8736,
+      "step": 348410
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.5061222927986525e-05,
+      "loss": 1.0315,
+      "step": 348420
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.506007499808678e-05,
+      "loss": 0.9146,
+      "step": 348430
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.505892706818703e-05,
+      "loss": 0.9266,
+      "step": 348440
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.505777913828728e-05,
+      "loss": 0.8767,
+      "step": 348450
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.505663120838754e-05,
+      "loss": 0.9202,
+      "step": 348460
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.505548327848779e-05,
+      "loss": 1.0288,
+      "step": 348470
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.5054335348588044e-05,
+      "loss": 0.8878,
+      "step": 348480
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.5053187418688294e-05,
+      "loss": 0.8932,
+      "step": 348490
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.505203948878855e-05,
+      "loss": 0.955,
+      "step": 348500
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.50508915588888e-05,
+      "loss": 0.8095,
+      "step": 348510
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.504974362898906e-05,
+      "loss": 1.0409,
+      "step": 348520
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.504859569908931e-05,
+      "loss": 0.8675,
+      "step": 348530
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.504744776918956e-05,
+      "loss": 0.9304,
+      "step": 348540
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.504629983928981e-05,
+      "loss": 0.9076,
+      "step": 348550
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.504515190939006e-05,
+      "loss": 0.9016,
+      "step": 348560
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.504400397949031e-05,
+      "loss": 0.9562,
+      "step": 348570
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.504285604959057e-05,
+      "loss": 0.9622,
+      "step": 348580
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.504170811969082e-05,
+      "loss": 0.9383,
+      "step": 348590
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.504056018979107e-05,
+      "loss": 0.9389,
+      "step": 348600
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 3.5039412259891326e-05,
+      "loss": 0.9204,
+      "step": 348610
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.5038264329991576e-05,
+      "loss": 0.919,
+      "step": 348620
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.503711640009183e-05,
+      "loss": 0.949,
+      "step": 348630
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.503596847019208e-05,
+      "loss": 0.8576,
+      "step": 348640
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.503482054029234e-05,
+      "loss": 0.9725,
+      "step": 348650
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.503367261039259e-05,
+      "loss": 0.921,
+      "step": 348660
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.5032524680492845e-05,
+      "loss": 1.022,
+      "step": 348670
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.5031376750593095e-05,
+      "loss": 0.9664,
+      "step": 348680
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.5030228820693345e-05,
+      "loss": 0.8831,
+      "step": 348690
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.50290808907936e-05,
+      "loss": 0.8381,
+      "step": 348700
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.502793296089385e-05,
+      "loss": 0.8942,
+      "step": 348710
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.50267850309941e-05,
+      "loss": 0.9518,
+      "step": 348720
+    },
+    {
+      "epoch": 16.0,
+      "eval_cer": 0.6686220447970015,
+      "eval_em": 0.007357212003872217,
+      "eval_f1": 0.007357212003872217,
+      "eval_loss": 1.24960196018219,
+      "eval_runtime": 1639.3782,
+      "eval_samples_per_second": 6.301,
+      "eval_steps_per_second": 0.788,
+      "eval_wer": 0.9926427879961278,
+      "step": 348722
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.502563710109436e-05,
+      "loss": 0.8785,
+      "step": 348730
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.502448917119461e-05,
+      "loss": 0.8535,
+      "step": 348740
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.502334124129486e-05,
+      "loss": 0.9412,
+      "step": 348750
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.5022193311395114e-05,
+      "loss": 0.8946,
+      "step": 348760
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.5021045381495364e-05,
+      "loss": 0.9667,
+      "step": 348770
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.501989745159562e-05,
+      "loss": 0.8782,
+      "step": 348780
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.501874952169587e-05,
+      "loss": 0.8232,
+      "step": 348790
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.501760159179613e-05,
+      "loss": 0.8711,
+      "step": 348800
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.501645366189638e-05,
+      "loss": 0.9672,
+      "step": 348810
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.501530573199663e-05,
+      "loss": 0.8308,
+      "step": 348820
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 3.5014157802096884e-05,
+      "loss": 0.8954,
+      "step": 348830
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.5013009872197134e-05,
+      "loss": 0.9108,
+      "step": 348840
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.5011861942297383e-05,
+      "loss": 0.806,
+      "step": 348850
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.501071401239764e-05,
+      "loss": 0.8694,
+      "step": 348860
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.500956608249789e-05,
+      "loss": 0.8641,
+      "step": 348870
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.500841815259814e-05,
+      "loss": 0.88,
+      "step": 348880
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.5007270222698396e-05,
+      "loss": 0.7639,
+      "step": 348890
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.5006122292798646e-05,
+      "loss": 0.8567,
+      "step": 348900
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.50049743628989e-05,
+      "loss": 0.8617,
+      "step": 348910
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.500382643299915e-05,
+      "loss": 0.8835,
+      "step": 348920
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.500267850309941e-05,
+      "loss": 0.8292,
+      "step": 348930
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.500153057319966e-05,
+      "loss": 0.878,
+      "step": 348940
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.5000382643299916e-05,
+      "loss": 0.856,
+      "step": 348950
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.4999234713400166e-05,
+      "loss": 0.8091,
+      "step": 348960
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.4998086783500416e-05,
+      "loss": 0.8976,
+      "step": 348970
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.499693885360067e-05,
+      "loss": 0.8848,
+      "step": 348980
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.499579092370092e-05,
+      "loss": 0.8543,
+      "step": 348990
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.499464299380117e-05,
+      "loss": 0.9002,
+      "step": 349000
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.499349506390143e-05,
+      "loss": 0.8881,
+      "step": 349010
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.499234713400168e-05,
+      "loss": 0.8927,
+      "step": 349020
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.499119920410193e-05,
+      "loss": 0.7738,
+      "step": 349030
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 3.4990051274202185e-05,
+      "loss": 0.8117,
+      "step": 349040
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.4988903344302435e-05,
+      "loss": 0.9442,
+      "step": 349050
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.498775541440269e-05,
+      "loss": 0.9561,
+      "step": 349060
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.498660748450294e-05,
+      "loss": 0.9033,
+      "step": 349070
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.49854595546032e-05,
+      "loss": 0.8604,
+      "step": 349080
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.498431162470345e-05,
+      "loss": 1.0437,
+      "step": 349090
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.4983163694803704e-05,
+      "loss": 0.8588,
+      "step": 349100
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.4982015764903954e-05,
+      "loss": 0.7767,
+      "step": 349110
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.4980867835004204e-05,
+      "loss": 0.9181,
+      "step": 349120
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.497971990510446e-05,
+      "loss": 0.825,
+      "step": 349130
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.497857197520471e-05,
+      "loss": 0.827,
+      "step": 349140
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.497742404530496e-05,
+      "loss": 0.919,
+      "step": 349150
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.497627611540522e-05,
+      "loss": 0.892,
+      "step": 349160
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.497512818550547e-05,
+      "loss": 0.8689,
+      "step": 349170
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.497398025560572e-05,
+      "loss": 0.8944,
+      "step": 349180
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.4972832325705973e-05,
+      "loss": 0.8697,
+      "step": 349190
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.497168439580622e-05,
+      "loss": 0.9656,
+      "step": 349200
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.497053646590648e-05,
+      "loss": 0.902,
+      "step": 349210
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.496938853600673e-05,
+      "loss": 0.8534,
+      "step": 349220
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.4968240606106986e-05,
+      "loss": 0.8112,
+      "step": 349230
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.4967092676207236e-05,
+      "loss": 0.8668,
+      "step": 349240
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.496594474630749e-05,
+      "loss": 0.871,
+      "step": 349250
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 3.496479681640774e-05,
+      "loss": 0.9258,
+      "step": 349260
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.496364888650799e-05,
+      "loss": 0.8833,
+      "step": 349270
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.496250095660825e-05,
+      "loss": 0.9126,
+      "step": 349280
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.49613530267085e-05,
+      "loss": 0.8453,
+      "step": 349290
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.496020509680875e-05,
+      "loss": 0.8443,
+      "step": 349300
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.4959057166909006e-05,
+      "loss": 0.9005,
+      "step": 349310
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.4957909237009255e-05,
+      "loss": 0.79,
+      "step": 349320
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.4956761307109505e-05,
+      "loss": 0.8699,
+      "step": 349330
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.495561337720976e-05,
+      "loss": 0.9412,
+      "step": 349340
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.495446544731001e-05,
+      "loss": 0.8473,
+      "step": 349350
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.495331751741027e-05,
+      "loss": 0.8979,
+      "step": 349360
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.495216958751052e-05,
+      "loss": 0.9231,
+      "step": 349370
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.4951021657610775e-05,
+      "loss": 0.8579,
+      "step": 349380
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.4949873727711025e-05,
+      "loss": 0.9335,
+      "step": 349390
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.494872579781128e-05,
+      "loss": 0.7872,
+      "step": 349400
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.494757786791153e-05,
+      "loss": 0.895,
+      "step": 349410
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.494642993801178e-05,
+      "loss": 1.0237,
+      "step": 349420
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.494528200811204e-05,
+      "loss": 0.7848,
+      "step": 349430
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.494413407821229e-05,
+      "loss": 0.9001,
+      "step": 349440
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.494298614831254e-05,
+      "loss": 0.9049,
+      "step": 349450
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.4941838218412794e-05,
+      "loss": 0.9114,
+      "step": 349460
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.4940690288513044e-05,
+      "loss": 0.86,
+      "step": 349470
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 3.4939542358613294e-05,
+      "loss": 0.8327,
+      "step": 349480
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.493839442871355e-05,
+      "loss": 0.9134,
+      "step": 349490
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.49372464988138e-05,
+      "loss": 0.8456,
+      "step": 349500
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.493609856891406e-05,
+      "loss": 0.8061,
+      "step": 349510
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.493495063901431e-05,
+      "loss": 0.8787,
+      "step": 349520
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.493380270911456e-05,
+      "loss": 0.8932,
+      "step": 349530
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.493265477921481e-05,
+      "loss": 0.8996,
+      "step": 349540
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.493150684931506e-05,
+      "loss": 0.905,
+      "step": 349550
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.493035891941532e-05,
+      "loss": 0.9711,
+      "step": 349560
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.492921098951557e-05,
+      "loss": 0.8737,
+      "step": 349570
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.492806305961582e-05,
+      "loss": 0.8184,
+      "step": 349580
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.4926915129716076e-05,
+      "loss": 0.9049,
+      "step": 349590
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.4925767199816326e-05,
+      "loss": 0.9493,
+      "step": 349600
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.4924619269916576e-05,
+      "loss": 0.9197,
+      "step": 349610
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.492347134001683e-05,
+      "loss": 0.8791,
+      "step": 349620
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.492232341011708e-05,
+      "loss": 0.9027,
+      "step": 349630
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.492117548021734e-05,
+      "loss": 0.8419,
+      "step": 349640
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.492002755031759e-05,
+      "loss": 0.8033,
+      "step": 349650
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.4918879620417845e-05,
+      "loss": 0.9073,
+      "step": 349660
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.4917731690518095e-05,
+      "loss": 0.8983,
+      "step": 349670
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.491658376061835e-05,
+      "loss": 0.9428,
+      "step": 349680
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.49154358307186e-05,
+      "loss": 0.8402,
+      "step": 349690
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 3.491428790081885e-05,
+      "loss": 0.9018,
+      "step": 349700
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.491313997091911e-05,
+      "loss": 0.8564,
+      "step": 349710
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.491199204101936e-05,
+      "loss": 0.9149,
+      "step": 349720
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.491084411111961e-05,
+      "loss": 0.8639,
+      "step": 349730
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.4909696181219864e-05,
+      "loss": 0.8829,
+      "step": 349740
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.4908548251320114e-05,
+      "loss": 0.8497,
+      "step": 349750
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.4907400321420364e-05,
+      "loss": 0.9727,
+      "step": 349760
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.490625239152062e-05,
+      "loss": 0.7543,
+      "step": 349770
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.490510446162087e-05,
+      "loss": 0.7654,
+      "step": 349780
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.490395653172113e-05,
+      "loss": 0.8477,
+      "step": 349790
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.490280860182138e-05,
+      "loss": 0.9045,
+      "step": 349800
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.4901660671921634e-05,
+      "loss": 0.9766,
+      "step": 349810
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.4900512742021884e-05,
+      "loss": 0.8122,
+      "step": 349820
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.489936481212214e-05,
+      "loss": 0.8616,
+      "step": 349830
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.489821688222239e-05,
+      "loss": 0.8922,
+      "step": 349840
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.489706895232264e-05,
+      "loss": 0.8718,
+      "step": 349850
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.4895921022422897e-05,
+      "loss": 0.8826,
+      "step": 349860
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.4894773092523146e-05,
+      "loss": 0.8207,
+      "step": 349870
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.4893625162623396e-05,
+      "loss": 0.8765,
+      "step": 349880
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.489247723272365e-05,
+      "loss": 0.8405,
+      "step": 349890
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.48913293028239e-05,
+      "loss": 0.8501,
+      "step": 349900
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.489018137292415e-05,
+      "loss": 0.9006,
+      "step": 349910
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 3.488903344302441e-05,
+      "loss": 0.9026,
+      "step": 349920
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.488788551312466e-05,
+      "loss": 0.8246,
+      "step": 349930
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.4886737583224916e-05,
+      "loss": 0.8672,
+      "step": 349940
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.4885589653325166e-05,
+      "loss": 0.8428,
+      "step": 349950
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.488444172342542e-05,
+      "loss": 0.8814,
+      "step": 349960
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.488329379352567e-05,
+      "loss": 0.9154,
+      "step": 349970
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.488214586362593e-05,
+      "loss": 0.8698,
+      "step": 349980
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.488099793372618e-05,
+      "loss": 0.7542,
+      "step": 349990
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.487985000382643e-05,
+      "loss": 0.9446,
+      "step": 350000
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.4878702073926685e-05,
+      "loss": 0.8276,
+      "step": 350010
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.4877554144026935e-05,
+      "loss": 0.8739,
+      "step": 350020
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.4876406214127185e-05,
+      "loss": 0.8587,
+      "step": 350030
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.487525828422744e-05,
+      "loss": 0.8725,
+      "step": 350040
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.487411035432769e-05,
+      "loss": 0.8565,
+      "step": 350050
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.487296242442794e-05,
+      "loss": 0.9808,
+      "step": 350060
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.48718144945282e-05,
+      "loss": 0.9093,
+      "step": 350070
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.487066656462845e-05,
+      "loss": 0.9246,
+      "step": 350080
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.4869518634728704e-05,
+      "loss": 0.8117,
+      "step": 350090
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.4868370704828954e-05,
+      "loss": 0.9317,
+      "step": 350100
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.486722277492921e-05,
+      "loss": 0.882,
+      "step": 350110
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.486607484502946e-05,
+      "loss": 0.8548,
+      "step": 350120
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 3.486492691512972e-05,
+      "loss": 0.8873,
+      "step": 350130
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.486377898522997e-05,
+      "loss": 0.9154,
+      "step": 350140
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.486263105533022e-05,
+      "loss": 0.9743,
+      "step": 350150
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.4861483125430474e-05,
+      "loss": 0.8632,
+      "step": 350160
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.4860335195530723e-05,
+      "loss": 0.8803,
+      "step": 350170
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.485918726563097e-05,
+      "loss": 0.8611,
+      "step": 350180
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.485803933573122e-05,
+      "loss": 0.9386,
+      "step": 350190
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.485689140583148e-05,
+      "loss": 0.9259,
+      "step": 350200
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.485574347593173e-05,
+      "loss": 0.9075,
+      "step": 350210
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.4854595546031986e-05,
+      "loss": 0.8425,
+      "step": 350220
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.4853447616132236e-05,
+      "loss": 0.864,
+      "step": 350230
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.485229968623249e-05,
+      "loss": 0.9867,
+      "step": 350240
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.485115175633274e-05,
+      "loss": 0.8125,
+      "step": 350250
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.4850003826433e-05,
+      "loss": 0.8954,
+      "step": 350260
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.484885589653325e-05,
+      "loss": 0.8607,
+      "step": 350270
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.48477079666335e-05,
+      "loss": 0.8472,
+      "step": 350280
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.4846560036733756e-05,
+      "loss": 0.9109,
+      "step": 350290
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.4845412106834005e-05,
+      "loss": 0.816,
+      "step": 350300
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.4844264176934255e-05,
+      "loss": 0.8378,
+      "step": 350310
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.484311624703451e-05,
+      "loss": 0.9847,
+      "step": 350320
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.484196831713476e-05,
+      "loss": 0.8859,
+      "step": 350330
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.484082038723501e-05,
+      "loss": 0.916,
+      "step": 350340
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 3.483967245733527e-05,
+      "loss": 0.7796,
+      "step": 350350
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.483852452743552e-05,
+      "loss": 0.9069,
+      "step": 350360
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.4837376597535775e-05,
+      "loss": 0.8519,
+      "step": 350370
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.4836228667636025e-05,
+      "loss": 0.971,
+      "step": 350380
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.483508073773628e-05,
+      "loss": 0.9661,
+      "step": 350390
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.483393280783653e-05,
+      "loss": 0.9657,
+      "step": 350400
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.483278487793679e-05,
+      "loss": 0.9497,
+      "step": 350410
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.483163694803704e-05,
+      "loss": 0.814,
+      "step": 350420
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.483048901813729e-05,
+      "loss": 0.8824,
+      "step": 350430
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.4829341088237544e-05,
+      "loss": 0.8463,
+      "step": 350440
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.4828193158337794e-05,
+      "loss": 0.9011,
+      "step": 350450
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.4827045228438044e-05,
+      "loss": 0.9352,
+      "step": 350460
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.48258972985383e-05,
+      "loss": 0.8481,
+      "step": 350470
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.482474936863855e-05,
+      "loss": 0.8893,
+      "step": 350480
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.48236014387388e-05,
+      "loss": 0.9678,
+      "step": 350490
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.482245350883906e-05,
+      "loss": 0.8882,
+      "step": 350500
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.4821305578939307e-05,
+      "loss": 0.9022,
+      "step": 350510
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.482015764903956e-05,
+      "loss": 0.8792,
+      "step": 350520
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.481900971913981e-05,
+      "loss": 0.8435,
+      "step": 350530
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.481786178924007e-05,
+      "loss": 0.8747,
+      "step": 350540
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.481671385934032e-05,
+      "loss": 0.8166,
+      "step": 350550
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.4815565929440576e-05,
+      "loss": 0.8839,
+      "step": 350560
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 3.4814417999540826e-05,
+      "loss": 0.8896,
+      "step": 350570
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.4813270069641076e-05,
+      "loss": 0.9348,
+      "step": 350580
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.481212213974133e-05,
+      "loss": 0.8091,
+      "step": 350590
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.481097420984158e-05,
+      "loss": 0.8682,
+      "step": 350600
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.480982627994183e-05,
+      "loss": 0.8437,
+      "step": 350610
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.480867835004209e-05,
+      "loss": 0.8823,
+      "step": 350620
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.480753042014234e-05,
+      "loss": 0.896,
+      "step": 350630
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.480638249024259e-05,
+      "loss": 0.8694,
+      "step": 350640
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.4805234560342845e-05,
+      "loss": 0.8697,
+      "step": 350650
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.4804086630443095e-05,
+      "loss": 0.9875,
+      "step": 350660
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.480293870054335e-05,
+      "loss": 0.8504,
+      "step": 350670
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.48017907706436e-05,
+      "loss": 0.86,
+      "step": 350680
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.480064284074386e-05,
+      "loss": 0.8861,
+      "step": 350690
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.479949491084411e-05,
+      "loss": 0.8871,
+      "step": 350700
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.4798346980944365e-05,
+      "loss": 0.9406,
+      "step": 350710
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.4797199051044614e-05,
+      "loss": 0.9298,
+      "step": 350720
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.4796051121144864e-05,
+      "loss": 0.9157,
+      "step": 350730
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.479490319124512e-05,
+      "loss": 0.9656,
+      "step": 350740
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.479375526134537e-05,
+      "loss": 0.9097,
+      "step": 350750
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.479260733144562e-05,
+      "loss": 0.914,
+      "step": 350760
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.479145940154588e-05,
+      "loss": 0.8873,
+      "step": 350770
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.479031147164613e-05,
+      "loss": 0.89,
+      "step": 350780
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.478916354174638e-05,
+      "loss": 0.8964,
+      "step": 350790
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.4788015611846634e-05,
+      "loss": 0.8868,
+      "step": 350800
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.4786867681946883e-05,
+      "loss": 0.9043,
+      "step": 350810
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.478571975204714e-05,
+      "loss": 0.9567,
+      "step": 350820
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.478457182214739e-05,
+      "loss": 0.9525,
+      "step": 350830
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.4783423892247647e-05,
+      "loss": 0.951,
+      "step": 350840
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.4782275962347896e-05,
+      "loss": 0.8811,
+      "step": 350850
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.4781128032448146e-05,
+      "loss": 0.8588,
+      "step": 350860
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.47799801025484e-05,
+      "loss": 0.8714,
+      "step": 350870
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.477883217264865e-05,
+      "loss": 0.7762,
+      "step": 350880
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.47776842427489e-05,
+      "loss": 0.8909,
+      "step": 350890
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.477653631284916e-05,
+      "loss": 0.8607,
+      "step": 350900
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.477538838294941e-05,
+      "loss": 0.9329,
+      "step": 350910
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.477424045304966e-05,
+      "loss": 0.9011,
+      "step": 350920
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.4773092523149916e-05,
+      "loss": 0.902,
+      "step": 350930
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.4771944593250165e-05,
+      "loss": 0.817,
+      "step": 350940
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.477079666335042e-05,
+      "loss": 0.9098,
+      "step": 350950
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.476964873345067e-05,
+      "loss": 0.8484,
+      "step": 350960
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.476850080355093e-05,
+      "loss": 0.9261,
+      "step": 350970
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.476735287365118e-05,
+      "loss": 0.8979,
+      "step": 350980
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.4766204943751435e-05,
+      "loss": 0.8828,
+      "step": 350990
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.4765057013851685e-05,
+      "loss": 0.8801,
+      "step": 351000
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 3.4763909083951935e-05,
+      "loss": 0.8717,
+      "step": 351010
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.476276115405219e-05,
+      "loss": 0.7973,
+      "step": 351020
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.476161322415244e-05,
+      "loss": 0.8787,
+      "step": 351030
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.476046529425269e-05,
+      "loss": 0.8539,
+      "step": 351040
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.475931736435295e-05,
+      "loss": 0.9249,
+      "step": 351050
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.47581694344532e-05,
+      "loss": 0.8709,
+      "step": 351060
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.475702150455345e-05,
+      "loss": 0.8383,
+      "step": 351070
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.4755873574653704e-05,
+      "loss": 0.9693,
+      "step": 351080
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.4754725644753954e-05,
+      "loss": 0.8656,
+      "step": 351090
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.475357771485421e-05,
+      "loss": 0.7656,
+      "step": 351100
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.475242978495446e-05,
+      "loss": 0.8934,
+      "step": 351110
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.475128185505472e-05,
+      "loss": 0.8766,
+      "step": 351120
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.475013392515497e-05,
+      "loss": 0.8184,
+      "step": 351130
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.4748985995255224e-05,
+      "loss": 0.834,
+      "step": 351140
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.474783806535547e-05,
+      "loss": 0.8688,
+      "step": 351150
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.474669013545572e-05,
+      "loss": 0.8064,
+      "step": 351160
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.474554220555598e-05,
+      "loss": 0.8733,
+      "step": 351170
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.474439427565623e-05,
+      "loss": 0.8956,
+      "step": 351180
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.474324634575648e-05,
+      "loss": 0.8972,
+      "step": 351190
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.4742098415856736e-05,
+      "loss": 0.9298,
+      "step": 351200
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.4740950485956986e-05,
+      "loss": 0.9201,
+      "step": 351210
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 3.4739802556057236e-05,
+      "loss": 0.8946,
+      "step": 351220
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.473865462615749e-05,
+      "loss": 0.8143,
+      "step": 351230
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.473750669625774e-05,
+      "loss": 0.8671,
+      "step": 351240
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.4736358766358e-05,
+      "loss": 0.8903,
+      "step": 351250
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.473521083645825e-05,
+      "loss": 0.8943,
+      "step": 351260
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.4734062906558506e-05,
+      "loss": 0.8515,
+      "step": 351270
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.4732914976658755e-05,
+      "loss": 0.8647,
+      "step": 351280
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.473176704675901e-05,
+      "loss": 0.9109,
+      "step": 351290
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.473061911685926e-05,
+      "loss": 0.8029,
+      "step": 351300
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.472947118695951e-05,
+      "loss": 0.9804,
+      "step": 351310
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.472832325705977e-05,
+      "loss": 0.85,
+      "step": 351320
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.472717532716002e-05,
+      "loss": 0.8775,
+      "step": 351330
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.472602739726027e-05,
+      "loss": 0.9195,
+      "step": 351340
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.4724879467360525e-05,
+      "loss": 0.8802,
+      "step": 351350
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.4723731537460775e-05,
+      "loss": 0.8822,
+      "step": 351360
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.4722583607561024e-05,
+      "loss": 0.8884,
+      "step": 351370
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.472143567766128e-05,
+      "loss": 0.9244,
+      "step": 351380
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.472028774776153e-05,
+      "loss": 0.9387,
+      "step": 351390
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.471913981786179e-05,
+      "loss": 0.9291,
+      "step": 351400
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.471799188796204e-05,
+      "loss": 0.8734,
+      "step": 351410
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.4716843958062294e-05,
+      "loss": 0.886,
+      "step": 351420
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.4715696028162544e-05,
+      "loss": 0.7854,
+      "step": 351430
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 3.47145480982628e-05,
+      "loss": 0.8583,
+      "step": 351440
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.471340016836305e-05,
+      "loss": 0.891,
+      "step": 351450
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.47122522384633e-05,
+      "loss": 0.847,
+      "step": 351460
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.471110430856356e-05,
+      "loss": 0.9044,
+      "step": 351470
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.470995637866381e-05,
+      "loss": 0.884,
+      "step": 351480
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.4708808448764057e-05,
+      "loss": 0.8587,
+      "step": 351490
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.470766051886431e-05,
+      "loss": 0.8539,
+      "step": 351500
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.470651258896456e-05,
+      "loss": 0.8667,
+      "step": 351510
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.470536465906481e-05,
+      "loss": 0.944,
+      "step": 351520
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.470421672916507e-05,
+      "loss": 0.8799,
+      "step": 351530
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.470306879926532e-05,
+      "loss": 0.8396,
+      "step": 351540
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.4701920869365576e-05,
+      "loss": 0.9311,
+      "step": 351550
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.4700772939465826e-05,
+      "loss": 0.7667,
+      "step": 351560
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.469962500956608e-05,
+      "loss": 0.8804,
+      "step": 351570
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.469847707966633e-05,
+      "loss": 0.925,
+      "step": 351580
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.469732914976658e-05,
+      "loss": 0.9319,
+      "step": 351590
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.469618121986684e-05,
+      "loss": 0.9336,
+      "step": 351600
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.469503328996709e-05,
+      "loss": 0.8842,
+      "step": 351610
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.469388536006734e-05,
+      "loss": 0.8563,
+      "step": 351620
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.4692737430167595e-05,
+      "loss": 0.8631,
+      "step": 351630
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.4691589500267845e-05,
+      "loss": 0.8696,
+      "step": 351640
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.4690441570368095e-05,
+      "loss": 0.9234,
+      "step": 351650
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 3.468929364046835e-05,
+      "loss": 0.9227,
+      "step": 351660
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.46881457105686e-05,
+      "loss": 0.8009,
+      "step": 351670
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.468699778066886e-05,
+      "loss": 0.898,
+      "step": 351680
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.468584985076911e-05,
+      "loss": 0.8336,
+      "step": 351690
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.4684701920869364e-05,
+      "loss": 0.7843,
+      "step": 351700
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.4683553990969614e-05,
+      "loss": 0.8921,
+      "step": 351710
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.468240606106987e-05,
+      "loss": 0.9336,
+      "step": 351720
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.468125813117012e-05,
+      "loss": 0.8216,
+      "step": 351730
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.468011020127037e-05,
+      "loss": 0.8658,
+      "step": 351740
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.467896227137063e-05,
+      "loss": 0.8221,
+      "step": 351750
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.467781434147088e-05,
+      "loss": 0.8171,
+      "step": 351760
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.467666641157113e-05,
+      "loss": 0.8673,
+      "step": 351770
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.4675518481671384e-05,
+      "loss": 0.8245,
+      "step": 351780
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.4674370551771633e-05,
+      "loss": 0.8756,
+      "step": 351790
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.467322262187188e-05,
+      "loss": 0.9428,
+      "step": 351800
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.467207469197214e-05,
+      "loss": 0.8086,
+      "step": 351810
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.467092676207239e-05,
+      "loss": 0.8392,
+      "step": 351820
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.4669778832172646e-05,
+      "loss": 0.8814,
+      "step": 351830
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.4668630902272896e-05,
+      "loss": 0.8087,
+      "step": 351840
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.466748297237315e-05,
+      "loss": 0.9055,
+      "step": 351850
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.46663350424734e-05,
+      "loss": 0.9091,
+      "step": 351860
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.466518711257366e-05,
+      "loss": 0.8769,
+      "step": 351870
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 3.466403918267391e-05,
+      "loss": 0.8792,
+      "step": 351880
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.466289125277416e-05,
+      "loss": 0.9144,
+      "step": 351890
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.4661743322874416e-05,
+      "loss": 0.8402,
+      "step": 351900
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.4660595392974666e-05,
+      "loss": 0.8037,
+      "step": 351910
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.4659447463074915e-05,
+      "loss": 0.9137,
+      "step": 351920
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.465829953317517e-05,
+      "loss": 0.8261,
+      "step": 351930
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.465715160327542e-05,
+      "loss": 0.8517,
+      "step": 351940
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.465600367337567e-05,
+      "loss": 0.8456,
+      "step": 351950
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.465485574347593e-05,
+      "loss": 0.8639,
+      "step": 351960
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.465370781357618e-05,
+      "loss": 0.8892,
+      "step": 351970
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.4652559883676435e-05,
+      "loss": 0.849,
+      "step": 351980
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.4651411953776685e-05,
+      "loss": 0.9212,
+      "step": 351990
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.465026402387694e-05,
+      "loss": 0.823,
+      "step": 352000
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.464911609397719e-05,
+      "loss": 0.9135,
+      "step": 352010
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.464796816407745e-05,
+      "loss": 0.9455,
+      "step": 352020
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.46468202341777e-05,
+      "loss": 0.7831,
+      "step": 352030
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.464567230427795e-05,
+      "loss": 0.9244,
+      "step": 352040
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.4644524374378204e-05,
+      "loss": 0.776,
+      "step": 352050
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.4643376444478454e-05,
+      "loss": 0.9496,
+      "step": 352060
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.4642228514578704e-05,
+      "loss": 0.9256,
+      "step": 352070
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.464108058467896e-05,
+      "loss": 0.9144,
+      "step": 352080
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.463993265477921e-05,
+      "loss": 0.8829,
+      "step": 352090
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 3.463878472487946e-05,
+      "loss": 0.8841,
+      "step": 352100
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.463763679497972e-05,
+      "loss": 0.9437,
+      "step": 352110
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.463648886507997e-05,
+      "loss": 0.9566,
+      "step": 352120
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.463534093518022e-05,
+      "loss": 0.8928,
+      "step": 352130
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.463419300528047e-05,
+      "loss": 0.953,
+      "step": 352140
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.463304507538073e-05,
+      "loss": 0.8852,
+      "step": 352150
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.463189714548098e-05,
+      "loss": 0.8938,
+      "step": 352160
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.4630749215581236e-05,
+      "loss": 0.8361,
+      "step": 352170
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.4629601285681486e-05,
+      "loss": 0.9132,
+      "step": 352180
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.4628453355781736e-05,
+      "loss": 0.8426,
+      "step": 352190
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.462730542588199e-05,
+      "loss": 0.837,
+      "step": 352200
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.462615749598224e-05,
+      "loss": 0.8236,
+      "step": 352210
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.462500956608249e-05,
+      "loss": 0.8254,
+      "step": 352220
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.462386163618274e-05,
+      "loss": 0.9273,
+      "step": 352230
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.4622713706283e-05,
+      "loss": 0.8778,
+      "step": 352240
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.462156577638325e-05,
+      "loss": 0.8454,
+      "step": 352250
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.4620417846483505e-05,
+      "loss": 0.9449,
+      "step": 352260
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.4619269916583755e-05,
+      "loss": 0.8571,
+      "step": 352270
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.461812198668401e-05,
+      "loss": 0.8487,
+      "step": 352280
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.461697405678426e-05,
+      "loss": 0.9922,
+      "step": 352290
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.461582612688452e-05,
+      "loss": 0.9111,
+      "step": 352300
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 3.461467819698477e-05,
+      "loss": 0.8834,
+      "step": 352310
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.461353026708502e-05,
+      "loss": 0.9656,
+      "step": 352320
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.4612382337185275e-05,
+      "loss": 0.7682,
+      "step": 352330
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.4611234407285525e-05,
+      "loss": 0.8473,
+      "step": 352340
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.4610086477385774e-05,
+      "loss": 0.8137,
+      "step": 352350
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.460893854748603e-05,
+      "loss": 0.8927,
+      "step": 352360
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.460779061758628e-05,
+      "loss": 0.9014,
+      "step": 352370
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.460664268768653e-05,
+      "loss": 0.9244,
+      "step": 352380
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.460549475778679e-05,
+      "loss": 0.9115,
+      "step": 352390
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.460434682788704e-05,
+      "loss": 0.8483,
+      "step": 352400
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.4603198897987294e-05,
+      "loss": 0.8496,
+      "step": 352410
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.4602050968087544e-05,
+      "loss": 0.8127,
+      "step": 352420
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.46009030381878e-05,
+      "loss": 0.8622,
+      "step": 352430
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.459975510828805e-05,
+      "loss": 0.9384,
+      "step": 352440
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.459860717838831e-05,
+      "loss": 0.8661,
+      "step": 352450
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.459745924848856e-05,
+      "loss": 0.9186,
+      "step": 352460
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.4596311318588807e-05,
+      "loss": 0.9073,
+      "step": 352470
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.459516338868906e-05,
+      "loss": 0.8887,
+      "step": 352480
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.459401545878931e-05,
+      "loss": 0.8197,
+      "step": 352490
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.459286752888956e-05,
+      "loss": 0.8768,
+      "step": 352500
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.459171959898982e-05,
+      "loss": 0.9307,
+      "step": 352510
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.459057166909007e-05,
+      "loss": 0.924,
+      "step": 352520
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 3.458942373919032e-05,
+      "loss": 0.8654,
+      "step": 352530
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.4588275809290576e-05,
+      "loss": 0.8684,
+      "step": 352540
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.4587127879390826e-05,
+      "loss": 0.8144,
+      "step": 352550
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.458597994949108e-05,
+      "loss": 0.9383,
+      "step": 352560
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.458483201959133e-05,
+      "loss": 0.8571,
+      "step": 352570
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.458368408969159e-05,
+      "loss": 0.8376,
+      "step": 352580
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.458253615979184e-05,
+      "loss": 0.8281,
+      "step": 352590
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.4581388229892095e-05,
+      "loss": 1.0144,
+      "step": 352600
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.4580240299992345e-05,
+      "loss": 0.814,
+      "step": 352610
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.4579092370092595e-05,
+      "loss": 0.9192,
+      "step": 352620
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.457794444019285e-05,
+      "loss": 0.8641,
+      "step": 352630
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.45767965102931e-05,
+      "loss": 0.9043,
+      "step": 352640
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.457564858039335e-05,
+      "loss": 0.9599,
+      "step": 352650
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.457450065049361e-05,
+      "loss": 0.8847,
+      "step": 352660
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.457335272059386e-05,
+      "loss": 0.9213,
+      "step": 352670
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.457220479069411e-05,
+      "loss": 1.0355,
+      "step": 352680
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.4571056860794364e-05,
+      "loss": 0.8753,
+      "step": 352690
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.4569908930894614e-05,
+      "loss": 0.8787,
+      "step": 352700
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.456876100099487e-05,
+      "loss": 0.9045,
+      "step": 352710
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.456761307109512e-05,
+      "loss": 0.9332,
+      "step": 352720
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.456646514119538e-05,
+      "loss": 0.8091,
+      "step": 352730
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.456531721129563e-05,
+      "loss": 0.8891,
+      "step": 352740
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 3.4564169281395884e-05,
+      "loss": 0.9171,
+      "step": 352750
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.4563021351496134e-05,
+      "loss": 0.886,
+      "step": 352760
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.4561873421596383e-05,
+      "loss": 0.9001,
+      "step": 352770
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.456072549169664e-05,
+      "loss": 0.8255,
+      "step": 352780
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.455957756179689e-05,
+      "loss": 0.8878,
+      "step": 352790
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.455842963189714e-05,
+      "loss": 0.9498,
+      "step": 352800
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.4557281701997396e-05,
+      "loss": 0.8783,
+      "step": 352810
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.4556133772097646e-05,
+      "loss": 0.7996,
+      "step": 352820
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.4554985842197896e-05,
+      "loss": 0.8329,
+      "step": 352830
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.455383791229815e-05,
+      "loss": 0.8985,
+      "step": 352840
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.45526899823984e-05,
+      "loss": 0.8839,
+      "step": 352850
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.455154205249866e-05,
+      "loss": 0.8784,
+      "step": 352860
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.455039412259891e-05,
+      "loss": 0.89,
+      "step": 352870
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.4549246192699166e-05,
+      "loss": 0.9168,
+      "step": 352880
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.4548098262799416e-05,
+      "loss": 0.8516,
+      "step": 352890
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.454695033289967e-05,
+      "loss": 0.9015,
+      "step": 352900
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.454580240299992e-05,
+      "loss": 0.8881,
+      "step": 352910
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.454465447310017e-05,
+      "loss": 0.8178,
+      "step": 352920
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.454350654320042e-05,
+      "loss": 0.9122,
+      "step": 352930
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.454235861330068e-05,
+      "loss": 0.9081,
+      "step": 352940
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.454121068340093e-05,
+      "loss": 0.9366,
+      "step": 352950
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.454006275350118e-05,
+      "loss": 0.7878,
+      "step": 352960
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 3.4538914823601435e-05,
+      "loss": 0.8481,
+      "step": 352970
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.4537766893701685e-05,
+      "loss": 0.7869,
+      "step": 352980
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.453661896380194e-05,
+      "loss": 0.9085,
+      "step": 352990
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.453547103390219e-05,
+      "loss": 0.8264,
+      "step": 353000
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.453432310400245e-05,
+      "loss": 0.8801,
+      "step": 353010
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.45331751741027e-05,
+      "loss": 0.8771,
+      "step": 353020
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.4532027244202954e-05,
+      "loss": 0.803,
+      "step": 353030
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.4530879314303204e-05,
+      "loss": 0.94,
+      "step": 353040
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.4529731384403454e-05,
+      "loss": 0.8689,
+      "step": 353050
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.452858345450371e-05,
+      "loss": 0.9021,
+      "step": 353060
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.452743552460396e-05,
+      "loss": 1.023,
+      "step": 353070
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.452628759470421e-05,
+      "loss": 0.9241,
+      "step": 353080
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.452513966480447e-05,
+      "loss": 1.0373,
+      "step": 353090
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.452399173490472e-05,
+      "loss": 0.9181,
+      "step": 353100
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.4522843805004967e-05,
+      "loss": 1.0198,
+      "step": 353110
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.452169587510522e-05,
+      "loss": 0.7908,
+      "step": 353120
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.452054794520547e-05,
+      "loss": 0.8691,
+      "step": 353130
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.451940001530573e-05,
+      "loss": 0.977,
+      "step": 353140
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.451825208540598e-05,
+      "loss": 0.9194,
+      "step": 353150
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.4517104155506236e-05,
+      "loss": 0.8547,
+      "step": 353160
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.4515956225606486e-05,
+      "loss": 0.8993,
+      "step": 353170
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.451480829570674e-05,
+      "loss": 0.9007,
+      "step": 353180
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.451366036580699e-05,
+      "loss": 0.8422,
+      "step": 353190
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.451251243590724e-05,
+      "loss": 0.8872,
+      "step": 353200
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.45113645060075e-05,
+      "loss": 0.9119,
+      "step": 353210
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.451021657610775e-05,
+      "loss": 0.9302,
+      "step": 353220
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.4509068646208e-05,
+      "loss": 0.8344,
+      "step": 353230
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.4507920716308255e-05,
+      "loss": 0.8353,
+      "step": 353240
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.4506772786408505e-05,
+      "loss": 0.9113,
+      "step": 353250
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.4505624856508755e-05,
+      "loss": 0.8738,
+      "step": 353260
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.450447692660901e-05,
+      "loss": 0.9247,
+      "step": 353270
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.450332899670926e-05,
+      "loss": 0.8505,
+      "step": 353280
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.450218106680952e-05,
+      "loss": 0.7879,
+      "step": 353290
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.450103313690977e-05,
+      "loss": 0.9122,
+      "step": 353300
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.4499885207010025e-05,
+      "loss": 0.8152,
+      "step": 353310
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.4498737277110275e-05,
+      "loss": 0.8633,
+      "step": 353320
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.449758934721053e-05,
+      "loss": 0.9108,
+      "step": 353330
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.449644141731078e-05,
+      "loss": 0.8796,
+      "step": 353340
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.449529348741103e-05,
+      "loss": 0.8531,
+      "step": 353350
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.449414555751129e-05,
+      "loss": 0.8543,
+      "step": 353360
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.449299762761154e-05,
+      "loss": 0.8784,
+      "step": 353370
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.449184969771179e-05,
+      "loss": 0.919,
+      "step": 353380
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.4490701767812044e-05,
+      "loss": 0.9899,
+      "step": 353390
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 3.4489553837912294e-05,
+      "loss": 0.9478,
+      "step": 353400
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.4488405908012544e-05,
+      "loss": 0.8638,
+      "step": 353410
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.44872579781128e-05,
+      "loss": 0.892,
+      "step": 353420
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.448611004821305e-05,
+      "loss": 0.8047,
+      "step": 353430
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.448496211831331e-05,
+      "loss": 0.9442,
+      "step": 353440
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.4483814188413557e-05,
+      "loss": 0.8656,
+      "step": 353450
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.448266625851381e-05,
+      "loss": 0.8198,
+      "step": 353460
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.448151832861406e-05,
+      "loss": 0.9238,
+      "step": 353470
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.448037039871432e-05,
+      "loss": 0.9006,
+      "step": 353480
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.447922246881457e-05,
+      "loss": 0.8626,
+      "step": 353490
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.447807453891482e-05,
+      "loss": 0.9682,
+      "step": 353500
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.4476926609015076e-05,
+      "loss": 0.9808,
+      "step": 353510
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.4475778679115326e-05,
+      "loss": 0.8343,
+      "step": 353520
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.4474630749215576e-05,
+      "loss": 0.9683,
+      "step": 353530
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.447348281931583e-05,
+      "loss": 0.8548,
+      "step": 353540
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.447233488941608e-05,
+      "loss": 0.883,
+      "step": 353550
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.447118695951633e-05,
+      "loss": 0.9218,
+      "step": 353560
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.447003902961659e-05,
+      "loss": 0.849,
+      "step": 353570
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.446889109971684e-05,
+      "loss": 0.8918,
+      "step": 353580
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.4467743169817095e-05,
+      "loss": 0.7556,
+      "step": 353590
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.4466595239917345e-05,
+      "loss": 0.8708,
+      "step": 353600
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.44654473100176e-05,
+      "loss": 0.8481,
+      "step": 353610
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 3.446429938011785e-05,
+      "loss": 0.8798,
+      "step": 353620
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.44631514502181e-05,
+      "loss": 0.8481,
+      "step": 353630
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.446200352031836e-05,
+      "loss": 0.9184,
+      "step": 353640
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.446085559041861e-05,
+      "loss": 0.8475,
+      "step": 353650
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.445970766051886e-05,
+      "loss": 0.8415,
+      "step": 353660
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.4458559730619114e-05,
+      "loss": 0.848,
+      "step": 353670
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.4457411800719364e-05,
+      "loss": 0.9901,
+      "step": 353680
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.4456263870819614e-05,
+      "loss": 0.8856,
+      "step": 353690
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.445511594091987e-05,
+      "loss": 0.8285,
+      "step": 353700
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.445396801102012e-05,
+      "loss": 0.9223,
+      "step": 353710
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.445282008112038e-05,
+      "loss": 0.8729,
+      "step": 353720
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.445167215122063e-05,
+      "loss": 0.7923,
+      "step": 353730
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.4450524221320884e-05,
+      "loss": 0.8777,
+      "step": 353740
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.4449376291421133e-05,
+      "loss": 0.9358,
+      "step": 353750
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.444822836152139e-05,
+      "loss": 0.8041,
+      "step": 353760
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.444708043162164e-05,
+      "loss": 0.8783,
+      "step": 353770
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.444593250172189e-05,
+      "loss": 0.8439,
+      "step": 353780
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.4444784571822146e-05,
+      "loss": 0.8491,
+      "step": 353790
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.4443636641922396e-05,
+      "loss": 0.9929,
+      "step": 353800
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.4442488712022646e-05,
+      "loss": 0.9413,
+      "step": 353810
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.44413407821229e-05,
+      "loss": 0.8379,
+      "step": 353820
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.444019285222315e-05,
+      "loss": 0.9332,
+      "step": 353830
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 3.44390449223234e-05,
+      "loss": 0.9048,
+      "step": 353840
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.443789699242366e-05,
+      "loss": 0.838,
+      "step": 353850
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.443674906252391e-05,
+      "loss": 0.8205,
+      "step": 353860
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.4435601132624166e-05,
+      "loss": 0.8563,
+      "step": 353870
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.4434453202724415e-05,
+      "loss": 0.8981,
+      "step": 353880
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.443330527282467e-05,
+      "loss": 0.8918,
+      "step": 353890
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.443215734292492e-05,
+      "loss": 0.9511,
+      "step": 353900
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.443100941302518e-05,
+      "loss": 0.8386,
+      "step": 353910
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.442986148312543e-05,
+      "loss": 0.8967,
+      "step": 353920
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.442871355322568e-05,
+      "loss": 0.9259,
+      "step": 353930
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.4427565623325935e-05,
+      "loss": 0.8366,
+      "step": 353940
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.4426417693426185e-05,
+      "loss": 0.7985,
+      "step": 353950
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.4425269763526435e-05,
+      "loss": 0.9042,
+      "step": 353960
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.442412183362669e-05,
+      "loss": 0.8865,
+      "step": 353970
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.442297390372694e-05,
+      "loss": 0.8386,
+      "step": 353980
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.442182597382719e-05,
+      "loss": 0.8365,
+      "step": 353990
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.442067804392745e-05,
+      "loss": 0.8969,
+      "step": 354000
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.44195301140277e-05,
+      "loss": 0.9344,
+      "step": 354010
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.4418382184127954e-05,
+      "loss": 0.9848,
+      "step": 354020
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.4417234254228204e-05,
+      "loss": 0.9119,
+      "step": 354030
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.441608632432846e-05,
+      "loss": 0.8516,
+      "step": 354040
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.441493839442871e-05,
+      "loss": 0.8603,
+      "step": 354050
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 3.441379046452897e-05,
+      "loss": 0.8714,
+      "step": 354060
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.441264253462922e-05,
+      "loss": 0.8507,
+      "step": 354070
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.441149460472947e-05,
+      "loss": 0.8728,
+      "step": 354080
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.441034667482972e-05,
+      "loss": 0.8426,
+      "step": 354090
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.440919874492997e-05,
+      "loss": 0.9016,
+      "step": 354100
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.440805081503022e-05,
+      "loss": 0.8572,
+      "step": 354110
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.440690288513048e-05,
+      "loss": 0.9162,
+      "step": 354120
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.440575495523073e-05,
+      "loss": 1.0017,
+      "step": 354130
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.440460702533098e-05,
+      "loss": 0.9061,
+      "step": 354140
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.4403459095431236e-05,
+      "loss": 0.9151,
+      "step": 354150
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.4402311165531486e-05,
+      "loss": 0.7891,
+      "step": 354160
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.440116323563174e-05,
+      "loss": 0.9159,
+      "step": 354170
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.440001530573199e-05,
+      "loss": 0.8226,
+      "step": 354180
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.439886737583225e-05,
+      "loss": 0.971,
+      "step": 354190
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.43977194459325e-05,
+      "loss": 0.7844,
+      "step": 354200
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.4396571516032756e-05,
+      "loss": 0.9236,
+      "step": 354210
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.4395423586133005e-05,
+      "loss": 0.8096,
+      "step": 354220
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.4394275656233255e-05,
+      "loss": 0.9218,
+      "step": 354230
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.439312772633351e-05,
+      "loss": 0.9381,
+      "step": 354240
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.439197979643376e-05,
+      "loss": 0.9942,
+      "step": 354250
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.439083186653401e-05,
+      "loss": 0.8874,
+      "step": 354260
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.438968393663427e-05,
+      "loss": 0.99,
+      "step": 354270
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 3.438853600673452e-05,
+      "loss": 0.8585,
+      "step": 354280
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.438738807683477e-05,
+      "loss": 0.8592,
+      "step": 354290
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.4386240146935025e-05,
+      "loss": 0.883,
+      "step": 354300
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.4385092217035274e-05,
+      "loss": 0.8851,
+      "step": 354310
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.438394428713553e-05,
+      "loss": 0.8527,
+      "step": 354320
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.438279635723578e-05,
+      "loss": 0.8094,
+      "step": 354330
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.438164842733604e-05,
+      "loss": 0.8778,
+      "step": 354340
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.438050049743629e-05,
+      "loss": 0.9337,
+      "step": 354350
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.437935256753654e-05,
+      "loss": 0.8553,
+      "step": 354360
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.4378204637636794e-05,
+      "loss": 0.8831,
+      "step": 354370
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.4377056707737044e-05,
+      "loss": 0.8386,
+      "step": 354380
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.4375908777837294e-05,
+      "loss": 0.8258,
+      "step": 354390
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.437476084793755e-05,
+      "loss": 0.8036,
+      "step": 354400
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.43736129180378e-05,
+      "loss": 0.7901,
+      "step": 354410
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.437246498813805e-05,
+      "loss": 0.7864,
+      "step": 354420
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.4371317058238307e-05,
+      "loss": 0.8448,
+      "step": 354430
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.4370169128338556e-05,
+      "loss": 0.8626,
+      "step": 354440
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.436902119843881e-05,
+      "loss": 0.839,
+      "step": 354450
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.436787326853906e-05,
+      "loss": 0.9237,
+      "step": 354460
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.436672533863932e-05,
+      "loss": 0.9079,
+      "step": 354470
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.436557740873957e-05,
+      "loss": 0.95,
+      "step": 354480
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 3.4364429478839826e-05,
+      "loss": 1.0202,
+      "step": 354490
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.4363281548940076e-05,
+      "loss": 0.9098,
+      "step": 354500
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.4362133619040326e-05,
+      "loss": 0.9274,
+      "step": 354510
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.436098568914058e-05,
+      "loss": 0.8346,
+      "step": 354520
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.435983775924083e-05,
+      "loss": 0.9494,
+      "step": 354530
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.435868982934108e-05,
+      "loss": 0.9354,
+      "step": 354540
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.435754189944134e-05,
+      "loss": 0.8553,
+      "step": 354550
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.435639396954159e-05,
+      "loss": 0.8689,
+      "step": 354560
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.435524603964184e-05,
+      "loss": 0.9189,
+      "step": 354570
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.4354098109742095e-05,
+      "loss": 0.9236,
+      "step": 354580
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.4352950179842345e-05,
+      "loss": 0.9524,
+      "step": 354590
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.43518022499426e-05,
+      "loss": 0.9572,
+      "step": 354600
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.435065432004285e-05,
+      "loss": 0.9878,
+      "step": 354610
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.434950639014311e-05,
+      "loss": 0.8487,
+      "step": 354620
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.434835846024336e-05,
+      "loss": 0.8865,
+      "step": 354630
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.4347210530343614e-05,
+      "loss": 0.9316,
+      "step": 354640
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.4346062600443864e-05,
+      "loss": 0.9757,
+      "step": 354650
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.4344914670544114e-05,
+      "loss": 0.8919,
+      "step": 354660
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.434376674064437e-05,
+      "loss": 0.9837,
+      "step": 354670
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.434261881074462e-05,
+      "loss": 0.8543,
+      "step": 354680
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.434147088084487e-05,
+      "loss": 0.8443,
+      "step": 354690
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.434032295094513e-05,
+      "loss": 0.9221,
+      "step": 354700
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 3.433917502104538e-05,
+      "loss": 0.8729,
+      "step": 354710
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.433802709114563e-05,
+      "loss": 0.9461,
+      "step": 354720
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.4336879161245883e-05,
+      "loss": 0.849,
+      "step": 354730
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.433573123134613e-05,
+      "loss": 0.9083,
+      "step": 354740
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.433458330144639e-05,
+      "loss": 0.8828,
+      "step": 354750
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.433343537154664e-05,
+      "loss": 0.9009,
+      "step": 354760
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.4332287441646896e-05,
+      "loss": 0.9019,
+      "step": 354770
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.4331139511747146e-05,
+      "loss": 0.8276,
+      "step": 354780
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.43299915818474e-05,
+      "loss": 0.946,
+      "step": 354790
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.432884365194765e-05,
+      "loss": 0.9055,
+      "step": 354800
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.43276957220479e-05,
+      "loss": 0.829,
+      "step": 354810
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.432654779214816e-05,
+      "loss": 0.8845,
+      "step": 354820
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.432539986224841e-05,
+      "loss": 0.8063,
+      "step": 354830
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.432425193234866e-05,
+      "loss": 0.8306,
+      "step": 354840
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.4323104002448916e-05,
+      "loss": 1.0121,
+      "step": 354850
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.4321956072549165e-05,
+      "loss": 0.8339,
+      "step": 354860
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.4320808142649415e-05,
+      "loss": 0.9143,
+      "step": 354870
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.431966021274967e-05,
+      "loss": 0.8628,
+      "step": 354880
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.431851228284992e-05,
+      "loss": 0.8524,
+      "step": 354890
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.431736435295018e-05,
+      "loss": 0.8725,
+      "step": 354900
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.431621642305043e-05,
+      "loss": 0.9065,
+      "step": 354910
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.4315068493150685e-05,
+      "loss": 0.8802,
+      "step": 354920
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 3.4313920563250935e-05,
+      "loss": 0.8608,
+      "step": 354930
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.431277263335119e-05,
+      "loss": 0.8373,
+      "step": 354940
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.431162470345144e-05,
+      "loss": 0.9759,
+      "step": 354950
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.431047677355169e-05,
+      "loss": 0.8646,
+      "step": 354960
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.430932884365194e-05,
+      "loss": 0.87,
+      "step": 354970
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.43081809137522e-05,
+      "loss": 0.8986,
+      "step": 354980
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.430703298385245e-05,
+      "loss": 0.8682,
+      "step": 354990
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.43058850539527e-05,
+      "loss": 0.8848,
+      "step": 355000
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.4304737124052954e-05,
+      "loss": 0.8321,
+      "step": 355010
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.4303589194153204e-05,
+      "loss": 0.9834,
+      "step": 355020
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.430244126425346e-05,
+      "loss": 0.9508,
+      "step": 355030
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.430129333435371e-05,
+      "loss": 0.8241,
+      "step": 355040
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.430014540445397e-05,
+      "loss": 0.9243,
+      "step": 355050
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.429899747455422e-05,
+      "loss": 0.9179,
+      "step": 355060
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.429784954465447e-05,
+      "loss": 0.8807,
+      "step": 355070
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.429670161475472e-05,
+      "loss": 0.9234,
+      "step": 355080
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.429555368485497e-05,
+      "loss": 0.9859,
+      "step": 355090
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.429440575495523e-05,
+      "loss": 0.8714,
+      "step": 355100
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.429325782505548e-05,
+      "loss": 0.9717,
+      "step": 355110
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.429210989515573e-05,
+      "loss": 0.8452,
+      "step": 355120
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.4290961965255986e-05,
+      "loss": 0.8429,
+      "step": 355130
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.4289814035356236e-05,
+      "loss": 0.85,
+      "step": 355140
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 3.4288666105456486e-05,
+      "loss": 0.9137,
+      "step": 355150
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.428751817555674e-05,
+      "loss": 0.9925,
+      "step": 355160
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.428637024565699e-05,
+      "loss": 0.9562,
+      "step": 355170
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.428522231575725e-05,
+      "loss": 0.9587,
+      "step": 355180
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.42840743858575e-05,
+      "loss": 0.9172,
+      "step": 355190
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.4282926455957755e-05,
+      "loss": 0.9116,
+      "step": 355200
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.4281778526058005e-05,
+      "loss": 0.8684,
+      "step": 355210
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.428063059615826e-05,
+      "loss": 0.9117,
+      "step": 355220
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.427948266625851e-05,
+      "loss": 0.8949,
+      "step": 355230
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.427833473635876e-05,
+      "loss": 0.842,
+      "step": 355240
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.427718680645902e-05,
+      "loss": 0.8233,
+      "step": 355250
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.427603887655927e-05,
+      "loss": 0.8569,
+      "step": 355260
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.427489094665952e-05,
+      "loss": 0.8288,
+      "step": 355270
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.4273743016759775e-05,
+      "loss": 0.877,
+      "step": 355280
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.4272595086860024e-05,
+      "loss": 0.8346,
+      "step": 355290
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.4271447156960274e-05,
+      "loss": 0.7609,
+      "step": 355300
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.427029922706053e-05,
+      "loss": 0.8592,
+      "step": 355310
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.426915129716078e-05,
+      "loss": 0.9017,
+      "step": 355320
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.426800336726104e-05,
+      "loss": 0.9373,
+      "step": 355330
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.426685543736129e-05,
+      "loss": 0.927,
+      "step": 355340
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.4265707507461544e-05,
+      "loss": 0.8922,
+      "step": 355350
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.4264559577561794e-05,
+      "loss": 0.9325,
+      "step": 355360
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.426341164766205e-05,
+      "loss": 0.844,
+      "step": 355370
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.42622637177623e-05,
+      "loss": 0.871,
+      "step": 355380
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.426111578786255e-05,
+      "loss": 0.8593,
+      "step": 355390
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.425996785796281e-05,
+      "loss": 0.8413,
+      "step": 355400
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.4258819928063057e-05,
+      "loss": 0.8637,
+      "step": 355410
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.4257671998163306e-05,
+      "loss": 0.935,
+      "step": 355420
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.425652406826356e-05,
+      "loss": 0.9109,
+      "step": 355430
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.425537613836381e-05,
+      "loss": 0.9071,
+      "step": 355440
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.425422820846406e-05,
+      "loss": 0.851,
+      "step": 355450
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.425308027856432e-05,
+      "loss": 0.8661,
+      "step": 355460
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.425193234866457e-05,
+      "loss": 0.873,
+      "step": 355470
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.4250784418764826e-05,
+      "loss": 0.8118,
+      "step": 355480
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.4249636488865076e-05,
+      "loss": 0.9067,
+      "step": 355490
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.424848855896533e-05,
+      "loss": 0.7945,
+      "step": 355500
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.424734062906558e-05,
+      "loss": 0.8401,
+      "step": 355510
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.424619269916584e-05,
+      "loss": 0.8903,
+      "step": 355520
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.424504476926609e-05,
+      "loss": 0.9546,
+      "step": 355530
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.424389683936634e-05,
+      "loss": 0.905,
+      "step": 355540
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.4242748909466595e-05,
+      "loss": 0.9586,
+      "step": 355550
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.4241600979566845e-05,
+      "loss": 0.8518,
+      "step": 355560
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.4240453049667095e-05,
+      "loss": 0.8117,
+      "step": 355570
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 3.423930511976735e-05,
+      "loss": 0.9973,
+      "step": 355580
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.42381571898676e-05,
+      "loss": 0.9297,
+      "step": 355590
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.423700925996785e-05,
+      "loss": 0.8302,
+      "step": 355600
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.423586133006811e-05,
+      "loss": 0.8702,
+      "step": 355610
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.423471340016836e-05,
+      "loss": 0.9148,
+      "step": 355620
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.4233565470268614e-05,
+      "loss": 1.018,
+      "step": 355630
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.4232417540368864e-05,
+      "loss": 1.0125,
+      "step": 355640
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.423126961046912e-05,
+      "loss": 0.9119,
+      "step": 355650
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.423012168056937e-05,
+      "loss": 0.8062,
+      "step": 355660
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.422897375066962e-05,
+      "loss": 0.9114,
+      "step": 355670
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.422782582076988e-05,
+      "loss": 0.9579,
+      "step": 355680
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.422667789087013e-05,
+      "loss": 0.9426,
+      "step": 355690
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.422552996097038e-05,
+      "loss": 0.8748,
+      "step": 355700
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.4224382031070633e-05,
+      "loss": 0.8792,
+      "step": 355710
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.422323410117088e-05,
+      "loss": 0.8899,
+      "step": 355720
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.422208617127113e-05,
+      "loss": 0.8576,
+      "step": 355730
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.422093824137139e-05,
+      "loss": 0.9993,
+      "step": 355740
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.421979031147164e-05,
+      "loss": 0.9213,
+      "step": 355750
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.4218642381571896e-05,
+      "loss": 0.8889,
+      "step": 355760
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.4217494451672146e-05,
+      "loss": 0.9592,
+      "step": 355770
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.42163465217724e-05,
+      "loss": 0.9399,
+      "step": 355780
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.421519859187265e-05,
+      "loss": 0.8826,
+      "step": 355790
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 3.421405066197291e-05,
+      "loss": 0.9293,
+      "step": 355800
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.421290273207316e-05,
+      "loss": 0.9071,
+      "step": 355810
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.421175480217341e-05,
+      "loss": 0.8127,
+      "step": 355820
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.4210606872273666e-05,
+      "loss": 0.9475,
+      "step": 355830
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.4209458942373915e-05,
+      "loss": 0.8797,
+      "step": 355840
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.4208311012474165e-05,
+      "loss": 0.8733,
+      "step": 355850
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.420716308257442e-05,
+      "loss": 0.8967,
+      "step": 355860
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.420601515267467e-05,
+      "loss": 0.8614,
+      "step": 355870
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.420486722277492e-05,
+      "loss": 0.8737,
+      "step": 355880
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.420371929287518e-05,
+      "loss": 1.058,
+      "step": 355890
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.420257136297543e-05,
+      "loss": 0.8832,
+      "step": 355900
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.4201423433075685e-05,
+      "loss": 0.9384,
+      "step": 355910
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.4200275503175935e-05,
+      "loss": 0.9442,
+      "step": 355920
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.419912757327619e-05,
+      "loss": 0.9218,
+      "step": 355930
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.419797964337644e-05,
+      "loss": 0.9022,
+      "step": 355940
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.41968317134767e-05,
+      "loss": 0.8927,
+      "step": 355950
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.419568378357695e-05,
+      "loss": 0.7665,
+      "step": 355960
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.41945358536772e-05,
+      "loss": 0.8907,
+      "step": 355970
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.4193387923777454e-05,
+      "loss": 0.93,
+      "step": 355980
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.4192239993877704e-05,
+      "loss": 0.828,
+      "step": 355990
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.4191092063977954e-05,
+      "loss": 1.0021,
+      "step": 356000
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.418994413407821e-05,
+      "loss": 0.8704,
+      "step": 356010
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 3.418879620417846e-05,
+      "loss": 0.9053,
+      "step": 356020
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.418764827427871e-05,
+      "loss": 0.9219,
+      "step": 356030
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.418650034437897e-05,
+      "loss": 0.8889,
+      "step": 356040
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.4185352414479217e-05,
+      "loss": 0.8044,
+      "step": 356050
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.418420448457947e-05,
+      "loss": 0.8199,
+      "step": 356060
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.418305655467972e-05,
+      "loss": 0.8906,
+      "step": 356070
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.418190862477998e-05,
+      "loss": 0.8552,
+      "step": 356080
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.418076069488023e-05,
+      "loss": 0.8799,
+      "step": 356090
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.4179612764980486e-05,
+      "loss": 0.9253,
+      "step": 356100
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.4178464835080736e-05,
+      "loss": 0.9704,
+      "step": 356110
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.4177316905180986e-05,
+      "loss": 1.0347,
+      "step": 356120
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.417616897528124e-05,
+      "loss": 0.9001,
+      "step": 356130
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.417502104538149e-05,
+      "loss": 0.882,
+      "step": 356140
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.417387311548174e-05,
+      "loss": 0.9014,
+      "step": 356150
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.4172725185582e-05,
+      "loss": 0.8858,
+      "step": 356160
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.417157725568225e-05,
+      "loss": 0.8758,
+      "step": 356170
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.41704293257825e-05,
+      "loss": 0.8539,
+      "step": 356180
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.4169281395882755e-05,
+      "loss": 0.8928,
+      "step": 356190
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.4168133465983005e-05,
+      "loss": 0.9115,
+      "step": 356200
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.416698553608326e-05,
+      "loss": 1.022,
+      "step": 356210
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.416583760618351e-05,
+      "loss": 0.9747,
+      "step": 356220
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.416468967628377e-05,
+      "loss": 0.8519,
+      "step": 356230
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 3.416354174638402e-05,
+      "loss": 0.9059,
+      "step": 356240
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.4162393816484275e-05,
+      "loss": 0.8568,
+      "step": 356250
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.4161245886584525e-05,
+      "loss": 0.8961,
+      "step": 356260
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.4160097956684774e-05,
+      "loss": 1.0115,
+      "step": 356270
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.415895002678503e-05,
+      "loss": 0.9019,
+      "step": 356280
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.415780209688528e-05,
+      "loss": 0.8295,
+      "step": 356290
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.415665416698553e-05,
+      "loss": 1.0264,
+      "step": 356300
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.415550623708579e-05,
+      "loss": 0.9277,
+      "step": 356310
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.415435830718604e-05,
+      "loss": 0.9314,
+      "step": 356320
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.415321037728629e-05,
+      "loss": 0.9247,
+      "step": 356330
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.4152062447386544e-05,
+      "loss": 0.9132,
+      "step": 356340
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.4150914517486794e-05,
+      "loss": 0.8914,
+      "step": 356350
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.414976658758705e-05,
+      "loss": 0.8123,
+      "step": 356360
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.41486186576873e-05,
+      "loss": 0.9817,
+      "step": 356370
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.414747072778756e-05,
+      "loss": 0.8879,
+      "step": 356380
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.4146322797887807e-05,
+      "loss": 0.9392,
+      "step": 356390
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.4145174867988056e-05,
+      "loss": 0.9062,
+      "step": 356400
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.414402693808831e-05,
+      "loss": 0.7966,
+      "step": 356410
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.414287900818856e-05,
+      "loss": 0.9201,
+      "step": 356420
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.414173107828881e-05,
+      "loss": 0.8008,
+      "step": 356430
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.414058314838907e-05,
+      "loss": 0.9602,
+      "step": 356440
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 3.413943521848932e-05,
+      "loss": 0.8779,
+      "step": 356450
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.413828728858957e-05,
+      "loss": 0.7734,
+      "step": 356460
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.4137139358689826e-05,
+      "loss": 0.8692,
+      "step": 356470
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.4135991428790076e-05,
+      "loss": 0.7609,
+      "step": 356480
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.413484349889033e-05,
+      "loss": 0.9202,
+      "step": 356490
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.413369556899058e-05,
+      "loss": 0.8396,
+      "step": 356500
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.413254763909084e-05,
+      "loss": 0.8597,
+      "step": 356510
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.413139970919109e-05,
+      "loss": 0.867,
+      "step": 356520
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.4130251779291345e-05,
+      "loss": 0.8566,
+      "step": 356530
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.4129103849391595e-05,
+      "loss": 0.8977,
+      "step": 356540
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.4127955919491845e-05,
+      "loss": 0.952,
+      "step": 356550
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.41268079895921e-05,
+      "loss": 0.9163,
+      "step": 356560
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.412566005969235e-05,
+      "loss": 0.9377,
+      "step": 356570
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.41245121297926e-05,
+      "loss": 0.8293,
+      "step": 356580
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.412336419989286e-05,
+      "loss": 0.7918,
+      "step": 356590
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.412221626999311e-05,
+      "loss": 1.0224,
+      "step": 356600
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.412106834009336e-05,
+      "loss": 0.8294,
+      "step": 356610
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.4119920410193614e-05,
+      "loss": 0.9033,
+      "step": 356620
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.4118772480293864e-05,
+      "loss": 0.9607,
+      "step": 356630
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.411762455039412e-05,
+      "loss": 0.9881,
+      "step": 356640
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.411647662049437e-05,
+      "loss": 0.8374,
+      "step": 356650
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.411532869059463e-05,
+      "loss": 0.9016,
+      "step": 356660
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 3.411418076069488e-05,
+      "loss": 0.9632,
+      "step": 356670
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.4113032830795134e-05,
+      "loss": 0.8538,
+      "step": 356680
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.4111884900895383e-05,
+      "loss": 0.9254,
+      "step": 356690
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.411073697099563e-05,
+      "loss": 0.8488,
+      "step": 356700
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.410958904109589e-05,
+      "loss": 0.812,
+      "step": 356710
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.410844111119614e-05,
+      "loss": 0.9057,
+      "step": 356720
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.410729318129639e-05,
+      "loss": 0.883,
+      "step": 356730
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.4106145251396646e-05,
+      "loss": 0.8881,
+      "step": 356740
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.4104997321496896e-05,
+      "loss": 0.9731,
+      "step": 356750
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.4103849391597146e-05,
+      "loss": 0.8815,
+      "step": 356760
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.41027014616974e-05,
+      "loss": 0.876,
+      "step": 356770
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.410155353179765e-05,
+      "loss": 0.8157,
+      "step": 356780
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.410040560189791e-05,
+      "loss": 0.8706,
+      "step": 356790
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.409925767199816e-05,
+      "loss": 0.8879,
+      "step": 356800
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.4098109742098416e-05,
+      "loss": 0.8753,
+      "step": 356810
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.4096961812198665e-05,
+      "loss": 0.8555,
+      "step": 356820
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.409581388229892e-05,
+      "loss": 0.8854,
+      "step": 356830
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.409466595239917e-05,
+      "loss": 0.8744,
+      "step": 356840
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.409351802249942e-05,
+      "loss": 0.8384,
+      "step": 356850
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.409237009259968e-05,
+      "loss": 0.8726,
+      "step": 356860
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.409122216269993e-05,
+      "loss": 0.8965,
+      "step": 356870
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.409007423280018e-05,
+      "loss": 0.9321,
+      "step": 356880
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 3.4088926302900435e-05,
+      "loss": 0.9585,
+      "step": 356890
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.4087778373000685e-05,
+      "loss": 0.8951,
+      "step": 356900
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.4086630443100934e-05,
+      "loss": 0.9258,
+      "step": 356910
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.408548251320119e-05,
+      "loss": 0.8079,
+      "step": 356920
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.408433458330144e-05,
+      "loss": 0.897,
+      "step": 356930
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.40831866534017e-05,
+      "loss": 0.9133,
+      "step": 356940
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.408203872350195e-05,
+      "loss": 0.8606,
+      "step": 356950
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.4080890793602204e-05,
+      "loss": 0.976,
+      "step": 356960
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.4079742863702454e-05,
+      "loss": 0.8121,
+      "step": 356970
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.407859493380271e-05,
+      "loss": 0.8829,
+      "step": 356980
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.407744700390296e-05,
+      "loss": 0.7749,
+      "step": 356990
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.407629907400321e-05,
+      "loss": 0.8965,
+      "step": 357000
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.407515114410347e-05,
+      "loss": 0.9548,
+      "step": 357010
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.407400321420372e-05,
+      "loss": 0.905,
+      "step": 357020
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.4072855284303967e-05,
+      "loss": 0.9493,
+      "step": 357030
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.4071707354404216e-05,
+      "loss": 0.9721,
+      "step": 357040
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.407055942450447e-05,
+      "loss": 0.9232,
+      "step": 357050
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.406941149460472e-05,
+      "loss": 0.8134,
+      "step": 357060
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.406826356470498e-05,
+      "loss": 0.8553,
+      "step": 357070
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.406711563480523e-05,
+      "loss": 1.0561,
+      "step": 357080
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.4065967704905486e-05,
+      "loss": 0.8762,
+      "step": 357090
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.4064819775005736e-05,
+      "loss": 0.9178,
+      "step": 357100
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 3.406367184510599e-05,
+      "loss": 0.9203,
+      "step": 357110
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.406252391520624e-05,
+      "loss": 0.8684,
+      "step": 357120
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.406137598530649e-05,
+      "loss": 0.8712,
+      "step": 357130
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.406022805540675e-05,
+      "loss": 0.9233,
+      "step": 357140
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.4059080125507e-05,
+      "loss": 0.9112,
+      "step": 357150
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.405793219560725e-05,
+      "loss": 0.8674,
+      "step": 357160
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.4056784265707505e-05,
+      "loss": 0.8907,
+      "step": 357170
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.4055636335807755e-05,
+      "loss": 0.8839,
+      "step": 357180
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.4054488405908005e-05,
+      "loss": 0.8727,
+      "step": 357190
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.405334047600826e-05,
+      "loss": 0.8772,
+      "step": 357200
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.405219254610851e-05,
+      "loss": 0.9511,
+      "step": 357210
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.405104461620877e-05,
+      "loss": 0.9468,
+      "step": 357220
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.404989668630902e-05,
+      "loss": 0.9557,
+      "step": 357230
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.4048748756409275e-05,
+      "loss": 0.9173,
+      "step": 357240
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.4047600826509524e-05,
+      "loss": 0.8651,
+      "step": 357250
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.404645289660978e-05,
+      "loss": 0.9742,
+      "step": 357260
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.404530496671003e-05,
+      "loss": 0.9012,
+      "step": 357270
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.404415703681028e-05,
+      "loss": 0.8527,
+      "step": 357280
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.404300910691054e-05,
+      "loss": 0.8689,
+      "step": 357290
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.404186117701079e-05,
+      "loss": 1.0212,
+      "step": 357300
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.404071324711104e-05,
+      "loss": 0.8633,
+      "step": 357310
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.4039565317211294e-05,
+      "loss": 0.9462,
+      "step": 357320
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 3.4038417387311544e-05,
+      "loss": 0.9016,
+      "step": 357330
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.403726945741179e-05,
+      "loss": 0.9095,
+      "step": 357340
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.403612152751205e-05,
+      "loss": 0.8804,
+      "step": 357350
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.40349735976123e-05,
+      "loss": 0.8882,
+      "step": 357360
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.4033825667712556e-05,
+      "loss": 0.9004,
+      "step": 357370
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.4032677737812806e-05,
+      "loss": 0.8859,
+      "step": 357380
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.403152980791306e-05,
+      "loss": 0.9198,
+      "step": 357390
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.403038187801331e-05,
+      "loss": 0.8291,
+      "step": 357400
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.402923394811357e-05,
+      "loss": 0.8689,
+      "step": 357410
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.402808601821382e-05,
+      "loss": 0.9189,
+      "step": 357420
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.402693808831407e-05,
+      "loss": 0.9467,
+      "step": 357430
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.4025790158414326e-05,
+      "loss": 0.9194,
+      "step": 357440
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.4024642228514576e-05,
+      "loss": 0.8587,
+      "step": 357450
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.4023494298614826e-05,
+      "loss": 0.8618,
+      "step": 357460
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.402234636871508e-05,
+      "loss": 0.7774,
+      "step": 357470
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.402119843881533e-05,
+      "loss": 0.9548,
+      "step": 357480
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.402005050891558e-05,
+      "loss": 0.8417,
+      "step": 357490
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.401890257901584e-05,
+      "loss": 0.8787,
+      "step": 357500
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.401775464911609e-05,
+      "loss": 0.8869,
+      "step": 357510
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.4016606719216345e-05,
+      "loss": 0.8631,
+      "step": 357520
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.4015458789316595e-05,
+      "loss": 0.8676,
+      "step": 357530
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 3.401431085941685e-05,
+      "loss": 0.8091,
+      "step": 357540
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.40131629295171e-05,
+      "loss": 0.8884,
+      "step": 357550
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.401201499961736e-05,
+      "loss": 0.8556,
+      "step": 357560
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.401086706971761e-05,
+      "loss": 0.9024,
+      "step": 357570
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.400971913981786e-05,
+      "loss": 0.9495,
+      "step": 357580
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.4008571209918114e-05,
+      "loss": 0.9437,
+      "step": 357590
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.4007423280018364e-05,
+      "loss": 0.926,
+      "step": 357600
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.4006275350118614e-05,
+      "loss": 0.9315,
+      "step": 357610
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.400512742021887e-05,
+      "loss": 0.9135,
+      "step": 357620
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.400397949031912e-05,
+      "loss": 0.8277,
+      "step": 357630
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.400283156041937e-05,
+      "loss": 0.8692,
+      "step": 357640
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.400168363051963e-05,
+      "loss": 0.8914,
+      "step": 357650
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.400053570061988e-05,
+      "loss": 0.9926,
+      "step": 357660
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.3999387770720133e-05,
+      "loss": 0.8682,
+      "step": 357670
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.399823984082038e-05,
+      "loss": 0.8741,
+      "step": 357680
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.399709191092064e-05,
+      "loss": 0.9504,
+      "step": 357690
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.399594398102089e-05,
+      "loss": 0.8756,
+      "step": 357700
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.3994796051121146e-05,
+      "loss": 0.993,
+      "step": 357710
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.3993648121221396e-05,
+      "loss": 0.9085,
+      "step": 357720
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.3992500191321646e-05,
+      "loss": 0.8286,
+      "step": 357730
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.3991352261421896e-05,
+      "loss": 0.9826,
+      "step": 357740
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.399020433152215e-05,
+      "loss": 0.918,
+      "step": 357750
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.39890564016224e-05,
+      "loss": 0.8841,
+      "step": 357760
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.398790847172265e-05,
+      "loss": 0.905,
+      "step": 357770
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.398676054182291e-05,
+      "loss": 0.8751,
+      "step": 357780
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.398561261192316e-05,
+      "loss": 0.8843,
+      "step": 357790
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.3984464682023415e-05,
+      "loss": 0.8737,
+      "step": 357800
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.3983316752123665e-05,
+      "loss": 0.9516,
+      "step": 357810
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.398216882222392e-05,
+      "loss": 0.9101,
+      "step": 357820
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.398102089232417e-05,
+      "loss": 0.8952,
+      "step": 357830
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.397987296242443e-05,
+      "loss": 0.84,
+      "step": 357840
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.397872503252468e-05,
+      "loss": 0.891,
+      "step": 357850
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.397757710262493e-05,
+      "loss": 0.9268,
+      "step": 357860
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.3976429172725185e-05,
+      "loss": 0.8647,
+      "step": 357870
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.3975281242825435e-05,
+      "loss": 0.8927,
+      "step": 357880
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.3974133312925684e-05,
+      "loss": 0.945,
+      "step": 357890
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.397298538302594e-05,
+      "loss": 0.8806,
+      "step": 357900
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.397183745312619e-05,
+      "loss": 0.9205,
+      "step": 357910
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.397068952322644e-05,
+      "loss": 0.9433,
+      "step": 357920
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.39695415933267e-05,
+      "loss": 0.8711,
+      "step": 357930
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.396839366342695e-05,
+      "loss": 0.8627,
+      "step": 357940
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.3967245733527204e-05,
+      "loss": 0.8925,
+      "step": 357950
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.3966097803627454e-05,
+      "loss": 0.8884,
+      "step": 357960
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.396494987372771e-05,
+      "loss": 0.8147,
+      "step": 357970
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 3.396380194382796e-05,
+      "loss": 0.8873,
+      "step": 357980
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.396265401392822e-05,
+      "loss": 0.9203,
+      "step": 357990
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.396150608402847e-05,
+      "loss": 0.9826,
+      "step": 358000
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.3960358154128717e-05,
+      "loss": 0.9669,
+      "step": 358010
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.395921022422897e-05,
+      "loss": 0.9393,
+      "step": 358020
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.395806229432922e-05,
+      "loss": 0.8558,
+      "step": 358030
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.395691436442947e-05,
+      "loss": 0.8861,
+      "step": 358040
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.395576643452973e-05,
+      "loss": 0.9328,
+      "step": 358050
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.395461850462998e-05,
+      "loss": 0.9369,
+      "step": 358060
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.395347057473023e-05,
+      "loss": 0.9623,
+      "step": 358070
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.3952322644830486e-05,
+      "loss": 0.9471,
+      "step": 358080
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.3951174714930736e-05,
+      "loss": 0.8258,
+      "step": 358090
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.395002678503099e-05,
+      "loss": 0.9096,
+      "step": 358100
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.394887885513124e-05,
+      "loss": 0.9505,
+      "step": 358110
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.39477309252315e-05,
+      "loss": 0.8829,
+      "step": 358120
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.394658299533175e-05,
+      "loss": 0.9168,
+      "step": 358130
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.3945435065432005e-05,
+      "loss": 0.9956,
+      "step": 358140
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.3944287135532255e-05,
+      "loss": 0.9067,
+      "step": 358150
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.3943139205632505e-05,
+      "loss": 0.9318,
+      "step": 358160
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.394199127573276e-05,
+      "loss": 0.9055,
+      "step": 358170
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.394084334583301e-05,
+      "loss": 0.8611,
+      "step": 358180
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.393969541593326e-05,
+      "loss": 0.8446,
+      "step": 358190
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 3.393854748603352e-05,
+      "loss": 0.8576,
+      "step": 358200
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.393739955613377e-05,
+      "loss": 0.841,
+      "step": 358210
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.393625162623402e-05,
+      "loss": 0.9021,
+      "step": 358220
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.3935103696334274e-05,
+      "loss": 0.8641,
+      "step": 358230
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.3933955766434524e-05,
+      "loss": 0.8786,
+      "step": 358240
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.393280783653478e-05,
+      "loss": 0.9185,
+      "step": 358250
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.393165990663503e-05,
+      "loss": 0.9083,
+      "step": 358260
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.393051197673529e-05,
+      "loss": 0.9852,
+      "step": 358270
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.392936404683554e-05,
+      "loss": 0.9442,
+      "step": 358280
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.3928216116935794e-05,
+      "loss": 0.9566,
+      "step": 358290
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.3927068187036044e-05,
+      "loss": 0.8671,
+      "step": 358300
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.3925920257136294e-05,
+      "loss": 0.9437,
+      "step": 358310
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.392477232723655e-05,
+      "loss": 0.8785,
+      "step": 358320
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.39236243973368e-05,
+      "loss": 0.8883,
+      "step": 358330
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.392247646743705e-05,
+      "loss": 0.9977,
+      "step": 358340
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.3921328537537306e-05,
+      "loss": 0.8677,
+      "step": 358350
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.3920180607637556e-05,
+      "loss": 0.8735,
+      "step": 358360
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.3919032677737806e-05,
+      "loss": 0.8782,
+      "step": 358370
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.391788474783806e-05,
+      "loss": 0.8738,
+      "step": 358380
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.391673681793831e-05,
+      "loss": 0.8897,
+      "step": 358390
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.391558888803857e-05,
+      "loss": 0.8831,
+      "step": 358400
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.391444095813882e-05,
+      "loss": 0.8749,
+      "step": 358410
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 3.3913293028239076e-05,
+      "loss": 0.9112,
+      "step": 358420
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.3912145098339326e-05,
+      "loss": 0.7984,
+      "step": 358430
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.3910997168439576e-05,
+      "loss": 0.9852,
+      "step": 358440
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.390984923853983e-05,
+      "loss": 0.7978,
+      "step": 358450
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.390870130864008e-05,
+      "loss": 0.798,
+      "step": 358460
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.390755337874033e-05,
+      "loss": 0.8456,
+      "step": 358470
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.390640544884059e-05,
+      "loss": 0.8319,
+      "step": 358480
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.390525751894084e-05,
+      "loss": 0.9066,
+      "step": 358490
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.390410958904109e-05,
+      "loss": 0.8542,
+      "step": 358500
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.3902961659141345e-05,
+      "loss": 0.8279,
+      "step": 358510
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.3901813729241595e-05,
+      "loss": 0.8839,
+      "step": 358520
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.390066579934185e-05,
+      "loss": 0.9202,
+      "step": 358530
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.38995178694421e-05,
+      "loss": 0.9672,
+      "step": 358540
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.389836993954236e-05,
+      "loss": 0.9463,
+      "step": 358550
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.389722200964261e-05,
+      "loss": 0.9248,
+      "step": 358560
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.3896074079742864e-05,
+      "loss": 0.8968,
+      "step": 358570
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.3894926149843114e-05,
+      "loss": 0.865,
+      "step": 358580
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.3893778219943364e-05,
+      "loss": 0.8834,
+      "step": 358590
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.389263029004362e-05,
+      "loss": 0.9754,
+      "step": 358600
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.389148236014387e-05,
+      "loss": 0.9319,
+      "step": 358610
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.389033443024412e-05,
+      "loss": 0.7862,
+      "step": 358620
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 3.388918650034438e-05,
+      "loss": 0.8601,
+      "step": 358630
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.388803857044463e-05,
+      "loss": 1.0208,
+      "step": 358640
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.388689064054488e-05,
+      "loss": 0.9493,
+      "step": 358650
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.388574271064513e-05,
+      "loss": 0.8985,
+      "step": 358660
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.388459478074538e-05,
+      "loss": 0.9302,
+      "step": 358670
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.388344685084564e-05,
+      "loss": 0.931,
+      "step": 358680
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.388229892094589e-05,
+      "loss": 0.8464,
+      "step": 358690
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.3881150991046146e-05,
+      "loss": 0.8089,
+      "step": 358700
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.3880003061146396e-05,
+      "loss": 0.928,
+      "step": 358710
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.387885513124665e-05,
+      "loss": 0.9515,
+      "step": 358720
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.38777072013469e-05,
+      "loss": 0.8759,
+      "step": 358730
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.387655927144715e-05,
+      "loss": 0.911,
+      "step": 358740
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.387541134154741e-05,
+      "loss": 0.8735,
+      "step": 358750
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.387426341164766e-05,
+      "loss": 0.8675,
+      "step": 358760
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.387311548174791e-05,
+      "loss": 0.9128,
+      "step": 358770
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.3871967551848165e-05,
+      "loss": 0.9188,
+      "step": 358780
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.3870819621948415e-05,
+      "loss": 0.9593,
+      "step": 358790
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.3869671692048665e-05,
+      "loss": 0.8759,
+      "step": 358800
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.386852376214892e-05,
+      "loss": 0.8301,
+      "step": 358810
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.386737583224917e-05,
+      "loss": 0.8855,
+      "step": 358820
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.386622790234943e-05,
+      "loss": 0.9109,
+      "step": 358830
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.386507997244968e-05,
+      "loss": 0.9106,
+      "step": 358840
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 3.3863932042549935e-05,
+      "loss": 0.9158,
+      "step": 358850
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.3862784112650185e-05,
+      "loss": 0.9016,
+      "step": 358860
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.386163618275044e-05,
+      "loss": 0.8915,
+      "step": 358870
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.386048825285069e-05,
+      "loss": 0.9914,
+      "step": 358880
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.385934032295094e-05,
+      "loss": 0.8957,
+      "step": 358890
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.38581923930512e-05,
+      "loss": 0.8892,
+      "step": 358900
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.385704446315145e-05,
+      "loss": 0.8764,
+      "step": 358910
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.38558965332517e-05,
+      "loss": 0.9266,
+      "step": 358920
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.3854748603351954e-05,
+      "loss": 1.0399,
+      "step": 358930
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.3853600673452204e-05,
+      "loss": 0.7811,
+      "step": 358940
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.3852452743552454e-05,
+      "loss": 0.9352,
+      "step": 358950
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.385130481365271e-05,
+      "loss": 0.8228,
+      "step": 358960
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.385015688375296e-05,
+      "loss": 0.9146,
+      "step": 358970
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.384900895385322e-05,
+      "loss": 0.8774,
+      "step": 358980
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.3847861023953467e-05,
+      "loss": 0.935,
+      "step": 358990
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.384671309405372e-05,
+      "loss": 0.826,
+      "step": 359000
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.384556516415397e-05,
+      "loss": 0.8692,
+      "step": 359010
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.384441723425423e-05,
+      "loss": 0.8583,
+      "step": 359020
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.384326930435448e-05,
+      "loss": 0.9607,
+      "step": 359030
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.384212137445473e-05,
+      "loss": 0.9308,
+      "step": 359040
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.3840973444554986e-05,
+      "loss": 0.8947,
+      "step": 359050
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.3839825514655236e-05,
+      "loss": 0.7721,
+      "step": 359060
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 3.3838677584755486e-05,
+      "loss": 0.863,
+      "step": 359070
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.383752965485574e-05,
+      "loss": 0.7932,
+      "step": 359080
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.383638172495599e-05,
+      "loss": 0.8826,
+      "step": 359090
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.383523379505624e-05,
+      "loss": 0.8818,
+      "step": 359100
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.38340858651565e-05,
+      "loss": 0.9637,
+      "step": 359110
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.383293793525675e-05,
+      "loss": 0.9215,
+      "step": 359120
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.3831790005357005e-05,
+      "loss": 0.848,
+      "step": 359130
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.3830642075457255e-05,
+      "loss": 0.8554,
+      "step": 359140
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.382949414555751e-05,
+      "loss": 0.7844,
+      "step": 359150
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.382834621565776e-05,
+      "loss": 0.985,
+      "step": 359160
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.382719828575801e-05,
+      "loss": 0.8835,
+      "step": 359170
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.382605035585827e-05,
+      "loss": 0.9127,
+      "step": 359180
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.382490242595852e-05,
+      "loss": 0.9639,
+      "step": 359190
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.382375449605877e-05,
+      "loss": 0.8923,
+      "step": 359200
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.3822606566159024e-05,
+      "loss": 0.8596,
+      "step": 359210
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.3821458636259274e-05,
+      "loss": 0.8574,
+      "step": 359220
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.3820310706359524e-05,
+      "loss": 0.9168,
+      "step": 359230
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.381916277645978e-05,
+      "loss": 0.8691,
+      "step": 359240
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.381801484656003e-05,
+      "loss": 0.8667,
+      "step": 359250
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.381686691666029e-05,
+      "loss": 0.9105,
+      "step": 359260
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.381571898676054e-05,
+      "loss": 0.8928,
+      "step": 359270
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.3814571056860794e-05,
+      "loss": 0.9566,
+      "step": 359280
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 3.3813423126961044e-05,
+      "loss": 1.011,
+      "step": 359290
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.38122751970613e-05,
+      "loss": 0.8653,
+      "step": 359300
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.381112726716155e-05,
+      "loss": 0.8632,
+      "step": 359310
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.38099793372618e-05,
+      "loss": 0.9545,
+      "step": 359320
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.3808831407362056e-05,
+      "loss": 0.9355,
+      "step": 359330
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.3807683477462306e-05,
+      "loss": 0.9133,
+      "step": 359340
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.3806535547562556e-05,
+      "loss": 0.8252,
+      "step": 359350
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.380538761766281e-05,
+      "loss": 0.9706,
+      "step": 359360
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.380423968776306e-05,
+      "loss": 0.8181,
+      "step": 359370
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.380309175786331e-05,
+      "loss": 0.9789,
+      "step": 359380
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.380194382796357e-05,
+      "loss": 0.9034,
+      "step": 359390
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.380079589806382e-05,
+      "loss": 0.8103,
+      "step": 359400
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.3799647968164076e-05,
+      "loss": 0.8959,
+      "step": 359410
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.3798500038264326e-05,
+      "loss": 0.8779,
+      "step": 359420
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.379735210836458e-05,
+      "loss": 0.9253,
+      "step": 359430
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.379620417846483e-05,
+      "loss": 0.8494,
+      "step": 359440
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.379505624856509e-05,
+      "loss": 0.9633,
+      "step": 359450
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.379390831866534e-05,
+      "loss": 0.9542,
+      "step": 359460
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.379276038876559e-05,
+      "loss": 0.9594,
+      "step": 359470
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.3791612458865845e-05,
+      "loss": 0.8459,
+      "step": 359480
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.3790464528966095e-05,
+      "loss": 0.9102,
+      "step": 359490
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.3789316599066345e-05,
+      "loss": 0.8393,
+      "step": 359500
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 3.37881686691666e-05,
+      "loss": 0.8073,
+      "step": 359510
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.378702073926685e-05,
+      "loss": 0.9962,
+      "step": 359520
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.37858728093671e-05,
+      "loss": 0.9661,
+      "step": 359530
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.378472487946736e-05,
+      "loss": 0.9536,
+      "step": 359540
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.378357694956761e-05,
+      "loss": 0.8518,
+      "step": 359550
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.3782429019667864e-05,
+      "loss": 0.867,
+      "step": 359560
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.3781281089768114e-05,
+      "loss": 0.9206,
+      "step": 359570
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.378013315986837e-05,
+      "loss": 0.8335,
+      "step": 359580
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.377898522996862e-05,
+      "loss": 0.8996,
+      "step": 359590
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.377783730006888e-05,
+      "loss": 0.8013,
+      "step": 359600
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.377668937016913e-05,
+      "loss": 0.8189,
+      "step": 359610
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.377554144026938e-05,
+      "loss": 0.8734,
+      "step": 359620
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.3774393510369633e-05,
+      "loss": 0.8647,
+      "step": 359630
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.377324558046988e-05,
+      "loss": 0.957,
+      "step": 359640
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.377209765057013e-05,
+      "loss": 0.831,
+      "step": 359650
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.377094972067039e-05,
+      "loss": 0.8686,
+      "step": 359660
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.376980179077064e-05,
+      "loss": 0.9379,
+      "step": 359670
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.376865386087089e-05,
+      "loss": 0.9816,
+      "step": 359680
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.3767505930971146e-05,
+      "loss": 0.99,
+      "step": 359690
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.3766358001071396e-05,
+      "loss": 0.8674,
+      "step": 359700
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.376521007117165e-05,
+      "loss": 0.958,
+      "step": 359710
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 3.37640621412719e-05,
+      "loss": 0.949,
+      "step": 359720
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.376291421137216e-05,
+      "loss": 0.8566,
+      "step": 359730
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.376176628147241e-05,
+      "loss": 0.9219,
+      "step": 359740
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.3760618351572666e-05,
+      "loss": 0.8796,
+      "step": 359750
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.3759470421672915e-05,
+      "loss": 0.8974,
+      "step": 359760
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.3758322491773165e-05,
+      "loss": 0.9105,
+      "step": 359770
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.3757174561873415e-05,
+      "loss": 0.9069,
+      "step": 359780
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.375602663197367e-05,
+      "loss": 0.8951,
+      "step": 359790
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.375487870207392e-05,
+      "loss": 0.8123,
+      "step": 359800
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.375373077217417e-05,
+      "loss": 0.8364,
+      "step": 359810
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.375258284227443e-05,
+      "loss": 0.9349,
+      "step": 359820
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.375143491237468e-05,
+      "loss": 0.7386,
+      "step": 359830
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.3750286982474935e-05,
+      "loss": 0.8998,
+      "step": 359840
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.3749139052575184e-05,
+      "loss": 0.9289,
+      "step": 359850
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.374799112267544e-05,
+      "loss": 0.8285,
+      "step": 359860
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.374684319277569e-05,
+      "loss": 0.8814,
+      "step": 359870
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.374569526287595e-05,
+      "loss": 0.8344,
+      "step": 359880
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.37445473329762e-05,
+      "loss": 0.8996,
+      "step": 359890
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.374339940307645e-05,
+      "loss": 0.8232,
+      "step": 359900
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.3742251473176704e-05,
+      "loss": 0.9528,
+      "step": 359910
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.3741103543276954e-05,
+      "loss": 0.9374,
+      "step": 359920
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.3739955613377204e-05,
+      "loss": 0.8444,
+      "step": 359930
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 3.373880768347746e-05,
+      "loss": 0.9169,
+      "step": 359940
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.373765975357771e-05,
+      "loss": 0.9191,
+      "step": 359950
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.373651182367796e-05,
+      "loss": 1.0104,
+      "step": 359960
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.3735363893778217e-05,
+      "loss": 0.9361,
+      "step": 359970
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.3734215963878466e-05,
+      "loss": 0.9594,
+      "step": 359980
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.373306803397872e-05,
+      "loss": 0.9007,
+      "step": 359990
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.373192010407897e-05,
+      "loss": 1.0357,
+      "step": 360000
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.373077217417923e-05,
+      "loss": 0.8602,
+      "step": 360010
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.372962424427948e-05,
+      "loss": 0.8872,
+      "step": 360020
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.3728476314379736e-05,
+      "loss": 0.8474,
+      "step": 360030
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.3727328384479986e-05,
+      "loss": 0.8542,
+      "step": 360040
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.3726180454580236e-05,
+      "loss": 0.9139,
+      "step": 360050
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.372503252468049e-05,
+      "loss": 0.8546,
+      "step": 360060
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.372388459478074e-05,
+      "loss": 0.8704,
+      "step": 360070
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.372273666488099e-05,
+      "loss": 0.9056,
+      "step": 360080
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.372158873498125e-05,
+      "loss": 0.8927,
+      "step": 360090
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.37204408050815e-05,
+      "loss": 0.8652,
+      "step": 360100
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.371929287518175e-05,
+      "loss": 0.9745,
+      "step": 360110
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.3718144945282005e-05,
+      "loss": 0.9242,
+      "step": 360120
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.3716997015382255e-05,
+      "loss": 0.9153,
+      "step": 360130
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.371584908548251e-05,
+      "loss": 0.9013,
+      "step": 360140
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.371470115558276e-05,
+      "loss": 0.8983,
+      "step": 360150
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.371355322568302e-05,
+      "loss": 0.8771,
+      "step": 360160
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.371240529578327e-05,
+      "loss": 0.8639,
+      "step": 360170
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.3711257365883524e-05,
+      "loss": 0.9094,
+      "step": 360180
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.3710109435983774e-05,
+      "loss": 0.9913,
+      "step": 360190
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.3708961506084024e-05,
+      "loss": 0.9746,
+      "step": 360200
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.370781357618428e-05,
+      "loss": 0.9746,
+      "step": 360210
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.370666564628453e-05,
+      "loss": 0.9044,
+      "step": 360220
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.370551771638478e-05,
+      "loss": 0.9332,
+      "step": 360230
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.370436978648504e-05,
+      "loss": 0.9086,
+      "step": 360240
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.370322185658529e-05,
+      "loss": 0.8754,
+      "step": 360250
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.370207392668554e-05,
+      "loss": 0.8986,
+      "step": 360260
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.3700925996785794e-05,
+      "loss": 0.8987,
+      "step": 360270
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.369977806688604e-05,
+      "loss": 0.9487,
+      "step": 360280
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.36986301369863e-05,
+      "loss": 0.9003,
+      "step": 360290
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.369748220708655e-05,
+      "loss": 0.8605,
+      "step": 360300
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.3696334277186806e-05,
+      "loss": 1.0276,
+      "step": 360310
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.3695186347287056e-05,
+      "loss": 0.9083,
+      "step": 360320
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.369403841738731e-05,
+      "loss": 0.8765,
+      "step": 360330
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.369289048748756e-05,
+      "loss": 0.8772,
+      "step": 360340
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.369174255758781e-05,
+      "loss": 0.9917,
+      "step": 360350
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.369059462768807e-05,
+      "loss": 0.9287,
+      "step": 360360
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.368944669778832e-05,
+      "loss": 0.8608,
+      "step": 360370
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 3.368829876788857e-05,
+      "loss": 0.9843,
+      "step": 360380
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.3687150837988826e-05,
+      "loss": 0.9797,
+      "step": 360390
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.3686002908089075e-05,
+      "loss": 0.9358,
+      "step": 360400
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.3684854978189325e-05,
+      "loss": 0.7583,
+      "step": 360410
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.368370704828958e-05,
+      "loss": 0.8329,
+      "step": 360420
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.368255911838983e-05,
+      "loss": 0.9022,
+      "step": 360430
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.368141118849009e-05,
+      "loss": 0.976,
+      "step": 360440
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.368026325859034e-05,
+      "loss": 0.8927,
+      "step": 360450
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.3679115328690595e-05,
+      "loss": 0.8342,
+      "step": 360460
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.3677967398790845e-05,
+      "loss": 0.907,
+      "step": 360470
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.3676819468891095e-05,
+      "loss": 1.0393,
+      "step": 360480
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.367567153899135e-05,
+      "loss": 0.8798,
+      "step": 360490
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.36745236090916e-05,
+      "loss": 0.8996,
+      "step": 360500
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.367337567919185e-05,
+      "loss": 0.9479,
+      "step": 360510
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.367222774929211e-05,
+      "loss": 0.8826,
+      "step": 360520
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.367107981939236e-05,
+      "loss": 0.911,
+      "step": 360530
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.366993188949261e-05,
+      "loss": 0.8632,
+      "step": 360540
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.3668783959592864e-05,
+      "loss": 0.9425,
+      "step": 360550
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.3667636029693114e-05,
+      "loss": 0.909,
+      "step": 360560
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.366648809979337e-05,
+      "loss": 0.9472,
+      "step": 360570
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.366534016989362e-05,
+      "loss": 0.869,
+      "step": 360580
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.366419223999388e-05,
+      "loss": 0.8432,
+      "step": 360590
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 3.366304431009413e-05,
+      "loss": 0.9578,
+      "step": 360600
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.3661896380194383e-05,
+      "loss": 0.8848,
+      "step": 360610
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.366074845029463e-05,
+      "loss": 0.8203,
+      "step": 360620
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.365960052039488e-05,
+      "loss": 0.91,
+      "step": 360630
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.365845259049514e-05,
+      "loss": 0.9329,
+      "step": 360640
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.365730466059539e-05,
+      "loss": 0.8601,
+      "step": 360650
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.365615673069564e-05,
+      "loss": 0.8296,
+      "step": 360660
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.3655008800795896e-05,
+      "loss": 0.9806,
+      "step": 360670
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.3653860870896146e-05,
+      "loss": 0.9007,
+      "step": 360680
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.3652712940996396e-05,
+      "loss": 0.8373,
+      "step": 360690
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.365156501109665e-05,
+      "loss": 1.0056,
+      "step": 360700
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.36504170811969e-05,
+      "loss": 0.8568,
+      "step": 360710
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.364926915129716e-05,
+      "loss": 0.9827,
+      "step": 360720
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.364812122139741e-05,
+      "loss": 0.902,
+      "step": 360730
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.3646973291497665e-05,
+      "loss": 0.854,
+      "step": 360740
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.3645825361597915e-05,
+      "loss": 0.771,
+      "step": 360750
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.364467743169817e-05,
+      "loss": 0.7857,
+      "step": 360760
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.364352950179842e-05,
+      "loss": 0.901,
+      "step": 360770
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.364238157189867e-05,
+      "loss": 0.9395,
+      "step": 360780
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.364123364199893e-05,
+      "loss": 0.8387,
+      "step": 360790
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.364008571209918e-05,
+      "loss": 0.9913,
+      "step": 360800
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 3.363893778219943e-05,
+      "loss": 0.9,
+      "step": 360810
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.3637789852299685e-05,
+      "loss": 0.99,
+      "step": 360820
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.3636641922399934e-05,
+      "loss": 0.8221,
+      "step": 360830
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.3635493992500184e-05,
+      "loss": 0.8534,
+      "step": 360840
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.363434606260044e-05,
+      "loss": 1.004,
+      "step": 360850
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.363319813270069e-05,
+      "loss": 0.9389,
+      "step": 360860
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.363205020280095e-05,
+      "loss": 0.8386,
+      "step": 360870
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.36309022729012e-05,
+      "loss": 0.8429,
+      "step": 360880
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.3629754343001454e-05,
+      "loss": 0.8745,
+      "step": 360890
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.3628606413101704e-05,
+      "loss": 0.9012,
+      "step": 360900
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.362745848320196e-05,
+      "loss": 0.8659,
+      "step": 360910
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.362631055330221e-05,
+      "loss": 0.9733,
+      "step": 360920
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.362516262340246e-05,
+      "loss": 0.9172,
+      "step": 360930
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.362401469350272e-05,
+      "loss": 0.9144,
+      "step": 360940
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.3622866763602967e-05,
+      "loss": 1.0274,
+      "step": 360950
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.3621718833703216e-05,
+      "loss": 0.9141,
+      "step": 360960
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.362057090380347e-05,
+      "loss": 0.9301,
+      "step": 360970
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.361942297390372e-05,
+      "loss": 0.943,
+      "step": 360980
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.361827504400397e-05,
+      "loss": 0.943,
+      "step": 360990
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.361712711410423e-05,
+      "loss": 0.798,
+      "step": 361000
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.361597918420448e-05,
+      "loss": 0.9203,
+      "step": 361010
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.3614831254304736e-05,
+      "loss": 0.9484,
+      "step": 361020
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 3.3613683324404986e-05,
+      "loss": 0.8576,
+      "step": 361030
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.361253539450524e-05,
+      "loss": 0.9537,
+      "step": 361040
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.361138746460549e-05,
+      "loss": 0.8902,
+      "step": 361050
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.361023953470575e-05,
+      "loss": 0.8594,
+      "step": 361060
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.3609091604806e-05,
+      "loss": 0.8772,
+      "step": 361070
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.360794367490625e-05,
+      "loss": 0.8623,
+      "step": 361080
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.3606795745006505e-05,
+      "loss": 0.934,
+      "step": 361090
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.3605647815106755e-05,
+      "loss": 0.8883,
+      "step": 361100
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.3604499885207005e-05,
+      "loss": 0.8298,
+      "step": 361110
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.360335195530726e-05,
+      "loss": 0.9816,
+      "step": 361120
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.360220402540751e-05,
+      "loss": 0.9013,
+      "step": 361130
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.360105609550776e-05,
+      "loss": 0.8771,
+      "step": 361140
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.359990816560802e-05,
+      "loss": 0.9182,
+      "step": 361150
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.359876023570827e-05,
+      "loss": 0.8952,
+      "step": 361160
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.3597612305808524e-05,
+      "loss": 0.9415,
+      "step": 361170
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.3596464375908774e-05,
+      "loss": 0.8877,
+      "step": 361180
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.359531644600903e-05,
+      "loss": 0.934,
+      "step": 361190
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.359416851610928e-05,
+      "loss": 0.9121,
+      "step": 361200
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.359302058620953e-05,
+      "loss": 0.8584,
+      "step": 361210
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.359187265630979e-05,
+      "loss": 0.8935,
+      "step": 361220
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.359072472641004e-05,
+      "loss": 0.8386,
+      "step": 361230
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.358957679651029e-05,
+      "loss": 0.8688,
+      "step": 361240
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 3.3588428866610544e-05,
+      "loss": 0.8291,
+      "step": 361250
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.358728093671079e-05,
+      "loss": 0.9286,
+      "step": 361260
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.358613300681104e-05,
+      "loss": 1.0227,
+      "step": 361270
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.35849850769113e-05,
+      "loss": 0.8961,
+      "step": 361280
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.358383714701155e-05,
+      "loss": 0.8196,
+      "step": 361290
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.3582689217111806e-05,
+      "loss": 0.9085,
+      "step": 361300
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.3581541287212056e-05,
+      "loss": 0.8432,
+      "step": 361310
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.358039335731231e-05,
+      "loss": 0.9824,
+      "step": 361320
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.357924542741256e-05,
+      "loss": 0.9027,
+      "step": 361330
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.357809749751282e-05,
+      "loss": 0.857,
+      "step": 361340
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.357694956761307e-05,
+      "loss": 0.909,
+      "step": 361350
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.357580163771332e-05,
+      "loss": 0.8436,
+      "step": 361360
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.3574653707813576e-05,
+      "loss": 0.9225,
+      "step": 361370
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.3573505777913825e-05,
+      "loss": 0.9013,
+      "step": 361380
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.3572357848014075e-05,
+      "loss": 0.952,
+      "step": 361390
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.357120991811433e-05,
+      "loss": 0.9031,
+      "step": 361400
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.357006198821458e-05,
+      "loss": 0.8936,
+      "step": 361410
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.356891405831483e-05,
+      "loss": 0.7881,
+      "step": 361420
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.356776612841509e-05,
+      "loss": 0.9084,
+      "step": 361430
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.356661819851534e-05,
+      "loss": 0.9062,
+      "step": 361440
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.3565470268615595e-05,
+      "loss": 0.9731,
+      "step": 361450
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.3564322338715845e-05,
+      "loss": 0.8495,
+      "step": 361460
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 3.35631744088161e-05,
+      "loss": 0.8751,
+      "step": 361470
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.356202647891635e-05,
+      "loss": 0.9523,
+      "step": 361480
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.356087854901661e-05,
+      "loss": 0.9401,
+      "step": 361490
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.355973061911686e-05,
+      "loss": 0.9172,
+      "step": 361500
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.355858268921711e-05,
+      "loss": 1.0515,
+      "step": 361510
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.3557434759317364e-05,
+      "loss": 0.8607,
+      "step": 361520
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.3556286829417614e-05,
+      "loss": 0.9705,
+      "step": 361530
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.3555138899517864e-05,
+      "loss": 1.0319,
+      "step": 361540
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.355399096961812e-05,
+      "loss": 0.9675,
+      "step": 361550
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.355284303971837e-05,
+      "loss": 1.0091,
+      "step": 361560
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.355169510981862e-05,
+      "loss": 0.9087,
+      "step": 361570
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.355054717991888e-05,
+      "loss": 0.8059,
+      "step": 361580
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.354939925001913e-05,
+      "loss": 0.9276,
+      "step": 361590
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.354825132011938e-05,
+      "loss": 0.8171,
+      "step": 361600
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.354710339021963e-05,
+      "loss": 0.911,
+      "step": 361610
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.354595546031989e-05,
+      "loss": 0.9039,
+      "step": 361620
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.354480753042014e-05,
+      "loss": 0.9346,
+      "step": 361630
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.3543659600520396e-05,
+      "loss": 0.9243,
+      "step": 361640
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.3542511670620646e-05,
+      "loss": 0.8952,
+      "step": 361650
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.3541363740720896e-05,
+      "loss": 0.8056,
+      "step": 361660
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.354021581082115e-05,
+      "loss": 0.8906,
+      "step": 361670
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.35390678809214e-05,
+      "loss": 0.9106,
+      "step": 361680
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 3.353791995102165e-05,
+      "loss": 0.8629,
+      "step": 361690
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.353677202112191e-05,
+      "loss": 0.9385,
+      "step": 361700
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.353562409122216e-05,
+      "loss": 0.8518,
+      "step": 361710
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.353447616132241e-05,
+      "loss": 0.8759,
+      "step": 361720
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.3533328231422665e-05,
+      "loss": 0.9066,
+      "step": 361730
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.3532180301522915e-05,
+      "loss": 0.9144,
+      "step": 361740
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.353103237162317e-05,
+      "loss": 0.9007,
+      "step": 361750
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.352988444172342e-05,
+      "loss": 0.8147,
+      "step": 361760
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.352873651182368e-05,
+      "loss": 0.8123,
+      "step": 361770
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.352758858192393e-05,
+      "loss": 0.8761,
+      "step": 361780
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.3526440652024185e-05,
+      "loss": 0.8496,
+      "step": 361790
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.3525292722124435e-05,
+      "loss": 0.9712,
+      "step": 361800
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.3524144792224684e-05,
+      "loss": 0.8668,
+      "step": 361810
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.352299686232494e-05,
+      "loss": 0.9036,
+      "step": 361820
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.352184893242519e-05,
+      "loss": 0.9159,
+      "step": 361830
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.352070100252544e-05,
+      "loss": 0.8758,
+      "step": 361840
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.351955307262569e-05,
+      "loss": 0.8456,
+      "step": 361850
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.351840514272595e-05,
+      "loss": 0.9303,
+      "step": 361860
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.35172572128262e-05,
+      "loss": 0.7826,
+      "step": 361870
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.3516109282926454e-05,
+      "loss": 0.894,
+      "step": 361880
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.3514961353026704e-05,
+      "loss": 0.9379,
+      "step": 361890
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 3.351381342312696e-05,
+      "loss": 0.8837,
+      "step": 361900
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.351266549322721e-05,
+      "loss": 0.914,
+      "step": 361910
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.351151756332747e-05,
+      "loss": 0.9476,
+      "step": 361920
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.3510369633427717e-05,
+      "loss": 0.9108,
+      "step": 361930
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.3509221703527966e-05,
+      "loss": 0.9754,
+      "step": 361940
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.350807377362822e-05,
+      "loss": 0.9288,
+      "step": 361950
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.350692584372847e-05,
+      "loss": 0.8712,
+      "step": 361960
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.350577791382872e-05,
+      "loss": 0.9189,
+      "step": 361970
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.350462998392898e-05,
+      "loss": 0.903,
+      "step": 361980
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.350348205402923e-05,
+      "loss": 0.9113,
+      "step": 361990
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.350233412412948e-05,
+      "loss": 0.9877,
+      "step": 362000
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.3501186194229736e-05,
+      "loss": 0.9614,
+      "step": 362010
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.3500038264329986e-05,
+      "loss": 0.8269,
+      "step": 362020
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.349889033443024e-05,
+      "loss": 0.8504,
+      "step": 362030
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.349774240453049e-05,
+      "loss": 0.8309,
+      "step": 362040
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.349659447463075e-05,
+      "loss": 0.8856,
+      "step": 362050
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.3495446544731e-05,
+      "loss": 1.0466,
+      "step": 362060
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.3494298614831255e-05,
+      "loss": 0.8101,
+      "step": 362070
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.3493150684931505e-05,
+      "loss": 0.8213,
+      "step": 362080
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.3492002755031755e-05,
+      "loss": 1.0373,
+      "step": 362090
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.349085482513201e-05,
+      "loss": 0.8995,
+      "step": 362100
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.348970689523226e-05,
+      "loss": 0.962,
+      "step": 362110
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 3.348855896533251e-05,
+      "loss": 0.901,
+      "step": 362120
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.348741103543277e-05,
+      "loss": 0.8418,
+      "step": 362130
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.348626310553302e-05,
+      "loss": 0.8868,
+      "step": 362140
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.348511517563327e-05,
+      "loss": 0.8304,
+      "step": 362150
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.3483967245733524e-05,
+      "loss": 0.9001,
+      "step": 362160
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.3482819315833774e-05,
+      "loss": 0.8631,
+      "step": 362170
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.348167138593403e-05,
+      "loss": 0.8241,
+      "step": 362180
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.348052345603428e-05,
+      "loss": 0.8267,
+      "step": 362190
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.347937552613454e-05,
+      "loss": 0.8918,
+      "step": 362200
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.347822759623479e-05,
+      "loss": 0.914,
+      "step": 362210
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.3477079666335044e-05,
+      "loss": 0.8495,
+      "step": 362220
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.3475931736435294e-05,
+      "loss": 0.8834,
+      "step": 362230
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.347478380653554e-05,
+      "loss": 0.8301,
+      "step": 362240
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.34736358766358e-05,
+      "loss": 0.8588,
+      "step": 362250
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.347248794673605e-05,
+      "loss": 0.932,
+      "step": 362260
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.34713400168363e-05,
+      "loss": 0.8682,
+      "step": 362270
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.3470192086936556e-05,
+      "loss": 0.8972,
+      "step": 362280
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.3469044157036806e-05,
+      "loss": 0.9114,
+      "step": 362290
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.3467896227137056e-05,
+      "loss": 0.9501,
+      "step": 362300
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.346674829723731e-05,
+      "loss": 0.8804,
+      "step": 362310
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.346560036733756e-05,
+      "loss": 0.9027,
+      "step": 362320
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.346445243743782e-05,
+      "loss": 0.8401,
+      "step": 362330
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 3.346330450753807e-05,
+      "loss": 1.0093,
+      "step": 362340
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.3462156577638326e-05,
+      "loss": 0.8688,
+      "step": 362350
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.3461008647738575e-05,
+      "loss": 0.8808,
+      "step": 362360
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.345986071783883e-05,
+      "loss": 0.8972,
+      "step": 362370
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.345871278793908e-05,
+      "loss": 0.9539,
+      "step": 362380
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.345756485803933e-05,
+      "loss": 1.0135,
+      "step": 362390
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.345641692813959e-05,
+      "loss": 0.9507,
+      "step": 362400
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.345526899823984e-05,
+      "loss": 0.9319,
+      "step": 362410
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.345412106834009e-05,
+      "loss": 0.8284,
+      "step": 362420
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.3452973138440345e-05,
+      "loss": 0.8706,
+      "step": 362430
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.3451825208540595e-05,
+      "loss": 0.8621,
+      "step": 362440
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.3450677278640845e-05,
+      "loss": 0.914,
+      "step": 362450
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.34495293487411e-05,
+      "loss": 0.9367,
+      "step": 362460
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.344838141884135e-05,
+      "loss": 0.8403,
+      "step": 362470
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.344723348894161e-05,
+      "loss": 0.8687,
+      "step": 362480
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.344608555904186e-05,
+      "loss": 0.932,
+      "step": 362490
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.3444937629142114e-05,
+      "loss": 0.9037,
+      "step": 362500
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.3443789699242364e-05,
+      "loss": 0.9897,
+      "step": 362510
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.3442641769342614e-05,
+      "loss": 0.9417,
+      "step": 362520
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.344149383944287e-05,
+      "loss": 0.8667,
+      "step": 362530
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.344034590954312e-05,
+      "loss": 0.866,
+      "step": 362540
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.343919797964337e-05,
+      "loss": 0.8638,
+      "step": 362550
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.343805004974363e-05,
+      "loss": 0.8805,
+      "step": 362560
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.343690211984388e-05,
+      "loss": 0.8637,
+      "step": 362570
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.3435754189944126e-05,
+      "loss": 0.8946,
+      "step": 362580
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.343460626004438e-05,
+      "loss": 0.9895,
+      "step": 362590
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.343345833014463e-05,
+      "loss": 0.8668,
+      "step": 362600
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.343231040024489e-05,
+      "loss": 0.9166,
+      "step": 362610
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.343116247034514e-05,
+      "loss": 0.9291,
+      "step": 362620
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.3430014540445396e-05,
+      "loss": 0.796,
+      "step": 362630
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.3428866610545646e-05,
+      "loss": 0.9285,
+      "step": 362640
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.34277186806459e-05,
+      "loss": 0.899,
+      "step": 362650
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.342657075074615e-05,
+      "loss": 0.937,
+      "step": 362660
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.34254228208464e-05,
+      "loss": 0.9613,
+      "step": 362670
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.342427489094666e-05,
+      "loss": 1.0145,
+      "step": 362680
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.342312696104691e-05,
+      "loss": 0.9677,
+      "step": 362690
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.342197903114716e-05,
+      "loss": 0.8478,
+      "step": 362700
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.3420831101247415e-05,
+      "loss": 0.8971,
+      "step": 362710
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.3419683171347665e-05,
+      "loss": 0.8896,
+      "step": 362720
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.3418535241447915e-05,
+      "loss": 0.9235,
+      "step": 362730
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.341738731154817e-05,
+      "loss": 0.9349,
+      "step": 362740
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.341623938164842e-05,
+      "loss": 0.8673,
+      "step": 362750
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.341509145174868e-05,
+      "loss": 1.0052,
+      "step": 362760
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.341394352184893e-05,
+      "loss": 0.8631,
+      "step": 362770
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 3.3412795591949185e-05,
+      "loss": 0.9372,
+      "step": 362780
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.3411647662049434e-05,
+      "loss": 0.8492,
+      "step": 362790
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.341049973214969e-05,
+      "loss": 0.9277,
+      "step": 362800
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.340935180224994e-05,
+      "loss": 0.888,
+      "step": 362810
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.340820387235019e-05,
+      "loss": 0.936,
+      "step": 362820
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.340705594245045e-05,
+      "loss": 0.9312,
+      "step": 362830
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.34059080125507e-05,
+      "loss": 0.8835,
+      "step": 362840
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.340476008265095e-05,
+      "loss": 0.8775,
+      "step": 362850
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.3403612152751204e-05,
+      "loss": 0.9494,
+      "step": 362860
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.3402464222851454e-05,
+      "loss": 0.8623,
+      "step": 362870
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.3401316292951703e-05,
+      "loss": 0.9048,
+      "step": 362880
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.340016836305196e-05,
+      "loss": 0.8167,
+      "step": 362890
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.339902043315221e-05,
+      "loss": 0.942,
+      "step": 362900
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.3397872503252467e-05,
+      "loss": 0.8808,
+      "step": 362910
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.3396724573352716e-05,
+      "loss": 0.9571,
+      "step": 362920
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.339557664345297e-05,
+      "loss": 0.901,
+      "step": 362930
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.339442871355322e-05,
+      "loss": 0.8924,
+      "step": 362940
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.339328078365348e-05,
+      "loss": 0.8524,
+      "step": 362950
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.339213285375373e-05,
+      "loss": 0.9064,
+      "step": 362960
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.339098492385398e-05,
+      "loss": 0.8685,
+      "step": 362970
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.3389836993954236e-05,
+      "loss": 0.9754,
+      "step": 362980
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 3.3388689064054486e-05,
+      "loss": 0.8729,
+      "step": 362990
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.3387541134154736e-05,
+      "loss": 0.9574,
+      "step": 363000
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.338639320425499e-05,
+      "loss": 0.9709,
+      "step": 363010
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.338524527435524e-05,
+      "loss": 0.8304,
+      "step": 363020
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.338409734445549e-05,
+      "loss": 0.9175,
+      "step": 363030
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.338294941455575e-05,
+      "loss": 0.9647,
+      "step": 363040
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.3381801484656e-05,
+      "loss": 0.9385,
+      "step": 363050
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.3380653554756255e-05,
+      "loss": 0.9972,
+      "step": 363060
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.3379505624856505e-05,
+      "loss": 0.8543,
+      "step": 363070
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.337835769495676e-05,
+      "loss": 0.9796,
+      "step": 363080
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.337720976505701e-05,
+      "loss": 0.964,
+      "step": 363090
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.337606183515727e-05,
+      "loss": 0.9057,
+      "step": 363100
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.337491390525752e-05,
+      "loss": 0.9699,
+      "step": 363110
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.337376597535777e-05,
+      "loss": 1.0044,
+      "step": 363120
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.3372618045458024e-05,
+      "loss": 0.9017,
+      "step": 363130
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.3371470115558274e-05,
+      "loss": 0.891,
+      "step": 363140
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.3370322185658524e-05,
+      "loss": 0.8904,
+      "step": 363150
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.336917425575878e-05,
+      "loss": 0.8918,
+      "step": 363160
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.336802632585903e-05,
+      "loss": 0.8613,
+      "step": 363170
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.336687839595928e-05,
+      "loss": 0.8578,
+      "step": 363180
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.336573046605954e-05,
+      "loss": 0.9039,
+      "step": 363190
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.336458253615979e-05,
+      "loss": 1.0451,
+      "step": 363200
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 3.3363434606260043e-05,
+      "loss": 0.9169,
+      "step": 363210
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.336228667636029e-05,
+      "loss": 0.8765,
+      "step": 363220
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.336113874646055e-05,
+      "loss": 0.9154,
+      "step": 363230
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.33599908165608e-05,
+      "loss": 0.9554,
+      "step": 363240
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.335884288666105e-05,
+      "loss": 0.9615,
+      "step": 363250
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.3357694956761306e-05,
+      "loss": 0.8753,
+      "step": 363260
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.3356547026861556e-05,
+      "loss": 0.9145,
+      "step": 363270
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.3355399096961806e-05,
+      "loss": 0.9288,
+      "step": 363280
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.335425116706206e-05,
+      "loss": 0.9091,
+      "step": 363290
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.335310323716231e-05,
+      "loss": 0.8846,
+      "step": 363300
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.335195530726256e-05,
+      "loss": 0.9061,
+      "step": 363310
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.335080737736282e-05,
+      "loss": 0.9256,
+      "step": 363320
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.334965944746307e-05,
+      "loss": 0.8092,
+      "step": 363330
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.3348511517563325e-05,
+      "loss": 0.857,
+      "step": 363340
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.3347363587663575e-05,
+      "loss": 0.9959,
+      "step": 363350
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.334621565776383e-05,
+      "loss": 0.9708,
+      "step": 363360
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.334506772786408e-05,
+      "loss": 0.8428,
+      "step": 363370
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.334391979796434e-05,
+      "loss": 0.9535,
+      "step": 363380
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.334277186806459e-05,
+      "loss": 0.9287,
+      "step": 363390
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.334162393816484e-05,
+      "loss": 0.9446,
+      "step": 363400
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.3340476008265095e-05,
+      "loss": 0.888,
+      "step": 363410
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.3339328078365345e-05,
+      "loss": 0.9254,
+      "step": 363420
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 3.3338180148465595e-05,
+      "loss": 0.8071,
+      "step": 363430
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.333703221856585e-05,
+      "loss": 0.9421,
+      "step": 363440
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.33358842886661e-05,
+      "loss": 0.9151,
+      "step": 363450
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.333473635876635e-05,
+      "loss": 0.8952,
+      "step": 363460
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.333358842886661e-05,
+      "loss": 0.9022,
+      "step": 363470
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.333244049896686e-05,
+      "loss": 0.8054,
+      "step": 363480
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.3331292569067114e-05,
+      "loss": 0.9025,
+      "step": 363490
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.3330144639167364e-05,
+      "loss": 0.8122,
+      "step": 363500
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.332899670926762e-05,
+      "loss": 1.0019,
+      "step": 363510
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.332784877936787e-05,
+      "loss": 1.001,
+      "step": 363520
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.332670084946813e-05,
+      "loss": 0.9971,
+      "step": 363530
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.332555291956838e-05,
+      "loss": 0.8964,
+      "step": 363540
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.332440498966863e-05,
+      "loss": 0.8913,
+      "step": 363550
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.332325705976888e-05,
+      "loss": 0.8635,
+      "step": 363560
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.332210912986913e-05,
+      "loss": 0.8661,
+      "step": 363570
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.332096119996938e-05,
+      "loss": 0.8574,
+      "step": 363580
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.331981327006964e-05,
+      "loss": 0.9175,
+      "step": 363590
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.331866534016989e-05,
+      "loss": 0.9157,
+      "step": 363600
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.331751741027014e-05,
+      "loss": 0.9197,
+      "step": 363610
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.3316369480370396e-05,
+      "loss": 0.8999,
+      "step": 363620
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.3315221550470646e-05,
+      "loss": 0.9573,
+      "step": 363630
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.33140736205709e-05,
+      "loss": 0.9041,
+      "step": 363640
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 3.331292569067115e-05,
+      "loss": 0.7871,
+      "step": 363650
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.331177776077141e-05,
+      "loss": 0.9033,
+      "step": 363660
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.331062983087166e-05,
+      "loss": 0.9062,
+      "step": 363670
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.3309481900971915e-05,
+      "loss": 0.9541,
+      "step": 363680
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.3308333971072165e-05,
+      "loss": 0.9077,
+      "step": 363690
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.3307186041172415e-05,
+      "loss": 1.0193,
+      "step": 363700
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.330603811127267e-05,
+      "loss": 0.9225,
+      "step": 363710
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.330489018137292e-05,
+      "loss": 0.8892,
+      "step": 363720
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.330374225147317e-05,
+      "loss": 0.8741,
+      "step": 363730
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.330259432157343e-05,
+      "loss": 0.8612,
+      "step": 363740
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.330144639167368e-05,
+      "loss": 0.9432,
+      "step": 363750
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.330029846177393e-05,
+      "loss": 0.8696,
+      "step": 363760
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.3299150531874184e-05,
+      "loss": 0.876,
+      "step": 363770
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.3298002601974434e-05,
+      "loss": 0.8654,
+      "step": 363780
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.329685467207469e-05,
+      "loss": 0.827,
+      "step": 363790
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.329570674217494e-05,
+      "loss": 0.8917,
+      "step": 363800
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.32945588122752e-05,
+      "loss": 0.9764,
+      "step": 363810
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.329341088237545e-05,
+      "loss": 0.8148,
+      "step": 363820
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.3292262952475704e-05,
+      "loss": 0.9401,
+      "step": 363830
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.3291115022575954e-05,
+      "loss": 0.814,
+      "step": 363840
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.3289967092676204e-05,
+      "loss": 0.7751,
+      "step": 363850
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.328881916277646e-05,
+      "loss": 0.8433,
+      "step": 363860
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 3.328767123287671e-05,
+      "loss": 0.902,
+      "step": 363870
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.328652330297696e-05,
+      "loss": 1.012,
+      "step": 363880
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.3285375373077217e-05,
+      "loss": 0.9386,
+      "step": 363890
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.3284227443177466e-05,
+      "loss": 0.9082,
+      "step": 363900
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.3283079513277716e-05,
+      "loss": 0.918,
+      "step": 363910
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.328193158337797e-05,
+      "loss": 0.8937,
+      "step": 363920
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.328078365347822e-05,
+      "loss": 0.9137,
+      "step": 363930
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.327963572357848e-05,
+      "loss": 0.918,
+      "step": 363940
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.327848779367873e-05,
+      "loss": 0.8911,
+      "step": 363950
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.3277339863778986e-05,
+      "loss": 0.8653,
+      "step": 363960
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.3276191933879236e-05,
+      "loss": 0.8282,
+      "step": 363970
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.3275044003979486e-05,
+      "loss": 0.8904,
+      "step": 363980
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.327389607407974e-05,
+      "loss": 0.8836,
+      "step": 363990
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.327274814417999e-05,
+      "loss": 0.8979,
+      "step": 364000
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.327160021428024e-05,
+      "loss": 0.8673,
+      "step": 364010
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.32704522843805e-05,
+      "loss": 1.0421,
+      "step": 364020
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.326930435448075e-05,
+      "loss": 0.9121,
+      "step": 364030
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.3268156424581e-05,
+      "loss": 0.898,
+      "step": 364040
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.3267008494681255e-05,
+      "loss": 0.9775,
+      "step": 364050
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.3265860564781505e-05,
+      "loss": 0.9281,
+      "step": 364060
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.326471263488176e-05,
+      "loss": 0.8914,
+      "step": 364070
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 3.326356470498201e-05,
+      "loss": 0.9744,
+      "step": 364080
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.326241677508227e-05,
+      "loss": 0.907,
+      "step": 364090
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.326126884518252e-05,
+      "loss": 0.9634,
+      "step": 364100
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.3260120915282774e-05,
+      "loss": 0.8369,
+      "step": 364110
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.3258972985383024e-05,
+      "loss": 0.9704,
+      "step": 364120
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.3257825055483274e-05,
+      "loss": 0.9195,
+      "step": 364130
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.325667712558353e-05,
+      "loss": 0.9849,
+      "step": 364140
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.325552919568378e-05,
+      "loss": 0.9072,
+      "step": 364150
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.325438126578403e-05,
+      "loss": 0.8607,
+      "step": 364160
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.325323333588429e-05,
+      "loss": 0.9964,
+      "step": 364170
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.325208540598454e-05,
+      "loss": 0.9631,
+      "step": 364180
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.325093747608479e-05,
+      "loss": 0.9727,
+      "step": 364190
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.324978954618504e-05,
+      "loss": 0.9014,
+      "step": 364200
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.324864161628529e-05,
+      "loss": 0.9354,
+      "step": 364210
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.324749368638555e-05,
+      "loss": 0.9384,
+      "step": 364220
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.32463457564858e-05,
+      "loss": 0.89,
+      "step": 364230
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.3245197826586056e-05,
+      "loss": 0.8851,
+      "step": 364240
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.3244049896686306e-05,
+      "loss": 0.9435,
+      "step": 364250
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.324290196678656e-05,
+      "loss": 0.895,
+      "step": 364260
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.324175403688681e-05,
+      "loss": 0.9464,
+      "step": 364270
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.324060610698706e-05,
+      "loss": 0.9244,
+      "step": 364280
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.323945817708732e-05,
+      "loss": 0.8999,
+      "step": 364290
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 3.323831024718757e-05,
+      "loss": 0.9345,
+      "step": 364300
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.323716231728782e-05,
+      "loss": 0.9789,
+      "step": 364310
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.3236014387388075e-05,
+      "loss": 1.0304,
+      "step": 364320
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.3234866457488325e-05,
+      "loss": 0.8454,
+      "step": 364330
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.3233718527588575e-05,
+      "loss": 0.8436,
+      "step": 364340
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.323257059768883e-05,
+      "loss": 0.9216,
+      "step": 364350
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.323142266778908e-05,
+      "loss": 0.8875,
+      "step": 364360
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.323027473788934e-05,
+      "loss": 0.9543,
+      "step": 364370
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.322912680798959e-05,
+      "loss": 0.9436,
+      "step": 364380
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.3227978878089845e-05,
+      "loss": 0.931,
+      "step": 364390
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.3226830948190095e-05,
+      "loss": 0.8816,
+      "step": 364400
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.322568301829035e-05,
+      "loss": 0.9811,
+      "step": 364410
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.32245350883906e-05,
+      "loss": 0.855,
+      "step": 364420
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.322338715849085e-05,
+      "loss": 0.9145,
+      "step": 364430
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.322223922859111e-05,
+      "loss": 0.9895,
+      "step": 364440
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.322109129869136e-05,
+      "loss": 0.9136,
+      "step": 364450
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.321994336879161e-05,
+      "loss": 0.8197,
+      "step": 364460
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.3218795438891864e-05,
+      "loss": 0.8571,
+      "step": 364470
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.3217647508992114e-05,
+      "loss": 0.8986,
+      "step": 364480
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.3216499579092364e-05,
+      "loss": 0.8545,
+      "step": 364490
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.321535164919262e-05,
+      "loss": 0.8313,
+      "step": 364500
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.321420371929287e-05,
+      "loss": 1.0279,
+      "step": 364510
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 3.321305578939313e-05,
+      "loss": 0.9216,
+      "step": 364520
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.321190785949338e-05,
+      "loss": 0.7642,
+      "step": 364530
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.321075992959363e-05,
+      "loss": 0.9065,
+      "step": 364540
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.320961199969388e-05,
+      "loss": 0.8655,
+      "step": 364550
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.320846406979414e-05,
+      "loss": 0.8498,
+      "step": 364560
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.320731613989439e-05,
+      "loss": 0.9444,
+      "step": 364570
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.320616820999464e-05,
+      "loss": 0.9076,
+      "step": 364580
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.320502028009489e-05,
+      "loss": 0.8356,
+      "step": 364590
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.3203872350195146e-05,
+      "loss": 0.8417,
+      "step": 364600
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.3202724420295396e-05,
+      "loss": 0.9249,
+      "step": 364610
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.3201576490395646e-05,
+      "loss": 1.0271,
+      "step": 364620
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.32004285604959e-05,
+      "loss": 0.8705,
+      "step": 364630
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.319928063059615e-05,
+      "loss": 0.8664,
+      "step": 364640
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.319813270069641e-05,
+      "loss": 0.892,
+      "step": 364650
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.319698477079666e-05,
+      "loss": 0.8701,
+      "step": 364660
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.3195836840896915e-05,
+      "loss": 0.8935,
+      "step": 364670
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.3194688910997165e-05,
+      "loss": 0.8745,
+      "step": 364680
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.319354098109742e-05,
+      "loss": 0.8712,
+      "step": 364690
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.319239305119767e-05,
+      "loss": 0.932,
+      "step": 364700
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.319124512129792e-05,
+      "loss": 0.8756,
+      "step": 364710
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.319009719139818e-05,
+      "loss": 0.8876,
+      "step": 364720
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.318894926149843e-05,
+      "loss": 0.8759,
+      "step": 364730
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 3.318780133159868e-05,
+      "loss": 0.8628,
+      "step": 364740
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.3186653401698934e-05,
+      "loss": 0.8775,
+      "step": 364750
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.3185505471799184e-05,
+      "loss": 0.874,
+      "step": 364760
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.3184357541899434e-05,
+      "loss": 0.8525,
+      "step": 364770
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.318320961199969e-05,
+      "loss": 0.958,
+      "step": 364780
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.318206168209994e-05,
+      "loss": 0.8519,
+      "step": 364790
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.31809137522002e-05,
+      "loss": 0.9591,
+      "step": 364800
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.317976582230045e-05,
+      "loss": 0.8894,
+      "step": 364810
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.3178617892400704e-05,
+      "loss": 1.0066,
+      "step": 364820
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.3177469962500954e-05,
+      "loss": 0.9372,
+      "step": 364830
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.317632203260121e-05,
+      "loss": 0.977,
+      "step": 364840
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.317517410270146e-05,
+      "loss": 0.8602,
+      "step": 364850
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.317402617280171e-05,
+      "loss": 0.9445,
+      "step": 364860
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.3172878242901967e-05,
+      "loss": 0.9902,
+      "step": 364870
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.3171730313002216e-05,
+      "loss": 1.0068,
+      "step": 364880
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.3170582383102466e-05,
+      "loss": 0.9167,
+      "step": 364890
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.316943445320272e-05,
+      "loss": 0.8726,
+      "step": 364900
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.316828652330297e-05,
+      "loss": 0.9254,
+      "step": 364910
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.316713859340322e-05,
+      "loss": 0.9373,
+      "step": 364920
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.316599066350348e-05,
+      "loss": 0.9471,
+      "step": 364930
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.316484273360373e-05,
+      "loss": 0.8753,
+      "step": 364940
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.3163694803703986e-05,
+      "loss": 0.8635,
+      "step": 364950
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.3162546873804236e-05,
+      "loss": 0.8582,
+      "step": 364960
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.316139894390449e-05,
+      "loss": 0.8228,
+      "step": 364970
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.316025101400474e-05,
+      "loss": 0.914,
+      "step": 364980
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.3159103084105e-05,
+      "loss": 0.8991,
+      "step": 364990
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.315795515420525e-05,
+      "loss": 0.9902,
+      "step": 365000
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.31568072243055e-05,
+      "loss": 0.935,
+      "step": 365010
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.3155659294405755e-05,
+      "loss": 0.9628,
+      "step": 365020
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.3154511364506005e-05,
+      "loss": 0.9114,
+      "step": 365030
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.3153363434606255e-05,
+      "loss": 0.9692,
+      "step": 365040
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.315221550470651e-05,
+      "loss": 0.8811,
+      "step": 365050
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.315106757480676e-05,
+      "loss": 0.9244,
+      "step": 365060
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.314991964490701e-05,
+      "loss": 0.9082,
+      "step": 365070
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.314877171500727e-05,
+      "loss": 0.953,
+      "step": 365080
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.314762378510752e-05,
+      "loss": 0.7888,
+      "step": 365090
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.3146475855207774e-05,
+      "loss": 0.8383,
+      "step": 365100
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.3145327925308024e-05,
+      "loss": 0.8551,
+      "step": 365110
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.314417999540828e-05,
+      "loss": 0.9961,
+      "step": 365120
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.314303206550853e-05,
+      "loss": 0.7985,
+      "step": 365130
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.314188413560879e-05,
+      "loss": 0.8898,
+      "step": 365140
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.314073620570904e-05,
+      "loss": 0.9407,
+      "step": 365150
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.313958827580929e-05,
+      "loss": 0.9347,
+      "step": 365160
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 3.3138440345909543e-05,
+      "loss": 0.9991,
+      "step": 365170
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.313729241600979e-05,
+      "loss": 0.9052,
+      "step": 365180
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.313614448611004e-05,
+      "loss": 0.9229,
+      "step": 365190
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.31349965562103e-05,
+      "loss": 0.8693,
+      "step": 365200
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.313384862631055e-05,
+      "loss": 0.8882,
+      "step": 365210
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.31327006964108e-05,
+      "loss": 0.8892,
+      "step": 365220
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.3131552766511056e-05,
+      "loss": 0.8696,
+      "step": 365230
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.3130404836611306e-05,
+      "loss": 0.8541,
+      "step": 365240
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.312925690671156e-05,
+      "loss": 0.8615,
+      "step": 365250
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.312810897681181e-05,
+      "loss": 0.9911,
+      "step": 365260
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.312696104691207e-05,
+      "loss": 0.8876,
+      "step": 365270
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.312581311701232e-05,
+      "loss": 0.9949,
+      "step": 365280
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.312466518711257e-05,
+      "loss": 0.9333,
+      "step": 365290
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.3123517257212825e-05,
+      "loss": 0.905,
+      "step": 365300
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.3122369327313075e-05,
+      "loss": 0.8911,
+      "step": 365310
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.3121221397413325e-05,
+      "loss": 0.8576,
+      "step": 365320
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.312007346751358e-05,
+      "loss": 0.9176,
+      "step": 365330
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.311892553761383e-05,
+      "loss": 0.9676,
+      "step": 365340
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.311777760771408e-05,
+      "loss": 0.9154,
+      "step": 365350
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.311662967781434e-05,
+      "loss": 0.8744,
+      "step": 365360
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.311548174791459e-05,
+      "loss": 0.9435,
+      "step": 365370
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.3114333818014845e-05,
+      "loss": 0.8156,
+      "step": 365380
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 3.3113185888115094e-05,
+      "loss": 0.9075,
+      "step": 365390
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.311203795821535e-05,
+      "loss": 0.8644,
+      "step": 365400
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.31108900283156e-05,
+      "loss": 0.8869,
+      "step": 365410
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.310974209841586e-05,
+      "loss": 0.9843,
+      "step": 365420
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.310859416851611e-05,
+      "loss": 0.8557,
+      "step": 365430
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.310744623861636e-05,
+      "loss": 0.9093,
+      "step": 365440
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.3106298308716614e-05,
+      "loss": 0.8291,
+      "step": 365450
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.3105150378816864e-05,
+      "loss": 0.9004,
+      "step": 365460
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.3104002448917114e-05,
+      "loss": 0.9232,
+      "step": 365470
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.310285451901737e-05,
+      "loss": 0.8478,
+      "step": 365480
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.310170658911762e-05,
+      "loss": 0.8092,
+      "step": 365490
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.310055865921787e-05,
+      "loss": 0.8586,
+      "step": 365500
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.309941072931813e-05,
+      "loss": 0.8984,
+      "step": 365510
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.3098262799418376e-05,
+      "loss": 0.9274,
+      "step": 365520
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.309711486951863e-05,
+      "loss": 0.858,
+      "step": 365530
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.309596693961888e-05,
+      "loss": 0.9189,
+      "step": 365540
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.309481900971914e-05,
+      "loss": 0.9319,
+      "step": 365550
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.309367107981939e-05,
+      "loss": 0.9289,
+      "step": 365560
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.3092523149919646e-05,
+      "loss": 1.0055,
+      "step": 365570
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.3091375220019896e-05,
+      "loss": 0.9125,
+      "step": 365580
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.3090227290120146e-05,
+      "loss": 0.868,
+      "step": 365590
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.30890793602204e-05,
+      "loss": 0.8808,
+      "step": 365600
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 3.308793143032065e-05,
+      "loss": 0.9307,
+      "step": 365610
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.30867835004209e-05,
+      "loss": 0.8445,
+      "step": 365620
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.308563557052116e-05,
+      "loss": 0.8354,
+      "step": 365630
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.308448764062141e-05,
+      "loss": 0.8948,
+      "step": 365640
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.308333971072166e-05,
+      "loss": 0.9584,
+      "step": 365650
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.3082191780821915e-05,
+      "loss": 0.8668,
+      "step": 365660
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.3081043850922165e-05,
+      "loss": 0.87,
+      "step": 365670
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.307989592102242e-05,
+      "loss": 0.8367,
+      "step": 365680
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.307874799112267e-05,
+      "loss": 0.8504,
+      "step": 365690
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.307760006122293e-05,
+      "loss": 0.875,
+      "step": 365700
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.307645213132318e-05,
+      "loss": 0.8457,
+      "step": 365710
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.3075304201423435e-05,
+      "loss": 0.9069,
+      "step": 365720
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.3074156271523684e-05,
+      "loss": 0.8329,
+      "step": 365730
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.3073008341623934e-05,
+      "loss": 0.9764,
+      "step": 365740
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.307186041172419e-05,
+      "loss": 0.9298,
+      "step": 365750
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.307071248182444e-05,
+      "loss": 0.9023,
+      "step": 365760
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.306956455192469e-05,
+      "loss": 0.8824,
+      "step": 365770
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.306841662202495e-05,
+      "loss": 1.0273,
+      "step": 365780
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.30672686921252e-05,
+      "loss": 1.0051,
+      "step": 365790
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.306612076222545e-05,
+      "loss": 0.9328,
+      "step": 365800
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.3064972832325704e-05,
+      "loss": 0.9613,
+      "step": 365810
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.3063824902425953e-05,
+      "loss": 0.8869,
+      "step": 365820
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 3.306267697252621e-05,
+      "loss": 1.0357,
+      "step": 365830
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.306152904262646e-05,
+      "loss": 0.8618,
+      "step": 365840
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.3060381112726717e-05,
+      "loss": 0.8856,
+      "step": 365850
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.3059233182826966e-05,
+      "loss": 0.96,
+      "step": 365860
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.305808525292722e-05,
+      "loss": 0.9748,
+      "step": 365870
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.305693732302747e-05,
+      "loss": 0.8626,
+      "step": 365880
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.305578939312772e-05,
+      "loss": 0.9166,
+      "step": 365890
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.305464146322798e-05,
+      "loss": 0.9335,
+      "step": 365900
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.305349353332823e-05,
+      "loss": 0.9828,
+      "step": 365910
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.305234560342848e-05,
+      "loss": 0.8639,
+      "step": 365920
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.3051197673528736e-05,
+      "loss": 0.9795,
+      "step": 365930
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.3050049743628986e-05,
+      "loss": 0.8431,
+      "step": 365940
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.3048901813729235e-05,
+      "loss": 0.9066,
+      "step": 365950
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.304775388382949e-05,
+      "loss": 0.9386,
+      "step": 365960
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.304660595392974e-05,
+      "loss": 0.9078,
+      "step": 365970
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.304545802403e-05,
+      "loss": 0.893,
+      "step": 365980
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.304431009413025e-05,
+      "loss": 0.9121,
+      "step": 365990
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.3043162164230505e-05,
+      "loss": 1.0083,
+      "step": 366000
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.3042014234330755e-05,
+      "loss": 0.8144,
+      "step": 366010
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.3040866304431005e-05,
+      "loss": 0.7822,
+      "step": 366020
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.303971837453126e-05,
+      "loss": 0.8876,
+      "step": 366030
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 3.303857044463151e-05,
+      "loss": 0.8695,
+      "step": 366040
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.303742251473176e-05,
+      "loss": 0.842,
+      "step": 366050
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.303627458483202e-05,
+      "loss": 0.9002,
+      "step": 366060
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.303512665493227e-05,
+      "loss": 0.8455,
+      "step": 366070
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.303397872503252e-05,
+      "loss": 0.866,
+      "step": 366080
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.3032830795132774e-05,
+      "loss": 0.9806,
+      "step": 366090
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.3031682865233024e-05,
+      "loss": 0.967,
+      "step": 366100
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.303053493533328e-05,
+      "loss": 0.9324,
+      "step": 366110
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.302938700543353e-05,
+      "loss": 0.7665,
+      "step": 366120
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.302823907553379e-05,
+      "loss": 0.8385,
+      "step": 366130
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.302709114563404e-05,
+      "loss": 0.8945,
+      "step": 366140
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.3025943215734293e-05,
+      "loss": 0.9997,
+      "step": 366150
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.302479528583454e-05,
+      "loss": 0.8579,
+      "step": 366160
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.302364735593479e-05,
+      "loss": 0.7854,
+      "step": 366170
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.302249942603505e-05,
+      "loss": 0.9818,
+      "step": 366180
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.30213514961353e-05,
+      "loss": 0.9172,
+      "step": 366190
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.302020356623555e-05,
+      "loss": 0.8926,
+      "step": 366200
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.3019055636335806e-05,
+      "loss": 0.7741,
+      "step": 366210
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.3017907706436056e-05,
+      "loss": 0.8896,
+      "step": 366220
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.3016759776536306e-05,
+      "loss": 1.0313,
+      "step": 366230
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.301561184663656e-05,
+      "loss": 0.9722,
+      "step": 366240
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.301446391673681e-05,
+      "loss": 0.8831,
+      "step": 366250
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 3.301331598683707e-05,
+      "loss": 0.924,
+      "step": 366260
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.301216805693732e-05,
+      "loss": 0.9286,
+      "step": 366270
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.3011020127037575e-05,
+      "loss": 0.9739,
+      "step": 366280
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.3009872197137825e-05,
+      "loss": 0.8972,
+      "step": 366290
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.300872426723808e-05,
+      "loss": 0.8983,
+      "step": 366300
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.300757633733833e-05,
+      "loss": 0.9009,
+      "step": 366310
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.300642840743858e-05,
+      "loss": 0.8969,
+      "step": 366320
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.300528047753884e-05,
+      "loss": 0.8055,
+      "step": 366330
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.300413254763909e-05,
+      "loss": 0.9337,
+      "step": 366340
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.300298461773934e-05,
+      "loss": 0.927,
+      "step": 366350
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.3001836687839595e-05,
+      "loss": 0.9667,
+      "step": 366360
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.3000688757939844e-05,
+      "loss": 0.9829,
+      "step": 366370
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.2999540828040094e-05,
+      "loss": 0.8778,
+      "step": 366380
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.299839289814035e-05,
+      "loss": 0.9685,
+      "step": 366390
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.29972449682406e-05,
+      "loss": 0.9602,
+      "step": 366400
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.299609703834086e-05,
+      "loss": 0.8672,
+      "step": 366410
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.299494910844111e-05,
+      "loss": 0.9499,
+      "step": 366420
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.2993801178541364e-05,
+      "loss": 0.959,
+      "step": 366430
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.2992653248641614e-05,
+      "loss": 0.901,
+      "step": 366440
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.299150531874187e-05,
+      "loss": 0.894,
+      "step": 366450
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.299035738884212e-05,
+      "loss": 0.8803,
+      "step": 366460
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.298920945894237e-05,
+      "loss": 0.908,
+      "step": 366470
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 3.298806152904263e-05,
+      "loss": 1.0118,
+      "step": 366480
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.298691359914288e-05,
+      "loss": 0.9164,
+      "step": 366490
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.2985765669243126e-05,
+      "loss": 0.9323,
+      "step": 366500
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.298461773934338e-05,
+      "loss": 0.9059,
+      "step": 366510
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.298346980944363e-05,
+      "loss": 0.9064,
+      "step": 366520
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.298232187954388e-05,
+      "loss": 0.9056,
+      "step": 366530
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.298117394964414e-05,
+      "loss": 0.8982,
+      "step": 366540
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.298002601974439e-05,
+      "loss": 0.8714,
+      "step": 366550
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.2978878089844646e-05,
+      "loss": 0.9072,
+      "step": 366560
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.2977730159944896e-05,
+      "loss": 0.8746,
+      "step": 366570
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.297658223004515e-05,
+      "loss": 0.9271,
+      "step": 366580
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.29754343001454e-05,
+      "loss": 0.9167,
+      "step": 366590
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.297428637024566e-05,
+      "loss": 0.875,
+      "step": 366600
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.297313844034591e-05,
+      "loss": 0.9294,
+      "step": 366610
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.297199051044616e-05,
+      "loss": 0.9563,
+      "step": 366620
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.2970842580546415e-05,
+      "loss": 0.8856,
+      "step": 366630
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.2969694650646665e-05,
+      "loss": 0.9019,
+      "step": 366640
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.2968546720746915e-05,
+      "loss": 0.9429,
+      "step": 366650
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.2967398790847165e-05,
+      "loss": 0.8564,
+      "step": 366660
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.296625086094742e-05,
+      "loss": 0.8801,
+      "step": 366670
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.296510293104767e-05,
+      "loss": 0.9289,
+      "step": 366680
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.296395500114793e-05,
+      "loss": 0.9302,
+      "step": 366690
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 3.296280707124818e-05,
+      "loss": 0.8116,
+      "step": 366700
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.2961659141348434e-05,
+      "loss": 0.9258,
+      "step": 366710
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.2960511211448684e-05,
+      "loss": 0.8181,
+      "step": 366720
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.295936328154894e-05,
+      "loss": 0.9281,
+      "step": 366730
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.295821535164919e-05,
+      "loss": 0.9768,
+      "step": 366740
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.295706742174944e-05,
+      "loss": 0.9696,
+      "step": 366750
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.29559194918497e-05,
+      "loss": 0.8262,
+      "step": 366760
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.295477156194995e-05,
+      "loss": 0.8689,
+      "step": 366770
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.29536236320502e-05,
+      "loss": 0.8746,
+      "step": 366780
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.2952475702150454e-05,
+      "loss": 0.8812,
+      "step": 366790
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.2951327772250703e-05,
+      "loss": 0.8627,
+      "step": 366800
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.295017984235095e-05,
+      "loss": 0.9173,
+      "step": 366810
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.294903191245121e-05,
+      "loss": 0.9374,
+      "step": 366820
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.294788398255146e-05,
+      "loss": 0.8455,
+      "step": 366830
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.2946736052651716e-05,
+      "loss": 0.987,
+      "step": 366840
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.2945588122751966e-05,
+      "loss": 0.8602,
+      "step": 366850
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.294444019285222e-05,
+      "loss": 0.9181,
+      "step": 366860
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.294329226295247e-05,
+      "loss": 0.8877,
+      "step": 366870
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.294214433305273e-05,
+      "loss": 0.8494,
+      "step": 366880
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.294099640315298e-05,
+      "loss": 0.8753,
+      "step": 366890
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.293984847325323e-05,
+      "loss": 0.8808,
+      "step": 366900
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.2938700543353486e-05,
+      "loss": 0.9823,
+      "step": 366910
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 3.2937552613453736e-05,
+      "loss": 0.9109,
+      "step": 366920
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.2936404683553985e-05,
+      "loss": 0.893,
+      "step": 366930
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.293525675365424e-05,
+      "loss": 0.8878,
+      "step": 366940
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.293410882375449e-05,
+      "loss": 0.9028,
+      "step": 366950
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.293296089385474e-05,
+      "loss": 0.9202,
+      "step": 366960
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.2931812963955e-05,
+      "loss": 0.8793,
+      "step": 366970
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.293066503405525e-05,
+      "loss": 0.9019,
+      "step": 366980
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.2929517104155505e-05,
+      "loss": 0.9033,
+      "step": 366990
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.2928369174255755e-05,
+      "loss": 0.9503,
+      "step": 367000
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.292722124435601e-05,
+      "loss": 0.9741,
+      "step": 367010
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.292607331445626e-05,
+      "loss": 0.9917,
+      "step": 367020
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.292492538455652e-05,
+      "loss": 0.9474,
+      "step": 367030
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.292377745465677e-05,
+      "loss": 0.8598,
+      "step": 367040
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.292262952475702e-05,
+      "loss": 0.9369,
+      "step": 367050
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.2921481594857274e-05,
+      "loss": 0.8619,
+      "step": 367060
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.2920333664957524e-05,
+      "loss": 0.9482,
+      "step": 367070
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.2919185735057774e-05,
+      "loss": 0.8132,
+      "step": 367080
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.291803780515803e-05,
+      "loss": 0.904,
+      "step": 367090
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.291688987525828e-05,
+      "loss": 0.8899,
+      "step": 367100
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.291574194535853e-05,
+      "loss": 0.8495,
+      "step": 367110
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.291459401545879e-05,
+      "loss": 0.9154,
+      "step": 367120
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 3.291344608555904e-05,
+      "loss": 0.9055,
+      "step": 367130
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.291229815565929e-05,
+      "loss": 0.9103,
+      "step": 367140
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.291115022575954e-05,
+      "loss": 0.9487,
+      "step": 367150
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.29100022958598e-05,
+      "loss": 0.8144,
+      "step": 367160
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.290885436596005e-05,
+      "loss": 0.8859,
+      "step": 367170
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.2907706436060306e-05,
+      "loss": 0.9055,
+      "step": 367180
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.2906558506160556e-05,
+      "loss": 0.9049,
+      "step": 367190
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.2905410576260806e-05,
+      "loss": 0.9263,
+      "step": 367200
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.290426264636106e-05,
+      "loss": 0.771,
+      "step": 367210
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.290311471646131e-05,
+      "loss": 0.8712,
+      "step": 367220
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.290196678656156e-05,
+      "loss": 0.9829,
+      "step": 367230
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.290081885666182e-05,
+      "loss": 0.928,
+      "step": 367240
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.289967092676207e-05,
+      "loss": 0.7854,
+      "step": 367250
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.289852299686232e-05,
+      "loss": 0.9099,
+      "step": 367260
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.2897375066962575e-05,
+      "loss": 0.9082,
+      "step": 367270
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.2896227137062825e-05,
+      "loss": 0.9517,
+      "step": 367280
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.289507920716308e-05,
+      "loss": 0.8702,
+      "step": 367290
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.289393127726333e-05,
+      "loss": 0.8695,
+      "step": 367300
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.289278334736359e-05,
+      "loss": 0.8298,
+      "step": 367310
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.289163541746384e-05,
+      "loss": 0.8982,
+      "step": 367320
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.289048748756409e-05,
+      "loss": 0.9197,
+      "step": 367330
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.2889339557664345e-05,
+      "loss": 1.034,
+      "step": 367340
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.2888191627764594e-05,
+      "loss": 0.8521,
+      "step": 367350
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.2887043697864844e-05,
+      "loss": 0.8778,
+      "step": 367360
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.28858957679651e-05,
+      "loss": 0.8795,
+      "step": 367370
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.288474783806535e-05,
+      "loss": 0.958,
+      "step": 367380
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.28835999081656e-05,
+      "loss": 0.9315,
+      "step": 367390
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.288245197826586e-05,
+      "loss": 0.9406,
+      "step": 367400
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.288130404836611e-05,
+      "loss": 0.9583,
+      "step": 367410
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.2880156118466364e-05,
+      "loss": 0.8771,
+      "step": 367420
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.2879008188566614e-05,
+      "loss": 0.8818,
+      "step": 367430
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.287786025866687e-05,
+      "loss": 0.8702,
+      "step": 367440
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.287671232876712e-05,
+      "loss": 0.8929,
+      "step": 367450
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.287556439886738e-05,
+      "loss": 0.8858,
+      "step": 367460
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.287441646896763e-05,
+      "loss": 0.8904,
+      "step": 367470
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.2873268539067876e-05,
+      "loss": 0.8898,
+      "step": 367480
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.287212060916813e-05,
+      "loss": 0.8876,
+      "step": 367490
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.287097267926838e-05,
+      "loss": 0.9112,
+      "step": 367500
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.286982474936863e-05,
+      "loss": 0.8192,
+      "step": 367510
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.286867681946889e-05,
+      "loss": 0.852,
+      "step": 367520
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.286752888956914e-05,
+      "loss": 0.946,
+      "step": 367530
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.286638095966939e-05,
+      "loss": 0.921,
+      "step": 367540
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.2865233029769646e-05,
+      "loss": 0.8965,
+      "step": 367550
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.2864085099869896e-05,
+      "loss": 0.9026,
+      "step": 367560
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 3.286293716997015e-05,
+      "loss": 0.9524,
+      "step": 367570
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.28617892400704e-05,
+      "loss": 0.8052,
+      "step": 367580
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.286064131017066e-05,
+      "loss": 0.9394,
+      "step": 367590
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.285949338027091e-05,
+      "loss": 0.8949,
+      "step": 367600
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.2858345450371165e-05,
+      "loss": 0.8805,
+      "step": 367610
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.2857197520471415e-05,
+      "loss": 0.9394,
+      "step": 367620
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.2856049590571665e-05,
+      "loss": 0.9232,
+      "step": 367630
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.285490166067192e-05,
+      "loss": 0.9425,
+      "step": 367640
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.285375373077217e-05,
+      "loss": 0.8607,
+      "step": 367650
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.285260580087242e-05,
+      "loss": 0.8936,
+      "step": 367660
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.285145787097268e-05,
+      "loss": 0.8955,
+      "step": 367670
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.285030994107293e-05,
+      "loss": 0.9452,
+      "step": 367680
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.284916201117318e-05,
+      "loss": 0.9271,
+      "step": 367690
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.2848014081273434e-05,
+      "loss": 0.8551,
+      "step": 367700
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.2846866151373684e-05,
+      "loss": 0.9008,
+      "step": 367710
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.284571822147394e-05,
+      "loss": 0.873,
+      "step": 367720
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.284457029157419e-05,
+      "loss": 0.9454,
+      "step": 367730
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.284342236167445e-05,
+      "loss": 0.8933,
+      "step": 367740
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.28422744317747e-05,
+      "loss": 0.8556,
+      "step": 367750
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.2841126501874954e-05,
+      "loss": 0.9777,
+      "step": 367760
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.2839978571975204e-05,
+      "loss": 0.9906,
+      "step": 367770
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.2838830642075453e-05,
+      "loss": 0.8478,
+      "step": 367780
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 3.283768271217571e-05,
+      "loss": 0.9701,
+      "step": 367790
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.283653478227596e-05,
+      "loss": 0.9216,
+      "step": 367800
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.283538685237621e-05,
+      "loss": 0.9453,
+      "step": 367810
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.2834238922476466e-05,
+      "loss": 0.978,
+      "step": 367820
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.2833090992576716e-05,
+      "loss": 0.8357,
+      "step": 367830
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.2831943062676966e-05,
+      "loss": 0.9513,
+      "step": 367840
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.283079513277722e-05,
+      "loss": 0.8795,
+      "step": 367850
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.282964720287747e-05,
+      "loss": 0.9612,
+      "step": 367860
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.282849927297773e-05,
+      "loss": 1.0623,
+      "step": 367870
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.282735134307798e-05,
+      "loss": 0.9383,
+      "step": 367880
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.2826203413178236e-05,
+      "loss": 0.8738,
+      "step": 367890
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.2825055483278486e-05,
+      "loss": 0.9874,
+      "step": 367900
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.282390755337874e-05,
+      "loss": 0.9323,
+      "step": 367910
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.282275962347899e-05,
+      "loss": 0.9344,
+      "step": 367920
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.282161169357924e-05,
+      "loss": 0.9519,
+      "step": 367930
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.28204637636795e-05,
+      "loss": 0.8488,
+      "step": 367940
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.281931583377975e-05,
+      "loss": 0.9228,
+      "step": 367950
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.281816790388e-05,
+      "loss": 0.9132,
+      "step": 367960
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.2817019973980255e-05,
+      "loss": 0.8928,
+      "step": 367970
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.2815872044080505e-05,
+      "loss": 0.9019,
+      "step": 367980
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.2814724114180755e-05,
+      "loss": 0.8741,
+      "step": 367990
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.281357618428101e-05,
+      "loss": 0.9152,
+      "step": 368000
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 3.281242825438126e-05,
+      "loss": 0.8953,
+      "step": 368010
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.281128032448152e-05,
+      "loss": 0.832,
+      "step": 368020
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.281013239458177e-05,
+      "loss": 0.871,
+      "step": 368030
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.2808984464682024e-05,
+      "loss": 0.9386,
+      "step": 368040
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.2807836534782274e-05,
+      "loss": 0.8993,
+      "step": 368050
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.2806688604882524e-05,
+      "loss": 0.9429,
+      "step": 368060
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.280554067498278e-05,
+      "loss": 0.8593,
+      "step": 368070
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.280439274508303e-05,
+      "loss": 0.9234,
+      "step": 368080
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.280324481518328e-05,
+      "loss": 0.9521,
+      "step": 368090
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.280209688528354e-05,
+      "loss": 0.9138,
+      "step": 368100
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.280094895538379e-05,
+      "loss": 0.8663,
+      "step": 368110
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.2799801025484037e-05,
+      "loss": 0.8861,
+      "step": 368120
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.279865309558429e-05,
+      "loss": 0.9414,
+      "step": 368130
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.279750516568454e-05,
+      "loss": 0.878,
+      "step": 368140
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.27963572357848e-05,
+      "loss": 0.8698,
+      "step": 368150
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.279520930588505e-05,
+      "loss": 0.8578,
+      "step": 368160
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.2794061375985306e-05,
+      "loss": 0.9922,
+      "step": 368170
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.2792913446085556e-05,
+      "loss": 0.9482,
+      "step": 368180
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.279176551618581e-05,
+      "loss": 0.9192,
+      "step": 368190
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.279061758628606e-05,
+      "loss": 0.8919,
+      "step": 368200
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.278946965638631e-05,
+      "loss": 0.9528,
+      "step": 368210
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 3.278832172648657e-05,
+      "loss": 0.8032,
+      "step": 368220
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.278717379658682e-05,
+      "loss": 0.9548,
+      "step": 368230
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.278602586668707e-05,
+      "loss": 0.9713,
+      "step": 368240
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.2784877936787325e-05,
+      "loss": 0.879,
+      "step": 368250
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.2783730006887575e-05,
+      "loss": 0.9613,
+      "step": 368260
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.2782582076987825e-05,
+      "loss": 0.9092,
+      "step": 368270
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.278143414708808e-05,
+      "loss": 0.8901,
+      "step": 368280
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.278028621718833e-05,
+      "loss": 0.9634,
+      "step": 368290
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.277913828728859e-05,
+      "loss": 0.83,
+      "step": 368300
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.277799035738884e-05,
+      "loss": 0.9145,
+      "step": 368310
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.2776842427489095e-05,
+      "loss": 0.8584,
+      "step": 368320
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.2775694497589344e-05,
+      "loss": 0.9382,
+      "step": 368330
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.27745465676896e-05,
+      "loss": 0.9259,
+      "step": 368340
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.277339863778985e-05,
+      "loss": 0.9638,
+      "step": 368350
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.27722507078901e-05,
+      "loss": 0.8535,
+      "step": 368360
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.277110277799036e-05,
+      "loss": 0.9501,
+      "step": 368370
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.276995484809061e-05,
+      "loss": 0.8539,
+      "step": 368380
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.276880691819086e-05,
+      "loss": 0.8664,
+      "step": 368390
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.2767658988291114e-05,
+      "loss": 0.9933,
+      "step": 368400
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.2766511058391364e-05,
+      "loss": 0.9874,
+      "step": 368410
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.2765363128491613e-05,
+      "loss": 0.7763,
+      "step": 368420
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.276421519859187e-05,
+      "loss": 0.9024,
+      "step": 368430
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 3.276306726869212e-05,
+      "loss": 0.8687,
+      "step": 368440
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.2761919338792377e-05,
+      "loss": 0.9124,
+      "step": 368450
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.2760771408892626e-05,
+      "loss": 0.8899,
+      "step": 368460
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.275962347899288e-05,
+      "loss": 0.9497,
+      "step": 368470
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.275847554909313e-05,
+      "loss": 0.8778,
+      "step": 368480
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.275732761919339e-05,
+      "loss": 0.8951,
+      "step": 368490
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.275617968929364e-05,
+      "loss": 0.8676,
+      "step": 368500
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.275503175939389e-05,
+      "loss": 0.9098,
+      "step": 368510
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.2753883829494146e-05,
+      "loss": 0.9365,
+      "step": 368520
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.2752735899594396e-05,
+      "loss": 0.8513,
+      "step": 368530
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.2751587969694646e-05,
+      "loss": 0.8671,
+      "step": 368540
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.27504400397949e-05,
+      "loss": 0.9225,
+      "step": 368550
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.274929210989515e-05,
+      "loss": 0.842,
+      "step": 368560
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.27481441799954e-05,
+      "loss": 0.9829,
+      "step": 368570
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.274699625009566e-05,
+      "loss": 0.8754,
+      "step": 368580
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.274584832019591e-05,
+      "loss": 0.9085,
+      "step": 368590
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.2744700390296165e-05,
+      "loss": 0.9214,
+      "step": 368600
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.2743552460396415e-05,
+      "loss": 0.9058,
+      "step": 368610
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.274240453049667e-05,
+      "loss": 0.9492,
+      "step": 368620
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.274125660059692e-05,
+      "loss": 0.9546,
+      "step": 368630
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.274010867069718e-05,
+      "loss": 0.8757,
+      "step": 368640
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.273896074079743e-05,
+      "loss": 0.8963,
+      "step": 368650
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 3.273781281089768e-05,
+      "loss": 0.9883,
+      "step": 368660
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.2736664880997934e-05,
+      "loss": 0.9068,
+      "step": 368670
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.2735516951098184e-05,
+      "loss": 0.9292,
+      "step": 368680
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.2734369021198434e-05,
+      "loss": 0.9474,
+      "step": 368690
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.2733221091298684e-05,
+      "loss": 0.962,
+      "step": 368700
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.273207316139894e-05,
+      "loss": 0.8146,
+      "step": 368710
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.273092523149919e-05,
+      "loss": 0.8136,
+      "step": 368720
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.272977730159945e-05,
+      "loss": 0.8488,
+      "step": 368730
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.27286293716997e-05,
+      "loss": 0.9432,
+      "step": 368740
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.2727481441799954e-05,
+      "loss": 0.9646,
+      "step": 368750
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.2726333511900203e-05,
+      "loss": 0.9203,
+      "step": 368760
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.272518558200046e-05,
+      "loss": 0.9245,
+      "step": 368770
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.272403765210071e-05,
+      "loss": 0.9785,
+      "step": 368780
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.272288972220096e-05,
+      "loss": 0.8313,
+      "step": 368790
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.2721741792301216e-05,
+      "loss": 0.9082,
+      "step": 368800
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.2720593862401466e-05,
+      "loss": 1.0148,
+      "step": 368810
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.2719445932501716e-05,
+      "loss": 0.8074,
+      "step": 368820
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.271829800260197e-05,
+      "loss": 1.0013,
+      "step": 368830
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.271715007270222e-05,
+      "loss": 0.8723,
+      "step": 368840
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.271600214280247e-05,
+      "loss": 0.8971,
+      "step": 368850
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.271485421290273e-05,
+      "loss": 1.0026,
+      "step": 368860
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.271370628300298e-05,
+      "loss": 0.8417,
+      "step": 368870
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 3.2712558353103236e-05,
+      "loss": 0.9038,
+      "step": 368880
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.2711410423203485e-05,
+      "loss": 0.9066,
+      "step": 368890
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.271026249330374e-05,
+      "loss": 0.9299,
+      "step": 368900
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.270911456340399e-05,
+      "loss": 0.9427,
+      "step": 368910
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.270796663350425e-05,
+      "loss": 0.8819,
+      "step": 368920
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.27068187036045e-05,
+      "loss": 1.0388,
+      "step": 368930
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.270567077370475e-05,
+      "loss": 1.0098,
+      "step": 368940
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.2704522843805005e-05,
+      "loss": 0.7866,
+      "step": 368950
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.2703374913905255e-05,
+      "loss": 0.9055,
+      "step": 368960
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.2702226984005505e-05,
+      "loss": 0.8331,
+      "step": 368970
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.270107905410576e-05,
+      "loss": 0.9164,
+      "step": 368980
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.269993112420601e-05,
+      "loss": 0.9629,
+      "step": 368990
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.269878319430626e-05,
+      "loss": 0.9081,
+      "step": 369000
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.269763526440652e-05,
+      "loss": 0.9425,
+      "step": 369010
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.269648733450677e-05,
+      "loss": 0.973,
+      "step": 369020
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.2695339404607024e-05,
+      "loss": 0.9358,
+      "step": 369030
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.2694191474707274e-05,
+      "loss": 0.9182,
+      "step": 369040
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.269304354480753e-05,
+      "loss": 0.8982,
+      "step": 369050
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.269189561490778e-05,
+      "loss": 0.8735,
+      "step": 369060
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.269074768500804e-05,
+      "loss": 0.8985,
+      "step": 369070
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.268959975510829e-05,
+      "loss": 0.9308,
+      "step": 369080
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.268845182520854e-05,
+      "loss": 0.8493,
+      "step": 369090
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 3.268730389530879e-05,
+      "loss": 0.8966,
+      "step": 369100
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.268615596540904e-05,
+      "loss": 0.8843,
+      "step": 369110
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.268500803550929e-05,
+      "loss": 0.8837,
+      "step": 369120
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.268386010560955e-05,
+      "loss": 0.9093,
+      "step": 369130
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.26827121757098e-05,
+      "loss": 0.9319,
+      "step": 369140
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.268156424581005e-05,
+      "loss": 0.8894,
+      "step": 369150
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.2680416315910306e-05,
+      "loss": 0.8122,
+      "step": 369160
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.2679268386010556e-05,
+      "loss": 0.9061,
+      "step": 369170
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.267812045611081e-05,
+      "loss": 0.9308,
+      "step": 369180
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.267697252621106e-05,
+      "loss": 0.9485,
+      "step": 369190
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.267582459631132e-05,
+      "loss": 0.8921,
+      "step": 369200
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.267467666641157e-05,
+      "loss": 0.8834,
+      "step": 369210
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.2673528736511825e-05,
+      "loss": 0.8483,
+      "step": 369220
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.2672380806612075e-05,
+      "loss": 0.8669,
+      "step": 369230
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.2671232876712325e-05,
+      "loss": 0.8883,
+      "step": 369240
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.267008494681258e-05,
+      "loss": 0.9287,
+      "step": 369250
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.266893701691283e-05,
+      "loss": 0.8351,
+      "step": 369260
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.266778908701308e-05,
+      "loss": 0.9552,
+      "step": 369270
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.266664115711334e-05,
+      "loss": 0.9405,
+      "step": 369280
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.266549322721359e-05,
+      "loss": 0.9172,
+      "step": 369290
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.266434529731384e-05,
+      "loss": 0.9138,
+      "step": 369300
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 3.2663197367414094e-05,
+      "loss": 0.9676,
+      "step": 369310
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.2662049437514344e-05,
+      "loss": 1.0425,
+      "step": 369320
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.26609015076146e-05,
+      "loss": 0.8206,
+      "step": 369330
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.265975357771485e-05,
+      "loss": 0.9077,
+      "step": 369340
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.265860564781511e-05,
+      "loss": 0.8852,
+      "step": 369350
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.265745771791536e-05,
+      "loss": 0.8644,
+      "step": 369360
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.2656309788015614e-05,
+      "loss": 0.9603,
+      "step": 369370
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.2655161858115864e-05,
+      "loss": 0.84,
+      "step": 369380
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.2654013928216114e-05,
+      "loss": 0.8809,
+      "step": 369390
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.2652865998316363e-05,
+      "loss": 0.8784,
+      "step": 369400
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.265171806841662e-05,
+      "loss": 0.8011,
+      "step": 369410
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.265057013851687e-05,
+      "loss": 0.8548,
+      "step": 369420
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.264942220861712e-05,
+      "loss": 0.8922,
+      "step": 369430
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.2648274278717376e-05,
+      "loss": 0.8359,
+      "step": 369440
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.2647126348817626e-05,
+      "loss": 0.9733,
+      "step": 369450
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.264597841891788e-05,
+      "loss": 0.8858,
+      "step": 369460
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.264483048901813e-05,
+      "loss": 0.8993,
+      "step": 369470
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.264368255911839e-05,
+      "loss": 0.8611,
+      "step": 369480
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.264253462921864e-05,
+      "loss": 0.8706,
+      "step": 369490
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.2641386699318896e-05,
+      "loss": 0.9066,
+      "step": 369500
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.2640238769419146e-05,
+      "loss": 0.9678,
+      "step": 369510
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.2639090839519396e-05,
+      "loss": 0.8247,
+      "step": 369520
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 3.263794290961965e-05,
+      "loss": 0.9353,
+      "step": 369530
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.26367949797199e-05,
+      "loss": 0.9437,
+      "step": 369540
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.263564704982015e-05,
+      "loss": 0.8475,
+      "step": 369550
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.263449911992041e-05,
+      "loss": 0.8931,
+      "step": 369560
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.263335119002066e-05,
+      "loss": 0.8134,
+      "step": 369570
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.263220326012091e-05,
+      "loss": 0.8332,
+      "step": 369580
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.2631055330221165e-05,
+      "loss": 0.9507,
+      "step": 369590
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.2629907400321415e-05,
+      "loss": 0.9553,
+      "step": 369600
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.262875947042167e-05,
+      "loss": 0.9765,
+      "step": 369610
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.262761154052192e-05,
+      "loss": 0.8666,
+      "step": 369620
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.262646361062218e-05,
+      "loss": 0.9265,
+      "step": 369630
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.262531568072243e-05,
+      "loss": 0.9492,
+      "step": 369640
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.2624167750822684e-05,
+      "loss": 0.8322,
+      "step": 369650
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.2623019820922934e-05,
+      "loss": 0.8215,
+      "step": 369660
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.2621871891023184e-05,
+      "loss": 1.016,
+      "step": 369670
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.262072396112344e-05,
+      "loss": 0.8696,
+      "step": 369680
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.261957603122369e-05,
+      "loss": 0.809,
+      "step": 369690
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.261842810132394e-05,
+      "loss": 0.8892,
+      "step": 369700
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.26172801714242e-05,
+      "loss": 0.9889,
+      "step": 369710
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.261613224152445e-05,
+      "loss": 0.8316,
+      "step": 369720
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.26149843116247e-05,
+      "loss": 0.9223,
+      "step": 369730
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.2613836381724953e-05,
+      "loss": 0.8761,
+      "step": 369740
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.26126884518252e-05,
+      "loss": 0.9876,
+      "step": 369750
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.261154052192546e-05,
+      "loss": 0.7908,
+      "step": 369760
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.261039259202571e-05,
+      "loss": 0.9573,
+      "step": 369770
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.2609244662125966e-05,
+      "loss": 0.941,
+      "step": 369780
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.2608096732226216e-05,
+      "loss": 0.9068,
+      "step": 369790
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.260694880232647e-05,
+      "loss": 0.9039,
+      "step": 369800
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.260580087242672e-05,
+      "loss": 0.8059,
+      "step": 369810
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.260465294252697e-05,
+      "loss": 0.9033,
+      "step": 369820
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.260350501262723e-05,
+      "loss": 0.8336,
+      "step": 369830
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.260235708272748e-05,
+      "loss": 0.8349,
+      "step": 369840
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.260120915282773e-05,
+      "loss": 0.8449,
+      "step": 369850
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.2600061222927986e-05,
+      "loss": 0.9035,
+      "step": 369860
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.2598913293028235e-05,
+      "loss": 0.8159,
+      "step": 369870
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.2597765363128485e-05,
+      "loss": 0.8974,
+      "step": 369880
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.259661743322874e-05,
+      "loss": 0.9012,
+      "step": 369890
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.259546950332899e-05,
+      "loss": 0.9688,
+      "step": 369900
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.259432157342925e-05,
+      "loss": 0.9147,
+      "step": 369910
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.25931736435295e-05,
+      "loss": 0.9633,
+      "step": 369920
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.2592025713629755e-05,
+      "loss": 0.9101,
+      "step": 369930
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.2590877783730005e-05,
+      "loss": 0.9268,
+      "step": 369940
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.258972985383026e-05,
+      "loss": 0.825,
+      "step": 369950
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.258858192393051e-05,
+      "loss": 0.9167,
+      "step": 369960
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 3.258743399403076e-05,
+      "loss": 0.9464,
+      "step": 369970
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.258628606413102e-05,
+      "loss": 0.9076,
+      "step": 369980
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.258513813423127e-05,
+      "loss": 1.0231,
+      "step": 369990
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.258399020433152e-05,
+      "loss": 0.8421,
+      "step": 370000
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.2582842274431774e-05,
+      "loss": 0.9047,
+      "step": 370010
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.2581694344532024e-05,
+      "loss": 0.8732,
+      "step": 370020
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.2580546414632274e-05,
+      "loss": 0.9519,
+      "step": 370030
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.257939848473253e-05,
+      "loss": 1.0186,
+      "step": 370040
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.257825055483278e-05,
+      "loss": 0.8763,
+      "step": 370050
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.257710262493304e-05,
+      "loss": 0.9121,
+      "step": 370060
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.257595469503329e-05,
+      "loss": 0.9441,
+      "step": 370070
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.257480676513354e-05,
+      "loss": 0.8976,
+      "step": 370080
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.257365883523379e-05,
+      "loss": 0.8907,
+      "step": 370090
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.257251090533404e-05,
+      "loss": 0.9031,
+      "step": 370100
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.25713629754343e-05,
+      "loss": 0.9083,
+      "step": 370110
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.257021504553455e-05,
+      "loss": 0.9946,
+      "step": 370120
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.25690671156348e-05,
+      "loss": 0.9785,
+      "step": 370130
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.2567919185735056e-05,
+      "loss": 0.93,
+      "step": 370140
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.2566771255835306e-05,
+      "loss": 0.9397,
+      "step": 370150
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.2565623325935556e-05,
+      "loss": 0.9592,
+      "step": 370160
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.256447539603581e-05,
+      "loss": 0.9837,
+      "step": 370170
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.256332746613606e-05,
+      "loss": 0.882,
+      "step": 370180
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 3.256217953623632e-05,
+      "loss": 0.9572,
+      "step": 370190
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.256103160633657e-05,
+      "loss": 0.851,
+      "step": 370200
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.2559883676436825e-05,
+      "loss": 0.8959,
+      "step": 370210
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.2558735746537075e-05,
+      "loss": 0.8773,
+      "step": 370220
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.255758781663733e-05,
+      "loss": 0.8724,
+      "step": 370230
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.255643988673758e-05,
+      "loss": 0.9855,
+      "step": 370240
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.255529195683783e-05,
+      "loss": 0.8653,
+      "step": 370250
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.255414402693809e-05,
+      "loss": 0.9219,
+      "step": 370260
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.255299609703834e-05,
+      "loss": 0.872,
+      "step": 370270
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.255184816713859e-05,
+      "loss": 0.9009,
+      "step": 370280
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.2550700237238844e-05,
+      "loss": 0.9387,
+      "step": 370290
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.2549552307339094e-05,
+      "loss": 0.9329,
+      "step": 370300
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.2548404377439344e-05,
+      "loss": 0.9356,
+      "step": 370310
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.25472564475396e-05,
+      "loss": 0.9307,
+      "step": 370320
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.254610851763985e-05,
+      "loss": 0.9237,
+      "step": 370330
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.254496058774011e-05,
+      "loss": 0.9827,
+      "step": 370340
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.254381265784036e-05,
+      "loss": 1.0836,
+      "step": 370350
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.2542664727940614e-05,
+      "loss": 0.8842,
+      "step": 370360
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.2541516798040864e-05,
+      "loss": 0.8913,
+      "step": 370370
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.254036886814112e-05,
+      "loss": 0.9036,
+      "step": 370380
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.253922093824137e-05,
+      "loss": 0.9065,
+      "step": 370390
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 3.253807300834162e-05,
+      "loss": 0.983,
+      "step": 370400
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.2536925078441877e-05,
+      "loss": 0.8907,
+      "step": 370410
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.2535777148542126e-05,
+      "loss": 0.8969,
+      "step": 370420
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.2534629218642376e-05,
+      "loss": 0.9072,
+      "step": 370430
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.253348128874263e-05,
+      "loss": 0.9718,
+      "step": 370440
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.253233335884288e-05,
+      "loss": 0.8566,
+      "step": 370450
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.253118542894313e-05,
+      "loss": 0.9388,
+      "step": 370460
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.253003749904339e-05,
+      "loss": 0.9149,
+      "step": 370470
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.252888956914364e-05,
+      "loss": 0.9211,
+      "step": 370480
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.2527741639243896e-05,
+      "loss": 0.832,
+      "step": 370490
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.2526593709344146e-05,
+      "loss": 0.8394,
+      "step": 370500
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.25254457794444e-05,
+      "loss": 0.9162,
+      "step": 370510
+    },
+    {
+      "epoch": 17.0,
+      "eval_cer": 0.6684674753222445,
+      "eval_em": 0.007454017424975799,
+      "eval_f1": 0.007454017424975799,
+      "eval_loss": 1.2607142925262451,
+      "eval_runtime": 1515.2677,
+      "eval_samples_per_second": 6.817,
+      "eval_steps_per_second": 0.853,
+      "eval_wer": 0.9925459825750242,
+      "step": 370518
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.252429784954465e-05,
+      "loss": 0.9288,
+      "step": 370520
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.252314991964491e-05,
+      "loss": 0.8305,
+      "step": 370530
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.252200198974516e-05,
+      "loss": 0.7884,
+      "step": 370540
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.252085405984541e-05,
+      "loss": 0.8488,
+      "step": 370550
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.2519706129945665e-05,
+      "loss": 0.9273,
+      "step": 370560
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.2518558200045915e-05,
+      "loss": 0.8945,
+      "step": 370570
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.2517410270146165e-05,
+      "loss": 0.9097,
+      "step": 370580
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.251626234024642e-05,
+      "loss": 0.8583,
+      "step": 370590
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.251511441034667e-05,
+      "loss": 0.7504,
+      "step": 370600
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.251396648044692e-05,
+      "loss": 0.8945,
+      "step": 370610
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 3.251281855054718e-05,
+      "loss": 0.9053,
+      "step": 370620
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.251167062064743e-05,
+      "loss": 0.7759,
+      "step": 370630
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.2510522690747684e-05,
+      "loss": 0.9616,
+      "step": 370640
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.2509374760847934e-05,
+      "loss": 0.8688,
+      "step": 370650
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.250822683094819e-05,
+      "loss": 0.8353,
+      "step": 370660
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.250707890104844e-05,
+      "loss": 0.8531,
+      "step": 370670
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.25059309711487e-05,
+      "loss": 0.8321,
+      "step": 370680
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.250478304124895e-05,
+      "loss": 0.9059,
+      "step": 370690
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.25036351113492e-05,
+      "loss": 0.8299,
+      "step": 370700
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.2502487181449454e-05,
+      "loss": 0.8041,
+      "step": 370710
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.2501339251549703e-05,
+      "loss": 0.7876,
+      "step": 370720
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.250019132164995e-05,
+      "loss": 0.8589,
+      "step": 370730
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.249904339175021e-05,
+      "loss": 0.8955,
+      "step": 370740
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.249789546185046e-05,
+      "loss": 0.8347,
+      "step": 370750
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.249674753195071e-05,
+      "loss": 0.8788,
+      "step": 370760
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.2495599602050966e-05,
+      "loss": 0.8344,
+      "step": 370770
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.2494451672151216e-05,
+      "loss": 0.821,
+      "step": 370780
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.249330374225147e-05,
+      "loss": 0.8342,
+      "step": 370790
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.249215581235172e-05,
+      "loss": 0.8455,
+      "step": 370800
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.249100788245198e-05,
+      "loss": 0.8689,
+      "step": 370810
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.248985995255223e-05,
+      "loss": 0.9132,
+      "step": 370820
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.248871202265248e-05,
+      "loss": 0.8163,
+      "step": 370830
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 3.2487564092752736e-05,
+      "loss": 0.8945,
+      "step": 370840
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.2486416162852985e-05,
+      "loss": 0.7849,
+      "step": 370850
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.2485268232953235e-05,
+      "loss": 0.8369,
+      "step": 370860
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.248412030305349e-05,
+      "loss": 0.8667,
+      "step": 370870
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.248297237315374e-05,
+      "loss": 0.84,
+      "step": 370880
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.248182444325399e-05,
+      "loss": 0.9128,
+      "step": 370890
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.248067651335425e-05,
+      "loss": 0.8824,
+      "step": 370900
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.24795285834545e-05,
+      "loss": 0.7407,
+      "step": 370910
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.2478380653554755e-05,
+      "loss": 0.8387,
+      "step": 370920
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.2477232723655005e-05,
+      "loss": 0.8483,
+      "step": 370930
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.247608479375526e-05,
+      "loss": 0.8787,
+      "step": 370940
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.247493686385551e-05,
+      "loss": 0.883,
+      "step": 370950
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.247378893395577e-05,
+      "loss": 0.8227,
+      "step": 370960
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.247264100405602e-05,
+      "loss": 0.8569,
+      "step": 370970
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.247149307415627e-05,
+      "loss": 0.8908,
+      "step": 370980
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.2470345144256524e-05,
+      "loss": 0.8478,
+      "step": 370990
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.2469197214356774e-05,
+      "loss": 0.8492,
+      "step": 371000
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.2468049284457024e-05,
+      "loss": 0.8826,
+      "step": 371010
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.246690135455728e-05,
+      "loss": 0.8711,
+      "step": 371020
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.246575342465753e-05,
+      "loss": 0.8385,
+      "step": 371030
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.246460549475778e-05,
+      "loss": 0.8094,
+      "step": 371040
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.246345756485804e-05,
+      "loss": 0.8396,
+      "step": 371050
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 3.2462309634958287e-05,
+      "loss": 0.8647,
+      "step": 371060
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.246116170505854e-05,
+      "loss": 0.8125,
+      "step": 371070
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.246001377515879e-05,
+      "loss": 0.8929,
+      "step": 371080
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.245886584525905e-05,
+      "loss": 0.9063,
+      "step": 371090
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.24577179153593e-05,
+      "loss": 0.9148,
+      "step": 371100
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.2456569985459556e-05,
+      "loss": 0.8099,
+      "step": 371110
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.2455422055559806e-05,
+      "loss": 0.9073,
+      "step": 371120
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.2454274125660056e-05,
+      "loss": 0.8357,
+      "step": 371130
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.245312619576031e-05,
+      "loss": 0.9327,
+      "step": 371140
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.245197826586056e-05,
+      "loss": 0.812,
+      "step": 371150
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.245083033596081e-05,
+      "loss": 0.8133,
+      "step": 371160
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.244968240606107e-05,
+      "loss": 0.8889,
+      "step": 371170
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.244853447616132e-05,
+      "loss": 0.8252,
+      "step": 371180
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.244738654626157e-05,
+      "loss": 0.8468,
+      "step": 371190
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.2446238616361825e-05,
+      "loss": 0.8188,
+      "step": 371200
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.2445090686462075e-05,
+      "loss": 0.8213,
+      "step": 371210
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.244394275656233e-05,
+      "loss": 0.8736,
+      "step": 371220
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.244279482666258e-05,
+      "loss": 0.8412,
+      "step": 371230
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.244164689676284e-05,
+      "loss": 0.7835,
+      "step": 371240
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.244049896686309e-05,
+      "loss": 0.9237,
+      "step": 371250
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.2439351036963345e-05,
+      "loss": 0.8294,
+      "step": 371260
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.2438203107063594e-05,
+      "loss": 0.8823,
+      "step": 371270
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.2437055177163844e-05,
+      "loss": 0.8429,
+      "step": 371280
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.24359072472641e-05,
+      "loss": 0.8397,
+      "step": 371290
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.243475931736435e-05,
+      "loss": 0.8024,
+      "step": 371300
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.24336113874646e-05,
+      "loss": 0.8976,
+      "step": 371310
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.243246345756486e-05,
+      "loss": 0.8433,
+      "step": 371320
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.243131552766511e-05,
+      "loss": 0.9163,
+      "step": 371330
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.243016759776536e-05,
+      "loss": 0.9755,
+      "step": 371340
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.2429019667865614e-05,
+      "loss": 0.9239,
+      "step": 371350
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.2427871737965863e-05,
+      "loss": 0.8345,
+      "step": 371360
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.242672380806612e-05,
+      "loss": 0.8201,
+      "step": 371370
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.242557587816637e-05,
+      "loss": 0.8649,
+      "step": 371380
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.2424427948266627e-05,
+      "loss": 0.8221,
+      "step": 371390
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.2423280018366876e-05,
+      "loss": 0.896,
+      "step": 371400
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.242213208846713e-05,
+      "loss": 0.8682,
+      "step": 371410
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.242098415856738e-05,
+      "loss": 0.9056,
+      "step": 371420
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.241983622866763e-05,
+      "loss": 0.8886,
+      "step": 371430
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.241868829876789e-05,
+      "loss": 0.8203,
+      "step": 371440
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.241754036886814e-05,
+      "loss": 0.9919,
+      "step": 371450
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.241639243896839e-05,
+      "loss": 0.7999,
+      "step": 371460
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.241524450906864e-05,
+      "loss": 0.8667,
+      "step": 371470
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.2414096579168896e-05,
+      "loss": 0.7922,
+      "step": 371480
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 3.2412948649269145e-05,
+      "loss": 0.8944,
+      "step": 371490
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.24118007193694e-05,
+      "loss": 0.8038,
+      "step": 371500
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.241065278946965e-05,
+      "loss": 0.8126,
+      "step": 371510
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.240950485956991e-05,
+      "loss": 0.9624,
+      "step": 371520
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.240835692967016e-05,
+      "loss": 0.7567,
+      "step": 371530
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.2407208999770415e-05,
+      "loss": 0.8441,
+      "step": 371540
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.2406061069870665e-05,
+      "loss": 0.8484,
+      "step": 371550
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.2404913139970915e-05,
+      "loss": 0.8572,
+      "step": 371560
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.240376521007117e-05,
+      "loss": 0.9326,
+      "step": 371570
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.240261728017142e-05,
+      "loss": 0.7984,
+      "step": 371580
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.240146935027167e-05,
+      "loss": 0.856,
+      "step": 371590
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.240032142037193e-05,
+      "loss": 0.811,
+      "step": 371600
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.239917349047218e-05,
+      "loss": 0.7992,
+      "step": 371610
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.239802556057243e-05,
+      "loss": 0.8881,
+      "step": 371620
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.2396877630672684e-05,
+      "loss": 0.7435,
+      "step": 371630
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.2395729700772934e-05,
+      "loss": 0.8844,
+      "step": 371640
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.239458177087319e-05,
+      "loss": 0.9024,
+      "step": 371650
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.239343384097344e-05,
+      "loss": 0.8551,
+      "step": 371660
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.23922859110737e-05,
+      "loss": 0.9742,
+      "step": 371670
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.239113798117395e-05,
+      "loss": 0.8626,
+      "step": 371680
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.2389990051274204e-05,
+      "loss": 0.9215,
+      "step": 371690
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.2388842121374453e-05,
+      "loss": 0.8348,
+      "step": 371700
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 3.23876941914747e-05,
+      "loss": 0.8577,
+      "step": 371710
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.238654626157496e-05,
+      "loss": 0.883,
+      "step": 371720
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.238539833167521e-05,
+      "loss": 0.8765,
+      "step": 371730
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.238425040177546e-05,
+      "loss": 0.8271,
+      "step": 371740
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.2383102471875716e-05,
+      "loss": 0.9416,
+      "step": 371750
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.2381954541975966e-05,
+      "loss": 0.8421,
+      "step": 371760
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.2380806612076216e-05,
+      "loss": 0.8983,
+      "step": 371770
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.237965868217647e-05,
+      "loss": 0.8066,
+      "step": 371780
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.237851075227672e-05,
+      "loss": 0.8233,
+      "step": 371790
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.237736282237698e-05,
+      "loss": 0.8709,
+      "step": 371800
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.237621489247723e-05,
+      "loss": 0.8942,
+      "step": 371810
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.2375066962577486e-05,
+      "loss": 0.906,
+      "step": 371820
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.2373919032677735e-05,
+      "loss": 0.8648,
+      "step": 371830
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.237277110277799e-05,
+      "loss": 0.7737,
+      "step": 371840
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.237162317287824e-05,
+      "loss": 0.7855,
+      "step": 371850
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.237047524297849e-05,
+      "loss": 0.9033,
+      "step": 371860
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.236932731307875e-05,
+      "loss": 0.9314,
+      "step": 371870
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.2368179383179e-05,
+      "loss": 0.8516,
+      "step": 371880
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.236703145327925e-05,
+      "loss": 0.9251,
+      "step": 371890
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.2365883523379505e-05,
+      "loss": 0.8022,
+      "step": 371900
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.2364735593479755e-05,
+      "loss": 0.7714,
+      "step": 371910
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.2363587663580004e-05,
+      "loss": 0.8655,
+      "step": 371920
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 3.236243973368026e-05,
+      "loss": 0.8581,
+      "step": 371930
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.236129180378051e-05,
+      "loss": 0.8821,
+      "step": 371940
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.236014387388077e-05,
+      "loss": 0.9141,
+      "step": 371950
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.235899594398102e-05,
+      "loss": 0.8847,
+      "step": 371960
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.2357848014081274e-05,
+      "loss": 0.8344,
+      "step": 371970
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.2356700084181524e-05,
+      "loss": 0.8159,
+      "step": 371980
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.235555215428178e-05,
+      "loss": 0.8364,
+      "step": 371990
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.235440422438203e-05,
+      "loss": 0.8639,
+      "step": 372000
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.235325629448228e-05,
+      "loss": 0.887,
+      "step": 372010
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.235210836458254e-05,
+      "loss": 0.8667,
+      "step": 372020
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.235096043468279e-05,
+      "loss": 0.8194,
+      "step": 372030
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.2349812504783037e-05,
+      "loss": 0.8607,
+      "step": 372040
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.234866457488329e-05,
+      "loss": 0.9705,
+      "step": 372050
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.234751664498354e-05,
+      "loss": 0.8652,
+      "step": 372060
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.234636871508379e-05,
+      "loss": 0.8376,
+      "step": 372070
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.234522078518405e-05,
+      "loss": 0.8516,
+      "step": 372080
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.23440728552843e-05,
+      "loss": 0.7957,
+      "step": 372090
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.2342924925384556e-05,
+      "loss": 0.7746,
+      "step": 372100
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.2341776995484806e-05,
+      "loss": 0.8694,
+      "step": 372110
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.234062906558506e-05,
+      "loss": 0.8401,
+      "step": 372120
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.233948113568531e-05,
+      "loss": 0.7907,
+      "step": 372130
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.233833320578556e-05,
+      "loss": 0.8488,
+      "step": 372140
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 3.233718527588582e-05,
+      "loss": 0.9714,
+      "step": 372150
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.233603734598607e-05,
+      "loss": 0.8835,
+      "step": 372160
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.233488941608632e-05,
+      "loss": 0.8752,
+      "step": 372170
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.2333741486186575e-05,
+      "loss": 0.8943,
+      "step": 372180
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.2332593556286825e-05,
+      "loss": 0.758,
+      "step": 372190
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.2331445626387075e-05,
+      "loss": 0.9589,
+      "step": 372200
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.233029769648733e-05,
+      "loss": 0.9085,
+      "step": 372210
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.232914976658758e-05,
+      "loss": 0.7986,
+      "step": 372220
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.232800183668784e-05,
+      "loss": 0.9227,
+      "step": 372230
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.232685390678809e-05,
+      "loss": 0.8471,
+      "step": 372240
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.2325705976888344e-05,
+      "loss": 0.8533,
+      "step": 372250
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.2324558046988594e-05,
+      "loss": 0.8377,
+      "step": 372260
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.232341011708885e-05,
+      "loss": 0.803,
+      "step": 372270
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.23222621871891e-05,
+      "loss": 0.9143,
+      "step": 372280
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.232111425728935e-05,
+      "loss": 0.8824,
+      "step": 372290
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.231996632738961e-05,
+      "loss": 0.8709,
+      "step": 372300
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.231881839748986e-05,
+      "loss": 0.8743,
+      "step": 372310
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.231767046759011e-05,
+      "loss": 0.8193,
+      "step": 372320
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.2316522537690364e-05,
+      "loss": 0.7903,
+      "step": 372330
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.2315374607790613e-05,
+      "loss": 0.8941,
+      "step": 372340
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.231422667789086e-05,
+      "loss": 0.8626,
+      "step": 372350
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.231307874799112e-05,
+      "loss": 0.8908,
+      "step": 372360
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 3.231193081809137e-05,
+      "loss": 0.8943,
+      "step": 372370
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.2310782888191626e-05,
+      "loss": 0.9013,
+      "step": 372380
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.2309634958291876e-05,
+      "loss": 0.8426,
+      "step": 372390
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.230848702839213e-05,
+      "loss": 0.8557,
+      "step": 372400
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.230733909849238e-05,
+      "loss": 0.8618,
+      "step": 372410
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.230619116859264e-05,
+      "loss": 0.8052,
+      "step": 372420
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.230504323869289e-05,
+      "loss": 0.8045,
+      "step": 372430
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.230389530879314e-05,
+      "loss": 0.7595,
+      "step": 372440
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.2302747378893396e-05,
+      "loss": 0.8038,
+      "step": 372450
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.2301599448993646e-05,
+      "loss": 0.9377,
+      "step": 372460
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.2300451519093895e-05,
+      "loss": 0.8556,
+      "step": 372470
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.229930358919415e-05,
+      "loss": 0.8673,
+      "step": 372480
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.22981556592944e-05,
+      "loss": 0.899,
+      "step": 372490
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.229700772939465e-05,
+      "loss": 0.8778,
+      "step": 372500
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.229585979949491e-05,
+      "loss": 0.9276,
+      "step": 372510
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.229471186959516e-05,
+      "loss": 0.7944,
+      "step": 372520
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.2293563939695415e-05,
+      "loss": 0.938,
+      "step": 372530
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.2292416009795665e-05,
+      "loss": 0.8628,
+      "step": 372540
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.229126807989592e-05,
+      "loss": 0.8322,
+      "step": 372550
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.229012014999617e-05,
+      "loss": 0.8493,
+      "step": 372560
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.228897222009643e-05,
+      "loss": 0.8776,
+      "step": 372570
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 3.228782429019668e-05,
+      "loss": 0.8513,
+      "step": 372580
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.228667636029693e-05,
+      "loss": 0.8825,
+      "step": 372590
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.2285528430397184e-05,
+      "loss": 0.84,
+      "step": 372600
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.2284380500497434e-05,
+      "loss": 0.9778,
+      "step": 372610
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.2283232570597684e-05,
+      "loss": 0.9178,
+      "step": 372620
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.228208464069794e-05,
+      "loss": 0.8389,
+      "step": 372630
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.228093671079819e-05,
+      "loss": 0.8847,
+      "step": 372640
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.227978878089844e-05,
+      "loss": 0.9014,
+      "step": 372650
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.22786408509987e-05,
+      "loss": 0.8447,
+      "step": 372660
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.227749292109895e-05,
+      "loss": 0.8831,
+      "step": 372670
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.22763449911992e-05,
+      "loss": 0.8169,
+      "step": 372680
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.227519706129945e-05,
+      "loss": 0.8205,
+      "step": 372690
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.227404913139971e-05,
+      "loss": 0.867,
+      "step": 372700
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.227290120149996e-05,
+      "loss": 0.7203,
+      "step": 372710
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.2271753271600216e-05,
+      "loss": 0.8186,
+      "step": 372720
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.2270605341700466e-05,
+      "loss": 0.873,
+      "step": 372730
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.2269457411800716e-05,
+      "loss": 0.9074,
+      "step": 372740
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.226830948190097e-05,
+      "loss": 0.7971,
+      "step": 372750
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.226716155200122e-05,
+      "loss": 0.8428,
+      "step": 372760
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.226601362210147e-05,
+      "loss": 0.782,
+      "step": 372770
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.226486569220173e-05,
+      "loss": 0.8465,
+      "step": 372780
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.226371776230198e-05,
+      "loss": 0.9406,
+      "step": 372790
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 3.226256983240223e-05,
+      "loss": 0.9307,
+      "step": 372800
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.2261421902502485e-05,
+      "loss": 0.9192,
+      "step": 372810
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.2260273972602735e-05,
+      "loss": 0.8052,
+      "step": 372820
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.225912604270299e-05,
+      "loss": 0.7949,
+      "step": 372830
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.225797811280324e-05,
+      "loss": 0.7717,
+      "step": 372840
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.22568301829035e-05,
+      "loss": 0.8671,
+      "step": 372850
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.225568225300375e-05,
+      "loss": 0.8354,
+      "step": 372860
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.2254534323104e-05,
+      "loss": 0.8932,
+      "step": 372870
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.2253386393204255e-05,
+      "loss": 0.9446,
+      "step": 372880
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.2252238463304505e-05,
+      "loss": 0.8155,
+      "step": 372890
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.2251090533404754e-05,
+      "loss": 0.9343,
+      "step": 372900
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.224994260350501e-05,
+      "loss": 0.8505,
+      "step": 372910
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.224879467360526e-05,
+      "loss": 0.9267,
+      "step": 372920
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.224764674370551e-05,
+      "loss": 0.9017,
+      "step": 372930
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.224649881380577e-05,
+      "loss": 0.8454,
+      "step": 372940
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.224535088390602e-05,
+      "loss": 0.7645,
+      "step": 372950
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.2244202954006274e-05,
+      "loss": 0.8741,
+      "step": 372960
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.2243055024106524e-05,
+      "loss": 0.7906,
+      "step": 372970
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.224190709420678e-05,
+      "loss": 0.822,
+      "step": 372980
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.224075916430703e-05,
+      "loss": 0.9143,
+      "step": 372990
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.223961123440729e-05,
+      "loss": 0.8296,
+      "step": 373000
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.223846330450754e-05,
+      "loss": 0.8113,
+      "step": 373010
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 3.2237315374607787e-05,
+      "loss": 0.8672,
+      "step": 373020
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.223616744470804e-05,
+      "loss": 0.8604,
+      "step": 373030
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.223501951480829e-05,
+      "loss": 0.8089,
+      "step": 373040
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.223387158490854e-05,
+      "loss": 0.8637,
+      "step": 373050
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.22327236550088e-05,
+      "loss": 0.8205,
+      "step": 373060
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.223157572510905e-05,
+      "loss": 0.824,
+      "step": 373070
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.22304277952093e-05,
+      "loss": 0.9473,
+      "step": 373080
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.2229279865309556e-05,
+      "loss": 0.8069,
+      "step": 373090
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.2228131935409806e-05,
+      "loss": 0.889,
+      "step": 373100
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.222698400551006e-05,
+      "loss": 0.8458,
+      "step": 373110
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.222583607561031e-05,
+      "loss": 0.8174,
+      "step": 373120
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.222468814571057e-05,
+      "loss": 0.9015,
+      "step": 373130
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.222354021581082e-05,
+      "loss": 0.7324,
+      "step": 373140
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.2222392285911075e-05,
+      "loss": 0.8746,
+      "step": 373150
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.2221244356011325e-05,
+      "loss": 0.8899,
+      "step": 373160
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.2220096426111575e-05,
+      "loss": 0.7929,
+      "step": 373170
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.221894849621183e-05,
+      "loss": 0.9144,
+      "step": 373180
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.221780056631208e-05,
+      "loss": 0.8496,
+      "step": 373190
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.221665263641233e-05,
+      "loss": 0.8588,
+      "step": 373200
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.221550470651259e-05,
+      "loss": 0.8768,
+      "step": 373210
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.221435677661284e-05,
+      "loss": 0.9374,
+      "step": 373220
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.221320884671309e-05,
+      "loss": 0.8459,
+      "step": 373230
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 3.2212060916813344e-05,
+      "loss": 0.8086,
+      "step": 373240
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.2210912986913594e-05,
+      "loss": 0.7845,
+      "step": 373250
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.220976505701385e-05,
+      "loss": 0.8409,
+      "step": 373260
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.22086171271141e-05,
+      "loss": 0.8472,
+      "step": 373270
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.220746919721436e-05,
+      "loss": 0.9317,
+      "step": 373280
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.220632126731461e-05,
+      "loss": 0.9153,
+      "step": 373290
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.2205173337414864e-05,
+      "loss": 0.8584,
+      "step": 373300
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.2204025407515114e-05,
+      "loss": 0.8672,
+      "step": 373310
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.2202877477615363e-05,
+      "loss": 0.8525,
+      "step": 373320
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.220172954771562e-05,
+      "loss": 0.8863,
+      "step": 373330
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.220058161781587e-05,
+      "loss": 0.8223,
+      "step": 373340
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.219943368791612e-05,
+      "loss": 0.9098,
+      "step": 373350
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.2198285758016376e-05,
+      "loss": 0.8814,
+      "step": 373360
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.2197137828116626e-05,
+      "loss": 0.825,
+      "step": 373370
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.2195989898216876e-05,
+      "loss": 0.8681,
+      "step": 373380
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.219484196831713e-05,
+      "loss": 0.9238,
+      "step": 373390
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.219369403841738e-05,
+      "loss": 0.9233,
+      "step": 373400
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.219254610851764e-05,
+      "loss": 0.796,
+      "step": 373410
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.219139817861789e-05,
+      "loss": 0.8612,
+      "step": 373420
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.2190250248718146e-05,
+      "loss": 0.8835,
+      "step": 373430
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.2189102318818396e-05,
+      "loss": 0.873,
+      "step": 373440
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.218795438891865e-05,
+      "loss": 0.9024,
+      "step": 373450
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 3.21868064590189e-05,
+      "loss": 0.8456,
+      "step": 373460
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.218565852911915e-05,
+      "loss": 0.9558,
+      "step": 373470
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.218451059921941e-05,
+      "loss": 0.9178,
+      "step": 373480
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.218336266931966e-05,
+      "loss": 0.8572,
+      "step": 373490
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.218221473941991e-05,
+      "loss": 0.9264,
+      "step": 373500
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.218106680952016e-05,
+      "loss": 0.9166,
+      "step": 373510
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.2179918879620415e-05,
+      "loss": 0.7794,
+      "step": 373520
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.2178770949720665e-05,
+      "loss": 0.9147,
+      "step": 373530
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.217762301982092e-05,
+      "loss": 0.947,
+      "step": 373540
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.217647508992117e-05,
+      "loss": 0.8096,
+      "step": 373550
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.217532716002143e-05,
+      "loss": 0.8765,
+      "step": 373560
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.217417923012168e-05,
+      "loss": 0.8903,
+      "step": 373570
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.2173031300221934e-05,
+      "loss": 0.9165,
+      "step": 373580
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.2171883370322184e-05,
+      "loss": 0.8999,
+      "step": 373590
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.2170735440422434e-05,
+      "loss": 0.873,
+      "step": 373600
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.216958751052269e-05,
+      "loss": 0.828,
+      "step": 373610
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.216843958062294e-05,
+      "loss": 0.9145,
+      "step": 373620
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.216729165072319e-05,
+      "loss": 0.859,
+      "step": 373630
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.216614372082345e-05,
+      "loss": 0.8682,
+      "step": 373640
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.21649957909237e-05,
+      "loss": 0.8816,
+      "step": 373650
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.2163847861023947e-05,
+      "loss": 0.7637,
+      "step": 373660
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 3.21626999311242e-05,
+      "loss": 0.938,
+      "step": 373670
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.216155200122445e-05,
+      "loss": 0.9124,
+      "step": 373680
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.216040407132471e-05,
+      "loss": 0.8727,
+      "step": 373690
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.215925614142496e-05,
+      "loss": 0.8976,
+      "step": 373700
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.2158108211525216e-05,
+      "loss": 0.9084,
+      "step": 373710
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.2156960281625466e-05,
+      "loss": 0.8996,
+      "step": 373720
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.215581235172572e-05,
+      "loss": 0.8925,
+      "step": 373730
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.215466442182597e-05,
+      "loss": 0.8297,
+      "step": 373740
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.215351649192622e-05,
+      "loss": 0.8439,
+      "step": 373750
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.215236856202648e-05,
+      "loss": 0.8507,
+      "step": 373760
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.215122063212673e-05,
+      "loss": 0.8458,
+      "step": 373770
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.215007270222698e-05,
+      "loss": 0.7736,
+      "step": 373780
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.2148924772327235e-05,
+      "loss": 0.9064,
+      "step": 373790
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.2147776842427485e-05,
+      "loss": 0.8074,
+      "step": 373800
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.2146628912527735e-05,
+      "loss": 0.8144,
+      "step": 373810
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.214548098262799e-05,
+      "loss": 0.8489,
+      "step": 373820
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.214433305272824e-05,
+      "loss": 0.8783,
+      "step": 373830
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.21431851228285e-05,
+      "loss": 0.9063,
+      "step": 373840
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.214203719292875e-05,
+      "loss": 0.8609,
+      "step": 373850
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.2140889263029005e-05,
+      "loss": 0.8845,
+      "step": 373860
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.2139741333129255e-05,
+      "loss": 0.9015,
+      "step": 373870
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.213859340322951e-05,
+      "loss": 0.7664,
+      "step": 373880
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 3.213744547332976e-05,
+      "loss": 0.8302,
+      "step": 373890
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.213629754343001e-05,
+      "loss": 0.9089,
+      "step": 373900
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.213514961353027e-05,
+      "loss": 0.7927,
+      "step": 373910
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.213400168363052e-05,
+      "loss": 0.9196,
+      "step": 373920
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.213285375373077e-05,
+      "loss": 0.8509,
+      "step": 373930
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.2131705823831024e-05,
+      "loss": 0.7985,
+      "step": 373940
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.2130557893931274e-05,
+      "loss": 0.8315,
+      "step": 373950
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.2129409964031524e-05,
+      "loss": 0.8163,
+      "step": 373960
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.212826203413178e-05,
+      "loss": 0.8214,
+      "step": 373970
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.212711410423203e-05,
+      "loss": 0.9125,
+      "step": 373980
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.212596617433229e-05,
+      "loss": 0.8127,
+      "step": 373990
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.2124818244432537e-05,
+      "loss": 0.8586,
+      "step": 374000
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.212367031453279e-05,
+      "loss": 0.8994,
+      "step": 374010
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.212252238463304e-05,
+      "loss": 0.922,
+      "step": 374020
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.21213744547333e-05,
+      "loss": 0.8957,
+      "step": 374030
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.212022652483355e-05,
+      "loss": 0.9094,
+      "step": 374040
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.21190785949338e-05,
+      "loss": 0.8851,
+      "step": 374050
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.2117930665034056e-05,
+      "loss": 0.8818,
+      "step": 374060
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.2116782735134306e-05,
+      "loss": 0.9101,
+      "step": 374070
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.2115634805234556e-05,
+      "loss": 0.8521,
+      "step": 374080
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.211448687533481e-05,
+      "loss": 0.7796,
+      "step": 374090
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.211333894543506e-05,
+      "loss": 0.8904,
+      "step": 374100
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 3.211219101553531e-05,
+      "loss": 0.8114,
+      "step": 374110
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.211104308563557e-05,
+      "loss": 0.8914,
+      "step": 374120
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.210989515573582e-05,
+      "loss": 0.8116,
+      "step": 374130
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.2108747225836075e-05,
+      "loss": 0.9321,
+      "step": 374140
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.2107599295936325e-05,
+      "loss": 0.796,
+      "step": 374150
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.210645136603658e-05,
+      "loss": 0.8441,
+      "step": 374160
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.210530343613683e-05,
+      "loss": 0.8342,
+      "step": 374170
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.210415550623709e-05,
+      "loss": 0.8587,
+      "step": 374180
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.210300757633734e-05,
+      "loss": 0.8926,
+      "step": 374190
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.210185964643759e-05,
+      "loss": 0.9128,
+      "step": 374200
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.210071171653784e-05,
+      "loss": 0.9159,
+      "step": 374210
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.2099563786638094e-05,
+      "loss": 0.9238,
+      "step": 374220
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.2098415856738344e-05,
+      "loss": 0.9058,
+      "step": 374230
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.2097267926838594e-05,
+      "loss": 0.9514,
+      "step": 374240
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.209611999693885e-05,
+      "loss": 0.8539,
+      "step": 374250
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.20949720670391e-05,
+      "loss": 0.7914,
+      "step": 374260
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.209382413713936e-05,
+      "loss": 0.8794,
+      "step": 374270
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.209267620723961e-05,
+      "loss": 0.9082,
+      "step": 374280
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.2091528277339864e-05,
+      "loss": 0.9414,
+      "step": 374290
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.2090380347440113e-05,
+      "loss": 0.8169,
+      "step": 374300
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.208923241754037e-05,
+      "loss": 0.8443,
+      "step": 374310
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.208808448764062e-05,
+      "loss": 0.7986,
+      "step": 374320
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 3.208693655774087e-05,
+      "loss": 0.9502,
+      "step": 374330
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.2085788627841126e-05,
+      "loss": 0.8687,
+      "step": 374340
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.2084640697941376e-05,
+      "loss": 0.9055,
+      "step": 374350
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.2083492768041626e-05,
+      "loss": 0.8076,
+      "step": 374360
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.208234483814188e-05,
+      "loss": 0.8063,
+      "step": 374370
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.208119690824213e-05,
+      "loss": 0.8764,
+      "step": 374380
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.208004897834238e-05,
+      "loss": 0.7712,
+      "step": 374390
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.207890104844264e-05,
+      "loss": 0.7587,
+      "step": 374400
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.207775311854289e-05,
+      "loss": 0.8849,
+      "step": 374410
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.2076605188643146e-05,
+      "loss": 0.8701,
+      "step": 374420
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.2075457258743395e-05,
+      "loss": 0.8191,
+      "step": 374430
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.207430932884365e-05,
+      "loss": 0.9377,
+      "step": 374440
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.20731613989439e-05,
+      "loss": 0.962,
+      "step": 374450
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.207201346904416e-05,
+      "loss": 0.8518,
+      "step": 374460
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.207086553914441e-05,
+      "loss": 0.8196,
+      "step": 374470
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.206971760924466e-05,
+      "loss": 0.9594,
+      "step": 374480
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.2068569679344915e-05,
+      "loss": 0.8235,
+      "step": 374490
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.2067421749445165e-05,
+      "loss": 0.847,
+      "step": 374500
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.2066273819545415e-05,
+      "loss": 0.9281,
+      "step": 374510
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.206512588964567e-05,
+      "loss": 0.859,
+      "step": 374520
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.206397795974592e-05,
+      "loss": 0.8359,
+      "step": 374530
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.206283002984617e-05,
+      "loss": 0.9109,
+      "step": 374540
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 3.206168209994643e-05,
+      "loss": 0.902,
+      "step": 374550
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.206053417004668e-05,
+      "loss": 0.8552,
+      "step": 374560
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.2059386240146934e-05,
+      "loss": 0.9313,
+      "step": 374570
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.2058238310247184e-05,
+      "loss": 0.9392,
+      "step": 374580
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.205709038034744e-05,
+      "loss": 0.9192,
+      "step": 374590
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.205594245044769e-05,
+      "loss": 0.8585,
+      "step": 374600
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.205479452054795e-05,
+      "loss": 0.942,
+      "step": 374610
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.20536465906482e-05,
+      "loss": 0.8866,
+      "step": 374620
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.205249866074845e-05,
+      "loss": 0.8744,
+      "step": 374630
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.20513507308487e-05,
+      "loss": 0.717,
+      "step": 374640
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.205020280094895e-05,
+      "loss": 0.7948,
+      "step": 374650
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.20490548710492e-05,
+      "loss": 0.851,
+      "step": 374660
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.204790694114946e-05,
+      "loss": 0.95,
+      "step": 374670
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.204675901124971e-05,
+      "loss": 0.9259,
+      "step": 374680
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.204561108134996e-05,
+      "loss": 0.8526,
+      "step": 374690
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.2044463151450216e-05,
+      "loss": 0.8811,
+      "step": 374700
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.2043315221550466e-05,
+      "loss": 0.9064,
+      "step": 374710
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.204216729165072e-05,
+      "loss": 0.8262,
+      "step": 374720
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.204101936175097e-05,
+      "loss": 0.8845,
+      "step": 374730
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.203987143185123e-05,
+      "loss": 0.8886,
+      "step": 374740
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.203872350195148e-05,
+      "loss": 0.8934,
+      "step": 374750
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 3.2037575572051736e-05,
+      "loss": 0.9723,
+      "step": 374760
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.2036427642151985e-05,
+      "loss": 0.8595,
+      "step": 374770
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.2035279712252235e-05,
+      "loss": 0.885,
+      "step": 374780
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.203413178235249e-05,
+      "loss": 0.8496,
+      "step": 374790
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.203298385245274e-05,
+      "loss": 0.8728,
+      "step": 374800
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.203183592255299e-05,
+      "loss": 0.9132,
+      "step": 374810
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.203068799265325e-05,
+      "loss": 0.838,
+      "step": 374820
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.20295400627535e-05,
+      "loss": 0.824,
+      "step": 374830
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.202839213285375e-05,
+      "loss": 0.8956,
+      "step": 374840
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.2027244202954005e-05,
+      "loss": 0.8855,
+      "step": 374850
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.2026096273054254e-05,
+      "loss": 0.8431,
+      "step": 374860
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.202494834315451e-05,
+      "loss": 0.8036,
+      "step": 374870
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.202380041325476e-05,
+      "loss": 0.8623,
+      "step": 374880
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.202265248335502e-05,
+      "loss": 0.8756,
+      "step": 374890
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.202150455345527e-05,
+      "loss": 0.8836,
+      "step": 374900
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.202035662355552e-05,
+      "loss": 0.9568,
+      "step": 374910
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.2019208693655774e-05,
+      "loss": 0.9611,
+      "step": 374920
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.2018060763756024e-05,
+      "loss": 0.7672,
+      "step": 374930
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.2016912833856274e-05,
+      "loss": 0.9281,
+      "step": 374940
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.201576490395653e-05,
+      "loss": 0.8731,
+      "step": 374950
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.201461697405678e-05,
+      "loss": 0.8779,
+      "step": 374960
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.201346904415703e-05,
+      "loss": 0.8496,
+      "step": 374970
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 3.2012321114257287e-05,
+      "loss": 0.8609,
+      "step": 374980
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.2011173184357536e-05,
+      "loss": 0.8529,
+      "step": 374990
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.201002525445779e-05,
+      "loss": 0.8508,
+      "step": 375000
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.200887732455804e-05,
+      "loss": 0.8996,
+      "step": 375010
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.20077293946583e-05,
+      "loss": 0.8879,
+      "step": 375020
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.200658146475855e-05,
+      "loss": 0.8042,
+      "step": 375030
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.2005433534858806e-05,
+      "loss": 0.9069,
+      "step": 375040
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.2004285604959056e-05,
+      "loss": 0.8581,
+      "step": 375050
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.2003137675059306e-05,
+      "loss": 0.8447,
+      "step": 375060
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.200198974515956e-05,
+      "loss": 0.8734,
+      "step": 375070
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.200084181525981e-05,
+      "loss": 0.8738,
+      "step": 375080
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.199969388536006e-05,
+      "loss": 0.8247,
+      "step": 375090
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.199854595546032e-05,
+      "loss": 0.7727,
+      "step": 375100
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.199739802556057e-05,
+      "loss": 0.8716,
+      "step": 375110
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.199625009566082e-05,
+      "loss": 0.8792,
+      "step": 375120
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.1995102165761075e-05,
+      "loss": 0.8385,
+      "step": 375130
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.1993954235861325e-05,
+      "loss": 0.8039,
+      "step": 375140
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.199280630596158e-05,
+      "loss": 0.9028,
+      "step": 375150
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.199165837606183e-05,
+      "loss": 0.7924,
+      "step": 375160
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.199051044616209e-05,
+      "loss": 0.7698,
+      "step": 375170
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.198936251626234e-05,
+      "loss": 0.8246,
+      "step": 375180
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.1988214586362594e-05,
+      "loss": 0.8888,
+      "step": 375190
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 3.1987066656462844e-05,
+      "loss": 0.834,
+      "step": 375200
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.1985918726563094e-05,
+      "loss": 0.9277,
+      "step": 375210
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.198477079666335e-05,
+      "loss": 0.8281,
+      "step": 375220
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.19836228667636e-05,
+      "loss": 0.8284,
+      "step": 375230
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.198247493686385e-05,
+      "loss": 0.8903,
+      "step": 375240
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.198132700696411e-05,
+      "loss": 0.8935,
+      "step": 375250
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.198017907706436e-05,
+      "loss": 0.9268,
+      "step": 375260
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.197903114716461e-05,
+      "loss": 0.7591,
+      "step": 375270
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.1977883217264863e-05,
+      "loss": 0.9028,
+      "step": 375280
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.197673528736511e-05,
+      "loss": 0.872,
+      "step": 375290
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.197558735746537e-05,
+      "loss": 0.862,
+      "step": 375300
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.197443942756562e-05,
+      "loss": 0.8452,
+      "step": 375310
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.1973291497665876e-05,
+      "loss": 0.8784,
+      "step": 375320
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.1972143567766126e-05,
+      "loss": 0.8533,
+      "step": 375330
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.197099563786638e-05,
+      "loss": 0.843,
+      "step": 375340
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.196984770796663e-05,
+      "loss": 0.9131,
+      "step": 375350
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.196869977806688e-05,
+      "loss": 0.8615,
+      "step": 375360
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.196755184816714e-05,
+      "loss": 0.9375,
+      "step": 375370
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.196640391826739e-05,
+      "loss": 0.8713,
+      "step": 375380
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.196525598836764e-05,
+      "loss": 0.9941,
+      "step": 375390
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.1964108058467896e-05,
+      "loss": 0.8394,
+      "step": 375400
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.1962960128568145e-05,
+      "loss": 0.9462,
+      "step": 375410
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 3.1961812198668395e-05,
+      "loss": 0.8179,
+      "step": 375420
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.196066426876865e-05,
+      "loss": 0.8266,
+      "step": 375430
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.19595163388689e-05,
+      "loss": 0.9189,
+      "step": 375440
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.195836840896916e-05,
+      "loss": 0.9276,
+      "step": 375450
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.195722047906941e-05,
+      "loss": 0.8545,
+      "step": 375460
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.1956072549169665e-05,
+      "loss": 0.9925,
+      "step": 375470
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.1954924619269915e-05,
+      "loss": 0.9125,
+      "step": 375480
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.195377668937017e-05,
+      "loss": 0.9387,
+      "step": 375490
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.195262875947042e-05,
+      "loss": 0.8733,
+      "step": 375500
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.195148082957067e-05,
+      "loss": 0.7678,
+      "step": 375510
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.195033289967093e-05,
+      "loss": 0.8404,
+      "step": 375520
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.194918496977118e-05,
+      "loss": 0.8517,
+      "step": 375530
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.194803703987143e-05,
+      "loss": 0.885,
+      "step": 375540
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.1946889109971684e-05,
+      "loss": 0.9595,
+      "step": 375550
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.1945741180071934e-05,
+      "loss": 0.9459,
+      "step": 375560
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.1944593250172184e-05,
+      "loss": 0.8513,
+      "step": 375570
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.194344532027244e-05,
+      "loss": 0.7439,
+      "step": 375580
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.194229739037269e-05,
+      "loss": 0.8673,
+      "step": 375590
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.194114946047295e-05,
+      "loss": 0.8908,
+      "step": 375600
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.19400015305732e-05,
+      "loss": 0.9441,
+      "step": 375610
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.193885360067345e-05,
+      "loss": 0.8783,
+      "step": 375620
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 3.19377056707737e-05,
+      "loss": 0.8733,
+      "step": 375630
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.193655774087395e-05,
+      "loss": 0.8384,
+      "step": 375640
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.193540981097421e-05,
+      "loss": 1.0021,
+      "step": 375650
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.193426188107446e-05,
+      "loss": 0.8443,
+      "step": 375660
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.193311395117471e-05,
+      "loss": 0.8495,
+      "step": 375670
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.1931966021274966e-05,
+      "loss": 0.8205,
+      "step": 375680
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.1930818091375216e-05,
+      "loss": 0.8097,
+      "step": 375690
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.1929670161475466e-05,
+      "loss": 0.8745,
+      "step": 375700
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.192852223157572e-05,
+      "loss": 0.8825,
+      "step": 375710
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.192737430167597e-05,
+      "loss": 0.89,
+      "step": 375720
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.192622637177623e-05,
+      "loss": 0.9042,
+      "step": 375730
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.192507844187648e-05,
+      "loss": 0.9592,
+      "step": 375740
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.1923930511976735e-05,
+      "loss": 0.8616,
+      "step": 375750
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.1922782582076985e-05,
+      "loss": 0.9188,
+      "step": 375760
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.192163465217724e-05,
+      "loss": 0.8664,
+      "step": 375770
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.192048672227749e-05,
+      "loss": 0.8433,
+      "step": 375780
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.191933879237774e-05,
+      "loss": 0.8436,
+      "step": 375790
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.1918190862478e-05,
+      "loss": 0.7797,
+      "step": 375800
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.191704293257825e-05,
+      "loss": 0.9088,
+      "step": 375810
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.19158950026785e-05,
+      "loss": 0.9211,
+      "step": 375820
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.1914747072778755e-05,
+      "loss": 0.9232,
+      "step": 375830
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.1913599142879004e-05,
+      "loss": 0.8824,
+      "step": 375840
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 3.1912451212979254e-05,
+      "loss": 0.8797,
+      "step": 375850
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.191130328307951e-05,
+      "loss": 0.9316,
+      "step": 375860
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.191015535317976e-05,
+      "loss": 0.8807,
+      "step": 375870
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.190900742328002e-05,
+      "loss": 0.8487,
+      "step": 375880
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.190785949338027e-05,
+      "loss": 0.8873,
+      "step": 375890
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.1906711563480524e-05,
+      "loss": 0.8024,
+      "step": 375900
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.1905563633580774e-05,
+      "loss": 0.8613,
+      "step": 375910
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.190441570368103e-05,
+      "loss": 0.9806,
+      "step": 375920
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.190326777378128e-05,
+      "loss": 0.9125,
+      "step": 375930
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.190211984388153e-05,
+      "loss": 0.8487,
+      "step": 375940
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.190097191398179e-05,
+      "loss": 0.8636,
+      "step": 375950
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.1899823984082037e-05,
+      "loss": 0.8714,
+      "step": 375960
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.1898676054182286e-05,
+      "loss": 0.8671,
+      "step": 375970
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.189752812428254e-05,
+      "loss": 0.913,
+      "step": 375980
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.189638019438279e-05,
+      "loss": 0.9076,
+      "step": 375990
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.189523226448304e-05,
+      "loss": 0.8962,
+      "step": 376000
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.18940843345833e-05,
+      "loss": 0.9108,
+      "step": 376010
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.189293640468355e-05,
+      "loss": 0.9084,
+      "step": 376020
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.1891788474783806e-05,
+      "loss": 0.8324,
+      "step": 376030
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.1890640544884056e-05,
+      "loss": 0.9103,
+      "step": 376040
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.188949261498431e-05,
+      "loss": 0.9144,
+      "step": 376050
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.188834468508456e-05,
+      "loss": 0.7775,
+      "step": 376060
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 3.188719675518482e-05,
+      "loss": 0.8907,
+      "step": 376070
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.188604882528507e-05,
+      "loss": 1.0128,
+      "step": 376080
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.188490089538532e-05,
+      "loss": 0.9362,
+      "step": 376090
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.1883752965485575e-05,
+      "loss": 0.8405,
+      "step": 376100
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.1882605035585825e-05,
+      "loss": 0.8284,
+      "step": 376110
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.1881457105686075e-05,
+      "loss": 0.7935,
+      "step": 376120
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.188030917578633e-05,
+      "loss": 0.8073,
+      "step": 376130
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.187916124588658e-05,
+      "loss": 0.9154,
+      "step": 376140
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.187801331598683e-05,
+      "loss": 1.0136,
+      "step": 376150
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.187686538608709e-05,
+      "loss": 0.8055,
+      "step": 376160
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.187571745618734e-05,
+      "loss": 0.8582,
+      "step": 376170
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.1874569526287594e-05,
+      "loss": 0.8371,
+      "step": 376180
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.1873421596387844e-05,
+      "loss": 0.989,
+      "step": 376190
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.18722736664881e-05,
+      "loss": 0.8976,
+      "step": 376200
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.187112573658835e-05,
+      "loss": 0.8256,
+      "step": 376210
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.186997780668861e-05,
+      "loss": 0.9147,
+      "step": 376220
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.186882987678886e-05,
+      "loss": 0.7883,
+      "step": 376230
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.186768194688911e-05,
+      "loss": 0.9023,
+      "step": 376240
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.186653401698936e-05,
+      "loss": 0.964,
+      "step": 376250
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.1865386087089613e-05,
+      "loss": 0.8908,
+      "step": 376260
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.186423815718986e-05,
+      "loss": 0.8768,
+      "step": 376270
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.186309022729011e-05,
+      "loss": 0.791,
+      "step": 376280
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 3.186194229739037e-05,
+      "loss": 0.8855,
+      "step": 376290
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.186079436749062e-05,
+      "loss": 0.9205,
+      "step": 376300
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.1859646437590876e-05,
+      "loss": 0.9213,
+      "step": 376310
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.1858498507691126e-05,
+      "loss": 0.8688,
+      "step": 376320
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.185735057779138e-05,
+      "loss": 0.7833,
+      "step": 376330
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.185620264789163e-05,
+      "loss": 0.8089,
+      "step": 376340
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.185505471799189e-05,
+      "loss": 0.8557,
+      "step": 376350
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.185390678809214e-05,
+      "loss": 0.8385,
+      "step": 376360
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.185275885819239e-05,
+      "loss": 0.917,
+      "step": 376370
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.1851610928292646e-05,
+      "loss": 0.8639,
+      "step": 376380
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.1850462998392895e-05,
+      "loss": 0.8293,
+      "step": 376390
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.1849315068493145e-05,
+      "loss": 0.9018,
+      "step": 376400
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.18481671385934e-05,
+      "loss": 0.9398,
+      "step": 376410
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.184701920869365e-05,
+      "loss": 0.7937,
+      "step": 376420
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.18458712787939e-05,
+      "loss": 0.8402,
+      "step": 376430
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.184472334889416e-05,
+      "loss": 0.8735,
+      "step": 376440
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.184357541899441e-05,
+      "loss": 0.8653,
+      "step": 376450
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.1842427489094665e-05,
+      "loss": 0.807,
+      "step": 376460
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.1841279559194915e-05,
+      "loss": 0.8671,
+      "step": 376470
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.184013162929517e-05,
+      "loss": 0.8108,
+      "step": 376480
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.183898369939542e-05,
+      "loss": 0.878,
+      "step": 376490
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.183783576949568e-05,
+      "loss": 0.89,
+      "step": 376500
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 3.183668783959593e-05,
+      "loss": 0.8811,
+      "step": 376510
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.183553990969618e-05,
+      "loss": 0.8201,
+      "step": 376520
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.1834391979796434e-05,
+      "loss": 0.894,
+      "step": 376530
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.1833244049896684e-05,
+      "loss": 0.8699,
+      "step": 376540
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.1832096119996934e-05,
+      "loss": 0.8096,
+      "step": 376550
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.183094819009719e-05,
+      "loss": 0.9219,
+      "step": 376560
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.182980026019744e-05,
+      "loss": 0.8832,
+      "step": 376570
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.182865233029769e-05,
+      "loss": 0.8682,
+      "step": 376580
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.182750440039795e-05,
+      "loss": 0.8677,
+      "step": 376590
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.1826356470498197e-05,
+      "loss": 0.8689,
+      "step": 376600
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.182520854059845e-05,
+      "loss": 0.8435,
+      "step": 376610
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.18240606106987e-05,
+      "loss": 0.7922,
+      "step": 376620
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.182291268079896e-05,
+      "loss": 0.9174,
+      "step": 376630
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.182176475089921e-05,
+      "loss": 0.943,
+      "step": 376640
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.1820616820999466e-05,
+      "loss": 0.782,
+      "step": 376650
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.1819468891099716e-05,
+      "loss": 0.848,
+      "step": 376660
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.1818320961199966e-05,
+      "loss": 0.7898,
+      "step": 376670
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.181717303130022e-05,
+      "loss": 0.8405,
+      "step": 376680
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.181602510140047e-05,
+      "loss": 0.9244,
+      "step": 376690
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.181487717150072e-05,
+      "loss": 0.8613,
+      "step": 376700
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.181372924160098e-05,
+      "loss": 0.9223,
+      "step": 376710
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 3.181258131170123e-05,
+      "loss": 0.8454,
+      "step": 376720
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.181143338180148e-05,
+      "loss": 0.8504,
+      "step": 376730
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.1810285451901735e-05,
+      "loss": 0.8539,
+      "step": 376740
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.1809137522001985e-05,
+      "loss": 0.9041,
+      "step": 376750
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.180798959210224e-05,
+      "loss": 0.8987,
+      "step": 376760
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.180684166220249e-05,
+      "loss": 0.8898,
+      "step": 376770
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.180569373230275e-05,
+      "loss": 0.8686,
+      "step": 376780
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.1804545802403e-05,
+      "loss": 0.8636,
+      "step": 376790
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.1803397872503255e-05,
+      "loss": 0.869,
+      "step": 376800
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.1802249942603505e-05,
+      "loss": 0.8548,
+      "step": 376810
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.1801102012703754e-05,
+      "loss": 0.8663,
+      "step": 376820
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.179995408280401e-05,
+      "loss": 0.8946,
+      "step": 376830
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.179880615290426e-05,
+      "loss": 0.7844,
+      "step": 376840
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.179765822300451e-05,
+      "loss": 0.8431,
+      "step": 376850
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.179651029310477e-05,
+      "loss": 0.9042,
+      "step": 376860
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.179536236320502e-05,
+      "loss": 0.8816,
+      "step": 376870
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.179421443330527e-05,
+      "loss": 0.8975,
+      "step": 376880
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.1793066503405524e-05,
+      "loss": 0.9019,
+      "step": 376890
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.1791918573505774e-05,
+      "loss": 0.7733,
+      "step": 376900
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.179077064360603e-05,
+      "loss": 0.9166,
+      "step": 376910
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.178962271370628e-05,
+      "loss": 0.8565,
+      "step": 376920
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.178847478380654e-05,
+      "loss": 0.8584,
+      "step": 376930
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 3.1787326853906787e-05,
+      "loss": 0.8732,
+      "step": 376940
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.1786178924007036e-05,
+      "loss": 0.9612,
+      "step": 376950
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.178503099410729e-05,
+      "loss": 0.8788,
+      "step": 376960
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.178388306420754e-05,
+      "loss": 0.8776,
+      "step": 376970
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.178273513430779e-05,
+      "loss": 0.8712,
+      "step": 376980
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.178158720440805e-05,
+      "loss": 0.8705,
+      "step": 376990
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.17804392745083e-05,
+      "loss": 0.8463,
+      "step": 377000
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.177929134460855e-05,
+      "loss": 0.8657,
+      "step": 377010
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.1778143414708806e-05,
+      "loss": 0.8712,
+      "step": 377020
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.1776995484809056e-05,
+      "loss": 0.9844,
+      "step": 377030
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.177584755490931e-05,
+      "loss": 0.9039,
+      "step": 377040
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.177469962500956e-05,
+      "loss": 0.834,
+      "step": 377050
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.177355169510982e-05,
+      "loss": 0.849,
+      "step": 377060
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.177240376521007e-05,
+      "loss": 0.8449,
+      "step": 377070
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.1771255835310325e-05,
+      "loss": 0.8755,
+      "step": 377080
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.1770107905410575e-05,
+      "loss": 0.8364,
+      "step": 377090
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.1768959975510825e-05,
+      "loss": 0.7924,
+      "step": 377100
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.176781204561108e-05,
+      "loss": 0.8309,
+      "step": 377110
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.176666411571133e-05,
+      "loss": 0.8359,
+      "step": 377120
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.176551618581158e-05,
+      "loss": 0.8988,
+      "step": 377130
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.176436825591184e-05,
+      "loss": 0.935,
+      "step": 377140
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.176322032601209e-05,
+      "loss": 0.8107,
+      "step": 377150
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 3.176207239611234e-05,
+      "loss": 0.8394,
+      "step": 377160
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 3.1760924466212594e-05,
+      "loss": 0.9094,
+      "step": 377170
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 3.1759776536312844e-05,
+      "loss": 0.8285,
+      "step": 377180
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 3.17586286064131e-05,
+      "loss": 0.9464,
+      "step": 377190
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 3.175748067651335e-05,
+      "loss": 0.8988,
+      "step": 377200
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 3.175633274661361e-05,
+      "loss": 0.8991,
+      "step": 377210
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 3.175518481671386e-05,
+      "loss": 0.8312,
+      "step": 377220
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 3.1754036886814114e-05,
+      "loss": 0.8203,
+      "step": 377230
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 3.1752888956914363e-05,
+      "loss": 0.7838,
+      "step": 377240
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 3.175174102701461e-05,
+      "loss": 0.9047,
+      "step": 377250
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 3.175059309711487e-05,
+      "loss": 0.907,
+      "step": 377260
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 3.174944516721512e-05,
+      "loss": 0.8966,
+      "step": 377270
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 3.174829723731537e-05,
+      "loss": 0.8653,
+      "step": 377280
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 3.1747149307415626e-05,
+      "loss": 0.8352,
+      "step": 377290
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 3.1746001377515876e-05,
+      "loss": 0.8705,
+      "step": 377300
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 3.1744853447616126e-05,
+      "loss": 0.8976,
+      "step": 377310
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 3.174370551771638e-05,
+      "loss": 0.9211,
+      "step": 377320
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 3.174255758781663e-05,
+      "loss": 0.9114,
+      "step": 377330
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 3.174140965791689e-05,
+      "loss": 0.8417,
+      "step": 377340
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 3.174026172801714e-05,
+      "loss": 0.8446,
+      "step": 377350
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 3.1739113798117396e-05,
+      "loss": 0.9024,
+      "step": 377360
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 3.1737965868217645e-05,
+      "loss": 0.8385,
+      "step": 377370
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 3.17368179383179e-05,
+      "loss": 0.8246,
+      "step": 377380
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3.173567000841815e-05,
+      "loss": 0.9127,
+      "step": 377390
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3.17345220785184e-05,
+      "loss": 0.8636,
+      "step": 377400
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3.173337414861866e-05,
+      "loss": 0.8841,
+      "step": 377410
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3.173222621871891e-05,
+      "loss": 0.8872,
+      "step": 377420
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3.173107828881916e-05,
+      "loss": 0.9043,
+      "step": 377430
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3.1729930358919415e-05,
+      "loss": 0.8041,
+      "step": 377440
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3.1728782429019665e-05,
+      "loss": 0.8734,
+      "step": 377450
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3.1727634499119914e-05,
+      "loss": 0.8601,
+      "step": 377460
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3.172648656922017e-05,
+      "loss": 0.8249,
+      "step": 377470
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3.172533863932042e-05,
+      "loss": 0.8264,
+      "step": 377480
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3.172419070942068e-05,
+      "loss": 0.9455,
+      "step": 377490
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3.172304277952093e-05,
+      "loss": 0.8049,
+      "step": 377500
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3.1721894849621184e-05,
+      "loss": 0.9516,
+      "step": 377510
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3.1720746919721434e-05,
+      "loss": 0.8499,
+      "step": 377520
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3.171959898982169e-05,
+      "loss": 0.8566,
+      "step": 377530
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3.171845105992194e-05,
+      "loss": 0.8099,
+      "step": 377540
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3.171730313002219e-05,
+      "loss": 0.9284,
+      "step": 377550
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3.171615520012245e-05,
+      "loss": 0.9752,
+      "step": 377560
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3.17150072702227e-05,
+      "loss": 0.8474,
+      "step": 377570
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3.1713859340322947e-05,
+      "loss": 0.8416,
+      "step": 377580
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3.17127114104232e-05,
+      "loss": 0.9008,
+      "step": 377590
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 3.171156348052345e-05,
+      "loss": 0.8128,
+      "step": 377600
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 3.17104155506237e-05,
+      "loss": 0.7626,
+      "step": 377610
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 3.170926762072396e-05,
+      "loss": 0.8837,
+      "step": 377620
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 3.170811969082421e-05,
+      "loss": 0.8517,
+      "step": 377630
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 3.1706971760924466e-05,
+      "loss": 0.7737,
+      "step": 377640
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 3.1705823831024716e-05,
+      "loss": 0.8933,
+      "step": 377650
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 3.170467590112497e-05,
+      "loss": 0.7874,
+      "step": 377660
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 3.170352797122522e-05,
+      "loss": 0.8659,
+      "step": 377670
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 3.170238004132547e-05,
+      "loss": 0.8751,
+      "step": 377680
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 3.170123211142573e-05,
+      "loss": 0.889,
+      "step": 377690
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 3.170008418152598e-05,
+      "loss": 0.9263,
+      "step": 377700
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 3.169893625162623e-05,
+      "loss": 0.7634,
+      "step": 377710
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 3.1697788321726485e-05,
+      "loss": 0.9156,
+      "step": 377720
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 3.1696640391826735e-05,
+      "loss": 1.0387,
+      "step": 377730
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 3.1695492461926985e-05,
+      "loss": 0.8058,
+      "step": 377740
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 3.169434453202724e-05,
+      "loss": 0.888,
+      "step": 377750
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 3.169319660212749e-05,
+      "loss": 0.9287,
+      "step": 377760
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 3.169204867222775e-05,
+      "loss": 0.8203,
+      "step": 377770
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 3.1690900742328e-05,
+      "loss": 0.8662,
+      "step": 377780
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 3.1689752812428255e-05,
+      "loss": 0.9641,
+      "step": 377790
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 3.1688604882528504e-05,
+      "loss": 0.9044,
+      "step": 377800
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 3.168745695262876e-05,
+      "loss": 0.9151,
+      "step": 377810
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3.168630902272901e-05,
+      "loss": 0.8405,
+      "step": 377820
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3.168516109282926e-05,
+      "loss": 0.8369,
+      "step": 377830
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3.168401316292952e-05,
+      "loss": 0.8401,
+      "step": 377840
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3.168286523302977e-05,
+      "loss": 0.8767,
+      "step": 377850
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3.168171730313002e-05,
+      "loss": 0.8884,
+      "step": 377860
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3.1680569373230274e-05,
+      "loss": 0.8058,
+      "step": 377870
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3.1679421443330524e-05,
+      "loss": 0.8646,
+      "step": 377880
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3.167827351343077e-05,
+      "loss": 0.903,
+      "step": 377890
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3.167712558353103e-05,
+      "loss": 0.903,
+      "step": 377900
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3.167597765363128e-05,
+      "loss": 0.8412,
+      "step": 377910
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3.1674829723731537e-05,
+      "loss": 0.8663,
+      "step": 377920
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3.1673681793831786e-05,
+      "loss": 0.8389,
+      "step": 377930
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3.167253386393204e-05,
+      "loss": 0.9727,
+      "step": 377940
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3.167138593403229e-05,
+      "loss": 0.8803,
+      "step": 377950
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3.167023800413255e-05,
+      "loss": 0.9598,
+      "step": 377960
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3.16690900742328e-05,
+      "loss": 0.8444,
+      "step": 377970
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3.166794214433305e-05,
+      "loss": 0.8201,
+      "step": 377980
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3.1666794214433306e-05,
+      "loss": 0.9305,
+      "step": 377990
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3.1665646284533556e-05,
+      "loss": 0.8553,
+      "step": 378000
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3.1664498354633806e-05,
+      "loss": 0.9692,
+      "step": 378010
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3.166335042473406e-05,
+      "loss": 0.8421,
+      "step": 378020
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 3.166220249483431e-05,
+      "loss": 0.7841,
+      "step": 378030
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3.166105456493456e-05,
+      "loss": 0.8911,
+      "step": 378040
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3.165990663503482e-05,
+      "loss": 0.8665,
+      "step": 378050
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3.165875870513507e-05,
+      "loss": 0.9096,
+      "step": 378060
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3.1657610775235325e-05,
+      "loss": 0.9398,
+      "step": 378070
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3.1656462845335575e-05,
+      "loss": 0.9077,
+      "step": 378080
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3.165531491543583e-05,
+      "loss": 0.839,
+      "step": 378090
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3.165416698553608e-05,
+      "loss": 0.8889,
+      "step": 378100
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3.165301905563634e-05,
+      "loss": 0.8345,
+      "step": 378110
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3.165187112573659e-05,
+      "loss": 0.9158,
+      "step": 378120
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3.165072319583684e-05,
+      "loss": 0.9479,
+      "step": 378130
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3.1649575265937094e-05,
+      "loss": 0.8785,
+      "step": 378140
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3.1648427336037344e-05,
+      "loss": 0.8731,
+      "step": 378150
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3.1647279406137594e-05,
+      "loss": 0.935,
+      "step": 378160
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3.164613147623785e-05,
+      "loss": 0.7732,
+      "step": 378170
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3.16449835463381e-05,
+      "loss": 0.9514,
+      "step": 378180
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3.164383561643835e-05,
+      "loss": 0.8191,
+      "step": 378190
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3.164268768653861e-05,
+      "loss": 0.6996,
+      "step": 378200
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3.164153975663886e-05,
+      "loss": 0.8848,
+      "step": 378210
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3.1640391826739113e-05,
+      "loss": 0.9372,
+      "step": 378220
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3.163924389683936e-05,
+      "loss": 0.9748,
+      "step": 378230
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3.163809596693962e-05,
+      "loss": 0.8783,
+      "step": 378240
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 3.163694803703987e-05,
+      "loss": 0.8446,
+      "step": 378250
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 3.1635800107140126e-05,
+      "loss": 0.8782,
+      "step": 378260
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 3.1634652177240376e-05,
+      "loss": 0.8874,
+      "step": 378270
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 3.1633504247340626e-05,
+      "loss": 0.8179,
+      "step": 378280
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 3.163235631744088e-05,
+      "loss": 0.9486,
+      "step": 378290
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 3.163120838754113e-05,
+      "loss": 0.8055,
+      "step": 378300
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 3.163006045764138e-05,
+      "loss": 0.9409,
+      "step": 378310
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 3.162891252774163e-05,
+      "loss": 0.8314,
+      "step": 378320
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 3.162776459784189e-05,
+      "loss": 0.8573,
+      "step": 378330
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 3.162661666794214e-05,
+      "loss": 0.808,
+      "step": 378340
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 3.1625468738042395e-05,
+      "loss": 0.8923,
+      "step": 378350
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 3.1624320808142645e-05,
+      "loss": 0.8119,
+      "step": 378360
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 3.16231728782429e-05,
+      "loss": 0.9345,
+      "step": 378370
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 3.162202494834315e-05,
+      "loss": 0.921,
+      "step": 378380
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 3.162087701844341e-05,
+      "loss": 0.8903,
+      "step": 378390
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 3.161972908854366e-05,
+      "loss": 0.9587,
+      "step": 378400
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 3.161858115864391e-05,
+      "loss": 0.9176,
+      "step": 378410
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 3.1617433228744165e-05,
+      "loss": 0.7986,
+      "step": 378420
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 3.1616285298844415e-05,
+      "loss": 0.8242,
+      "step": 378430
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 3.1615137368944664e-05,
+      "loss": 0.9188,
+      "step": 378440
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 3.161398943904492e-05,
+      "loss": 0.9317,
+      "step": 378450
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 3.161284150914517e-05,
+      "loss": 0.9215,
+      "step": 378460
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 3.161169357924542e-05,
+      "loss": 0.8856,
+      "step": 378470
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 3.161054564934568e-05,
+      "loss": 0.8495,
+      "step": 378480
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 3.160939771944593e-05,
+      "loss": 0.8613,
+      "step": 378490
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 3.1608249789546184e-05,
+      "loss": 0.8421,
+      "step": 378500
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 3.1607101859646434e-05,
+      "loss": 0.8289,
+      "step": 378510
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 3.160595392974669e-05,
+      "loss": 0.8758,
+      "step": 378520
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 3.160480599984694e-05,
+      "loss": 0.8506,
+      "step": 378530
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 3.16036580699472e-05,
+      "loss": 0.8667,
+      "step": 378540
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 3.160251014004745e-05,
+      "loss": 0.8778,
+      "step": 378550
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 3.1601362210147697e-05,
+      "loss": 0.9486,
+      "step": 378560
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 3.160021428024795e-05,
+      "loss": 0.8818,
+      "step": 378570
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 3.15990663503482e-05,
+      "loss": 0.8403,
+      "step": 378580
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 3.159791842044845e-05,
+      "loss": 0.8684,
+      "step": 378590
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 3.159677049054871e-05,
+      "loss": 0.9002,
+      "step": 378600
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 3.159562256064896e-05,
+      "loss": 0.8749,
+      "step": 378610
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 3.159447463074921e-05,
+      "loss": 0.7872,
+      "step": 378620
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 3.1593326700849466e-05,
+      "loss": 0.9037,
+      "step": 378630
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 3.1592178770949716e-05,
+      "loss": 0.9526,
+      "step": 378640
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 3.159103084104997e-05,
+      "loss": 0.8873,
+      "step": 378650
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 3.158988291115022e-05,
+      "loss": 0.8389,
+      "step": 378660
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 3.158873498125048e-05,
+      "loss": 0.8367,
+      "step": 378670
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 3.158758705135073e-05,
+      "loss": 0.8965,
+      "step": 378680
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 3.1586439121450985e-05,
+      "loss": 0.8169,
+      "step": 378690
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 3.1585291191551235e-05,
+      "loss": 0.903,
+      "step": 378700
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 3.1584143261651485e-05,
+      "loss": 1.0028,
+      "step": 378710
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 3.158299533175174e-05,
+      "loss": 0.7624,
+      "step": 378720
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 3.158184740185199e-05,
+      "loss": 0.9358,
+      "step": 378730
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 3.158069947195224e-05,
+      "loss": 0.8858,
+      "step": 378740
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 3.15795515420525e-05,
+      "loss": 0.7224,
+      "step": 378750
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 3.157840361215275e-05,
+      "loss": 0.8371,
+      "step": 378760
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 3.1577255682253e-05,
+      "loss": 0.8762,
+      "step": 378770
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 3.1576107752353254e-05,
+      "loss": 0.8859,
+      "step": 378780
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 3.1574959822453504e-05,
+      "loss": 0.8836,
+      "step": 378790
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 3.157381189255376e-05,
+      "loss": 0.8136,
+      "step": 378800
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 3.157266396265401e-05,
+      "loss": 0.8818,
+      "step": 378810
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 3.157151603275427e-05,
+      "loss": 0.8173,
+      "step": 378820
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 3.157036810285452e-05,
+      "loss": 0.8288,
+      "step": 378830
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 3.1569220172954774e-05,
+      "loss": 0.8852,
+      "step": 378840
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 3.1568072243055024e-05,
+      "loss": 0.9047,
+      "step": 378850
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 3.1566924313155274e-05,
+      "loss": 0.9738,
+      "step": 378860
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 3.156577638325553e-05,
+      "loss": 0.8497,
+      "step": 378870
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 3.156462845335578e-05,
+      "loss": 0.8489,
+      "step": 378880
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 3.156348052345603e-05,
+      "loss": 0.8619,
+      "step": 378890
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 3.1562332593556286e-05,
+      "loss": 0.7952,
+      "step": 378900
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3.1561184663656536e-05,
+      "loss": 0.9163,
+      "step": 378910
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3.1560036733756786e-05,
+      "loss": 0.9322,
+      "step": 378920
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3.155888880385704e-05,
+      "loss": 0.9778,
+      "step": 378930
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3.155774087395729e-05,
+      "loss": 0.7828,
+      "step": 378940
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3.155659294405755e-05,
+      "loss": 0.8585,
+      "step": 378950
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3.15554450141578e-05,
+      "loss": 0.8451,
+      "step": 378960
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3.1554297084258056e-05,
+      "loss": 0.8368,
+      "step": 378970
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3.1553149154358306e-05,
+      "loss": 0.782,
+      "step": 378980
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3.155200122445856e-05,
+      "loss": 0.8338,
+      "step": 378990
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3.155085329455881e-05,
+      "loss": 0.8047,
+      "step": 379000
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3.154970536465906e-05,
+      "loss": 0.9191,
+      "step": 379010
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3.154855743475931e-05,
+      "loss": 0.844,
+      "step": 379020
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3.154740950485957e-05,
+      "loss": 0.8247,
+      "step": 379030
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3.154626157495982e-05,
+      "loss": 0.881,
+      "step": 379040
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3.154511364506007e-05,
+      "loss": 0.9203,
+      "step": 379050
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3.1543965715160325e-05,
+      "loss": 0.8623,
+      "step": 379060
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3.1542817785260575e-05,
+      "loss": 0.9263,
+      "step": 379070
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3.154166985536083e-05,
+      "loss": 0.885,
+      "step": 379080
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3.154052192546108e-05,
+      "loss": 0.9306,
+      "step": 379090
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3.153937399556134e-05,
+      "loss": 0.8696,
+      "step": 379100
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3.153822606566159e-05,
+      "loss": 0.8815,
+      "step": 379110
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 3.1537078135761844e-05,
+      "loss": 0.84,
+      "step": 379120
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3.1535930205862094e-05,
+      "loss": 0.8458,
+      "step": 379130
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3.1534782275962344e-05,
+      "loss": 0.9252,
+      "step": 379140
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3.15336343460626e-05,
+      "loss": 0.8674,
+      "step": 379150
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3.153248641616285e-05,
+      "loss": 0.9074,
+      "step": 379160
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3.15313384862631e-05,
+      "loss": 0.8192,
+      "step": 379170
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3.153019055636336e-05,
+      "loss": 0.9373,
+      "step": 379180
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3.152904262646361e-05,
+      "loss": 0.8854,
+      "step": 379190
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3.152789469656386e-05,
+      "loss": 0.9278,
+      "step": 379200
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3.152674676666411e-05,
+      "loss": 0.8759,
+      "step": 379210
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3.152559883676436e-05,
+      "loss": 0.852,
+      "step": 379220
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3.152445090686462e-05,
+      "loss": 0.8379,
+      "step": 379230
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3.152330297696487e-05,
+      "loss": 0.8849,
+      "step": 379240
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3.1522155047065126e-05,
+      "loss": 0.8426,
+      "step": 379250
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3.1521007117165376e-05,
+      "loss": 0.8294,
+      "step": 379260
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3.151985918726563e-05,
+      "loss": 0.8496,
+      "step": 379270
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3.151871125736588e-05,
+      "loss": 0.8314,
+      "step": 379280
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3.151756332746613e-05,
+      "loss": 0.8042,
+      "step": 379290
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3.151641539756639e-05,
+      "loss": 0.8215,
+      "step": 379300
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3.151526746766664e-05,
+      "loss": 0.8887,
+      "step": 379310
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3.151411953776689e-05,
+      "loss": 0.8655,
+      "step": 379320
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3.1512971607867145e-05,
+      "loss": 0.8774,
+      "step": 379330
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 3.1511823677967395e-05,
+      "loss": 0.8743,
+      "step": 379340
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 3.1510675748067645e-05,
+      "loss": 0.8225,
+      "step": 379350
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 3.15095278181679e-05,
+      "loss": 0.8496,
+      "step": 379360
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 3.150837988826815e-05,
+      "loss": 0.8968,
+      "step": 379370
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 3.150723195836841e-05,
+      "loss": 0.8132,
+      "step": 379380
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 3.150608402846866e-05,
+      "loss": 0.9173,
+      "step": 379390
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 3.1504936098568915e-05,
+      "loss": 0.8909,
+      "step": 379400
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 3.1503788168669165e-05,
+      "loss": 0.8393,
+      "step": 379410
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 3.150264023876942e-05,
+      "loss": 0.86,
+      "step": 379420
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 3.150149230886967e-05,
+      "loss": 0.8945,
+      "step": 379430
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 3.150034437896992e-05,
+      "loss": 0.9294,
+      "step": 379440
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 3.149919644907018e-05,
+      "loss": 0.8879,
+      "step": 379450
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 3.149804851917043e-05,
+      "loss": 0.9727,
+      "step": 379460
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 3.149690058927068e-05,
+      "loss": 0.835,
+      "step": 379470
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 3.1495752659370934e-05,
+      "loss": 0.8737,
+      "step": 379480
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 3.1494604729471184e-05,
+      "loss": 0.8682,
+      "step": 379490
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 3.1493456799571434e-05,
+      "loss": 0.7989,
+      "step": 379500
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 3.149230886967169e-05,
+      "loss": 0.9628,
+      "step": 379510
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 3.149116093977194e-05,
+      "loss": 0.8434,
+      "step": 379520
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 3.14900130098722e-05,
+      "loss": 0.8532,
+      "step": 379530
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 3.1488865079972447e-05,
+      "loss": 0.8742,
+      "step": 379540
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 3.14877171500727e-05,
+      "loss": 0.8127,
+      "step": 379550
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 3.148656922017295e-05,
+      "loss": 0.912,
+      "step": 379560
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 3.148542129027321e-05,
+      "loss": 0.8248,
+      "step": 379570
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 3.148427336037346e-05,
+      "loss": 0.8074,
+      "step": 379580
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 3.148312543047371e-05,
+      "loss": 0.8579,
+      "step": 379590
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 3.1481977500573966e-05,
+      "loss": 0.9229,
+      "step": 379600
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 3.1480829570674216e-05,
+      "loss": 0.8811,
+      "step": 379610
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 3.1479681640774466e-05,
+      "loss": 0.7869,
+      "step": 379620
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 3.147853371087472e-05,
+      "loss": 0.8522,
+      "step": 379630
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 3.147738578097497e-05,
+      "loss": 0.859,
+      "step": 379640
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 3.147623785107522e-05,
+      "loss": 0.7865,
+      "step": 379650
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 3.147508992117548e-05,
+      "loss": 0.8566,
+      "step": 379660
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 3.147394199127573e-05,
+      "loss": 0.853,
+      "step": 379670
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 3.1472794061375985e-05,
+      "loss": 0.8468,
+      "step": 379680
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 3.1471646131476235e-05,
+      "loss": 0.8181,
+      "step": 379690
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 3.147049820157649e-05,
+      "loss": 0.876,
+      "step": 379700
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 3.146935027167674e-05,
+      "loss": 0.877,
+      "step": 379710
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 3.146820234177699e-05,
+      "loss": 0.9302,
+      "step": 379720
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 3.146705441187725e-05,
+      "loss": 0.9111,
+      "step": 379730
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 3.14659064819775e-05,
+      "loss": 0.8537,
+      "step": 379740
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 3.146475855207775e-05,
+      "loss": 0.9066,
+      "step": 379750
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 3.1463610622178004e-05,
+      "loss": 0.9515,
+      "step": 379760
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 3.1462462692278254e-05,
+      "loss": 0.8657,
+      "step": 379770
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 3.1461314762378504e-05,
+      "loss": 0.8339,
+      "step": 379780
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3.146016683247876e-05,
+      "loss": 0.8757,
+      "step": 379790
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3.145901890257901e-05,
+      "loss": 0.8475,
+      "step": 379800
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3.145787097267927e-05,
+      "loss": 0.8548,
+      "step": 379810
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3.145672304277952e-05,
+      "loss": 0.7808,
+      "step": 379820
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3.1455575112879774e-05,
+      "loss": 0.9205,
+      "step": 379830
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3.1454427182980024e-05,
+      "loss": 0.871,
+      "step": 379840
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3.145327925308028e-05,
+      "loss": 0.9204,
+      "step": 379850
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3.145213132318053e-05,
+      "loss": 0.9139,
+      "step": 379860
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3.145098339328078e-05,
+      "loss": 0.8325,
+      "step": 379870
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3.1449835463381036e-05,
+      "loss": 0.9357,
+      "step": 379880
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3.1448687533481286e-05,
+      "loss": 0.8849,
+      "step": 379890
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3.1447539603581536e-05,
+      "loss": 0.8594,
+      "step": 379900
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3.144639167368179e-05,
+      "loss": 0.7597,
+      "step": 379910
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3.144524374378204e-05,
+      "loss": 0.8151,
+      "step": 379920
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3.144409581388229e-05,
+      "loss": 0.889,
+      "step": 379930
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3.144294788398255e-05,
+      "loss": 0.9827,
+      "step": 379940
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3.14417999540828e-05,
+      "loss": 0.9234,
+      "step": 379950
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3.1440652024183056e-05,
+      "loss": 0.9061,
+      "step": 379960
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3.1439504094283306e-05,
+      "loss": 0.9168,
+      "step": 379970
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3.143835616438356e-05,
+      "loss": 0.7889,
+      "step": 379980
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 3.143720823448381e-05,
+      "loss": 0.9305,
+      "step": 379990
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 3.143606030458407e-05,
+      "loss": 0.9077,
+      "step": 380000
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 3.143491237468432e-05,
+      "loss": 0.9066,
+      "step": 380010
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 3.143376444478457e-05,
+      "loss": 0.8873,
+      "step": 380020
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 3.1432616514884825e-05,
+      "loss": 0.776,
+      "step": 380030
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 3.1431468584985075e-05,
+      "loss": 0.8939,
+      "step": 380040
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 3.1430320655085325e-05,
+      "loss": 0.7607,
+      "step": 380050
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 3.142917272518558e-05,
+      "loss": 0.8063,
+      "step": 380060
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 3.142802479528583e-05,
+      "loss": 0.8592,
+      "step": 380070
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 3.142687686538608e-05,
+      "loss": 0.8887,
+      "step": 380080
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 3.142572893548634e-05,
+      "loss": 0.8905,
+      "step": 380090
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 3.142458100558659e-05,
+      "loss": 0.9314,
+      "step": 380100
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 3.1423433075686844e-05,
+      "loss": 0.8241,
+      "step": 380110
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 3.1422285145787094e-05,
+      "loss": 0.7961,
+      "step": 380120
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 3.142113721588735e-05,
+      "loss": 0.9428,
+      "step": 380130
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 3.14199892859876e-05,
+      "loss": 0.8738,
+      "step": 380140
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 3.141884135608786e-05,
+      "loss": 0.8178,
+      "step": 380150
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 3.141769342618811e-05,
+      "loss": 0.9194,
+      "step": 380160
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 3.141654549628836e-05,
+      "loss": 0.9396,
+      "step": 380170
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 3.1415397566388613e-05,
+      "loss": 0.9202,
+      "step": 380180
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 3.141424963648886e-05,
+      "loss": 0.9497,
+      "step": 380190
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 3.141310170658911e-05,
+      "loss": 0.8954,
+      "step": 380200
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 3.141195377668937e-05,
+      "loss": 0.8892,
+      "step": 380210
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3.141080584678962e-05,
+      "loss": 0.9993,
+      "step": 380220
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3.140965791688987e-05,
+      "loss": 0.8816,
+      "step": 380230
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3.1408509986990126e-05,
+      "loss": 0.8826,
+      "step": 380240
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3.1407362057090376e-05,
+      "loss": 0.9199,
+      "step": 380250
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3.140621412719063e-05,
+      "loss": 0.9852,
+      "step": 380260
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3.140506619729088e-05,
+      "loss": 0.844,
+      "step": 380270
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3.140391826739114e-05,
+      "loss": 0.8523,
+      "step": 380280
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3.140277033749139e-05,
+      "loss": 0.8264,
+      "step": 380290
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3.1401622407591646e-05,
+      "loss": 0.7679,
+      "step": 380300
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3.1400474477691895e-05,
+      "loss": 0.8876,
+      "step": 380310
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3.1399326547792145e-05,
+      "loss": 0.862,
+      "step": 380320
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3.13981786178924e-05,
+      "loss": 0.8473,
+      "step": 380330
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3.139703068799265e-05,
+      "loss": 0.8495,
+      "step": 380340
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3.13958827580929e-05,
+      "loss": 0.8841,
+      "step": 380350
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3.139473482819316e-05,
+      "loss": 0.8199,
+      "step": 380360
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3.139358689829341e-05,
+      "loss": 0.8935,
+      "step": 380370
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3.139243896839366e-05,
+      "loss": 0.821,
+      "step": 380380
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3.139129103849391e-05,
+      "loss": 0.9334,
+      "step": 380390
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3.1390143108594164e-05,
+      "loss": 0.9085,
+      "step": 380400
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3.138899517869442e-05,
+      "loss": 0.7702,
+      "step": 380410
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3.138784724879467e-05,
+      "loss": 0.7957,
+      "step": 380420
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 3.138669931889493e-05,
+      "loss": 0.9326,
+      "step": 380430
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 3.138555138899518e-05,
+      "loss": 0.8371,
+      "step": 380440
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 3.138440345909543e-05,
+      "loss": 0.9332,
+      "step": 380450
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 3.1383255529195684e-05,
+      "loss": 0.9091,
+      "step": 380460
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 3.1382107599295934e-05,
+      "loss": 0.9266,
+      "step": 380470
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 3.1380959669396184e-05,
+      "loss": 0.8596,
+      "step": 380480
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 3.137981173949644e-05,
+      "loss": 0.8997,
+      "step": 380490
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 3.137866380959669e-05,
+      "loss": 0.8748,
+      "step": 380500
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 3.137751587969694e-05,
+      "loss": 0.9294,
+      "step": 380510
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 3.1376367949797197e-05,
+      "loss": 0.8756,
+      "step": 380520
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 3.1375220019897446e-05,
+      "loss": 0.9429,
+      "step": 380530
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 3.1374072089997696e-05,
+      "loss": 0.8564,
+      "step": 380540
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 3.137292416009795e-05,
+      "loss": 0.8048,
+      "step": 380550
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 3.137177623019821e-05,
+      "loss": 0.8244,
+      "step": 380560
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 3.137062830029846e-05,
+      "loss": 0.9033,
+      "step": 380570
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 3.1369480370398716e-05,
+      "loss": 0.8649,
+      "step": 380580
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 3.1368332440498966e-05,
+      "loss": 0.8986,
+      "step": 380590
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 3.1367184510599216e-05,
+      "loss": 0.9432,
+      "step": 380600
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 3.136603658069947e-05,
+      "loss": 0.8238,
+      "step": 380610
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 3.136488865079972e-05,
+      "loss": 0.8165,
+      "step": 380620
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 3.136374072089997e-05,
+      "loss": 0.8158,
+      "step": 380630
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 3.136259279100023e-05,
+      "loss": 0.9318,
+      "step": 380640
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 3.136144486110048e-05,
+      "loss": 0.8938,
+      "step": 380650
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 3.136029693120073e-05,
+      "loss": 0.918,
+      "step": 380660
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 3.1359149001300985e-05,
+      "loss": 0.9353,
+      "step": 380670
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 3.1358001071401235e-05,
+      "loss": 0.8924,
+      "step": 380680
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 3.135685314150149e-05,
+      "loss": 0.8694,
+      "step": 380690
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 3.135570521160174e-05,
+      "loss": 0.8322,
+      "step": 380700
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 3.1354557281702e-05,
+      "loss": 0.8919,
+      "step": 380710
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 3.135340935180225e-05,
+      "loss": 0.8921,
+      "step": 380720
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 3.1352261421902504e-05,
+      "loss": 0.8099,
+      "step": 380730
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 3.1351113492002754e-05,
+      "loss": 0.8156,
+      "step": 380740
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 3.1349965562103004e-05,
+      "loss": 0.8249,
+      "step": 380750
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 3.134881763220326e-05,
+      "loss": 0.8355,
+      "step": 380760
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 3.134766970230351e-05,
+      "loss": 0.8787,
+      "step": 380770
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 3.134652177240376e-05,
+      "loss": 0.7695,
+      "step": 380780
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 3.134537384250402e-05,
+      "loss": 0.9175,
+      "step": 380790
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 3.134422591260427e-05,
+      "loss": 0.8775,
+      "step": 380800
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 3.134307798270452e-05,
+      "loss": 0.8425,
+      "step": 380810
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 3.1341930052804774e-05,
+      "loss": 0.8637,
+      "step": 380820
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 3.134078212290502e-05,
+      "loss": 0.8127,
+      "step": 380830
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 3.133963419300528e-05,
+      "loss": 0.8389,
+      "step": 380840
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 3.133848626310553e-05,
+      "loss": 0.8378,
+      "step": 380850
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 3.1337338333205786e-05,
+      "loss": 0.9307,
+      "step": 380860
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 3.1336190403306036e-05,
+      "loss": 0.8465,
+      "step": 380870
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3.133504247340629e-05,
+      "loss": 0.8503,
+      "step": 380880
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3.133389454350654e-05,
+      "loss": 0.8403,
+      "step": 380890
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3.133274661360679e-05,
+      "loss": 0.8488,
+      "step": 380900
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3.133159868370705e-05,
+      "loss": 0.9839,
+      "step": 380910
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3.13304507538073e-05,
+      "loss": 0.8816,
+      "step": 380920
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3.132930282390755e-05,
+      "loss": 1.0537,
+      "step": 380930
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3.1328154894007806e-05,
+      "loss": 0.8409,
+      "step": 380940
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3.1327006964108056e-05,
+      "loss": 0.8442,
+      "step": 380950
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3.1325859034208305e-05,
+      "loss": 0.8481,
+      "step": 380960
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3.132471110430856e-05,
+      "loss": 0.8991,
+      "step": 380970
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3.132356317440881e-05,
+      "loss": 0.9334,
+      "step": 380980
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3.132241524450907e-05,
+      "loss": 0.7824,
+      "step": 380990
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3.132126731460932e-05,
+      "loss": 0.8648,
+      "step": 381000
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3.1320119384709575e-05,
+      "loss": 0.9869,
+      "step": 381010
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3.1318971454809825e-05,
+      "loss": 0.9159,
+      "step": 381020
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3.131782352491008e-05,
+      "loss": 0.7531,
+      "step": 381030
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3.131667559501033e-05,
+      "loss": 0.9219,
+      "step": 381040
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3.131552766511058e-05,
+      "loss": 0.876,
+      "step": 381050
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3.131437973521083e-05,
+      "loss": 0.8332,
+      "step": 381060
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3.131323180531109e-05,
+      "loss": 0.9164,
+      "step": 381070
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 3.131208387541134e-05,
+      "loss": 0.849,
+      "step": 381080
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 3.131093594551159e-05,
+      "loss": 0.8886,
+      "step": 381090
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 3.1309788015611844e-05,
+      "loss": 0.9125,
+      "step": 381100
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 3.1308640085712094e-05,
+      "loss": 0.8575,
+      "step": 381110
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 3.1307492155812344e-05,
+      "loss": 0.9407,
+      "step": 381120
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 3.13063442259126e-05,
+      "loss": 0.9522,
+      "step": 381130
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 3.130519629601286e-05,
+      "loss": 0.8274,
+      "step": 381140
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 3.130404836611311e-05,
+      "loss": 0.8482,
+      "step": 381150
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 3.1302900436213363e-05,
+      "loss": 0.8239,
+      "step": 381160
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 3.130175250631361e-05,
+      "loss": 0.8915,
+      "step": 381170
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 3.130060457641386e-05,
+      "loss": 0.914,
+      "step": 381180
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 3.129945664651412e-05,
+      "loss": 0.9263,
+      "step": 381190
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 3.129830871661437e-05,
+      "loss": 0.8094,
+      "step": 381200
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 3.129716078671462e-05,
+      "loss": 0.9096,
+      "step": 381210
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 3.1296012856814876e-05,
+      "loss": 0.8446,
+      "step": 381220
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 3.1294864926915126e-05,
+      "loss": 0.8755,
+      "step": 381230
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 3.1293716997015376e-05,
+      "loss": 0.849,
+      "step": 381240
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 3.129256906711563e-05,
+      "loss": 0.8394,
+      "step": 381250
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 3.129142113721588e-05,
+      "loss": 0.983,
+      "step": 381260
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 3.129027320731613e-05,
+      "loss": 0.9719,
+      "step": 381270
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 3.128912527741639e-05,
+      "loss": 0.9047,
+      "step": 381280
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 3.1287977347516645e-05,
+      "loss": 0.8362,
+      "step": 381290
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 3.1286829417616895e-05,
+      "loss": 0.8578,
+      "step": 381300
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3.128568148771715e-05,
+      "loss": 0.8398,
+      "step": 381310
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3.12845335578174e-05,
+      "loss": 0.8331,
+      "step": 381320
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3.128338562791765e-05,
+      "loss": 0.9333,
+      "step": 381330
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3.128223769801791e-05,
+      "loss": 0.8425,
+      "step": 381340
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3.128108976811816e-05,
+      "loss": 0.8904,
+      "step": 381350
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3.127994183821841e-05,
+      "loss": 0.8806,
+      "step": 381360
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3.1278793908318665e-05,
+      "loss": 0.8503,
+      "step": 381370
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3.1277645978418914e-05,
+      "loss": 0.8346,
+      "step": 381380
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3.1276498048519164e-05,
+      "loss": 0.8627,
+      "step": 381390
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3.127535011861942e-05,
+      "loss": 0.8448,
+      "step": 381400
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3.127420218871967e-05,
+      "loss": 0.7781,
+      "step": 381410
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3.127305425881992e-05,
+      "loss": 0.8733,
+      "step": 381420
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3.127190632892018e-05,
+      "loss": 0.8485,
+      "step": 381430
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3.1270758399020434e-05,
+      "loss": 0.9132,
+      "step": 381440
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3.1269610469120684e-05,
+      "loss": 0.8416,
+      "step": 381450
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3.126846253922094e-05,
+      "loss": 0.8494,
+      "step": 381460
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3.126731460932119e-05,
+      "loss": 0.8598,
+      "step": 381470
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3.126616667942144e-05,
+      "loss": 0.8457,
+      "step": 381480
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3.12650187495217e-05,
+      "loss": 0.886,
+      "step": 381490
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3.1263870819621947e-05,
+      "loss": 0.8331,
+      "step": 381500
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3.1262722889722196e-05,
+      "loss": 0.8507,
+      "step": 381510
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 3.126157495982245e-05,
+      "loss": 0.9185,
+      "step": 381520
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 3.12604270299227e-05,
+      "loss": 0.8201,
+      "step": 381530
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 3.125927910002295e-05,
+      "loss": 0.8808,
+      "step": 381540
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 3.125813117012321e-05,
+      "loss": 0.9381,
+      "step": 381550
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 3.125698324022346e-05,
+      "loss": 0.8411,
+      "step": 381560
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 3.125583531032371e-05,
+      "loss": 0.8303,
+      "step": 381570
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 3.1254687380423966e-05,
+      "loss": 0.9281,
+      "step": 381580
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 3.125353945052422e-05,
+      "loss": 0.9059,
+      "step": 381590
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 3.125239152062447e-05,
+      "loss": 0.9105,
+      "step": 381600
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 3.125124359072473e-05,
+      "loss": 0.8391,
+      "step": 381610
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 3.125009566082498e-05,
+      "loss": 0.8069,
+      "step": 381620
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 3.124894773092523e-05,
+      "loss": 0.9014,
+      "step": 381630
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 3.1247799801025485e-05,
+      "loss": 0.7967,
+      "step": 381640
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 3.1246651871125735e-05,
+      "loss": 0.9203,
+      "step": 381650
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 3.1245503941225985e-05,
+      "loss": 0.9038,
+      "step": 381660
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 3.124435601132624e-05,
+      "loss": 0.8583,
+      "step": 381670
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 3.124320808142649e-05,
+      "loss": 0.8926,
+      "step": 381680
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 3.124206015152674e-05,
+      "loss": 0.8335,
+      "step": 381690
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 3.1240912221627e-05,
+      "loss": 0.8117,
+      "step": 381700
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 3.123976429172725e-05,
+      "loss": 0.8529,
+      "step": 381710
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 3.12386163618275e-05,
+      "loss": 0.8897,
+      "step": 381720
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 3.1237468431927754e-05,
+      "loss": 0.8078,
+      "step": 381730
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 3.123632050202801e-05,
+      "loss": 0.8927,
+      "step": 381740
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.123517257212826e-05,
+      "loss": 0.8755,
+      "step": 381750
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.123402464222851e-05,
+      "loss": 0.8336,
+      "step": 381760
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.123287671232877e-05,
+      "loss": 0.8797,
+      "step": 381770
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.123172878242902e-05,
+      "loss": 0.9049,
+      "step": 381780
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.123058085252927e-05,
+      "loss": 0.8198,
+      "step": 381790
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.1229432922629524e-05,
+      "loss": 0.87,
+      "step": 381800
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.122828499272977e-05,
+      "loss": 0.8834,
+      "step": 381810
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.122713706283002e-05,
+      "loss": 0.8871,
+      "step": 381820
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.122598913293028e-05,
+      "loss": 0.8193,
+      "step": 381830
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.122484120303053e-05,
+      "loss": 0.9247,
+      "step": 381840
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.122369327313078e-05,
+      "loss": 0.8531,
+      "step": 381850
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.1222545343231036e-05,
+      "loss": 0.8837,
+      "step": 381860
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.1221397413331286e-05,
+      "loss": 0.7919,
+      "step": 381870
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.122024948343154e-05,
+      "loss": 0.8645,
+      "step": 381880
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.12191015535318e-05,
+      "loss": 0.9411,
+      "step": 381890
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.121795362363205e-05,
+      "loss": 0.8734,
+      "step": 381900
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.12168056937323e-05,
+      "loss": 0.878,
+      "step": 381910
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.1215657763832556e-05,
+      "loss": 0.8143,
+      "step": 381920
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.1214509833932805e-05,
+      "loss": 0.8166,
+      "step": 381930
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.1213361904033055e-05,
+      "loss": 0.9459,
+      "step": 381940
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.121221397413331e-05,
+      "loss": 0.7719,
+      "step": 381950
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.121106604423356e-05,
+      "loss": 0.8249,
+      "step": 381960
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3.120991811433381e-05,
+      "loss": 0.8255,
+      "step": 381970
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3.120877018443407e-05,
+      "loss": 0.8854,
+      "step": 381980
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3.120762225453432e-05,
+      "loss": 0.8795,
+      "step": 381990
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3.120647432463457e-05,
+      "loss": 0.8601,
+      "step": 382000
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3.1205326394734825e-05,
+      "loss": 0.8631,
+      "step": 382010
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3.1204178464835075e-05,
+      "loss": 0.8926,
+      "step": 382020
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3.120303053493533e-05,
+      "loss": 0.8713,
+      "step": 382030
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3.120188260503559e-05,
+      "loss": 0.816,
+      "step": 382040
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3.120073467513584e-05,
+      "loss": 0.8478,
+      "step": 382050
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3.119958674523609e-05,
+      "loss": 0.927,
+      "step": 382060
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3.1198438815336344e-05,
+      "loss": 0.9257,
+      "step": 382070
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3.1197290885436594e-05,
+      "loss": 0.8512,
+      "step": 382080
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3.1196142955536844e-05,
+      "loss": 0.8388,
+      "step": 382090
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3.11949950256371e-05,
+      "loss": 0.8643,
+      "step": 382100
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3.119384709573735e-05,
+      "loss": 1.0067,
+      "step": 382110
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3.11926991658376e-05,
+      "loss": 0.7754,
+      "step": 382120
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3.119155123593786e-05,
+      "loss": 0.8734,
+      "step": 382130
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3.119040330603811e-05,
+      "loss": 0.801,
+      "step": 382140
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3.1189255376138357e-05,
+      "loss": 0.8121,
+      "step": 382150
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3.118810744623861e-05,
+      "loss": 0.8219,
+      "step": 382160
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3.118695951633886e-05,
+      "loss": 0.8751,
+      "step": 382170
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.118581158643912e-05,
+      "loss": 0.7771,
+      "step": 382180
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.1184663656539376e-05,
+      "loss": 0.8123,
+      "step": 382190
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.1183515726639626e-05,
+      "loss": 0.8062,
+      "step": 382200
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.1182367796739876e-05,
+      "loss": 0.8083,
+      "step": 382210
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.118121986684013e-05,
+      "loss": 0.919,
+      "step": 382220
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.118007193694038e-05,
+      "loss": 0.9197,
+      "step": 382230
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.117892400704063e-05,
+      "loss": 0.8728,
+      "step": 382240
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.117777607714089e-05,
+      "loss": 0.8331,
+      "step": 382250
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.117662814724114e-05,
+      "loss": 0.845,
+      "step": 382260
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.117548021734139e-05,
+      "loss": 0.9251,
+      "step": 382270
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.1174332287441645e-05,
+      "loss": 0.8463,
+      "step": 382280
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.1173184357541895e-05,
+      "loss": 0.8806,
+      "step": 382290
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.1172036427642145e-05,
+      "loss": 0.8439,
+      "step": 382300
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.11708884977424e-05,
+      "loss": 0.8532,
+      "step": 382310
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.116974056784265e-05,
+      "loss": 0.9053,
+      "step": 382320
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.116859263794291e-05,
+      "loss": 0.9201,
+      "step": 382330
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.1167444708043165e-05,
+      "loss": 0.905,
+      "step": 382340
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.1166296778143415e-05,
+      "loss": 0.9,
+      "step": 382350
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.1165148848243664e-05,
+      "loss": 0.9195,
+      "step": 382360
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.116400091834392e-05,
+      "loss": 0.8872,
+      "step": 382370
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.116285298844417e-05,
+      "loss": 0.8437,
+      "step": 382380
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.116170505854442e-05,
+      "loss": 0.8932,
+      "step": 382390
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.116055712864468e-05,
+      "loss": 1.0094,
+      "step": 382400
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.115940919874493e-05,
+      "loss": 0.9097,
+      "step": 382410
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.115826126884518e-05,
+      "loss": 0.7939,
+      "step": 382420
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.115711333894543e-05,
+      "loss": 0.8736,
+      "step": 382430
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.1155965409045684e-05,
+      "loss": 0.8419,
+      "step": 382440
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.1154817479145933e-05,
+      "loss": 0.823,
+      "step": 382450
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.115366954924619e-05,
+      "loss": 0.8871,
+      "step": 382460
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.115252161934644e-05,
+      "loss": 0.8348,
+      "step": 382470
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.1151373689446697e-05,
+      "loss": 0.8542,
+      "step": 382480
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.1150225759546946e-05,
+      "loss": 0.8346,
+      "step": 382490
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.11490778296472e-05,
+      "loss": 0.8507,
+      "step": 382500
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.114792989974745e-05,
+      "loss": 0.8334,
+      "step": 382510
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.11467819698477e-05,
+      "loss": 0.8665,
+      "step": 382520
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.114563403994796e-05,
+      "loss": 0.8763,
+      "step": 382530
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.114448611004821e-05,
+      "loss": 0.766,
+      "step": 382540
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.114333818014846e-05,
+      "loss": 0.7645,
+      "step": 382550
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.1142190250248716e-05,
+      "loss": 0.9517,
+      "step": 382560
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.1141042320348966e-05,
+      "loss": 0.804,
+      "step": 382570
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.1139894390449215e-05,
+      "loss": 0.7678,
+      "step": 382580
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.113874646054947e-05,
+      "loss": 0.909,
+      "step": 382590
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.113759853064972e-05,
+      "loss": 0.8691,
+      "step": 382600
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.113645060074998e-05,
+      "loss": 0.8163,
+      "step": 382610
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.113530267085023e-05,
+      "loss": 0.8072,
+      "step": 382620
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.1134154740950485e-05,
+      "loss": 0.8734,
+      "step": 382630
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.1133006811050735e-05,
+      "loss": 0.8793,
+      "step": 382640
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.113185888115099e-05,
+      "loss": 0.8931,
+      "step": 382650
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.113071095125124e-05,
+      "loss": 0.8713,
+      "step": 382660
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.112956302135149e-05,
+      "loss": 0.8922,
+      "step": 382670
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.112841509145175e-05,
+      "loss": 0.9416,
+      "step": 382680
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.1127267161552e-05,
+      "loss": 0.9229,
+      "step": 382690
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.112611923165225e-05,
+      "loss": 0.8612,
+      "step": 382700
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.1124971301752504e-05,
+      "loss": 0.8341,
+      "step": 382710
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.1123823371852754e-05,
+      "loss": 0.8952,
+      "step": 382720
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.1122675441953004e-05,
+      "loss": 0.8838,
+      "step": 382730
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.112152751205326e-05,
+      "loss": 0.8623,
+      "step": 382740
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.112037958215351e-05,
+      "loss": 0.7967,
+      "step": 382750
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.111923165225377e-05,
+      "loss": 0.8686,
+      "step": 382760
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.111808372235402e-05,
+      "loss": 0.923,
+      "step": 382770
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.1116935792454274e-05,
+      "loss": 0.7971,
+      "step": 382780
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.111578786255452e-05,
+      "loss": 0.9441,
+      "step": 382790
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.111463993265478e-05,
+      "loss": 0.8564,
+      "step": 382800
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.111349200275503e-05,
+      "loss": 0.886,
+      "step": 382810
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.111234407285528e-05,
+      "loss": 0.8886,
+      "step": 382820
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.1111196142955536e-05,
+      "loss": 0.777,
+      "step": 382830
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.1110048213055786e-05,
+      "loss": 0.8265,
+      "step": 382840
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.1108900283156036e-05,
+      "loss": 0.8842,
+      "step": 382850
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.110775235325629e-05,
+      "loss": 0.7834,
+      "step": 382860
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.110660442335654e-05,
+      "loss": 0.8635,
+      "step": 382870
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.110545649345679e-05,
+      "loss": 0.9323,
+      "step": 382880
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.110430856355705e-05,
+      "loss": 0.8319,
+      "step": 382890
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.11031606336573e-05,
+      "loss": 0.8758,
+      "step": 382900
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.1102012703757555e-05,
+      "loss": 0.8352,
+      "step": 382910
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.1100864773857805e-05,
+      "loss": 0.7811,
+      "step": 382920
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.109971684395806e-05,
+      "loss": 0.8463,
+      "step": 382930
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.109856891405831e-05,
+      "loss": 0.8925,
+      "step": 382940
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.109742098415857e-05,
+      "loss": 0.8219,
+      "step": 382950
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.109627305425882e-05,
+      "loss": 0.9238,
+      "step": 382960
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.109512512435907e-05,
+      "loss": 0.9038,
+      "step": 382970
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.1093977194459325e-05,
+      "loss": 0.9702,
+      "step": 382980
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.1092829264559575e-05,
+      "loss": 0.8446,
+      "step": 382990
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.1091681334659825e-05,
+      "loss": 1.02,
+      "step": 383000
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.109053340476008e-05,
+      "loss": 0.9684,
+      "step": 383010
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.108938547486033e-05,
+      "loss": 0.8496,
+      "step": 383020
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.108823754496058e-05,
+      "loss": 0.8403,
+      "step": 383030
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.108708961506084e-05,
+      "loss": 0.8186,
+      "step": 383040
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.108594168516109e-05,
+      "loss": 0.8196,
+      "step": 383050
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 3.1084793755261344e-05,
+      "loss": 0.9373,
+      "step": 383060
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 3.1083645825361594e-05,
+      "loss": 0.8552,
+      "step": 383070
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 3.108249789546185e-05,
+      "loss": 0.9173,
+      "step": 383080
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 3.10813499655621e-05,
+      "loss": 0.9992,
+      "step": 383090
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 3.108020203566236e-05,
+      "loss": 0.8963,
+      "step": 383100
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 3.107905410576261e-05,
+      "loss": 0.9312,
+      "step": 383110
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 3.107790617586286e-05,
+      "loss": 0.8621,
+      "step": 383120
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 3.1076758245963106e-05,
+      "loss": 0.9139,
+      "step": 383130
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 3.107561031606336e-05,
+      "loss": 0.8226,
+      "step": 383140
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 3.107446238616361e-05,
+      "loss": 0.9047,
+      "step": 383150
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 3.107331445626386e-05,
+      "loss": 0.8146,
+      "step": 383160
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 3.107216652636412e-05,
+      "loss": 0.9266,
+      "step": 383170
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 3.107101859646437e-05,
+      "loss": 0.8277,
+      "step": 383180
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 3.1069870666564626e-05,
+      "loss": 0.9135,
+      "step": 383190
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 3.1068722736664876e-05,
+      "loss": 0.857,
+      "step": 383200
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 3.106757480676513e-05,
+      "loss": 0.8677,
+      "step": 383210
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 3.106642687686538e-05,
+      "loss": 0.8078,
+      "step": 383220
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 3.106527894696564e-05,
+      "loss": 0.8991,
+      "step": 383230
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 3.106413101706589e-05,
+      "loss": 0.8268,
+      "step": 383240
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 3.106298308716614e-05,
+      "loss": 0.9503,
+      "step": 383250
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 3.1061835157266395e-05,
+      "loss": 0.8707,
+      "step": 383260
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.1060687227366645e-05,
+      "loss": 0.8547,
+      "step": 383270
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.1059539297466895e-05,
+      "loss": 0.8658,
+      "step": 383280
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.105839136756715e-05,
+      "loss": 0.9857,
+      "step": 383290
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.10572434376674e-05,
+      "loss": 0.8661,
+      "step": 383300
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.105609550776765e-05,
+      "loss": 0.8901,
+      "step": 383310
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.105494757786791e-05,
+      "loss": 0.9191,
+      "step": 383320
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.105379964796816e-05,
+      "loss": 0.8782,
+      "step": 383330
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.1052651718068414e-05,
+      "loss": 0.9142,
+      "step": 383340
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.1051503788168664e-05,
+      "loss": 0.9064,
+      "step": 383350
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.105035585826892e-05,
+      "loss": 0.8275,
+      "step": 383360
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.104920792836917e-05,
+      "loss": 0.9198,
+      "step": 383370
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.104805999846943e-05,
+      "loss": 0.9651,
+      "step": 383380
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.104691206856968e-05,
+      "loss": 0.8723,
+      "step": 383390
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.104576413866993e-05,
+      "loss": 0.8538,
+      "step": 383400
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.1044616208770184e-05,
+      "loss": 0.9714,
+      "step": 383410
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.1043468278870434e-05,
+      "loss": 0.8345,
+      "step": 383420
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.1042320348970683e-05,
+      "loss": 0.8372,
+      "step": 383430
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.104117241907094e-05,
+      "loss": 0.9497,
+      "step": 383440
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.104002448917119e-05,
+      "loss": 0.8297,
+      "step": 383450
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.103887655927144e-05,
+      "loss": 0.9412,
+      "step": 383460
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.1037728629371696e-05,
+      "loss": 0.8702,
+      "step": 383470
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.1036580699471946e-05,
+      "loss": 0.9404,
+      "step": 383480
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.10354327695722e-05,
+      "loss": 0.8669,
+      "step": 383490
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.103428483967245e-05,
+      "loss": 0.8266,
+      "step": 383500
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.103313690977271e-05,
+      "loss": 0.8138,
+      "step": 383510
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.103198897987296e-05,
+      "loss": 0.8454,
+      "step": 383520
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.1030841049973216e-05,
+      "loss": 0.8956,
+      "step": 383530
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.1029693120073466e-05,
+      "loss": 0.8425,
+      "step": 383540
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.1028545190173716e-05,
+      "loss": 0.7881,
+      "step": 383550
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.102739726027397e-05,
+      "loss": 0.8641,
+      "step": 383560
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.102624933037422e-05,
+      "loss": 0.9112,
+      "step": 383570
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.102510140047447e-05,
+      "loss": 0.8704,
+      "step": 383580
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.102395347057473e-05,
+      "loss": 0.9645,
+      "step": 383590
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.102280554067498e-05,
+      "loss": 0.8446,
+      "step": 383600
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.102165761077523e-05,
+      "loss": 0.7958,
+      "step": 383610
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.1020509680875485e-05,
+      "loss": 0.8079,
+      "step": 383620
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.1019361750975735e-05,
+      "loss": 0.8281,
+      "step": 383630
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.101821382107599e-05,
+      "loss": 0.8255,
+      "step": 383640
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.101706589117624e-05,
+      "loss": 0.9328,
+      "step": 383650
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.10159179612765e-05,
+      "loss": 1.0012,
+      "step": 383660
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.101477003137675e-05,
+      "loss": 0.9481,
+      "step": 383670
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.1013622101477004e-05,
+      "loss": 0.9065,
+      "step": 383680
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.1012474171577254e-05,
+      "loss": 0.9493,
+      "step": 383690
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.1011326241677504e-05,
+      "loss": 0.8864,
+      "step": 383700
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.101017831177776e-05,
+      "loss": 0.8303,
+      "step": 383710
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.100903038187801e-05,
+      "loss": 0.8776,
+      "step": 383720
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.100788245197826e-05,
+      "loss": 0.8733,
+      "step": 383730
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.100673452207852e-05,
+      "loss": 0.8817,
+      "step": 383740
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.100558659217877e-05,
+      "loss": 0.8961,
+      "step": 383750
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.100443866227902e-05,
+      "loss": 0.8542,
+      "step": 383760
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.100329073237927e-05,
+      "loss": 0.9263,
+      "step": 383770
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.100214280247952e-05,
+      "loss": 0.8511,
+      "step": 383780
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.100099487257978e-05,
+      "loss": 0.9373,
+      "step": 383790
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.099984694268003e-05,
+      "loss": 0.8999,
+      "step": 383800
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.0998699012780286e-05,
+      "loss": 0.9689,
+      "step": 383810
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.0997551082880536e-05,
+      "loss": 0.823,
+      "step": 383820
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.0996403152980786e-05,
+      "loss": 0.8093,
+      "step": 383830
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.099525522308104e-05,
+      "loss": 0.8543,
+      "step": 383840
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.099410729318129e-05,
+      "loss": 0.9265,
+      "step": 383850
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.099295936328154e-05,
+      "loss": 0.8264,
+      "step": 383860
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.09918114333818e-05,
+      "loss": 0.8271,
+      "step": 383870
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.099066350348205e-05,
+      "loss": 0.8523,
+      "step": 383880
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.09895155735823e-05,
+      "loss": 0.8939,
+      "step": 383890
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.0988367643682555e-05,
+      "loss": 0.7961,
+      "step": 383900
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.0987219713782805e-05,
+      "loss": 0.8137,
+      "step": 383910
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.098607178388306e-05,
+      "loss": 0.9354,
+      "step": 383920
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.098492385398331e-05,
+      "loss": 0.9478,
+      "step": 383930
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.098377592408357e-05,
+      "loss": 0.8385,
+      "step": 383940
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.098262799418382e-05,
+      "loss": 0.9071,
+      "step": 383950
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.0981480064284075e-05,
+      "loss": 1.0339,
+      "step": 383960
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.0980332134384325e-05,
+      "loss": 0.8825,
+      "step": 383970
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.0979184204484575e-05,
+      "loss": 0.8525,
+      "step": 383980
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.097803627458483e-05,
+      "loss": 0.8536,
+      "step": 383990
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.097688834468508e-05,
+      "loss": 0.7954,
+      "step": 384000
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.097574041478533e-05,
+      "loss": 0.8278,
+      "step": 384010
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.097459248488559e-05,
+      "loss": 0.8103,
+      "step": 384020
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.097344455498584e-05,
+      "loss": 0.961,
+      "step": 384030
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.097229662508609e-05,
+      "loss": 0.9343,
+      "step": 384040
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.0971148695186344e-05,
+      "loss": 0.8997,
+      "step": 384050
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.0970000765286594e-05,
+      "loss": 0.7883,
+      "step": 384060
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.096885283538685e-05,
+      "loss": 0.9903,
+      "step": 384070
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.09677049054871e-05,
+      "loss": 0.8925,
+      "step": 384080
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.096655697558736e-05,
+      "loss": 0.941,
+      "step": 384090
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.096540904568761e-05,
+      "loss": 0.8315,
+      "step": 384100
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.096426111578786e-05,
+      "loss": 0.9742,
+      "step": 384110
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.096311318588811e-05,
+      "loss": 0.9694,
+      "step": 384120
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.096196525598836e-05,
+      "loss": 0.9499,
+      "step": 384130
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.096081732608862e-05,
+      "loss": 0.8159,
+      "step": 384140
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3.095966939618887e-05,
+      "loss": 0.9156,
+      "step": 384150
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3.095852146628912e-05,
+      "loss": 0.8484,
+      "step": 384160
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3.0957373536389376e-05,
+      "loss": 0.8613,
+      "step": 384170
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3.0956225606489626e-05,
+      "loss": 0.82,
+      "step": 384180
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3.0955077676589876e-05,
+      "loss": 1.004,
+      "step": 384190
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3.095392974669013e-05,
+      "loss": 0.8687,
+      "step": 384200
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3.095278181679038e-05,
+      "loss": 0.8309,
+      "step": 384210
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3.095163388689064e-05,
+      "loss": 0.8795,
+      "step": 384220
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3.095048595699089e-05,
+      "loss": 1.0522,
+      "step": 384230
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3.0949338027091145e-05,
+      "loss": 0.861,
+      "step": 384240
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3.0948190097191395e-05,
+      "loss": 0.8766,
+      "step": 384250
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3.094704216729165e-05,
+      "loss": 0.8793,
+      "step": 384260
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3.09458942373919e-05,
+      "loss": 0.8008,
+      "step": 384270
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3.094474630749215e-05,
+      "loss": 0.9058,
+      "step": 384280
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3.094359837759241e-05,
+      "loss": 0.8456,
+      "step": 384290
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3.094245044769266e-05,
+      "loss": 0.8532,
+      "step": 384300
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3.094130251779291e-05,
+      "loss": 1.0358,
+      "step": 384310
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3.0940154587893164e-05,
+      "loss": 0.8424,
+      "step": 384320
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3.0939006657993414e-05,
+      "loss": 0.8899,
+      "step": 384330
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3.0937858728093664e-05,
+      "loss": 0.8973,
+      "step": 384340
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3.093671079819392e-05,
+      "loss": 0.8531,
+      "step": 384350
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.093556286829417e-05,
+      "loss": 0.7554,
+      "step": 384360
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.093441493839443e-05,
+      "loss": 0.8288,
+      "step": 384370
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.093326700849468e-05,
+      "loss": 0.8989,
+      "step": 384380
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.0932119078594934e-05,
+      "loss": 0.8646,
+      "step": 384390
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.0930971148695184e-05,
+      "loss": 0.8709,
+      "step": 384400
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.092982321879544e-05,
+      "loss": 0.8932,
+      "step": 384410
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.092867528889569e-05,
+      "loss": 0.9023,
+      "step": 384420
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.092752735899594e-05,
+      "loss": 0.8531,
+      "step": 384430
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.0926379429096197e-05,
+      "loss": 0.9886,
+      "step": 384440
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.0925231499196446e-05,
+      "loss": 0.7988,
+      "step": 384450
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.0924083569296696e-05,
+      "loss": 0.8874,
+      "step": 384460
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.092293563939695e-05,
+      "loss": 0.8788,
+      "step": 384470
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.09217877094972e-05,
+      "loss": 0.8658,
+      "step": 384480
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.092063977959745e-05,
+      "loss": 0.8962,
+      "step": 384490
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.091949184969771e-05,
+      "loss": 0.8345,
+      "step": 384500
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.091834391979796e-05,
+      "loss": 0.8303,
+      "step": 384510
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.0917195989898216e-05,
+      "loss": 0.918,
+      "step": 384520
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.0916048059998466e-05,
+      "loss": 0.936,
+      "step": 384530
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.091490013009872e-05,
+      "loss": 0.8614,
+      "step": 384540
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.091375220019897e-05,
+      "loss": 0.827,
+      "step": 384550
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.091260427029922e-05,
+      "loss": 0.8628,
+      "step": 384560
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.091145634039948e-05,
+      "loss": 0.9128,
+      "step": 384570
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.091030841049973e-05,
+      "loss": 0.8672,
+      "step": 384580
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.090916048059998e-05,
+      "loss": 0.9438,
+      "step": 384590
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.0908012550700235e-05,
+      "loss": 0.8514,
+      "step": 384600
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.0906864620800485e-05,
+      "loss": 0.7397,
+      "step": 384610
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.0905716690900735e-05,
+      "loss": 0.8538,
+      "step": 384620
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.090456876100099e-05,
+      "loss": 0.9226,
+      "step": 384630
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.090342083110124e-05,
+      "loss": 0.8342,
+      "step": 384640
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.09022729012015e-05,
+      "loss": 0.9084,
+      "step": 384650
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.090112497130175e-05,
+      "loss": 0.8859,
+      "step": 384660
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.0899977041402004e-05,
+      "loss": 0.8969,
+      "step": 384670
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.0898829111502254e-05,
+      "loss": 0.9292,
+      "step": 384680
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.089768118160251e-05,
+      "loss": 0.8615,
+      "step": 384690
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.089653325170276e-05,
+      "loss": 0.8927,
+      "step": 384700
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.089538532180301e-05,
+      "loss": 0.8429,
+      "step": 384710
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.089423739190327e-05,
+      "loss": 0.8617,
+      "step": 384720
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.089308946200352e-05,
+      "loss": 0.973,
+      "step": 384730
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.089194153210377e-05,
+      "loss": 0.8826,
+      "step": 384740
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.089079360220402e-05,
+      "loss": 0.867,
+      "step": 384750
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.088964567230427e-05,
+      "loss": 0.9515,
+      "step": 384760
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.088849774240452e-05,
+      "loss": 0.8394,
+      "step": 384770
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.088734981250478e-05,
+      "loss": 0.8367,
+      "step": 384780
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.088620188260503e-05,
+      "loss": 0.8635,
+      "step": 384790
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.0885053952705286e-05,
+      "loss": 0.9355,
+      "step": 384800
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.0883906022805536e-05,
+      "loss": 1.0398,
+      "step": 384810
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.088275809290579e-05,
+      "loss": 0.7821,
+      "step": 384820
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.088161016300604e-05,
+      "loss": 0.8759,
+      "step": 384830
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.08804622331063e-05,
+      "loss": 0.8655,
+      "step": 384840
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.087931430320655e-05,
+      "loss": 0.7999,
+      "step": 384850
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.08781663733068e-05,
+      "loss": 0.82,
+      "step": 384860
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.0877018443407055e-05,
+      "loss": 0.8738,
+      "step": 384870
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.0875870513507305e-05,
+      "loss": 0.8568,
+      "step": 384880
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.0874722583607555e-05,
+      "loss": 0.8624,
+      "step": 384890
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.087357465370781e-05,
+      "loss": 0.975,
+      "step": 384900
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.087242672380806e-05,
+      "loss": 0.9327,
+      "step": 384910
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.087127879390831e-05,
+      "loss": 0.8584,
+      "step": 384920
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.087013086400857e-05,
+      "loss": 0.8909,
+      "step": 384930
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.086898293410882e-05,
+      "loss": 0.8561,
+      "step": 384940
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.0867835004209075e-05,
+      "loss": 0.786,
+      "step": 384950
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.0866687074309325e-05,
+      "loss": 0.902,
+      "step": 384960
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.086553914440958e-05,
+      "loss": 0.8709,
+      "step": 384970
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.086439121450983e-05,
+      "loss": 0.8401,
+      "step": 384980
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.086324328461009e-05,
+      "loss": 0.8552,
+      "step": 384990
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.086209535471034e-05,
+      "loss": 0.837,
+      "step": 385000
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.086094742481059e-05,
+      "loss": 0.8809,
+      "step": 385010
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 3.0859799494910844e-05,
+      "loss": 0.8414,
+      "step": 385020
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 3.0858651565011094e-05,
+      "loss": 0.8607,
+      "step": 385030
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 3.0857503635111344e-05,
+      "loss": 0.8362,
+      "step": 385040
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 3.08563557052116e-05,
+      "loss": 0.8619,
+      "step": 385050
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 3.085520777531185e-05,
+      "loss": 0.8926,
+      "step": 385060
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 3.08540598454121e-05,
+      "loss": 0.8866,
+      "step": 385070
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 3.085291191551236e-05,
+      "loss": 0.8387,
+      "step": 385080
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 3.0851763985612606e-05,
+      "loss": 0.9114,
+      "step": 385090
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 3.085061605571286e-05,
+      "loss": 0.8869,
+      "step": 385100
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 3.084946812581311e-05,
+      "loss": 0.7562,
+      "step": 385110
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 3.084832019591337e-05,
+      "loss": 0.8027,
+      "step": 385120
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 3.084717226601362e-05,
+      "loss": 0.8034,
+      "step": 385130
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 3.0846024336113876e-05,
+      "loss": 0.8456,
+      "step": 385140
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 3.0844876406214126e-05,
+      "loss": 0.8239,
+      "step": 385150
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 3.0843728476314376e-05,
+      "loss": 0.9026,
+      "step": 385160
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 3.0842580546414626e-05,
+      "loss": 0.8921,
+      "step": 385170
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 3.084143261651488e-05,
+      "loss": 0.8101,
+      "step": 385180
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 3.084028468661513e-05,
+      "loss": 0.9707,
+      "step": 385190
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 3.083913675671538e-05,
+      "loss": 0.877,
+      "step": 385200
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 3.083798882681564e-05,
+      "loss": 0.921,
+      "step": 385210
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 3.083684089691589e-05,
+      "loss": 0.8795,
+      "step": 385220
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.0835692967016145e-05,
+      "loss": 0.8448,
+      "step": 385230
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.0834545037116395e-05,
+      "loss": 0.8796,
+      "step": 385240
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.083339710721665e-05,
+      "loss": 0.8262,
+      "step": 385250
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.08322491773169e-05,
+      "loss": 0.8017,
+      "step": 385260
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.083110124741716e-05,
+      "loss": 0.8626,
+      "step": 385270
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.082995331751741e-05,
+      "loss": 0.783,
+      "step": 385280
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.082880538761766e-05,
+      "loss": 0.9243,
+      "step": 385290
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.0827657457717914e-05,
+      "loss": 0.8158,
+      "step": 385300
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.0826509527818164e-05,
+      "loss": 0.8502,
+      "step": 385310
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.0825361597918414e-05,
+      "loss": 0.8215,
+      "step": 385320
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.082421366801867e-05,
+      "loss": 0.7972,
+      "step": 385330
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.082306573811892e-05,
+      "loss": 0.9149,
+      "step": 385340
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.082191780821917e-05,
+      "loss": 0.8593,
+      "step": 385350
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.082076987831943e-05,
+      "loss": 0.8309,
+      "step": 385360
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.081962194841968e-05,
+      "loss": 0.9652,
+      "step": 385370
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.0818474018519934e-05,
+      "loss": 0.9324,
+      "step": 385380
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.0817326088620183e-05,
+      "loss": 0.8491,
+      "step": 385390
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.081617815872044e-05,
+      "loss": 0.8703,
+      "step": 385400
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.081503022882069e-05,
+      "loss": 0.8219,
+      "step": 385410
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.0813882298920947e-05,
+      "loss": 0.8464,
+      "step": 385420
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.0812734369021196e-05,
+      "loss": 0.8946,
+      "step": 385430
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.0811586439121446e-05,
+      "loss": 0.92,
+      "step": 385440
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.08104385092217e-05,
+      "loss": 0.8808,
+      "step": 385450
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.080929057932195e-05,
+      "loss": 0.9359,
+      "step": 385460
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.08081426494222e-05,
+      "loss": 1.0571,
+      "step": 385470
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.080699471952246e-05,
+      "loss": 0.7941,
+      "step": 385480
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.080584678962271e-05,
+      "loss": 0.9004,
+      "step": 385490
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.080469885972296e-05,
+      "loss": 0.9176,
+      "step": 385500
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.0803550929823216e-05,
+      "loss": 0.8178,
+      "step": 385510
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.0802402999923465e-05,
+      "loss": 0.8406,
+      "step": 385520
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.080125507002372e-05,
+      "loss": 0.9157,
+      "step": 385530
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.080010714012397e-05,
+      "loss": 0.8411,
+      "step": 385540
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.079895921022423e-05,
+      "loss": 0.8097,
+      "step": 385550
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.079781128032448e-05,
+      "loss": 0.8239,
+      "step": 385560
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.0796663350424735e-05,
+      "loss": 0.8458,
+      "step": 385570
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.0795515420524985e-05,
+      "loss": 0.8584,
+      "step": 385580
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.0794367490625235e-05,
+      "loss": 0.8525,
+      "step": 385590
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.079321956072549e-05,
+      "loss": 0.8764,
+      "step": 385600
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.079207163082574e-05,
+      "loss": 0.8543,
+      "step": 385610
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.079092370092599e-05,
+      "loss": 0.9147,
+      "step": 385620
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.078977577102625e-05,
+      "loss": 0.9474,
+      "step": 385630
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.07886278411265e-05,
+      "loss": 0.8828,
+      "step": 385640
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.078747991122675e-05,
+      "loss": 0.8206,
+      "step": 385650
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.0786331981327004e-05,
+      "loss": 0.8358,
+      "step": 385660
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.0785184051427254e-05,
+      "loss": 0.938,
+      "step": 385670
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.078403612152751e-05,
+      "loss": 0.8638,
+      "step": 385680
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.078288819162776e-05,
+      "loss": 0.9445,
+      "step": 385690
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.078174026172802e-05,
+      "loss": 0.8735,
+      "step": 385700
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.078059233182827e-05,
+      "loss": 0.8243,
+      "step": 385710
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.0779444401928523e-05,
+      "loss": 0.91,
+      "step": 385720
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.077829647202877e-05,
+      "loss": 0.8779,
+      "step": 385730
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.077714854212902e-05,
+      "loss": 0.8525,
+      "step": 385740
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.077600061222928e-05,
+      "loss": 0.8662,
+      "step": 385750
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.077485268232953e-05,
+      "loss": 0.9127,
+      "step": 385760
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.077370475242978e-05,
+      "loss": 0.7874,
+      "step": 385770
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.0772556822530036e-05,
+      "loss": 0.9044,
+      "step": 385780
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.0771408892630286e-05,
+      "loss": 0.8616,
+      "step": 385790
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.0770260962730536e-05,
+      "loss": 0.9069,
+      "step": 385800
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.076911303283079e-05,
+      "loss": 1.0109,
+      "step": 385810
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.076796510293104e-05,
+      "loss": 0.9055,
+      "step": 385820
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.07668171730313e-05,
+      "loss": 0.8576,
+      "step": 385830
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.076566924313155e-05,
+      "loss": 0.8185,
+      "step": 385840
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.0764521313231805e-05,
+      "loss": 0.8728,
+      "step": 385850
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.0763373383332055e-05,
+      "loss": 0.8955,
+      "step": 385860
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.0762225453432305e-05,
+      "loss": 0.9761,
+      "step": 385870
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.076107752353256e-05,
+      "loss": 0.8899,
+      "step": 385880
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.075992959363281e-05,
+      "loss": 0.9131,
+      "step": 385890
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.075878166373306e-05,
+      "loss": 0.8175,
+      "step": 385900
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.075763373383332e-05,
+      "loss": 0.8665,
+      "step": 385910
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.075648580393357e-05,
+      "loss": 0.8215,
+      "step": 385920
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.075533787403382e-05,
+      "loss": 0.8796,
+      "step": 385930
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.0754189944134074e-05,
+      "loss": 0.9235,
+      "step": 385940
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.0753042014234324e-05,
+      "loss": 0.9157,
+      "step": 385950
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.075189408433458e-05,
+      "loss": 0.8473,
+      "step": 385960
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.075074615443483e-05,
+      "loss": 0.9414,
+      "step": 385970
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.074959822453509e-05,
+      "loss": 0.9384,
+      "step": 385980
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.074845029463534e-05,
+      "loss": 0.9023,
+      "step": 385990
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.0747302364735594e-05,
+      "loss": 0.9464,
+      "step": 386000
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.0746154434835844e-05,
+      "loss": 0.8181,
+      "step": 386010
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.0745006504936094e-05,
+      "loss": 0.8572,
+      "step": 386020
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.074385857503635e-05,
+      "loss": 0.9069,
+      "step": 386030
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.07427106451366e-05,
+      "loss": 0.8853,
+      "step": 386040
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.074156271523685e-05,
+      "loss": 0.9264,
+      "step": 386050
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.074041478533711e-05,
+      "loss": 0.8591,
+      "step": 386060
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.0739266855437356e-05,
+      "loss": 0.8931,
+      "step": 386070
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.0738118925537606e-05,
+      "loss": 0.9649,
+      "step": 386080
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.073697099563786e-05,
+      "loss": 0.8196,
+      "step": 386090
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.073582306573811e-05,
+      "loss": 0.9298,
+      "step": 386100
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3.073467513583837e-05,
+      "loss": 0.8527,
+      "step": 386110
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3.073352720593862e-05,
+      "loss": 0.8671,
+      "step": 386120
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3.0732379276038876e-05,
+      "loss": 0.9626,
+      "step": 386130
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3.0731231346139126e-05,
+      "loss": 0.8643,
+      "step": 386140
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3.073008341623938e-05,
+      "loss": 0.9805,
+      "step": 386150
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3.072893548633963e-05,
+      "loss": 0.9168,
+      "step": 386160
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3.072778755643988e-05,
+      "loss": 0.9057,
+      "step": 386170
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3.072663962654014e-05,
+      "loss": 0.7918,
+      "step": 386180
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3.072549169664039e-05,
+      "loss": 0.8103,
+      "step": 386190
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3.072434376674064e-05,
+      "loss": 0.8216,
+      "step": 386200
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3.0723195836840895e-05,
+      "loss": 0.8621,
+      "step": 386210
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3.0722047906941145e-05,
+      "loss": 0.8482,
+      "step": 386220
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3.0720899977041395e-05,
+      "loss": 0.8453,
+      "step": 386230
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3.071975204714165e-05,
+      "loss": 0.8108,
+      "step": 386240
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3.07186041172419e-05,
+      "loss": 0.8392,
+      "step": 386250
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3.071745618734216e-05,
+      "loss": 0.8148,
+      "step": 386260
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3.071630825744241e-05,
+      "loss": 0.8694,
+      "step": 386270
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3.0715160327542664e-05,
+      "loss": 0.9215,
+      "step": 386280
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3.0714012397642914e-05,
+      "loss": 0.7795,
+      "step": 386290
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3.071286446774317e-05,
+      "loss": 0.8368,
+      "step": 386300
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3.071171653784342e-05,
+      "loss": 0.9592,
+      "step": 386310
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.071056860794367e-05,
+      "loss": 0.8105,
+      "step": 386320
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.070942067804393e-05,
+      "loss": 0.7506,
+      "step": 386330
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.070827274814418e-05,
+      "loss": 0.8896,
+      "step": 386340
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.070712481824443e-05,
+      "loss": 0.8934,
+      "step": 386350
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.0705976888344684e-05,
+      "loss": 0.9914,
+      "step": 386360
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.0704828958444933e-05,
+      "loss": 0.8819,
+      "step": 386370
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.070368102854518e-05,
+      "loss": 0.8662,
+      "step": 386380
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.070253309864544e-05,
+      "loss": 0.8895,
+      "step": 386390
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.070138516874569e-05,
+      "loss": 0.9683,
+      "step": 386400
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.0700237238845946e-05,
+      "loss": 0.954,
+      "step": 386410
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.0699089308946196e-05,
+      "loss": 0.8978,
+      "step": 386420
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.069794137904645e-05,
+      "loss": 0.9615,
+      "step": 386430
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.06967934491467e-05,
+      "loss": 0.8547,
+      "step": 386440
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.069564551924696e-05,
+      "loss": 0.9542,
+      "step": 386450
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.069449758934721e-05,
+      "loss": 0.8923,
+      "step": 386460
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.069334965944746e-05,
+      "loss": 0.7963,
+      "step": 386470
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.0692201729547716e-05,
+      "loss": 0.8913,
+      "step": 386480
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.0691053799647966e-05,
+      "loss": 0.8055,
+      "step": 386490
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.0689905869748215e-05,
+      "loss": 0.8866,
+      "step": 386500
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.068875793984847e-05,
+      "loss": 0.8711,
+      "step": 386510
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.068761000994872e-05,
+      "loss": 0.8877,
+      "step": 386520
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.068646208004897e-05,
+      "loss": 0.8713,
+      "step": 386530
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.068531415014923e-05,
+      "loss": 1.0011,
+      "step": 386540
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.068416622024948e-05,
+      "loss": 0.9522,
+      "step": 386550
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.0683018290349735e-05,
+      "loss": 0.8402,
+      "step": 386560
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.0681870360449985e-05,
+      "loss": 0.85,
+      "step": 386570
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.068072243055024e-05,
+      "loss": 0.8747,
+      "step": 386580
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.067957450065049e-05,
+      "loss": 0.9701,
+      "step": 386590
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.067842657075074e-05,
+      "loss": 0.8472,
+      "step": 386600
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.0677278640851e-05,
+      "loss": 0.8519,
+      "step": 386610
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.067613071095125e-05,
+      "loss": 0.954,
+      "step": 386620
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.06749827810515e-05,
+      "loss": 0.8726,
+      "step": 386630
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.0673834851151754e-05,
+      "loss": 0.8597,
+      "step": 386640
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.0672686921252004e-05,
+      "loss": 0.8697,
+      "step": 386650
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.0671538991352254e-05,
+      "loss": 0.8816,
+      "step": 386660
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.067039106145251e-05,
+      "loss": 0.8951,
+      "step": 386670
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.066924313155276e-05,
+      "loss": 0.8468,
+      "step": 386680
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.066809520165302e-05,
+      "loss": 0.9084,
+      "step": 386690
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.066694727175327e-05,
+      "loss": 0.8478,
+      "step": 386700
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.066579934185352e-05,
+      "loss": 0.9448,
+      "step": 386710
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.066465141195377e-05,
+      "loss": 0.8746,
+      "step": 386720
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.066350348205403e-05,
+      "loss": 0.9839,
+      "step": 386730
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.066235555215428e-05,
+      "loss": 0.8964,
+      "step": 386740
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.066120762225453e-05,
+      "loss": 0.8471,
+      "step": 386750
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.0660059692354786e-05,
+      "loss": 0.8304,
+      "step": 386760
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.0658911762455036e-05,
+      "loss": 0.8464,
+      "step": 386770
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.0657763832555286e-05,
+      "loss": 0.9151,
+      "step": 386780
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.065661590265554e-05,
+      "loss": 0.9514,
+      "step": 386790
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.065546797275579e-05,
+      "loss": 0.8879,
+      "step": 386800
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.065432004285604e-05,
+      "loss": 0.8947,
+      "step": 386810
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.06531721129563e-05,
+      "loss": 0.8746,
+      "step": 386820
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.065202418305655e-05,
+      "loss": 0.7841,
+      "step": 386830
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.0650876253156805e-05,
+      "loss": 0.8514,
+      "step": 386840
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.0649728323257055e-05,
+      "loss": 0.9769,
+      "step": 386850
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.064858039335731e-05,
+      "loss": 0.9361,
+      "step": 386860
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.064743246345756e-05,
+      "loss": 0.898,
+      "step": 386870
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.064628453355782e-05,
+      "loss": 0.81,
+      "step": 386880
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.064513660365807e-05,
+      "loss": 0.9219,
+      "step": 386890
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.064398867375832e-05,
+      "loss": 0.8357,
+      "step": 386900
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.0642840743858575e-05,
+      "loss": 0.8403,
+      "step": 386910
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.0641692813958824e-05,
+      "loss": 0.8801,
+      "step": 386920
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.0640544884059074e-05,
+      "loss": 0.863,
+      "step": 386930
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.063939695415933e-05,
+      "loss": 0.8442,
+      "step": 386940
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.063824902425958e-05,
+      "loss": 0.9289,
+      "step": 386950
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.063710109435983e-05,
+      "loss": 0.9142,
+      "step": 386960
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.063595316446009e-05,
+      "loss": 0.838,
+      "step": 386970
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.063480523456034e-05,
+      "loss": 0.8658,
+      "step": 386980
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.0633657304660594e-05,
+      "loss": 0.9007,
+      "step": 386990
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.0632509374760844e-05,
+      "loss": 0.8231,
+      "step": 387000
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.06313614448611e-05,
+      "loss": 0.9011,
+      "step": 387010
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.063021351496135e-05,
+      "loss": 0.9084,
+      "step": 387020
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.062906558506161e-05,
+      "loss": 0.8928,
+      "step": 387030
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.062791765516186e-05,
+      "loss": 0.8463,
+      "step": 387040
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.0626769725262106e-05,
+      "loss": 0.7815,
+      "step": 387050
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.062562179536236e-05,
+      "loss": 0.9079,
+      "step": 387060
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.062447386546261e-05,
+      "loss": 0.807,
+      "step": 387070
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.062332593556286e-05,
+      "loss": 0.9333,
+      "step": 387080
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.062217800566312e-05,
+      "loss": 0.8117,
+      "step": 387090
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.062103007576337e-05,
+      "loss": 0.9418,
+      "step": 387100
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.061988214586362e-05,
+      "loss": 0.8579,
+      "step": 387110
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.0618734215963876e-05,
+      "loss": 0.8694,
+      "step": 387120
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.0617586286064126e-05,
+      "loss": 0.8984,
+      "step": 387130
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.061643835616438e-05,
+      "loss": 0.8301,
+      "step": 387140
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.061529042626463e-05,
+      "loss": 0.8386,
+      "step": 387150
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.061414249636489e-05,
+      "loss": 0.8321,
+      "step": 387160
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.061299456646514e-05,
+      "loss": 0.8187,
+      "step": 387170
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.0611846636565395e-05,
+      "loss": 0.8166,
+      "step": 387180
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.0610698706665645e-05,
+      "loss": 0.8582,
+      "step": 387190
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.0609550776765895e-05,
+      "loss": 0.7819,
+      "step": 387200
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.060840284686615e-05,
+      "loss": 0.8712,
+      "step": 387210
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.06072549169664e-05,
+      "loss": 0.8007,
+      "step": 387220
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.060610698706665e-05,
+      "loss": 0.9129,
+      "step": 387230
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.06049590571669e-05,
+      "loss": 0.9015,
+      "step": 387240
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.060381112726716e-05,
+      "loss": 0.8687,
+      "step": 387250
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.060266319736741e-05,
+      "loss": 0.8944,
+      "step": 387260
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.0601515267467664e-05,
+      "loss": 0.8837,
+      "step": 387270
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.0600367337567914e-05,
+      "loss": 0.8556,
+      "step": 387280
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.059921940766817e-05,
+      "loss": 0.857,
+      "step": 387290
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.059807147776842e-05,
+      "loss": 0.8295,
+      "step": 387300
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.059692354786868e-05,
+      "loss": 0.8257,
+      "step": 387310
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.059577561796893e-05,
+      "loss": 0.7877,
+      "step": 387320
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.059462768806918e-05,
+      "loss": 0.9284,
+      "step": 387330
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.0593479758169434e-05,
+      "loss": 0.9016,
+      "step": 387340
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.0592331828269683e-05,
+      "loss": 0.884,
+      "step": 387350
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.059118389836993e-05,
+      "loss": 0.8569,
+      "step": 387360
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.059003596847019e-05,
+      "loss": 0.8752,
+      "step": 387370
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.058888803857044e-05,
+      "loss": 0.9285,
+      "step": 387380
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.058774010867069e-05,
+      "loss": 0.8775,
+      "step": 387390
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.0586592178770946e-05,
+      "loss": 0.9619,
+      "step": 387400
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.0585444248871196e-05,
+      "loss": 0.9263,
+      "step": 387410
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.058429631897145e-05,
+      "loss": 0.8095,
+      "step": 387420
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.05831483890717e-05,
+      "loss": 0.974,
+      "step": 387430
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.058200045917196e-05,
+      "loss": 0.8773,
+      "step": 387440
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.058085252927221e-05,
+      "loss": 0.8762,
+      "step": 387450
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.0579704599372466e-05,
+      "loss": 0.8554,
+      "step": 387460
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.0578556669472716e-05,
+      "loss": 0.8335,
+      "step": 387470
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.0577408739572965e-05,
+      "loss": 0.8846,
+      "step": 387480
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.057626080967322e-05,
+      "loss": 0.8786,
+      "step": 387490
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.057511287977347e-05,
+      "loss": 0.8537,
+      "step": 387500
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.057396494987372e-05,
+      "loss": 0.9208,
+      "step": 387510
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.057281701997398e-05,
+      "loss": 0.7829,
+      "step": 387520
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.057166909007423e-05,
+      "loss": 0.8487,
+      "step": 387530
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.057052116017448e-05,
+      "loss": 0.8343,
+      "step": 387540
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.0569373230274735e-05,
+      "loss": 0.8412,
+      "step": 387550
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.0568225300374985e-05,
+      "loss": 0.8281,
+      "step": 387560
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.056707737047524e-05,
+      "loss": 0.7703,
+      "step": 387570
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.056592944057549e-05,
+      "loss": 0.8877,
+      "step": 387580
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.056478151067575e-05,
+      "loss": 0.8735,
+      "step": 387590
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.0563633580776e-05,
+      "loss": 0.8707,
+      "step": 387600
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.0562485650876254e-05,
+      "loss": 0.8221,
+      "step": 387610
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.0561337720976504e-05,
+      "loss": 0.9715,
+      "step": 387620
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.0560189791076754e-05,
+      "loss": 0.8987,
+      "step": 387630
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.055904186117701e-05,
+      "loss": 0.9035,
+      "step": 387640
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.055789393127726e-05,
+      "loss": 0.8465,
+      "step": 387650
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.055674600137751e-05,
+      "loss": 0.913,
+      "step": 387660
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.055559807147777e-05,
+      "loss": 0.8093,
+      "step": 387670
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.055445014157802e-05,
+      "loss": 0.9101,
+      "step": 387680
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.0553302211678267e-05,
+      "loss": 0.9213,
+      "step": 387690
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.055215428177852e-05,
+      "loss": 0.8538,
+      "step": 387700
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.055100635187877e-05,
+      "loss": 0.8031,
+      "step": 387710
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.054985842197903e-05,
+      "loss": 0.8034,
+      "step": 387720
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.054871049207928e-05,
+      "loss": 0.9154,
+      "step": 387730
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.0547562562179536e-05,
+      "loss": 0.9019,
+      "step": 387740
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.0546414632279786e-05,
+      "loss": 0.8587,
+      "step": 387750
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.054526670238004e-05,
+      "loss": 0.8465,
+      "step": 387760
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.054411877248029e-05,
+      "loss": 0.8507,
+      "step": 387770
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.054297084258054e-05,
+      "loss": 0.8444,
+      "step": 387780
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.05418229126808e-05,
+      "loss": 0.8586,
+      "step": 387790
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.054067498278105e-05,
+      "loss": 0.9025,
+      "step": 387800
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.05395270528813e-05,
+      "loss": 0.8652,
+      "step": 387810
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.0538379122981555e-05,
+      "loss": 0.8779,
+      "step": 387820
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.0537231193081805e-05,
+      "loss": 0.9186,
+      "step": 387830
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.0536083263182055e-05,
+      "loss": 0.8975,
+      "step": 387840
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.053493533328231e-05,
+      "loss": 0.905,
+      "step": 387850
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.053378740338256e-05,
+      "loss": 0.8881,
+      "step": 387860
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.053263947348282e-05,
+      "loss": 0.8736,
+      "step": 387870
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.053149154358307e-05,
+      "loss": 0.9456,
+      "step": 387880
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.0530343613683325e-05,
+      "loss": 0.9068,
+      "step": 387890
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.0529195683783574e-05,
+      "loss": 0.9134,
+      "step": 387900
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.052804775388383e-05,
+      "loss": 0.7908,
+      "step": 387910
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.052689982398408e-05,
+      "loss": 0.9646,
+      "step": 387920
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.052575189408433e-05,
+      "loss": 0.8013,
+      "step": 387930
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.052460396418458e-05,
+      "loss": 0.8078,
+      "step": 387940
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.052345603428484e-05,
+      "loss": 0.8599,
+      "step": 387950
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.052230810438509e-05,
+      "loss": 0.8452,
+      "step": 387960
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.052116017448534e-05,
+      "loss": 0.8416,
+      "step": 387970
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.0520012244585594e-05,
+      "loss": 0.7907,
+      "step": 387980
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.0518864314685844e-05,
+      "loss": 0.9258,
+      "step": 387990
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.05177163847861e-05,
+      "loss": 0.9289,
+      "step": 388000
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.0516568454886353e-05,
+      "loss": 0.93,
+      "step": 388010
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.0515420524986603e-05,
+      "loss": 0.8831,
+      "step": 388020
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.0514272595086856e-05,
+      "loss": 0.887,
+      "step": 388030
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.051312466518711e-05,
+      "loss": 0.8489,
+      "step": 388040
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.0511976735287363e-05,
+      "loss": 0.8384,
+      "step": 388050
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.0510828805387613e-05,
+      "loss": 0.8487,
+      "step": 388060
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.050968087548787e-05,
+      "loss": 0.9029,
+      "step": 388070
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.050853294558812e-05,
+      "loss": 0.9053,
+      "step": 388080
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.0507385015688373e-05,
+      "loss": 1.0078,
+      "step": 388090
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.0506237085788626e-05,
+      "loss": 0.8434,
+      "step": 388100
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.050508915588888e-05,
+      "loss": 0.8906,
+      "step": 388110
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.050394122598913e-05,
+      "loss": 0.8377,
+      "step": 388120
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.0502793296089386e-05,
+      "loss": 0.8443,
+      "step": 388130
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.0501645366189635e-05,
+      "loss": 1.0019,
+      "step": 388140
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.0500497436289885e-05,
+      "loss": 0.8832,
+      "step": 388150
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.0499349506390142e-05,
+      "loss": 0.8736,
+      "step": 388160
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.0498201576490392e-05,
+      "loss": 0.8626,
+      "step": 388170
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.0497053646590645e-05,
+      "loss": 0.9331,
+      "step": 388180
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.0495905716690898e-05,
+      "loss": 0.8656,
+      "step": 388190
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.049475778679115e-05,
+      "loss": 0.895,
+      "step": 388200
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.04936098568914e-05,
+      "loss": 0.8346,
+      "step": 388210
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.0492461926991658e-05,
+      "loss": 0.8518,
+      "step": 388220
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.0491313997091908e-05,
+      "loss": 0.8442,
+      "step": 388230
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.049016606719216e-05,
+      "loss": 0.916,
+      "step": 388240
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.0489018137292414e-05,
+      "loss": 0.9402,
+      "step": 388250
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.0487870207392667e-05,
+      "loss": 0.8706,
+      "step": 388260
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.0486722277492917e-05,
+      "loss": 1.0206,
+      "step": 388270
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.0485574347593174e-05,
+      "loss": 0.8847,
+      "step": 388280
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.0484426417693424e-05,
+      "loss": 0.8578,
+      "step": 388290
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.0483278487793674e-05,
+      "loss": 0.861,
+      "step": 388300
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.048213055789393e-05,
+      "loss": 0.9166,
+      "step": 388310
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.048098262799418e-05,
+      "loss": 0.9293,
+      "step": 388320
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.0479834698094433e-05,
+      "loss": 0.9221,
+      "step": 388330
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.0478686768194687e-05,
+      "loss": 0.8487,
+      "step": 388340
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.047753883829494e-05,
+      "loss": 0.8808,
+      "step": 388350
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.047639090839519e-05,
+      "loss": 0.9412,
+      "step": 388360
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.0475242978495446e-05,
+      "loss": 0.8258,
+      "step": 388370
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.0474095048595696e-05,
+      "loss": 0.8422,
+      "step": 388380
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.047294711869595e-05,
+      "loss": 0.9603,
+      "step": 388390
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.0471799188796203e-05,
+      "loss": 0.8393,
+      "step": 388400
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.0470651258896456e-05,
+      "loss": 0.9485,
+      "step": 388410
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.0469503328996706e-05,
+      "loss": 0.8924,
+      "step": 388420
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.0468355399096962e-05,
+      "loss": 0.883,
+      "step": 388430
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.0467207469197212e-05,
+      "loss": 0.9031,
+      "step": 388440
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.0466059539297462e-05,
+      "loss": 0.8159,
+      "step": 388450
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.046491160939772e-05,
+      "loss": 0.8645,
+      "step": 388460
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.046376367949797e-05,
+      "loss": 0.9518,
+      "step": 388470
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.0462615749598222e-05,
+      "loss": 0.9502,
+      "step": 388480
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.0461467819698475e-05,
+      "loss": 0.8045,
+      "step": 388490
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3.046031988979873e-05,
+      "loss": 0.9014,
+      "step": 388500
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3.0459171959898978e-05,
+      "loss": 0.9023,
+      "step": 388510
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3.0458024029999235e-05,
+      "loss": 0.8881,
+      "step": 388520
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3.0456876100099485e-05,
+      "loss": 0.8671,
+      "step": 388530
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3.0455728170199738e-05,
+      "loss": 0.9678,
+      "step": 388540
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3.045458024029999e-05,
+      "loss": 0.8465,
+      "step": 388550
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3.0453432310400244e-05,
+      "loss": 0.9127,
+      "step": 388560
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3.0452284380500494e-05,
+      "loss": 0.8155,
+      "step": 388570
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3.045113645060075e-05,
+      "loss": 0.875,
+      "step": 388580
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3.0449988520701e-05,
+      "loss": 0.8589,
+      "step": 388590
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3.044884059080125e-05,
+      "loss": 0.8746,
+      "step": 388600
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3.0447692660901504e-05,
+      "loss": 0.8316,
+      "step": 388610
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3.0446544731001757e-05,
+      "loss": 0.9044,
+      "step": 388620
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3.044539680110201e-05,
+      "loss": 0.8742,
+      "step": 388630
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3.044424887120226e-05,
+      "loss": 0.8717,
+      "step": 388640
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3.0443100941302517e-05,
+      "loss": 0.9579,
+      "step": 388650
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3.0441953011402767e-05,
+      "loss": 0.8493,
+      "step": 388660
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3.044080508150302e-05,
+      "loss": 0.9145,
+      "step": 388670
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3.0439657151603273e-05,
+      "loss": 0.8571,
+      "step": 388680
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3.0438509221703526e-05,
+      "loss": 0.8716,
+      "step": 388690
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3.0437361291803776e-05,
+      "loss": 0.7925,
+      "step": 388700
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3.0436213361904033e-05,
+      "loss": 0.9021,
+      "step": 388710
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.0435065432004283e-05,
+      "loss": 0.8977,
+      "step": 388720
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.0433917502104533e-05,
+      "loss": 0.9262,
+      "step": 388730
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.043276957220479e-05,
+      "loss": 0.8765,
+      "step": 388740
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.043162164230504e-05,
+      "loss": 0.8477,
+      "step": 388750
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.0430473712405292e-05,
+      "loss": 0.8747,
+      "step": 388760
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.0429325782505546e-05,
+      "loss": 0.9237,
+      "step": 388770
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.04281778526058e-05,
+      "loss": 0.8667,
+      "step": 388780
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.042702992270605e-05,
+      "loss": 0.8593,
+      "step": 388790
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.0425881992806305e-05,
+      "loss": 0.9576,
+      "step": 388800
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.0424734062906555e-05,
+      "loss": 0.7779,
+      "step": 388810
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.042358613300681e-05,
+      "loss": 0.8895,
+      "step": 388820
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.042243820310706e-05,
+      "loss": 0.9572,
+      "step": 388830
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.0421290273207315e-05,
+      "loss": 0.8765,
+      "step": 388840
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.0420142343307565e-05,
+      "loss": 0.8821,
+      "step": 388850
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.041899441340782e-05,
+      "loss": 0.8645,
+      "step": 388860
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.041784648350807e-05,
+      "loss": 0.8629,
+      "step": 388870
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.041669855360832e-05,
+      "loss": 0.7106,
+      "step": 388880
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.0415550623708578e-05,
+      "loss": 0.8458,
+      "step": 388890
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.0414402693808828e-05,
+      "loss": 0.7947,
+      "step": 388900
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.041325476390908e-05,
+      "loss": 1.0506,
+      "step": 388910
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.0412106834009334e-05,
+      "loss": 0.879,
+      "step": 388920
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.0410958904109587e-05,
+      "loss": 0.8373,
+      "step": 388930
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.0409810974209837e-05,
+      "loss": 0.8261,
+      "step": 388940
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.0408663044310094e-05,
+      "loss": 0.8199,
+      "step": 388950
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.0407515114410344e-05,
+      "loss": 0.8957,
+      "step": 388960
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.0406367184510597e-05,
+      "loss": 0.8072,
+      "step": 388970
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.040521925461085e-05,
+      "loss": 0.8332,
+      "step": 388980
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.0404071324711103e-05,
+      "loss": 0.9728,
+      "step": 388990
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.0402923394811353e-05,
+      "loss": 0.84,
+      "step": 389000
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.040177546491161e-05,
+      "loss": 0.8766,
+      "step": 389010
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.040062753501186e-05,
+      "loss": 0.901,
+      "step": 389020
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.039947960511211e-05,
+      "loss": 0.9573,
+      "step": 389030
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.0398331675212366e-05,
+      "loss": 0.8443,
+      "step": 389040
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.0397183745312616e-05,
+      "loss": 0.8133,
+      "step": 389050
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.039603581541287e-05,
+      "loss": 0.885,
+      "step": 389060
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.0394887885513123e-05,
+      "loss": 0.9024,
+      "step": 389070
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.0393739955613376e-05,
+      "loss": 0.855,
+      "step": 389080
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.0392592025713626e-05,
+      "loss": 0.8433,
+      "step": 389090
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.0391444095813882e-05,
+      "loss": 0.8804,
+      "step": 389100
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.0390296165914132e-05,
+      "loss": 0.8375,
+      "step": 389110
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.0389148236014385e-05,
+      "loss": 0.8599,
+      "step": 389120
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.038800030611464e-05,
+      "loss": 0.9483,
+      "step": 389130
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.0386852376214892e-05,
+      "loss": 0.8277,
+      "step": 389140
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.0385704446315142e-05,
+      "loss": 0.9031,
+      "step": 389150
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 3.03845565164154e-05,
+      "loss": 0.9303,
+      "step": 389160
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 3.0383408586515648e-05,
+      "loss": 0.8402,
+      "step": 389170
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 3.0382260656615898e-05,
+      "loss": 0.8793,
+      "step": 389180
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 3.0381112726716155e-05,
+      "loss": 0.8086,
+      "step": 389190
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 3.0379964796816405e-05,
+      "loss": 0.898,
+      "step": 389200
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 3.0378816866916658e-05,
+      "loss": 0.9132,
+      "step": 389210
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 3.037766893701691e-05,
+      "loss": 0.8284,
+      "step": 389220
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 3.0376521007117164e-05,
+      "loss": 0.9515,
+      "step": 389230
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 3.0375373077217414e-05,
+      "loss": 0.8581,
+      "step": 389240
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 3.037422514731767e-05,
+      "loss": 0.8412,
+      "step": 389250
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 3.037307721741792e-05,
+      "loss": 0.9624,
+      "step": 389260
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 3.0371929287518174e-05,
+      "loss": 0.9099,
+      "step": 389270
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 3.0370781357618427e-05,
+      "loss": 0.934,
+      "step": 389280
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 3.036963342771868e-05,
+      "loss": 0.8112,
+      "step": 389290
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 3.036848549781893e-05,
+      "loss": 0.8621,
+      "step": 389300
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 3.036733756791918e-05,
+      "loss": 0.7794,
+      "step": 389310
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 3.0366189638019437e-05,
+      "loss": 0.9204,
+      "step": 389320
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 3.0365041708119687e-05,
+      "loss": 0.8631,
+      "step": 389330
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 3.036389377821994e-05,
+      "loss": 0.8007,
+      "step": 389340
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 3.0362745848320193e-05,
+      "loss": 0.8476,
+      "step": 389350
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 3.0361597918420446e-05,
+      "loss": 0.7668,
+      "step": 389360
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 3.0360449988520696e-05,
+      "loss": 0.9509,
+      "step": 389370
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 3.0359302058620953e-05,
+      "loss": 0.9362,
+      "step": 389380
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 3.0358154128721203e-05,
+      "loss": 0.841,
+      "step": 389390
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 3.0357006198821456e-05,
+      "loss": 0.7703,
+      "step": 389400
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 3.035585826892171e-05,
+      "loss": 0.8226,
+      "step": 389410
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 3.0354710339021962e-05,
+      "loss": 0.8358,
+      "step": 389420
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 3.0353562409122212e-05,
+      "loss": 0.9104,
+      "step": 389430
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 3.035241447922247e-05,
+      "loss": 0.8616,
+      "step": 389440
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 3.035126654932272e-05,
+      "loss": 0.8745,
+      "step": 389450
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 3.035011861942297e-05,
+      "loss": 0.8994,
+      "step": 389460
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 3.0348970689523225e-05,
+      "loss": 0.9269,
+      "step": 389470
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 3.0347822759623475e-05,
+      "loss": 0.871,
+      "step": 389480
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 3.0346674829723728e-05,
+      "loss": 0.9062,
+      "step": 389490
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 3.034552689982398e-05,
+      "loss": 0.888,
+      "step": 389500
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 3.0344378969924235e-05,
+      "loss": 0.8487,
+      "step": 389510
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 3.0343231040024485e-05,
+      "loss": 0.8405,
+      "step": 389520
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 3.034208311012474e-05,
+      "loss": 0.9617,
+      "step": 389530
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 3.034093518022499e-05,
+      "loss": 0.8132,
+      "step": 389540
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 3.0339787250325244e-05,
+      "loss": 0.9293,
+      "step": 389550
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 3.0338639320425498e-05,
+      "loss": 0.8227,
+      "step": 389560
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 3.033749139052575e-05,
+      "loss": 0.9269,
+      "step": 389570
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 3.0336343460626e-05,
+      "loss": 0.8956,
+      "step": 389580
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3.0335195530726257e-05,
+      "loss": 0.8316,
+      "step": 389590
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3.0334047600826507e-05,
+      "loss": 0.8566,
+      "step": 389600
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3.0332899670926757e-05,
+      "loss": 0.7665,
+      "step": 389610
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3.0331751741027014e-05,
+      "loss": 0.8261,
+      "step": 389620
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3.0330603811127263e-05,
+      "loss": 0.961,
+      "step": 389630
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3.0329455881227517e-05,
+      "loss": 0.8745,
+      "step": 389640
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3.032830795132777e-05,
+      "loss": 0.959,
+      "step": 389650
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3.0327160021428023e-05,
+      "loss": 0.8955,
+      "step": 389660
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3.0326012091528273e-05,
+      "loss": 0.9042,
+      "step": 389670
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3.032486416162853e-05,
+      "loss": 0.7669,
+      "step": 389680
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3.032371623172878e-05,
+      "loss": 0.8451,
+      "step": 389690
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3.0322568301829033e-05,
+      "loss": 0.843,
+      "step": 389700
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3.0321420371929286e-05,
+      "loss": 0.8599,
+      "step": 389710
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3.032027244202954e-05,
+      "loss": 0.8862,
+      "step": 389720
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3.031912451212979e-05,
+      "loss": 0.8885,
+      "step": 389730
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3.0317976582230046e-05,
+      "loss": 0.8862,
+      "step": 389740
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3.0316828652330296e-05,
+      "loss": 0.9179,
+      "step": 389750
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3.0315680722430545e-05,
+      "loss": 0.8109,
+      "step": 389760
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3.0314532792530802e-05,
+      "loss": 0.8913,
+      "step": 389770
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3.0313384862631052e-05,
+      "loss": 0.8148,
+      "step": 389780
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3.0312236932731305e-05,
+      "loss": 0.8132,
+      "step": 389790
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 3.031108900283156e-05,
+      "loss": 0.9043,
+      "step": 389800
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3.030994107293181e-05,
+      "loss": 0.9069,
+      "step": 389810
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3.030879314303206e-05,
+      "loss": 0.8869,
+      "step": 389820
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3.0307645213132318e-05,
+      "loss": 0.9352,
+      "step": 389830
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3.0306497283232568e-05,
+      "loss": 0.8841,
+      "step": 389840
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3.030534935333282e-05,
+      "loss": 0.8204,
+      "step": 389850
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3.0304201423433074e-05,
+      "loss": 0.8402,
+      "step": 389860
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3.0303053493533328e-05,
+      "loss": 0.8806,
+      "step": 389870
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3.0301905563633578e-05,
+      "loss": 0.7971,
+      "step": 389880
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3.0300757633733834e-05,
+      "loss": 0.8382,
+      "step": 389890
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3.0299609703834084e-05,
+      "loss": 0.8967,
+      "step": 389900
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3.0298461773934334e-05,
+      "loss": 0.8237,
+      "step": 389910
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3.029731384403459e-05,
+      "loss": 0.9771,
+      "step": 389920
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3.029616591413484e-05,
+      "loss": 0.863,
+      "step": 389930
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3.0295017984235094e-05,
+      "loss": 0.8539,
+      "step": 389940
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3.0293870054335347e-05,
+      "loss": 0.9143,
+      "step": 389950
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3.02927221244356e-05,
+      "loss": 0.8084,
+      "step": 389960
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3.029157419453585e-05,
+      "loss": 0.9259,
+      "step": 389970
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3.02904262646361e-05,
+      "loss": 0.9,
+      "step": 389980
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3.0289278334736356e-05,
+      "loss": 0.8211,
+      "step": 389990
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3.028813040483661e-05,
+      "loss": 0.9456,
+      "step": 390000
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3.028698247493686e-05,
+      "loss": 0.8562,
+      "step": 390010
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 3.0285834545037116e-05,
+      "loss": 0.9155,
+      "step": 390020
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3.0284686615137366e-05,
+      "loss": 0.8662,
+      "step": 390030
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3.0283538685237616e-05,
+      "loss": 0.8521,
+      "step": 390040
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3.0282390755337873e-05,
+      "loss": 0.8918,
+      "step": 390050
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3.0281242825438122e-05,
+      "loss": 0.7683,
+      "step": 390060
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3.0280094895538376e-05,
+      "loss": 0.9031,
+      "step": 390070
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3.027894696563863e-05,
+      "loss": 0.8538,
+      "step": 390080
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3.0277799035738882e-05,
+      "loss": 0.9535,
+      "step": 390090
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3.0276651105839132e-05,
+      "loss": 0.8885,
+      "step": 390100
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3.027550317593939e-05,
+      "loss": 1.0006,
+      "step": 390110
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3.027435524603964e-05,
+      "loss": 0.9075,
+      "step": 390120
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3.0273207316139888e-05,
+      "loss": 0.8339,
+      "step": 390130
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3.0272059386240145e-05,
+      "loss": 0.868,
+      "step": 390140
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3.0270911456340398e-05,
+      "loss": 0.8496,
+      "step": 390150
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3.0269763526440648e-05,
+      "loss": 0.8066,
+      "step": 390160
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3.0268615596540905e-05,
+      "loss": 0.8549,
+      "step": 390170
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3.0267467666641155e-05,
+      "loss": 0.904,
+      "step": 390180
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3.0266319736741404e-05,
+      "loss": 0.8775,
+      "step": 390190
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3.026517180684166e-05,
+      "loss": 0.8982,
+      "step": 390200
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3.026402387694191e-05,
+      "loss": 0.9065,
+      "step": 390210
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3.0262875947042164e-05,
+      "loss": 0.8767,
+      "step": 390220
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3.0261728017142417e-05,
+      "loss": 0.8076,
+      "step": 390230
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 3.026058008724267e-05,
+      "loss": 0.9142,
+      "step": 390240
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 3.025943215734292e-05,
+      "loss": 0.8526,
+      "step": 390250
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 3.0258284227443177e-05,
+      "loss": 0.9547,
+      "step": 390260
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 3.0257136297543427e-05,
+      "loss": 0.8869,
+      "step": 390270
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 3.0255988367643677e-05,
+      "loss": 0.928,
+      "step": 390280
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 3.0254840437743933e-05,
+      "loss": 0.9468,
+      "step": 390290
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 3.0253692507844187e-05,
+      "loss": 0.8824,
+      "step": 390300
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 3.0252544577944437e-05,
+      "loss": 0.8375,
+      "step": 390310
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 3.0251396648044693e-05,
+      "loss": 0.9102,
+      "step": 390320
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 3.0250248718144943e-05,
+      "loss": 0.8123,
+      "step": 390330
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 3.0249100788245193e-05,
+      "loss": 0.9147,
+      "step": 390340
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 3.024795285834545e-05,
+      "loss": 0.8272,
+      "step": 390350
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 3.02468049284457e-05,
+      "loss": 0.8295,
+      "step": 390360
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 3.0245656998545953e-05,
+      "loss": 0.9625,
+      "step": 390370
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 3.0244509068646206e-05,
+      "loss": 0.9476,
+      "step": 390380
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 3.024336113874646e-05,
+      "loss": 0.9653,
+      "step": 390390
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 3.024221320884671e-05,
+      "loss": 0.9288,
+      "step": 390400
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 3.0241065278946966e-05,
+      "loss": 0.9629,
+      "step": 390410
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 3.0239917349047215e-05,
+      "loss": 0.8026,
+      "step": 390420
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 3.0238769419147465e-05,
+      "loss": 0.8705,
+      "step": 390430
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 3.0237621489247722e-05,
+      "loss": 0.9111,
+      "step": 390440
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 3.0236473559347975e-05,
+      "loss": 0.9272,
+      "step": 390450
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 3.0235325629448225e-05,
+      "loss": 0.8939,
+      "step": 390460
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 3.023417769954848e-05,
+      "loss": 0.9027,
+      "step": 390470
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 3.023302976964873e-05,
+      "loss": 0.8975,
+      "step": 390480
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 3.023188183974898e-05,
+      "loss": 0.8368,
+      "step": 390490
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 3.0230733909849238e-05,
+      "loss": 0.8526,
+      "step": 390500
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 3.0229585979949488e-05,
+      "loss": 0.9407,
+      "step": 390510
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 3.022843805004974e-05,
+      "loss": 0.9577,
+      "step": 390520
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 3.0227290120149994e-05,
+      "loss": 0.8785,
+      "step": 390530
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 3.0226142190250248e-05,
+      "loss": 0.9419,
+      "step": 390540
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 3.0224994260350497e-05,
+      "loss": 0.8568,
+      "step": 390550
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 3.0223846330450754e-05,
+      "loss": 0.8613,
+      "step": 390560
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 3.0222698400551004e-05,
+      "loss": 0.864,
+      "step": 390570
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 3.0221550470651254e-05,
+      "loss": 0.8885,
+      "step": 390580
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 3.022040254075151e-05,
+      "loss": 0.9062,
+      "step": 390590
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 3.0219254610851764e-05,
+      "loss": 0.8215,
+      "step": 390600
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 3.0218106680952013e-05,
+      "loss": 0.8579,
+      "step": 390610
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 3.021695875105227e-05,
+      "loss": 0.8884,
+      "step": 390620
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 3.021581082115252e-05,
+      "loss": 0.8069,
+      "step": 390630
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 3.021466289125277e-05,
+      "loss": 0.8628,
+      "step": 390640
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 3.0213514961353026e-05,
+      "loss": 0.8386,
+      "step": 390650
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 3.0212367031453276e-05,
+      "loss": 0.8962,
+      "step": 390660
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 3.021121910155353e-05,
+      "loss": 0.8854,
+      "step": 390670
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3.021007117165378e-05,
+      "loss": 0.9271,
+      "step": 390680
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3.0208923241754036e-05,
+      "loss": 0.8214,
+      "step": 390690
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3.0207775311854286e-05,
+      "loss": 0.913,
+      "step": 390700
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3.0206627381954536e-05,
+      "loss": 0.8226,
+      "step": 390710
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3.0205479452054792e-05,
+      "loss": 0.9522,
+      "step": 390720
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3.0204331522155042e-05,
+      "loss": 0.8063,
+      "step": 390730
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3.0203183592255295e-05,
+      "loss": 0.8922,
+      "step": 390740
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3.0202035662355552e-05,
+      "loss": 0.9726,
+      "step": 390750
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3.0200887732455802e-05,
+      "loss": 0.8676,
+      "step": 390760
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3.0199739802556052e-05,
+      "loss": 0.8269,
+      "step": 390770
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3.019859187265631e-05,
+      "loss": 0.8459,
+      "step": 390780
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3.0197443942756558e-05,
+      "loss": 0.943,
+      "step": 390790
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3.019629601285681e-05,
+      "loss": 0.9141,
+      "step": 390800
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3.0195148082957065e-05,
+      "loss": 0.9088,
+      "step": 390810
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3.0194000153057318e-05,
+      "loss": 0.9861,
+      "step": 390820
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3.0192852223157568e-05,
+      "loss": 0.927,
+      "step": 390830
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3.0191704293257824e-05,
+      "loss": 0.8344,
+      "step": 390840
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3.0190556363358074e-05,
+      "loss": 0.8778,
+      "step": 390850
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3.0189408433458324e-05,
+      "loss": 0.82,
+      "step": 390860
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3.018826050355858e-05,
+      "loss": 0.9184,
+      "step": 390870
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3.018711257365883e-05,
+      "loss": 0.8512,
+      "step": 390880
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 3.0185964643759084e-05,
+      "loss": 0.9582,
+      "step": 390890
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 3.018481671385934e-05,
+      "loss": 0.8561,
+      "step": 390900
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 3.018366878395959e-05,
+      "loss": 0.8356,
+      "step": 390910
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 3.018252085405984e-05,
+      "loss": 0.9626,
+      "step": 390920
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 3.0181372924160097e-05,
+      "loss": 0.9508,
+      "step": 390930
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 3.0180224994260347e-05,
+      "loss": 0.8898,
+      "step": 390940
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 3.01790770643606e-05,
+      "loss": 0.8383,
+      "step": 390950
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 3.0177929134460853e-05,
+      "loss": 0.9043,
+      "step": 390960
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 3.0176781204561106e-05,
+      "loss": 0.9484,
+      "step": 390970
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 3.0175633274661356e-05,
+      "loss": 0.9448,
+      "step": 390980
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 3.0174485344761613e-05,
+      "loss": 0.8786,
+      "step": 390990
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 3.0173337414861863e-05,
+      "loss": 0.8184,
+      "step": 391000
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 3.0172189484962113e-05,
+      "loss": 0.9832,
+      "step": 391010
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 3.017104155506237e-05,
+      "loss": 0.8638,
+      "step": 391020
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 3.016989362516262e-05,
+      "loss": 0.7945,
+      "step": 391030
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 3.0168745695262872e-05,
+      "loss": 0.9261,
+      "step": 391040
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 3.016759776536313e-05,
+      "loss": 0.9259,
+      "step": 391050
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 3.016644983546338e-05,
+      "loss": 0.8369,
+      "step": 391060
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 3.016530190556363e-05,
+      "loss": 0.9031,
+      "step": 391070
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 3.0164153975663885e-05,
+      "loss": 0.9613,
+      "step": 391080
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 3.0163006045764135e-05,
+      "loss": 0.9046,
+      "step": 391090
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 3.016185811586439e-05,
+      "loss": 0.7943,
+      "step": 391100
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 3.016071018596464e-05,
+      "loss": 0.9439,
+      "step": 391110
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3.0159562256064895e-05,
+      "loss": 0.917,
+      "step": 391120
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3.0158414326165145e-05,
+      "loss": 0.8723,
+      "step": 391130
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3.01572663962654e-05,
+      "loss": 0.9141,
+      "step": 391140
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3.015611846636565e-05,
+      "loss": 0.7747,
+      "step": 391150
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3.01549705364659e-05,
+      "loss": 0.8521,
+      "step": 391160
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3.0153822606566158e-05,
+      "loss": 0.9733,
+      "step": 391170
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3.0152674676666408e-05,
+      "loss": 0.8228,
+      "step": 391180
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3.015152674676666e-05,
+      "loss": 0.8687,
+      "step": 391190
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3.0150378816866917e-05,
+      "loss": 0.8493,
+      "step": 391200
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3.0149230886967167e-05,
+      "loss": 0.9119,
+      "step": 391210
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3.0148082957067417e-05,
+      "loss": 0.8821,
+      "step": 391220
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3.0146935027167674e-05,
+      "loss": 0.8328,
+      "step": 391230
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3.0145787097267924e-05,
+      "loss": 0.914,
+      "step": 391240
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3.0144639167368177e-05,
+      "loss": 0.857,
+      "step": 391250
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3.014349123746843e-05,
+      "loss": 0.8361,
+      "step": 391260
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3.0142343307568683e-05,
+      "loss": 0.9463,
+      "step": 391270
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3.0141195377668933e-05,
+      "loss": 0.9538,
+      "step": 391280
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3.014004744776919e-05,
+      "loss": 0.9333,
+      "step": 391290
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3.013889951786944e-05,
+      "loss": 0.885,
+      "step": 391300
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3.013775158796969e-05,
+      "loss": 0.8526,
+      "step": 391310
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3.0136603658069946e-05,
+      "loss": 0.8523,
+      "step": 391320
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 3.0135455728170196e-05,
+      "loss": 0.925,
+      "step": 391330
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 3.013430779827045e-05,
+      "loss": 0.9434,
+      "step": 391340
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 3.01331598683707e-05,
+      "loss": 0.8933,
+      "step": 391350
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 3.0132011938470956e-05,
+      "loss": 0.8862,
+      "step": 391360
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 3.0130864008571206e-05,
+      "loss": 0.8722,
+      "step": 391370
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 3.012971607867146e-05,
+      "loss": 0.8342,
+      "step": 391380
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 3.0128568148771712e-05,
+      "loss": 0.9308,
+      "step": 391390
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 3.0127420218871965e-05,
+      "loss": 0.8872,
+      "step": 391400
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 3.0126272288972215e-05,
+      "loss": 0.961,
+      "step": 391410
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 3.0125124359072472e-05,
+      "loss": 0.9543,
+      "step": 391420
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 3.0123976429172722e-05,
+      "loss": 0.8561,
+      "step": 391430
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 3.012282849927297e-05,
+      "loss": 0.926,
+      "step": 391440
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 3.0121680569373228e-05,
+      "loss": 0.9593,
+      "step": 391450
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 3.0120532639473478e-05,
+      "loss": 0.884,
+      "step": 391460
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 3.011938470957373e-05,
+      "loss": 0.8573,
+      "step": 391470
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 3.0118236779673985e-05,
+      "loss": 0.8797,
+      "step": 391480
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 3.0117088849774238e-05,
+      "loss": 0.7866,
+      "step": 391490
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 3.0115940919874488e-05,
+      "loss": 0.7526,
+      "step": 391500
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 3.0114792989974744e-05,
+      "loss": 0.8283,
+      "step": 391510
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 3.0113645060074994e-05,
+      "loss": 0.9229,
+      "step": 391520
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 3.0112497130175247e-05,
+      "loss": 0.908,
+      "step": 391530
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 3.01113492002755e-05,
+      "loss": 0.8859,
+      "step": 391540
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 3.0110201270375754e-05,
+      "loss": 0.8638,
+      "step": 391550
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 3.0109053340476004e-05,
+      "loss": 0.863,
+      "step": 391560
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 3.010790541057626e-05,
+      "loss": 0.873,
+      "step": 391570
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 3.010675748067651e-05,
+      "loss": 0.9592,
+      "step": 391580
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 3.010560955077676e-05,
+      "loss": 0.905,
+      "step": 391590
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 3.0104461620877017e-05,
+      "loss": 0.8732,
+      "step": 391600
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 3.0103313690977267e-05,
+      "loss": 0.863,
+      "step": 391610
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 3.010216576107752e-05,
+      "loss": 0.8151,
+      "step": 391620
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 3.0101017831177773e-05,
+      "loss": 0.9695,
+      "step": 391630
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 3.0099869901278026e-05,
+      "loss": 0.8429,
+      "step": 391640
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 3.0098721971378276e-05,
+      "loss": 0.8393,
+      "step": 391650
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 3.0097574041478533e-05,
+      "loss": 0.8841,
+      "step": 391660
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 3.0096426111578783e-05,
+      "loss": 0.8565,
+      "step": 391670
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 3.0095278181679036e-05,
+      "loss": 0.7624,
+      "step": 391680
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 3.009413025177929e-05,
+      "loss": 0.8721,
+      "step": 391690
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 3.0092982321879542e-05,
+      "loss": 0.8285,
+      "step": 391700
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 3.0091834391979792e-05,
+      "loss": 0.8759,
+      "step": 391710
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 3.009068646208005e-05,
+      "loss": 0.805,
+      "step": 391720
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 3.00895385321803e-05,
+      "loss": 0.8496,
+      "step": 391730
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 3.008839060228055e-05,
+      "loss": 0.776,
+      "step": 391740
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 3.0087242672380805e-05,
+      "loss": 0.8605,
+      "step": 391750
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 3.0086094742481055e-05,
+      "loss": 0.964,
+      "step": 391760
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3.0084946812581308e-05,
+      "loss": 0.8891,
+      "step": 391770
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3.008379888268156e-05,
+      "loss": 0.9614,
+      "step": 391780
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3.0082650952781815e-05,
+      "loss": 0.9377,
+      "step": 391790
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3.0081503022882065e-05,
+      "loss": 0.9234,
+      "step": 391800
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3.008035509298232e-05,
+      "loss": 0.9619,
+      "step": 391810
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3.007920716308257e-05,
+      "loss": 0.9962,
+      "step": 391820
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3.0078059233182824e-05,
+      "loss": 0.8189,
+      "step": 391830
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3.0076911303283078e-05,
+      "loss": 0.8517,
+      "step": 391840
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3.007576337338333e-05,
+      "loss": 0.924,
+      "step": 391850
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3.007461544348358e-05,
+      "loss": 0.8645,
+      "step": 391860
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3.0073467513583837e-05,
+      "loss": 0.8929,
+      "step": 391870
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3.0072319583684087e-05,
+      "loss": 0.9682,
+      "step": 391880
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3.0071171653784337e-05,
+      "loss": 0.948,
+      "step": 391890
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3.0070023723884594e-05,
+      "loss": 0.8806,
+      "step": 391900
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3.0068875793984843e-05,
+      "loss": 0.9309,
+      "step": 391910
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3.0067727864085097e-05,
+      "loss": 0.8094,
+      "step": 391920
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3.006657993418535e-05,
+      "loss": 0.8815,
+      "step": 391930
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3.0065432004285603e-05,
+      "loss": 0.8914,
+      "step": 391940
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3.0064284074385853e-05,
+      "loss": 0.8197,
+      "step": 391950
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3.006313614448611e-05,
+      "loss": 0.8735,
+      "step": 391960
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3.006198821458636e-05,
+      "loss": 0.8397,
+      "step": 391970
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 3.0060840284686613e-05,
+      "loss": 0.8145,
+      "step": 391980
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 3.0059692354786866e-05,
+      "loss": 0.8626,
+      "step": 391990
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 3.005854442488712e-05,
+      "loss": 0.8794,
+      "step": 392000
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 3.005739649498737e-05,
+      "loss": 0.9481,
+      "step": 392010
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 3.0056248565087626e-05,
+      "loss": 0.8798,
+      "step": 392020
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 3.0055100635187876e-05,
+      "loss": 0.8589,
+      "step": 392030
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 3.0053952705288125e-05,
+      "loss": 0.8392,
+      "step": 392040
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 3.005280477538838e-05,
+      "loss": 0.8795,
+      "step": 392050
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 3.0051656845488632e-05,
+      "loss": 0.9828,
+      "step": 392060
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 3.0050508915588885e-05,
+      "loss": 0.8569,
+      "step": 392070
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 3.0049360985689135e-05,
+      "loss": 0.8193,
+      "step": 392080
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 3.004821305578939e-05,
+      "loss": 0.9586,
+      "step": 392090
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 3.004706512588964e-05,
+      "loss": 0.8842,
+      "step": 392100
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 3.0045917195989895e-05,
+      "loss": 0.8596,
+      "step": 392110
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 3.0044769266090148e-05,
+      "loss": 0.9036,
+      "step": 392120
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 3.00436213361904e-05,
+      "loss": 0.892,
+      "step": 392130
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 3.004247340629065e-05,
+      "loss": 0.8608,
+      "step": 392140
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 3.0041325476390908e-05,
+      "loss": 0.8416,
+      "step": 392150
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 3.0040177546491158e-05,
+      "loss": 0.8886,
+      "step": 392160
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 3.0039029616591407e-05,
+      "loss": 0.8844,
+      "step": 392170
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 3.0037881686691664e-05,
+      "loss": 0.9134,
+      "step": 392180
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 3.0036733756791914e-05,
+      "loss": 0.9542,
+      "step": 392190
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 3.0035585826892167e-05,
+      "loss": 0.8579,
+      "step": 392200
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.003443789699242e-05,
+      "loss": 0.9265,
+      "step": 392210
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.0033289967092674e-05,
+      "loss": 0.9256,
+      "step": 392220
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.0032142037192924e-05,
+      "loss": 0.8685,
+      "step": 392230
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.003099410729318e-05,
+      "loss": 0.9501,
+      "step": 392240
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.002984617739343e-05,
+      "loss": 0.9408,
+      "step": 392250
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.0028698247493683e-05,
+      "loss": 0.8465,
+      "step": 392260
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.0027550317593936e-05,
+      "loss": 0.8769,
+      "step": 392270
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.002640238769419e-05,
+      "loss": 0.8914,
+      "step": 392280
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.002525445779444e-05,
+      "loss": 0.8717,
+      "step": 392290
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.0024106527894696e-05,
+      "loss": 0.7968,
+      "step": 392300
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.0022958597994946e-05,
+      "loss": 0.8935,
+      "step": 392310
+    },
+    {
+      "epoch": 18.0,
+      "eval_cer": 0.6683384719669845,
+      "eval_em": 0.007357212003872217,
+      "eval_f1": 0.007357212003872217,
+      "eval_loss": 1.2729239463806152,
+      "eval_runtime": 1596.7272,
+      "eval_samples_per_second": 6.469,
+      "eval_steps_per_second": 0.809,
+      "eval_wer": 0.9926427879961278,
+      "step": 392313
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.0021810668095196e-05,
+      "loss": 0.8283,
+      "step": 392320
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.0020662738195453e-05,
+      "loss": 0.8651,
+      "step": 392330
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.0019514808295702e-05,
+      "loss": 0.7574,
+      "step": 392340
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.0018366878395956e-05,
+      "loss": 0.8848,
+      "step": 392350
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.001721894849621e-05,
+      "loss": 0.8912,
+      "step": 392360
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.0016071018596462e-05,
+      "loss": 0.8737,
+      "step": 392370
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.0014923088696712e-05,
+      "loss": 0.7684,
+      "step": 392380
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.001377515879697e-05,
+      "loss": 0.7721,
+      "step": 392390
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.001262722889722e-05,
+      "loss": 0.874,
+      "step": 392400
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.0011479298997472e-05,
+      "loss": 0.8504,
+      "step": 392410
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 3.0010331369097725e-05,
+      "loss": 0.8065,
+      "step": 392420
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.0009183439197978e-05,
+      "loss": 0.8485,
+      "step": 392430
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.0008035509298228e-05,
+      "loss": 0.8056,
+      "step": 392440
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.0006887579398485e-05,
+      "loss": 0.8288,
+      "step": 392450
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.0005739649498735e-05,
+      "loss": 0.9132,
+      "step": 392460
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.0004591719598984e-05,
+      "loss": 0.7862,
+      "step": 392470
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.000344378969924e-05,
+      "loss": 0.7715,
+      "step": 392480
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.000229585979949e-05,
+      "loss": 0.8149,
+      "step": 392490
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.0001147929899744e-05,
+      "loss": 0.7874,
+      "step": 392500
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 0.8273,
+      "step": 392510
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 2.999885207010025e-05,
+      "loss": 0.8976,
+      "step": 392520
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 2.99977041402005e-05,
+      "loss": 0.865,
+      "step": 392530
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 2.9996556210300757e-05,
+      "loss": 0.7378,
+      "step": 392540
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 2.9995408280401007e-05,
+      "loss": 0.7494,
+      "step": 392550
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 2.999426035050126e-05,
+      "loss": 0.8176,
+      "step": 392560
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 2.9993112420601513e-05,
+      "loss": 0.866,
+      "step": 392570
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 2.9991964490701767e-05,
+      "loss": 0.8415,
+      "step": 392580
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 2.9990816560802017e-05,
+      "loss": 0.8448,
+      "step": 392590
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 2.9989668630902273e-05,
+      "loss": 0.8624,
+      "step": 392600
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 2.9988520701002523e-05,
+      "loss": 0.8903,
+      "step": 392610
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 2.9987372771102773e-05,
+      "loss": 0.8605,
+      "step": 392620
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 2.998622484120303e-05,
+      "loss": 0.7438,
+      "step": 392630
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 2.998507691130328e-05,
+      "loss": 0.7585,
+      "step": 392640
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9983928981403533e-05,
+      "loss": 0.785,
+      "step": 392650
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9982781051503786e-05,
+      "loss": 0.8862,
+      "step": 392660
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.998163312160404e-05,
+      "loss": 0.871,
+      "step": 392670
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.998048519170429e-05,
+      "loss": 0.8484,
+      "step": 392680
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9979337261804546e-05,
+      "loss": 0.819,
+      "step": 392690
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9978189331904795e-05,
+      "loss": 0.7885,
+      "step": 392700
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.997704140200505e-05,
+      "loss": 0.8127,
+      "step": 392710
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.99758934721053e-05,
+      "loss": 0.7924,
+      "step": 392720
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9974745542205555e-05,
+      "loss": 0.759,
+      "step": 392730
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9973597612305805e-05,
+      "loss": 0.8039,
+      "step": 392740
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9972449682406055e-05,
+      "loss": 0.8099,
+      "step": 392750
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.997130175250631e-05,
+      "loss": 0.8288,
+      "step": 392760
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.997015382260656e-05,
+      "loss": 0.7722,
+      "step": 392770
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9969005892706815e-05,
+      "loss": 0.8231,
+      "step": 392780
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9967857962807068e-05,
+      "loss": 0.8209,
+      "step": 392790
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.996671003290732e-05,
+      "loss": 0.8943,
+      "step": 392800
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.996556210300757e-05,
+      "loss": 0.8736,
+      "step": 392810
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9964414173107828e-05,
+      "loss": 0.8251,
+      "step": 392820
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9963266243208077e-05,
+      "loss": 0.8031,
+      "step": 392830
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.996211831330833e-05,
+      "loss": 0.8813,
+      "step": 392840
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.9960970383408584e-05,
+      "loss": 0.9124,
+      "step": 392850
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.9959822453508837e-05,
+      "loss": 0.8937,
+      "step": 392860
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.9958674523609087e-05,
+      "loss": 0.7773,
+      "step": 392870
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.9957526593709344e-05,
+      "loss": 0.8189,
+      "step": 392880
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.9956378663809593e-05,
+      "loss": 0.7821,
+      "step": 392890
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.9955230733909843e-05,
+      "loss": 0.8981,
+      "step": 392900
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.99540828040101e-05,
+      "loss": 0.891,
+      "step": 392910
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.995293487411035e-05,
+      "loss": 0.7899,
+      "step": 392920
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.9951786944210603e-05,
+      "loss": 0.8998,
+      "step": 392930
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.9950639014310856e-05,
+      "loss": 0.7828,
+      "step": 392940
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.994949108441111e-05,
+      "loss": 0.8518,
+      "step": 392950
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.994834315451136e-05,
+      "loss": 0.8591,
+      "step": 392960
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.9947195224611616e-05,
+      "loss": 0.8094,
+      "step": 392970
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.9946047294711866e-05,
+      "loss": 0.8001,
+      "step": 392980
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.994489936481212e-05,
+      "loss": 0.8332,
+      "step": 392990
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.9943751434912372e-05,
+      "loss": 0.843,
+      "step": 393000
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.9942603505012626e-05,
+      "loss": 0.8466,
+      "step": 393010
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.9941455575112875e-05,
+      "loss": 0.7827,
+      "step": 393020
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.9940307645213132e-05,
+      "loss": 0.9052,
+      "step": 393030
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.9939159715313382e-05,
+      "loss": 0.9533,
+      "step": 393040
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.9938011785413632e-05,
+      "loss": 0.8771,
+      "step": 393050
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.993686385551389e-05,
+      "loss": 0.7783,
+      "step": 393060
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.9935715925614138e-05,
+      "loss": 0.8714,
+      "step": 393070
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.993456799571439e-05,
+      "loss": 0.8322,
+      "step": 393080
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.9933420065814645e-05,
+      "loss": 0.8923,
+      "step": 393090
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.9932272135914898e-05,
+      "loss": 0.756,
+      "step": 393100
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.9931124206015148e-05,
+      "loss": 0.8585,
+      "step": 393110
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.9929976276115405e-05,
+      "loss": 0.8657,
+      "step": 393120
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.9928828346215654e-05,
+      "loss": 0.8587,
+      "step": 393130
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.9927680416315908e-05,
+      "loss": 0.8135,
+      "step": 393140
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.992653248641616e-05,
+      "loss": 0.7769,
+      "step": 393150
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.9925384556516414e-05,
+      "loss": 0.917,
+      "step": 393160
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.9924236626616664e-05,
+      "loss": 0.8276,
+      "step": 393170
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.992308869671692e-05,
+      "loss": 0.7988,
+      "step": 393180
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.992194076681717e-05,
+      "loss": 0.8966,
+      "step": 393190
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.992079283691742e-05,
+      "loss": 0.8047,
+      "step": 393200
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.9919644907017677e-05,
+      "loss": 0.8454,
+      "step": 393210
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.9918496977117927e-05,
+      "loss": 0.7878,
+      "step": 393220
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.991734904721818e-05,
+      "loss": 0.841,
+      "step": 393230
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.9916201117318433e-05,
+      "loss": 0.8434,
+      "step": 393240
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.9915053187418686e-05,
+      "loss": 0.8492,
+      "step": 393250
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.9913905257518936e-05,
+      "loss": 0.8037,
+      "step": 393260
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.9912757327619193e-05,
+      "loss": 0.8209,
+      "step": 393270
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.9911609397719443e-05,
+      "loss": 0.9294,
+      "step": 393280
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.9910461467819696e-05,
+      "loss": 0.7472,
+      "step": 393290
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 2.990931353791995e-05,
+      "loss": 0.8102,
+      "step": 393300
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 2.9908165608020203e-05,
+      "loss": 0.8377,
+      "step": 393310
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 2.9907017678120452e-05,
+      "loss": 0.8243,
+      "step": 393320
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 2.990586974822071e-05,
+      "loss": 0.7967,
+      "step": 393330
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 2.990472181832096e-05,
+      "loss": 0.8377,
+      "step": 393340
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 2.990357388842121e-05,
+      "loss": 0.8333,
+      "step": 393350
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 2.9902425958521465e-05,
+      "loss": 0.7866,
+      "step": 393360
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 2.9901278028621715e-05,
+      "loss": 0.828,
+      "step": 393370
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 2.990013009872197e-05,
+      "loss": 0.7505,
+      "step": 393380
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 2.9898982168822222e-05,
+      "loss": 0.8611,
+      "step": 393390
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 2.9897834238922475e-05,
+      "loss": 0.8712,
+      "step": 393400
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 2.9896686309022725e-05,
+      "loss": 0.7741,
+      "step": 393410
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 2.9895538379122978e-05,
+      "loss": 0.8366,
+      "step": 393420
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 2.989439044922323e-05,
+      "loss": 0.8615,
+      "step": 393430
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 2.9893242519323485e-05,
+      "loss": 0.7667,
+      "step": 393440
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 2.9892094589423734e-05,
+      "loss": 0.8365,
+      "step": 393450
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 2.989094665952399e-05,
+      "loss": 0.8493,
+      "step": 393460
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 2.988979872962424e-05,
+      "loss": 0.9017,
+      "step": 393470
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 2.988865079972449e-05,
+      "loss": 0.8187,
+      "step": 393480
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 2.9887502869824747e-05,
+      "loss": 0.9023,
+      "step": 393490
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 2.9886354939924997e-05,
+      "loss": 0.9121,
+      "step": 393500
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 2.988520701002525e-05,
+      "loss": 0.8095,
+      "step": 393510
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 2.9884059080125504e-05,
+      "loss": 0.8727,
+      "step": 393520
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 2.9882911150225757e-05,
+      "loss": 0.8363,
+      "step": 393530
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 2.9881763220326007e-05,
+      "loss": 0.8628,
+      "step": 393540
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 2.9880615290426263e-05,
+      "loss": 0.8759,
+      "step": 393550
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 2.9879467360526513e-05,
+      "loss": 0.8244,
+      "step": 393560
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 2.9878319430626767e-05,
+      "loss": 0.8503,
+      "step": 393570
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 2.987717150072702e-05,
+      "loss": 0.76,
+      "step": 393580
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 2.9876023570827273e-05,
+      "loss": 0.8186,
+      "step": 393590
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 2.9874875640927523e-05,
+      "loss": 0.8598,
+      "step": 393600
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 2.987372771102778e-05,
+      "loss": 0.8065,
+      "step": 393610
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 2.987257978112803e-05,
+      "loss": 0.8718,
+      "step": 393620
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 2.987143185122828e-05,
+      "loss": 0.7604,
+      "step": 393630
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 2.9870283921328536e-05,
+      "loss": 0.8609,
+      "step": 393640
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 2.9869135991428786e-05,
+      "loss": 0.8134,
+      "step": 393650
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 2.986798806152904e-05,
+      "loss": 0.849,
+      "step": 393660
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 2.9866840131629292e-05,
+      "loss": 0.8336,
+      "step": 393670
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 2.9865692201729545e-05,
+      "loss": 0.7894,
+      "step": 393680
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 2.9864544271829795e-05,
+      "loss": 0.8521,
+      "step": 393690
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 2.9863396341930052e-05,
+      "loss": 0.8435,
+      "step": 393700
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 2.9862248412030302e-05,
+      "loss": 0.8735,
+      "step": 393710
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 2.9861100482130555e-05,
+      "loss": 0.755,
+      "step": 393720
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 2.9859952552230808e-05,
+      "loss": 0.8506,
+      "step": 393730
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 2.985880462233106e-05,
+      "loss": 0.9019,
+      "step": 393740
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 2.985765669243131e-05,
+      "loss": 0.88,
+      "step": 393750
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 2.9856508762531568e-05,
+      "loss": 0.7396,
+      "step": 393760
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 2.9855360832631818e-05,
+      "loss": 0.8417,
+      "step": 393770
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 2.9854212902732068e-05,
+      "loss": 0.7408,
+      "step": 393780
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 2.9853064972832324e-05,
+      "loss": 0.8207,
+      "step": 393790
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 2.9851917042932574e-05,
+      "loss": 0.8185,
+      "step": 393800
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 2.9850769113032827e-05,
+      "loss": 0.7967,
+      "step": 393810
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 2.984962118313308e-05,
+      "loss": 0.8776,
+      "step": 393820
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 2.9848473253233334e-05,
+      "loss": 0.7406,
+      "step": 393830
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 2.9847325323333584e-05,
+      "loss": 0.8122,
+      "step": 393840
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 2.984617739343384e-05,
+      "loss": 0.8427,
+      "step": 393850
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 2.984502946353409e-05,
+      "loss": 0.811,
+      "step": 393860
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 2.9843881533634343e-05,
+      "loss": 0.842,
+      "step": 393870
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 2.9842733603734597e-05,
+      "loss": 0.9031,
+      "step": 393880
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 2.984158567383485e-05,
+      "loss": 0.8983,
+      "step": 393890
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 2.98404377439351e-05,
+      "loss": 0.7803,
+      "step": 393900
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 2.9839289814035356e-05,
+      "loss": 0.7514,
+      "step": 393910
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 2.9838141884135606e-05,
+      "loss": 0.9289,
+      "step": 393920
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 2.9836993954235856e-05,
+      "loss": 0.7639,
+      "step": 393930
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 2.9835846024336113e-05,
+      "loss": 0.8787,
+      "step": 393940
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 2.9834698094436363e-05,
+      "loss": 0.9356,
+      "step": 393950
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 2.9833550164536616e-05,
+      "loss": 0.8038,
+      "step": 393960
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 2.983240223463687e-05,
+      "loss": 0.8854,
+      "step": 393970
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 2.9831254304737122e-05,
+      "loss": 0.8424,
+      "step": 393980
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 2.9830106374837372e-05,
+      "loss": 0.8432,
+      "step": 393990
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 2.982895844493763e-05,
+      "loss": 0.8506,
+      "step": 394000
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 2.982781051503788e-05,
+      "loss": 0.8407,
+      "step": 394010
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 2.9826662585138132e-05,
+      "loss": 0.9455,
+      "step": 394020
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 2.9825514655238385e-05,
+      "loss": 0.7982,
+      "step": 394030
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 2.982436672533864e-05,
+      "loss": 0.8,
+      "step": 394040
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 2.9823218795438888e-05,
+      "loss": 0.8289,
+      "step": 394050
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 2.9822070865539145e-05,
+      "loss": 0.7624,
+      "step": 394060
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 2.9820922935639395e-05,
+      "loss": 0.8094,
+      "step": 394070
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 2.9819775005739645e-05,
+      "loss": 0.8305,
+      "step": 394080
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 2.9818627075839898e-05,
+      "loss": 0.9178,
+      "step": 394090
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 2.981747914594015e-05,
+      "loss": 0.8004,
+      "step": 394100
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 2.9816331216040404e-05,
+      "loss": 0.8491,
+      "step": 394110
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 2.9815183286140654e-05,
+      "loss": 0.8329,
+      "step": 394120
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 2.981403535624091e-05,
+      "loss": 0.8022,
+      "step": 394130
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 2.981288742634116e-05,
+      "loss": 0.8888,
+      "step": 394140
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 2.9811739496441414e-05,
+      "loss": 0.9092,
+      "step": 394150
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 2.9810591566541667e-05,
+      "loss": 0.9321,
+      "step": 394160
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 2.980944363664192e-05,
+      "loss": 0.8227,
+      "step": 394170
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 2.980829570674217e-05,
+      "loss": 0.9241,
+      "step": 394180
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 2.9807147776842427e-05,
+      "loss": 0.8373,
+      "step": 394190
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 2.9805999846942677e-05,
+      "loss": 0.8218,
+      "step": 394200
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 2.9804851917042927e-05,
+      "loss": 0.8287,
+      "step": 394210
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 2.9803703987143183e-05,
+      "loss": 0.8301,
+      "step": 394220
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 2.9802556057243433e-05,
+      "loss": 0.8315,
+      "step": 394230
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 2.9801408127343686e-05,
+      "loss": 0.8004,
+      "step": 394240
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 2.980026019744394e-05,
+      "loss": 0.8313,
+      "step": 394250
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 2.9799112267544193e-05,
+      "loss": 0.8194,
+      "step": 394260
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 2.9797964337644443e-05,
+      "loss": 0.8362,
+      "step": 394270
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 2.97968164077447e-05,
+      "loss": 0.864,
+      "step": 394280
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 2.979566847784495e-05,
+      "loss": 0.7773,
+      "step": 394290
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 2.9794520547945202e-05,
+      "loss": 0.8309,
+      "step": 394300
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 2.9793372618045456e-05,
+      "loss": 0.7835,
+      "step": 394310
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 2.979222468814571e-05,
+      "loss": 0.8745,
+      "step": 394320
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 2.979107675824596e-05,
+      "loss": 0.8531,
+      "step": 394330
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 2.9789928828346215e-05,
+      "loss": 0.9047,
+      "step": 394340
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 2.9788780898446465e-05,
+      "loss": 0.7714,
+      "step": 394350
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 2.9787632968546715e-05,
+      "loss": 0.8862,
+      "step": 394360
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 2.9786485038646972e-05,
+      "loss": 0.8615,
+      "step": 394370
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 2.978533710874722e-05,
+      "loss": 0.8536,
+      "step": 394380
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 2.9784189178847475e-05,
+      "loss": 0.7609,
+      "step": 394390
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 2.9783041248947728e-05,
+      "loss": 0.8309,
+      "step": 394400
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 2.978189331904798e-05,
+      "loss": 0.8576,
+      "step": 394410
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 2.978074538914823e-05,
+      "loss": 0.7972,
+      "step": 394420
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 2.9779597459248488e-05,
+      "loss": 0.7987,
+      "step": 394430
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 2.9778449529348738e-05,
+      "loss": 0.8137,
+      "step": 394440
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 2.977730159944899e-05,
+      "loss": 0.8887,
+      "step": 394450
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 2.9776153669549244e-05,
+      "loss": 0.823,
+      "step": 394460
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 2.9775005739649497e-05,
+      "loss": 0.8118,
+      "step": 394470
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 2.9773857809749747e-05,
+      "loss": 0.8966,
+      "step": 394480
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 2.9772709879850004e-05,
+      "loss": 0.9159,
+      "step": 394490
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 2.9771561949950254e-05,
+      "loss": 0.8651,
+      "step": 394500
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 2.9770414020050504e-05,
+      "loss": 0.8584,
+      "step": 394510
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 2.976926609015076e-05,
+      "loss": 0.8105,
+      "step": 394520
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 2.976811816025101e-05,
+      "loss": 0.7329,
+      "step": 394530
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 2.9766970230351263e-05,
+      "loss": 0.9342,
+      "step": 394540
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 2.9765822300451517e-05,
+      "loss": 0.7871,
+      "step": 394550
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 2.976467437055177e-05,
+      "loss": 0.7237,
+      "step": 394560
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 2.976352644065202e-05,
+      "loss": 0.859,
+      "step": 394570
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 2.9762378510752276e-05,
+      "loss": 0.882,
+      "step": 394580
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 2.9761230580852526e-05,
+      "loss": 0.8242,
+      "step": 394590
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 2.976008265095278e-05,
+      "loss": 0.8271,
+      "step": 394600
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 2.9758934721053033e-05,
+      "loss": 0.8979,
+      "step": 394610
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 2.9757786791153286e-05,
+      "loss": 0.8858,
+      "step": 394620
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 2.9756638861253536e-05,
+      "loss": 0.8369,
+      "step": 394630
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 2.9755490931353792e-05,
+      "loss": 0.8553,
+      "step": 394640
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 2.9754343001454042e-05,
+      "loss": 0.7534,
+      "step": 394650
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 2.9753195071554292e-05,
+      "loss": 0.8487,
+      "step": 394660
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 2.975204714165455e-05,
+      "loss": 0.9084,
+      "step": 394670
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 2.97508992117548e-05,
+      "loss": 0.8975,
+      "step": 394680
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 2.9749751281855052e-05,
+      "loss": 0.8752,
+      "step": 394690
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 2.9748603351955305e-05,
+      "loss": 0.8588,
+      "step": 394700
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 2.9747455422055558e-05,
+      "loss": 0.7809,
+      "step": 394710
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 2.9746307492155808e-05,
+      "loss": 0.8607,
+      "step": 394720
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 2.9745159562256065e-05,
+      "loss": 0.8916,
+      "step": 394730
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 2.9744011632356315e-05,
+      "loss": 0.7817,
+      "step": 394740
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 2.9742863702456568e-05,
+      "loss": 0.8569,
+      "step": 394750
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 2.974171577255682e-05,
+      "loss": 0.8019,
+      "step": 394760
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 2.9740567842657074e-05,
+      "loss": 0.8179,
+      "step": 394770
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 2.9739419912757324e-05,
+      "loss": 0.831,
+      "step": 394780
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 2.9738271982857574e-05,
+      "loss": 0.8017,
+      "step": 394790
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 2.973712405295783e-05,
+      "loss": 0.7291,
+      "step": 394800
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 2.973597612305808e-05,
+      "loss": 0.8691,
+      "step": 394810
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 2.9734828193158334e-05,
+      "loss": 0.8292,
+      "step": 394820
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 2.9733680263258587e-05,
+      "loss": 0.7541,
+      "step": 394830
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 2.973253233335884e-05,
+      "loss": 0.8822,
+      "step": 394840
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 2.973138440345909e-05,
+      "loss": 0.8722,
+      "step": 394850
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 2.9730236473559347e-05,
+      "loss": 0.816,
+      "step": 394860
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 2.9729088543659597e-05,
+      "loss": 0.9018,
+      "step": 394870
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 2.972794061375985e-05,
+      "loss": 0.8271,
+      "step": 394880
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 2.9726792683860103e-05,
+      "loss": 0.7725,
+      "step": 394890
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 2.9725644753960356e-05,
+      "loss": 0.938,
+      "step": 394900
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 2.9724496824060606e-05,
+      "loss": 0.8048,
+      "step": 394910
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 2.9723348894160863e-05,
+      "loss": 0.8464,
+      "step": 394920
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 2.9722200964261113e-05,
+      "loss": 0.839,
+      "step": 394930
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 2.9721053034361362e-05,
+      "loss": 0.8209,
+      "step": 394940
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 2.971990510446162e-05,
+      "loss": 0.8095,
+      "step": 394950
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 2.971875717456187e-05,
+      "loss": 0.7913,
+      "step": 394960
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 2.9717609244662122e-05,
+      "loss": 0.7705,
+      "step": 394970
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 2.9716461314762375e-05,
+      "loss": 0.8917,
+      "step": 394980
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 2.971531338486263e-05,
+      "loss": 0.799,
+      "step": 394990
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 2.971416545496288e-05,
+      "loss": 0.7837,
+      "step": 395000
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 2.9713017525063135e-05,
+      "loss": 0.8862,
+      "step": 395010
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 2.9711869595163385e-05,
+      "loss": 0.785,
+      "step": 395020
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 2.9710721665263638e-05,
+      "loss": 0.8795,
+      "step": 395030
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 2.970957373536389e-05,
+      "loss": 0.8717,
+      "step": 395040
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 2.9708425805464145e-05,
+      "loss": 0.7848,
+      "step": 395050
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 2.9707277875564395e-05,
+      "loss": 0.8566,
+      "step": 395060
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 2.970612994566465e-05,
+      "loss": 0.8854,
+      "step": 395070
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 2.97049820157649e-05,
+      "loss": 0.8158,
+      "step": 395080
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 2.970383408586515e-05,
+      "loss": 0.902,
+      "step": 395090
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 2.9702686155965408e-05,
+      "loss": 0.8101,
+      "step": 395100
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 2.9701538226065657e-05,
+      "loss": 0.8197,
+      "step": 395110
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 2.970039029616591e-05,
+      "loss": 0.8367,
+      "step": 395120
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 2.9699242366266164e-05,
+      "loss": 0.7916,
+      "step": 395130
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 2.9698094436366417e-05,
+      "loss": 0.8154,
+      "step": 395140
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 2.9696946506466667e-05,
+      "loss": 0.873,
+      "step": 395150
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 2.9695798576566924e-05,
+      "loss": 0.8382,
+      "step": 395160
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 2.9694650646667174e-05,
+      "loss": 0.9014,
+      "step": 395170
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 2.9693502716767427e-05,
+      "loss": 0.8218,
+      "step": 395180
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 2.969235478686768e-05,
+      "loss": 0.814,
+      "step": 395190
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 2.9691206856967933e-05,
+      "loss": 0.8881,
+      "step": 395200
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 2.9690058927068183e-05,
+      "loss": 0.8554,
+      "step": 395210
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 2.968891099716844e-05,
+      "loss": 0.8533,
+      "step": 395220
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 2.968776306726869e-05,
+      "loss": 0.8079,
+      "step": 395230
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 2.968661513736894e-05,
+      "loss": 0.8452,
+      "step": 395240
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 2.9685467207469196e-05,
+      "loss": 0.8507,
+      "step": 395250
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 2.9684319277569446e-05,
+      "loss": 0.7407,
+      "step": 395260
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 2.96831713476697e-05,
+      "loss": 0.8573,
+      "step": 395270
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 2.9682023417769952e-05,
+      "loss": 0.7692,
+      "step": 395280
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 2.9680875487870206e-05,
+      "loss": 0.8565,
+      "step": 395290
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 2.9679727557970455e-05,
+      "loss": 0.8743,
+      "step": 395300
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 2.9678579628070712e-05,
+      "loss": 0.7766,
+      "step": 395310
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 2.9677431698170962e-05,
+      "loss": 0.8418,
+      "step": 395320
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 2.9676283768271215e-05,
+      "loss": 0.8412,
+      "step": 395330
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 2.967513583837147e-05,
+      "loss": 0.8371,
+      "step": 395340
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 2.9673987908471722e-05,
+      "loss": 0.8288,
+      "step": 395350
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 2.967283997857197e-05,
+      "loss": 0.7914,
+      "step": 395360
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 2.9671692048672228e-05,
+      "loss": 0.8283,
+      "step": 395370
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 2.9670544118772478e-05,
+      "loss": 0.8916,
+      "step": 395380
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 2.9669396188872728e-05,
+      "loss": 0.8086,
+      "step": 395390
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 2.9668248258972985e-05,
+      "loss": 0.8584,
+      "step": 395400
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 2.9667100329073234e-05,
+      "loss": 0.8853,
+      "step": 395410
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 2.9665952399173488e-05,
+      "loss": 0.8086,
+      "step": 395420
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 2.966480446927374e-05,
+      "loss": 0.7985,
+      "step": 395430
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 2.9663656539373994e-05,
+      "loss": 0.8319,
+      "step": 395440
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 2.9662508609474244e-05,
+      "loss": 0.8496,
+      "step": 395450
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 2.96613606795745e-05,
+      "loss": 0.8497,
+      "step": 395460
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 2.966021274967475e-05,
+      "loss": 0.8847,
+      "step": 395470
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 2.9659064819775004e-05,
+      "loss": 0.8232,
+      "step": 395480
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 2.9657916889875254e-05,
+      "loss": 0.9385,
+      "step": 395490
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 2.965676895997551e-05,
+      "loss": 0.7872,
+      "step": 395500
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 2.965562103007576e-05,
+      "loss": 0.945,
+      "step": 395510
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 2.965447310017601e-05,
+      "loss": 0.7913,
+      "step": 395520
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 2.9653325170276267e-05,
+      "loss": 0.7748,
+      "step": 395530
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 2.9652177240376516e-05,
+      "loss": 0.8483,
+      "step": 395540
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 2.965102931047677e-05,
+      "loss": 0.8179,
+      "step": 395550
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 2.9649881380577023e-05,
+      "loss": 0.841,
+      "step": 395560
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 2.9648733450677276e-05,
+      "loss": 0.8084,
+      "step": 395570
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 2.9647585520777526e-05,
+      "loss": 0.8218,
+      "step": 395580
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 2.9646437590877783e-05,
+      "loss": 0.8573,
+      "step": 395590
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 2.9645289660978032e-05,
+      "loss": 0.8047,
+      "step": 395600
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 2.9644141731078286e-05,
+      "loss": 0.8972,
+      "step": 395610
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 2.964299380117854e-05,
+      "loss": 0.8416,
+      "step": 395620
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 2.9641845871278792e-05,
+      "loss": 0.799,
+      "step": 395630
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 2.9640697941379042e-05,
+      "loss": 0.842,
+      "step": 395640
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 2.96395500114793e-05,
+      "loss": 0.842,
+      "step": 395650
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 2.963840208157955e-05,
+      "loss": 0.9341,
+      "step": 395660
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 2.96372541516798e-05,
+      "loss": 0.9072,
+      "step": 395670
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 2.9636106221780055e-05,
+      "loss": 0.8888,
+      "step": 395680
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 2.9634958291880305e-05,
+      "loss": 0.8049,
+      "step": 395690
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 2.9633810361980558e-05,
+      "loss": 0.9258,
+      "step": 395700
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 2.963266243208081e-05,
+      "loss": 0.774,
+      "step": 395710
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 2.9631514502181065e-05,
+      "loss": 0.779,
+      "step": 395720
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 2.9630366572281314e-05,
+      "loss": 0.8878,
+      "step": 395730
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 2.962921864238157e-05,
+      "loss": 0.8703,
+      "step": 395740
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 2.962807071248182e-05,
+      "loss": 0.8183,
+      "step": 395750
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 2.9626922782582074e-05,
+      "loss": 0.8579,
+      "step": 395760
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 2.9625774852682327e-05,
+      "loss": 0.7572,
+      "step": 395770
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 2.962462692278258e-05,
+      "loss": 0.858,
+      "step": 395780
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 2.962347899288283e-05,
+      "loss": 0.8693,
+      "step": 395790
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 2.9622331062983087e-05,
+      "loss": 0.8396,
+      "step": 395800
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 2.9621183133083337e-05,
+      "loss": 0.8219,
+      "step": 395810
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 2.9620035203183587e-05,
+      "loss": 0.8472,
+      "step": 395820
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 2.9618887273283843e-05,
+      "loss": 0.7695,
+      "step": 395830
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 2.9617739343384093e-05,
+      "loss": 0.7888,
+      "step": 395840
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 2.9616591413484347e-05,
+      "loss": 0.8324,
+      "step": 395850
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 2.96154434835846e-05,
+      "loss": 0.814,
+      "step": 395860
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 2.9614295553684853e-05,
+      "loss": 0.8163,
+      "step": 395870
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 2.9613147623785103e-05,
+      "loss": 0.8955,
+      "step": 395880
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 2.961199969388536e-05,
+      "loss": 0.8479,
+      "step": 395890
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 2.961085176398561e-05,
+      "loss": 0.8743,
+      "step": 395900
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 2.9609703834085863e-05,
+      "loss": 0.8677,
+      "step": 395910
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 2.9608555904186116e-05,
+      "loss": 0.8361,
+      "step": 395920
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 2.960740797428637e-05,
+      "loss": 0.8154,
+      "step": 395930
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 2.960626004438662e-05,
+      "loss": 0.8555,
+      "step": 395940
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 2.9605112114486876e-05,
+      "loss": 0.8814,
+      "step": 395950
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 2.9603964184587125e-05,
+      "loss": 0.82,
+      "step": 395960
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 2.9602816254687375e-05,
+      "loss": 0.8327,
+      "step": 395970
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 2.9601668324787632e-05,
+      "loss": 0.8821,
+      "step": 395980
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 2.9600520394887882e-05,
+      "loss": 0.9276,
+      "step": 395990
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 2.9599372464988135e-05,
+      "loss": 0.8764,
+      "step": 396000
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 2.9598224535088388e-05,
+      "loss": 0.9049,
+      "step": 396010
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 2.959707660518864e-05,
+      "loss": 0.8208,
+      "step": 396020
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 2.959592867528889e-05,
+      "loss": 0.8136,
+      "step": 396030
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 2.9594780745389148e-05,
+      "loss": 0.9067,
+      "step": 396040
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 2.9593632815489398e-05,
+      "loss": 0.7982,
+      "step": 396050
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 2.959248488558965e-05,
+      "loss": 0.7785,
+      "step": 396060
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 2.9591336955689904e-05,
+      "loss": 0.8694,
+      "step": 396070
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 2.9590189025790158e-05,
+      "loss": 0.8599,
+      "step": 396080
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 2.9589041095890407e-05,
+      "loss": 0.9213,
+      "step": 396090
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 2.9587893165990664e-05,
+      "loss": 0.815,
+      "step": 396100
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 2.9586745236090914e-05,
+      "loss": 0.7948,
+      "step": 396110
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 2.9585597306191164e-05,
+      "loss": 0.7848,
+      "step": 396120
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 2.958444937629142e-05,
+      "loss": 0.7455,
+      "step": 396130
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 2.958330144639167e-05,
+      "loss": 0.8776,
+      "step": 396140
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 2.9582153516491924e-05,
+      "loss": 0.8737,
+      "step": 396150
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 2.9581005586592173e-05,
+      "loss": 0.9002,
+      "step": 396160
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 2.957985765669243e-05,
+      "loss": 0.9717,
+      "step": 396170
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 2.957870972679268e-05,
+      "loss": 0.897,
+      "step": 396180
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 2.9577561796892933e-05,
+      "loss": 0.9156,
+      "step": 396190
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 2.9576413866993186e-05,
+      "loss": 0.8613,
+      "step": 396200
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 2.957526593709344e-05,
+      "loss": 0.8994,
+      "step": 396210
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 2.957411800719369e-05,
+      "loss": 0.8276,
+      "step": 396220
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 2.9572970077293946e-05,
+      "loss": 0.7265,
+      "step": 396230
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 2.9571822147394196e-05,
+      "loss": 0.8998,
+      "step": 396240
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 2.9570674217494446e-05,
+      "loss": 0.8523,
+      "step": 396250
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 2.9569526287594702e-05,
+      "loss": 0.8889,
+      "step": 396260
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 2.9568378357694952e-05,
+      "loss": 0.7907,
+      "step": 396270
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 2.9567230427795205e-05,
+      "loss": 0.8157,
+      "step": 396280
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 2.956608249789546e-05,
+      "loss": 0.7887,
+      "step": 396290
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 2.9564934567995712e-05,
+      "loss": 0.8511,
+      "step": 396300
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 2.9563786638095962e-05,
+      "loss": 0.783,
+      "step": 396310
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 2.956263870819622e-05,
+      "loss": 0.8065,
+      "step": 396320
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 2.956149077829647e-05,
+      "loss": 0.7129,
+      "step": 396330
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 2.956034284839672e-05,
+      "loss": 0.7682,
+      "step": 396340
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 2.9559194918496975e-05,
+      "loss": 0.902,
+      "step": 396350
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 2.9558046988597228e-05,
+      "loss": 0.7864,
+      "step": 396360
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 2.9556899058697478e-05,
+      "loss": 0.8036,
+      "step": 396370
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 2.9555751128797735e-05,
+      "loss": 0.8461,
+      "step": 396380
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 2.9554603198897984e-05,
+      "loss": 0.8101,
+      "step": 396390
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 2.9553455268998234e-05,
+      "loss": 0.8253,
+      "step": 396400
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 2.955230733909849e-05,
+      "loss": 0.8798,
+      "step": 396410
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 2.955115940919874e-05,
+      "loss": 0.8189,
+      "step": 396420
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 2.9550011479298994e-05,
+      "loss": 0.8645,
+      "step": 396430
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 2.9548863549399247e-05,
+      "loss": 0.8162,
+      "step": 396440
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 2.95477156194995e-05,
+      "loss": 0.888,
+      "step": 396450
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 2.954656768959975e-05,
+      "loss": 0.8206,
+      "step": 396460
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 2.9545419759700007e-05,
+      "loss": 0.8796,
+      "step": 396470
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 2.9544271829800257e-05,
+      "loss": 0.9248,
+      "step": 396480
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 2.954312389990051e-05,
+      "loss": 0.8302,
+      "step": 396490
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 2.9541975970000763e-05,
+      "loss": 0.8536,
+      "step": 396500
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 2.9540828040101017e-05,
+      "loss": 0.7443,
+      "step": 396510
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 2.9539680110201266e-05,
+      "loss": 0.7463,
+      "step": 396520
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 2.9538532180301523e-05,
+      "loss": 0.8023,
+      "step": 396530
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 2.9537384250401773e-05,
+      "loss": 0.8306,
+      "step": 396540
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 2.9536236320502023e-05,
+      "loss": 0.8879,
+      "step": 396550
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 2.953508839060228e-05,
+      "loss": 0.8439,
+      "step": 396560
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 2.953394046070253e-05,
+      "loss": 0.8982,
+      "step": 396570
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 2.9532792530802782e-05,
+      "loss": 0.8161,
+      "step": 396580
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 2.9531644600903036e-05,
+      "loss": 0.8775,
+      "step": 396590
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 2.953049667100329e-05,
+      "loss": 0.9453,
+      "step": 396600
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 2.952934874110354e-05,
+      "loss": 0.8313,
+      "step": 396610
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 2.9528200811203795e-05,
+      "loss": 0.8084,
+      "step": 396620
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 2.9527052881304045e-05,
+      "loss": 0.7882,
+      "step": 396630
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 2.95259049514043e-05,
+      "loss": 0.7827,
+      "step": 396640
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 2.9524757021504552e-05,
+      "loss": 0.8986,
+      "step": 396650
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 2.9523609091604805e-05,
+      "loss": 0.9077,
+      "step": 396660
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 2.9522461161705055e-05,
+      "loss": 0.7649,
+      "step": 396670
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 2.952131323180531e-05,
+      "loss": 0.9072,
+      "step": 396680
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 2.952016530190556e-05,
+      "loss": 0.7986,
+      "step": 396690
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 2.951901737200581e-05,
+      "loss": 0.839,
+      "step": 396700
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 2.9517869442106068e-05,
+      "loss": 0.8836,
+      "step": 396710
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 2.9516721512206318e-05,
+      "loss": 0.8963,
+      "step": 396720
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 2.951557358230657e-05,
+      "loss": 0.762,
+      "step": 396730
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 2.9514425652406824e-05,
+      "loss": 0.8289,
+      "step": 396740
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 2.9513277722507077e-05,
+      "loss": 0.8773,
+      "step": 396750
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 2.9512129792607327e-05,
+      "loss": 0.8036,
+      "step": 396760
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 2.9510981862707584e-05,
+      "loss": 0.8178,
+      "step": 396770
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 2.9509833932807834e-05,
+      "loss": 0.8797,
+      "step": 396780
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 2.9508686002908087e-05,
+      "loss": 0.8351,
+      "step": 396790
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 2.950753807300834e-05,
+      "loss": 0.9523,
+      "step": 396800
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 2.9506390143108593e-05,
+      "loss": 0.8648,
+      "step": 396810
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 2.9505242213208843e-05,
+      "loss": 0.8818,
+      "step": 396820
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 2.95040942833091e-05,
+      "loss": 0.8222,
+      "step": 396830
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 2.950294635340935e-05,
+      "loss": 0.9186,
+      "step": 396840
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 2.95017984235096e-05,
+      "loss": 0.8417,
+      "step": 396850
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 2.9500650493609853e-05,
+      "loss": 0.8097,
+      "step": 396860
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 2.9499502563710106e-05,
+      "loss": 0.8496,
+      "step": 396870
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 2.949835463381036e-05,
+      "loss": 0.8424,
+      "step": 396880
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 2.949720670391061e-05,
+      "loss": 0.8251,
+      "step": 396890
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 2.9496058774010866e-05,
+      "loss": 0.8874,
+      "step": 396900
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 2.9494910844111116e-05,
+      "loss": 0.8695,
+      "step": 396910
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 2.949376291421137e-05,
+      "loss": 0.91,
+      "step": 396920
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 2.9492614984311622e-05,
+      "loss": 0.8411,
+      "step": 396930
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 2.9491467054411875e-05,
+      "loss": 0.8409,
+      "step": 396940
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 2.9490319124512125e-05,
+      "loss": 0.788,
+      "step": 396950
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 2.9489171194612382e-05,
+      "loss": 0.8534,
+      "step": 396960
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 2.9488023264712632e-05,
+      "loss": 0.7513,
+      "step": 396970
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 2.948687533481288e-05,
+      "loss": 0.8774,
+      "step": 396980
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 2.9485727404913138e-05,
+      "loss": 0.9235,
+      "step": 396990
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 2.9484579475013388e-05,
+      "loss": 0.8197,
+      "step": 397000
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 2.948343154511364e-05,
+      "loss": 0.9021,
+      "step": 397010
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 2.9482283615213895e-05,
+      "loss": 0.7395,
+      "step": 397020
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 2.9481135685314148e-05,
+      "loss": 0.7959,
+      "step": 397030
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 2.9479987755414398e-05,
+      "loss": 0.8822,
+      "step": 397040
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 2.9478839825514654e-05,
+      "loss": 0.9013,
+      "step": 397050
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 2.9477691895614904e-05,
+      "loss": 0.741,
+      "step": 397060
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 2.9476543965715157e-05,
+      "loss": 0.7989,
+      "step": 397070
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 2.947539603581541e-05,
+      "loss": 0.8807,
+      "step": 397080
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 2.9474248105915664e-05,
+      "loss": 0.8353,
+      "step": 397090
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 2.9473100176015914e-05,
+      "loss": 0.8429,
+      "step": 397100
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 2.947195224611617e-05,
+      "loss": 0.8341,
+      "step": 397110
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 2.947080431621642e-05,
+      "loss": 0.9413,
+      "step": 397120
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 2.946965638631667e-05,
+      "loss": 0.9002,
+      "step": 397130
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 2.9468508456416927e-05,
+      "loss": 0.7851,
+      "step": 397140
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 2.9467360526517177e-05,
+      "loss": 0.8392,
+      "step": 397150
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 2.946621259661743e-05,
+      "loss": 0.8197,
+      "step": 397160
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 2.9465064666717683e-05,
+      "loss": 0.8492,
+      "step": 397170
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 2.9463916736817936e-05,
+      "loss": 0.8529,
+      "step": 397180
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 2.9462768806918186e-05,
+      "loss": 0.7648,
+      "step": 397190
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 2.9461620877018443e-05,
+      "loss": 0.932,
+      "step": 397200
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 2.9460472947118693e-05,
+      "loss": 0.8745,
+      "step": 397210
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 2.9459325017218946e-05,
+      "loss": 0.8539,
+      "step": 397220
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 2.94581770873192e-05,
+      "loss": 0.8077,
+      "step": 397230
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 2.9457029157419452e-05,
+      "loss": 0.8658,
+      "step": 397240
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 2.9455881227519702e-05,
+      "loss": 0.8588,
+      "step": 397250
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 2.945473329761996e-05,
+      "loss": 0.8357,
+      "step": 397260
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 2.945358536772021e-05,
+      "loss": 0.8112,
+      "step": 397270
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 2.945243743782046e-05,
+      "loss": 0.765,
+      "step": 397280
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 2.9451289507920715e-05,
+      "loss": 0.8672,
+      "step": 397290
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 2.9450141578020965e-05,
+      "loss": 0.8749,
+      "step": 397300
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 2.944899364812122e-05,
+      "loss": 0.8177,
+      "step": 397310
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 2.944784571822147e-05,
+      "loss": 0.7769,
+      "step": 397320
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 2.9446697788321725e-05,
+      "loss": 0.835,
+      "step": 397330
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 2.9445549858421975e-05,
+      "loss": 0.7769,
+      "step": 397340
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 2.944440192852223e-05,
+      "loss": 0.8253,
+      "step": 397350
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 2.944325399862248e-05,
+      "loss": 0.864,
+      "step": 397360
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 2.9442106068722734e-05,
+      "loss": 0.8526,
+      "step": 397370
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 2.9440958138822988e-05,
+      "loss": 0.7598,
+      "step": 397380
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 2.943981020892324e-05,
+      "loss": 0.9538,
+      "step": 397390
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 2.943866227902349e-05,
+      "loss": 0.8606,
+      "step": 397400
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 2.9437514349123747e-05,
+      "loss": 0.79,
+      "step": 397410
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 2.9436366419223997e-05,
+      "loss": 0.7898,
+      "step": 397420
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 2.9435218489324247e-05,
+      "loss": 0.8593,
+      "step": 397430
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 2.9434070559424504e-05,
+      "loss": 0.8438,
+      "step": 397440
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 2.9432922629524754e-05,
+      "loss": 0.7442,
+      "step": 397450
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 2.9431774699625007e-05,
+      "loss": 0.883,
+      "step": 397460
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 2.943062676972526e-05,
+      "loss": 0.8408,
+      "step": 397470
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 2.9429478839825513e-05,
+      "loss": 0.8099,
+      "step": 397480
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 2.9428330909925763e-05,
+      "loss": 0.7771,
+      "step": 397490
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 2.942718298002602e-05,
+      "loss": 0.8322,
+      "step": 397500
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 2.942603505012627e-05,
+      "loss": 0.7587,
+      "step": 397510
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 2.9424887120226523e-05,
+      "loss": 0.8967,
+      "step": 397520
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 2.9423739190326773e-05,
+      "loss": 0.8524,
+      "step": 397530
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 2.942259126042703e-05,
+      "loss": 0.8663,
+      "step": 397540
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 2.942144333052728e-05,
+      "loss": 0.8788,
+      "step": 397550
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 2.942029540062753e-05,
+      "loss": 0.7537,
+      "step": 397560
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 2.9419147470727786e-05,
+      "loss": 0.7953,
+      "step": 397570
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 2.9417999540828036e-05,
+      "loss": 0.8392,
+      "step": 397580
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 2.941685161092829e-05,
+      "loss": 0.7876,
+      "step": 397590
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 2.9415703681028542e-05,
+      "loss": 0.8763,
+      "step": 397600
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 2.9414555751128795e-05,
+      "loss": 0.8417,
+      "step": 397610
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 2.9413407821229045e-05,
+      "loss": 0.8053,
+      "step": 397620
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 2.9412259891329302e-05,
+      "loss": 0.8181,
+      "step": 397630
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 2.941111196142955e-05,
+      "loss": 0.8983,
+      "step": 397640
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 2.9409964031529805e-05,
+      "loss": 0.7154,
+      "step": 397650
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 2.9408816101630058e-05,
+      "loss": 0.8031,
+      "step": 397660
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 2.940766817173031e-05,
+      "loss": 0.8308,
+      "step": 397670
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 2.940652024183056e-05,
+      "loss": 0.9026,
+      "step": 397680
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 2.9405372311930818e-05,
+      "loss": 0.9114,
+      "step": 397690
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 2.9404224382031068e-05,
+      "loss": 0.815,
+      "step": 397700
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 2.9403076452131318e-05,
+      "loss": 0.8502,
+      "step": 397710
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 2.9401928522231574e-05,
+      "loss": 0.8614,
+      "step": 397720
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 2.9400780592331824e-05,
+      "loss": 0.8622,
+      "step": 397730
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 2.9399632662432077e-05,
+      "loss": 0.7758,
+      "step": 397740
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 2.939848473253233e-05,
+      "loss": 0.8835,
+      "step": 397750
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 2.9397336802632584e-05,
+      "loss": 0.7648,
+      "step": 397760
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 2.9396188872732834e-05,
+      "loss": 0.809,
+      "step": 397770
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 2.939504094283309e-05,
+      "loss": 0.8514,
+      "step": 397780
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 2.939389301293334e-05,
+      "loss": 0.8517,
+      "step": 397790
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 2.9392745083033593e-05,
+      "loss": 1.004,
+      "step": 397800
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 2.9391597153133847e-05,
+      "loss": 0.8856,
+      "step": 397810
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 2.93904492232341e-05,
+      "loss": 0.8149,
+      "step": 397820
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 2.938930129333435e-05,
+      "loss": 0.8467,
+      "step": 397830
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 2.9388153363434606e-05,
+      "loss": 0.9402,
+      "step": 397840
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 2.9387005433534856e-05,
+      "loss": 0.8593,
+      "step": 397850
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 2.9385857503635106e-05,
+      "loss": 0.7461,
+      "step": 397860
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 2.9384709573735363e-05,
+      "loss": 0.8712,
+      "step": 397870
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 2.9383561643835612e-05,
+      "loss": 0.9475,
+      "step": 397880
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 2.9382413713935866e-05,
+      "loss": 0.8128,
+      "step": 397890
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 2.938126578403612e-05,
+      "loss": 0.8903,
+      "step": 397900
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 2.9380117854136372e-05,
+      "loss": 0.8522,
+      "step": 397910
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 2.9378969924236622e-05,
+      "loss": 0.8461,
+      "step": 397920
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 2.937782199433688e-05,
+      "loss": 1.0087,
+      "step": 397930
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 2.937667406443713e-05,
+      "loss": 0.8797,
+      "step": 397940
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 2.9375526134537382e-05,
+      "loss": 0.8515,
+      "step": 397950
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 2.9374378204637635e-05,
+      "loss": 0.8728,
+      "step": 397960
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 2.9373230274737888e-05,
+      "loss": 0.8383,
+      "step": 397970
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 2.9372082344838138e-05,
+      "loss": 0.776,
+      "step": 397980
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 2.9370934414938395e-05,
+      "loss": 0.81,
+      "step": 397990
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 2.9369786485038645e-05,
+      "loss": 0.8137,
+      "step": 398000
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 2.9368638555138894e-05,
+      "loss": 0.8249,
+      "step": 398010
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 2.936749062523915e-05,
+      "loss": 0.8348,
+      "step": 398020
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 2.93663426953394e-05,
+      "loss": 0.7415,
+      "step": 398030
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 2.9365194765439654e-05,
+      "loss": 0.8652,
+      "step": 398040
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 2.9364046835539907e-05,
+      "loss": 0.8765,
+      "step": 398050
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 2.936289890564016e-05,
+      "loss": 0.8975,
+      "step": 398060
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 2.936175097574041e-05,
+      "loss": 0.9219,
+      "step": 398070
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 2.9360603045840667e-05,
+      "loss": 0.8135,
+      "step": 398080
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 2.9359455115940917e-05,
+      "loss": 0.7812,
+      "step": 398090
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 2.935830718604117e-05,
+      "loss": 0.8193,
+      "step": 398100
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 2.9357159256141423e-05,
+      "loss": 0.8189,
+      "step": 398110
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 2.9356011326241677e-05,
+      "loss": 0.8831,
+      "step": 398120
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 2.9354863396341927e-05,
+      "loss": 0.8974,
+      "step": 398130
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 2.9353715466442183e-05,
+      "loss": 0.8608,
+      "step": 398140
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 2.9352567536542433e-05,
+      "loss": 0.8088,
+      "step": 398150
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 2.9351419606642683e-05,
+      "loss": 0.8832,
+      "step": 398160
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 2.935027167674294e-05,
+      "loss": 0.8052,
+      "step": 398170
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 2.934912374684319e-05,
+      "loss": 0.8633,
+      "step": 398180
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 2.9347975816943443e-05,
+      "loss": 0.8193,
+      "step": 398190
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 2.9346827887043696e-05,
+      "loss": 0.8091,
+      "step": 398200
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 2.934567995714395e-05,
+      "loss": 0.7827,
+      "step": 398210
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 2.93445320272442e-05,
+      "loss": 0.8246,
+      "step": 398220
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 2.9343384097344452e-05,
+      "loss": 0.8645,
+      "step": 398230
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 2.9342236167444705e-05,
+      "loss": 0.8885,
+      "step": 398240
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 2.934108823754496e-05,
+      "loss": 0.9483,
+      "step": 398250
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 2.933994030764521e-05,
+      "loss": 0.8305,
+      "step": 398260
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 2.9338792377745465e-05,
+      "loss": 0.7856,
+      "step": 398270
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 2.9337644447845715e-05,
+      "loss": 0.808,
+      "step": 398280
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 2.9336496517945965e-05,
+      "loss": 0.8879,
+      "step": 398290
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 2.933534858804622e-05,
+      "loss": 0.7323,
+      "step": 398300
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 2.933420065814647e-05,
+      "loss": 0.8572,
+      "step": 398310
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 2.9333052728246725e-05,
+      "loss": 0.9105,
+      "step": 398320
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 2.9331904798346978e-05,
+      "loss": 0.8289,
+      "step": 398330
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 2.933075686844723e-05,
+      "loss": 0.8819,
+      "step": 398340
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 2.932960893854748e-05,
+      "loss": 0.8736,
+      "step": 398350
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 2.9328461008647738e-05,
+      "loss": 0.8242,
+      "step": 398360
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 2.9327313078747987e-05,
+      "loss": 0.8373,
+      "step": 398370
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 2.932616514884824e-05,
+      "loss": 0.7417,
+      "step": 398380
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 2.9325017218948494e-05,
+      "loss": 0.8679,
+      "step": 398390
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 2.9323869289048747e-05,
+      "loss": 0.7753,
+      "step": 398400
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 2.9322721359148997e-05,
+      "loss": 0.8268,
+      "step": 398410
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 2.9321573429249254e-05,
+      "loss": 0.825,
+      "step": 398420
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 2.9320425499349504e-05,
+      "loss": 0.9138,
+      "step": 398430
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 2.9319277569449753e-05,
+      "loss": 0.8259,
+      "step": 398440
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 2.931812963955001e-05,
+      "loss": 0.8169,
+      "step": 398450
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 2.931698170965026e-05,
+      "loss": 0.7421,
+      "step": 398460
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 2.9315833779750513e-05,
+      "loss": 0.903,
+      "step": 398470
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 2.9314685849850766e-05,
+      "loss": 0.8177,
+      "step": 398480
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 2.931353791995102e-05,
+      "loss": 0.8411,
+      "step": 398490
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 2.931238999005127e-05,
+      "loss": 0.8839,
+      "step": 398500
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 2.9311242060151526e-05,
+      "loss": 0.8043,
+      "step": 398510
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 2.9310094130251776e-05,
+      "loss": 0.8622,
+      "step": 398520
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 2.930894620035203e-05,
+      "loss": 0.8596,
+      "step": 398530
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 2.9307798270452282e-05,
+      "loss": 0.7996,
+      "step": 398540
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 2.9306650340552536e-05,
+      "loss": 0.8778,
+      "step": 398550
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 2.9305502410652786e-05,
+      "loss": 0.8505,
+      "step": 398560
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 2.9304354480753042e-05,
+      "loss": 0.9356,
+      "step": 398570
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 2.9303206550853292e-05,
+      "loss": 0.8962,
+      "step": 398580
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 2.9302058620953542e-05,
+      "loss": 0.7648,
+      "step": 398590
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 2.93009106910538e-05,
+      "loss": 0.8485,
+      "step": 398600
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 2.929976276115405e-05,
+      "loss": 0.857,
+      "step": 398610
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 2.92986148312543e-05,
+      "loss": 0.8972,
+      "step": 398620
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 2.9297466901354555e-05,
+      "loss": 0.8904,
+      "step": 398630
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 2.9296318971454808e-05,
+      "loss": 0.869,
+      "step": 398640
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 2.9295171041555058e-05,
+      "loss": 0.8408,
+      "step": 398650
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 2.9294023111655315e-05,
+      "loss": 0.8659,
+      "step": 398660
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 2.9292875181755564e-05,
+      "loss": 0.8723,
+      "step": 398670
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 2.9291727251855818e-05,
+      "loss": 0.8033,
+      "step": 398680
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 2.929057932195607e-05,
+      "loss": 0.8067,
+      "step": 398690
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 2.9289431392056324e-05,
+      "loss": 0.9002,
+      "step": 398700
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 2.9288283462156574e-05,
+      "loss": 0.8624,
+      "step": 398710
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 2.928713553225683e-05,
+      "loss": 0.8287,
+      "step": 398720
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 2.928598760235708e-05,
+      "loss": 0.8174,
+      "step": 398730
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 2.928483967245733e-05,
+      "loss": 0.8397,
+      "step": 398740
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 2.9283691742557587e-05,
+      "loss": 0.8862,
+      "step": 398750
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 2.9282543812657837e-05,
+      "loss": 0.8672,
+      "step": 398760
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 2.928139588275809e-05,
+      "loss": 0.8599,
+      "step": 398770
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 2.9280247952858343e-05,
+      "loss": 0.8837,
+      "step": 398780
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 2.9279100022958597e-05,
+      "loss": 0.7939,
+      "step": 398790
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 2.9277952093058846e-05,
+      "loss": 0.8112,
+      "step": 398800
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 2.9276804163159103e-05,
+      "loss": 0.8797,
+      "step": 398810
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 2.9275656233259353e-05,
+      "loss": 0.8314,
+      "step": 398820
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 2.9274508303359606e-05,
+      "loss": 0.8885,
+      "step": 398830
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 2.927336037345986e-05,
+      "loss": 0.7909,
+      "step": 398840
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 2.9272212443560113e-05,
+      "loss": 0.8987,
+      "step": 398850
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 2.9271064513660362e-05,
+      "loss": 0.9012,
+      "step": 398860
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 2.926991658376062e-05,
+      "loss": 0.7948,
+      "step": 398870
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 2.926876865386087e-05,
+      "loss": 0.7517,
+      "step": 398880
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 2.926762072396112e-05,
+      "loss": 0.8206,
+      "step": 398890
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 2.9266472794061372e-05,
+      "loss": 0.8226,
+      "step": 398900
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 2.9265324864161625e-05,
+      "loss": 0.776,
+      "step": 398910
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 2.926417693426188e-05,
+      "loss": 0.8573,
+      "step": 398920
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 2.926302900436213e-05,
+      "loss": 0.8727,
+      "step": 398930
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 2.9261881074462385e-05,
+      "loss": 0.891,
+      "step": 398940
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 2.9260733144562635e-05,
+      "loss": 0.8436,
+      "step": 398950
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 2.9259585214662888e-05,
+      "loss": 0.8719,
+      "step": 398960
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 2.925843728476314e-05,
+      "loss": 0.7765,
+      "step": 398970
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 2.9257289354863395e-05,
+      "loss": 0.7482,
+      "step": 398980
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 2.9256141424963644e-05,
+      "loss": 0.8213,
+      "step": 398990
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 2.92549934950639e-05,
+      "loss": 0.8111,
+      "step": 399000
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 2.925384556516415e-05,
+      "loss": 0.7502,
+      "step": 399010
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 2.92526976352644e-05,
+      "loss": 0.8351,
+      "step": 399020
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 2.9251549705364657e-05,
+      "loss": 0.8893,
+      "step": 399030
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 2.9250401775464907e-05,
+      "loss": 0.8414,
+      "step": 399040
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 2.924925384556516e-05,
+      "loss": 0.8574,
+      "step": 399050
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 2.9248105915665414e-05,
+      "loss": 0.9239,
+      "step": 399060
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 2.9246957985765667e-05,
+      "loss": 0.7825,
+      "step": 399070
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 2.9245810055865917e-05,
+      "loss": 0.8609,
+      "step": 399080
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 2.9244662125966173e-05,
+      "loss": 0.8107,
+      "step": 399090
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 2.9243514196066423e-05,
+      "loss": 0.8632,
+      "step": 399100
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 2.9242366266166677e-05,
+      "loss": 0.8768,
+      "step": 399110
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 2.924121833626693e-05,
+      "loss": 0.8628,
+      "step": 399120
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 2.9240070406367183e-05,
+      "loss": 0.7792,
+      "step": 399130
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 2.9238922476467433e-05,
+      "loss": 0.835,
+      "step": 399140
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 2.923777454656769e-05,
+      "loss": 0.8726,
+      "step": 399150
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 2.923662661666794e-05,
+      "loss": 0.8957,
+      "step": 399160
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 2.923547868676819e-05,
+      "loss": 0.8319,
+      "step": 399170
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 2.9234330756868446e-05,
+      "loss": 0.94,
+      "step": 399180
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 2.9233182826968696e-05,
+      "loss": 0.8101,
+      "step": 399190
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 2.923203489706895e-05,
+      "loss": 0.8463,
+      "step": 399200
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 2.9230886967169202e-05,
+      "loss": 0.9094,
+      "step": 399210
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 2.9229739037269455e-05,
+      "loss": 0.8945,
+      "step": 399220
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 2.9228591107369705e-05,
+      "loss": 0.8418,
+      "step": 399230
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 2.9227443177469962e-05,
+      "loss": 0.8054,
+      "step": 399240
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 2.9226295247570212e-05,
+      "loss": 0.8634,
+      "step": 399250
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 2.9225147317670465e-05,
+      "loss": 0.8399,
+      "step": 399260
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 2.9223999387770718e-05,
+      "loss": 0.8343,
+      "step": 399270
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 2.922285145787097e-05,
+      "loss": 0.89,
+      "step": 399280
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 2.922170352797122e-05,
+      "loss": 0.8411,
+      "step": 399290
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 2.9220555598071478e-05,
+      "loss": 0.8527,
+      "step": 399300
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 2.9219407668171728e-05,
+      "loss": 0.891,
+      "step": 399310
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 2.9218259738271978e-05,
+      "loss": 0.7857,
+      "step": 399320
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 2.9217111808372234e-05,
+      "loss": 0.7831,
+      "step": 399330
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 2.9215963878472484e-05,
+      "loss": 0.8337,
+      "step": 399340
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 2.9214815948572737e-05,
+      "loss": 0.809,
+      "step": 399350
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 2.921366801867299e-05,
+      "loss": 0.7886,
+      "step": 399360
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 2.9212520088773244e-05,
+      "loss": 0.8602,
+      "step": 399370
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 2.9211372158873494e-05,
+      "loss": 0.8974,
+      "step": 399380
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 2.921022422897375e-05,
+      "loss": 0.8815,
+      "step": 399390
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 2.9209076299074e-05,
+      "loss": 0.9424,
+      "step": 399400
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 2.9207928369174254e-05,
+      "loss": 0.8064,
+      "step": 399410
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 2.9206780439274507e-05,
+      "loss": 0.8096,
+      "step": 399420
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 2.920563250937476e-05,
+      "loss": 0.8086,
+      "step": 399430
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 2.920448457947501e-05,
+      "loss": 0.8962,
+      "step": 399440
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 2.9203336649575267e-05,
+      "loss": 0.8255,
+      "step": 399450
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 2.9202188719675516e-05,
+      "loss": 0.8763,
+      "step": 399460
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 2.9201040789775766e-05,
+      "loss": 0.8598,
+      "step": 399470
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 2.9199892859876023e-05,
+      "loss": 0.7744,
+      "step": 399480
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 2.9198744929976273e-05,
+      "loss": 0.8393,
+      "step": 399490
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 2.9197597000076526e-05,
+      "loss": 0.8882,
+      "step": 399500
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 2.919644907017678e-05,
+      "loss": 0.8828,
+      "step": 399510
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 2.9195301140277032e-05,
+      "loss": 0.8557,
+      "step": 399520
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 2.9194153210377282e-05,
+      "loss": 0.8414,
+      "step": 399530
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 2.919300528047754e-05,
+      "loss": 0.8974,
+      "step": 399540
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 2.919185735057779e-05,
+      "loss": 0.8677,
+      "step": 399550
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 2.9190709420678042e-05,
+      "loss": 0.8289,
+      "step": 399560
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 2.9189561490778295e-05,
+      "loss": 0.8999,
+      "step": 399570
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 2.918841356087855e-05,
+      "loss": 0.9227,
+      "step": 399580
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 2.91872656309788e-05,
+      "loss": 0.8839,
+      "step": 399590
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 2.9186117701079048e-05,
+      "loss": 0.8415,
+      "step": 399600
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 2.9184969771179305e-05,
+      "loss": 0.8764,
+      "step": 399610
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 2.9183821841279555e-05,
+      "loss": 0.9117,
+      "step": 399620
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 2.9182673911379808e-05,
+      "loss": 0.8054,
+      "step": 399630
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 2.918152598148006e-05,
+      "loss": 0.8241,
+      "step": 399640
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 2.9180378051580314e-05,
+      "loss": 0.8483,
+      "step": 399650
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 2.9179230121680564e-05,
+      "loss": 0.8328,
+      "step": 399660
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 2.917808219178082e-05,
+      "loss": 0.8478,
+      "step": 399670
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 2.917693426188107e-05,
+      "loss": 0.8495,
+      "step": 399680
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 2.9175786331981324e-05,
+      "loss": 0.8268,
+      "step": 399690
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 2.9174638402081577e-05,
+      "loss": 0.8669,
+      "step": 399700
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 2.917349047218183e-05,
+      "loss": 0.8562,
+      "step": 399710
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 2.917234254228208e-05,
+      "loss": 0.9713,
+      "step": 399720
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 2.9171194612382337e-05,
+      "loss": 0.9121,
+      "step": 399730
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 2.9170046682482587e-05,
+      "loss": 0.7876,
+      "step": 399740
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 2.9168898752582837e-05,
+      "loss": 1.0003,
+      "step": 399750
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 2.9167750822683093e-05,
+      "loss": 0.8297,
+      "step": 399760
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 2.9166602892783343e-05,
+      "loss": 0.8675,
+      "step": 399770
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 2.9165454962883596e-05,
+      "loss": 0.7619,
+      "step": 399780
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 2.916430703298385e-05,
+      "loss": 0.8081,
+      "step": 399790
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 2.9163159103084103e-05,
+      "loss": 0.9259,
+      "step": 399800
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 2.9162011173184353e-05,
+      "loss": 0.8283,
+      "step": 399810
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 2.916086324328461e-05,
+      "loss": 0.801,
+      "step": 399820
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 2.915971531338486e-05,
+      "loss": 0.8405,
+      "step": 399830
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 2.9158567383485112e-05,
+      "loss": 0.8329,
+      "step": 399840
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 2.9157419453585366e-05,
+      "loss": 0.8583,
+      "step": 399850
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 2.915627152368562e-05,
+      "loss": 0.804,
+      "step": 399860
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 2.915512359378587e-05,
+      "loss": 0.9075,
+      "step": 399870
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 2.9153975663886125e-05,
+      "loss": 0.8372,
+      "step": 399880
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 2.9152827733986375e-05,
+      "loss": 0.8547,
+      "step": 399890
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 2.9151679804086625e-05,
+      "loss": 0.8137,
+      "step": 399900
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 2.9150531874186882e-05,
+      "loss": 0.7854,
+      "step": 399910
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 2.914938394428713e-05,
+      "loss": 0.8085,
+      "step": 399920
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 2.9148236014387385e-05,
+      "loss": 0.8946,
+      "step": 399930
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 2.9147088084487638e-05,
+      "loss": 0.893,
+      "step": 399940
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 2.914594015458789e-05,
+      "loss": 0.8358,
+      "step": 399950
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 2.914479222468814e-05,
+      "loss": 0.7927,
+      "step": 399960
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 2.9143644294788398e-05,
+      "loss": 0.7973,
+      "step": 399970
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 2.9142496364888648e-05,
+      "loss": 0.9015,
+      "step": 399980
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 2.91413484349889e-05,
+      "loss": 0.8387,
+      "step": 399990
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 2.9140200505089154e-05,
+      "loss": 0.86,
+      "step": 400000
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 2.9139052575189407e-05,
+      "loss": 0.8368,
+      "step": 400010
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 2.9137904645289657e-05,
+      "loss": 0.8336,
+      "step": 400020
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 2.9136756715389914e-05,
+      "loss": 0.7539,
+      "step": 400030
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 2.9135608785490164e-05,
+      "loss": 0.8193,
+      "step": 400040
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 2.9134460855590414e-05,
+      "loss": 0.8229,
+      "step": 400050
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 2.913331292569067e-05,
+      "loss": 0.8624,
+      "step": 400060
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 2.913216499579092e-05,
+      "loss": 0.7606,
+      "step": 400070
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 2.9131017065891173e-05,
+      "loss": 0.8485,
+      "step": 400080
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 2.9129869135991427e-05,
+      "loss": 0.9242,
+      "step": 400090
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 2.912872120609168e-05,
+      "loss": 0.9108,
+      "step": 400100
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 2.912757327619193e-05,
+      "loss": 0.9253,
+      "step": 400110
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 2.9126425346292186e-05,
+      "loss": 0.8335,
+      "step": 400120
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 2.9125277416392436e-05,
+      "loss": 0.7922,
+      "step": 400130
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 2.912412948649269e-05,
+      "loss": 0.8827,
+      "step": 400140
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 2.9122981556592943e-05,
+      "loss": 0.8191,
+      "step": 400150
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 2.9121833626693196e-05,
+      "loss": 0.8215,
+      "step": 400160
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 2.9120685696793446e-05,
+      "loss": 0.8722,
+      "step": 400170
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 2.9119537766893702e-05,
+      "loss": 0.8466,
+      "step": 400180
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 2.9118389836993952e-05,
+      "loss": 0.8352,
+      "step": 400190
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 2.9117241907094202e-05,
+      "loss": 0.896,
+      "step": 400200
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 2.911609397719446e-05,
+      "loss": 0.8717,
+      "step": 400210
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 2.911494604729471e-05,
+      "loss": 0.8035,
+      "step": 400220
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 2.9113798117394962e-05,
+      "loss": 0.8528,
+      "step": 400230
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 2.9112650187495215e-05,
+      "loss": 0.9058,
+      "step": 400240
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 2.9111502257595468e-05,
+      "loss": 0.8302,
+      "step": 400250
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 2.9110354327695718e-05,
+      "loss": 0.844,
+      "step": 400260
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 2.910920639779597e-05,
+      "loss": 0.7853,
+      "step": 400270
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 2.9108058467896225e-05,
+      "loss": 0.8536,
+      "step": 400280
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 2.9106910537996478e-05,
+      "loss": 0.8498,
+      "step": 400290
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 2.9105762608096728e-05,
+      "loss": 0.9143,
+      "step": 400300
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 2.9104614678196984e-05,
+      "loss": 0.7444,
+      "step": 400310
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 2.9103466748297234e-05,
+      "loss": 0.8394,
+      "step": 400320
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 2.9102318818397484e-05,
+      "loss": 0.8145,
+      "step": 400330
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 2.910117088849774e-05,
+      "loss": 0.8912,
+      "step": 400340
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 2.910002295859799e-05,
+      "loss": 0.902,
+      "step": 400350
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 2.9098875028698244e-05,
+      "loss": 0.8551,
+      "step": 400360
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 2.9097727098798497e-05,
+      "loss": 0.767,
+      "step": 400370
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 2.909657916889875e-05,
+      "loss": 0.8714,
+      "step": 400380
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 2.9095431238999e-05,
+      "loss": 0.8369,
+      "step": 400390
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 2.9094283309099257e-05,
+      "loss": 0.8551,
+      "step": 400400
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 2.9093135379199507e-05,
+      "loss": 0.9621,
+      "step": 400410
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 2.909198744929976e-05,
+      "loss": 0.8623,
+      "step": 400420
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 2.9090839519400013e-05,
+      "loss": 0.8101,
+      "step": 400430
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 2.9089691589500266e-05,
+      "loss": 0.9828,
+      "step": 400440
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 2.9088543659600516e-05,
+      "loss": 0.8067,
+      "step": 400450
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 2.9087395729700773e-05,
+      "loss": 0.8091,
+      "step": 400460
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 2.9086247799801023e-05,
+      "loss": 0.8735,
+      "step": 400470
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 2.9085099869901273e-05,
+      "loss": 0.7827,
+      "step": 400480
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 2.908395194000153e-05,
+      "loss": 0.9952,
+      "step": 400490
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 2.908280401010178e-05,
+      "loss": 0.9031,
+      "step": 400500
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 2.9081656080202032e-05,
+      "loss": 0.8506,
+      "step": 400510
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 2.9080508150302286e-05,
+      "loss": 0.9408,
+      "step": 400520
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 2.907936022040254e-05,
+      "loss": 0.8307,
+      "step": 400530
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 2.907821229050279e-05,
+      "loss": 0.8532,
+      "step": 400540
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 2.9077064360603045e-05,
+      "loss": 0.7694,
+      "step": 400550
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 2.9075916430703295e-05,
+      "loss": 0.8452,
+      "step": 400560
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 2.907476850080355e-05,
+      "loss": 0.8637,
+      "step": 400570
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 2.90736205709038e-05,
+      "loss": 0.9591,
+      "step": 400580
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 2.9072472641004055e-05,
+      "loss": 0.8566,
+      "step": 400590
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 2.9071324711104305e-05,
+      "loss": 0.8857,
+      "step": 400600
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 2.907017678120456e-05,
+      "loss": 0.9596,
+      "step": 400610
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 2.906902885130481e-05,
+      "loss": 0.8847,
+      "step": 400620
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 2.906788092140506e-05,
+      "loss": 0.8585,
+      "step": 400630
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 2.9066732991505318e-05,
+      "loss": 0.8502,
+      "step": 400640
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 2.9065585061605568e-05,
+      "loss": 0.8597,
+      "step": 400650
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 2.906443713170582e-05,
+      "loss": 0.91,
+      "step": 400660
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 2.9063289201806074e-05,
+      "loss": 0.9316,
+      "step": 400670
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 2.9062141271906327e-05,
+      "loss": 0.897,
+      "step": 400680
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 2.9060993342006577e-05,
+      "loss": 0.8996,
+      "step": 400690
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 2.9059845412106834e-05,
+      "loss": 0.7934,
+      "step": 400700
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 2.9058697482207084e-05,
+      "loss": 0.7699,
+      "step": 400710
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 2.9057549552307337e-05,
+      "loss": 0.8495,
+      "step": 400720
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 2.905640162240759e-05,
+      "loss": 0.7454,
+      "step": 400730
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 2.9055253692507843e-05,
+      "loss": 0.8424,
+      "step": 400740
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 2.9054105762608093e-05,
+      "loss": 0.9107,
+      "step": 400750
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 2.905295783270835e-05,
+      "loss": 0.9192,
+      "step": 400760
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 2.90518099028086e-05,
+      "loss": 0.9264,
+      "step": 400770
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 2.905066197290885e-05,
+      "loss": 0.915,
+      "step": 400780
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 2.9049514043009106e-05,
+      "loss": 0.8007,
+      "step": 400790
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 2.9048366113109356e-05,
+      "loss": 0.7688,
+      "step": 400800
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 2.904721818320961e-05,
+      "loss": 0.8534,
+      "step": 400810
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 2.9046070253309862e-05,
+      "loss": 0.8651,
+      "step": 400820
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 2.9044922323410116e-05,
+      "loss": 0.8277,
+      "step": 400830
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 2.9043774393510366e-05,
+      "loss": 0.8109,
+      "step": 400840
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 2.9042626463610622e-05,
+      "loss": 0.8021,
+      "step": 400850
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 2.9041478533710872e-05,
+      "loss": 0.8823,
+      "step": 400860
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 2.9040330603811125e-05,
+      "loss": 0.8073,
+      "step": 400870
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 2.903918267391138e-05,
+      "loss": 0.9141,
+      "step": 400880
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 2.9038034744011632e-05,
+      "loss": 0.8583,
+      "step": 400890
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 2.903688681411188e-05,
+      "loss": 0.8584,
+      "step": 400900
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 2.9035738884212138e-05,
+      "loss": 0.8937,
+      "step": 400910
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 2.9034590954312388e-05,
+      "loss": 0.8032,
+      "step": 400920
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 2.9033443024412638e-05,
+      "loss": 0.8832,
+      "step": 400930
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 2.9032295094512895e-05,
+      "loss": 0.8105,
+      "step": 400940
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 2.9031147164613144e-05,
+      "loss": 0.8191,
+      "step": 400950
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 2.9029999234713398e-05,
+      "loss": 0.9146,
+      "step": 400960
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 2.9028851304813648e-05,
+      "loss": 0.8774,
+      "step": 400970
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 2.9027703374913904e-05,
+      "loss": 0.9187,
+      "step": 400980
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 2.9026555445014154e-05,
+      "loss": 0.7659,
+      "step": 400990
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 2.9025407515114407e-05,
+      "loss": 0.8633,
+      "step": 401000
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 2.902425958521466e-05,
+      "loss": 0.8437,
+      "step": 401010
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 2.9023111655314914e-05,
+      "loss": 0.8124,
+      "step": 401020
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 2.9021963725415164e-05,
+      "loss": 0.8944,
+      "step": 401030
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 2.902081579551542e-05,
+      "loss": 0.8499,
+      "step": 401040
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 2.901966786561567e-05,
+      "loss": 0.8868,
+      "step": 401050
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 2.901851993571592e-05,
+      "loss": 0.862,
+      "step": 401060
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 2.9017372005816177e-05,
+      "loss": 0.9554,
+      "step": 401070
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 2.9016224075916426e-05,
+      "loss": 0.8694,
+      "step": 401080
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 2.901507614601668e-05,
+      "loss": 0.841,
+      "step": 401090
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 2.9013928216116933e-05,
+      "loss": 0.7828,
+      "step": 401100
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 2.9012780286217186e-05,
+      "loss": 0.8419,
+      "step": 401110
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 2.9011632356317436e-05,
+      "loss": 0.8924,
+      "step": 401120
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 2.9010484426417693e-05,
+      "loss": 0.8099,
+      "step": 401130
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 2.9009336496517942e-05,
+      "loss": 0.6861,
+      "step": 401140
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 2.9008188566618196e-05,
+      "loss": 0.8719,
+      "step": 401150
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 2.900704063671845e-05,
+      "loss": 0.863,
+      "step": 401160
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 2.9005892706818702e-05,
+      "loss": 0.8269,
+      "step": 401170
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 2.9004744776918952e-05,
+      "loss": 0.8192,
+      "step": 401180
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 2.900359684701921e-05,
+      "loss": 0.8253,
+      "step": 401190
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 2.900244891711946e-05,
+      "loss": 0.8126,
+      "step": 401200
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 2.900130098721971e-05,
+      "loss": 0.8081,
+      "step": 401210
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 2.9000153057319965e-05,
+      "loss": 0.8466,
+      "step": 401220
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 2.8999005127420215e-05,
+      "loss": 0.7128,
+      "step": 401230
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 2.8997857197520468e-05,
+      "loss": 0.9887,
+      "step": 401240
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 2.899670926762072e-05,
+      "loss": 0.8532,
+      "step": 401250
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 2.8995561337720975e-05,
+      "loss": 0.895,
+      "step": 401260
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 2.8994413407821224e-05,
+      "loss": 0.892,
+      "step": 401270
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 2.899326547792148e-05,
+      "loss": 0.8734,
+      "step": 401280
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 2.899211754802173e-05,
+      "loss": 0.864,
+      "step": 401290
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 2.8990969618121984e-05,
+      "loss": 0.8379,
+      "step": 401300
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 2.8989821688222237e-05,
+      "loss": 0.9231,
+      "step": 401310
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 2.898867375832249e-05,
+      "loss": 0.8593,
+      "step": 401320
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 2.898752582842274e-05,
+      "loss": 0.9292,
+      "step": 401330
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 2.8986377898522997e-05,
+      "loss": 0.9107,
+      "step": 401340
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 2.8985229968623247e-05,
+      "loss": 0.8289,
+      "step": 401350
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 2.8984082038723497e-05,
+      "loss": 0.8262,
+      "step": 401360
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 2.8982934108823754e-05,
+      "loss": 0.7696,
+      "step": 401370
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 2.8981786178924003e-05,
+      "loss": 0.8783,
+      "step": 401380
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 2.8980638249024257e-05,
+      "loss": 0.8759,
+      "step": 401390
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 2.897949031912451e-05,
+      "loss": 0.8957,
+      "step": 401400
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 2.8978342389224763e-05,
+      "loss": 0.8909,
+      "step": 401410
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 2.8977194459325013e-05,
+      "loss": 0.7942,
+      "step": 401420
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 2.897604652942527e-05,
+      "loss": 0.8655,
+      "step": 401430
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 2.897489859952552e-05,
+      "loss": 0.9343,
+      "step": 401440
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 2.8973750669625773e-05,
+      "loss": 0.8671,
+      "step": 401450
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 2.8972602739726026e-05,
+      "loss": 0.9207,
+      "step": 401460
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 2.897145480982628e-05,
+      "loss": 0.8358,
+      "step": 401470
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 2.897030687992653e-05,
+      "loss": 0.8265,
+      "step": 401480
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 2.8969158950026786e-05,
+      "loss": 0.8796,
+      "step": 401490
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 2.8968011020127036e-05,
+      "loss": 0.8329,
+      "step": 401500
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 2.8966863090227285e-05,
+      "loss": 0.8207,
+      "step": 401510
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 2.8965715160327542e-05,
+      "loss": 0.8543,
+      "step": 401520
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 2.8964567230427792e-05,
+      "loss": 0.8533,
+      "step": 401530
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 2.8963419300528045e-05,
+      "loss": 0.8376,
+      "step": 401540
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 2.89622713706283e-05,
+      "loss": 0.8594,
+      "step": 401550
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 2.896112344072855e-05,
+      "loss": 0.8613,
+      "step": 401560
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 2.89599755108288e-05,
+      "loss": 0.844,
+      "step": 401570
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 2.8958827580929058e-05,
+      "loss": 0.7774,
+      "step": 401580
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 2.8957679651029308e-05,
+      "loss": 0.8534,
+      "step": 401590
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 2.895653172112956e-05,
+      "loss": 0.8314,
+      "step": 401600
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 2.8955383791229814e-05,
+      "loss": 0.8488,
+      "step": 401610
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 2.8954235861330068e-05,
+      "loss": 0.8496,
+      "step": 401620
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 2.8953087931430317e-05,
+      "loss": 0.8222,
+      "step": 401630
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 2.8951940001530567e-05,
+      "loss": 0.8785,
+      "step": 401640
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 2.8950792071630824e-05,
+      "loss": 0.8358,
+      "step": 401650
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 2.8949644141731074e-05,
+      "loss": 0.8493,
+      "step": 401660
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 2.8948496211831327e-05,
+      "loss": 0.8652,
+      "step": 401670
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 2.894734828193158e-05,
+      "loss": 0.8573,
+      "step": 401680
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 2.8946200352031834e-05,
+      "loss": 0.8341,
+      "step": 401690
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 2.8945052422132083e-05,
+      "loss": 0.8374,
+      "step": 401700
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 2.894390449223234e-05,
+      "loss": 0.9207,
+      "step": 401710
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 2.894275656233259e-05,
+      "loss": 0.8704,
+      "step": 401720
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 2.8941608632432843e-05,
+      "loss": 0.7866,
+      "step": 401730
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 2.8940460702533096e-05,
+      "loss": 0.7331,
+      "step": 401740
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 2.893931277263335e-05,
+      "loss": 0.8955,
+      "step": 401750
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 2.89381648427336e-05,
+      "loss": 0.8133,
+      "step": 401760
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 2.8937016912833856e-05,
+      "loss": 0.844,
+      "step": 401770
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 2.8935868982934106e-05,
+      "loss": 0.8507,
+      "step": 401780
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 2.8934721053034356e-05,
+      "loss": 0.9007,
+      "step": 401790
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 2.8933573123134612e-05,
+      "loss": 0.8701,
+      "step": 401800
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 2.8932425193234862e-05,
+      "loss": 0.9988,
+      "step": 401810
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 2.8931277263335116e-05,
+      "loss": 0.8705,
+      "step": 401820
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 2.893012933343537e-05,
+      "loss": 0.8886,
+      "step": 401830
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 2.8928981403535622e-05,
+      "loss": 0.9419,
+      "step": 401840
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 2.8927833473635872e-05,
+      "loss": 0.8631,
+      "step": 401850
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 2.892668554373613e-05,
+      "loss": 0.8584,
+      "step": 401860
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 2.892553761383638e-05,
+      "loss": 0.8449,
+      "step": 401870
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 2.892438968393663e-05,
+      "loss": 0.8248,
+      "step": 401880
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 2.8923241754036885e-05,
+      "loss": 0.9384,
+      "step": 401890
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 2.8922093824137138e-05,
+      "loss": 0.8706,
+      "step": 401900
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 2.8920945894237388e-05,
+      "loss": 0.9542,
+      "step": 401910
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 2.8919797964337645e-05,
+      "loss": 0.8623,
+      "step": 401920
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 2.8918650034437894e-05,
+      "loss": 0.8594,
+      "step": 401930
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 2.8917502104538144e-05,
+      "loss": 0.881,
+      "step": 401940
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 2.89163541746384e-05,
+      "loss": 0.897,
+      "step": 401950
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 2.891520624473865e-05,
+      "loss": 0.771,
+      "step": 401960
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 2.8914058314838904e-05,
+      "loss": 0.8859,
+      "step": 401970
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 2.8912910384939157e-05,
+      "loss": 0.8899,
+      "step": 401980
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 2.891176245503941e-05,
+      "loss": 0.8092,
+      "step": 401990
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 2.891061452513966e-05,
+      "loss": 0.7507,
+      "step": 402000
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 2.8909466595239917e-05,
+      "loss": 0.74,
+      "step": 402010
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 2.8908318665340167e-05,
+      "loss": 0.8061,
+      "step": 402020
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 2.890717073544042e-05,
+      "loss": 0.8164,
+      "step": 402030
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 2.8906022805540673e-05,
+      "loss": 0.9576,
+      "step": 402040
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 2.8904874875640927e-05,
+      "loss": 0.7736,
+      "step": 402050
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 2.8903726945741176e-05,
+      "loss": 0.8677,
+      "step": 402060
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 2.8902579015841433e-05,
+      "loss": 0.8571,
+      "step": 402070
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 2.8901431085941683e-05,
+      "loss": 0.8196,
+      "step": 402080
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 2.8900283156041933e-05,
+      "loss": 0.7947,
+      "step": 402090
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 2.889913522614219e-05,
+      "loss": 0.818,
+      "step": 402100
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 2.889798729624244e-05,
+      "loss": 0.8667,
+      "step": 402110
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 2.8896839366342692e-05,
+      "loss": 0.7799,
+      "step": 402120
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 2.8895691436442946e-05,
+      "loss": 0.9057,
+      "step": 402130
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 2.88945435065432e-05,
+      "loss": 0.7291,
+      "step": 402140
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 2.889339557664345e-05,
+      "loss": 0.8329,
+      "step": 402150
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 2.8892247646743705e-05,
+      "loss": 0.8181,
+      "step": 402160
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 2.8891099716843955e-05,
+      "loss": 0.87,
+      "step": 402170
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 2.888995178694421e-05,
+      "loss": 0.8038,
+      "step": 402180
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 2.8888803857044462e-05,
+      "loss": 0.8867,
+      "step": 402190
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 2.8887655927144715e-05,
+      "loss": 0.7677,
+      "step": 402200
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 2.8886507997244965e-05,
+      "loss": 0.8926,
+      "step": 402210
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 2.888536006734522e-05,
+      "loss": 0.8952,
+      "step": 402220
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 2.888421213744547e-05,
+      "loss": 0.937,
+      "step": 402230
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 2.888306420754572e-05,
+      "loss": 0.8031,
+      "step": 402240
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 2.8881916277645978e-05,
+      "loss": 0.8475,
+      "step": 402250
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 2.8880768347746228e-05,
+      "loss": 0.8996,
+      "step": 402260
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 2.887962041784648e-05,
+      "loss": 0.8707,
+      "step": 402270
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 2.8878472487946734e-05,
+      "loss": 0.7957,
+      "step": 402280
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 2.8877324558046987e-05,
+      "loss": 0.8793,
+      "step": 402290
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 2.8876176628147237e-05,
+      "loss": 0.8713,
+      "step": 402300
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 2.8875028698247494e-05,
+      "loss": 0.834,
+      "step": 402310
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 2.8873880768347744e-05,
+      "loss": 0.8617,
+      "step": 402320
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 2.8872732838447997e-05,
+      "loss": 0.9393,
+      "step": 402330
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 2.8871584908548247e-05,
+      "loss": 0.8659,
+      "step": 402340
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 2.8870436978648504e-05,
+      "loss": 0.8699,
+      "step": 402350
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 2.8869289048748753e-05,
+      "loss": 0.8562,
+      "step": 402360
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 2.8868141118849003e-05,
+      "loss": 0.8075,
+      "step": 402370
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 2.886699318894926e-05,
+      "loss": 0.8547,
+      "step": 402380
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 2.886584525904951e-05,
+      "loss": 0.8406,
+      "step": 402390
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 2.8864697329149763e-05,
+      "loss": 0.8214,
+      "step": 402400
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 2.8863549399250016e-05,
+      "loss": 0.9107,
+      "step": 402410
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 2.886240146935027e-05,
+      "loss": 0.8877,
+      "step": 402420
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 2.886125353945052e-05,
+      "loss": 0.9149,
+      "step": 402430
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 2.8860105609550776e-05,
+      "loss": 0.8516,
+      "step": 402440
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 2.8858957679651026e-05,
+      "loss": 0.8026,
+      "step": 402450
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 2.885780974975128e-05,
+      "loss": 0.8532,
+      "step": 402460
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 2.8856661819851532e-05,
+      "loss": 0.8612,
+      "step": 402470
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 2.8855513889951786e-05,
+      "loss": 0.7482,
+      "step": 402480
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 2.8854365960052035e-05,
+      "loss": 0.8571,
+      "step": 402490
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 2.8853218030152292e-05,
+      "loss": 0.9421,
+      "step": 402500
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 2.8852070100252542e-05,
+      "loss": 0.8219,
+      "step": 402510
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 2.8850922170352792e-05,
+      "loss": 0.8523,
+      "step": 402520
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 2.884977424045305e-05,
+      "loss": 0.8644,
+      "step": 402530
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 2.8848626310553298e-05,
+      "loss": 0.8032,
+      "step": 402540
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 2.884747838065355e-05,
+      "loss": 0.8327,
+      "step": 402550
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 2.8846330450753805e-05,
+      "loss": 0.7811,
+      "step": 402560
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 2.8845182520854058e-05,
+      "loss": 0.8333,
+      "step": 402570
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 2.8844034590954308e-05,
+      "loss": 0.7797,
+      "step": 402580
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 2.8842886661054564e-05,
+      "loss": 0.8915,
+      "step": 402590
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 2.8841738731154814e-05,
+      "loss": 0.8344,
+      "step": 402600
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 2.8840590801255067e-05,
+      "loss": 0.7615,
+      "step": 402610
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 2.883944287135532e-05,
+      "loss": 0.9478,
+      "step": 402620
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 2.8838294941455574e-05,
+      "loss": 0.768,
+      "step": 402630
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 2.8837147011555824e-05,
+      "loss": 0.8616,
+      "step": 402640
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 2.883599908165608e-05,
+      "loss": 0.8249,
+      "step": 402650
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 2.883485115175633e-05,
+      "loss": 0.8464,
+      "step": 402660
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 2.883370322185658e-05,
+      "loss": 0.7806,
+      "step": 402670
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 2.8832555291956837e-05,
+      "loss": 0.7511,
+      "step": 402680
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 2.8831407362057087e-05,
+      "loss": 0.8087,
+      "step": 402690
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 2.883025943215734e-05,
+      "loss": 0.9469,
+      "step": 402700
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 2.8829111502257593e-05,
+      "loss": 0.8357,
+      "step": 402710
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 2.8827963572357846e-05,
+      "loss": 0.8414,
+      "step": 402720
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 2.8826815642458096e-05,
+      "loss": 0.8147,
+      "step": 402730
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 2.8825667712558353e-05,
+      "loss": 0.9097,
+      "step": 402740
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 2.8824519782658603e-05,
+      "loss": 0.8741,
+      "step": 402750
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 2.8823371852758856e-05,
+      "loss": 0.7756,
+      "step": 402760
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 2.882222392285911e-05,
+      "loss": 0.8506,
+      "step": 402770
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 2.8821075992959362e-05,
+      "loss": 0.8916,
+      "step": 402780
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 2.8819928063059612e-05,
+      "loss": 0.854,
+      "step": 402790
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 2.881878013315987e-05,
+      "loss": 0.8488,
+      "step": 402800
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 2.881763220326012e-05,
+      "loss": 0.878,
+      "step": 402810
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 2.881648427336037e-05,
+      "loss": 0.7901,
+      "step": 402820
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 2.8815336343460625e-05,
+      "loss": 0.8435,
+      "step": 402830
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 2.8814188413560875e-05,
+      "loss": 0.8437,
+      "step": 402840
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 2.881304048366113e-05,
+      "loss": 0.8465,
+      "step": 402850
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 2.881189255376138e-05,
+      "loss": 0.857,
+      "step": 402860
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 2.8810744623861635e-05,
+      "loss": 0.8228,
+      "step": 402870
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 2.8809596693961885e-05,
+      "loss": 0.7687,
+      "step": 402880
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 2.880844876406214e-05,
+      "loss": 0.8164,
+      "step": 402890
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 2.880730083416239e-05,
+      "loss": 0.8278,
+      "step": 402900
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 2.8806152904262644e-05,
+      "loss": 0.8221,
+      "step": 402910
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 2.8805004974362898e-05,
+      "loss": 0.9003,
+      "step": 402920
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 2.880385704446315e-05,
+      "loss": 0.8593,
+      "step": 402930
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 2.88027091145634e-05,
+      "loss": 0.8652,
+      "step": 402940
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 2.8801561184663657e-05,
+      "loss": 0.7869,
+      "step": 402950
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 2.8800413254763907e-05,
+      "loss": 0.802,
+      "step": 402960
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 2.8799265324864157e-05,
+      "loss": 0.8298,
+      "step": 402970
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 2.8798117394964414e-05,
+      "loss": 0.8324,
+      "step": 402980
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 2.8796969465064664e-05,
+      "loss": 0.8581,
+      "step": 402990
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 2.8795821535164917e-05,
+      "loss": 0.908,
+      "step": 403000
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 2.879467360526517e-05,
+      "loss": 0.7869,
+      "step": 403010
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 2.8793525675365423e-05,
+      "loss": 0.8993,
+      "step": 403020
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 2.8792377745465673e-05,
+      "loss": 0.8425,
+      "step": 403030
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 2.8791229815565926e-05,
+      "loss": 0.8531,
+      "step": 403040
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 2.879008188566618e-05,
+      "loss": 0.9408,
+      "step": 403050
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 2.8788933955766433e-05,
+      "loss": 0.844,
+      "step": 403060
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 2.8787786025866683e-05,
+      "loss": 0.7365,
+      "step": 403070
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 2.878663809596694e-05,
+      "loss": 0.9032,
+      "step": 403080
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 2.878549016606719e-05,
+      "loss": 0.8995,
+      "step": 403090
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 2.878434223616744e-05,
+      "loss": 0.8086,
+      "step": 403100
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 2.8783194306267696e-05,
+      "loss": 0.9066,
+      "step": 403110
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 2.8782046376367946e-05,
+      "loss": 0.7973,
+      "step": 403120
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 2.87808984464682e-05,
+      "loss": 0.9061,
+      "step": 403130
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 2.8779750516568452e-05,
+      "loss": 0.829,
+      "step": 403140
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 2.8778602586668705e-05,
+      "loss": 0.8991,
+      "step": 403150
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 2.8777454656768955e-05,
+      "loss": 0.907,
+      "step": 403160
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 2.8776306726869212e-05,
+      "loss": 0.841,
+      "step": 403170
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 2.877515879696946e-05,
+      "loss": 0.8914,
+      "step": 403180
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 2.8774010867069715e-05,
+      "loss": 0.7811,
+      "step": 403190
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 2.8772862937169968e-05,
+      "loss": 0.8218,
+      "step": 403200
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 2.877171500727022e-05,
+      "loss": 0.9237,
+      "step": 403210
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 2.877056707737047e-05,
+      "loss": 0.8889,
+      "step": 403220
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 2.8769419147470728e-05,
+      "loss": 0.8676,
+      "step": 403230
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 2.8768271217570978e-05,
+      "loss": 0.801,
+      "step": 403240
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 2.8767123287671228e-05,
+      "loss": 0.8289,
+      "step": 403250
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 2.8765975357771484e-05,
+      "loss": 0.7686,
+      "step": 403260
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 2.8764827427871734e-05,
+      "loss": 0.7983,
+      "step": 403270
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 2.8763679497971987e-05,
+      "loss": 0.7836,
+      "step": 403280
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 2.876253156807224e-05,
+      "loss": 0.8847,
+      "step": 403290
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 2.8761383638172494e-05,
+      "loss": 0.8371,
+      "step": 403300
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 2.8760235708272744e-05,
+      "loss": 0.8274,
+      "step": 403310
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 2.8759087778373e-05,
+      "loss": 0.7846,
+      "step": 403320
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 2.875793984847325e-05,
+      "loss": 0.9295,
+      "step": 403330
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 2.8756791918573503e-05,
+      "loss": 0.867,
+      "step": 403340
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 2.8755643988673757e-05,
+      "loss": 0.8706,
+      "step": 403350
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 2.875449605877401e-05,
+      "loss": 0.869,
+      "step": 403360
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 2.875334812887426e-05,
+      "loss": 0.7754,
+      "step": 403370
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 2.8752200198974516e-05,
+      "loss": 0.7875,
+      "step": 403380
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 2.8751052269074766e-05,
+      "loss": 0.9158,
+      "step": 403390
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 2.8749904339175016e-05,
+      "loss": 0.8636,
+      "step": 403400
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 2.8748756409275273e-05,
+      "loss": 0.8749,
+      "step": 403410
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 2.8747608479375523e-05,
+      "loss": 0.7855,
+      "step": 403420
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 2.8746460549475776e-05,
+      "loss": 0.9388,
+      "step": 403430
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 2.874531261957603e-05,
+      "loss": 0.8397,
+      "step": 403440
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 2.8744164689676282e-05,
+      "loss": 0.8011,
+      "step": 403450
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 2.8743016759776532e-05,
+      "loss": 0.8911,
+      "step": 403460
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 2.874186882987679e-05,
+      "loss": 0.847,
+      "step": 403470
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 2.874072089997704e-05,
+      "loss": 0.835,
+      "step": 403480
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 2.8739572970077292e-05,
+      "loss": 0.8317,
+      "step": 403490
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 2.8738425040177545e-05,
+      "loss": 0.8072,
+      "step": 403500
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 2.87372771102778e-05,
+      "loss": 0.8795,
+      "step": 403510
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 2.8736129180378048e-05,
+      "loss": 0.8756,
+      "step": 403520
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 2.8734981250478305e-05,
+      "loss": 0.8447,
+      "step": 403530
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 2.8733833320578555e-05,
+      "loss": 0.8265,
+      "step": 403540
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 2.8732685390678805e-05,
+      "loss": 0.7774,
+      "step": 403550
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 2.873153746077906e-05,
+      "loss": 0.841,
+      "step": 403560
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 2.873038953087931e-05,
+      "loss": 0.8369,
+      "step": 403570
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 2.8729241600979564e-05,
+      "loss": 0.8553,
+      "step": 403580
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 2.8728093671079817e-05,
+      "loss": 0.8377,
+      "step": 403590
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 2.872694574118007e-05,
+      "loss": 0.8443,
+      "step": 403600
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 2.872579781128032e-05,
+      "loss": 0.8335,
+      "step": 403610
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 2.8724649881380577e-05,
+      "loss": 0.9773,
+      "step": 403620
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 2.8723501951480827e-05,
+      "loss": 0.8157,
+      "step": 403630
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 2.872235402158108e-05,
+      "loss": 0.7391,
+      "step": 403640
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 2.8721206091681334e-05,
+      "loss": 0.914,
+      "step": 403650
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 2.8720058161781587e-05,
+      "loss": 0.8126,
+      "step": 403660
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 2.8718910231881837e-05,
+      "loss": 0.8338,
+      "step": 403670
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 2.8717762301982093e-05,
+      "loss": 0.8172,
+      "step": 403680
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 2.8716614372082343e-05,
+      "loss": 0.8576,
+      "step": 403690
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 2.8715466442182593e-05,
+      "loss": 0.8967,
+      "step": 403700
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 2.8714318512282846e-05,
+      "loss": 0.9292,
+      "step": 403710
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 2.87131705823831e-05,
+      "loss": 0.9406,
+      "step": 403720
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 2.8712022652483353e-05,
+      "loss": 0.8917,
+      "step": 403730
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 2.8710874722583603e-05,
+      "loss": 0.8211,
+      "step": 403740
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 2.870972679268386e-05,
+      "loss": 0.8211,
+      "step": 403750
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 2.870857886278411e-05,
+      "loss": 0.8022,
+      "step": 403760
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 2.8707430932884362e-05,
+      "loss": 0.825,
+      "step": 403770
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 2.8706283002984616e-05,
+      "loss": 0.8799,
+      "step": 403780
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 2.870513507308487e-05,
+      "loss": 0.8223,
+      "step": 403790
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 2.870398714318512e-05,
+      "loss": 0.8413,
+      "step": 403800
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 2.8702839213285375e-05,
+      "loss": 0.7942,
+      "step": 403810
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 2.8701691283385625e-05,
+      "loss": 0.6992,
+      "step": 403820
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 2.8700543353485875e-05,
+      "loss": 0.882,
+      "step": 403830
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 2.869939542358613e-05,
+      "loss": 0.9029,
+      "step": 403840
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 2.869824749368638e-05,
+      "loss": 0.85,
+      "step": 403850
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 2.8697099563786635e-05,
+      "loss": 0.8451,
+      "step": 403860
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 2.8695951633886888e-05,
+      "loss": 0.8678,
+      "step": 403870
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 2.869480370398714e-05,
+      "loss": 0.8535,
+      "step": 403880
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 2.869365577408739e-05,
+      "loss": 0.8008,
+      "step": 403890
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 2.8692507844187648e-05,
+      "loss": 0.8958,
+      "step": 403900
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 2.8691359914287898e-05,
+      "loss": 0.7821,
+      "step": 403910
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 2.869021198438815e-05,
+      "loss": 0.8782,
+      "step": 403920
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 2.8689064054488404e-05,
+      "loss": 0.9101,
+      "step": 403930
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 2.8687916124588657e-05,
+      "loss": 0.8075,
+      "step": 403940
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 2.8686768194688907e-05,
+      "loss": 0.8369,
+      "step": 403950
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 2.8685620264789164e-05,
+      "loss": 0.9817,
+      "step": 403960
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 2.8684472334889414e-05,
+      "loss": 0.7919,
+      "step": 403970
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 2.8683324404989663e-05,
+      "loss": 0.786,
+      "step": 403980
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 2.868217647508992e-05,
+      "loss": 0.8171,
+      "step": 403990
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 2.868102854519017e-05,
+      "loss": 0.9385,
+      "step": 404000
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 2.8679880615290423e-05,
+      "loss": 0.8125,
+      "step": 404010
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 2.8678732685390676e-05,
+      "loss": 0.8454,
+      "step": 404020
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 2.867758475549093e-05,
+      "loss": 1.0256,
+      "step": 404030
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 2.867643682559118e-05,
+      "loss": 0.8103,
+      "step": 404040
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 2.8675288895691436e-05,
+      "loss": 0.8351,
+      "step": 404050
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 2.8674140965791686e-05,
+      "loss": 0.8277,
+      "step": 404060
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 2.867299303589194e-05,
+      "loss": 0.7918,
+      "step": 404070
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 2.8671845105992192e-05,
+      "loss": 0.8023,
+      "step": 404080
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 2.8670697176092446e-05,
+      "loss": 0.9066,
+      "step": 404090
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 2.8669549246192696e-05,
+      "loss": 0.8727,
+      "step": 404100
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 2.8668401316292952e-05,
+      "loss": 0.8359,
+      "step": 404110
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 2.8667253386393202e-05,
+      "loss": 0.7856,
+      "step": 404120
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 2.8666105456493452e-05,
+      "loss": 0.8966,
+      "step": 404130
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 2.866495752659371e-05,
+      "loss": 0.8324,
+      "step": 404140
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 2.866380959669396e-05,
+      "loss": 0.9169,
+      "step": 404150
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 2.866266166679421e-05,
+      "loss": 0.7642,
+      "step": 404160
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 2.8661513736894465e-05,
+      "loss": 0.7884,
+      "step": 404170
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 2.8660365806994718e-05,
+      "loss": 0.7319,
+      "step": 404180
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 2.8659217877094968e-05,
+      "loss": 0.8611,
+      "step": 404190
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 2.8658069947195225e-05,
+      "loss": 0.8964,
+      "step": 404200
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 2.8656922017295474e-05,
+      "loss": 0.9295,
+      "step": 404210
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 2.8655774087395728e-05,
+      "loss": 0.9036,
+      "step": 404220
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 2.865462615749598e-05,
+      "loss": 0.8849,
+      "step": 404230
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 2.8653478227596234e-05,
+      "loss": 0.8748,
+      "step": 404240
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 2.8652330297696484e-05,
+      "loss": 0.8831,
+      "step": 404250
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 2.865118236779674e-05,
+      "loss": 0.7947,
+      "step": 404260
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 2.865003443789699e-05,
+      "loss": 0.7664,
+      "step": 404270
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 2.864888650799724e-05,
+      "loss": 0.8105,
+      "step": 404280
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 2.8647738578097497e-05,
+      "loss": 0.8803,
+      "step": 404290
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 2.8646590648197747e-05,
+      "loss": 0.9481,
+      "step": 404300
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 2.8645442718298e-05,
+      "loss": 0.9076,
+      "step": 404310
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 2.8644294788398253e-05,
+      "loss": 0.8578,
+      "step": 404320
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 2.8643146858498507e-05,
+      "loss": 0.9113,
+      "step": 404330
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 2.8641998928598756e-05,
+      "loss": 0.8449,
+      "step": 404340
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 2.8640850998699013e-05,
+      "loss": 0.8318,
+      "step": 404350
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 2.8639703068799263e-05,
+      "loss": 0.8268,
+      "step": 404360
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 2.8638555138899516e-05,
+      "loss": 0.8056,
+      "step": 404370
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 2.863740720899977e-05,
+      "loss": 0.8398,
+      "step": 404380
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 2.8636259279100023e-05,
+      "loss": 0.9312,
+      "step": 404390
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 2.8635111349200273e-05,
+      "loss": 0.8046,
+      "step": 404400
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 2.8633963419300522e-05,
+      "loss": 0.8857,
+      "step": 404410
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 2.863281548940078e-05,
+      "loss": 0.9115,
+      "step": 404420
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 2.863166755950103e-05,
+      "loss": 0.8081,
+      "step": 404430
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 2.8630519629601282e-05,
+      "loss": 0.8238,
+      "step": 404440
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 2.8629371699701535e-05,
+      "loss": 0.7349,
+      "step": 404450
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 2.862822376980179e-05,
+      "loss": 0.8083,
+      "step": 404460
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 2.862707583990204e-05,
+      "loss": 0.7919,
+      "step": 404470
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 2.8625927910002295e-05,
+      "loss": 0.8312,
+      "step": 404480
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 2.8624779980102545e-05,
+      "loss": 0.8256,
+      "step": 404490
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 2.8623632050202798e-05,
+      "loss": 0.8137,
+      "step": 404500
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 2.862248412030305e-05,
+      "loss": 0.8292,
+      "step": 404510
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 2.8621336190403305e-05,
+      "loss": 0.775,
+      "step": 404520
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 2.8620188260503555e-05,
+      "loss": 0.9581,
+      "step": 404530
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 2.861904033060381e-05,
+      "loss": 0.8919,
+      "step": 404540
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 2.861789240070406e-05,
+      "loss": 0.8961,
+      "step": 404550
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 2.861674447080431e-05,
+      "loss": 0.9408,
+      "step": 404560
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 2.8615596540904567e-05,
+      "loss": 0.894,
+      "step": 404570
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 2.8614448611004817e-05,
+      "loss": 0.9293,
+      "step": 404580
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 2.861330068110507e-05,
+      "loss": 0.8309,
+      "step": 404590
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 2.8612152751205324e-05,
+      "loss": 0.9275,
+      "step": 404600
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 2.8611004821305577e-05,
+      "loss": 0.8119,
+      "step": 404610
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 2.8609856891405827e-05,
+      "loss": 0.9275,
+      "step": 404620
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 2.8608708961506084e-05,
+      "loss": 0.8428,
+      "step": 404630
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 2.8607561031606333e-05,
+      "loss": 0.8619,
+      "step": 404640
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 2.8606413101706587e-05,
+      "loss": 0.7889,
+      "step": 404650
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 2.860526517180684e-05,
+      "loss": 0.9056,
+      "step": 404660
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 2.8604117241907093e-05,
+      "loss": 0.8395,
+      "step": 404670
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 2.8602969312007343e-05,
+      "loss": 0.8971,
+      "step": 404680
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 2.86018213821076e-05,
+      "loss": 0.8311,
+      "step": 404690
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 2.860067345220785e-05,
+      "loss": 0.9237,
+      "step": 404700
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 2.85995255223081e-05,
+      "loss": 0.7994,
+      "step": 404710
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 2.8598377592408356e-05,
+      "loss": 0.838,
+      "step": 404720
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 2.8597229662508606e-05,
+      "loss": 0.8165,
+      "step": 404730
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 2.859608173260886e-05,
+      "loss": 0.8805,
+      "step": 404740
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 2.8594933802709112e-05,
+      "loss": 0.8469,
+      "step": 404750
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 2.8593785872809366e-05,
+      "loss": 0.8345,
+      "step": 404760
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 2.8592637942909615e-05,
+      "loss": 0.8482,
+      "step": 404770
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 2.8591490013009872e-05,
+      "loss": 0.8558,
+      "step": 404780
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 2.8590342083110122e-05,
+      "loss": 0.8389,
+      "step": 404790
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 2.8589194153210375e-05,
+      "loss": 0.7686,
+      "step": 404800
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 2.858804622331063e-05,
+      "loss": 0.849,
+      "step": 404810
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 2.858689829341088e-05,
+      "loss": 0.8643,
+      "step": 404820
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 2.858575036351113e-05,
+      "loss": 0.8724,
+      "step": 404830
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 2.8584602433611388e-05,
+      "loss": 0.8602,
+      "step": 404840
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 2.8583454503711638e-05,
+      "loss": 0.7439,
+      "step": 404850
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 2.8582306573811888e-05,
+      "loss": 0.8148,
+      "step": 404860
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 2.8581158643912144e-05,
+      "loss": 0.8841,
+      "step": 404870
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 2.8580010714012394e-05,
+      "loss": 0.8701,
+      "step": 404880
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 2.8578862784112648e-05,
+      "loss": 0.8406,
+      "step": 404890
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 2.85777148542129e-05,
+      "loss": 0.8427,
+      "step": 404900
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 2.8576566924313154e-05,
+      "loss": 0.8441,
+      "step": 404910
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 2.8575418994413404e-05,
+      "loss": 0.867,
+      "step": 404920
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 2.857427106451366e-05,
+      "loss": 0.8244,
+      "step": 404930
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 2.857312313461391e-05,
+      "loss": 0.821,
+      "step": 404940
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 2.8571975204714164e-05,
+      "loss": 0.8576,
+      "step": 404950
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 2.8570827274814417e-05,
+      "loss": 0.8496,
+      "step": 404960
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 2.856967934491467e-05,
+      "loss": 0.9286,
+      "step": 404970
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 2.856853141501492e-05,
+      "loss": 0.892,
+      "step": 404980
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 2.8567383485115177e-05,
+      "loss": 0.8128,
+      "step": 404990
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 2.8566235555215426e-05,
+      "loss": 0.7429,
+      "step": 405000
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 2.8565087625315676e-05,
+      "loss": 0.7958,
+      "step": 405010
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 2.8563939695415933e-05,
+      "loss": 0.8551,
+      "step": 405020
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 2.8562791765516183e-05,
+      "loss": 0.8279,
+      "step": 405030
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 2.8561643835616436e-05,
+      "loss": 0.8719,
+      "step": 405040
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 2.856049590571669e-05,
+      "loss": 0.8604,
+      "step": 405050
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 2.8559347975816942e-05,
+      "loss": 0.9017,
+      "step": 405060
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 2.8558200045917192e-05,
+      "loss": 0.9195,
+      "step": 405070
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 2.8557052116017446e-05,
+      "loss": 0.9192,
+      "step": 405080
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 2.85559041861177e-05,
+      "loss": 0.7759,
+      "step": 405090
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 2.8554756256217952e-05,
+      "loss": 0.8685,
+      "step": 405100
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 2.8553608326318202e-05,
+      "loss": 0.9128,
+      "step": 405110
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 2.855246039641846e-05,
+      "loss": 0.8611,
+      "step": 405120
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 2.855131246651871e-05,
+      "loss": 0.9376,
+      "step": 405130
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 2.8550164536618958e-05,
+      "loss": 0.8069,
+      "step": 405140
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 2.8549016606719215e-05,
+      "loss": 0.8557,
+      "step": 405150
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 2.8547868676819465e-05,
+      "loss": 0.8838,
+      "step": 405160
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 2.8546720746919718e-05,
+      "loss": 0.8811,
+      "step": 405170
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 2.854557281701997e-05,
+      "loss": 0.9558,
+      "step": 405180
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 2.8544424887120224e-05,
+      "loss": 0.8407,
+      "step": 405190
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 2.8543276957220474e-05,
+      "loss": 0.8439,
+      "step": 405200
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 2.854212902732073e-05,
+      "loss": 0.7721,
+      "step": 405210
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 2.854098109742098e-05,
+      "loss": 0.8754,
+      "step": 405220
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 2.8539833167521234e-05,
+      "loss": 0.856,
+      "step": 405230
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 2.8538685237621487e-05,
+      "loss": 0.9061,
+      "step": 405240
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 2.853753730772174e-05,
+      "loss": 0.8019,
+      "step": 405250
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 2.853638937782199e-05,
+      "loss": 0.8114,
+      "step": 405260
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 2.8535241447922247e-05,
+      "loss": 0.8879,
+      "step": 405270
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 2.8534093518022497e-05,
+      "loss": 0.8026,
+      "step": 405280
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 2.8532945588122747e-05,
+      "loss": 0.8153,
+      "step": 405290
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 2.8531797658223003e-05,
+      "loss": 0.8019,
+      "step": 405300
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 2.8530649728323253e-05,
+      "loss": 0.8298,
+      "step": 405310
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 2.8529501798423506e-05,
+      "loss": 0.8367,
+      "step": 405320
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 2.852835386852376e-05,
+      "loss": 0.8848,
+      "step": 405330
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 2.8527205938624013e-05,
+      "loss": 0.8077,
+      "step": 405340
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 2.8526058008724263e-05,
+      "loss": 0.8857,
+      "step": 405350
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 2.852491007882452e-05,
+      "loss": 0.9074,
+      "step": 405360
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 2.852376214892477e-05,
+      "loss": 0.8799,
+      "step": 405370
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 2.8522614219025023e-05,
+      "loss": 0.8106,
+      "step": 405380
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 2.8521466289125276e-05,
+      "loss": 0.8227,
+      "step": 405390
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 2.852031835922553e-05,
+      "loss": 0.8421,
+      "step": 405400
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 2.851917042932578e-05,
+      "loss": 0.8222,
+      "step": 405410
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 2.8518022499426035e-05,
+      "loss": 0.8651,
+      "step": 405420
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 2.8516874569526285e-05,
+      "loss": 0.8163,
+      "step": 405430
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 2.8515726639626535e-05,
+      "loss": 0.9465,
+      "step": 405440
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 2.8514578709726792e-05,
+      "loss": 0.7861,
+      "step": 405450
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 2.851343077982704e-05,
+      "loss": 0.8586,
+      "step": 405460
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 2.8512282849927295e-05,
+      "loss": 0.9177,
+      "step": 405470
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 2.8511134920027548e-05,
+      "loss": 0.8766,
+      "step": 405480
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 2.85099869901278e-05,
+      "loss": 0.8961,
+      "step": 405490
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 2.850883906022805e-05,
+      "loss": 0.9181,
+      "step": 405500
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 2.8507691130328308e-05,
+      "loss": 0.826,
+      "step": 405510
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 2.8506543200428558e-05,
+      "loss": 0.91,
+      "step": 405520
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 2.850539527052881e-05,
+      "loss": 0.8824,
+      "step": 405530
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 2.8504247340629064e-05,
+      "loss": 0.8779,
+      "step": 405540
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 2.8503099410729317e-05,
+      "loss": 0.8703,
+      "step": 405550
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 2.8501951480829567e-05,
+      "loss": 0.9119,
+      "step": 405560
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 2.8500803550929824e-05,
+      "loss": 0.879,
+      "step": 405570
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 2.8499655621030074e-05,
+      "loss": 0.7878,
+      "step": 405580
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 2.8498507691130324e-05,
+      "loss": 0.9143,
+      "step": 405590
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 2.849735976123058e-05,
+      "loss": 0.8232,
+      "step": 405600
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 2.849621183133083e-05,
+      "loss": 0.7993,
+      "step": 405610
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 2.8495063901431083e-05,
+      "loss": 0.8955,
+      "step": 405620
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 2.8493915971531337e-05,
+      "loss": 0.8942,
+      "step": 405630
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 2.849276804163159e-05,
+      "loss": 0.8292,
+      "step": 405640
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 2.849162011173184e-05,
+      "loss": 0.8571,
+      "step": 405650
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 2.8490472181832096e-05,
+      "loss": 0.8803,
+      "step": 405660
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 2.8489324251932346e-05,
+      "loss": 0.8295,
+      "step": 405670
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 2.84881763220326e-05,
+      "loss": 0.8124,
+      "step": 405680
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 2.8487028392132853e-05,
+      "loss": 0.8261,
+      "step": 405690
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 2.8485880462233106e-05,
+      "loss": 0.8432,
+      "step": 405700
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 2.8484732532333356e-05,
+      "loss": 0.7849,
+      "step": 405710
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 2.8483584602433612e-05,
+      "loss": 0.8675,
+      "step": 405720
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 2.8482436672533862e-05,
+      "loss": 0.9161,
+      "step": 405730
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 2.8481288742634112e-05,
+      "loss": 0.9177,
+      "step": 405740
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 2.848014081273437e-05,
+      "loss": 0.8651,
+      "step": 405750
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 2.847899288283462e-05,
+      "loss": 0.8009,
+      "step": 405760
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 2.8477844952934872e-05,
+      "loss": 0.8175,
+      "step": 405770
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 2.8476697023035122e-05,
+      "loss": 0.8827,
+      "step": 405780
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 2.847554909313538e-05,
+      "loss": 0.8399,
+      "step": 405790
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 2.8474401163235628e-05,
+      "loss": 0.8174,
+      "step": 405800
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 2.847325323333588e-05,
+      "loss": 0.7937,
+      "step": 405810
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 2.8472105303436135e-05,
+      "loss": 0.7534,
+      "step": 405820
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 2.8470957373536388e-05,
+      "loss": 0.8954,
+      "step": 405830
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 2.8469809443636638e-05,
+      "loss": 0.8592,
+      "step": 405840
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 2.8468661513736894e-05,
+      "loss": 0.8726,
+      "step": 405850
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 2.8467513583837144e-05,
+      "loss": 0.8199,
+      "step": 405860
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 2.8466365653937394e-05,
+      "loss": 0.9308,
+      "step": 405870
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 2.846521772403765e-05,
+      "loss": 0.8785,
+      "step": 405880
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 2.84640697941379e-05,
+      "loss": 0.8468,
+      "step": 405890
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 2.8462921864238154e-05,
+      "loss": 0.8558,
+      "step": 405900
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 2.8461773934338407e-05,
+      "loss": 0.9136,
+      "step": 405910
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 2.846062600443866e-05,
+      "loss": 0.838,
+      "step": 405920
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 2.845947807453891e-05,
+      "loss": 0.8045,
+      "step": 405930
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 2.8458330144639167e-05,
+      "loss": 0.8423,
+      "step": 405940
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 2.8457182214739417e-05,
+      "loss": 0.8895,
+      "step": 405950
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 2.845603428483967e-05,
+      "loss": 0.8036,
+      "step": 405960
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 2.8454886354939923e-05,
+      "loss": 0.8669,
+      "step": 405970
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 2.8453738425040176e-05,
+      "loss": 0.7809,
+      "step": 405980
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 2.8452590495140426e-05,
+      "loss": 0.8442,
+      "step": 405990
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 2.8451442565240683e-05,
+      "loss": 0.8975,
+      "step": 406000
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 2.8450294635340933e-05,
+      "loss": 0.7889,
+      "step": 406010
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 2.8449146705441183e-05,
+      "loss": 0.8843,
+      "step": 406020
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 2.844799877554144e-05,
+      "loss": 0.8042,
+      "step": 406030
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 2.844685084564169e-05,
+      "loss": 0.8089,
+      "step": 406040
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 2.8445702915741942e-05,
+      "loss": 0.8603,
+      "step": 406050
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 2.8444554985842196e-05,
+      "loss": 0.965,
+      "step": 406060
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 2.844340705594245e-05,
+      "loss": 0.7452,
+      "step": 406070
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 2.84422591260427e-05,
+      "loss": 0.7917,
+      "step": 406080
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 2.8441111196142955e-05,
+      "loss": 0.9017,
+      "step": 406090
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 2.8439963266243205e-05,
+      "loss": 0.9061,
+      "step": 406100
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 2.843881533634346e-05,
+      "loss": 0.9859,
+      "step": 406110
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 2.843766740644371e-05,
+      "loss": 0.8682,
+      "step": 406120
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 2.8436519476543965e-05,
+      "loss": 0.8194,
+      "step": 406130
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 2.8435371546644215e-05,
+      "loss": 0.8083,
+      "step": 406140
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 2.843422361674447e-05,
+      "loss": 0.9241,
+      "step": 406150
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 2.843307568684472e-05,
+      "loss": 0.8544,
+      "step": 406160
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 2.843192775694497e-05,
+      "loss": 0.8663,
+      "step": 406170
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 2.8430779827045228e-05,
+      "loss": 0.8511,
+      "step": 406180
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 2.8429631897145478e-05,
+      "loss": 0.811,
+      "step": 406190
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 2.842848396724573e-05,
+      "loss": 0.9146,
+      "step": 406200
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 2.8427336037345984e-05,
+      "loss": 0.869,
+      "step": 406210
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 2.8426188107446237e-05,
+      "loss": 0.9012,
+      "step": 406220
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 2.8425040177546487e-05,
+      "loss": 0.8665,
+      "step": 406230
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 2.8423892247646744e-05,
+      "loss": 0.8017,
+      "step": 406240
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 2.8422744317746994e-05,
+      "loss": 0.8535,
+      "step": 406250
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 2.8421596387847247e-05,
+      "loss": 0.8285,
+      "step": 406260
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 2.84204484579475e-05,
+      "loss": 0.8886,
+      "step": 406270
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 2.8419300528047753e-05,
+      "loss": 0.9283,
+      "step": 406280
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 2.8418152598148003e-05,
+      "loss": 0.8583,
+      "step": 406290
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 2.841700466824826e-05,
+      "loss": 0.8431,
+      "step": 406300
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 2.841585673834851e-05,
+      "loss": 0.7568,
+      "step": 406310
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 2.841470880844876e-05,
+      "loss": 0.8273,
+      "step": 406320
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 2.8413560878549016e-05,
+      "loss": 0.9054,
+      "step": 406330
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 2.8412412948649266e-05,
+      "loss": 0.8338,
+      "step": 406340
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 2.841126501874952e-05,
+      "loss": 0.8657,
+      "step": 406350
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 2.8410117088849773e-05,
+      "loss": 0.796,
+      "step": 406360
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 2.8408969158950026e-05,
+      "loss": 0.8461,
+      "step": 406370
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 2.8407821229050276e-05,
+      "loss": 0.9198,
+      "step": 406380
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 2.8406673299150532e-05,
+      "loss": 0.8014,
+      "step": 406390
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 2.8405525369250782e-05,
+      "loss": 0.9282,
+      "step": 406400
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 2.8404377439351035e-05,
+      "loss": 0.9003,
+      "step": 406410
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 2.840322950945129e-05,
+      "loss": 0.8225,
+      "step": 406420
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 2.8402081579551542e-05,
+      "loss": 0.8965,
+      "step": 406430
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 2.840093364965179e-05,
+      "loss": 0.9308,
+      "step": 406440
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 2.839978571975204e-05,
+      "loss": 0.8989,
+      "step": 406450
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 2.8398637789852298e-05,
+      "loss": 0.8867,
+      "step": 406460
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 2.8397489859952548e-05,
+      "loss": 0.8571,
+      "step": 406470
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 2.83963419300528e-05,
+      "loss": 0.8092,
+      "step": 406480
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 2.8395194000153055e-05,
+      "loss": 0.8072,
+      "step": 406490
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 2.8394046070253308e-05,
+      "loss": 0.8612,
+      "step": 406500
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 2.8392898140353558e-05,
+      "loss": 0.835,
+      "step": 406510
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 2.8391750210453814e-05,
+      "loss": 0.8113,
+      "step": 406520
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 2.8390602280554064e-05,
+      "loss": 0.9607,
+      "step": 406530
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 2.8389454350654317e-05,
+      "loss": 0.9199,
+      "step": 406540
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 2.838830642075457e-05,
+      "loss": 0.8638,
+      "step": 406550
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 2.8387158490854824e-05,
+      "loss": 0.8789,
+      "step": 406560
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 2.8386010560955074e-05,
+      "loss": 0.8107,
+      "step": 406570
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 2.838486263105533e-05,
+      "loss": 0.8809,
+      "step": 406580
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 2.838371470115558e-05,
+      "loss": 0.8409,
+      "step": 406590
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 2.838256677125583e-05,
+      "loss": 0.8404,
+      "step": 406600
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 2.8381418841356087e-05,
+      "loss": 0.8221,
+      "step": 406610
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 2.8380270911456336e-05,
+      "loss": 0.8382,
+      "step": 406620
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 2.837912298155659e-05,
+      "loss": 0.792,
+      "step": 406630
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 2.8377975051656843e-05,
+      "loss": 0.8462,
+      "step": 406640
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 2.8376827121757096e-05,
+      "loss": 0.9733,
+      "step": 406650
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 2.8375679191857346e-05,
+      "loss": 0.913,
+      "step": 406660
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 2.8374531261957603e-05,
+      "loss": 0.8595,
+      "step": 406670
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 2.8373383332057853e-05,
+      "loss": 0.9403,
+      "step": 406680
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 2.8372235402158106e-05,
+      "loss": 0.8379,
+      "step": 406690
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 2.837108747225836e-05,
+      "loss": 0.8965,
+      "step": 406700
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 2.8369939542358612e-05,
+      "loss": 0.802,
+      "step": 406710
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 2.8368791612458862e-05,
+      "loss": 0.7885,
+      "step": 406720
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 2.836764368255912e-05,
+      "loss": 0.8115,
+      "step": 406730
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 2.836649575265937e-05,
+      "loss": 0.9032,
+      "step": 406740
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 2.836534782275962e-05,
+      "loss": 0.8701,
+      "step": 406750
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 2.8364199892859875e-05,
+      "loss": 0.7958,
+      "step": 406760
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 2.8363051962960125e-05,
+      "loss": 0.7972,
+      "step": 406770
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 2.8361904033060378e-05,
+      "loss": 0.8898,
+      "step": 406780
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 2.836075610316063e-05,
+      "loss": 0.9285,
+      "step": 406790
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 2.8359608173260885e-05,
+      "loss": 0.9003,
+      "step": 406800
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 2.8358460243361135e-05,
+      "loss": 0.7897,
+      "step": 406810
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 2.835731231346139e-05,
+      "loss": 0.8696,
+      "step": 406820
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 2.835616438356164e-05,
+      "loss": 0.8158,
+      "step": 406830
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 2.8355016453661894e-05,
+      "loss": 0.8686,
+      "step": 406840
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 2.8353868523762148e-05,
+      "loss": 0.9143,
+      "step": 406850
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 2.83527205938624e-05,
+      "loss": 0.9966,
+      "step": 406860
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 2.835157266396265e-05,
+      "loss": 0.8397,
+      "step": 406870
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 2.8350424734062907e-05,
+      "loss": 0.7591,
+      "step": 406880
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 2.8349276804163157e-05,
+      "loss": 0.8133,
+      "step": 406890
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 2.8348128874263407e-05,
+      "loss": 0.9181,
+      "step": 406900
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 2.8346980944363664e-05,
+      "loss": 0.8629,
+      "step": 406910
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 2.8345833014463913e-05,
+      "loss": 0.8649,
+      "step": 406920
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 2.8344685084564167e-05,
+      "loss": 0.8544,
+      "step": 406930
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 2.834353715466442e-05,
+      "loss": 0.8347,
+      "step": 406940
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 2.8342389224764673e-05,
+      "loss": 0.9535,
+      "step": 406950
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 2.8341241294864923e-05,
+      "loss": 0.8889,
+      "step": 406960
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 2.834009336496518e-05,
+      "loss": 0.8149,
+      "step": 406970
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 2.833894543506543e-05,
+      "loss": 0.9868,
+      "step": 406980
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 2.8337797505165683e-05,
+      "loss": 0.8769,
+      "step": 406990
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 2.8336649575265936e-05,
+      "loss": 0.8493,
+      "step": 407000
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 2.833550164536619e-05,
+      "loss": 0.9129,
+      "step": 407010
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 2.833435371546644e-05,
+      "loss": 0.9121,
+      "step": 407020
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 2.8333205785566696e-05,
+      "loss": 0.8068,
+      "step": 407030
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 2.8332057855666946e-05,
+      "loss": 0.8471,
+      "step": 407040
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 2.8330909925767195e-05,
+      "loss": 0.7902,
+      "step": 407050
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 2.8329761995867452e-05,
+      "loss": 0.7992,
+      "step": 407060
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 2.8328614065967702e-05,
+      "loss": 0.9263,
+      "step": 407070
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 2.8327466136067955e-05,
+      "loss": 0.8654,
+      "step": 407080
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 2.832631820616821e-05,
+      "loss": 0.8634,
+      "step": 407090
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 2.832517027626846e-05,
+      "loss": 0.8031,
+      "step": 407100
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 2.832402234636871e-05,
+      "loss": 0.8592,
+      "step": 407110
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 2.8322874416468968e-05,
+      "loss": 0.8893,
+      "step": 407120
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 2.8321726486569218e-05,
+      "loss": 0.8174,
+      "step": 407130
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 2.832057855666947e-05,
+      "loss": 0.7903,
+      "step": 407140
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 2.831943062676972e-05,
+      "loss": 0.8239,
+      "step": 407150
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 2.8318282696869978e-05,
+      "loss": 0.8048,
+      "step": 407160
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 2.8317134766970228e-05,
+      "loss": 0.7186,
+      "step": 407170
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 2.8315986837070477e-05,
+      "loss": 0.7688,
+      "step": 407180
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 2.8314838907170734e-05,
+      "loss": 0.8029,
+      "step": 407190
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 2.8313690977270984e-05,
+      "loss": 0.8716,
+      "step": 407200
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 2.8312543047371237e-05,
+      "loss": 0.9421,
+      "step": 407210
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 2.831139511747149e-05,
+      "loss": 0.9118,
+      "step": 407220
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 2.8310247187571744e-05,
+      "loss": 0.8363,
+      "step": 407230
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 2.8309099257671993e-05,
+      "loss": 0.8186,
+      "step": 407240
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 2.830795132777225e-05,
+      "loss": 0.8924,
+      "step": 407250
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 2.83068033978725e-05,
+      "loss": 0.8847,
+      "step": 407260
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 2.8305655467972753e-05,
+      "loss": 0.8762,
+      "step": 407270
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 2.8304507538073006e-05,
+      "loss": 0.8731,
+      "step": 407280
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 2.830335960817326e-05,
+      "loss": 0.8855,
+      "step": 407290
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 2.830221167827351e-05,
+      "loss": 0.8754,
+      "step": 407300
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 2.8301063748373766e-05,
+      "loss": 0.8907,
+      "step": 407310
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 2.8299915818474016e-05,
+      "loss": 0.7618,
+      "step": 407320
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 2.8298767888574266e-05,
+      "loss": 0.8381,
+      "step": 407330
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 2.8297619958674523e-05,
+      "loss": 0.7601,
+      "step": 407340
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 2.8296472028774772e-05,
+      "loss": 0.907,
+      "step": 407350
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 2.8295324098875026e-05,
+      "loss": 0.908,
+      "step": 407360
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 2.829417616897528e-05,
+      "loss": 0.8287,
+      "step": 407370
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 2.8293028239075532e-05,
+      "loss": 0.9013,
+      "step": 407380
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 2.8291880309175782e-05,
+      "loss": 0.919,
+      "step": 407390
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 2.829073237927604e-05,
+      "loss": 0.8822,
+      "step": 407400
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 2.828958444937629e-05,
+      "loss": 0.914,
+      "step": 407410
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 2.828843651947654e-05,
+      "loss": 0.8497,
+      "step": 407420
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 2.8287288589576795e-05,
+      "loss": 0.9074,
+      "step": 407430
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 2.8286140659677048e-05,
+      "loss": 0.7732,
+      "step": 407440
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 2.8284992729777298e-05,
+      "loss": 0.8401,
+      "step": 407450
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 2.8283844799877555e-05,
+      "loss": 0.7206,
+      "step": 407460
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 2.8282696869977804e-05,
+      "loss": 0.8777,
+      "step": 407470
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 2.8281548940078054e-05,
+      "loss": 0.8882,
+      "step": 407480
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 2.828040101017831e-05,
+      "loss": 0.9037,
+      "step": 407490
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 2.827925308027856e-05,
+      "loss": 0.8718,
+      "step": 407500
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 2.8278105150378814e-05,
+      "loss": 0.8773,
+      "step": 407510
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 2.8276957220479067e-05,
+      "loss": 0.8798,
+      "step": 407520
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 2.827580929057932e-05,
+      "loss": 0.7951,
+      "step": 407530
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 2.827466136067957e-05,
+      "loss": 0.8418,
+      "step": 407540
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 2.8273513430779827e-05,
+      "loss": 0.85,
+      "step": 407550
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 2.8272365500880077e-05,
+      "loss": 0.8139,
+      "step": 407560
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 2.827121757098033e-05,
+      "loss": 0.8058,
+      "step": 407570
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 2.8270069641080583e-05,
+      "loss": 0.8,
+      "step": 407580
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 2.8268921711180837e-05,
+      "loss": 0.8926,
+      "step": 407590
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 2.8267773781281086e-05,
+      "loss": 0.8787,
+      "step": 407600
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 2.8266625851381343e-05,
+      "loss": 0.7619,
+      "step": 407610
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 2.8265477921481593e-05,
+      "loss": 0.8958,
+      "step": 407620
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 2.8264329991581843e-05,
+      "loss": 0.8253,
+      "step": 407630
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 2.82631820616821e-05,
+      "loss": 0.8508,
+      "step": 407640
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 2.826203413178235e-05,
+      "loss": 0.8694,
+      "step": 407650
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 2.8260886201882603e-05,
+      "loss": 0.9179,
+      "step": 407660
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 2.8259738271982856e-05,
+      "loss": 0.8176,
+      "step": 407670
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 2.825859034208311e-05,
+      "loss": 0.8621,
+      "step": 407680
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 2.825744241218336e-05,
+      "loss": 0.7343,
+      "step": 407690
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 2.8256294482283616e-05,
+      "loss": 0.917,
+      "step": 407700
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 2.8255146552383865e-05,
+      "loss": 0.798,
+      "step": 407710
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 2.825399862248412e-05,
+      "loss": 0.9046,
+      "step": 407720
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 2.8252850692584372e-05,
+      "loss": 0.8907,
+      "step": 407730
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 2.8251702762684625e-05,
+      "loss": 0.873,
+      "step": 407740
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 2.8250554832784875e-05,
+      "loss": 0.773,
+      "step": 407750
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 2.824940690288513e-05,
+      "loss": 0.8801,
+      "step": 407760
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 2.824825897298538e-05,
+      "loss": 0.8762,
+      "step": 407770
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 2.824711104308563e-05,
+      "loss": 0.9051,
+      "step": 407780
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 2.8245963113185888e-05,
+      "loss": 0.9228,
+      "step": 407790
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 2.8244815183286138e-05,
+      "loss": 0.8759,
+      "step": 407800
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 2.824366725338639e-05,
+      "loss": 0.8633,
+      "step": 407810
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 2.824251932348664e-05,
+      "loss": 0.8795,
+      "step": 407820
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 2.8241371393586898e-05,
+      "loss": 0.8844,
+      "step": 407830
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 2.8240223463687147e-05,
+      "loss": 0.7943,
+      "step": 407840
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 2.8239075533787397e-05,
+      "loss": 0.8714,
+      "step": 407850
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 2.8237927603887654e-05,
+      "loss": 0.8281,
+      "step": 407860
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 2.8236779673987907e-05,
+      "loss": 0.8606,
+      "step": 407870
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 2.8235631744088157e-05,
+      "loss": 0.8579,
+      "step": 407880
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 2.8234483814188414e-05,
+      "loss": 0.9065,
+      "step": 407890
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 2.8233335884288663e-05,
+      "loss": 0.9252,
+      "step": 407900
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 2.8232187954388913e-05,
+      "loss": 0.8234,
+      "step": 407910
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 2.823104002448917e-05,
+      "loss": 0.8051,
+      "step": 407920
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 2.822989209458942e-05,
+      "loss": 0.8375,
+      "step": 407930
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 2.8228744164689673e-05,
+      "loss": 0.8512,
+      "step": 407940
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 2.8227596234789926e-05,
+      "loss": 0.8654,
+      "step": 407950
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 2.822644830489018e-05,
+      "loss": 0.9004,
+      "step": 407960
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 2.822530037499043e-05,
+      "loss": 0.866,
+      "step": 407970
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 2.8224152445090686e-05,
+      "loss": 0.7977,
+      "step": 407980
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 2.8223004515190936e-05,
+      "loss": 0.7938,
+      "step": 407990
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 2.8221856585291186e-05,
+      "loss": 0.8348,
+      "step": 408000
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 2.8220708655391442e-05,
+      "loss": 0.7832,
+      "step": 408010
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 2.8219560725491696e-05,
+      "loss": 0.8269,
+      "step": 408020
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 2.8218412795591945e-05,
+      "loss": 0.8705,
+      "step": 408030
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 2.8217264865692202e-05,
+      "loss": 0.8863,
+      "step": 408040
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 2.8216116935792452e-05,
+      "loss": 0.8177,
+      "step": 408050
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 2.8214969005892702e-05,
+      "loss": 1.008,
+      "step": 408060
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 2.821382107599296e-05,
+      "loss": 0.8368,
+      "step": 408070
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 2.8212673146093208e-05,
+      "loss": 0.8864,
+      "step": 408080
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 2.821152521619346e-05,
+      "loss": 0.8963,
+      "step": 408090
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 2.8210377286293715e-05,
+      "loss": 0.8215,
+      "step": 408100
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 2.8209229356393968e-05,
+      "loss": 0.9361,
+      "step": 408110
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 2.8208081426494218e-05,
+      "loss": 0.8733,
+      "step": 408120
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 2.8206933496594474e-05,
+      "loss": 0.8463,
+      "step": 408130
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 2.8205785566694724e-05,
+      "loss": 0.807,
+      "step": 408140
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 2.8204637636794974e-05,
+      "loss": 0.8474,
+      "step": 408150
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 2.820348970689523e-05,
+      "loss": 0.8589,
+      "step": 408160
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 2.8202341776995484e-05,
+      "loss": 0.8558,
+      "step": 408170
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 2.8201193847095734e-05,
+      "loss": 0.962,
+      "step": 408180
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 2.820004591719599e-05,
+      "loss": 0.7891,
+      "step": 408190
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 2.819889798729624e-05,
+      "loss": 0.8332,
+      "step": 408200
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 2.819775005739649e-05,
+      "loss": 0.7616,
+      "step": 408210
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 2.8196602127496747e-05,
+      "loss": 0.863,
+      "step": 408220
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 2.8195454197596997e-05,
+      "loss": 0.8437,
+      "step": 408230
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 2.819430626769725e-05,
+      "loss": 0.7974,
+      "step": 408240
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 2.8193158337797503e-05,
+      "loss": 0.8453,
+      "step": 408250
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 2.8192010407897756e-05,
+      "loss": 0.8549,
+      "step": 408260
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 2.8190862477998006e-05,
+      "loss": 0.8997,
+      "step": 408270
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 2.8189714548098263e-05,
+      "loss": 0.8181,
+      "step": 408280
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 2.8188566618198513e-05,
+      "loss": 0.8978,
+      "step": 408290
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 2.8187418688298763e-05,
+      "loss": 0.9075,
+      "step": 408300
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 2.818627075839902e-05,
+      "loss": 0.85,
+      "step": 408310
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 2.8185122828499273e-05,
+      "loss": 0.8531,
+      "step": 408320
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 2.8183974898599522e-05,
+      "loss": 0.883,
+      "step": 408330
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 2.818282696869978e-05,
+      "loss": 0.8615,
+      "step": 408340
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 2.818167903880003e-05,
+      "loss": 0.8524,
+      "step": 408350
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 2.818053110890028e-05,
+      "loss": 0.787,
+      "step": 408360
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 2.8179383179000535e-05,
+      "loss": 0.8146,
+      "step": 408370
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 2.8178235249100785e-05,
+      "loss": 0.7844,
+      "step": 408380
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 2.817708731920104e-05,
+      "loss": 0.8554,
+      "step": 408390
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 2.817593938930129e-05,
+      "loss": 0.8968,
+      "step": 408400
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 2.8174791459401545e-05,
+      "loss": 0.8204,
+      "step": 408410
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 2.8173643529501795e-05,
+      "loss": 0.8714,
+      "step": 408420
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 2.817249559960205e-05,
+      "loss": 0.8241,
+      "step": 408430
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 2.81713476697023e-05,
+      "loss": 0.924,
+      "step": 408440
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 2.817019973980255e-05,
+      "loss": 0.8172,
+      "step": 408450
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 2.8169051809902808e-05,
+      "loss": 0.802,
+      "step": 408460
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 2.816790388000306e-05,
+      "loss": 0.8798,
+      "step": 408470
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 2.816675595010331e-05,
+      "loss": 0.8446,
+      "step": 408480
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 2.8165608020203567e-05,
+      "loss": 0.8272,
+      "step": 408490
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 2.8164460090303817e-05,
+      "loss": 0.8862,
+      "step": 408500
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 2.8163312160404067e-05,
+      "loss": 0.8048,
+      "step": 408510
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 2.816216423050432e-05,
+      "loss": 0.8527,
+      "step": 408520
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 2.8161016300604574e-05,
+      "loss": 0.8389,
+      "step": 408530
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 2.8159868370704827e-05,
+      "loss": 0.8621,
+      "step": 408540
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 2.8158720440805077e-05,
+      "loss": 0.7358,
+      "step": 408550
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 2.8157572510905333e-05,
+      "loss": 0.7946,
+      "step": 408560
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 2.8156424581005583e-05,
+      "loss": 0.8619,
+      "step": 408570
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 2.8155276651105833e-05,
+      "loss": 0.9305,
+      "step": 408580
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 2.815412872120609e-05,
+      "loss": 0.8023,
+      "step": 408590
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 2.815298079130634e-05,
+      "loss": 0.8386,
+      "step": 408600
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 2.8151832861406593e-05,
+      "loss": 0.9602,
+      "step": 408610
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 2.815068493150685e-05,
+      "loss": 0.946,
+      "step": 408620
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 2.81495370016071e-05,
+      "loss": 0.8213,
+      "step": 408630
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 2.814838907170735e-05,
+      "loss": 0.9186,
+      "step": 408640
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 2.8147241141807606e-05,
+      "loss": 0.8039,
+      "step": 408650
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 2.8146093211907856e-05,
+      "loss": 0.8004,
+      "step": 408660
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 2.814494528200811e-05,
+      "loss": 0.8006,
+      "step": 408670
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 2.8143797352108362e-05,
+      "loss": 0.9191,
+      "step": 408680
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 2.8142649422208615e-05,
+      "loss": 0.8359,
+      "step": 408690
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 2.8141501492308865e-05,
+      "loss": 0.7811,
+      "step": 408700
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 2.8140353562409122e-05,
+      "loss": 0.8621,
+      "step": 408710
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 2.8139205632509372e-05,
+      "loss": 0.8471,
+      "step": 408720
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 2.813805770260962e-05,
+      "loss": 0.9492,
+      "step": 408730
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 2.8136909772709878e-05,
+      "loss": 0.8595,
+      "step": 408740
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 2.8135761842810128e-05,
+      "loss": 0.9577,
+      "step": 408750
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 2.813461391291038e-05,
+      "loss": 0.8416,
+      "step": 408760
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 2.8133465983010638e-05,
+      "loss": 0.7734,
+      "step": 408770
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 2.8132318053110888e-05,
+      "loss": 0.8906,
+      "step": 408780
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 2.8131170123211138e-05,
+      "loss": 0.8471,
+      "step": 408790
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 2.8130022193311394e-05,
+      "loss": 0.8929,
+      "step": 408800
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 2.8128874263411644e-05,
+      "loss": 0.7473,
+      "step": 408810
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 2.8127726333511897e-05,
+      "loss": 0.8679,
+      "step": 408820
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 2.812657840361215e-05,
+      "loss": 0.7766,
+      "step": 408830
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 2.8125430473712404e-05,
+      "loss": 0.9082,
+      "step": 408840
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 2.8124282543812654e-05,
+      "loss": 0.7773,
+      "step": 408850
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 2.812313461391291e-05,
+      "loss": 0.8793,
+      "step": 408860
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 2.812198668401316e-05,
+      "loss": 0.7646,
+      "step": 408870
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 2.812083875411341e-05,
+      "loss": 0.8639,
+      "step": 408880
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 2.8119690824213667e-05,
+      "loss": 0.7681,
+      "step": 408890
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 2.811854289431392e-05,
+      "loss": 0.9073,
+      "step": 408900
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 2.811739496441417e-05,
+      "loss": 0.8434,
+      "step": 408910
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 2.8116247034514426e-05,
+      "loss": 0.8417,
+      "step": 408920
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 2.8115099104614676e-05,
+      "loss": 0.8514,
+      "step": 408930
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 2.8113951174714926e-05,
+      "loss": 0.8223,
+      "step": 408940
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 2.8112803244815183e-05,
+      "loss": 0.8643,
+      "step": 408950
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 2.8111655314915433e-05,
+      "loss": 0.8891,
+      "step": 408960
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 2.8110507385015686e-05,
+      "loss": 0.7639,
+      "step": 408970
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 2.810935945511594e-05,
+      "loss": 0.8414,
+      "step": 408980
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 2.8108211525216192e-05,
+      "loss": 0.8155,
+      "step": 408990
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 2.8107063595316442e-05,
+      "loss": 0.8695,
+      "step": 409000
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 2.81059156654167e-05,
+      "loss": 0.8749,
+      "step": 409010
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 2.810476773551695e-05,
+      "loss": 0.7803,
+      "step": 409020
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 2.81036198056172e-05,
+      "loss": 0.879,
+      "step": 409030
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 2.8102471875717455e-05,
+      "loss": 0.8987,
+      "step": 409040
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 2.810132394581771e-05,
+      "loss": 0.9307,
+      "step": 409050
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 2.8100176015917958e-05,
+      "loss": 0.8634,
+      "step": 409060
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 2.8099028086018215e-05,
+      "loss": 0.8304,
+      "step": 409070
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 2.8097880156118465e-05,
+      "loss": 0.9157,
+      "step": 409080
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 2.8096732226218715e-05,
+      "loss": 0.8474,
+      "step": 409090
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 2.809558429631897e-05,
+      "loss": 0.8197,
+      "step": 409100
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 2.809443636641922e-05,
+      "loss": 0.785,
+      "step": 409110
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 2.8093288436519474e-05,
+      "loss": 0.8643,
+      "step": 409120
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 2.8092140506619728e-05,
+      "loss": 0.8826,
+      "step": 409130
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 2.809099257671998e-05,
+      "loss": 0.8865,
+      "step": 409140
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 2.808984464682023e-05,
+      "loss": 0.7818,
+      "step": 409150
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 2.8088696716920487e-05,
+      "loss": 0.8849,
+      "step": 409160
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 2.8087548787020737e-05,
+      "loss": 0.9572,
+      "step": 409170
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 2.8086400857120987e-05,
+      "loss": 0.8709,
+      "step": 409180
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 2.808525292722124e-05,
+      "loss": 0.8544,
+      "step": 409190
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 2.8084104997321497e-05,
+      "loss": 0.8493,
+      "step": 409200
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 2.8082957067421747e-05,
+      "loss": 0.8539,
+      "step": 409210
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 2.8081809137521997e-05,
+      "loss": 0.8807,
+      "step": 409220
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 2.8080661207622253e-05,
+      "loss": 0.8137,
+      "step": 409230
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 2.8079513277722503e-05,
+      "loss": 0.9547,
+      "step": 409240
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 2.8078365347822756e-05,
+      "loss": 0.8715,
+      "step": 409250
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 2.807721741792301e-05,
+      "loss": 0.8708,
+      "step": 409260
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 2.8076069488023263e-05,
+      "loss": 0.8609,
+      "step": 409270
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 2.8074921558123513e-05,
+      "loss": 0.9047,
+      "step": 409280
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 2.807377362822377e-05,
+      "loss": 0.7739,
+      "step": 409290
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 2.807262569832402e-05,
+      "loss": 0.868,
+      "step": 409300
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 2.807147776842427e-05,
+      "loss": 0.7668,
+      "step": 409310
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 2.8070329838524526e-05,
+      "loss": 0.8661,
+      "step": 409320
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 2.8069181908624775e-05,
+      "loss": 0.7757,
+      "step": 409330
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 2.806803397872503e-05,
+      "loss": 0.8934,
+      "step": 409340
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 2.8066886048825285e-05,
+      "loss": 0.881,
+      "step": 409350
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 2.8065738118925535e-05,
+      "loss": 0.7792,
+      "step": 409360
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 2.8064590189025785e-05,
+      "loss": 0.8188,
+      "step": 409370
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 2.806344225912604e-05,
+      "loss": 0.7677,
+      "step": 409380
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 2.806229432922629e-05,
+      "loss": 0.8668,
+      "step": 409390
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 2.8061146399326545e-05,
+      "loss": 0.8228,
+      "step": 409400
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 2.8059998469426798e-05,
+      "loss": 0.7773,
+      "step": 409410
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 2.805885053952705e-05,
+      "loss": 0.8823,
+      "step": 409420
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 2.80577026096273e-05,
+      "loss": 0.7815,
+      "step": 409430
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 2.8056554679727558e-05,
+      "loss": 0.7689,
+      "step": 409440
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 2.8055406749827808e-05,
+      "loss": 0.8402,
+      "step": 409450
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 2.8054258819928057e-05,
+      "loss": 0.8812,
+      "step": 409460
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 2.8053110890028314e-05,
+      "loss": 0.8377,
+      "step": 409470
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 2.8051962960128564e-05,
+      "loss": 0.8666,
+      "step": 409480
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 2.8050815030228817e-05,
+      "loss": 0.873,
+      "step": 409490
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 2.8049667100329074e-05,
+      "loss": 0.9561,
+      "step": 409500
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 2.8048519170429324e-05,
+      "loss": 0.813,
+      "step": 409510
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 2.8047371240529574e-05,
+      "loss": 0.8374,
+      "step": 409520
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 2.804622331062983e-05,
+      "loss": 0.7917,
+      "step": 409530
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 2.804507538073008e-05,
+      "loss": 0.9194,
+      "step": 409540
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 2.8043927450830333e-05,
+      "loss": 0.8379,
+      "step": 409550
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 2.8042779520930586e-05,
+      "loss": 0.8218,
+      "step": 409560
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 2.804163159103084e-05,
+      "loss": 0.7845,
+      "step": 409570
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 2.804048366113109e-05,
+      "loss": 0.8177,
+      "step": 409580
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 2.8039335731231346e-05,
+      "loss": 0.8799,
+      "step": 409590
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 2.8038187801331596e-05,
+      "loss": 0.7818,
+      "step": 409600
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 2.8037039871431846e-05,
+      "loss": 0.8755,
+      "step": 409610
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 2.8035891941532103e-05,
+      "loss": 0.8603,
+      "step": 409620
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 2.8034744011632352e-05,
+      "loss": 0.8138,
+      "step": 409630
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 2.8033596081732606e-05,
+      "loss": 0.8867,
+      "step": 409640
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 2.8032448151832862e-05,
+      "loss": 0.7782,
+      "step": 409650
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 2.8031300221933112e-05,
+      "loss": 0.8294,
+      "step": 409660
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 2.8030152292033362e-05,
+      "loss": 0.7814,
+      "step": 409670
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 2.802900436213362e-05,
+      "loss": 0.8077,
+      "step": 409680
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 2.802785643223387e-05,
+      "loss": 0.8169,
+      "step": 409690
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 2.8026708502334122e-05,
+      "loss": 0.9056,
+      "step": 409700
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 2.8025560572434375e-05,
+      "loss": 0.8473,
+      "step": 409710
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 2.8024412642534628e-05,
+      "loss": 0.8187,
+      "step": 409720
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 2.8023264712634878e-05,
+      "loss": 0.879,
+      "step": 409730
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 2.8022116782735135e-05,
+      "loss": 0.8804,
+      "step": 409740
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 2.8020968852835385e-05,
+      "loss": 0.9088,
+      "step": 409750
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 2.8019820922935634e-05,
+      "loss": 0.9159,
+      "step": 409760
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 2.801867299303589e-05,
+      "loss": 0.8931,
+      "step": 409770
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 2.801752506313614e-05,
+      "loss": 0.8642,
+      "step": 409780
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 2.8016377133236394e-05,
+      "loss": 0.8795,
+      "step": 409790
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 2.801522920333665e-05,
+      "loss": 0.8284,
+      "step": 409800
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 2.80140812734369e-05,
+      "loss": 0.8536,
+      "step": 409810
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 2.801293334353715e-05,
+      "loss": 0.8466,
+      "step": 409820
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 2.8011785413637407e-05,
+      "loss": 0.8406,
+      "step": 409830
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 2.8010637483737657e-05,
+      "loss": 0.8749,
+      "step": 409840
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 2.800948955383791e-05,
+      "loss": 0.8817,
+      "step": 409850
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.8008341623938163e-05,
+      "loss": 0.8317,
+      "step": 409860
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.8007193694038417e-05,
+      "loss": 0.8794,
+      "step": 409870
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.8006045764138667e-05,
+      "loss": 0.8317,
+      "step": 409880
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.8004897834238916e-05,
+      "loss": 0.8443,
+      "step": 409890
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.8003749904339173e-05,
+      "loss": 0.8609,
+      "step": 409900
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.8002601974439423e-05,
+      "loss": 0.8076,
+      "step": 409910
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.8001454044539676e-05,
+      "loss": 0.8202,
+      "step": 409920
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.800030611463993e-05,
+      "loss": 0.8428,
+      "step": 409930
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.7999158184740183e-05,
+      "loss": 0.8318,
+      "step": 409940
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.7998010254840432e-05,
+      "loss": 0.9155,
+      "step": 409950
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.799686232494069e-05,
+      "loss": 0.8417,
+      "step": 409960
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.799571439504094e-05,
+      "loss": 0.9534,
+      "step": 409970
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.7994566465141192e-05,
+      "loss": 0.922,
+      "step": 409980
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.7993418535241445e-05,
+      "loss": 0.9536,
+      "step": 409990
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.79922706053417e-05,
+      "loss": 0.857,
+      "step": 410000
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.799112267544195e-05,
+      "loss": 0.7906,
+      "step": 410010
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.7989974745542205e-05,
+      "loss": 0.8463,
+      "step": 410020
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.7988826815642455e-05,
+      "loss": 0.8493,
+      "step": 410030
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.7987678885742705e-05,
+      "loss": 0.7873,
+      "step": 410040
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.798653095584296e-05,
+      "loss": 0.8264,
+      "step": 410050
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.798538302594321e-05,
+      "loss": 0.841,
+      "step": 410060
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.7984235096043465e-05,
+      "loss": 0.8226,
+      "step": 410070
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.7983087166143718e-05,
+      "loss": 0.9068,
+      "step": 410080
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.798193923624397e-05,
+      "loss": 0.9058,
+      "step": 410090
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.798079130634422e-05,
+      "loss": 0.8503,
+      "step": 410100
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.7979643376444478e-05,
+      "loss": 0.8653,
+      "step": 410110
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.7978495446544727e-05,
+      "loss": 0.813,
+      "step": 410120
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.797734751664498e-05,
+      "loss": 0.8333,
+      "step": 410130
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.7976199586745234e-05,
+      "loss": 0.8338,
+      "step": 410140
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.7975051656845487e-05,
+      "loss": 0.9012,
+      "step": 410150
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.7973903726945737e-05,
+      "loss": 0.8286,
+      "step": 410160
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.7972755797045994e-05,
+      "loss": 0.8903,
+      "step": 410170
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.7971607867146243e-05,
+      "loss": 0.8635,
+      "step": 410180
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.7970459937246493e-05,
+      "loss": 0.8227,
+      "step": 410190
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.796931200734675e-05,
+      "loss": 0.8595,
+      "step": 410200
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.7968164077447e-05,
+      "loss": 0.8959,
+      "step": 410210
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.7967016147547253e-05,
+      "loss": 0.807,
+      "step": 410220
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.7965868217647506e-05,
+      "loss": 0.8756,
+      "step": 410230
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.796472028774776e-05,
+      "loss": 0.8438,
+      "step": 410240
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.796357235784801e-05,
+      "loss": 0.9108,
+      "step": 410250
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.7962424427948266e-05,
+      "loss": 0.849,
+      "step": 410260
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.7961276498048516e-05,
+      "loss": 0.8498,
+      "step": 410270
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.796012856814877e-05,
+      "loss": 0.8863,
+      "step": 410280
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.7958980638249022e-05,
+      "loss": 0.8768,
+      "step": 410290
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.7957832708349276e-05,
+      "loss": 0.8262,
+      "step": 410300
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.7956684778449525e-05,
+      "loss": 0.8512,
+      "step": 410310
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.7955536848549782e-05,
+      "loss": 0.9054,
+      "step": 410320
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.7954388918650032e-05,
+      "loss": 0.9328,
+      "step": 410330
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.7953240988750282e-05,
+      "loss": 0.8118,
+      "step": 410340
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.795209305885054e-05,
+      "loss": 0.9175,
+      "step": 410350
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.7950945128950788e-05,
+      "loss": 0.777,
+      "step": 410360
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.794979719905104e-05,
+      "loss": 0.854,
+      "step": 410370
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.7948649269151295e-05,
+      "loss": 0.9221,
+      "step": 410380
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.7947501339251548e-05,
+      "loss": 0.806,
+      "step": 410390
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.7946353409351798e-05,
+      "loss": 0.8828,
+      "step": 410400
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.7945205479452054e-05,
+      "loss": 0.8991,
+      "step": 410410
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.7944057549552304e-05,
+      "loss": 0.8753,
+      "step": 410420
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.7942909619652558e-05,
+      "loss": 0.8754,
+      "step": 410430
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.794176168975281e-05,
+      "loss": 0.9285,
+      "step": 410440
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.7940613759853064e-05,
+      "loss": 0.8017,
+      "step": 410450
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.7939465829953314e-05,
+      "loss": 0.8989,
+      "step": 410460
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.793831790005357e-05,
+      "loss": 0.8553,
+      "step": 410470
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.793716997015382e-05,
+      "loss": 0.7876,
+      "step": 410480
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.793602204025407e-05,
+      "loss": 0.9243,
+      "step": 410490
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.7934874110354327e-05,
+      "loss": 0.7609,
+      "step": 410500
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.7933726180454577e-05,
+      "loss": 0.762,
+      "step": 410510
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.793257825055483e-05,
+      "loss": 0.8532,
+      "step": 410520
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.7931430320655083e-05,
+      "loss": 0.8241,
+      "step": 410530
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.7930282390755336e-05,
+      "loss": 0.9122,
+      "step": 410540
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.7929134460855586e-05,
+      "loss": 0.9167,
+      "step": 410550
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.792798653095584e-05,
+      "loss": 0.8786,
+      "step": 410560
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.7926838601056093e-05,
+      "loss": 0.8805,
+      "step": 410570
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.7925690671156346e-05,
+      "loss": 0.8894,
+      "step": 410580
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.7924542741256596e-05,
+      "loss": 0.7835,
+      "step": 410590
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.7923394811356853e-05,
+      "loss": 0.8527,
+      "step": 410600
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.7922246881457102e-05,
+      "loss": 0.9191,
+      "step": 410610
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.7921098951557352e-05,
+      "loss": 0.864,
+      "step": 410620
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.791995102165761e-05,
+      "loss": 0.8176,
+      "step": 410630
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.791880309175786e-05,
+      "loss": 0.9137,
+      "step": 410640
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.7917655161858112e-05,
+      "loss": 0.8975,
+      "step": 410650
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.7916507231958365e-05,
+      "loss": 0.8306,
+      "step": 410660
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.791535930205862e-05,
+      "loss": 0.7887,
+      "step": 410670
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.791421137215887e-05,
+      "loss": 0.9045,
+      "step": 410680
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.7913063442259125e-05,
+      "loss": 0.8743,
+      "step": 410690
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.7911915512359375e-05,
+      "loss": 0.8385,
+      "step": 410700
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.7910767582459628e-05,
+      "loss": 0.8292,
+      "step": 410710
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.790961965255988e-05,
+      "loss": 0.9405,
+      "step": 410720
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.7908471722660135e-05,
+      "loss": 0.8383,
+      "step": 410730
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.7907323792760384e-05,
+      "loss": 0.9082,
+      "step": 410740
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.790617586286064e-05,
+      "loss": 0.8822,
+      "step": 410750
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.790502793296089e-05,
+      "loss": 0.8384,
+      "step": 410760
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.790388000306114e-05,
+      "loss": 0.8415,
+      "step": 410770
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.7902732073161397e-05,
+      "loss": 0.9523,
+      "step": 410780
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.7901584143261647e-05,
+      "loss": 0.8822,
+      "step": 410790
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.79004362133619e-05,
+      "loss": 0.8481,
+      "step": 410800
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.7899288283462154e-05,
+      "loss": 0.9097,
+      "step": 410810
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.7898140353562407e-05,
+      "loss": 0.8566,
+      "step": 410820
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.7896992423662657e-05,
+      "loss": 0.8989,
+      "step": 410830
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.7895844493762913e-05,
+      "loss": 0.8448,
+      "step": 410840
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.7894696563863163e-05,
+      "loss": 0.7984,
+      "step": 410850
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.7893548633963417e-05,
+      "loss": 0.7909,
+      "step": 410860
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.789240070406367e-05,
+      "loss": 0.8579,
+      "step": 410870
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.7891252774163923e-05,
+      "loss": 0.8643,
+      "step": 410880
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.7890104844264173e-05,
+      "loss": 0.9157,
+      "step": 410890
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.788895691436443e-05,
+      "loss": 0.8332,
+      "step": 410900
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.788780898446468e-05,
+      "loss": 0.8278,
+      "step": 410910
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.788666105456493e-05,
+      "loss": 0.9091,
+      "step": 410920
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.7885513124665186e-05,
+      "loss": 0.8131,
+      "step": 410930
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.7884365194765436e-05,
+      "loss": 0.8226,
+      "step": 410940
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.788321726486569e-05,
+      "loss": 0.8802,
+      "step": 410950
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.7882069334965942e-05,
+      "loss": 0.8987,
+      "step": 410960
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.7880921405066195e-05,
+      "loss": 0.8855,
+      "step": 410970
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.7879773475166445e-05,
+      "loss": 0.8148,
+      "step": 410980
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.7878625545266702e-05,
+      "loss": 0.8653,
+      "step": 410990
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.7877477615366952e-05,
+      "loss": 0.8941,
+      "step": 411000
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.7876329685467205e-05,
+      "loss": 0.9877,
+      "step": 411010
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.7875181755567458e-05,
+      "loss": 0.8431,
+      "step": 411020
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.787403382566771e-05,
+      "loss": 0.819,
+      "step": 411030
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.787288589576796e-05,
+      "loss": 0.8741,
+      "step": 411040
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.7871737965868218e-05,
+      "loss": 0.9,
+      "step": 411050
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.7870590035968468e-05,
+      "loss": 0.8362,
+      "step": 411060
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.7869442106068718e-05,
+      "loss": 0.9592,
+      "step": 411070
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.7868294176168974e-05,
+      "loss": 0.934,
+      "step": 411080
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.7867146246269224e-05,
+      "loss": 0.8816,
+      "step": 411090
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.7865998316369477e-05,
+      "loss": 0.8819,
+      "step": 411100
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.786485038646973e-05,
+      "loss": 0.8008,
+      "step": 411110
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.7863702456569984e-05,
+      "loss": 0.7879,
+      "step": 411120
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.7862554526670234e-05,
+      "loss": 0.8689,
+      "step": 411130
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.786140659677049e-05,
+      "loss": 0.8617,
+      "step": 411140
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.786025866687074e-05,
+      "loss": 0.8409,
+      "step": 411150
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.7859110736970993e-05,
+      "loss": 0.8733,
+      "step": 411160
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.7857962807071247e-05,
+      "loss": 0.7969,
+      "step": 411170
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.78568148771715e-05,
+      "loss": 0.8659,
+      "step": 411180
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.785566694727175e-05,
+      "loss": 0.8173,
+      "step": 411190
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.7854519017372006e-05,
+      "loss": 0.8736,
+      "step": 411200
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.7853371087472256e-05,
+      "loss": 0.8944,
+      "step": 411210
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.7852223157572506e-05,
+      "loss": 0.8659,
+      "step": 411220
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.7851075227672763e-05,
+      "loss": 0.897,
+      "step": 411230
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.7849927297773013e-05,
+      "loss": 0.7982,
+      "step": 411240
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.7848779367873266e-05,
+      "loss": 0.8874,
+      "step": 411250
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.7847631437973516e-05,
+      "loss": 0.9061,
+      "step": 411260
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.7846483508073772e-05,
+      "loss": 0.91,
+      "step": 411270
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.7845335578174022e-05,
+      "loss": 0.8051,
+      "step": 411280
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.7844187648274275e-05,
+      "loss": 0.8836,
+      "step": 411290
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.784303971837453e-05,
+      "loss": 0.9609,
+      "step": 411300
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.7841891788474782e-05,
+      "loss": 0.884,
+      "step": 411310
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.7840743858575032e-05,
+      "loss": 0.8068,
+      "step": 411320
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.783959592867529e-05,
+      "loss": 0.9079,
+      "step": 411330
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.7838447998775538e-05,
+      "loss": 0.9034,
+      "step": 411340
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.7837300068875788e-05,
+      "loss": 0.8564,
+      "step": 411350
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.7836152138976045e-05,
+      "loss": 0.8833,
+      "step": 411360
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.7835004209076295e-05,
+      "loss": 0.9117,
+      "step": 411370
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.7833856279176548e-05,
+      "loss": 0.8176,
+      "step": 411380
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.78327083492768e-05,
+      "loss": 0.8966,
+      "step": 411390
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.7831560419377054e-05,
+      "loss": 0.7512,
+      "step": 411400
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.7830412489477304e-05,
+      "loss": 0.864,
+      "step": 411410
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.782926455957756e-05,
+      "loss": 0.8956,
+      "step": 411420
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.782811662967781e-05,
+      "loss": 0.8732,
+      "step": 411430
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.7826968699778064e-05,
+      "loss": 0.8956,
+      "step": 411440
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.7825820769878317e-05,
+      "loss": 0.8322,
+      "step": 411450
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.782467283997857e-05,
+      "loss": 0.9197,
+      "step": 411460
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.782352491007882e-05,
+      "loss": 0.874,
+      "step": 411470
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.7822376980179077e-05,
+      "loss": 0.8299,
+      "step": 411480
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.7821229050279327e-05,
+      "loss": 0.92,
+      "step": 411490
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.7820081120379577e-05,
+      "loss": 0.8684,
+      "step": 411500
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.7818933190479833e-05,
+      "loss": 0.8428,
+      "step": 411510
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.7817785260580083e-05,
+      "loss": 0.8983,
+      "step": 411520
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.7816637330680336e-05,
+      "loss": 0.8586,
+      "step": 411530
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.781548940078059e-05,
+      "loss": 0.8069,
+      "step": 411540
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.7814341470880843e-05,
+      "loss": 0.864,
+      "step": 411550
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.7813193540981093e-05,
+      "loss": 0.8679,
+      "step": 411560
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.781204561108135e-05,
+      "loss": 0.8848,
+      "step": 411570
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.78108976811816e-05,
+      "loss": 0.8516,
+      "step": 411580
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.7809749751281852e-05,
+      "loss": 0.8399,
+      "step": 411590
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.7808601821382106e-05,
+      "loss": 0.8658,
+      "step": 411600
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.780745389148236e-05,
+      "loss": 0.8393,
+      "step": 411610
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.780630596158261e-05,
+      "loss": 0.868,
+      "step": 411620
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.7805158031682865e-05,
+      "loss": 0.8252,
+      "step": 411630
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.7804010101783115e-05,
+      "loss": 0.8394,
+      "step": 411640
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.7802862171883365e-05,
+      "loss": 0.8133,
+      "step": 411650
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.780171424198362e-05,
+      "loss": 0.8561,
+      "step": 411660
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.780056631208387e-05,
+      "loss": 0.8609,
+      "step": 411670
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.7799418382184125e-05,
+      "loss": 0.861,
+      "step": 411680
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.7798270452284378e-05,
+      "loss": 0.8701,
+      "step": 411690
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.779712252238463e-05,
+      "loss": 0.8612,
+      "step": 411700
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.779597459248488e-05,
+      "loss": 0.8645,
+      "step": 411710
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.7794826662585138e-05,
+      "loss": 0.9049,
+      "step": 411720
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.7793678732685388e-05,
+      "loss": 0.8705,
+      "step": 411730
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.779253080278564e-05,
+      "loss": 0.8235,
+      "step": 411740
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.7791382872885894e-05,
+      "loss": 0.8401,
+      "step": 411750
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.7790234942986147e-05,
+      "loss": 0.8564,
+      "step": 411760
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.7789087013086397e-05,
+      "loss": 0.808,
+      "step": 411770
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.7787939083186654e-05,
+      "loss": 0.8052,
+      "step": 411780
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.7786791153286904e-05,
+      "loss": 0.7592,
+      "step": 411790
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.7785643223387154e-05,
+      "loss": 0.8213,
+      "step": 411800
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.778449529348741e-05,
+      "loss": 0.849,
+      "step": 411810
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.778334736358766e-05,
+      "loss": 0.8137,
+      "step": 411820
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.7782199433687913e-05,
+      "loss": 0.8688,
+      "step": 411830
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.7781051503788167e-05,
+      "loss": 0.8795,
+      "step": 411840
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.777990357388842e-05,
+      "loss": 0.8774,
+      "step": 411850
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.777875564398867e-05,
+      "loss": 0.8026,
+      "step": 411860
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.7777607714088926e-05,
+      "loss": 0.8271,
+      "step": 411870
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.7776459784189176e-05,
+      "loss": 0.8456,
+      "step": 411880
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.777531185428943e-05,
+      "loss": 0.8142,
+      "step": 411890
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.7774163924389683e-05,
+      "loss": 0.8563,
+      "step": 411900
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.7773015994489936e-05,
+      "loss": 0.8852,
+      "step": 411910
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.7771868064590186e-05,
+      "loss": 0.8491,
+      "step": 411920
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.7770720134690442e-05,
+      "loss": 0.8089,
+      "step": 411930
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.7769572204790692e-05,
+      "loss": 0.8501,
+      "step": 411940
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.7768424274890942e-05,
+      "loss": 0.7848,
+      "step": 411950
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.7767276344991195e-05,
+      "loss": 0.9007,
+      "step": 411960
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.776612841509145e-05,
+      "loss": 0.9001,
+      "step": 411970
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.7764980485191702e-05,
+      "loss": 0.7742,
+      "step": 411980
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.776383255529195e-05,
+      "loss": 0.8158,
+      "step": 411990
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.7762684625392208e-05,
+      "loss": 0.8655,
+      "step": 412000
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.7761536695492458e-05,
+      "loss": 0.8706,
+      "step": 412010
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.776038876559271e-05,
+      "loss": 0.8123,
+      "step": 412020
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.7759240835692965e-05,
+      "loss": 0.8828,
+      "step": 412030
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.7758092905793218e-05,
+      "loss": 0.8709,
+      "step": 412040
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.7756944975893468e-05,
+      "loss": 0.9202,
+      "step": 412050
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.7755797045993724e-05,
+      "loss": 0.8509,
+      "step": 412060
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.7754649116093974e-05,
+      "loss": 0.7861,
+      "step": 412070
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.7753501186194224e-05,
+      "loss": 0.857,
+      "step": 412080
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.775235325629448e-05,
+      "loss": 0.8253,
+      "step": 412090
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.775120532639473e-05,
+      "loss": 0.9344,
+      "step": 412100
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.7750057396494984e-05,
+      "loss": 0.8849,
+      "step": 412110
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.7748909466595237e-05,
+      "loss": 0.7923,
+      "step": 412120
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.774776153669549e-05,
+      "loss": 0.76,
+      "step": 412130
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.774661360679574e-05,
+      "loss": 0.7915,
+      "step": 412140
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.7745465676895997e-05,
+      "loss": 0.8785,
+      "step": 412150
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.7744317746996247e-05,
+      "loss": 0.9127,
+      "step": 412160
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.77431698170965e-05,
+      "loss": 0.8504,
+      "step": 412170
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.7742021887196753e-05,
+      "loss": 0.9168,
+      "step": 412180
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.7740873957297006e-05,
+      "loss": 0.8762,
+      "step": 412190
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.7739726027397256e-05,
+      "loss": 0.8077,
+      "step": 412200
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.7738578097497513e-05,
+      "loss": 0.815,
+      "step": 412210
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.7737430167597763e-05,
+      "loss": 0.8698,
+      "step": 412220
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.7736282237698012e-05,
+      "loss": 0.7982,
+      "step": 412230
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.773513430779827e-05,
+      "loss": 0.8873,
+      "step": 412240
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.773398637789852e-05,
+      "loss": 0.8219,
+      "step": 412250
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.7732838447998772e-05,
+      "loss": 0.9482,
+      "step": 412260
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.7731690518099025e-05,
+      "loss": 0.835,
+      "step": 412270
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.773054258819928e-05,
+      "loss": 0.8597,
+      "step": 412280
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.772939465829953e-05,
+      "loss": 0.7992,
+      "step": 412290
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.7728246728399785e-05,
+      "loss": 0.8887,
+      "step": 412300
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.7727098798500035e-05,
+      "loss": 0.8208,
+      "step": 412310
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.7725950868600288e-05,
+      "loss": 0.7594,
+      "step": 412320
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.772480293870054e-05,
+      "loss": 0.8912,
+      "step": 412330
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.7723655008800795e-05,
+      "loss": 0.8032,
+      "step": 412340
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.7722507078901045e-05,
+      "loss": 0.8042,
+      "step": 412350
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.77213591490013e-05,
+      "loss": 0.8659,
+      "step": 412360
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.772021121910155e-05,
+      "loss": 0.8735,
+      "step": 412370
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.77190632892018e-05,
+      "loss": 0.8869,
+      "step": 412380
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.7717915359302058e-05,
+      "loss": 0.9501,
+      "step": 412390
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.7716767429402307e-05,
+      "loss": 0.8859,
+      "step": 412400
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.771561949950256e-05,
+      "loss": 0.8338,
+      "step": 412410
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.7714471569602814e-05,
+      "loss": 0.9399,
+      "step": 412420
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.7713323639703067e-05,
+      "loss": 0.8134,
+      "step": 412430
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.7712175709803317e-05,
+      "loss": 0.9055,
+      "step": 412440
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.7711027779903574e-05,
+      "loss": 0.8739,
+      "step": 412450
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.7709879850003823e-05,
+      "loss": 0.8403,
+      "step": 412460
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.7708731920104077e-05,
+      "loss": 0.7854,
+      "step": 412470
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.770758399020433e-05,
+      "loss": 0.8225,
+      "step": 412480
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.7706436060304583e-05,
+      "loss": 0.9204,
+      "step": 412490
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.7705288130404833e-05,
+      "loss": 0.845,
+      "step": 412500
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.770414020050509e-05,
+      "loss": 0.7634,
+      "step": 412510
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.770299227060534e-05,
+      "loss": 0.8919,
+      "step": 412520
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.770184434070559e-05,
+      "loss": 0.7966,
+      "step": 412530
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.7700696410805846e-05,
+      "loss": 0.9095,
+      "step": 412540
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.7699548480906096e-05,
+      "loss": 0.9388,
+      "step": 412550
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.769840055100635e-05,
+      "loss": 0.9057,
+      "step": 412560
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.7697252621106602e-05,
+      "loss": 0.8874,
+      "step": 412570
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.7696104691206856e-05,
+      "loss": 0.8697,
+      "step": 412580
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.7694956761307105e-05,
+      "loss": 0.8704,
+      "step": 412590
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.7693808831407362e-05,
+      "loss": 0.8758,
+      "step": 412600
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.7692660901507612e-05,
+      "loss": 0.8467,
+      "step": 412610
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.7691512971607865e-05,
+      "loss": 0.8806,
+      "step": 412620
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.7690365041708115e-05,
+      "loss": 0.8723,
+      "step": 412630
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.768921711180837e-05,
+      "loss": 0.9433,
+      "step": 412640
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.768806918190862e-05,
+      "loss": 0.8521,
+      "step": 412650
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.768692125200887e-05,
+      "loss": 0.8469,
+      "step": 412660
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.7685773322109128e-05,
+      "loss": 0.8186,
+      "step": 412670
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.7684625392209378e-05,
+      "loss": 0.7792,
+      "step": 412680
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.768347746230963e-05,
+      "loss": 0.9098,
+      "step": 412690
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.7682329532409884e-05,
+      "loss": 0.8684,
+      "step": 412700
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.7681181602510138e-05,
+      "loss": 0.7636,
+      "step": 412710
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.7680033672610387e-05,
+      "loss": 0.841,
+      "step": 412720
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.7678885742710644e-05,
+      "loss": 0.8743,
+      "step": 412730
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.7677737812810894e-05,
+      "loss": 0.8978,
+      "step": 412740
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.7676589882911147e-05,
+      "loss": 0.8212,
+      "step": 412750
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.76754419530114e-05,
+      "loss": 0.854,
+      "step": 412760
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.7674294023111654e-05,
+      "loss": 0.8219,
+      "step": 412770
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.7673146093211904e-05,
+      "loss": 0.8749,
+      "step": 412780
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.767199816331216e-05,
+      "loss": 0.9329,
+      "step": 412790
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.767085023341241e-05,
+      "loss": 0.8326,
+      "step": 412800
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.766970230351266e-05,
+      "loss": 0.852,
+      "step": 412810
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.7668554373612916e-05,
+      "loss": 0.7766,
+      "step": 412820
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.7667406443713166e-05,
+      "loss": 0.9253,
+      "step": 412830
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.766625851381342e-05,
+      "loss": 0.8202,
+      "step": 412840
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.7665110583913673e-05,
+      "loss": 0.7911,
+      "step": 412850
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.7663962654013926e-05,
+      "loss": 0.8981,
+      "step": 412860
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.7662814724114176e-05,
+      "loss": 0.9002,
+      "step": 412870
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.7661666794214433e-05,
+      "loss": 0.8018,
+      "step": 412880
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.7660518864314682e-05,
+      "loss": 0.8405,
+      "step": 412890
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.7659370934414936e-05,
+      "loss": 0.82,
+      "step": 412900
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.765822300451519e-05,
+      "loss": 0.9316,
+      "step": 412910
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.7657075074615442e-05,
+      "loss": 0.8742,
+      "step": 412920
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.7655927144715692e-05,
+      "loss": 0.8149,
+      "step": 412930
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.765477921481595e-05,
+      "loss": 0.82,
+      "step": 412940
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.76536312849162e-05,
+      "loss": 0.8745,
+      "step": 412950
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.765248335501645e-05,
+      "loss": 0.8935,
+      "step": 412960
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.7651335425116705e-05,
+      "loss": 0.883,
+      "step": 412970
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.7650187495216955e-05,
+      "loss": 0.8716,
+      "step": 412980
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.7649039565317208e-05,
+      "loss": 0.8692,
+      "step": 412990
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.764789163541746e-05,
+      "loss": 0.8816,
+      "step": 413000
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.7646743705517715e-05,
+      "loss": 0.7809,
+      "step": 413010
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.7645595775617964e-05,
+      "loss": 0.8816,
+      "step": 413020
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.764444784571822e-05,
+      "loss": 0.8429,
+      "step": 413030
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.764329991581847e-05,
+      "loss": 0.797,
+      "step": 413040
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.7642151985918724e-05,
+      "loss": 0.9586,
+      "step": 413050
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.7641004056018977e-05,
+      "loss": 0.8905,
+      "step": 413060
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.763985612611923e-05,
+      "loss": 0.8563,
+      "step": 413070
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.763870819621948e-05,
+      "loss": 0.8576,
+      "step": 413080
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.7637560266319737e-05,
+      "loss": 0.8992,
+      "step": 413090
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.7636412336419987e-05,
+      "loss": 0.8369,
+      "step": 413100
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.7635264406520237e-05,
+      "loss": 0.8367,
+      "step": 413110
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.7634116476620493e-05,
+      "loss": 0.8635,
+      "step": 413120
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.7632968546720743e-05,
+      "loss": 0.9281,
+      "step": 413130
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.7631820616820997e-05,
+      "loss": 0.9453,
+      "step": 413140
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.763067268692125e-05,
+      "loss": 0.7978,
+      "step": 413150
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.7629524757021503e-05,
+      "loss": 0.8233,
+      "step": 413160
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.7628376827121753e-05,
+      "loss": 0.8605,
+      "step": 413170
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.762722889722201e-05,
+      "loss": 0.9425,
+      "step": 413180
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.762608096732226e-05,
+      "loss": 0.8736,
+      "step": 413190
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.7624933037422513e-05,
+      "loss": 0.9032,
+      "step": 413200
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.7623785107522766e-05,
+      "loss": 0.8713,
+      "step": 413210
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.762263717762302e-05,
+      "loss": 0.8524,
+      "step": 413220
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.762148924772327e-05,
+      "loss": 0.913,
+      "step": 413230
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.7620341317823526e-05,
+      "loss": 0.8248,
+      "step": 413240
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.7619193387923775e-05,
+      "loss": 0.912,
+      "step": 413250
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.7618045458024025e-05,
+      "loss": 0.7916,
+      "step": 413260
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.7616897528124282e-05,
+      "loss": 0.8353,
+      "step": 413270
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.7615749598224532e-05,
+      "loss": 0.8986,
+      "step": 413280
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.7614601668324785e-05,
+      "loss": 0.9117,
+      "step": 413290
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.7613453738425038e-05,
+      "loss": 0.8415,
+      "step": 413300
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.761230580852529e-05,
+      "loss": 0.891,
+      "step": 413310
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.761115787862554e-05,
+      "loss": 0.8976,
+      "step": 413320
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.7610009948725795e-05,
+      "loss": 0.847,
+      "step": 413330
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.7608862018826048e-05,
+      "loss": 0.9108,
+      "step": 413340
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.76077140889263e-05,
+      "loss": 0.818,
+      "step": 413350
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.760656615902655e-05,
+      "loss": 0.8189,
+      "step": 413360
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.7605418229126808e-05,
+      "loss": 0.8593,
+      "step": 413370
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.7604270299227057e-05,
+      "loss": 0.8292,
+      "step": 413380
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.7603122369327307e-05,
+      "loss": 0.7933,
+      "step": 413390
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.7601974439427564e-05,
+      "loss": 0.8089,
+      "step": 413400
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.7600826509527814e-05,
+      "loss": 0.9004,
+      "step": 413410
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.7599678579628067e-05,
+      "loss": 0.8192,
+      "step": 413420
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.759853064972832e-05,
+      "loss": 0.8603,
+      "step": 413430
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.7597382719828573e-05,
+      "loss": 0.8047,
+      "step": 413440
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.7596234789928823e-05,
+      "loss": 0.901,
+      "step": 413450
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.759508686002908e-05,
+      "loss": 0.902,
+      "step": 413460
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.759393893012933e-05,
+      "loss": 0.878,
+      "step": 413470
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.7592791000229583e-05,
+      "loss": 0.8956,
+      "step": 413480
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.7591643070329836e-05,
+      "loss": 0.8086,
+      "step": 413490
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.759049514043009e-05,
+      "loss": 0.9381,
+      "step": 413500
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.758934721053034e-05,
+      "loss": 0.8489,
+      "step": 413510
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.7588199280630596e-05,
+      "loss": 0.8574,
+      "step": 413520
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.7587051350730846e-05,
+      "loss": 0.8243,
+      "step": 413530
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.7585903420831096e-05,
+      "loss": 0.8269,
+      "step": 413540
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.7584755490931352e-05,
+      "loss": 0.8712,
+      "step": 413550
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.7583607561031602e-05,
+      "loss": 0.9893,
+      "step": 413560
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.7582459631131855e-05,
+      "loss": 0.8925,
+      "step": 413570
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.758131170123211e-05,
+      "loss": 0.8421,
+      "step": 413580
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.7580163771332362e-05,
+      "loss": 0.8549,
+      "step": 413590
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.7579015841432612e-05,
+      "loss": 0.816,
+      "step": 413600
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.757786791153287e-05,
+      "loss": 0.8561,
+      "step": 413610
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.7576719981633118e-05,
+      "loss": 0.8476,
+      "step": 413620
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.757557205173337e-05,
+      "loss": 0.9004,
+      "step": 413630
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.7574424121833625e-05,
+      "loss": 0.8292,
+      "step": 413640
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.7573276191933878e-05,
+      "loss": 0.9135,
+      "step": 413650
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.7572128262034128e-05,
+      "loss": 0.835,
+      "step": 413660
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.7570980332134385e-05,
+      "loss": 0.7822,
+      "step": 413670
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.7569832402234634e-05,
+      "loss": 0.8051,
+      "step": 413680
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.7568684472334884e-05,
+      "loss": 0.9041,
+      "step": 413690
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.756753654243514e-05,
+      "loss": 0.813,
+      "step": 413700
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.756638861253539e-05,
+      "loss": 0.8285,
+      "step": 413710
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.7565240682635644e-05,
+      "loss": 0.8874,
+      "step": 413720
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.7564092752735897e-05,
+      "loss": 0.825,
+      "step": 413730
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.756294482283615e-05,
+      "loss": 0.8486,
+      "step": 413740
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.75617968929364e-05,
+      "loss": 0.8595,
+      "step": 413750
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.7560648963036657e-05,
+      "loss": 0.8742,
+      "step": 413760
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.7559501033136907e-05,
+      "loss": 0.8638,
+      "step": 413770
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.755835310323716e-05,
+      "loss": 0.8546,
+      "step": 413780
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.7557205173337413e-05,
+      "loss": 0.9415,
+      "step": 413790
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.7556057243437666e-05,
+      "loss": 0.8506,
+      "step": 413800
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.7554909313537916e-05,
+      "loss": 0.8802,
+      "step": 413810
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.7553761383638173e-05,
+      "loss": 0.8391,
+      "step": 413820
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.7552613453738423e-05,
+      "loss": 0.8803,
+      "step": 413830
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.7551465523838673e-05,
+      "loss": 0.8576,
+      "step": 413840
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.755031759393893e-05,
+      "loss": 0.9207,
+      "step": 413850
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.754916966403918e-05,
+      "loss": 0.941,
+      "step": 413860
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.7548021734139432e-05,
+      "loss": 0.8204,
+      "step": 413870
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.7546873804239686e-05,
+      "loss": 0.8399,
+      "step": 413880
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.754572587433994e-05,
+      "loss": 0.8468,
+      "step": 413890
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.754457794444019e-05,
+      "loss": 0.8682,
+      "step": 413900
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.7543430014540445e-05,
+      "loss": 0.8191,
+      "step": 413910
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.7542282084640695e-05,
+      "loss": 0.8774,
+      "step": 413920
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.754113415474095e-05,
+      "loss": 0.8454,
+      "step": 413930
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.7539986224841202e-05,
+      "loss": 0.7662,
+      "step": 413940
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.7538838294941455e-05,
+      "loss": 0.8371,
+      "step": 413950
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.7537690365041705e-05,
+      "loss": 0.8259,
+      "step": 413960
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.753654243514196e-05,
+      "loss": 0.845,
+      "step": 413970
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.753539450524221e-05,
+      "loss": 0.8519,
+      "step": 413980
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.753424657534246e-05,
+      "loss": 0.8535,
+      "step": 413990
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.7533098645442714e-05,
+      "loss": 0.8225,
+      "step": 414000
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.7531950715542968e-05,
+      "loss": 0.8728,
+      "step": 414010
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.753080278564322e-05,
+      "loss": 0.9107,
+      "step": 414020
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.752965485574347e-05,
+      "loss": 0.825,
+      "step": 414030
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.7528506925843727e-05,
+      "loss": 0.8788,
+      "step": 414040
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.7527358995943977e-05,
+      "loss": 0.8361,
+      "step": 414050
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.752621106604423e-05,
+      "loss": 0.8328,
+      "step": 414060
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.7525063136144484e-05,
+      "loss": 0.871,
+      "step": 414070
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.7523915206244737e-05,
+      "loss": 0.8066,
+      "step": 414080
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.7522767276344987e-05,
+      "loss": 0.8547,
+      "step": 414090
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.7521619346445243e-05,
+      "loss": 0.8788,
+      "step": 414100
+    },
+    {
+      "epoch": 19.0,
+      "eval_cer": 0.6686694242111152,
+      "eval_em": 0.007357212003872217,
+      "eval_f1": 0.007357212003872217,
+      "eval_loss": 1.2890383005142212,
+      "eval_runtime": 1582.6676,
+      "eval_samples_per_second": 6.527,
+      "eval_steps_per_second": 0.816,
+      "eval_wer": 0.9926427879961278,
+      "step": 414109
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.7520471416545493e-05,
+      "loss": 0.8454,
+      "step": 414110
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.7519323486645743e-05,
+      "loss": 0.7388,
+      "step": 414120
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.7518175556746e-05,
+      "loss": 0.8484,
+      "step": 414130
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.751702762684625e-05,
+      "loss": 0.8203,
+      "step": 414140
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.7515879696946503e-05,
+      "loss": 0.7803,
+      "step": 414150
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.7514731767046756e-05,
+      "loss": 0.8412,
+      "step": 414160
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.751358383714701e-05,
+      "loss": 0.7694,
+      "step": 414170
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.751243590724726e-05,
+      "loss": 0.789,
+      "step": 414180
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.7511287977347516e-05,
+      "loss": 0.8205,
+      "step": 414190
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.7510140047447766e-05,
+      "loss": 0.8146,
+      "step": 414200
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.750899211754802e-05,
+      "loss": 0.7963,
+      "step": 414210
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.7507844187648272e-05,
+      "loss": 0.8007,
+      "step": 414220
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.7506696257748525e-05,
+      "loss": 0.7641,
+      "step": 414230
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.7505548327848775e-05,
+      "loss": 0.7715,
+      "step": 414240
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.7504400397949032e-05,
+      "loss": 0.8097,
+      "step": 414250
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.7503252468049282e-05,
+      "loss": 0.8079,
+      "step": 414260
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.750210453814953e-05,
+      "loss": 0.7554,
+      "step": 414270
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.7500956608249788e-05,
+      "loss": 0.8661,
+      "step": 414280
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.7499808678350038e-05,
+      "loss": 0.8197,
+      "step": 414290
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.749866074845029e-05,
+      "loss": 0.8462,
+      "step": 414300
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.7497512818550545e-05,
+      "loss": 0.8307,
+      "step": 414310
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.7496364888650798e-05,
+      "loss": 0.8271,
+      "step": 414320
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.7495216958751048e-05,
+      "loss": 0.7745,
+      "step": 414330
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.7494069028851304e-05,
+      "loss": 0.8914,
+      "step": 414340
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.7492921098951554e-05,
+      "loss": 0.8724,
+      "step": 414350
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.7491773169051807e-05,
+      "loss": 0.9213,
+      "step": 414360
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.749062523915206e-05,
+      "loss": 0.8516,
+      "step": 414370
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.7489477309252314e-05,
+      "loss": 0.7761,
+      "step": 414380
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.7488329379352564e-05,
+      "loss": 0.83,
+      "step": 414390
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.748718144945282e-05,
+      "loss": 0.7972,
+      "step": 414400
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.748603351955307e-05,
+      "loss": 0.8881,
+      "step": 414410
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.748488558965332e-05,
+      "loss": 0.7752,
+      "step": 414420
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 2.7483737659753577e-05,
+      "loss": 0.8327,
+      "step": 414430
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.7482589729853827e-05,
+      "loss": 0.7671,
+      "step": 414440
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.748144179995408e-05,
+      "loss": 0.8168,
+      "step": 414450
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.7480293870054333e-05,
+      "loss": 0.9196,
+      "step": 414460
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.7479145940154586e-05,
+      "loss": 0.8506,
+      "step": 414470
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.7477998010254836e-05,
+      "loss": 0.8037,
+      "step": 414480
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.7476850080355093e-05,
+      "loss": 0.8211,
+      "step": 414490
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.7475702150455343e-05,
+      "loss": 0.7569,
+      "step": 414500
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.7474554220555596e-05,
+      "loss": 0.7339,
+      "step": 414510
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.747340629065585e-05,
+      "loss": 0.7209,
+      "step": 414520
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.7472258360756102e-05,
+      "loss": 0.8485,
+      "step": 414530
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.7471110430856352e-05,
+      "loss": 0.7352,
+      "step": 414540
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.746996250095661e-05,
+      "loss": 0.7851,
+      "step": 414550
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.746881457105686e-05,
+      "loss": 0.8494,
+      "step": 414560
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.746766664115711e-05,
+      "loss": 0.8658,
+      "step": 414570
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.7466518711257365e-05,
+      "loss": 0.807,
+      "step": 414580
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.7465370781357615e-05,
+      "loss": 0.8327,
+      "step": 414590
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.7464222851457868e-05,
+      "loss": 0.8618,
+      "step": 414600
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.746307492155812e-05,
+      "loss": 0.8545,
+      "step": 414610
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.7461926991658375e-05,
+      "loss": 0.8593,
+      "step": 414620
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.7460779061758625e-05,
+      "loss": 0.8577,
+      "step": 414630
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.745963113185888e-05,
+      "loss": 0.8306,
+      "step": 414640
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.745848320195913e-05,
+      "loss": 0.8289,
+      "step": 414650
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.7457335272059384e-05,
+      "loss": 0.8675,
+      "step": 414660
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.7456187342159638e-05,
+      "loss": 0.8643,
+      "step": 414670
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.745503941225989e-05,
+      "loss": 0.7962,
+      "step": 414680
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.745389148236014e-05,
+      "loss": 0.8332,
+      "step": 414690
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.745274355246039e-05,
+      "loss": 0.8351,
+      "step": 414700
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.7451595622560647e-05,
+      "loss": 0.8566,
+      "step": 414710
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.7450447692660897e-05,
+      "loss": 0.8628,
+      "step": 414720
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.744929976276115e-05,
+      "loss": 0.8091,
+      "step": 414730
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.7448151832861404e-05,
+      "loss": 0.771,
+      "step": 414740
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.7447003902961657e-05,
+      "loss": 0.8056,
+      "step": 414750
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.7445855973061907e-05,
+      "loss": 0.8327,
+      "step": 414760
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.7444708043162163e-05,
+      "loss": 0.8395,
+      "step": 414770
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.7443560113262413e-05,
+      "loss": 0.8801,
+      "step": 414780
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.7442412183362666e-05,
+      "loss": 0.8496,
+      "step": 414790
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.744126425346292e-05,
+      "loss": 0.7798,
+      "step": 414800
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.7440116323563173e-05,
+      "loss": 0.8291,
+      "step": 414810
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.7438968393663423e-05,
+      "loss": 0.7393,
+      "step": 414820
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.743782046376368e-05,
+      "loss": 0.7352,
+      "step": 414830
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.743667253386393e-05,
+      "loss": 0.8424,
+      "step": 414840
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.743552460396418e-05,
+      "loss": 0.811,
+      "step": 414850
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.7434376674064436e-05,
+      "loss": 0.7777,
+      "step": 414860
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 2.7433228744164686e-05,
+      "loss": 0.8796,
+      "step": 414870
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 2.743208081426494e-05,
+      "loss": 0.8379,
+      "step": 414880
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 2.7430932884365192e-05,
+      "loss": 0.833,
+      "step": 414890
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 2.7429784954465445e-05,
+      "loss": 0.8998,
+      "step": 414900
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 2.7428637024565695e-05,
+      "loss": 0.8071,
+      "step": 414910
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 2.7427489094665952e-05,
+      "loss": 0.6988,
+      "step": 414920
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 2.74263411647662e-05,
+      "loss": 0.8181,
+      "step": 414930
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 2.7425193234866455e-05,
+      "loss": 0.8156,
+      "step": 414940
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 2.7424045304966708e-05,
+      "loss": 0.815,
+      "step": 414950
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 2.742289737506696e-05,
+      "loss": 0.8578,
+      "step": 414960
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 2.742174944516721e-05,
+      "loss": 0.8233,
+      "step": 414970
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 2.7420601515267468e-05,
+      "loss": 0.8245,
+      "step": 414980
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 2.7419453585367718e-05,
+      "loss": 0.8351,
+      "step": 414990
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 2.7418305655467967e-05,
+      "loss": 0.8827,
+      "step": 415000
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 2.7417157725568224e-05,
+      "loss": 0.832,
+      "step": 415010
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 2.7416009795668474e-05,
+      "loss": 0.8198,
+      "step": 415020
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 2.7414861865768727e-05,
+      "loss": 0.7611,
+      "step": 415030
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 2.741371393586898e-05,
+      "loss": 0.8433,
+      "step": 415040
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 2.7412566005969234e-05,
+      "loss": 0.8229,
+      "step": 415050
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 2.7411418076069484e-05,
+      "loss": 0.886,
+      "step": 415060
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 2.741027014616974e-05,
+      "loss": 0.8162,
+      "step": 415070
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 2.740912221626999e-05,
+      "loss": 0.8068,
+      "step": 415080
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 2.7407974286370243e-05,
+      "loss": 0.834,
+      "step": 415090
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 2.7406826356470497e-05,
+      "loss": 0.878,
+      "step": 415100
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 2.740567842657075e-05,
+      "loss": 0.8914,
+      "step": 415110
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 2.7404530496671e-05,
+      "loss": 0.814,
+      "step": 415120
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 2.7403382566771256e-05,
+      "loss": 0.8061,
+      "step": 415130
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 2.7402234636871506e-05,
+      "loss": 0.8207,
+      "step": 415140
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 2.7401086706971756e-05,
+      "loss": 0.8279,
+      "step": 415150
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 2.7399938777072013e-05,
+      "loss": 0.8454,
+      "step": 415160
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 2.7398790847172262e-05,
+      "loss": 0.7843,
+      "step": 415170
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 2.7397642917272516e-05,
+      "loss": 0.8116,
+      "step": 415180
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 2.739649498737277e-05,
+      "loss": 0.8031,
+      "step": 415190
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 2.7395347057473022e-05,
+      "loss": 0.7775,
+      "step": 415200
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 2.7394199127573272e-05,
+      "loss": 0.7655,
+      "step": 415210
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 2.739305119767353e-05,
+      "loss": 0.7986,
+      "step": 415220
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 2.739190326777378e-05,
+      "loss": 0.9113,
+      "step": 415230
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 2.7390755337874032e-05,
+      "loss": 0.8485,
+      "step": 415240
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 2.7389607407974285e-05,
+      "loss": 0.7934,
+      "step": 415250
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 2.7388459478074538e-05,
+      "loss": 0.863,
+      "step": 415260
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 2.7387311548174788e-05,
+      "loss": 0.7657,
+      "step": 415270
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 2.7386163618275045e-05,
+      "loss": 0.7708,
+      "step": 415280
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 2.7385015688375295e-05,
+      "loss": 0.7973,
+      "step": 415290
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 2.7383867758475544e-05,
+      "loss": 0.7885,
+      "step": 415300
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 2.73827198285758e-05,
+      "loss": 0.7628,
+      "step": 415310
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 2.738157189867605e-05,
+      "loss": 0.732,
+      "step": 415320
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 2.7380423968776304e-05,
+      "loss": 0.8947,
+      "step": 415330
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 2.7379276038876557e-05,
+      "loss": 0.7888,
+      "step": 415340
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 2.737812810897681e-05,
+      "loss": 0.8723,
+      "step": 415350
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 2.737698017907706e-05,
+      "loss": 0.8024,
+      "step": 415360
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 2.7375832249177314e-05,
+      "loss": 0.9006,
+      "step": 415370
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 2.7374684319277567e-05,
+      "loss": 0.7456,
+      "step": 415380
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 2.737353638937782e-05,
+      "loss": 0.771,
+      "step": 415390
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 2.737238845947807e-05,
+      "loss": 0.8138,
+      "step": 415400
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 2.7371240529578327e-05,
+      "loss": 0.9052,
+      "step": 415410
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 2.7370092599678577e-05,
+      "loss": 0.8116,
+      "step": 415420
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 2.7368944669778826e-05,
+      "loss": 0.7722,
+      "step": 415430
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 2.7367796739879083e-05,
+      "loss": 0.8106,
+      "step": 415440
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 2.7366648809979333e-05,
+      "loss": 0.7765,
+      "step": 415450
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 2.7365500880079586e-05,
+      "loss": 0.8297,
+      "step": 415460
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 2.736435295017984e-05,
+      "loss": 0.7408,
+      "step": 415470
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 2.7363205020280093e-05,
+      "loss": 0.8477,
+      "step": 415480
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 2.7362057090380342e-05,
+      "loss": 0.8494,
+      "step": 415490
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 2.73609091604806e-05,
+      "loss": 0.7673,
+      "step": 415500
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 2.735976123058085e-05,
+      "loss": 0.7936,
+      "step": 415510
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 2.7358613300681102e-05,
+      "loss": 0.7868,
+      "step": 415520
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.7357465370781355e-05,
+      "loss": 0.8384,
+      "step": 415530
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.735631744088161e-05,
+      "loss": 0.8995,
+      "step": 415540
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.735516951098186e-05,
+      "loss": 0.8007,
+      "step": 415550
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.7354021581082115e-05,
+      "loss": 0.8205,
+      "step": 415560
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.7352873651182365e-05,
+      "loss": 0.8457,
+      "step": 415570
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.7351725721282615e-05,
+      "loss": 0.8551,
+      "step": 415580
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.735057779138287e-05,
+      "loss": 0.7652,
+      "step": 415590
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.734942986148312e-05,
+      "loss": 0.9064,
+      "step": 415600
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.7348281931583375e-05,
+      "loss": 0.7967,
+      "step": 415610
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.7347134001683628e-05,
+      "loss": 0.8378,
+      "step": 415620
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.734598607178388e-05,
+      "loss": 0.9099,
+      "step": 415630
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.734483814188413e-05,
+      "loss": 0.9002,
+      "step": 415640
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.7343690211984388e-05,
+      "loss": 0.7931,
+      "step": 415650
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.7342542282084637e-05,
+      "loss": 0.7616,
+      "step": 415660
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.734139435218489e-05,
+      "loss": 0.8436,
+      "step": 415670
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.7340246422285144e-05,
+      "loss": 0.8465,
+      "step": 415680
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.7339098492385397e-05,
+      "loss": 0.8112,
+      "step": 415690
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.7337950562485647e-05,
+      "loss": 0.7915,
+      "step": 415700
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.7336802632585904e-05,
+      "loss": 0.7535,
+      "step": 415710
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.7335654702686154e-05,
+      "loss": 0.8339,
+      "step": 415720
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.7334506772786403e-05,
+      "loss": 0.7898,
+      "step": 415730
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.733335884288666e-05,
+      "loss": 0.7744,
+      "step": 415740
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.733221091298691e-05,
+      "loss": 0.8197,
+      "step": 415750
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.7331062983087163e-05,
+      "loss": 0.7256,
+      "step": 415760
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.7329915053187416e-05,
+      "loss": 0.8724,
+      "step": 415770
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.732876712328767e-05,
+      "loss": 0.7201,
+      "step": 415780
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.732761919338792e-05,
+      "loss": 0.7672,
+      "step": 415790
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.7326471263488176e-05,
+      "loss": 0.8386,
+      "step": 415800
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.7325323333588426e-05,
+      "loss": 0.8739,
+      "step": 415810
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.732417540368868e-05,
+      "loss": 0.7134,
+      "step": 415820
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.7323027473788932e-05,
+      "loss": 0.8347,
+      "step": 415830
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.7321879543889186e-05,
+      "loss": 0.8438,
+      "step": 415840
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.7320731613989436e-05,
+      "loss": 0.7983,
+      "step": 415850
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.7319583684089692e-05,
+      "loss": 0.8134,
+      "step": 415860
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.7318435754189942e-05,
+      "loss": 0.8102,
+      "step": 415870
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.7317287824290192e-05,
+      "loss": 0.7793,
+      "step": 415880
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.731613989439045e-05,
+      "loss": 0.7765,
+      "step": 415890
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.73149919644907e-05,
+      "loss": 0.8561,
+      "step": 415900
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.731384403459095e-05,
+      "loss": 0.8267,
+      "step": 415910
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.7312696104691205e-05,
+      "loss": 0.8609,
+      "step": 415920
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.7311548174791458e-05,
+      "loss": 0.9077,
+      "step": 415930
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.7310400244891708e-05,
+      "loss": 0.7875,
+      "step": 415940
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.7309252314991965e-05,
+      "loss": 0.9289,
+      "step": 415950
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 2.7308104385092214e-05,
+      "loss": 0.7757,
+      "step": 415960
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.7306956455192468e-05,
+      "loss": 0.8109,
+      "step": 415970
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.730580852529272e-05,
+      "loss": 0.8452,
+      "step": 415980
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.7304660595392974e-05,
+      "loss": 0.7793,
+      "step": 415990
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.7303512665493224e-05,
+      "loss": 0.7805,
+      "step": 416000
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.730236473559348e-05,
+      "loss": 0.8448,
+      "step": 416010
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.730121680569373e-05,
+      "loss": 0.8611,
+      "step": 416020
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.730006887579398e-05,
+      "loss": 0.8264,
+      "step": 416030
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.7298920945894237e-05,
+      "loss": 0.8646,
+      "step": 416040
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.7297773015994487e-05,
+      "loss": 0.7656,
+      "step": 416050
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.729662508609474e-05,
+      "loss": 0.7799,
+      "step": 416060
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.729547715619499e-05,
+      "loss": 0.7854,
+      "step": 416070
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.7294329226295247e-05,
+      "loss": 0.8389,
+      "step": 416080
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.7293181296395496e-05,
+      "loss": 0.8618,
+      "step": 416090
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.729203336649575e-05,
+      "loss": 0.8125,
+      "step": 416100
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.7290885436596003e-05,
+      "loss": 0.797,
+      "step": 416110
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.7289737506696256e-05,
+      "loss": 0.8022,
+      "step": 416120
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.7288589576796506e-05,
+      "loss": 0.9064,
+      "step": 416130
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.7287441646896763e-05,
+      "loss": 0.9044,
+      "step": 416140
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.7286293716997012e-05,
+      "loss": 0.7273,
+      "step": 416150
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.7285145787097262e-05,
+      "loss": 0.8171,
+      "step": 416160
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.728399785719752e-05,
+      "loss": 0.8586,
+      "step": 416170
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.728284992729777e-05,
+      "loss": 0.8008,
+      "step": 416180
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.7281701997398022e-05,
+      "loss": 0.8072,
+      "step": 416190
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.7280554067498275e-05,
+      "loss": 0.9103,
+      "step": 416200
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.727940613759853e-05,
+      "loss": 0.7904,
+      "step": 416210
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.727825820769878e-05,
+      "loss": 0.7775,
+      "step": 416220
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.7277110277799035e-05,
+      "loss": 0.8404,
+      "step": 416230
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.7275962347899285e-05,
+      "loss": 0.8386,
+      "step": 416240
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.7274814417999538e-05,
+      "loss": 0.7676,
+      "step": 416250
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.727366648809979e-05,
+      "loss": 0.7447,
+      "step": 416260
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.7272518558200045e-05,
+      "loss": 0.7501,
+      "step": 416270
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.7271370628300294e-05,
+      "loss": 0.7997,
+      "step": 416280
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.727022269840055e-05,
+      "loss": 0.8921,
+      "step": 416290
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.72690747685008e-05,
+      "loss": 0.8219,
+      "step": 416300
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.726792683860105e-05,
+      "loss": 0.8415,
+      "step": 416310
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.7266778908701307e-05,
+      "loss": 0.8194,
+      "step": 416320
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.7265630978801557e-05,
+      "loss": 0.8135,
+      "step": 416330
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.726448304890181e-05,
+      "loss": 0.8286,
+      "step": 416340
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.7263335119002064e-05,
+      "loss": 0.8157,
+      "step": 416350
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.7262187189102317e-05,
+      "loss": 0.7201,
+      "step": 416360
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.7261039259202567e-05,
+      "loss": 0.8211,
+      "step": 416370
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.7259891329302823e-05,
+      "loss": 0.6783,
+      "step": 416380
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 2.7258743399403073e-05,
+      "loss": 0.8289,
+      "step": 416390
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.7257595469503327e-05,
+      "loss": 0.786,
+      "step": 416400
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.725644753960358e-05,
+      "loss": 0.8582,
+      "step": 416410
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.7255299609703833e-05,
+      "loss": 0.7717,
+      "step": 416420
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.7254151679804083e-05,
+      "loss": 0.8879,
+      "step": 416430
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.725300374990434e-05,
+      "loss": 0.8519,
+      "step": 416440
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.725185582000459e-05,
+      "loss": 0.871,
+      "step": 416450
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.725070789010484e-05,
+      "loss": 0.8819,
+      "step": 416460
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.7249559960205096e-05,
+      "loss": 0.8248,
+      "step": 416470
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.7248412030305346e-05,
+      "loss": 0.8047,
+      "step": 416480
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.72472641004056e-05,
+      "loss": 0.8288,
+      "step": 416490
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.7246116170505852e-05,
+      "loss": 0.7656,
+      "step": 416500
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.7244968240606105e-05,
+      "loss": 0.8625,
+      "step": 416510
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.7243820310706355e-05,
+      "loss": 0.7973,
+      "step": 416520
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.7242672380806612e-05,
+      "loss": 0.823,
+      "step": 416530
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.7241524450906862e-05,
+      "loss": 0.8184,
+      "step": 416540
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.7240376521007115e-05,
+      "loss": 0.738,
+      "step": 416550
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.7239228591107368e-05,
+      "loss": 0.7506,
+      "step": 416560
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.723808066120762e-05,
+      "loss": 0.7994,
+      "step": 416570
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.723693273130787e-05,
+      "loss": 0.6973,
+      "step": 416580
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.7235784801408128e-05,
+      "loss": 0.7868,
+      "step": 416590
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.7234636871508378e-05,
+      "loss": 0.8734,
+      "step": 416600
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.7233488941608628e-05,
+      "loss": 0.7507,
+      "step": 416610
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.7232341011708884e-05,
+      "loss": 0.8288,
+      "step": 416620
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.7231193081809134e-05,
+      "loss": 0.8812,
+      "step": 416630
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.7230045151909387e-05,
+      "loss": 0.8101,
+      "step": 416640
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.722889722200964e-05,
+      "loss": 0.8336,
+      "step": 416650
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.7227749292109894e-05,
+      "loss": 0.748,
+      "step": 416660
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.7226601362210144e-05,
+      "loss": 0.7747,
+      "step": 416670
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.72254534323104e-05,
+      "loss": 0.8356,
+      "step": 416680
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.722430550241065e-05,
+      "loss": 0.8583,
+      "step": 416690
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.7223157572510904e-05,
+      "loss": 0.8737,
+      "step": 416700
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.7222009642611157e-05,
+      "loss": 0.7835,
+      "step": 416710
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.722086171271141e-05,
+      "loss": 0.7901,
+      "step": 416720
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.721971378281166e-05,
+      "loss": 0.8194,
+      "step": 416730
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.721856585291191e-05,
+      "loss": 0.8085,
+      "step": 416740
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.7217417923012166e-05,
+      "loss": 0.8633,
+      "step": 416750
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.7216269993112416e-05,
+      "loss": 0.8474,
+      "step": 416760
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.721512206321267e-05,
+      "loss": 0.82,
+      "step": 416770
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.7213974133312923e-05,
+      "loss": 0.8719,
+      "step": 416780
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.7212826203413176e-05,
+      "loss": 0.8473,
+      "step": 416790
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.7211678273513426e-05,
+      "loss": 0.8258,
+      "step": 416800
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.7210530343613682e-05,
+      "loss": 0.7335,
+      "step": 416810
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.7209382413713932e-05,
+      "loss": 0.8027,
+      "step": 416820
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 2.7208234483814185e-05,
+      "loss": 0.7591,
+      "step": 416830
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.720708655391444e-05,
+      "loss": 0.7551,
+      "step": 416840
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.7205938624014692e-05,
+      "loss": 0.7929,
+      "step": 416850
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.7204790694114942e-05,
+      "loss": 0.8159,
+      "step": 416860
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.72036427642152e-05,
+      "loss": 0.7455,
+      "step": 416870
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.720249483431545e-05,
+      "loss": 0.7474,
+      "step": 416880
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.7201346904415698e-05,
+      "loss": 0.925,
+      "step": 416890
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.7200198974515955e-05,
+      "loss": 0.7797,
+      "step": 416900
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.7199051044616205e-05,
+      "loss": 0.7752,
+      "step": 416910
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.7197903114716458e-05,
+      "loss": 0.8347,
+      "step": 416920
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.719675518481671e-05,
+      "loss": 0.8496,
+      "step": 416930
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.7195607254916964e-05,
+      "loss": 0.8845,
+      "step": 416940
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.7194459325017214e-05,
+      "loss": 0.8818,
+      "step": 416950
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.719331139511747e-05,
+      "loss": 0.7204,
+      "step": 416960
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.719216346521772e-05,
+      "loss": 0.7718,
+      "step": 416970
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.7191015535317974e-05,
+      "loss": 0.8141,
+      "step": 416980
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.7189867605418227e-05,
+      "loss": 0.9123,
+      "step": 416990
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.718871967551848e-05,
+      "loss": 0.779,
+      "step": 417000
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.718757174561873e-05,
+      "loss": 0.9032,
+      "step": 417010
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.7186423815718987e-05,
+      "loss": 0.7712,
+      "step": 417020
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.7185275885819237e-05,
+      "loss": 0.9299,
+      "step": 417030
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.7184127955919487e-05,
+      "loss": 0.7892,
+      "step": 417040
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.7182980026019743e-05,
+      "loss": 0.8479,
+      "step": 417050
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.7181832096119993e-05,
+      "loss": 0.8285,
+      "step": 417060
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.7180684166220246e-05,
+      "loss": 0.8674,
+      "step": 417070
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.71795362363205e-05,
+      "loss": 0.8056,
+      "step": 417080
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.7178388306420753e-05,
+      "loss": 0.876,
+      "step": 417090
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.7177240376521003e-05,
+      "loss": 0.8333,
+      "step": 417100
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.717609244662126e-05,
+      "loss": 0.7723,
+      "step": 417110
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.717494451672151e-05,
+      "loss": 0.8243,
+      "step": 417120
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.7173796586821762e-05,
+      "loss": 0.8087,
+      "step": 417130
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.7172648656922016e-05,
+      "loss": 0.7468,
+      "step": 417140
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.717150072702227e-05,
+      "loss": 0.6765,
+      "step": 417150
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.717035279712252e-05,
+      "loss": 0.8434,
+      "step": 417160
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.7169204867222775e-05,
+      "loss": 0.8261,
+      "step": 417170
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.7168056937323025e-05,
+      "loss": 0.8595,
+      "step": 417180
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.7166909007423275e-05,
+      "loss": 0.8209,
+      "step": 417190
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.7165761077523532e-05,
+      "loss": 0.8488,
+      "step": 417200
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.716461314762378e-05,
+      "loss": 0.8381,
+      "step": 417210
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.7163465217724035e-05,
+      "loss": 0.8539,
+      "step": 417220
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.7162317287824288e-05,
+      "loss": 0.808,
+      "step": 417230
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.716116935792454e-05,
+      "loss": 0.7021,
+      "step": 417240
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.716002142802479e-05,
+      "loss": 0.8535,
+      "step": 417250
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 2.7158873498125048e-05,
+      "loss": 0.7857,
+      "step": 417260
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.7157725568225298e-05,
+      "loss": 0.7866,
+      "step": 417270
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.715657763832555e-05,
+      "loss": 0.9181,
+      "step": 417280
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.7155429708425804e-05,
+      "loss": 0.8029,
+      "step": 417290
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.7154281778526057e-05,
+      "loss": 0.8327,
+      "step": 417300
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.7153133848626307e-05,
+      "loss": 0.8088,
+      "step": 417310
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.7151985918726564e-05,
+      "loss": 0.7198,
+      "step": 417320
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.7150837988826814e-05,
+      "loss": 0.7858,
+      "step": 417330
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.7149690058927064e-05,
+      "loss": 0.7592,
+      "step": 417340
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.714854212902732e-05,
+      "loss": 0.8333,
+      "step": 417350
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.714739419912757e-05,
+      "loss": 0.7847,
+      "step": 417360
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.7146246269227823e-05,
+      "loss": 0.7913,
+      "step": 417370
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.7145098339328077e-05,
+      "loss": 0.8523,
+      "step": 417380
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.714395040942833e-05,
+      "loss": 0.8173,
+      "step": 417390
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.714280247952858e-05,
+      "loss": 0.7928,
+      "step": 417400
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.7141654549628836e-05,
+      "loss": 0.8319,
+      "step": 417410
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.7140506619729086e-05,
+      "loss": 0.8809,
+      "step": 417420
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.713935868982934e-05,
+      "loss": 0.7843,
+      "step": 417430
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.713821075992959e-05,
+      "loss": 0.8461,
+      "step": 417440
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.7137062830029846e-05,
+      "loss": 0.8341,
+      "step": 417450
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.7135914900130096e-05,
+      "loss": 0.853,
+      "step": 417460
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.7134766970230346e-05,
+      "loss": 0.8636,
+      "step": 417470
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.7133619040330602e-05,
+      "loss": 0.8248,
+      "step": 417480
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.7132471110430852e-05,
+      "loss": 0.8862,
+      "step": 417490
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.7131323180531105e-05,
+      "loss": 0.8153,
+      "step": 417500
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.713017525063136e-05,
+      "loss": 0.7633,
+      "step": 417510
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.7129027320731612e-05,
+      "loss": 0.8144,
+      "step": 417520
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.712787939083186e-05,
+      "loss": 0.7343,
+      "step": 417530
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.7126731460932118e-05,
+      "loss": 0.7741,
+      "step": 417540
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.7125583531032368e-05,
+      "loss": 0.7616,
+      "step": 417550
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.712443560113262e-05,
+      "loss": 0.7701,
+      "step": 417560
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.7123287671232875e-05,
+      "loss": 0.7581,
+      "step": 417570
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.7122139741333128e-05,
+      "loss": 0.8351,
+      "step": 417580
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.7120991811433378e-05,
+      "loss": 0.8221,
+      "step": 417590
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.7119843881533634e-05,
+      "loss": 0.7742,
+      "step": 417600
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.7118695951633884e-05,
+      "loss": 0.8075,
+      "step": 417610
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.7117548021734134e-05,
+      "loss": 0.8628,
+      "step": 417620
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.711640009183439e-05,
+      "loss": 0.8404,
+      "step": 417630
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.711525216193464e-05,
+      "loss": 0.7282,
+      "step": 417640
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.7114104232034894e-05,
+      "loss": 0.8057,
+      "step": 417650
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.7112956302135147e-05,
+      "loss": 0.8055,
+      "step": 417660
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.71118083722354e-05,
+      "loss": 0.7665,
+      "step": 417670
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.711066044233565e-05,
+      "loss": 0.7453,
+      "step": 417680
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.7109512512435907e-05,
+      "loss": 0.8096,
+      "step": 417690
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 2.7108364582536157e-05,
+      "loss": 0.8581,
+      "step": 417700
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.710721665263641e-05,
+      "loss": 0.8587,
+      "step": 417710
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.7106068722736663e-05,
+      "loss": 0.8476,
+      "step": 417720
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.7104920792836916e-05,
+      "loss": 0.8053,
+      "step": 417730
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.7103772862937166e-05,
+      "loss": 0.7858,
+      "step": 417740
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.7102624933037423e-05,
+      "loss": 0.8418,
+      "step": 417750
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.7101477003137673e-05,
+      "loss": 0.7694,
+      "step": 417760
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.7100329073237923e-05,
+      "loss": 0.8383,
+      "step": 417770
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.709918114333818e-05,
+      "loss": 0.7882,
+      "step": 417780
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.709803321343843e-05,
+      "loss": 0.8014,
+      "step": 417790
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.7096885283538682e-05,
+      "loss": 0.7981,
+      "step": 417800
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.7095737353638935e-05,
+      "loss": 0.8429,
+      "step": 417810
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.709458942373919e-05,
+      "loss": 0.853,
+      "step": 417820
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.709344149383944e-05,
+      "loss": 0.7897,
+      "step": 417830
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.7092293563939695e-05,
+      "loss": 0.8732,
+      "step": 417840
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.7091145634039945e-05,
+      "loss": 0.8605,
+      "step": 417850
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.70899977041402e-05,
+      "loss": 0.928,
+      "step": 417860
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.708884977424045e-05,
+      "loss": 0.774,
+      "step": 417870
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.7087701844340705e-05,
+      "loss": 0.7807,
+      "step": 417880
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.7086553914440955e-05,
+      "loss": 0.8359,
+      "step": 417890
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.708540598454121e-05,
+      "loss": 0.8445,
+      "step": 417900
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.708425805464146e-05,
+      "loss": 0.8703,
+      "step": 417910
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.708311012474171e-05,
+      "loss": 0.833,
+      "step": 417920
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.7081962194841968e-05,
+      "loss": 0.8164,
+      "step": 417930
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.7080814264942217e-05,
+      "loss": 0.8334,
+      "step": 417940
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.707966633504247e-05,
+      "loss": 0.8752,
+      "step": 417950
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.7078518405142724e-05,
+      "loss": 0.8085,
+      "step": 417960
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.7077370475242977e-05,
+      "loss": 0.8976,
+      "step": 417970
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.7076222545343227e-05,
+      "loss": 0.7627,
+      "step": 417980
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.7075074615443484e-05,
+      "loss": 0.7634,
+      "step": 417990
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.7073926685543734e-05,
+      "loss": 0.7335,
+      "step": 418000
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.7072778755643987e-05,
+      "loss": 0.7868,
+      "step": 418010
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.707163082574424e-05,
+      "loss": 0.9134,
+      "step": 418020
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.7070482895844493e-05,
+      "loss": 0.8265,
+      "step": 418030
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.7069334965944743e-05,
+      "loss": 0.8431,
+      "step": 418040
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.7068187036045e-05,
+      "loss": 0.7365,
+      "step": 418050
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.706703910614525e-05,
+      "loss": 0.8403,
+      "step": 418060
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.70658911762455e-05,
+      "loss": 0.8625,
+      "step": 418070
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.7064743246345756e-05,
+      "loss": 0.8815,
+      "step": 418080
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.7063595316446006e-05,
+      "loss": 0.8811,
+      "step": 418090
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.706244738654626e-05,
+      "loss": 0.7609,
+      "step": 418100
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.706129945664651e-05,
+      "loss": 0.8529,
+      "step": 418110
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.7060151526746766e-05,
+      "loss": 0.8824,
+      "step": 418120
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.7059003596847016e-05,
+      "loss": 0.8027,
+      "step": 418130
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.705785566694727e-05,
+      "loss": 0.7747,
+      "step": 418140
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.7056707737047522e-05,
+      "loss": 0.7845,
+      "step": 418150
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.7055559807147775e-05,
+      "loss": 0.7912,
+      "step": 418160
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.7054411877248025e-05,
+      "loss": 0.7234,
+      "step": 418170
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.7053263947348282e-05,
+      "loss": 0.7815,
+      "step": 418180
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.705211601744853e-05,
+      "loss": 0.7996,
+      "step": 418190
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.705096808754878e-05,
+      "loss": 0.8689,
+      "step": 418200
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.7049820157649038e-05,
+      "loss": 0.7869,
+      "step": 418210
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.7048672227749288e-05,
+      "loss": 0.8061,
+      "step": 418220
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.704752429784954e-05,
+      "loss": 0.8079,
+      "step": 418230
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.7046376367949794e-05,
+      "loss": 0.8813,
+      "step": 418240
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.7045228438050048e-05,
+      "loss": 0.868,
+      "step": 418250
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.7044080508150298e-05,
+      "loss": 0.8395,
+      "step": 418260
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.7042932578250554e-05,
+      "loss": 0.7984,
+      "step": 418270
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.7041784648350804e-05,
+      "loss": 0.8016,
+      "step": 418280
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.7040636718451057e-05,
+      "loss": 0.8043,
+      "step": 418290
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.703948878855131e-05,
+      "loss": 0.7435,
+      "step": 418300
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.7038340858651564e-05,
+      "loss": 0.7876,
+      "step": 418310
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.7037192928751814e-05,
+      "loss": 0.8198,
+      "step": 418320
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.703604499885207e-05,
+      "loss": 0.7527,
+      "step": 418330
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.703489706895232e-05,
+      "loss": 0.834,
+      "step": 418340
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 2.703374913905257e-05,
+      "loss": 0.7896,
+      "step": 418350
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.7032601209152827e-05,
+      "loss": 0.9051,
+      "step": 418360
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.7031453279253076e-05,
+      "loss": 0.7781,
+      "step": 418370
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.703030534935333e-05,
+      "loss": 0.8259,
+      "step": 418380
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.7029157419453583e-05,
+      "loss": 0.8449,
+      "step": 418390
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.7028009489553836e-05,
+      "loss": 0.8052,
+      "step": 418400
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.7026861559654086e-05,
+      "loss": 0.7389,
+      "step": 418410
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.7025713629754343e-05,
+      "loss": 0.8616,
+      "step": 418420
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.7024565699854592e-05,
+      "loss": 0.7773,
+      "step": 418430
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.7023417769954846e-05,
+      "loss": 0.8235,
+      "step": 418440
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.70222698400551e-05,
+      "loss": 0.8918,
+      "step": 418450
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.7021121910155352e-05,
+      "loss": 0.7378,
+      "step": 418460
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.7019973980255602e-05,
+      "loss": 0.8858,
+      "step": 418470
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.701882605035586e-05,
+      "loss": 0.7592,
+      "step": 418480
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.701767812045611e-05,
+      "loss": 0.7827,
+      "step": 418490
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.701653019055636e-05,
+      "loss": 0.8677,
+      "step": 418500
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.7015382260656615e-05,
+      "loss": 0.8091,
+      "step": 418510
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.7014234330756865e-05,
+      "loss": 0.8637,
+      "step": 418520
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.7013086400857118e-05,
+      "loss": 0.7923,
+      "step": 418530
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.701193847095737e-05,
+      "loss": 0.7293,
+      "step": 418540
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.7010790541057625e-05,
+      "loss": 0.7414,
+      "step": 418550
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.7009642611157874e-05,
+      "loss": 0.8689,
+      "step": 418560
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.700849468125813e-05,
+      "loss": 0.7638,
+      "step": 418570
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.700734675135838e-05,
+      "loss": 0.8009,
+      "step": 418580
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.7006198821458634e-05,
+      "loss": 0.7509,
+      "step": 418590
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.7005050891558887e-05,
+      "loss": 0.8322,
+      "step": 418600
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.700390296165914e-05,
+      "loss": 0.9418,
+      "step": 418610
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.700275503175939e-05,
+      "loss": 0.9429,
+      "step": 418620
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.7001607101859647e-05,
+      "loss": 0.7529,
+      "step": 418630
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.7000459171959897e-05,
+      "loss": 0.9255,
+      "step": 418640
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.6999311242060147e-05,
+      "loss": 0.8283,
+      "step": 418650
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.6998163312160404e-05,
+      "loss": 0.7166,
+      "step": 418660
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.6997015382260653e-05,
+      "loss": 0.807,
+      "step": 418670
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.6995867452360907e-05,
+      "loss": 0.8586,
+      "step": 418680
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.699471952246116e-05,
+      "loss": 0.7696,
+      "step": 418690
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.6993571592561413e-05,
+      "loss": 0.8029,
+      "step": 418700
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.6992423662661663e-05,
+      "loss": 0.8716,
+      "step": 418710
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.699127573276192e-05,
+      "loss": 0.7605,
+      "step": 418720
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.699012780286217e-05,
+      "loss": 0.7537,
+      "step": 418730
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.6988979872962423e-05,
+      "loss": 0.8231,
+      "step": 418740
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.6987831943062676e-05,
+      "loss": 0.8522,
+      "step": 418750
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.698668401316293e-05,
+      "loss": 0.8235,
+      "step": 418760
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.698553608326318e-05,
+      "loss": 0.7995,
+      "step": 418770
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.6984388153363436e-05,
+      "loss": 0.7636,
+      "step": 418780
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 2.6983240223463685e-05,
+      "loss": 0.78,
+      "step": 418790
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.6982092293563935e-05,
+      "loss": 0.778,
+      "step": 418800
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.698094436366419e-05,
+      "loss": 0.8422,
+      "step": 418810
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.6979796433764442e-05,
+      "loss": 0.8272,
+      "step": 418820
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.6978648503864695e-05,
+      "loss": 0.8848,
+      "step": 418830
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.6977500573964945e-05,
+      "loss": 0.906,
+      "step": 418840
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.69763526440652e-05,
+      "loss": 0.847,
+      "step": 418850
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.697520471416545e-05,
+      "loss": 0.7157,
+      "step": 418860
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.6974056784265705e-05,
+      "loss": 0.9162,
+      "step": 418870
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.6972908854365958e-05,
+      "loss": 0.7921,
+      "step": 418880
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.697176092446621e-05,
+      "loss": 0.8347,
+      "step": 418890
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.697061299456646e-05,
+      "loss": 0.8178,
+      "step": 418900
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.6969465064666718e-05,
+      "loss": 0.8141,
+      "step": 418910
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.6968317134766967e-05,
+      "loss": 0.9004,
+      "step": 418920
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.6967169204867217e-05,
+      "loss": 0.7643,
+      "step": 418930
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.6966021274967474e-05,
+      "loss": 0.8107,
+      "step": 418940
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.6964873345067724e-05,
+      "loss": 0.8057,
+      "step": 418950
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.6963725415167977e-05,
+      "loss": 0.785,
+      "step": 418960
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.696257748526823e-05,
+      "loss": 0.821,
+      "step": 418970
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.6961429555368484e-05,
+      "loss": 0.7018,
+      "step": 418980
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.6960281625468733e-05,
+      "loss": 0.8312,
+      "step": 418990
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.695913369556899e-05,
+      "loss": 0.8993,
+      "step": 419000
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.695798576566924e-05,
+      "loss": 0.7775,
+      "step": 419010
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.6956837835769493e-05,
+      "loss": 0.7301,
+      "step": 419020
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.6955689905869746e-05,
+      "loss": 0.8883,
+      "step": 419030
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.695454197597e-05,
+      "loss": 0.7832,
+      "step": 419040
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.695339404607025e-05,
+      "loss": 0.7541,
+      "step": 419050
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.6952246116170506e-05,
+      "loss": 0.8559,
+      "step": 419060
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.6951098186270756e-05,
+      "loss": 0.8298,
+      "step": 419070
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.6949950256371006e-05,
+      "loss": 0.7668,
+      "step": 419080
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.6948802326471262e-05,
+      "loss": 0.7763,
+      "step": 419090
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.6947654396571512e-05,
+      "loss": 0.8172,
+      "step": 419100
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.6946506466671766e-05,
+      "loss": 0.7582,
+      "step": 419110
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.694535853677202e-05,
+      "loss": 0.8069,
+      "step": 419120
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.6944210606872272e-05,
+      "loss": 0.8321,
+      "step": 419130
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.6943062676972522e-05,
+      "loss": 0.7361,
+      "step": 419140
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.694191474707278e-05,
+      "loss": 0.815,
+      "step": 419150
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.694076681717303e-05,
+      "loss": 0.761,
+      "step": 419160
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.693961888727328e-05,
+      "loss": 0.7399,
+      "step": 419170
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.6938470957373535e-05,
+      "loss": 0.7991,
+      "step": 419180
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.6937323027473788e-05,
+      "loss": 0.8371,
+      "step": 419190
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.6936175097574038e-05,
+      "loss": 0.7944,
+      "step": 419200
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.6935027167674295e-05,
+      "loss": 0.8276,
+      "step": 419210
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.6933879237774544e-05,
+      "loss": 0.8364,
+      "step": 419220
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 2.6932731307874794e-05,
+      "loss": 0.7686,
+      "step": 419230
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.693158337797505e-05,
+      "loss": 0.8476,
+      "step": 419240
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.69304354480753e-05,
+      "loss": 0.8036,
+      "step": 419250
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.6929287518175554e-05,
+      "loss": 0.8199,
+      "step": 419260
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.6928139588275807e-05,
+      "loss": 0.8018,
+      "step": 419270
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.692699165837606e-05,
+      "loss": 0.7922,
+      "step": 419280
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.692584372847631e-05,
+      "loss": 0.8317,
+      "step": 419290
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.6924695798576567e-05,
+      "loss": 0.8544,
+      "step": 419300
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.6923547868676817e-05,
+      "loss": 0.7595,
+      "step": 419310
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.692239993877707e-05,
+      "loss": 0.7925,
+      "step": 419320
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.6921252008877323e-05,
+      "loss": 0.7662,
+      "step": 419330
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.6920104078977577e-05,
+      "loss": 0.8232,
+      "step": 419340
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.6918956149077826e-05,
+      "loss": 0.8755,
+      "step": 419350
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.6917808219178083e-05,
+      "loss": 0.8944,
+      "step": 419360
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.6916660289278333e-05,
+      "loss": 0.9207,
+      "step": 419370
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.6915512359378583e-05,
+      "loss": 0.854,
+      "step": 419380
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.691436442947884e-05,
+      "loss": 0.8662,
+      "step": 419390
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.691321649957909e-05,
+      "loss": 0.813,
+      "step": 419400
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.6912068569679342e-05,
+      "loss": 0.8263,
+      "step": 419410
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.6910920639779596e-05,
+      "loss": 0.8925,
+      "step": 419420
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.690977270987985e-05,
+      "loss": 0.7628,
+      "step": 419430
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.69086247799801e-05,
+      "loss": 0.8531,
+      "step": 419440
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 2.6907476850080355e-05,
+      "loss": 0.7874,
+      "step": 419450
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 2.6906328920180605e-05,
+      "loss": 0.8124,
+      "step": 419460
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 2.690518099028086e-05,
+      "loss": 0.7856,
+      "step": 419470
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 2.6904033060381112e-05,
+      "loss": 0.7935,
+      "step": 419480
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 2.6902885130481365e-05,
+      "loss": 0.8182,
+      "step": 419490
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 2.6901737200581615e-05,
+      "loss": 0.8067,
+      "step": 419500
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 2.6900589270681865e-05,
+      "loss": 0.7748,
+      "step": 419510
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 2.689944134078212e-05,
+      "loss": 0.7832,
+      "step": 419520
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 2.689829341088237e-05,
+      "loss": 0.8207,
+      "step": 419530
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 2.6897145480982624e-05,
+      "loss": 0.8061,
+      "step": 419540
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 2.6895997551082878e-05,
+      "loss": 0.8398,
+      "step": 419550
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 2.689484962118313e-05,
+      "loss": 0.8362,
+      "step": 419560
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 2.689370169128338e-05,
+      "loss": 0.783,
+      "step": 419570
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 2.6892553761383637e-05,
+      "loss": 0.9366,
+      "step": 419580
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 2.6891405831483887e-05,
+      "loss": 0.7911,
+      "step": 419590
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 2.689025790158414e-05,
+      "loss": 0.7788,
+      "step": 419600
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 2.6889109971684394e-05,
+      "loss": 0.8715,
+      "step": 419610
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 2.6887962041784647e-05,
+      "loss": 0.7395,
+      "step": 419620
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 2.6886814111884897e-05,
+      "loss": 0.8437,
+      "step": 419630
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 2.6885666181985153e-05,
+      "loss": 0.7668,
+      "step": 419640
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 2.6884518252085403e-05,
+      "loss": 0.8138,
+      "step": 419650
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 2.6883370322185653e-05,
+      "loss": 0.7676,
+      "step": 419660
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 2.688222239228591e-05,
+      "loss": 0.825,
+      "step": 419670
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 2.688107446238616e-05,
+      "loss": 0.9142,
+      "step": 419680
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 2.6879926532486413e-05,
+      "loss": 0.8392,
+      "step": 419690
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 2.6878778602586666e-05,
+      "loss": 0.8197,
+      "step": 419700
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 2.687763067268692e-05,
+      "loss": 0.8309,
+      "step": 419710
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 2.687648274278717e-05,
+      "loss": 0.7513,
+      "step": 419720
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 2.6875334812887426e-05,
+      "loss": 0.8189,
+      "step": 419730
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 2.6874186882987676e-05,
+      "loss": 0.9012,
+      "step": 419740
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 2.687303895308793e-05,
+      "loss": 0.818,
+      "step": 419750
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 2.6871891023188182e-05,
+      "loss": 0.7453,
+      "step": 419760
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 2.6870743093288435e-05,
+      "loss": 0.808,
+      "step": 419770
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 2.6869595163388685e-05,
+      "loss": 0.7985,
+      "step": 419780
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 2.6868447233488942e-05,
+      "loss": 0.8662,
+      "step": 419790
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 2.6867299303589192e-05,
+      "loss": 0.8808,
+      "step": 419800
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 2.686615137368944e-05,
+      "loss": 0.8207,
+      "step": 419810
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 2.68650034437897e-05,
+      "loss": 0.8306,
+      "step": 419820
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 2.6863855513889948e-05,
+      "loss": 0.8216,
+      "step": 419830
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 2.68627075839902e-05,
+      "loss": 0.7805,
+      "step": 419840
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 2.6861559654090455e-05,
+      "loss": 0.7686,
+      "step": 419850
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 2.6860411724190708e-05,
+      "loss": 0.8075,
+      "step": 419860
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 2.6859263794290958e-05,
+      "loss": 0.8165,
+      "step": 419870
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 2.6858115864391214e-05,
+      "loss": 0.8348,
+      "step": 419880
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 2.6856967934491464e-05,
+      "loss": 0.839,
+      "step": 419890
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 2.6855820004591717e-05,
+      "loss": 0.8107,
+      "step": 419900
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 2.685467207469197e-05,
+      "loss": 0.8761,
+      "step": 419910
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 2.6853524144792224e-05,
+      "loss": 0.8548,
+      "step": 419920
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 2.6852376214892474e-05,
+      "loss": 0.8306,
+      "step": 419930
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 2.685122828499273e-05,
+      "loss": 0.7731,
+      "step": 419940
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 2.685008035509298e-05,
+      "loss": 0.8313,
+      "step": 419950
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 2.684893242519323e-05,
+      "loss": 0.8458,
+      "step": 419960
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 2.6847784495293487e-05,
+      "loss": 0.7833,
+      "step": 419970
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 2.6846636565393737e-05,
+      "loss": 0.8463,
+      "step": 419980
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 2.684548863549399e-05,
+      "loss": 0.7919,
+      "step": 419990
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 2.6844340705594243e-05,
+      "loss": 0.8433,
+      "step": 420000
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 2.6843192775694496e-05,
+      "loss": 0.7446,
+      "step": 420010
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 2.6842044845794746e-05,
+      "loss": 0.849,
+      "step": 420020
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 2.6840896915895003e-05,
+      "loss": 0.8402,
+      "step": 420030
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 2.6839748985995253e-05,
+      "loss": 0.8686,
+      "step": 420040
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 2.6838601056095506e-05,
+      "loss": 0.8055,
+      "step": 420050
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 2.683745312619576e-05,
+      "loss": 0.7982,
+      "step": 420060
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 2.6836305196296012e-05,
+      "loss": 0.8091,
+      "step": 420070
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 2.6835157266396262e-05,
+      "loss": 0.8646,
+      "step": 420080
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 2.683400933649652e-05,
+      "loss": 0.7475,
+      "step": 420090
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 2.683286140659677e-05,
+      "loss": 0.7942,
+      "step": 420100
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 2.683171347669702e-05,
+      "loss": 0.7605,
+      "step": 420110
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 2.6830565546797275e-05,
+      "loss": 0.8547,
+      "step": 420120
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 2.6829417616897525e-05,
+      "loss": 0.8273,
+      "step": 420130
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 2.682826968699778e-05,
+      "loss": 0.8105,
+      "step": 420140
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 2.682712175709803e-05,
+      "loss": 0.7644,
+      "step": 420150
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 2.6825973827198285e-05,
+      "loss": 0.8175,
+      "step": 420160
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 2.6824825897298535e-05,
+      "loss": 0.881,
+      "step": 420170
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 2.6823677967398788e-05,
+      "loss": 0.7412,
+      "step": 420180
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 2.682253003749904e-05,
+      "loss": 0.7827,
+      "step": 420190
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 2.6821382107599294e-05,
+      "loss": 0.8462,
+      "step": 420200
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 2.6820234177699544e-05,
+      "loss": 0.9493,
+      "step": 420210
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 2.68190862477998e-05,
+      "loss": 0.7833,
+      "step": 420220
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 2.681793831790005e-05,
+      "loss": 0.7524,
+      "step": 420230
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 2.68167903880003e-05,
+      "loss": 0.8058,
+      "step": 420240
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 2.6815642458100557e-05,
+      "loss": 0.8722,
+      "step": 420250
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 2.6814494528200807e-05,
+      "loss": 0.7831,
+      "step": 420260
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 2.681334659830106e-05,
+      "loss": 0.8143,
+      "step": 420270
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 2.6812198668401314e-05,
+      "loss": 0.8701,
+      "step": 420280
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 2.6811050738501567e-05,
+      "loss": 0.7971,
+      "step": 420290
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 2.6809902808601817e-05,
+      "loss": 0.8615,
+      "step": 420300
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 2.6808754878702073e-05,
+      "loss": 0.7492,
+      "step": 420310
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 2.6807606948802323e-05,
+      "loss": 0.8529,
+      "step": 420320
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 2.6806459018902576e-05,
+      "loss": 0.8693,
+      "step": 420330
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 2.680531108900283e-05,
+      "loss": 0.8076,
+      "step": 420340
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 2.6804163159103083e-05,
+      "loss": 0.7734,
+      "step": 420350
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 2.6803015229203333e-05,
+      "loss": 0.8277,
+      "step": 420360
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 2.680186729930359e-05,
+      "loss": 0.8083,
+      "step": 420370
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 2.680071936940384e-05,
+      "loss": 0.8674,
+      "step": 420380
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 2.679957143950409e-05,
+      "loss": 0.8561,
+      "step": 420390
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 2.6798423509604346e-05,
+      "loss": 0.8436,
+      "step": 420400
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 2.6797275579704596e-05,
+      "loss": 0.8529,
+      "step": 420410
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 2.679612764980485e-05,
+      "loss": 0.9091,
+      "step": 420420
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 2.6794979719905102e-05,
+      "loss": 0.8713,
+      "step": 420430
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 2.6793831790005355e-05,
+      "loss": 0.8664,
+      "step": 420440
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 2.6792683860105605e-05,
+      "loss": 0.7302,
+      "step": 420450
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 2.6791535930205862e-05,
+      "loss": 0.8412,
+      "step": 420460
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 2.679038800030611e-05,
+      "loss": 0.8063,
+      "step": 420470
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 2.6789240070406365e-05,
+      "loss": 0.7695,
+      "step": 420480
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 2.6788092140506618e-05,
+      "loss": 0.7995,
+      "step": 420490
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 2.678694421060687e-05,
+      "loss": 0.8369,
+      "step": 420500
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 2.678579628070712e-05,
+      "loss": 0.8301,
+      "step": 420510
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 2.6784648350807378e-05,
+      "loss": 0.8415,
+      "step": 420520
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 2.6783500420907628e-05,
+      "loss": 0.8452,
+      "step": 420530
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 2.6782352491007878e-05,
+      "loss": 0.8151,
+      "step": 420540
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 2.6781204561108134e-05,
+      "loss": 0.7445,
+      "step": 420550
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 2.6780056631208384e-05,
+      "loss": 0.8184,
+      "step": 420560
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 2.6778908701308637e-05,
+      "loss": 0.9848,
+      "step": 420570
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 2.677776077140889e-05,
+      "loss": 0.8265,
+      "step": 420580
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 2.6776612841509144e-05,
+      "loss": 0.8534,
+      "step": 420590
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 2.6775464911609394e-05,
+      "loss": 0.8682,
+      "step": 420600
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 2.677431698170965e-05,
+      "loss": 0.7211,
+      "step": 420610
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 2.67731690518099e-05,
+      "loss": 0.7672,
+      "step": 420620
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 2.6772021121910153e-05,
+      "loss": 0.8421,
+      "step": 420630
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 2.6770873192010407e-05,
+      "loss": 0.8235,
+      "step": 420640
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 2.676972526211066e-05,
+      "loss": 0.8333,
+      "step": 420650
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 2.676857733221091e-05,
+      "loss": 0.7279,
+      "step": 420660
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 2.6767429402311166e-05,
+      "loss": 0.8231,
+      "step": 420670
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 2.6766281472411416e-05,
+      "loss": 0.83,
+      "step": 420680
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 2.6765133542511666e-05,
+      "loss": 0.7878,
+      "step": 420690
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 2.6763985612611923e-05,
+      "loss": 0.8393,
+      "step": 420700
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 2.6762837682712173e-05,
+      "loss": 0.8337,
+      "step": 420710
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 2.6761689752812426e-05,
+      "loss": 0.8362,
+      "step": 420720
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 2.676054182291268e-05,
+      "loss": 0.8084,
+      "step": 420730
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 2.6759393893012932e-05,
+      "loss": 0.8525,
+      "step": 420740
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 2.6758245963113182e-05,
+      "loss": 0.7627,
+      "step": 420750
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 2.675709803321344e-05,
+      "loss": 0.8133,
+      "step": 420760
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 2.675595010331369e-05,
+      "loss": 0.8563,
+      "step": 420770
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 2.6754802173413942e-05,
+      "loss": 0.8094,
+      "step": 420780
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 2.6753654243514195e-05,
+      "loss": 0.9081,
+      "step": 420790
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 2.6752506313614448e-05,
+      "loss": 0.8679,
+      "step": 420800
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 2.6751358383714698e-05,
+      "loss": 0.8216,
+      "step": 420810
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 2.6750210453814955e-05,
+      "loss": 0.9431,
+      "step": 420820
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 2.6749062523915205e-05,
+      "loss": 0.8026,
+      "step": 420830
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 2.6747914594015454e-05,
+      "loss": 0.8019,
+      "step": 420840
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 2.674676666411571e-05,
+      "loss": 0.7774,
+      "step": 420850
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 2.674561873421596e-05,
+      "loss": 0.9373,
+      "step": 420860
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 2.6744470804316214e-05,
+      "loss": 0.8094,
+      "step": 420870
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 2.6743322874416464e-05,
+      "loss": 0.9357,
+      "step": 420880
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 2.674217494451672e-05,
+      "loss": 0.8355,
+      "step": 420890
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 2.674102701461697e-05,
+      "loss": 0.8834,
+      "step": 420900
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 2.6739879084717224e-05,
+      "loss": 0.7615,
+      "step": 420910
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 2.6738731154817477e-05,
+      "loss": 0.8148,
+      "step": 420920
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 2.673758322491773e-05,
+      "loss": 0.8047,
+      "step": 420930
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 2.673643529501798e-05,
+      "loss": 0.8764,
+      "step": 420940
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 2.6735287365118237e-05,
+      "loss": 0.8432,
+      "step": 420950
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 2.6734139435218487e-05,
+      "loss": 0.8346,
+      "step": 420960
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 2.6732991505318736e-05,
+      "loss": 0.9127,
+      "step": 420970
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 2.6731843575418993e-05,
+      "loss": 0.7835,
+      "step": 420980
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 2.6730695645519243e-05,
+      "loss": 0.7402,
+      "step": 420990
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 2.6729547715619496e-05,
+      "loss": 0.7654,
+      "step": 421000
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 2.672839978571975e-05,
+      "loss": 0.8695,
+      "step": 421010
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 2.6727251855820003e-05,
+      "loss": 0.8798,
+      "step": 421020
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 2.6726103925920253e-05,
+      "loss": 0.9386,
+      "step": 421030
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 2.672495599602051e-05,
+      "loss": 0.8321,
+      "step": 421040
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 2.672380806612076e-05,
+      "loss": 0.7592,
+      "step": 421050
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 2.6722660136221012e-05,
+      "loss": 0.8636,
+      "step": 421060
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 2.6721512206321266e-05,
+      "loss": 0.7318,
+      "step": 421070
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 2.672036427642152e-05,
+      "loss": 0.8067,
+      "step": 421080
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 2.671921634652177e-05,
+      "loss": 0.813,
+      "step": 421090
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 2.6718068416622025e-05,
+      "loss": 0.804,
+      "step": 421100
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 2.6716920486722275e-05,
+      "loss": 0.7707,
+      "step": 421110
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 2.6715772556822525e-05,
+      "loss": 0.8032,
+      "step": 421120
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 2.671462462692278e-05,
+      "loss": 0.7837,
+      "step": 421130
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 2.671347669702303e-05,
+      "loss": 0.7444,
+      "step": 421140
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 2.6712328767123285e-05,
+      "loss": 0.8596,
+      "step": 421150
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 2.6711180837223538e-05,
+      "loss": 0.8778,
+      "step": 421160
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 2.671003290732379e-05,
+      "loss": 0.9032,
+      "step": 421170
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 2.670888497742404e-05,
+      "loss": 0.7809,
+      "step": 421180
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 2.6707737047524298e-05,
+      "loss": 0.9099,
+      "step": 421190
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 2.6706589117624548e-05,
+      "loss": 0.7961,
+      "step": 421200
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 2.67054411877248e-05,
+      "loss": 0.8092,
+      "step": 421210
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 2.6704293257825054e-05,
+      "loss": 0.8045,
+      "step": 421220
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 2.6703145327925307e-05,
+      "loss": 0.8112,
+      "step": 421230
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 2.6701997398025557e-05,
+      "loss": 0.8118,
+      "step": 421240
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 2.6700849468125814e-05,
+      "loss": 0.8644,
+      "step": 421250
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 2.6699701538226064e-05,
+      "loss": 0.8045,
+      "step": 421260
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 2.6698553608326313e-05,
+      "loss": 0.7494,
+      "step": 421270
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 2.669740567842657e-05,
+      "loss": 0.8194,
+      "step": 421280
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 2.669625774852682e-05,
+      "loss": 0.9103,
+      "step": 421290
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 2.6695109818627073e-05,
+      "loss": 0.8088,
+      "step": 421300
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 2.6693961888727326e-05,
+      "loss": 0.8124,
+      "step": 421310
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 2.669281395882758e-05,
+      "loss": 0.8331,
+      "step": 421320
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 2.669166602892783e-05,
+      "loss": 0.7632,
+      "step": 421330
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 2.6690518099028086e-05,
+      "loss": 0.741,
+      "step": 421340
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 2.6689370169128336e-05,
+      "loss": 0.768,
+      "step": 421350
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 2.668822223922859e-05,
+      "loss": 0.8201,
+      "step": 421360
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 2.6687074309328842e-05,
+      "loss": 0.8421,
+      "step": 421370
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 2.6685926379429096e-05,
+      "loss": 0.8263,
+      "step": 421380
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 2.6684778449529346e-05,
+      "loss": 0.8845,
+      "step": 421390
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 2.6683630519629602e-05,
+      "loss": 0.8411,
+      "step": 421400
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 2.6682482589729852e-05,
+      "loss": 0.8087,
+      "step": 421410
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 2.6681334659830102e-05,
+      "loss": 0.8262,
+      "step": 421420
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 2.668018672993036e-05,
+      "loss": 0.8589,
+      "step": 421430
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 2.667903880003061e-05,
+      "loss": 0.7682,
+      "step": 421440
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 2.667789087013086e-05,
+      "loss": 0.8136,
+      "step": 421450
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 2.6676742940231115e-05,
+      "loss": 0.8144,
+      "step": 421460
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 2.6675595010331368e-05,
+      "loss": 0.8375,
+      "step": 421470
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 2.6674447080431618e-05,
+      "loss": 0.8066,
+      "step": 421480
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 2.6673299150531875e-05,
+      "loss": 0.9004,
+      "step": 421490
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 2.6672151220632124e-05,
+      "loss": 0.7561,
+      "step": 421500
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 2.6671003290732378e-05,
+      "loss": 0.8298,
+      "step": 421510
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 2.666985536083263e-05,
+      "loss": 0.8665,
+      "step": 421520
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 2.6668707430932884e-05,
+      "loss": 0.7621,
+      "step": 421530
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 2.6667559501033134e-05,
+      "loss": 0.7867,
+      "step": 421540
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 2.6666411571133384e-05,
+      "loss": 0.8074,
+      "step": 421550
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 2.666526364123364e-05,
+      "loss": 0.8682,
+      "step": 421560
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 2.666411571133389e-05,
+      "loss": 0.889,
+      "step": 421570
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 2.6662967781434144e-05,
+      "loss": 0.7844,
+      "step": 421580
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 2.6661819851534397e-05,
+      "loss": 0.7923,
+      "step": 421590
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 2.666067192163465e-05,
+      "loss": 0.7894,
+      "step": 421600
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 2.66595239917349e-05,
+      "loss": 0.7691,
+      "step": 421610
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 2.6658376061835157e-05,
+      "loss": 0.8913,
+      "step": 421620
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 2.6657228131935406e-05,
+      "loss": 0.8133,
+      "step": 421630
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 2.665608020203566e-05,
+      "loss": 0.8902,
+      "step": 421640
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 2.6654932272135913e-05,
+      "loss": 0.8127,
+      "step": 421650
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 2.6653784342236166e-05,
+      "loss": 0.8215,
+      "step": 421660
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 2.6652636412336416e-05,
+      "loss": 0.8605,
+      "step": 421670
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 2.6651488482436673e-05,
+      "loss": 0.845,
+      "step": 421680
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 2.6650340552536923e-05,
+      "loss": 0.776,
+      "step": 421690
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 2.6649192622637172e-05,
+      "loss": 0.8528,
+      "step": 421700
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 2.664804469273743e-05,
+      "loss": 0.9214,
+      "step": 421710
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 2.664689676283768e-05,
+      "loss": 0.8439,
+      "step": 421720
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 2.6645748832937932e-05,
+      "loss": 0.8052,
+      "step": 421730
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 2.6644600903038185e-05,
+      "loss": 0.8322,
+      "step": 421740
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 2.664345297313844e-05,
+      "loss": 0.8293,
+      "step": 421750
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 2.664230504323869e-05,
+      "loss": 0.7819,
+      "step": 421760
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 2.6641157113338945e-05,
+      "loss": 0.8445,
+      "step": 421770
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 2.6640009183439195e-05,
+      "loss": 0.8459,
+      "step": 421780
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 2.6638861253539448e-05,
+      "loss": 0.809,
+      "step": 421790
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 2.66377133236397e-05,
+      "loss": 0.8324,
+      "step": 421800
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 2.6636565393739955e-05,
+      "loss": 0.8596,
+      "step": 421810
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 2.6635417463840204e-05,
+      "loss": 0.8609,
+      "step": 421820
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 2.663426953394046e-05,
+      "loss": 0.8459,
+      "step": 421830
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 2.663312160404071e-05,
+      "loss": 0.8384,
+      "step": 421840
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 2.663197367414096e-05,
+      "loss": 0.8337,
+      "step": 421850
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 2.6630825744241217e-05,
+      "loss": 0.86,
+      "step": 421860
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 2.6629677814341467e-05,
+      "loss": 0.8086,
+      "step": 421870
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 2.662852988444172e-05,
+      "loss": 0.9162,
+      "step": 421880
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 2.6627381954541974e-05,
+      "loss": 0.7964,
+      "step": 421890
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 2.6626234024642227e-05,
+      "loss": 0.9399,
+      "step": 421900
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 2.6625086094742477e-05,
+      "loss": 0.8989,
+      "step": 421910
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 2.6623938164842734e-05,
+      "loss": 0.818,
+      "step": 421920
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 2.6622790234942983e-05,
+      "loss": 0.8033,
+      "step": 421930
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 2.6621642305043237e-05,
+      "loss": 0.7901,
+      "step": 421940
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 2.662049437514349e-05,
+      "loss": 0.8336,
+      "step": 421950
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 2.6619346445243743e-05,
+      "loss": 0.8555,
+      "step": 421960
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 2.6618198515343993e-05,
+      "loss": 0.8422,
+      "step": 421970
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 2.661705058544425e-05,
+      "loss": 0.8441,
+      "step": 421980
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 2.66159026555445e-05,
+      "loss": 0.7549,
+      "step": 421990
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 2.661475472564475e-05,
+      "loss": 0.8269,
+      "step": 422000
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 2.6613606795745006e-05,
+      "loss": 0.7479,
+      "step": 422010
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 2.6612458865845256e-05,
+      "loss": 0.8514,
+      "step": 422020
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 2.661131093594551e-05,
+      "loss": 0.835,
+      "step": 422030
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 2.6610163006045762e-05,
+      "loss": 0.8638,
+      "step": 422040
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 2.6609015076146016e-05,
+      "loss": 0.8518,
+      "step": 422050
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 2.6607867146246265e-05,
+      "loss": 0.7268,
+      "step": 422060
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 2.6606719216346522e-05,
+      "loss": 0.8117,
+      "step": 422070
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 2.6605571286446772e-05,
+      "loss": 0.8175,
+      "step": 422080
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 2.6604423356547025e-05,
+      "loss": 0.7165,
+      "step": 422090
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 2.660327542664728e-05,
+      "loss": 0.8625,
+      "step": 422100
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 2.660212749674753e-05,
+      "loss": 0.8438,
+      "step": 422110
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 2.660097956684778e-05,
+      "loss": 0.8268,
+      "step": 422120
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 2.6599831636948038e-05,
+      "loss": 0.8957,
+      "step": 422130
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 2.6598683707048288e-05,
+      "loss": 0.8886,
+      "step": 422140
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 2.6597535777148538e-05,
+      "loss": 0.8501,
+      "step": 422150
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 2.6596387847248794e-05,
+      "loss": 0.8011,
+      "step": 422160
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 2.6595239917349044e-05,
+      "loss": 0.8311,
+      "step": 422170
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 2.6594091987449298e-05,
+      "loss": 0.775,
+      "step": 422180
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 2.659294405754955e-05,
+      "loss": 0.8358,
+      "step": 422190
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 2.6591796127649804e-05,
+      "loss": 0.7826,
+      "step": 422200
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 2.6590648197750054e-05,
+      "loss": 0.7592,
+      "step": 422210
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 2.658950026785031e-05,
+      "loss": 0.7914,
+      "step": 422220
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 2.658835233795056e-05,
+      "loss": 0.7516,
+      "step": 422230
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 2.6587204408050814e-05,
+      "loss": 0.8619,
+      "step": 422240
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 2.6586056478151063e-05,
+      "loss": 0.833,
+      "step": 422250
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 2.658490854825132e-05,
+      "loss": 0.8921,
+      "step": 422260
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 2.658376061835157e-05,
+      "loss": 0.8781,
+      "step": 422270
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 2.658261268845182e-05,
+      "loss": 0.8064,
+      "step": 422280
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 2.6581464758552076e-05,
+      "loss": 0.8884,
+      "step": 422290
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 2.6580316828652326e-05,
+      "loss": 0.7746,
+      "step": 422300
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 2.657916889875258e-05,
+      "loss": 0.8438,
+      "step": 422310
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 2.6578020968852833e-05,
+      "loss": 0.7694,
+      "step": 422320
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 2.6576873038953086e-05,
+      "loss": 0.8416,
+      "step": 422330
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 2.6575725109053336e-05,
+      "loss": 0.7834,
+      "step": 422340
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 2.6574577179153592e-05,
+      "loss": 0.7903,
+      "step": 422350
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 2.6573429249253842e-05,
+      "loss": 0.8154,
+      "step": 422360
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 2.6572281319354096e-05,
+      "loss": 0.8429,
+      "step": 422370
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 2.657113338945435e-05,
+      "loss": 0.7652,
+      "step": 422380
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 2.6569985459554602e-05,
+      "loss": 0.8786,
+      "step": 422390
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 2.6568837529654852e-05,
+      "loss": 0.8668,
+      "step": 422400
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 2.656768959975511e-05,
+      "loss": 0.8056,
+      "step": 422410
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 2.656654166985536e-05,
+      "loss": 0.8063,
+      "step": 422420
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 2.6565393739955608e-05,
+      "loss": 0.9092,
+      "step": 422430
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 2.6564245810055865e-05,
+      "loss": 0.7938,
+      "step": 422440
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 2.6563097880156115e-05,
+      "loss": 0.8607,
+      "step": 422450
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 2.6561949950256368e-05,
+      "loss": 0.8048,
+      "step": 422460
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 2.656080202035662e-05,
+      "loss": 0.7643,
+      "step": 422470
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 2.6559654090456874e-05,
+      "loss": 0.79,
+      "step": 422480
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 2.6558506160557124e-05,
+      "loss": 0.7596,
+      "step": 422490
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 2.655735823065738e-05,
+      "loss": 0.8713,
+      "step": 422500
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 2.655621030075763e-05,
+      "loss": 0.7826,
+      "step": 422510
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 2.6555062370857884e-05,
+      "loss": 0.7927,
+      "step": 422520
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 2.6553914440958137e-05,
+      "loss": 0.9289,
+      "step": 422530
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 2.655276651105839e-05,
+      "loss": 0.8085,
+      "step": 422540
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 2.655161858115864e-05,
+      "loss": 0.7912,
+      "step": 422550
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 2.6550470651258897e-05,
+      "loss": 0.8683,
+      "step": 422560
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 2.6549322721359147e-05,
+      "loss": 0.8591,
+      "step": 422570
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 2.6548174791459397e-05,
+      "loss": 0.8728,
+      "step": 422580
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 2.6547026861559653e-05,
+      "loss": 0.7892,
+      "step": 422590
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 2.6545878931659903e-05,
+      "loss": 0.8583,
+      "step": 422600
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 2.6544731001760156e-05,
+      "loss": 0.7736,
+      "step": 422610
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 2.654358307186041e-05,
+      "loss": 0.8393,
+      "step": 422620
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 2.6542435141960663e-05,
+      "loss": 0.7628,
+      "step": 422630
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 2.6541287212060913e-05,
+      "loss": 0.9845,
+      "step": 422640
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 2.654013928216117e-05,
+      "loss": 0.8034,
+      "step": 422650
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 2.653899135226142e-05,
+      "loss": 0.8919,
+      "step": 422660
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 2.6537843422361672e-05,
+      "loss": 0.8469,
+      "step": 422670
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 2.6536695492461926e-05,
+      "loss": 0.8532,
+      "step": 422680
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 2.653554756256218e-05,
+      "loss": 0.8477,
+      "step": 422690
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 2.653439963266243e-05,
+      "loss": 0.9215,
+      "step": 422700
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 2.6533251702762685e-05,
+      "loss": 0.7468,
+      "step": 422710
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 2.6532103772862935e-05,
+      "loss": 0.8278,
+      "step": 422720
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 2.6530955842963185e-05,
+      "loss": 0.7626,
+      "step": 422730
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 2.6529807913063442e-05,
+      "loss": 0.8004,
+      "step": 422740
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 2.652865998316369e-05,
+      "loss": 0.8669,
+      "step": 422750
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 2.6527512053263945e-05,
+      "loss": 0.8671,
+      "step": 422760
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 2.6526364123364198e-05,
+      "loss": 0.8392,
+      "step": 422770
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 2.652521619346445e-05,
+      "loss": 0.881,
+      "step": 422780
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 2.65240682635647e-05,
+      "loss": 0.8068,
+      "step": 422790
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 2.6522920333664958e-05,
+      "loss": 0.8195,
+      "step": 422800
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 2.6521772403765208e-05,
+      "loss": 0.8339,
+      "step": 422810
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 2.652062447386546e-05,
+      "loss": 0.7373,
+      "step": 422820
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 2.6519476543965714e-05,
+      "loss": 0.8325,
+      "step": 422830
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 2.6518328614065967e-05,
+      "loss": 0.8603,
+      "step": 422840
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 2.6517180684166217e-05,
+      "loss": 0.8109,
+      "step": 422850
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 2.6516032754266474e-05,
+      "loss": 0.8259,
+      "step": 422860
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 2.6514884824366724e-05,
+      "loss": 0.8023,
+      "step": 422870
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 2.6513736894466974e-05,
+      "loss": 0.8284,
+      "step": 422880
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 2.651258896456723e-05,
+      "loss": 0.8716,
+      "step": 422890
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 2.651144103466748e-05,
+      "loss": 0.7501,
+      "step": 422900
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 2.6510293104767733e-05,
+      "loss": 0.7925,
+      "step": 422910
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 2.6509145174867983e-05,
+      "loss": 0.8517,
+      "step": 422920
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 2.650799724496824e-05,
+      "loss": 0.7724,
+      "step": 422930
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 2.650684931506849e-05,
+      "loss": 0.8062,
+      "step": 422940
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 2.6505701385168743e-05,
+      "loss": 0.874,
+      "step": 422950
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 2.6504553455268996e-05,
+      "loss": 0.8492,
+      "step": 422960
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 2.650340552536925e-05,
+      "loss": 0.8678,
+      "step": 422970
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 2.65022575954695e-05,
+      "loss": 0.7943,
+      "step": 422980
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 2.6501109665569756e-05,
+      "loss": 0.7579,
+      "step": 422990
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 2.6499961735670006e-05,
+      "loss": 0.9096,
+      "step": 423000
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 2.6498813805770256e-05,
+      "loss": 0.8983,
+      "step": 423010
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 2.6497665875870512e-05,
+      "loss": 0.9664,
+      "step": 423020
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 2.6496517945970762e-05,
+      "loss": 0.8326,
+      "step": 423030
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 2.6495370016071015e-05,
+      "loss": 0.8749,
+      "step": 423040
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 2.649422208617127e-05,
+      "loss": 0.8498,
+      "step": 423050
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 2.6493074156271522e-05,
+      "loss": 0.8638,
+      "step": 423060
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 2.6491926226371772e-05,
+      "loss": 0.801,
+      "step": 423070
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 2.649077829647203e-05,
+      "loss": 0.8206,
+      "step": 423080
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 2.6489630366572278e-05,
+      "loss": 0.8006,
+      "step": 423090
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 2.648848243667253e-05,
+      "loss": 0.902,
+      "step": 423100
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 2.6487334506772785e-05,
+      "loss": 0.7375,
+      "step": 423110
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 2.6486186576873038e-05,
+      "loss": 0.7518,
+      "step": 423120
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 2.6485038646973288e-05,
+      "loss": 0.8498,
+      "step": 423130
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 2.6483890717073544e-05,
+      "loss": 0.8385,
+      "step": 423140
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 2.6482742787173794e-05,
+      "loss": 0.8217,
+      "step": 423150
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 2.6481594857274044e-05,
+      "loss": 0.7936,
+      "step": 423160
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 2.64804469273743e-05,
+      "loss": 0.8121,
+      "step": 423170
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 2.647929899747455e-05,
+      "loss": 0.8402,
+      "step": 423180
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 2.6478151067574804e-05,
+      "loss": 0.7822,
+      "step": 423190
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 2.6477003137675057e-05,
+      "loss": 0.8108,
+      "step": 423200
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 2.647585520777531e-05,
+      "loss": 0.8634,
+      "step": 423210
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 2.647470727787556e-05,
+      "loss": 0.8117,
+      "step": 423220
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 2.6473559347975817e-05,
+      "loss": 0.8246,
+      "step": 423230
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 2.6472411418076067e-05,
+      "loss": 0.8551,
+      "step": 423240
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 2.647126348817632e-05,
+      "loss": 0.8482,
+      "step": 423250
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 2.6470115558276573e-05,
+      "loss": 0.8612,
+      "step": 423260
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 2.6468967628376826e-05,
+      "loss": 0.8116,
+      "step": 423270
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 2.6467819698477076e-05,
+      "loss": 0.7993,
+      "step": 423280
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 2.6466671768577333e-05,
+      "loss": 0.7625,
+      "step": 423290
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 2.6465523838677583e-05,
+      "loss": 0.8693,
+      "step": 423300
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 2.6464375908777833e-05,
+      "loss": 0.8454,
+      "step": 423310
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 2.646322797887809e-05,
+      "loss": 0.8864,
+      "step": 423320
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 2.646208004897834e-05,
+      "loss": 0.8182,
+      "step": 423330
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 2.6460932119078592e-05,
+      "loss": 0.8115,
+      "step": 423340
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 2.6459784189178846e-05,
+      "loss": 0.7582,
+      "step": 423350
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 2.64586362592791e-05,
+      "loss": 0.8302,
+      "step": 423360
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 2.645748832937935e-05,
+      "loss": 0.787,
+      "step": 423370
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 2.6456340399479605e-05,
+      "loss": 0.7849,
+      "step": 423380
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 2.6455192469579855e-05,
+      "loss": 0.8215,
+      "step": 423390
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 2.645404453968011e-05,
+      "loss": 0.8095,
+      "step": 423400
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 2.645289660978036e-05,
+      "loss": 0.8195,
+      "step": 423410
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 2.6451748679880615e-05,
+      "loss": 0.816,
+      "step": 423420
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 2.6450600749980865e-05,
+      "loss": 0.8115,
+      "step": 423430
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 2.644945282008112e-05,
+      "loss": 0.7655,
+      "step": 423440
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 2.644830489018137e-05,
+      "loss": 0.7481,
+      "step": 423450
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 2.644715696028162e-05,
+      "loss": 0.8135,
+      "step": 423460
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 2.6446009030381878e-05,
+      "loss": 0.8612,
+      "step": 423470
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 2.6444861100482128e-05,
+      "loss": 0.8441,
+      "step": 423480
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 2.644371317058238e-05,
+      "loss": 0.7609,
+      "step": 423490
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 2.6442565240682634e-05,
+      "loss": 0.7834,
+      "step": 423500
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 2.6441417310782887e-05,
+      "loss": 0.7903,
+      "step": 423510
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 2.6440269380883137e-05,
+      "loss": 0.8021,
+      "step": 423520
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 2.6439121450983394e-05,
+      "loss": 0.8396,
+      "step": 423530
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 2.6437973521083644e-05,
+      "loss": 0.8851,
+      "step": 423540
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 2.6436825591183897e-05,
+      "loss": 0.9158,
+      "step": 423550
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 2.643567766128415e-05,
+      "loss": 0.8514,
+      "step": 423560
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 2.6434529731384403e-05,
+      "loss": 0.7808,
+      "step": 423570
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 2.6433381801484653e-05,
+      "loss": 0.7556,
+      "step": 423580
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 2.643223387158491e-05,
+      "loss": 0.9345,
+      "step": 423590
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.643108594168516e-05,
+      "loss": 0.7993,
+      "step": 423600
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.642993801178541e-05,
+      "loss": 0.931,
+      "step": 423610
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.6428790081885663e-05,
+      "loss": 0.8718,
+      "step": 423620
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.6427642151985916e-05,
+      "loss": 0.8579,
+      "step": 423630
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.642649422208617e-05,
+      "loss": 0.7821,
+      "step": 423640
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.642534629218642e-05,
+      "loss": 0.9012,
+      "step": 423650
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.6424198362286676e-05,
+      "loss": 0.803,
+      "step": 423660
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.6423050432386926e-05,
+      "loss": 0.7913,
+      "step": 423670
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.642190250248718e-05,
+      "loss": 0.8842,
+      "step": 423680
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.6420754572587432e-05,
+      "loss": 0.8139,
+      "step": 423690
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.6419606642687685e-05,
+      "loss": 0.8345,
+      "step": 423700
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.6418458712787935e-05,
+      "loss": 0.8323,
+      "step": 423710
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.6417310782888192e-05,
+      "loss": 0.8375,
+      "step": 423720
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.641616285298844e-05,
+      "loss": 0.846,
+      "step": 423730
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.641501492308869e-05,
+      "loss": 0.7141,
+      "step": 423740
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.6413866993188948e-05,
+      "loss": 0.7532,
+      "step": 423750
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.6412719063289198e-05,
+      "loss": 0.8108,
+      "step": 423760
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.641157113338945e-05,
+      "loss": 0.8025,
+      "step": 423770
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.6410423203489704e-05,
+      "loss": 0.8196,
+      "step": 423780
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.6409275273589958e-05,
+      "loss": 0.7901,
+      "step": 423790
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.6408127343690208e-05,
+      "loss": 0.8506,
+      "step": 423800
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.6406979413790464e-05,
+      "loss": 0.802,
+      "step": 423810
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.6405831483890714e-05,
+      "loss": 0.7933,
+      "step": 423820
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.6404683553990967e-05,
+      "loss": 0.8593,
+      "step": 423830
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.640353562409122e-05,
+      "loss": 0.7785,
+      "step": 423840
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.6402387694191474e-05,
+      "loss": 0.8994,
+      "step": 423850
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.6401239764291724e-05,
+      "loss": 0.8621,
+      "step": 423860
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.640009183439198e-05,
+      "loss": 0.8186,
+      "step": 423870
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.639894390449223e-05,
+      "loss": 0.9301,
+      "step": 423880
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.639779597459248e-05,
+      "loss": 0.8133,
+      "step": 423890
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.6396648044692737e-05,
+      "loss": 0.7894,
+      "step": 423900
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.6395500114792986e-05,
+      "loss": 0.9133,
+      "step": 423910
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.639435218489324e-05,
+      "loss": 0.8078,
+      "step": 423920
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.6393204254993493e-05,
+      "loss": 0.7975,
+      "step": 423930
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.6392056325093746e-05,
+      "loss": 0.9299,
+      "step": 423940
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.6390908395193996e-05,
+      "loss": 0.8676,
+      "step": 423950
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.6389760465294253e-05,
+      "loss": 0.9193,
+      "step": 423960
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.6388612535394503e-05,
+      "loss": 0.7764,
+      "step": 423970
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.6387464605494756e-05,
+      "loss": 0.776,
+      "step": 423980
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.638631667559501e-05,
+      "loss": 0.8112,
+      "step": 423990
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.6385168745695262e-05,
+      "loss": 0.8365,
+      "step": 424000
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.6384020815795512e-05,
+      "loss": 0.8591,
+      "step": 424010
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.638287288589577e-05,
+      "loss": 0.8798,
+      "step": 424020
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.638172495599602e-05,
+      "loss": 0.7756,
+      "step": 424030
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.638057702609627e-05,
+      "loss": 0.8906,
+      "step": 424040
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.6379429096196525e-05,
+      "loss": 0.8198,
+      "step": 424050
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.6378281166296775e-05,
+      "loss": 0.7492,
+      "step": 424060
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.6377133236397028e-05,
+      "loss": 0.7881,
+      "step": 424070
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.637598530649728e-05,
+      "loss": 0.8298,
+      "step": 424080
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.6374837376597535e-05,
+      "loss": 0.9082,
+      "step": 424090
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.6373689446697785e-05,
+      "loss": 0.8132,
+      "step": 424100
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.637254151679804e-05,
+      "loss": 0.7751,
+      "step": 424110
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.637139358689829e-05,
+      "loss": 0.8194,
+      "step": 424120
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.6370245656998544e-05,
+      "loss": 0.8804,
+      "step": 424130
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.6369097727098797e-05,
+      "loss": 0.8156,
+      "step": 424140
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.636794979719905e-05,
+      "loss": 0.8902,
+      "step": 424150
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.63668018672993e-05,
+      "loss": 0.7886,
+      "step": 424160
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.6365653937399557e-05,
+      "loss": 0.8193,
+      "step": 424170
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.6364506007499807e-05,
+      "loss": 0.7474,
+      "step": 424180
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.6363358077600057e-05,
+      "loss": 0.8111,
+      "step": 424190
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.6362210147700314e-05,
+      "loss": 0.8161,
+      "step": 424200
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.6361062217800563e-05,
+      "loss": 0.783,
+      "step": 424210
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.6359914287900817e-05,
+      "loss": 0.7985,
+      "step": 424220
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.635876635800107e-05,
+      "loss": 0.9558,
+      "step": 424230
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 2.6357618428101323e-05,
+      "loss": 0.8289,
+      "step": 424240
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.6356470498201573e-05,
+      "loss": 0.9059,
+      "step": 424250
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.635532256830183e-05,
+      "loss": 0.7734,
+      "step": 424260
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.635417463840208e-05,
+      "loss": 0.8255,
+      "step": 424270
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.6353026708502333e-05,
+      "loss": 0.8182,
+      "step": 424280
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.6351878778602583e-05,
+      "loss": 0.7815,
+      "step": 424290
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.635073084870284e-05,
+      "loss": 0.8518,
+      "step": 424300
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.634958291880309e-05,
+      "loss": 0.8042,
+      "step": 424310
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.634843498890334e-05,
+      "loss": 0.7976,
+      "step": 424320
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.6347287059003596e-05,
+      "loss": 0.8406,
+      "step": 424330
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.6346139129103845e-05,
+      "loss": 0.8417,
+      "step": 424340
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.63449911992041e-05,
+      "loss": 0.8093,
+      "step": 424350
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.6343843269304352e-05,
+      "loss": 0.8749,
+      "step": 424360
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.6342695339404605e-05,
+      "loss": 0.7884,
+      "step": 424370
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.6341547409504855e-05,
+      "loss": 0.8407,
+      "step": 424380
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.634039947960511e-05,
+      "loss": 0.9216,
+      "step": 424390
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.633925154970536e-05,
+      "loss": 0.8668,
+      "step": 424400
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.6338103619805615e-05,
+      "loss": 0.8966,
+      "step": 424410
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.6336955689905868e-05,
+      "loss": 0.8099,
+      "step": 424420
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.633580776000612e-05,
+      "loss": 0.8131,
+      "step": 424430
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.633465983010637e-05,
+      "loss": 0.7696,
+      "step": 424440
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.6333511900206628e-05,
+      "loss": 0.8712,
+      "step": 424450
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.6332363970306878e-05,
+      "loss": 0.8537,
+      "step": 424460
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.6331216040407127e-05,
+      "loss": 0.8287,
+      "step": 424470
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.6330068110507384e-05,
+      "loss": 0.8439,
+      "step": 424480
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.6328920180607634e-05,
+      "loss": 0.796,
+      "step": 424490
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.6327772250707887e-05,
+      "loss": 0.828,
+      "step": 424500
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.632662432080814e-05,
+      "loss": 0.8171,
+      "step": 424510
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.6325476390908394e-05,
+      "loss": 0.873,
+      "step": 424520
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.6324328461008643e-05,
+      "loss": 0.9061,
+      "step": 424530
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.63231805311089e-05,
+      "loss": 0.7825,
+      "step": 424540
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.632203260120915e-05,
+      "loss": 0.8707,
+      "step": 424550
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.6320884671309403e-05,
+      "loss": 0.8506,
+      "step": 424560
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.6319736741409656e-05,
+      "loss": 0.8927,
+      "step": 424570
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.631858881150991e-05,
+      "loss": 0.7731,
+      "step": 424580
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.631744088161016e-05,
+      "loss": 0.8463,
+      "step": 424590
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.6316292951710416e-05,
+      "loss": 0.8752,
+      "step": 424600
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.6315145021810666e-05,
+      "loss": 0.7747,
+      "step": 424610
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.6313997091910916e-05,
+      "loss": 0.8224,
+      "step": 424620
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.6312849162011172e-05,
+      "loss": 0.7321,
+      "step": 424630
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.6311701232111422e-05,
+      "loss": 0.8567,
+      "step": 424640
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.6310553302211676e-05,
+      "loss": 0.8525,
+      "step": 424650
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.630940537231193e-05,
+      "loss": 0.8299,
+      "step": 424660
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.6308257442412182e-05,
+      "loss": 0.8283,
+      "step": 424670
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.6307109512512432e-05,
+      "loss": 0.7609,
+      "step": 424680
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.630596158261269e-05,
+      "loss": 0.8204,
+      "step": 424690
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.630481365271294e-05,
+      "loss": 0.8226,
+      "step": 424700
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.630366572281319e-05,
+      "loss": 0.8589,
+      "step": 424710
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.6302517792913445e-05,
+      "loss": 0.9493,
+      "step": 424720
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.6301369863013698e-05,
+      "loss": 0.8711,
+      "step": 424730
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.6300221933113948e-05,
+      "loss": 0.777,
+      "step": 424740
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.6299074003214205e-05,
+      "loss": 0.7611,
+      "step": 424750
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.6297926073314454e-05,
+      "loss": 0.7808,
+      "step": 424760
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.6296778143414704e-05,
+      "loss": 0.7442,
+      "step": 424770
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.629563021351496e-05,
+      "loss": 0.7889,
+      "step": 424780
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.629448228361521e-05,
+      "loss": 0.7564,
+      "step": 424790
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.6293334353715464e-05,
+      "loss": 0.8731,
+      "step": 424800
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.6292186423815717e-05,
+      "loss": 0.8589,
+      "step": 424810
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.629103849391597e-05,
+      "loss": 0.8828,
+      "step": 424820
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.628989056401622e-05,
+      "loss": 0.7994,
+      "step": 424830
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.6288742634116477e-05,
+      "loss": 0.8646,
+      "step": 424840
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.6287594704216727e-05,
+      "loss": 0.884,
+      "step": 424850
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.628644677431698e-05,
+      "loss": 0.7528,
+      "step": 424860
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.6285298844417233e-05,
+      "loss": 0.8796,
+      "step": 424870
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.6284150914517487e-05,
+      "loss": 0.8618,
+      "step": 424880
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 2.6283002984617736e-05,
+      "loss": 0.8422,
+      "step": 424890
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.6281855054717993e-05,
+      "loss": 0.8818,
+      "step": 424900
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.6280707124818243e-05,
+      "loss": 0.8632,
+      "step": 424910
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.6279559194918493e-05,
+      "loss": 0.7954,
+      "step": 424920
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.627841126501875e-05,
+      "loss": 0.7446,
+      "step": 424930
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.6277263335119e-05,
+      "loss": 0.796,
+      "step": 424940
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.6276115405219253e-05,
+      "loss": 0.8878,
+      "step": 424950
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.6274967475319506e-05,
+      "loss": 0.7588,
+      "step": 424960
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.627381954541976e-05,
+      "loss": 0.8291,
+      "step": 424970
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.627267161552001e-05,
+      "loss": 0.7588,
+      "step": 424980
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.6271523685620262e-05,
+      "loss": 0.8578,
+      "step": 424990
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.6270375755720515e-05,
+      "loss": 0.8135,
+      "step": 425000
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.626922782582077e-05,
+      "loss": 0.893,
+      "step": 425010
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.626807989592102e-05,
+      "loss": 0.8984,
+      "step": 425020
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.6266931966021275e-05,
+      "loss": 0.9333,
+      "step": 425030
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.6265784036121525e-05,
+      "loss": 0.854,
+      "step": 425040
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.6264636106221775e-05,
+      "loss": 0.7959,
+      "step": 425050
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.626348817632203e-05,
+      "loss": 0.8106,
+      "step": 425060
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.626234024642228e-05,
+      "loss": 0.9013,
+      "step": 425070
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.6261192316522535e-05,
+      "loss": 0.8158,
+      "step": 425080
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.6260044386622788e-05,
+      "loss": 0.8016,
+      "step": 425090
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.625889645672304e-05,
+      "loss": 0.8079,
+      "step": 425100
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.625774852682329e-05,
+      "loss": 0.9317,
+      "step": 425110
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.6256600596923547e-05,
+      "loss": 0.7691,
+      "step": 425120
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.6255452667023797e-05,
+      "loss": 0.8442,
+      "step": 425130
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.625430473712405e-05,
+      "loss": 0.8312,
+      "step": 425140
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.6253156807224304e-05,
+      "loss": 0.8435,
+      "step": 425150
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.6252008877324557e-05,
+      "loss": 0.8247,
+      "step": 425160
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.6250860947424807e-05,
+      "loss": 0.8222,
+      "step": 425170
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.6249713017525064e-05,
+      "loss": 0.8117,
+      "step": 425180
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.6248565087625313e-05,
+      "loss": 0.791,
+      "step": 425190
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.6247417157725563e-05,
+      "loss": 0.8965,
+      "step": 425200
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.624626922782582e-05,
+      "loss": 0.853,
+      "step": 425210
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.624512129792607e-05,
+      "loss": 0.7486,
+      "step": 425220
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.6243973368026323e-05,
+      "loss": 0.7792,
+      "step": 425230
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.6242825438126576e-05,
+      "loss": 0.7944,
+      "step": 425240
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.624167750822683e-05,
+      "loss": 0.831,
+      "step": 425250
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.624052957832708e-05,
+      "loss": 0.8237,
+      "step": 425260
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.6239381648427336e-05,
+      "loss": 0.7779,
+      "step": 425270
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.6238233718527586e-05,
+      "loss": 0.7978,
+      "step": 425280
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.623708578862784e-05,
+      "loss": 0.8008,
+      "step": 425290
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.6235937858728092e-05,
+      "loss": 0.8858,
+      "step": 425300
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.6234789928828346e-05,
+      "loss": 0.759,
+      "step": 425310
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.6233641998928595e-05,
+      "loss": 0.7768,
+      "step": 425320
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 2.6232494069028852e-05,
+      "loss": 0.7679,
+      "step": 425330
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.6231346139129102e-05,
+      "loss": 0.8864,
+      "step": 425340
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.6230198209229352e-05,
+      "loss": 0.905,
+      "step": 425350
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.622905027932961e-05,
+      "loss": 0.7706,
+      "step": 425360
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.6227902349429858e-05,
+      "loss": 0.7801,
+      "step": 425370
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.622675441953011e-05,
+      "loss": 0.8208,
+      "step": 425380
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.6225606489630365e-05,
+      "loss": 0.8362,
+      "step": 425390
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.6224458559730618e-05,
+      "loss": 0.8138,
+      "step": 425400
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.6223310629830868e-05,
+      "loss": 0.8657,
+      "step": 425410
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.6222162699931124e-05,
+      "loss": 0.8416,
+      "step": 425420
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.6221014770031374e-05,
+      "loss": 0.8837,
+      "step": 425430
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.6219866840131628e-05,
+      "loss": 0.7937,
+      "step": 425440
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.621871891023188e-05,
+      "loss": 0.8482,
+      "step": 425450
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.6217570980332134e-05,
+      "loss": 0.8119,
+      "step": 425460
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.6216423050432384e-05,
+      "loss": 0.8434,
+      "step": 425470
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.621527512053264e-05,
+      "loss": 0.778,
+      "step": 425480
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.621412719063289e-05,
+      "loss": 0.8012,
+      "step": 425490
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.621297926073314e-05,
+      "loss": 0.8291,
+      "step": 425500
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.6211831330833397e-05,
+      "loss": 0.8461,
+      "step": 425510
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.6210683400933647e-05,
+      "loss": 0.8103,
+      "step": 425520
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.62095354710339e-05,
+      "loss": 0.8332,
+      "step": 425530
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.6208387541134153e-05,
+      "loss": 0.8529,
+      "step": 425540
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.6207239611234406e-05,
+      "loss": 0.7664,
+      "step": 425550
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.6206091681334656e-05,
+      "loss": 0.8419,
+      "step": 425560
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.6204943751434913e-05,
+      "loss": 0.8794,
+      "step": 425570
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.6203795821535163e-05,
+      "loss": 0.868,
+      "step": 425580
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.6202647891635416e-05,
+      "loss": 0.8534,
+      "step": 425590
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.620149996173567e-05,
+      "loss": 0.8889,
+      "step": 425600
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.6200352031835922e-05,
+      "loss": 0.8345,
+      "step": 425610
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.6199204101936172e-05,
+      "loss": 0.7904,
+      "step": 425620
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.619805617203643e-05,
+      "loss": 0.7966,
+      "step": 425630
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.619690824213668e-05,
+      "loss": 0.7968,
+      "step": 425640
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.619576031223693e-05,
+      "loss": 0.8564,
+      "step": 425650
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.6194612382337182e-05,
+      "loss": 0.8119,
+      "step": 425660
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.6193464452437435e-05,
+      "loss": 0.8758,
+      "step": 425670
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.619231652253769e-05,
+      "loss": 0.8627,
+      "step": 425680
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.6191168592637938e-05,
+      "loss": 0.7561,
+      "step": 425690
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.6190020662738195e-05,
+      "loss": 0.6926,
+      "step": 425700
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.6188872732838445e-05,
+      "loss": 0.9003,
+      "step": 425710
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.6187724802938698e-05,
+      "loss": 0.7604,
+      "step": 425720
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.618657687303895e-05,
+      "loss": 0.8074,
+      "step": 425730
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.6185428943139204e-05,
+      "loss": 0.7599,
+      "step": 425740
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.6184281013239454e-05,
+      "loss": 0.8872,
+      "step": 425750
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.618313308333971e-05,
+      "loss": 0.8855,
+      "step": 425760
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.618198515343996e-05,
+      "loss": 0.8429,
+      "step": 425770
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.618083722354021e-05,
+      "loss": 0.8476,
+      "step": 425780
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.6179689293640467e-05,
+      "loss": 0.7975,
+      "step": 425790
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.6178541363740717e-05,
+      "loss": 0.7542,
+      "step": 425800
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.617739343384097e-05,
+      "loss": 0.7972,
+      "step": 425810
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.6176245503941224e-05,
+      "loss": 0.7566,
+      "step": 425820
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.6175097574041477e-05,
+      "loss": 0.8079,
+      "step": 425830
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.6173949644141727e-05,
+      "loss": 0.834,
+      "step": 425840
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.6172801714241983e-05,
+      "loss": 0.9259,
+      "step": 425850
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.6171653784342233e-05,
+      "loss": 0.847,
+      "step": 425860
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.6170505854442486e-05,
+      "loss": 0.7853,
+      "step": 425870
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.616935792454274e-05,
+      "loss": 0.8401,
+      "step": 425880
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.6168209994642993e-05,
+      "loss": 0.797,
+      "step": 425890
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.6167062064743243e-05,
+      "loss": 0.8036,
+      "step": 425900
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.61659141348435e-05,
+      "loss": 0.8465,
+      "step": 425910
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.616476620494375e-05,
+      "loss": 0.7857,
+      "step": 425920
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.6163618275044e-05,
+      "loss": 0.7714,
+      "step": 425930
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.6162470345144256e-05,
+      "loss": 0.7925,
+      "step": 425940
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.6161322415244506e-05,
+      "loss": 0.8527,
+      "step": 425950
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.616017448534476e-05,
+      "loss": 0.8209,
+      "step": 425960
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.6159026555445012e-05,
+      "loss": 0.8078,
+      "step": 425970
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 2.6157878625545265e-05,
+      "loss": 0.7877,
+      "step": 425980
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.6156730695645515e-05,
+      "loss": 0.8404,
+      "step": 425990
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.6155582765745772e-05,
+      "loss": 0.767,
+      "step": 426000
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.615443483584602e-05,
+      "loss": 0.8441,
+      "step": 426010
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.6153286905946275e-05,
+      "loss": 0.8399,
+      "step": 426020
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.6152138976046528e-05,
+      "loss": 0.836,
+      "step": 426030
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.615099104614678e-05,
+      "loss": 0.8963,
+      "step": 426040
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.614984311624703e-05,
+      "loss": 0.7544,
+      "step": 426050
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.6148695186347288e-05,
+      "loss": 0.8892,
+      "step": 426060
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.6147547256447538e-05,
+      "loss": 0.8656,
+      "step": 426070
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.6146399326547788e-05,
+      "loss": 0.8,
+      "step": 426080
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.6145251396648044e-05,
+      "loss": 0.8161,
+      "step": 426090
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.6144103466748294e-05,
+      "loss": 0.8127,
+      "step": 426100
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.6142955536848547e-05,
+      "loss": 0.7741,
+      "step": 426110
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.61418076069488e-05,
+      "loss": 0.7696,
+      "step": 426120
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.6140659677049054e-05,
+      "loss": 0.8954,
+      "step": 426130
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.6139511747149304e-05,
+      "loss": 0.8534,
+      "step": 426140
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.613836381724956e-05,
+      "loss": 0.8382,
+      "step": 426150
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.613721588734981e-05,
+      "loss": 0.8819,
+      "step": 426160
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.6136067957450063e-05,
+      "loss": 0.796,
+      "step": 426170
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.6134920027550317e-05,
+      "loss": 0.8529,
+      "step": 426180
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.613377209765057e-05,
+      "loss": 0.8384,
+      "step": 426190
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.613262416775082e-05,
+      "loss": 0.8363,
+      "step": 426200
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.6131476237851076e-05,
+      "loss": 0.7463,
+      "step": 426210
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.6130328307951326e-05,
+      "loss": 0.8723,
+      "step": 426220
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.6129180378051576e-05,
+      "loss": 0.7905,
+      "step": 426230
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.6128032448151833e-05,
+      "loss": 0.8029,
+      "step": 426240
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.6126884518252083e-05,
+      "loss": 0.8841,
+      "step": 426250
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.6125736588352336e-05,
+      "loss": 0.8974,
+      "step": 426260
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.612458865845259e-05,
+      "loss": 0.8405,
+      "step": 426270
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.6123440728552842e-05,
+      "loss": 0.8938,
+      "step": 426280
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.6122292798653092e-05,
+      "loss": 0.7556,
+      "step": 426290
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.612114486875335e-05,
+      "loss": 0.8127,
+      "step": 426300
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.61199969388536e-05,
+      "loss": 0.807,
+      "step": 426310
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.6118849008953852e-05,
+      "loss": 0.8104,
+      "step": 426320
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.6117701079054105e-05,
+      "loss": 0.8071,
+      "step": 426330
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.611655314915436e-05,
+      "loss": 0.8613,
+      "step": 426340
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.6115405219254608e-05,
+      "loss": 0.8798,
+      "step": 426350
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.6114257289354858e-05,
+      "loss": 0.8061,
+      "step": 426360
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.6113109359455115e-05,
+      "loss": 0.855,
+      "step": 426370
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.6111961429555365e-05,
+      "loss": 0.7389,
+      "step": 426380
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.6110813499655618e-05,
+      "loss": 0.7872,
+      "step": 426390
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.610966556975587e-05,
+      "loss": 0.7892,
+      "step": 426400
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.6108517639856124e-05,
+      "loss": 0.7737,
+      "step": 426410
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.6107369709956374e-05,
+      "loss": 0.766,
+      "step": 426420
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.610622178005663e-05,
+      "loss": 0.7472,
+      "step": 426430
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.610507385015688e-05,
+      "loss": 0.8766,
+      "step": 426440
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.610392592025713e-05,
+      "loss": 0.8265,
+      "step": 426450
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.6102777990357387e-05,
+      "loss": 0.8563,
+      "step": 426460
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.610163006045764e-05,
+      "loss": 0.7998,
+      "step": 426470
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.610048213055789e-05,
+      "loss": 0.858,
+      "step": 426480
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.6099334200658147e-05,
+      "loss": 0.7869,
+      "step": 426490
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.6098186270758397e-05,
+      "loss": 0.6858,
+      "step": 426500
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.6097038340858647e-05,
+      "loss": 0.8628,
+      "step": 426510
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.6095890410958903e-05,
+      "loss": 0.8545,
+      "step": 426520
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.6094742481059153e-05,
+      "loss": 0.8524,
+      "step": 426530
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.6093594551159406e-05,
+      "loss": 0.7795,
+      "step": 426540
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.609244662125966e-05,
+      "loss": 0.812,
+      "step": 426550
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.6091298691359913e-05,
+      "loss": 0.8196,
+      "step": 426560
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.6090150761460163e-05,
+      "loss": 0.7584,
+      "step": 426570
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.608900283156042e-05,
+      "loss": 0.9029,
+      "step": 426580
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.608785490166067e-05,
+      "loss": 0.8171,
+      "step": 426590
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.608670697176092e-05,
+      "loss": 0.9388,
+      "step": 426600
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.6085559041861176e-05,
+      "loss": 0.8934,
+      "step": 426610
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.608441111196143e-05,
+      "loss": 0.814,
+      "step": 426620
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.608326318206168e-05,
+      "loss": 0.7753,
+      "step": 426630
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 2.6082115252161935e-05,
+      "loss": 0.885,
+      "step": 426640
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.6080967322262185e-05,
+      "loss": 0.8789,
+      "step": 426650
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.6079819392362435e-05,
+      "loss": 0.7936,
+      "step": 426660
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.607867146246269e-05,
+      "loss": 0.9022,
+      "step": 426670
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.607752353256294e-05,
+      "loss": 0.859,
+      "step": 426680
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.6076375602663195e-05,
+      "loss": 0.735,
+      "step": 426690
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.6075227672763448e-05,
+      "loss": 0.831,
+      "step": 426700
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.60740797428637e-05,
+      "loss": 0.8455,
+      "step": 426710
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.607293181296395e-05,
+      "loss": 0.7746,
+      "step": 426720
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.6071783883064208e-05,
+      "loss": 0.7689,
+      "step": 426730
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.6070635953164458e-05,
+      "loss": 0.7587,
+      "step": 426740
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.6069488023264707e-05,
+      "loss": 0.8574,
+      "step": 426750
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.6068340093364964e-05,
+      "loss": 0.8169,
+      "step": 426760
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.6067192163465217e-05,
+      "loss": 0.9037,
+      "step": 426770
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.6066044233565467e-05,
+      "loss": 0.7412,
+      "step": 426780
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.6064896303665724e-05,
+      "loss": 0.8774,
+      "step": 426790
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.6063748373765974e-05,
+      "loss": 0.8146,
+      "step": 426800
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.6062600443866223e-05,
+      "loss": 0.817,
+      "step": 426810
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.606145251396648e-05,
+      "loss": 0.7674,
+      "step": 426820
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.606030458406673e-05,
+      "loss": 0.9154,
+      "step": 426830
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.6059156654166983e-05,
+      "loss": 0.7664,
+      "step": 426840
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.6058008724267236e-05,
+      "loss": 0.8199,
+      "step": 426850
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.605686079436749e-05,
+      "loss": 0.8975,
+      "step": 426860
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.605571286446774e-05,
+      "loss": 0.8545,
+      "step": 426870
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.6054564934567996e-05,
+      "loss": 0.713,
+      "step": 426880
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.6053417004668246e-05,
+      "loss": 0.8829,
+      "step": 426890
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.6052269074768496e-05,
+      "loss": 0.8164,
+      "step": 426900
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.6051121144868753e-05,
+      "loss": 0.8939,
+      "step": 426910
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.6049973214969006e-05,
+      "loss": 0.8468,
+      "step": 426920
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.6048825285069256e-05,
+      "loss": 0.906,
+      "step": 426930
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.6047677355169512e-05,
+      "loss": 0.7979,
+      "step": 426940
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.6046529425269762e-05,
+      "loss": 0.7835,
+      "step": 426950
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.6045381495370012e-05,
+      "loss": 0.7759,
+      "step": 426960
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.604423356547027e-05,
+      "loss": 0.8595,
+      "step": 426970
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.604308563557052e-05,
+      "loss": 0.8828,
+      "step": 426980
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.604193770567077e-05,
+      "loss": 0.8526,
+      "step": 426990
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.6040789775771025e-05,
+      "loss": 0.8624,
+      "step": 427000
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.6039641845871278e-05,
+      "loss": 0.7784,
+      "step": 427010
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.6038493915971528e-05,
+      "loss": 0.7829,
+      "step": 427020
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.6037345986071785e-05,
+      "loss": 0.8823,
+      "step": 427030
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.6036198056172035e-05,
+      "loss": 0.8669,
+      "step": 427040
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.6035050126272284e-05,
+      "loss": 0.7558,
+      "step": 427050
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.6033902196372538e-05,
+      "loss": 0.7732,
+      "step": 427060
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.6032754266472794e-05,
+      "loss": 0.9249,
+      "step": 427070
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.6031606336573044e-05,
+      "loss": 0.8086,
+      "step": 427080
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.6030458406673294e-05,
+      "loss": 0.7652,
+      "step": 427090
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.602931047677355e-05,
+      "loss": 0.8948,
+      "step": 427100
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.60281625468738e-05,
+      "loss": 0.7841,
+      "step": 427110
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.6027014616974054e-05,
+      "loss": 0.8656,
+      "step": 427120
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.6025866687074307e-05,
+      "loss": 0.783,
+      "step": 427130
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.602471875717456e-05,
+      "loss": 0.8203,
+      "step": 427140
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.602357082727481e-05,
+      "loss": 0.8723,
+      "step": 427150
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.6022422897375067e-05,
+      "loss": 0.8709,
+      "step": 427160
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.6021274967475316e-05,
+      "loss": 0.8184,
+      "step": 427170
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.6020127037575566e-05,
+      "loss": 0.7898,
+      "step": 427180
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.6018979107675823e-05,
+      "loss": 0.7927,
+      "step": 427190
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.6017831177776073e-05,
+      "loss": 0.9251,
+      "step": 427200
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.6016683247876326e-05,
+      "loss": 0.8833,
+      "step": 427210
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.6015535317976583e-05,
+      "loss": 0.7597,
+      "step": 427220
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.6014387388076833e-05,
+      "loss": 0.8016,
+      "step": 427230
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.6013239458177082e-05,
+      "loss": 0.8455,
+      "step": 427240
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.601209152827734e-05,
+      "loss": 0.7689,
+      "step": 427250
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.601094359837759e-05,
+      "loss": 0.8475,
+      "step": 427260
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.6009795668477842e-05,
+      "loss": 0.8999,
+      "step": 427270
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.6008647738578095e-05,
+      "loss": 0.8286,
+      "step": 427280
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 2.600749980867835e-05,
+      "loss": 0.7587,
+      "step": 427290
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.60063518787786e-05,
+      "loss": 0.8848,
+      "step": 427300
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.6005203948878855e-05,
+      "loss": 0.8251,
+      "step": 427310
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.6004056018979105e-05,
+      "loss": 0.7722,
+      "step": 427320
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.6002908089079355e-05,
+      "loss": 0.8015,
+      "step": 427330
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.600176015917961e-05,
+      "loss": 0.8982,
+      "step": 427340
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.600061222927986e-05,
+      "loss": 0.7289,
+      "step": 427350
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.5999464299380115e-05,
+      "loss": 0.8516,
+      "step": 427360
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.599831636948037e-05,
+      "loss": 0.8582,
+      "step": 427370
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.599716843958062e-05,
+      "loss": 0.8953,
+      "step": 427380
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.599602050968087e-05,
+      "loss": 0.8822,
+      "step": 427390
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.5994872579781128e-05,
+      "loss": 0.7875,
+      "step": 427400
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.5993724649881377e-05,
+      "loss": 0.8211,
+      "step": 427410
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.599257671998163e-05,
+      "loss": 0.8822,
+      "step": 427420
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.5991428790081884e-05,
+      "loss": 0.8498,
+      "step": 427430
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.5990280860182137e-05,
+      "loss": 0.793,
+      "step": 427440
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.5989132930282387e-05,
+      "loss": 0.9413,
+      "step": 427450
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.5987985000382644e-05,
+      "loss": 0.8689,
+      "step": 427460
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.5986837070482893e-05,
+      "loss": 0.8676,
+      "step": 427470
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.5985689140583143e-05,
+      "loss": 0.7772,
+      "step": 427480
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.59845412106834e-05,
+      "loss": 0.812,
+      "step": 427490
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.598339328078365e-05,
+      "loss": 0.8688,
+      "step": 427500
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.5982245350883903e-05,
+      "loss": 0.8582,
+      "step": 427510
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.598109742098416e-05,
+      "loss": 0.7635,
+      "step": 427520
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.597994949108441e-05,
+      "loss": 0.799,
+      "step": 427530
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.597880156118466e-05,
+      "loss": 0.8398,
+      "step": 427540
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.5977653631284916e-05,
+      "loss": 0.8597,
+      "step": 427550
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.5976505701385166e-05,
+      "loss": 0.878,
+      "step": 427560
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.597535777148542e-05,
+      "loss": 0.8832,
+      "step": 427570
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.5974209841585672e-05,
+      "loss": 0.8821,
+      "step": 427580
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.5973061911685926e-05,
+      "loss": 0.979,
+      "step": 427590
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.5971913981786175e-05,
+      "loss": 0.7987,
+      "step": 427600
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.5970766051886432e-05,
+      "loss": 0.8745,
+      "step": 427610
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.5969618121986682e-05,
+      "loss": 0.8983,
+      "step": 427620
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.5968470192086932e-05,
+      "loss": 0.8887,
+      "step": 427630
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.596732226218719e-05,
+      "loss": 0.8191,
+      "step": 427640
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.5966174332287438e-05,
+      "loss": 0.8337,
+      "step": 427650
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.596502640238769e-05,
+      "loss": 0.9583,
+      "step": 427660
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.5963878472487948e-05,
+      "loss": 0.867,
+      "step": 427670
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.5962730542588198e-05,
+      "loss": 0.8412,
+      "step": 427680
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.5961582612688448e-05,
+      "loss": 0.8457,
+      "step": 427690
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.5960434682788704e-05,
+      "loss": 0.8048,
+      "step": 427700
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.5959286752888954e-05,
+      "loss": 0.83,
+      "step": 427710
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.5958138822989208e-05,
+      "loss": 0.8474,
+      "step": 427720
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 2.5956990893089457e-05,
+      "loss": 0.8262,
+      "step": 427730
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.5955842963189714e-05,
+      "loss": 0.8857,
+      "step": 427740
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.5954695033289964e-05,
+      "loss": 0.8741,
+      "step": 427750
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.5953547103390214e-05,
+      "loss": 0.7984,
+      "step": 427760
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.595239917349047e-05,
+      "loss": 0.7629,
+      "step": 427770
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.595125124359072e-05,
+      "loss": 0.8074,
+      "step": 427780
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.5950103313690973e-05,
+      "loss": 0.7868,
+      "step": 427790
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.5948955383791227e-05,
+      "loss": 0.8387,
+      "step": 427800
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.594780745389148e-05,
+      "loss": 0.811,
+      "step": 427810
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.594665952399173e-05,
+      "loss": 0.8169,
+      "step": 427820
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.5945511594091986e-05,
+      "loss": 0.7765,
+      "step": 427830
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.5944363664192236e-05,
+      "loss": 0.775,
+      "step": 427840
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.594321573429249e-05,
+      "loss": 0.8155,
+      "step": 427850
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.5942067804392743e-05,
+      "loss": 0.8309,
+      "step": 427860
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.5940919874492996e-05,
+      "loss": 0.8743,
+      "step": 427870
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.5939771944593246e-05,
+      "loss": 0.8245,
+      "step": 427880
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.5938624014693503e-05,
+      "loss": 0.8367,
+      "step": 427890
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.5937476084793752e-05,
+      "loss": 0.7489,
+      "step": 427900
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.5936328154894002e-05,
+      "loss": 0.8454,
+      "step": 427910
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.593518022499426e-05,
+      "loss": 0.8482,
+      "step": 427920
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.593403229509451e-05,
+      "loss": 0.8623,
+      "step": 427930
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 2.5932884365194762e-05,
+      "loss": 0.8561,
+      "step": 427940
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 2.5931736435295015e-05,
+      "loss": 0.889,
+      "step": 427950
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 2.593058850539527e-05,
+      "loss": 0.8393,
+      "step": 427960
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 2.5929440575495518e-05,
+      "loss": 0.8662,
+      "step": 427970
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 2.5928292645595775e-05,
+      "loss": 0.839,
+      "step": 427980
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 2.5927144715696025e-05,
+      "loss": 0.8088,
+      "step": 427990
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 2.5925996785796278e-05,
+      "loss": 0.8024,
+      "step": 428000
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 2.592484885589653e-05,
+      "loss": 0.7798,
+      "step": 428010
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 2.5923700925996785e-05,
+      "loss": 0.8013,
+      "step": 428020
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 2.5922552996097034e-05,
+      "loss": 0.786,
+      "step": 428030
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 2.592140506619729e-05,
+      "loss": 0.7777,
+      "step": 428040
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 2.592025713629754e-05,
+      "loss": 0.8011,
+      "step": 428050
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 2.591910920639779e-05,
+      "loss": 0.8927,
+      "step": 428060
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 2.5917961276498047e-05,
+      "loss": 0.7704,
+      "step": 428070
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 2.5916813346598297e-05,
+      "loss": 0.8253,
+      "step": 428080
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 2.591566541669855e-05,
+      "loss": 0.784,
+      "step": 428090
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 2.5914517486798804e-05,
+      "loss": 0.878,
+      "step": 428100
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 2.5913369556899057e-05,
+      "loss": 0.7765,
+      "step": 428110
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 2.5912221626999307e-05,
+      "loss": 0.9076,
+      "step": 428120
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 2.5911073697099563e-05,
+      "loss": 0.829,
+      "step": 428130
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 2.5909925767199813e-05,
+      "loss": 0.8632,
+      "step": 428140
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 2.5908777837300066e-05,
+      "loss": 0.8198,
+      "step": 428150
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 2.590762990740032e-05,
+      "loss": 0.7765,
+      "step": 428160
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 2.5906481977500573e-05,
+      "loss": 0.8454,
+      "step": 428170
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 2.5905334047600823e-05,
+      "loss": 0.8484,
+      "step": 428180
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 2.590418611770108e-05,
+      "loss": 0.8351,
+      "step": 428190
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 2.590303818780133e-05,
+      "loss": 0.869,
+      "step": 428200
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 2.590189025790158e-05,
+      "loss": 0.8197,
+      "step": 428210
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 2.5900742328001836e-05,
+      "loss": 0.8288,
+      "step": 428220
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 2.5899594398102086e-05,
+      "loss": 0.7803,
+      "step": 428230
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 2.589844646820234e-05,
+      "loss": 0.8268,
+      "step": 428240
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 2.5897298538302592e-05,
+      "loss": 0.8199,
+      "step": 428250
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 2.5896150608402845e-05,
+      "loss": 0.8609,
+      "step": 428260
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 2.5895002678503095e-05,
+      "loss": 0.8106,
+      "step": 428270
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 2.5893854748603352e-05,
+      "loss": 0.7782,
+      "step": 428280
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 2.5892706818703602e-05,
+      "loss": 0.8588,
+      "step": 428290
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 2.5891558888803855e-05,
+      "loss": 0.8047,
+      "step": 428300
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 2.5890410958904108e-05,
+      "loss": 0.8996,
+      "step": 428310
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 2.588926302900436e-05,
+      "loss": 0.7753,
+      "step": 428320
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 2.588811509910461e-05,
+      "loss": 0.7649,
+      "step": 428330
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 2.5886967169204868e-05,
+      "loss": 0.8042,
+      "step": 428340
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 2.5885819239305118e-05,
+      "loss": 0.8732,
+      "step": 428350
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 2.5884671309405368e-05,
+      "loss": 0.818,
+      "step": 428360
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 2.5883523379505624e-05,
+      "loss": 0.7945,
+      "step": 428370
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 2.5882375449605874e-05,
+      "loss": 0.8355,
+      "step": 428380
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 2.5881227519706127e-05,
+      "loss": 0.7879,
+      "step": 428390
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 2.588007958980638e-05,
+      "loss": 0.8336,
+      "step": 428400
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 2.5878931659906634e-05,
+      "loss": 0.796,
+      "step": 428410
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 2.5877783730006884e-05,
+      "loss": 0.9215,
+      "step": 428420
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 2.5876635800107137e-05,
+      "loss": 0.9,
+      "step": 428430
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 2.587548787020739e-05,
+      "loss": 0.8381,
+      "step": 428440
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 2.5874339940307643e-05,
+      "loss": 0.7353,
+      "step": 428450
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 2.5873192010407893e-05,
+      "loss": 0.8809,
+      "step": 428460
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 2.587204408050815e-05,
+      "loss": 0.8124,
+      "step": 428470
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 2.58708961506084e-05,
+      "loss": 0.8104,
+      "step": 428480
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 2.586974822070865e-05,
+      "loss": 0.8411,
+      "step": 428490
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 2.5868600290808906e-05,
+      "loss": 0.8696,
+      "step": 428500
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 2.5867452360909156e-05,
+      "loss": 0.858,
+      "step": 428510
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 2.586630443100941e-05,
+      "loss": 0.8722,
+      "step": 428520
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 2.5865156501109663e-05,
+      "loss": 0.9215,
+      "step": 428530
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 2.5864008571209916e-05,
+      "loss": 0.8039,
+      "step": 428540
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 2.5862860641310166e-05,
+      "loss": 0.8398,
+      "step": 428550
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 2.5861712711410422e-05,
+      "loss": 0.8426,
+      "step": 428560
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 2.5860564781510672e-05,
+      "loss": 0.8339,
+      "step": 428570
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 2.5859416851610925e-05,
+      "loss": 0.7739,
+      "step": 428580
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 2.585826892171118e-05,
+      "loss": 0.865,
+      "step": 428590
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 2.5857120991811432e-05,
+      "loss": 0.7671,
+      "step": 428600
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 2.5855973061911682e-05,
+      "loss": 0.8212,
+      "step": 428610
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 2.585482513201194e-05,
+      "loss": 0.8445,
+      "step": 428620
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 2.5853677202112188e-05,
+      "loss": 0.932,
+      "step": 428630
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 2.5852529272212438e-05,
+      "loss": 0.7748,
+      "step": 428640
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 2.5851381342312695e-05,
+      "loss": 0.79,
+      "step": 428650
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 2.5850233412412945e-05,
+      "loss": 0.8456,
+      "step": 428660
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 2.5849085482513198e-05,
+      "loss": 0.8749,
+      "step": 428670
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 2.584793755261345e-05,
+      "loss": 0.9322,
+      "step": 428680
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 2.5846789622713704e-05,
+      "loss": 0.7892,
+      "step": 428690
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 2.5845641692813954e-05,
+      "loss": 0.8477,
+      "step": 428700
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 2.584449376291421e-05,
+      "loss": 0.8366,
+      "step": 428710
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 2.584334583301446e-05,
+      "loss": 0.7998,
+      "step": 428720
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 2.5842197903114714e-05,
+      "loss": 0.7801,
+      "step": 428730
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 2.5841049973214967e-05,
+      "loss": 0.8137,
+      "step": 428740
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 2.583990204331522e-05,
+      "loss": 0.7924,
+      "step": 428750
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 2.583875411341547e-05,
+      "loss": 0.8268,
+      "step": 428760
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 2.5837606183515727e-05,
+      "loss": 0.7844,
+      "step": 428770
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 2.5836458253615977e-05,
+      "loss": 0.8678,
+      "step": 428780
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 2.5835310323716227e-05,
+      "loss": 0.8227,
+      "step": 428790
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 2.5834162393816483e-05,
+      "loss": 0.7841,
+      "step": 428800
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 2.5833014463916733e-05,
+      "loss": 0.8234,
+      "step": 428810
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 2.5831866534016986e-05,
+      "loss": 0.7538,
+      "step": 428820
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 2.583071860411724e-05,
+      "loss": 0.8678,
+      "step": 428830
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 2.5829570674217493e-05,
+      "loss": 0.8433,
+      "step": 428840
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 2.5828422744317743e-05,
+      "loss": 0.801,
+      "step": 428850
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 2.5827274814418e-05,
+      "loss": 0.8654,
+      "step": 428860
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 2.582612688451825e-05,
+      "loss": 0.8555,
+      "step": 428870
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 2.5824978954618502e-05,
+      "loss": 0.8398,
+      "step": 428880
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 2.5823831024718756e-05,
+      "loss": 0.8337,
+      "step": 428890
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 2.582268309481901e-05,
+      "loss": 0.8174,
+      "step": 428900
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 2.582153516491926e-05,
+      "loss": 0.8989,
+      "step": 428910
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 2.5820387235019515e-05,
+      "loss": 0.847,
+      "step": 428920
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 2.5819239305119765e-05,
+      "loss": 0.84,
+      "step": 428930
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 2.5818091375220015e-05,
+      "loss": 0.7995,
+      "step": 428940
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 2.581694344532027e-05,
+      "loss": 0.8771,
+      "step": 428950
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 2.581579551542052e-05,
+      "loss": 0.7987,
+      "step": 428960
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 2.5814647585520775e-05,
+      "loss": 0.8411,
+      "step": 428970
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 2.5813499655621028e-05,
+      "loss": 0.811,
+      "step": 428980
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 2.581235172572128e-05,
+      "loss": 0.8473,
+      "step": 428990
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 2.581120379582153e-05,
+      "loss": 0.7722,
+      "step": 429000
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 2.5810055865921788e-05,
+      "loss": 0.894,
+      "step": 429010
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 2.5808907936022038e-05,
+      "loss": 0.8093,
+      "step": 429020
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 2.580776000612229e-05,
+      "loss": 0.7759,
+      "step": 429030
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 2.5806612076222544e-05,
+      "loss": 0.8602,
+      "step": 429040
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 2.5805464146322797e-05,
+      "loss": 0.8543,
+      "step": 429050
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 2.5804316216423047e-05,
+      "loss": 0.8301,
+      "step": 429060
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 2.5803168286523304e-05,
+      "loss": 0.7515,
+      "step": 429070
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 2.5802020356623554e-05,
+      "loss": 0.7527,
+      "step": 429080
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 2.5800872426723804e-05,
+      "loss": 0.842,
+      "step": 429090
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 2.5799724496824057e-05,
+      "loss": 0.8866,
+      "step": 429100
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 2.579857656692431e-05,
+      "loss": 0.8178,
+      "step": 429110
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 2.5797428637024563e-05,
+      "loss": 0.8775,
+      "step": 429120
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 2.5796280707124813e-05,
+      "loss": 0.9191,
+      "step": 429130
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 2.579513277722507e-05,
+      "loss": 0.802,
+      "step": 429140
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 2.579398484732532e-05,
+      "loss": 0.7856,
+      "step": 429150
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 2.5792836917425573e-05,
+      "loss": 0.8696,
+      "step": 429160
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 2.5791688987525826e-05,
+      "loss": 0.8453,
+      "step": 429170
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 2.579054105762608e-05,
+      "loss": 0.8907,
+      "step": 429180
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 2.578939312772633e-05,
+      "loss": 0.7833,
+      "step": 429190
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 2.5788245197826586e-05,
+      "loss": 0.8245,
+      "step": 429200
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 2.5787097267926836e-05,
+      "loss": 0.9909,
+      "step": 429210
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 2.5785949338027086e-05,
+      "loss": 0.8442,
+      "step": 429220
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 2.5784801408127342e-05,
+      "loss": 0.8236,
+      "step": 429230
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 2.5783653478227592e-05,
+      "loss": 0.8647,
+      "step": 429240
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 2.5782505548327845e-05,
+      "loss": 0.8636,
+      "step": 429250
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 2.57813576184281e-05,
+      "loss": 0.9435,
+      "step": 429260
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 2.5780209688528352e-05,
+      "loss": 0.8272,
+      "step": 429270
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 2.57790617586286e-05,
+      "loss": 0.7561,
+      "step": 429280
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 2.5777913828728858e-05,
+      "loss": 0.7737,
+      "step": 429290
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 2.5776765898829108e-05,
+      "loss": 0.8423,
+      "step": 429300
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 2.577561796892936e-05,
+      "loss": 0.7526,
+      "step": 429310
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 2.5774470039029615e-05,
+      "loss": 0.7876,
+      "step": 429320
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 2.5773322109129868e-05,
+      "loss": 0.8265,
+      "step": 429330
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 2.5772174179230118e-05,
+      "loss": 0.7631,
+      "step": 429340
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 2.5771026249330374e-05,
+      "loss": 0.7348,
+      "step": 429350
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 2.5769878319430624e-05,
+      "loss": 0.8293,
+      "step": 429360
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 2.5768730389530874e-05,
+      "loss": 0.8834,
+      "step": 429370
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 2.576758245963113e-05,
+      "loss": 0.7953,
+      "step": 429380
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 2.576643452973138e-05,
+      "loss": 0.799,
+      "step": 429390
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 2.5765286599831634e-05,
+      "loss": 0.7924,
+      "step": 429400
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 2.5764138669931887e-05,
+      "loss": 0.8982,
+      "step": 429410
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 2.576299074003214e-05,
+      "loss": 0.877,
+      "step": 429420
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 2.576184281013239e-05,
+      "loss": 0.7766,
+      "step": 429430
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 2.5760694880232647e-05,
+      "loss": 0.8686,
+      "step": 429440
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 2.5759546950332897e-05,
+      "loss": 0.8135,
+      "step": 429450
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 2.575839902043315e-05,
+      "loss": 0.7939,
+      "step": 429460
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 2.5757251090533403e-05,
+      "loss": 0.832,
+      "step": 429470
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 2.5756103160633656e-05,
+      "loss": 0.8818,
+      "step": 429480
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 2.5754955230733906e-05,
+      "loss": 0.8151,
+      "step": 429490
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 2.5753807300834163e-05,
+      "loss": 0.781,
+      "step": 429500
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 2.5752659370934413e-05,
+      "loss": 0.8482,
+      "step": 429510
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 2.5751511441034662e-05,
+      "loss": 0.8348,
+      "step": 429520
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 2.575036351113492e-05,
+      "loss": 0.8311,
+      "step": 429530
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 2.574921558123517e-05,
+      "loss": 0.8565,
+      "step": 429540
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 2.5748067651335422e-05,
+      "loss": 0.8598,
+      "step": 429550
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 2.5746919721435675e-05,
+      "loss": 0.8271,
+      "step": 429560
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 2.574577179153593e-05,
+      "loss": 0.9288,
+      "step": 429570
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 2.574462386163618e-05,
+      "loss": 0.8566,
+      "step": 429580
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 2.5743475931736435e-05,
+      "loss": 0.8144,
+      "step": 429590
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 2.5742328001836685e-05,
+      "loss": 0.841,
+      "step": 429600
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 2.5741180071936938e-05,
+      "loss": 0.8735,
+      "step": 429610
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 2.574003214203719e-05,
+      "loss": 0.7538,
+      "step": 429620
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 2.5738884212137445e-05,
+      "loss": 0.8701,
+      "step": 429630
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 2.5737736282237695e-05,
+      "loss": 0.8864,
+      "step": 429640
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 2.573658835233795e-05,
+      "loss": 0.909,
+      "step": 429650
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 2.57354404224382e-05,
+      "loss": 0.8126,
+      "step": 429660
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 2.573429249253845e-05,
+      "loss": 0.8117,
+      "step": 429670
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 2.5733144562638708e-05,
+      "loss": 0.8665,
+      "step": 429680
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 2.5731996632738957e-05,
+      "loss": 0.8626,
+      "step": 429690
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 2.573084870283921e-05,
+      "loss": 0.8129,
+      "step": 429700
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 2.5729700772939464e-05,
+      "loss": 0.923,
+      "step": 429710
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 2.5728552843039717e-05,
+      "loss": 0.8745,
+      "step": 429720
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 2.5727404913139967e-05,
+      "loss": 0.7675,
+      "step": 429730
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 2.5726256983240224e-05,
+      "loss": 0.7849,
+      "step": 429740
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 2.5725109053340473e-05,
+      "loss": 0.8422,
+      "step": 429750
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 2.5723961123440727e-05,
+      "loss": 0.8646,
+      "step": 429760
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 2.572281319354098e-05,
+      "loss": 0.77,
+      "step": 429770
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 2.5721665263641233e-05,
+      "loss": 0.7339,
+      "step": 429780
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 2.5720517333741483e-05,
+      "loss": 0.8166,
+      "step": 429790
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 2.5719369403841733e-05,
+      "loss": 0.8368,
+      "step": 429800
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 2.571822147394199e-05,
+      "loss": 0.7856,
+      "step": 429810
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 2.571707354404224e-05,
+      "loss": 0.8388,
+      "step": 429820
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 2.5715925614142493e-05,
+      "loss": 0.8064,
+      "step": 429830
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 2.5714777684242746e-05,
+      "loss": 0.8713,
+      "step": 429840
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 2.5713629754343e-05,
+      "loss": 0.8656,
+      "step": 429850
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 2.571248182444325e-05,
+      "loss": 0.8249,
+      "step": 429860
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 2.5711333894543506e-05,
+      "loss": 0.887,
+      "step": 429870
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 2.5710185964643755e-05,
+      "loss": 0.8702,
+      "step": 429880
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 2.570903803474401e-05,
+      "loss": 0.8323,
+      "step": 429890
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 2.5707890104844262e-05,
+      "loss": 0.8384,
+      "step": 429900
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 2.5706742174944515e-05,
+      "loss": 0.769,
+      "step": 429910
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.5705594245044765e-05,
+      "loss": 0.8994,
+      "step": 429920
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.570444631514502e-05,
+      "loss": 0.8049,
+      "step": 429930
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.570329838524527e-05,
+      "loss": 0.89,
+      "step": 429940
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.570215045534552e-05,
+      "loss": 0.7572,
+      "step": 429950
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.5701002525445778e-05,
+      "loss": 0.748,
+      "step": 429960
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.5699854595546028e-05,
+      "loss": 0.7857,
+      "step": 429970
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.569870666564628e-05,
+      "loss": 0.8373,
+      "step": 429980
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.5697558735746534e-05,
+      "loss": 0.754,
+      "step": 429990
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.5696410805846788e-05,
+      "loss": 0.8681,
+      "step": 430000
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.5695262875947037e-05,
+      "loss": 0.8797,
+      "step": 430010
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.5694114946047294e-05,
+      "loss": 0.8804,
+      "step": 430020
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.5692967016147544e-05,
+      "loss": 0.8522,
+      "step": 430030
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.5691819086247797e-05,
+      "loss": 0.8224,
+      "step": 430040
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.569067115634805e-05,
+      "loss": 0.8437,
+      "step": 430050
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.5689523226448304e-05,
+      "loss": 0.7955,
+      "step": 430060
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.5688375296548554e-05,
+      "loss": 0.8843,
+      "step": 430070
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.568722736664881e-05,
+      "loss": 0.8378,
+      "step": 430080
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.568607943674906e-05,
+      "loss": 0.8648,
+      "step": 430090
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.568493150684931e-05,
+      "loss": 0.8645,
+      "step": 430100
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.5683783576949566e-05,
+      "loss": 0.7624,
+      "step": 430110
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 2.5682635647049816e-05,
+      "loss": 0.8446,
+      "step": 430120
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.568148771715007e-05,
+      "loss": 0.8346,
+      "step": 430130
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.5680339787250323e-05,
+      "loss": 0.746,
+      "step": 430140
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.5679191857350576e-05,
+      "loss": 0.7826,
+      "step": 430150
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.5678043927450826e-05,
+      "loss": 0.7995,
+      "step": 430160
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.5676895997551083e-05,
+      "loss": 0.7448,
+      "step": 430170
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.5675748067651332e-05,
+      "loss": 0.9178,
+      "step": 430180
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.5674600137751586e-05,
+      "loss": 0.8571,
+      "step": 430190
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.567345220785184e-05,
+      "loss": 0.7832,
+      "step": 430200
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.5672304277952092e-05,
+      "loss": 0.925,
+      "step": 430210
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.5671156348052342e-05,
+      "loss": 0.8758,
+      "step": 430220
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.56700084181526e-05,
+      "loss": 0.7787,
+      "step": 430230
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.566886048825285e-05,
+      "loss": 0.8705,
+      "step": 430240
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.56677125583531e-05,
+      "loss": 0.7578,
+      "step": 430250
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.5666564628453355e-05,
+      "loss": 0.7989,
+      "step": 430260
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.5665416698553605e-05,
+      "loss": 0.826,
+      "step": 430270
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.5664268768653858e-05,
+      "loss": 0.8028,
+      "step": 430280
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.566312083875411e-05,
+      "loss": 0.8476,
+      "step": 430290
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.5661972908854365e-05,
+      "loss": 0.8556,
+      "step": 430300
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.5660824978954614e-05,
+      "loss": 0.8401,
+      "step": 430310
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.565967704905487e-05,
+      "loss": 0.8621,
+      "step": 430320
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.565852911915512e-05,
+      "loss": 0.795,
+      "step": 430330
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 2.5657381189255374e-05,
+      "loss": 0.8146,
+      "step": 430340
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.5656233259355627e-05,
+      "loss": 0.7983,
+      "step": 430350
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.565508532945588e-05,
+      "loss": 0.8642,
+      "step": 430360
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.565393739955613e-05,
+      "loss": 0.825,
+      "step": 430370
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.5652789469656387e-05,
+      "loss": 0.9052,
+      "step": 430380
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.5651641539756637e-05,
+      "loss": 0.8124,
+      "step": 430390
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.5650493609856887e-05,
+      "loss": 0.8037,
+      "step": 430400
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.5649345679957143e-05,
+      "loss": 0.7666,
+      "step": 430410
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.5648197750057393e-05,
+      "loss": 0.7919,
+      "step": 430420
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.5647049820157647e-05,
+      "loss": 0.8067,
+      "step": 430430
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.56459018902579e-05,
+      "loss": 0.8154,
+      "step": 430440
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.5644753960358153e-05,
+      "loss": 0.85,
+      "step": 430450
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.5643606030458403e-05,
+      "loss": 0.7998,
+      "step": 430460
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.5642458100558656e-05,
+      "loss": 0.8436,
+      "step": 430470
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.564131017065891e-05,
+      "loss": 0.7652,
+      "step": 430480
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.5640162240759163e-05,
+      "loss": 0.8378,
+      "step": 430490
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.5639014310859412e-05,
+      "loss": 0.9255,
+      "step": 430500
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.563786638095967e-05,
+      "loss": 0.7807,
+      "step": 430510
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.563671845105992e-05,
+      "loss": 0.7813,
+      "step": 430520
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.563557052116017e-05,
+      "loss": 0.8254,
+      "step": 430530
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.5634422591260425e-05,
+      "loss": 0.7574,
+      "step": 430540
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.5633274661360675e-05,
+      "loss": 0.8197,
+      "step": 430550
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 2.563212673146093e-05,
+      "loss": 0.8115,
+      "step": 430560
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.5630978801561182e-05,
+      "loss": 0.8797,
+      "step": 430570
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.5629830871661435e-05,
+      "loss": 0.8954,
+      "step": 430580
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.5628682941761685e-05,
+      "loss": 0.8252,
+      "step": 430590
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.562753501186194e-05,
+      "loss": 0.8495,
+      "step": 430600
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.562638708196219e-05,
+      "loss": 0.8023,
+      "step": 430610
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.5625239152062445e-05,
+      "loss": 0.8474,
+      "step": 430620
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.5624091222162698e-05,
+      "loss": 0.8183,
+      "step": 430630
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.562294329226295e-05,
+      "loss": 0.8182,
+      "step": 430640
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.56217953623632e-05,
+      "loss": 0.7839,
+      "step": 430650
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.5620647432463458e-05,
+      "loss": 0.8165,
+      "step": 430660
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.5619499502563707e-05,
+      "loss": 0.7934,
+      "step": 430670
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.5618351572663957e-05,
+      "loss": 0.8555,
+      "step": 430680
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.5617203642764214e-05,
+      "loss": 0.884,
+      "step": 430690
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.5616055712864464e-05,
+      "loss": 0.8565,
+      "step": 430700
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.5614907782964717e-05,
+      "loss": 0.7627,
+      "step": 430710
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.561375985306497e-05,
+      "loss": 0.8129,
+      "step": 430720
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.5612611923165223e-05,
+      "loss": 0.9319,
+      "step": 430730
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.5611463993265473e-05,
+      "loss": 0.9298,
+      "step": 430740
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.561031606336573e-05,
+      "loss": 0.8769,
+      "step": 430750
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.560916813346598e-05,
+      "loss": 0.9043,
+      "step": 430760
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.5608020203566233e-05,
+      "loss": 0.843,
+      "step": 430770
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 2.5606872273666486e-05,
+      "loss": 0.9509,
+      "step": 430780
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.560572434376674e-05,
+      "loss": 0.8103,
+      "step": 430790
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.560457641386699e-05,
+      "loss": 0.7728,
+      "step": 430800
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.5603428483967246e-05,
+      "loss": 0.8413,
+      "step": 430810
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.5602280554067496e-05,
+      "loss": 0.847,
+      "step": 430820
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.5601132624167746e-05,
+      "loss": 0.8093,
+      "step": 430830
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.5599984694268002e-05,
+      "loss": 0.8724,
+      "step": 430840
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.5598836764368252e-05,
+      "loss": 0.8035,
+      "step": 430850
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.5597688834468505e-05,
+      "loss": 0.8266,
+      "step": 430860
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.559654090456876e-05,
+      "loss": 0.7999,
+      "step": 430870
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.5595392974669012e-05,
+      "loss": 0.7659,
+      "step": 430880
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.5594245044769262e-05,
+      "loss": 0.7865,
+      "step": 430890
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.559309711486952e-05,
+      "loss": 0.8269,
+      "step": 430900
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.5591949184969768e-05,
+      "loss": 0.8773,
+      "step": 430910
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.559080125507002e-05,
+      "loss": 0.8769,
+      "step": 430920
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.5589653325170275e-05,
+      "loss": 0.8355,
+      "step": 430930
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.5588505395270528e-05,
+      "loss": 0.7947,
+      "step": 430940
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.5587357465370778e-05,
+      "loss": 0.8289,
+      "step": 430950
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.5586209535471034e-05,
+      "loss": 0.7994,
+      "step": 430960
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.5585061605571284e-05,
+      "loss": 0.8672,
+      "step": 430970
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.5583913675671534e-05,
+      "loss": 0.9066,
+      "step": 430980
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.558276574577179e-05,
+      "loss": 0.8811,
+      "step": 430990
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 2.558161781587204e-05,
+      "loss": 0.895,
+      "step": 431000
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.5580469885972294e-05,
+      "loss": 0.8659,
+      "step": 431010
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.5579321956072547e-05,
+      "loss": 0.7641,
+      "step": 431020
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.55781740261728e-05,
+      "loss": 0.8947,
+      "step": 431030
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.557702609627305e-05,
+      "loss": 0.8472,
+      "step": 431040
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.5575878166373307e-05,
+      "loss": 0.9015,
+      "step": 431050
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.5574730236473557e-05,
+      "loss": 0.828,
+      "step": 431060
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.557358230657381e-05,
+      "loss": 0.8772,
+      "step": 431070
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.5572434376674063e-05,
+      "loss": 0.7898,
+      "step": 431080
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.5571286446774316e-05,
+      "loss": 0.8274,
+      "step": 431090
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.5570138516874566e-05,
+      "loss": 0.7439,
+      "step": 431100
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.5568990586974823e-05,
+      "loss": 0.8277,
+      "step": 431110
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.5567842657075073e-05,
+      "loss": 0.8929,
+      "step": 431120
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.5566694727175323e-05,
+      "loss": 0.8251,
+      "step": 431130
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.556554679727558e-05,
+      "loss": 0.8575,
+      "step": 431140
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.556439886737583e-05,
+      "loss": 0.8475,
+      "step": 431150
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.5563250937476082e-05,
+      "loss": 0.8098,
+      "step": 431160
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.5562103007576332e-05,
+      "loss": 0.7659,
+      "step": 431170
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.556095507767659e-05,
+      "loss": 0.9084,
+      "step": 431180
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.555980714777684e-05,
+      "loss": 0.8736,
+      "step": 431190
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.5558659217877092e-05,
+      "loss": 0.8527,
+      "step": 431200
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.5557511287977345e-05,
+      "loss": 0.8474,
+      "step": 431210
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 2.55563633580776e-05,
+      "loss": 0.851,
+      "step": 431220
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 2.555521542817785e-05,
+      "loss": 0.8627,
+      "step": 431230
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 2.5554067498278105e-05,
+      "loss": 0.7983,
+      "step": 431240
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 2.5552919568378355e-05,
+      "loss": 0.8903,
+      "step": 431250
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 2.5551771638478605e-05,
+      "loss": 0.8181,
+      "step": 431260
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 2.555062370857886e-05,
+      "loss": 0.8963,
+      "step": 431270
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 2.554947577867911e-05,
+      "loss": 0.8502,
+      "step": 431280
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 2.5548327848779364e-05,
+      "loss": 0.953,
+      "step": 431290
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 2.5547179918879618e-05,
+      "loss": 0.8643,
+      "step": 431300
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 2.554603198897987e-05,
+      "loss": 0.7498,
+      "step": 431310
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 2.554488405908012e-05,
+      "loss": 0.8729,
+      "step": 431320
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 2.5543736129180377e-05,
+      "loss": 0.8441,
+      "step": 431330
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 2.5542588199280627e-05,
+      "loss": 0.8739,
+      "step": 431340
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 2.554144026938088e-05,
+      "loss": 0.8658,
+      "step": 431350
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 2.5540292339481134e-05,
+      "loss": 0.7822,
+      "step": 431360
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 2.5539144409581387e-05,
+      "loss": 0.912,
+      "step": 431370
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 2.5537996479681637e-05,
+      "loss": 0.7791,
+      "step": 431380
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 2.5536848549781893e-05,
+      "loss": 0.9017,
+      "step": 431390
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 2.5535700619882143e-05,
+      "loss": 0.7422,
+      "step": 431400
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 2.5534552689982393e-05,
+      "loss": 0.7739,
+      "step": 431410
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 2.553340476008265e-05,
+      "loss": 0.941,
+      "step": 431420
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 2.55322568301829e-05,
+      "loss": 0.8294,
+      "step": 431430
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 2.5531108900283153e-05,
+      "loss": 0.8361,
+      "step": 431440
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 2.5529960970383406e-05,
+      "loss": 0.7003,
+      "step": 431450
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 2.552881304048366e-05,
+      "loss": 0.8555,
+      "step": 431460
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 2.552766511058391e-05,
+      "loss": 0.8765,
+      "step": 431470
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 2.5526517180684166e-05,
+      "loss": 0.8926,
+      "step": 431480
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 2.5525369250784416e-05,
+      "loss": 0.811,
+      "step": 431490
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 2.552422132088467e-05,
+      "loss": 0.8743,
+      "step": 431500
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 2.5523073390984922e-05,
+      "loss": 0.743,
+      "step": 431510
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 2.5521925461085175e-05,
+      "loss": 0.823,
+      "step": 431520
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 2.5520777531185425e-05,
+      "loss": 0.8755,
+      "step": 431530
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 2.5519629601285682e-05,
+      "loss": 0.7266,
+      "step": 431540
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 2.5518481671385932e-05,
+      "loss": 0.8093,
+      "step": 431550
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 2.551733374148618e-05,
+      "loss": 0.9084,
+      "step": 431560
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 2.5516185811586438e-05,
+      "loss": 0.829,
+      "step": 431570
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 2.5515037881686688e-05,
+      "loss": 0.8602,
+      "step": 431580
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 2.551388995178694e-05,
+      "loss": 0.8638,
+      "step": 431590
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 2.5512742021887195e-05,
+      "loss": 0.8293,
+      "step": 431600
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 2.5511594091987448e-05,
+      "loss": 0.8378,
+      "step": 431610
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 2.5510446162087698e-05,
+      "loss": 0.8344,
+      "step": 431620
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 2.5509298232187954e-05,
+      "loss": 0.7963,
+      "step": 431630
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 2.5508150302288204e-05,
+      "loss": 0.8187,
+      "step": 431640
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 2.5507002372388457e-05,
+      "loss": 0.8644,
+      "step": 431650
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 2.550585444248871e-05,
+      "loss": 0.8468,
+      "step": 431660
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 2.5504706512588964e-05,
+      "loss": 0.8697,
+      "step": 431670
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 2.5503558582689214e-05,
+      "loss": 0.9056,
+      "step": 431680
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 2.550241065278947e-05,
+      "loss": 0.9082,
+      "step": 431690
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 2.550126272288972e-05,
+      "loss": 0.8675,
+      "step": 431700
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 2.550011479298997e-05,
+      "loss": 0.9674,
+      "step": 431710
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 2.5498966863090227e-05,
+      "loss": 0.8106,
+      "step": 431720
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 2.5497818933190477e-05,
+      "loss": 0.8168,
+      "step": 431730
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 2.549667100329073e-05,
+      "loss": 0.849,
+      "step": 431740
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 2.5495523073390983e-05,
+      "loss": 0.8395,
+      "step": 431750
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 2.5494375143491236e-05,
+      "loss": 0.7837,
+      "step": 431760
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 2.5493227213591486e-05,
+      "loss": 0.8123,
+      "step": 431770
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 2.5492079283691743e-05,
+      "loss": 0.8565,
+      "step": 431780
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 2.5490931353791993e-05,
+      "loss": 0.8426,
+      "step": 431790
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 2.5489783423892246e-05,
+      "loss": 0.8476,
+      "step": 431800
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 2.54886354939925e-05,
+      "loss": 0.8879,
+      "step": 431810
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 2.5487487564092752e-05,
+      "loss": 0.859,
+      "step": 431820
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 2.5486339634193002e-05,
+      "loss": 0.7811,
+      "step": 431830
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 2.5485191704293252e-05,
+      "loss": 0.8498,
+      "step": 431840
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 2.548404377439351e-05,
+      "loss": 0.8375,
+      "step": 431850
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 2.548289584449376e-05,
+      "loss": 0.8625,
+      "step": 431860
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 2.5481747914594012e-05,
+      "loss": 0.8938,
+      "step": 431870
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 2.5480599984694265e-05,
+      "loss": 0.8754,
+      "step": 431880
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 2.5479452054794518e-05,
+      "loss": 0.8894,
+      "step": 431890
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 2.5478304124894768e-05,
+      "loss": 0.9395,
+      "step": 431900
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 2.5477156194995025e-05,
+      "loss": 0.7998,
+      "step": 431910
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 2.5476008265095275e-05,
+      "loss": 0.8091,
+      "step": 431920
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 2.5474860335195528e-05,
+      "loss": 0.8137,
+      "step": 431930
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 2.547371240529578e-05,
+      "loss": 0.861,
+      "step": 431940
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 2.5472564475396034e-05,
+      "loss": 0.7263,
+      "step": 431950
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 2.5471416545496284e-05,
+      "loss": 0.9685,
+      "step": 431960
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 2.547026861559654e-05,
+      "loss": 0.825,
+      "step": 431970
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 2.546912068569679e-05,
+      "loss": 0.8476,
+      "step": 431980
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 2.546797275579704e-05,
+      "loss": 0.7788,
+      "step": 431990
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 2.5466824825897297e-05,
+      "loss": 0.8201,
+      "step": 432000
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 2.5465676895997547e-05,
+      "loss": 0.767,
+      "step": 432010
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 2.54645289660978e-05,
+      "loss": 0.8978,
+      "step": 432020
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 2.5463381036198053e-05,
+      "loss": 0.7099,
+      "step": 432030
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 2.5462233106298307e-05,
+      "loss": 0.8173,
+      "step": 432040
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 2.5461085176398557e-05,
+      "loss": 0.7905,
+      "step": 432050
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 2.5459937246498813e-05,
+      "loss": 0.762,
+      "step": 432060
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 2.5458789316599063e-05,
+      "loss": 0.7275,
+      "step": 432070
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 2.5457641386699316e-05,
+      "loss": 0.7959,
+      "step": 432080
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 2.545649345679957e-05,
+      "loss": 0.8186,
+      "step": 432090
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 2.5455345526899823e-05,
+      "loss": 0.833,
+      "step": 432100
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 2.5454197597000073e-05,
+      "loss": 0.8177,
+      "step": 432110
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 2.545304966710033e-05,
+      "loss": 0.7854,
+      "step": 432120
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 2.545190173720058e-05,
+      "loss": 0.882,
+      "step": 432130
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 2.545075380730083e-05,
+      "loss": 0.9526,
+      "step": 432140
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 2.5449605877401086e-05,
+      "loss": 0.7598,
+      "step": 432150
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 2.5448457947501335e-05,
+      "loss": 0.7975,
+      "step": 432160
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 2.544731001760159e-05,
+      "loss": 0.825,
+      "step": 432170
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 2.5446162087701842e-05,
+      "loss": 0.8837,
+      "step": 432180
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 2.5445014157802095e-05,
+      "loss": 0.876,
+      "step": 432190
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 2.5443866227902345e-05,
+      "loss": 0.768,
+      "step": 432200
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 2.5442718298002602e-05,
+      "loss": 0.8115,
+      "step": 432210
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 2.544157036810285e-05,
+      "loss": 0.739,
+      "step": 432220
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 2.5440422438203105e-05,
+      "loss": 0.8748,
+      "step": 432230
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 2.5439274508303358e-05,
+      "loss": 0.8346,
+      "step": 432240
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 2.543812657840361e-05,
+      "loss": 0.887,
+      "step": 432250
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 2.543697864850386e-05,
+      "loss": 0.7412,
+      "step": 432260
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 2.5435830718604118e-05,
+      "loss": 0.8162,
+      "step": 432270
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 2.5434682788704368e-05,
+      "loss": 0.9079,
+      "step": 432280
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 2.5433534858804617e-05,
+      "loss": 0.852,
+      "step": 432290
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 2.5432386928904874e-05,
+      "loss": 0.8308,
+      "step": 432300
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 2.5431238999005124e-05,
+      "loss": 0.8134,
+      "step": 432310
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 2.5430091069105377e-05,
+      "loss": 0.7548,
+      "step": 432320
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 2.542894313920563e-05,
+      "loss": 0.8502,
+      "step": 432330
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 2.5427795209305884e-05,
+      "loss": 0.7406,
+      "step": 432340
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 2.5426647279406134e-05,
+      "loss": 0.8337,
+      "step": 432350
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 2.542549934950639e-05,
+      "loss": 0.8299,
+      "step": 432360
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 2.542435141960664e-05,
+      "loss": 0.9017,
+      "step": 432370
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 2.5423203489706893e-05,
+      "loss": 0.8634,
+      "step": 432380
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 2.5422055559807147e-05,
+      "loss": 0.7751,
+      "step": 432390
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 2.54209076299074e-05,
+      "loss": 0.8706,
+      "step": 432400
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 2.541975970000765e-05,
+      "loss": 0.7931,
+      "step": 432410
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 2.5418611770107906e-05,
+      "loss": 0.7901,
+      "step": 432420
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 2.5417463840208156e-05,
+      "loss": 0.8359,
+      "step": 432430
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 2.5416315910308406e-05,
+      "loss": 0.8649,
+      "step": 432440
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 2.5415167980408663e-05,
+      "loss": 0.9116,
+      "step": 432450
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 2.5414020050508912e-05,
+      "loss": 0.7816,
+      "step": 432460
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 2.5412872120609166e-05,
+      "loss": 0.7938,
+      "step": 432470
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 2.541172419070942e-05,
+      "loss": 0.8957,
+      "step": 432480
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 2.5410576260809672e-05,
+      "loss": 0.8533,
+      "step": 432490
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 2.5409428330909922e-05,
+      "loss": 0.9212,
+      "step": 432500
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 2.540828040101018e-05,
+      "loss": 0.8737,
+      "step": 432510
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 2.540713247111043e-05,
+      "loss": 0.9418,
+      "step": 432520
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 2.5405984541210682e-05,
+      "loss": 0.8572,
+      "step": 432530
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 2.540483661131093e-05,
+      "loss": 0.7732,
+      "step": 432540
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 2.5403688681411188e-05,
+      "loss": 0.8251,
+      "step": 432550
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 2.5402540751511438e-05,
+      "loss": 0.8634,
+      "step": 432560
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 2.5401392821611688e-05,
+      "loss": 0.8638,
+      "step": 432570
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 2.5400244891711945e-05,
+      "loss": 0.8992,
+      "step": 432580
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 2.5399096961812194e-05,
+      "loss": 0.8679,
+      "step": 432590
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 2.5397949031912448e-05,
+      "loss": 0.8722,
+      "step": 432600
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 2.53968011020127e-05,
+      "loss": 0.7544,
+      "step": 432610
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 2.5395653172112954e-05,
+      "loss": 0.8527,
+      "step": 432620
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 2.5394505242213204e-05,
+      "loss": 0.8304,
+      "step": 432630
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 2.539335731231346e-05,
+      "loss": 0.8597,
+      "step": 432640
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 2.539220938241371e-05,
+      "loss": 0.9454,
+      "step": 432650
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 2.5391061452513964e-05,
+      "loss": 0.9115,
+      "step": 432660
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 2.5389913522614217e-05,
+      "loss": 0.825,
+      "step": 432670
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 2.538876559271447e-05,
+      "loss": 0.873,
+      "step": 432680
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 2.538761766281472e-05,
+      "loss": 0.9152,
+      "step": 432690
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 2.5386469732914977e-05,
+      "loss": 0.9211,
+      "step": 432700
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 2.5385321803015227e-05,
+      "loss": 0.847,
+      "step": 432710
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 2.5384173873115476e-05,
+      "loss": 0.8957,
+      "step": 432720
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 2.5383025943215733e-05,
+      "loss": 0.9006,
+      "step": 432730
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 2.5381878013315983e-05,
+      "loss": 0.81,
+      "step": 432740
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 2.5380730083416236e-05,
+      "loss": 0.9004,
+      "step": 432750
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 2.537958215351649e-05,
+      "loss": 0.7884,
+      "step": 432760
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 2.5378434223616743e-05,
+      "loss": 0.8205,
+      "step": 432770
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 2.5377286293716992e-05,
+      "loss": 0.8658,
+      "step": 432780
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 2.537613836381725e-05,
+      "loss": 0.9437,
+      "step": 432790
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 2.53749904339175e-05,
+      "loss": 0.8333,
+      "step": 432800
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 2.5373842504017752e-05,
+      "loss": 0.7784,
+      "step": 432810
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 2.5372694574118005e-05,
+      "loss": 0.7638,
+      "step": 432820
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 2.537154664421826e-05,
+      "loss": 0.8656,
+      "step": 432830
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 2.537039871431851e-05,
+      "loss": 0.8857,
+      "step": 432840
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 2.5369250784418765e-05,
+      "loss": 0.8432,
+      "step": 432850
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 2.5368102854519015e-05,
+      "loss": 0.8293,
+      "step": 432860
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 2.5366954924619265e-05,
+      "loss": 0.7938,
+      "step": 432870
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 2.536580699471952e-05,
+      "loss": 0.8582,
+      "step": 432880
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 2.536465906481977e-05,
+      "loss": 0.8724,
+      "step": 432890
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 2.5363511134920025e-05,
+      "loss": 0.9325,
+      "step": 432900
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 2.5362363205020278e-05,
+      "loss": 0.8366,
+      "step": 432910
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 2.536121527512053e-05,
+      "loss": 0.795,
+      "step": 432920
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 2.536006734522078e-05,
+      "loss": 0.8633,
+      "step": 432930
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 2.5358919415321038e-05,
+      "loss": 0.8543,
+      "step": 432940
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 2.5357771485421287e-05,
+      "loss": 0.854,
+      "step": 432950
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 2.535662355552154e-05,
+      "loss": 0.8204,
+      "step": 432960
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 2.5355475625621794e-05,
+      "loss": 0.7279,
+      "step": 432970
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 2.5354327695722047e-05,
+      "loss": 0.8576,
+      "step": 432980
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 2.5353179765822297e-05,
+      "loss": 0.8019,
+      "step": 432990
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 2.5352031835922554e-05,
+      "loss": 0.9119,
+      "step": 433000
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 2.5350883906022803e-05,
+      "loss": 0.7886,
+      "step": 433010
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 2.5349735976123053e-05,
+      "loss": 0.8276,
+      "step": 433020
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 2.534858804622331e-05,
+      "loss": 0.8528,
+      "step": 433030
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 2.534744011632356e-05,
+      "loss": 0.8556,
+      "step": 433040
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 2.5346292186423813e-05,
+      "loss": 0.8579,
+      "step": 433050
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 2.5345144256524066e-05,
+      "loss": 0.7958,
+      "step": 433060
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 2.534399632662432e-05,
+      "loss": 0.905,
+      "step": 433070
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 2.534284839672457e-05,
+      "loss": 0.8878,
+      "step": 433080
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 2.5341700466824826e-05,
+      "loss": 0.892,
+      "step": 433090
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 2.5340552536925076e-05,
+      "loss": 0.928,
+      "step": 433100
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 2.533940460702533e-05,
+      "loss": 0.8448,
+      "step": 433110
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 2.5338256677125582e-05,
+      "loss": 0.7714,
+      "step": 433120
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 2.5337108747225836e-05,
+      "loss": 0.8643,
+      "step": 433130
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 2.5335960817326085e-05,
+      "loss": 0.7755,
+      "step": 433140
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 2.5334812887426342e-05,
+      "loss": 0.9044,
+      "step": 433150
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 2.5333664957526592e-05,
+      "loss": 0.7613,
+      "step": 433160
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 2.5332517027626842e-05,
+      "loss": 0.7935,
+      "step": 433170
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 2.53313690977271e-05,
+      "loss": 0.7575,
+      "step": 433180
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 2.5330221167827348e-05,
+      "loss": 0.9081,
+      "step": 433190
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 2.53290732379276e-05,
+      "loss": 0.7479,
+      "step": 433200
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 2.532792530802785e-05,
+      "loss": 0.8079,
+      "step": 433210
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 2.5326777378128108e-05,
+      "loss": 0.8076,
+      "step": 433220
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 2.5325629448228358e-05,
+      "loss": 0.7849,
+      "step": 433230
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 2.532448151832861e-05,
+      "loss": 0.8216,
+      "step": 433240
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 2.5323333588428864e-05,
+      "loss": 0.77,
+      "step": 433250
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 2.5322185658529118e-05,
+      "loss": 0.8748,
+      "step": 433260
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 2.5321037728629367e-05,
+      "loss": 0.8379,
+      "step": 433270
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 2.5319889798729624e-05,
+      "loss": 0.9097,
+      "step": 433280
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 2.5318741868829874e-05,
+      "loss": 0.8243,
+      "step": 433290
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 2.5317593938930124e-05,
+      "loss": 0.7904,
+      "step": 433300
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 2.531644600903038e-05,
+      "loss": 0.8782,
+      "step": 433310
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 2.531529807913063e-05,
+      "loss": 0.7937,
+      "step": 433320
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 2.5314150149230884e-05,
+      "loss": 0.8894,
+      "step": 433330
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 2.5313002219331137e-05,
+      "loss": 0.7707,
+      "step": 433340
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 2.531185428943139e-05,
+      "loss": 0.7946,
+      "step": 433350
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 2.531070635953164e-05,
+      "loss": 0.7985,
+      "step": 433360
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 2.5309558429631897e-05,
+      "loss": 0.769,
+      "step": 433370
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 2.5308410499732146e-05,
+      "loss": 0.7707,
+      "step": 433380
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 2.53072625698324e-05,
+      "loss": 0.9301,
+      "step": 433390
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.5306114639932653e-05,
+      "loss": 0.8226,
+      "step": 433400
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.5304966710032906e-05,
+      "loss": 0.8305,
+      "step": 433410
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.5303818780133156e-05,
+      "loss": 0.9058,
+      "step": 433420
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.5302670850233413e-05,
+      "loss": 0.9166,
+      "step": 433430
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.5301522920333662e-05,
+      "loss": 0.7911,
+      "step": 433440
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.5300374990433912e-05,
+      "loss": 0.8344,
+      "step": 433450
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.529922706053417e-05,
+      "loss": 0.8389,
+      "step": 433460
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.529807913063442e-05,
+      "loss": 0.7895,
+      "step": 433470
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.5296931200734672e-05,
+      "loss": 0.8395,
+      "step": 433480
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.5295783270834925e-05,
+      "loss": 0.7526,
+      "step": 433490
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.529463534093518e-05,
+      "loss": 0.8241,
+      "step": 433500
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.529348741103543e-05,
+      "loss": 0.7459,
+      "step": 433510
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.5292339481135685e-05,
+      "loss": 0.807,
+      "step": 433520
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.5291191551235935e-05,
+      "loss": 0.8162,
+      "step": 433530
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.5290043621336188e-05,
+      "loss": 0.8181,
+      "step": 433540
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.528889569143644e-05,
+      "loss": 0.8176,
+      "step": 433550
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.5287747761536695e-05,
+      "loss": 0.8306,
+      "step": 433560
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.5286599831636944e-05,
+      "loss": 0.8389,
+      "step": 433570
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.52854519017372e-05,
+      "loss": 0.7806,
+      "step": 433580
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.528430397183745e-05,
+      "loss": 0.8878,
+      "step": 433590
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.52831560419377e-05,
+      "loss": 0.8395,
+      "step": 433600
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.5282008112037957e-05,
+      "loss": 0.8282,
+      "step": 433610
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 2.5280860182138207e-05,
+      "loss": 0.7665,
+      "step": 433620
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 2.527971225223846e-05,
+      "loss": 0.8477,
+      "step": 433630
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 2.5278564322338714e-05,
+      "loss": 0.8089,
+      "step": 433640
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 2.5277416392438967e-05,
+      "loss": 0.8625,
+      "step": 433650
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 2.5276268462539217e-05,
+      "loss": 0.9207,
+      "step": 433660
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 2.5275120532639473e-05,
+      "loss": 0.8156,
+      "step": 433670
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 2.5273972602739723e-05,
+      "loss": 0.8199,
+      "step": 433680
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 2.5272824672839977e-05,
+      "loss": 0.8324,
+      "step": 433690
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 2.527167674294023e-05,
+      "loss": 0.8825,
+      "step": 433700
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 2.5270528813040483e-05,
+      "loss": 0.8336,
+      "step": 433710
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 2.5269380883140733e-05,
+      "loss": 0.9158,
+      "step": 433720
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 2.526823295324099e-05,
+      "loss": 0.86,
+      "step": 433730
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 2.526708502334124e-05,
+      "loss": 0.7862,
+      "step": 433740
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 2.526593709344149e-05,
+      "loss": 0.8499,
+      "step": 433750
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 2.5264789163541746e-05,
+      "loss": 0.912,
+      "step": 433760
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 2.5263641233641996e-05,
+      "loss": 0.8583,
+      "step": 433770
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 2.526249330374225e-05,
+      "loss": 0.9872,
+      "step": 433780
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 2.5261345373842502e-05,
+      "loss": 0.8076,
+      "step": 433790
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 2.5260197443942755e-05,
+      "loss": 0.7904,
+      "step": 433800
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 2.5259049514043005e-05,
+      "loss": 0.7984,
+      "step": 433810
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 2.5257901584143262e-05,
+      "loss": 0.8339,
+      "step": 433820
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 2.5256753654243512e-05,
+      "loss": 0.7991,
+      "step": 433830
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 2.5255605724343765e-05,
+      "loss": 0.762,
+      "step": 433840
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 2.5254457794444018e-05,
+      "loss": 0.8772,
+      "step": 433850
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 2.525330986454427e-05,
+      "loss": 0.8754,
+      "step": 433860
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 2.525216193464452e-05,
+      "loss": 0.8256,
+      "step": 433870
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 2.5251014004744778e-05,
+      "loss": 0.8654,
+      "step": 433880
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 2.5249866074845028e-05,
+      "loss": 0.7947,
+      "step": 433890
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 2.5248718144945278e-05,
+      "loss": 0.8186,
+      "step": 433900
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 2.524757021504553e-05,
+      "loss": 0.8709,
+      "step": 433910
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 2.5246422285145784e-05,
+      "loss": 0.92,
+      "step": 433920
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 2.5245274355246037e-05,
+      "loss": 0.7971,
+      "step": 433930
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 2.5244126425346287e-05,
+      "loss": 0.7293,
+      "step": 433940
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 2.5242978495446544e-05,
+      "loss": 0.801,
+      "step": 433950
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 2.5241830565546794e-05,
+      "loss": 0.8394,
+      "step": 433960
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 2.5240682635647047e-05,
+      "loss": 0.8419,
+      "step": 433970
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 2.52395347057473e-05,
+      "loss": 0.7558,
+      "step": 433980
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 2.5238386775847553e-05,
+      "loss": 0.9157,
+      "step": 433990
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 2.5237238845947803e-05,
+      "loss": 0.8671,
+      "step": 434000
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 2.523609091604806e-05,
+      "loss": 0.8375,
+      "step": 434010
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 2.523494298614831e-05,
+      "loss": 0.88,
+      "step": 434020
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 2.523379505624856e-05,
+      "loss": 0.835,
+      "step": 434030
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 2.5232647126348816e-05,
+      "loss": 0.7461,
+      "step": 434040
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 2.5231499196449066e-05,
+      "loss": 0.8848,
+      "step": 434050
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 2.523035126654932e-05,
+      "loss": 0.8267,
+      "step": 434060
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 2.5229203336649573e-05,
+      "loss": 0.9286,
+      "step": 434070
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 2.5228055406749826e-05,
+      "loss": 0.7532,
+      "step": 434080
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 2.5226907476850076e-05,
+      "loss": 0.7569,
+      "step": 434090
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 2.5225759546950332e-05,
+      "loss": 0.8318,
+      "step": 434100
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 2.5224611617050582e-05,
+      "loss": 0.8783,
+      "step": 434110
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 2.5223463687150835e-05,
+      "loss": 0.8749,
+      "step": 434120
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 2.522231575725109e-05,
+      "loss": 0.8869,
+      "step": 434130
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 2.5221167827351342e-05,
+      "loss": 0.785,
+      "step": 434140
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 2.5220019897451592e-05,
+      "loss": 0.8822,
+      "step": 434150
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 2.521887196755185e-05,
+      "loss": 0.7529,
+      "step": 434160
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 2.5217724037652098e-05,
+      "loss": 0.8891,
+      "step": 434170
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 2.5216576107752348e-05,
+      "loss": 0.8423,
+      "step": 434180
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 2.5215428177852605e-05,
+      "loss": 0.8111,
+      "step": 434190
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 2.5214280247952855e-05,
+      "loss": 0.7922,
+      "step": 434200
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 2.5213132318053108e-05,
+      "loss": 0.8418,
+      "step": 434210
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 2.521198438815336e-05,
+      "loss": 0.7536,
+      "step": 434220
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 2.5210836458253614e-05,
+      "loss": 0.8815,
+      "step": 434230
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 2.5209688528353864e-05,
+      "loss": 0.8266,
+      "step": 434240
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 2.520854059845412e-05,
+      "loss": 0.8334,
+      "step": 434250
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 2.520739266855437e-05,
+      "loss": 0.8274,
+      "step": 434260
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 2.5206244738654624e-05,
+      "loss": 0.8136,
+      "step": 434270
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 2.5205096808754877e-05,
+      "loss": 0.8541,
+      "step": 434280
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 2.520394887885513e-05,
+      "loss": 0.8103,
+      "step": 434290
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 2.520280094895538e-05,
+      "loss": 0.8843,
+      "step": 434300
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 2.5201653019055637e-05,
+      "loss": 0.8983,
+      "step": 434310
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 2.5200505089155887e-05,
+      "loss": 0.7626,
+      "step": 434320
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 2.5199357159256137e-05,
+      "loss": 1.006,
+      "step": 434330
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 2.5198209229356393e-05,
+      "loss": 0.8965,
+      "step": 434340
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 2.5197061299456643e-05,
+      "loss": 0.8397,
+      "step": 434350
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 2.5195913369556896e-05,
+      "loss": 0.8142,
+      "step": 434360
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 2.519476543965715e-05,
+      "loss": 0.8346,
+      "step": 434370
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 2.5193617509757403e-05,
+      "loss": 0.7673,
+      "step": 434380
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 2.5192469579857653e-05,
+      "loss": 0.8384,
+      "step": 434390
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 2.519132164995791e-05,
+      "loss": 0.8202,
+      "step": 434400
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 2.519017372005816e-05,
+      "loss": 0.7912,
+      "step": 434410
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 2.5189025790158412e-05,
+      "loss": 0.977,
+      "step": 434420
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 2.5187877860258666e-05,
+      "loss": 1.0228,
+      "step": 434430
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 2.518672993035892e-05,
+      "loss": 0.8586,
+      "step": 434440
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 2.518558200045917e-05,
+      "loss": 0.8263,
+      "step": 434450
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 2.5184434070559425e-05,
+      "loss": 0.766,
+      "step": 434460
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 2.5183286140659675e-05,
+      "loss": 0.8342,
+      "step": 434470
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 2.5182138210759925e-05,
+      "loss": 0.7513,
+      "step": 434480
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.5180990280860182e-05,
+      "loss": 0.8308,
+      "step": 434490
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.517984235096043e-05,
+      "loss": 0.8603,
+      "step": 434500
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.5178694421060685e-05,
+      "loss": 0.8534,
+      "step": 434510
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.5177546491160938e-05,
+      "loss": 0.7866,
+      "step": 434520
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.517639856126119e-05,
+      "loss": 0.8306,
+      "step": 434530
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.517525063136144e-05,
+      "loss": 0.8021,
+      "step": 434540
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.5174102701461698e-05,
+      "loss": 0.8194,
+      "step": 434550
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.5172954771561948e-05,
+      "loss": 0.8272,
+      "step": 434560
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.51718068416622e-05,
+      "loss": 0.8635,
+      "step": 434570
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.517065891176245e-05,
+      "loss": 0.8491,
+      "step": 434580
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.5169510981862707e-05,
+      "loss": 0.8883,
+      "step": 434590
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.5168363051962957e-05,
+      "loss": 0.8343,
+      "step": 434600
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.5167215122063207e-05,
+      "loss": 0.8856,
+      "step": 434610
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.5166067192163464e-05,
+      "loss": 0.9379,
+      "step": 434620
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.5164919262263714e-05,
+      "loss": 0.8044,
+      "step": 434630
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.5163771332363967e-05,
+      "loss": 0.866,
+      "step": 434640
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.516262340246422e-05,
+      "loss": 0.908,
+      "step": 434650
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.5161475472564473e-05,
+      "loss": 0.8139,
+      "step": 434660
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.5160327542664723e-05,
+      "loss": 0.827,
+      "step": 434670
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.515917961276498e-05,
+      "loss": 0.8403,
+      "step": 434680
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.515803168286523e-05,
+      "loss": 0.8039,
+      "step": 434690
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.5156883752965483e-05,
+      "loss": 0.848,
+      "step": 434700
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.5155735823065736e-05,
+      "loss": 0.8446,
+      "step": 434710
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.515458789316599e-05,
+      "loss": 0.935,
+      "step": 434720
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.515343996326624e-05,
+      "loss": 0.9005,
+      "step": 434730
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.5152292033366496e-05,
+      "loss": 0.8325,
+      "step": 434740
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.5151144103466746e-05,
+      "loss": 0.8376,
+      "step": 434750
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.5149996173566996e-05,
+      "loss": 0.8583,
+      "step": 434760
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.5148848243667252e-05,
+      "loss": 0.7864,
+      "step": 434770
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.5147700313767502e-05,
+      "loss": 0.8576,
+      "step": 434780
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.5146552383867755e-05,
+      "loss": 0.7601,
+      "step": 434790
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.514540445396801e-05,
+      "loss": 0.8146,
+      "step": 434800
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.5144256524068262e-05,
+      "loss": 0.7801,
+      "step": 434810
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.514310859416851e-05,
+      "loss": 0.7558,
+      "step": 434820
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.5141960664268768e-05,
+      "loss": 0.8315,
+      "step": 434830
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.5140812734369018e-05,
+      "loss": 0.817,
+      "step": 434840
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.513966480446927e-05,
+      "loss": 0.7452,
+      "step": 434850
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.5138516874569525e-05,
+      "loss": 0.8228,
+      "step": 434860
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.5137368944669778e-05,
+      "loss": 0.8448,
+      "step": 434870
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.5136221014770028e-05,
+      "loss": 0.8893,
+      "step": 434880
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.5135073084870284e-05,
+      "loss": 0.8461,
+      "step": 434890
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.5133925154970534e-05,
+      "loss": 0.9011,
+      "step": 434900
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.5132777225070784e-05,
+      "loss": 0.7936,
+      "step": 434910
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 2.513162929517104e-05,
+      "loss": 0.8558,
+      "step": 434920
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.513048136527129e-05,
+      "loss": 0.8955,
+      "step": 434930
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.5129333435371544e-05,
+      "loss": 0.8984,
+      "step": 434940
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.5128185505471797e-05,
+      "loss": 0.7369,
+      "step": 434950
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.512703757557205e-05,
+      "loss": 0.8787,
+      "step": 434960
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.51258896456723e-05,
+      "loss": 0.7772,
+      "step": 434970
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.5124741715772557e-05,
+      "loss": 0.8301,
+      "step": 434980
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.5123593785872807e-05,
+      "loss": 0.7555,
+      "step": 434990
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.512244585597306e-05,
+      "loss": 0.9012,
+      "step": 435000
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.5121297926073313e-05,
+      "loss": 0.7563,
+      "step": 435010
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.5120149996173566e-05,
+      "loss": 0.7751,
+      "step": 435020
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.5119002066273816e-05,
+      "loss": 0.8285,
+      "step": 435030
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.5117854136374073e-05,
+      "loss": 0.922,
+      "step": 435040
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.5116706206474323e-05,
+      "loss": 0.7782,
+      "step": 435050
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.5115558276574573e-05,
+      "loss": 0.8517,
+      "step": 435060
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.511441034667483e-05,
+      "loss": 0.8541,
+      "step": 435070
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.511326241677508e-05,
+      "loss": 0.8368,
+      "step": 435080
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.5112114486875332e-05,
+      "loss": 0.8629,
+      "step": 435090
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.5110966556975585e-05,
+      "loss": 0.8821,
+      "step": 435100
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.510981862707584e-05,
+      "loss": 0.7978,
+      "step": 435110
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.510867069717609e-05,
+      "loss": 0.8485,
+      "step": 435120
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.5107522767276345e-05,
+      "loss": 0.8233,
+      "step": 435130
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 2.5106374837376595e-05,
+      "loss": 0.7423,
+      "step": 435140
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.5105226907476848e-05,
+      "loss": 0.8535,
+      "step": 435150
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.51040789775771e-05,
+      "loss": 0.8317,
+      "step": 435160
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.5102931047677355e-05,
+      "loss": 0.7529,
+      "step": 435170
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.5101783117777605e-05,
+      "loss": 0.8679,
+      "step": 435180
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.510063518787786e-05,
+      "loss": 0.8945,
+      "step": 435190
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.509948725797811e-05,
+      "loss": 0.7876,
+      "step": 435200
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.509833932807836e-05,
+      "loss": 0.9471,
+      "step": 435210
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.5097191398178618e-05,
+      "loss": 0.7736,
+      "step": 435220
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.5096043468278867e-05,
+      "loss": 0.8911,
+      "step": 435230
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.509489553837912e-05,
+      "loss": 0.8409,
+      "step": 435240
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.5093747608479374e-05,
+      "loss": 0.8326,
+      "step": 435250
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.5092599678579627e-05,
+      "loss": 0.8337,
+      "step": 435260
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.5091451748679877e-05,
+      "loss": 0.848,
+      "step": 435270
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.509030381878013e-05,
+      "loss": 0.8378,
+      "step": 435280
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.5089155888880384e-05,
+      "loss": 0.7065,
+      "step": 435290
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.5088007958980637e-05,
+      "loss": 0.941,
+      "step": 435300
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.5086860029080887e-05,
+      "loss": 0.8709,
+      "step": 435310
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.5085712099181143e-05,
+      "loss": 0.8736,
+      "step": 435320
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.5084564169281393e-05,
+      "loss": 0.8099,
+      "step": 435330
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.5083416239381643e-05,
+      "loss": 0.8714,
+      "step": 435340
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.50822683094819e-05,
+      "loss": 0.7802,
+      "step": 435350
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 2.508112037958215e-05,
+      "loss": 0.8774,
+      "step": 435360
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 2.5079972449682403e-05,
+      "loss": 0.8102,
+      "step": 435370
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 2.5078824519782656e-05,
+      "loss": 0.8272,
+      "step": 435380
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 2.507767658988291e-05,
+      "loss": 0.7952,
+      "step": 435390
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 2.507652865998316e-05,
+      "loss": 0.7656,
+      "step": 435400
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 2.5075380730083416e-05,
+      "loss": 0.8261,
+      "step": 435410
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 2.5074232800183666e-05,
+      "loss": 0.8082,
+      "step": 435420
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 2.507308487028392e-05,
+      "loss": 0.8357,
+      "step": 435430
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 2.5071936940384172e-05,
+      "loss": 0.8316,
+      "step": 435440
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 2.5070789010484425e-05,
+      "loss": 0.8022,
+      "step": 435450
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 2.5069641080584675e-05,
+      "loss": 0.8988,
+      "step": 435460
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 2.5068493150684932e-05,
+      "loss": 0.9061,
+      "step": 435470
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 2.506734522078518e-05,
+      "loss": 0.7438,
+      "step": 435480
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 2.506619729088543e-05,
+      "loss": 0.8076,
+      "step": 435490
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 2.5065049360985688e-05,
+      "loss": 0.8523,
+      "step": 435500
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 2.5063901431085938e-05,
+      "loss": 0.8667,
+      "step": 435510
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 2.506275350118619e-05,
+      "loss": 0.7683,
+      "step": 435520
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 2.5061605571286444e-05,
+      "loss": 0.8883,
+      "step": 435530
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 2.5060457641386698e-05,
+      "loss": 0.8373,
+      "step": 435540
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 2.5059309711486947e-05,
+      "loss": 0.798,
+      "step": 435550
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 2.5058161781587204e-05,
+      "loss": 0.8979,
+      "step": 435560
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 2.5057013851687454e-05,
+      "loss": 0.7937,
+      "step": 435570
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 2.5055865921787707e-05,
+      "loss": 0.8312,
+      "step": 435580
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 2.505471799188796e-05,
+      "loss": 0.9373,
+      "step": 435590
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 2.5053570061988214e-05,
+      "loss": 0.8915,
+      "step": 435600
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 2.5052422132088464e-05,
+      "loss": 0.7696,
+      "step": 435610
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 2.505127420218872e-05,
+      "loss": 0.857,
+      "step": 435620
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 2.505012627228897e-05,
+      "loss": 0.8448,
+      "step": 435630
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 2.504897834238922e-05,
+      "loss": 0.8404,
+      "step": 435640
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 2.5047830412489477e-05,
+      "loss": 0.8671,
+      "step": 435650
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 2.5046682482589726e-05,
+      "loss": 0.8575,
+      "step": 435660
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 2.504553455268998e-05,
+      "loss": 0.8064,
+      "step": 435670
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 2.5044386622790233e-05,
+      "loss": 0.8516,
+      "step": 435680
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 2.5043238692890486e-05,
+      "loss": 0.8491,
+      "step": 435690
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 2.5042090762990736e-05,
+      "loss": 0.7581,
+      "step": 435700
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 2.5040942833090993e-05,
+      "loss": 0.8332,
+      "step": 435710
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 2.5039794903191242e-05,
+      "loss": 0.8371,
+      "step": 435720
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 2.5038646973291496e-05,
+      "loss": 0.7608,
+      "step": 435730
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 2.503749904339175e-05,
+      "loss": 0.7433,
+      "step": 435740
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 2.5036351113492002e-05,
+      "loss": 0.7836,
+      "step": 435750
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 2.5035203183592252e-05,
+      "loss": 0.8033,
+      "step": 435760
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 2.503405525369251e-05,
+      "loss": 0.8841,
+      "step": 435770
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 2.503290732379276e-05,
+      "loss": 0.7765,
+      "step": 435780
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 2.503175939389301e-05,
+      "loss": 0.825,
+      "step": 435790
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 2.5030611463993265e-05,
+      "loss": 0.9819,
+      "step": 435800
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 2.5029463534093515e-05,
+      "loss": 0.7745,
+      "step": 435810
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 2.5028315604193768e-05,
+      "loss": 0.8531,
+      "step": 435820
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 2.502716767429402e-05,
+      "loss": 0.8806,
+      "step": 435830
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 2.5026019744394275e-05,
+      "loss": 0.869,
+      "step": 435840
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 2.5024871814494524e-05,
+      "loss": 0.8288,
+      "step": 435850
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 2.502372388459478e-05,
+      "loss": 0.8615,
+      "step": 435860
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 2.502257595469503e-05,
+      "loss": 0.8214,
+      "step": 435870
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 2.5021428024795284e-05,
+      "loss": 0.7433,
+      "step": 435880
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 2.5020280094895537e-05,
+      "loss": 0.8017,
+      "step": 435890
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 2.501913216499579e-05,
+      "loss": 0.9186,
+      "step": 435900
+    },
+    {
+      "epoch": 20.0,
+      "eval_cer": 0.6690240661623026,
+      "eval_em": 0.007357212003872217,
+      "eval_f1": 0.007357212003872217,
+      "eval_loss": 1.2976701259613037,
+      "eval_runtime": 1614.1779,
+      "eval_samples_per_second": 6.4,
+      "eval_steps_per_second": 0.8,
+      "eval_wer": 0.9926427879961278,
+      "step": 435904
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 2.501798423509604e-05,
+      "loss": 0.8431,
+      "step": 435910
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 2.5016836305196297e-05,
+      "loss": 0.7826,
+      "step": 435920
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 2.5015688375296547e-05,
+      "loss": 0.7573,
+      "step": 435930
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 2.5014540445396797e-05,
+      "loss": 0.9186,
+      "step": 435940
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 2.5013392515497053e-05,
+      "loss": 0.7256,
+      "step": 435950
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 2.5012244585597303e-05,
+      "loss": 0.8171,
+      "step": 435960
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 2.5011096655697557e-05,
+      "loss": 0.875,
+      "step": 435970
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 2.5009948725797806e-05,
+      "loss": 0.8015,
+      "step": 435980
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 2.5008800795898063e-05,
+      "loss": 0.7863,
+      "step": 435990
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 2.5007652865998313e-05,
+      "loss": 0.7666,
+      "step": 436000
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 2.5006504936098566e-05,
+      "loss": 0.7382,
+      "step": 436010
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 2.500535700619882e-05,
+      "loss": 0.766,
+      "step": 436020
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 2.5004209076299073e-05,
+      "loss": 0.8235,
+      "step": 436030
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 2.5003061146399322e-05,
+      "loss": 0.7664,
+      "step": 436040
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 2.500191321649958e-05,
+      "loss": 0.7828,
+      "step": 436050
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 2.500076528659983e-05,
+      "loss": 0.7802,
+      "step": 436060
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 2.499961735670008e-05,
+      "loss": 0.7959,
+      "step": 436070
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 2.4998469426800335e-05,
+      "loss": 0.7365,
+      "step": 436080
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 2.4997321496900585e-05,
+      "loss": 0.759,
+      "step": 436090
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 2.499617356700084e-05,
+      "loss": 0.8237,
+      "step": 436100
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 2.4995025637101092e-05,
+      "loss": 0.767,
+      "step": 436110
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 2.4993877707201345e-05,
+      "loss": 0.765,
+      "step": 436120
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 2.4992729777301595e-05,
+      "loss": 0.7834,
+      "step": 436130
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 2.499158184740185e-05,
+      "loss": 0.7808,
+      "step": 436140
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 2.49904339175021e-05,
+      "loss": 0.8002,
+      "step": 436150
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 2.4989285987602355e-05,
+      "loss": 0.7989,
+      "step": 436160
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 2.4988138057702608e-05,
+      "loss": 0.7514,
+      "step": 436170
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 2.498699012780286e-05,
+      "loss": 0.7474,
+      "step": 436180
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 2.498584219790311e-05,
+      "loss": 0.7502,
+      "step": 436190
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 2.4984694268003368e-05,
+      "loss": 0.8062,
+      "step": 436200
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 2.4983546338103617e-05,
+      "loss": 0.7754,
+      "step": 436210
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 2.4982398408203867e-05,
+      "loss": 0.8676,
+      "step": 436220
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 2.4981250478304124e-05,
+      "loss": 0.8148,
+      "step": 436230
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 2.4980102548404374e-05,
+      "loss": 0.7506,
+      "step": 436240
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 2.4978954618504627e-05,
+      "loss": 0.8033,
+      "step": 436250
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 2.497780668860488e-05,
+      "loss": 0.833,
+      "step": 436260
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 2.4976658758705134e-05,
+      "loss": 0.809,
+      "step": 436270
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 2.4975510828805383e-05,
+      "loss": 0.6978,
+      "step": 436280
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 2.497436289890564e-05,
+      "loss": 0.7915,
+      "step": 436290
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 2.497321496900589e-05,
+      "loss": 0.7694,
+      "step": 436300
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 2.4972067039106143e-05,
+      "loss": 0.7297,
+      "step": 436310
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 2.4970919109206396e-05,
+      "loss": 0.7388,
+      "step": 436320
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 2.496977117930665e-05,
+      "loss": 0.7554,
+      "step": 436330
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 2.49686232494069e-05,
+      "loss": 0.7899,
+      "step": 436340
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 2.4967475319507156e-05,
+      "loss": 0.7455,
+      "step": 436350
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 2.4966327389607406e-05,
+      "loss": 0.737,
+      "step": 436360
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 2.4965179459707656e-05,
+      "loss": 0.7753,
+      "step": 436370
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 2.4964031529807912e-05,
+      "loss": 0.8306,
+      "step": 436380
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 2.4962883599908162e-05,
+      "loss": 0.7143,
+      "step": 436390
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 2.4961735670008416e-05,
+      "loss": 0.7664,
+      "step": 436400
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 2.496058774010867e-05,
+      "loss": 0.7652,
+      "step": 436410
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 2.4959439810208922e-05,
+      "loss": 0.7659,
+      "step": 436420
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 2.4958291880309172e-05,
+      "loss": 0.8231,
+      "step": 436430
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 2.495714395040943e-05,
+      "loss": 0.8317,
+      "step": 436440
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 2.495599602050968e-05,
+      "loss": 0.8025,
+      "step": 436450
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 2.495484809060993e-05,
+      "loss": 0.7652,
+      "step": 436460
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 2.4953700160710185e-05,
+      "loss": 0.7645,
+      "step": 436470
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 2.4952552230810438e-05,
+      "loss": 0.8079,
+      "step": 436480
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 2.4951404300910688e-05,
+      "loss": 0.8402,
+      "step": 436490
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 2.4950256371010945e-05,
+      "loss": 0.7901,
+      "step": 436500
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 2.4949108441111194e-05,
+      "loss": 0.8033,
+      "step": 436510
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 2.4947960511211444e-05,
+      "loss": 0.7438,
+      "step": 436520
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 2.49468125813117e-05,
+      "loss": 0.8179,
+      "step": 436530
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 2.494566465141195e-05,
+      "loss": 0.8037,
+      "step": 436540
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 2.4944516721512204e-05,
+      "loss": 0.8541,
+      "step": 436550
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 2.4943368791612457e-05,
+      "loss": 0.8204,
+      "step": 436560
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 2.494222086171271e-05,
+      "loss": 0.8009,
+      "step": 436570
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 2.494107293181296e-05,
+      "loss": 0.8029,
+      "step": 436580
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 2.4939925001913217e-05,
+      "loss": 0.7436,
+      "step": 436590
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 2.4938777072013467e-05,
+      "loss": 0.8123,
+      "step": 436600
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 2.493762914211372e-05,
+      "loss": 0.8306,
+      "step": 436610
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 2.4936481212213973e-05,
+      "loss": 0.8741,
+      "step": 436620
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 2.4935333282314227e-05,
+      "loss": 0.7529,
+      "step": 436630
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 2.4934185352414476e-05,
+      "loss": 0.7694,
+      "step": 436640
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 2.4933037422514726e-05,
+      "loss": 0.764,
+      "step": 436650
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 2.4931889492614983e-05,
+      "loss": 0.7868,
+      "step": 436660
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 2.4930741562715233e-05,
+      "loss": 0.7488,
+      "step": 436670
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 2.4929593632815486e-05,
+      "loss": 0.8747,
+      "step": 436680
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 2.492844570291574e-05,
+      "loss": 0.7722,
+      "step": 436690
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 2.4927297773015992e-05,
+      "loss": 0.7538,
+      "step": 436700
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 2.4926149843116242e-05,
+      "loss": 0.79,
+      "step": 436710
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 2.49250019132165e-05,
+      "loss": 0.7887,
+      "step": 436720
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 2.492385398331675e-05,
+      "loss": 0.7323,
+      "step": 436730
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 2.4922706053417002e-05,
+      "loss": 0.8003,
+      "step": 436740
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 2.4921558123517255e-05,
+      "loss": 0.7074,
+      "step": 436750
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 2.492041019361751e-05,
+      "loss": 0.6828,
+      "step": 436760
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 2.491926226371776e-05,
+      "loss": 0.8263,
+      "step": 436770
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 2.4918114333818015e-05,
+      "loss": 0.7623,
+      "step": 436780
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 2.4916966403918265e-05,
+      "loss": 0.7414,
+      "step": 436790
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 2.4915818474018515e-05,
+      "loss": 0.7884,
+      "step": 436800
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 2.491467054411877e-05,
+      "loss": 0.8635,
+      "step": 436810
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 2.491352261421902e-05,
+      "loss": 0.8549,
+      "step": 436820
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 2.4912374684319274e-05,
+      "loss": 0.7676,
+      "step": 436830
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 2.4911226754419528e-05,
+      "loss": 0.7574,
+      "step": 436840
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 2.491007882451978e-05,
+      "loss": 0.8233,
+      "step": 436850
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 2.490893089462003e-05,
+      "loss": 0.8281,
+      "step": 436860
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 2.4907782964720287e-05,
+      "loss": 0.8777,
+      "step": 436870
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 2.4906635034820537e-05,
+      "loss": 0.7768,
+      "step": 436880
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 2.490548710492079e-05,
+      "loss": 0.7704,
+      "step": 436890
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 2.4904339175021044e-05,
+      "loss": 0.7669,
+      "step": 436900
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 2.4903191245121297e-05,
+      "loss": 0.7729,
+      "step": 436910
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 2.4902043315221547e-05,
+      "loss": 0.8278,
+      "step": 436920
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 2.4900895385321803e-05,
+      "loss": 0.8234,
+      "step": 436930
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 2.4899747455422053e-05,
+      "loss": 0.7256,
+      "step": 436940
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 2.4898599525522303e-05,
+      "loss": 0.8003,
+      "step": 436950
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 2.489745159562256e-05,
+      "loss": 0.7424,
+      "step": 436960
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 2.489630366572281e-05,
+      "loss": 0.8075,
+      "step": 436970
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 2.4895155735823063e-05,
+      "loss": 0.8608,
+      "step": 436980
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 2.4894007805923316e-05,
+      "loss": 0.7895,
+      "step": 436990
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 2.489285987602357e-05,
+      "loss": 0.8132,
+      "step": 437000
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 2.489171194612382e-05,
+      "loss": 0.811,
+      "step": 437010
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 2.4890564016224076e-05,
+      "loss": 0.7626,
+      "step": 437020
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 2.4889416086324326e-05,
+      "loss": 0.8194,
+      "step": 437030
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 2.488826815642458e-05,
+      "loss": 0.79,
+      "step": 437040
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 2.4887120226524832e-05,
+      "loss": 0.8572,
+      "step": 437050
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 2.4885972296625085e-05,
+      "loss": 0.7889,
+      "step": 437060
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 2.4884824366725335e-05,
+      "loss": 0.6805,
+      "step": 437070
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 2.4883676436825592e-05,
+      "loss": 0.8075,
+      "step": 437080
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 2.4882528506925842e-05,
+      "loss": 0.8393,
+      "step": 437090
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 2.488138057702609e-05,
+      "loss": 0.7596,
+      "step": 437100
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 2.4880232647126348e-05,
+      "loss": 0.7652,
+      "step": 437110
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 2.4879084717226598e-05,
+      "loss": 0.7732,
+      "step": 437120
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 2.487793678732685e-05,
+      "loss": 0.8011,
+      "step": 437130
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 2.4876788857427105e-05,
+      "loss": 0.7806,
+      "step": 437140
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 2.4875640927527358e-05,
+      "loss": 0.8207,
+      "step": 437150
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 2.4874492997627608e-05,
+      "loss": 0.7739,
+      "step": 437160
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 2.4873345067727864e-05,
+      "loss": 0.779,
+      "step": 437170
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 2.4872197137828114e-05,
+      "loss": 0.8069,
+      "step": 437180
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 2.4871049207928367e-05,
+      "loss": 0.7535,
+      "step": 437190
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 2.486990127802862e-05,
+      "loss": 0.7526,
+      "step": 437200
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 2.4868753348128874e-05,
+      "loss": 0.7707,
+      "step": 437210
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 2.4867605418229124e-05,
+      "loss": 0.8162,
+      "step": 437220
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 2.486645748832938e-05,
+      "loss": 0.7818,
+      "step": 437230
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 2.486530955842963e-05,
+      "loss": 0.883,
+      "step": 437240
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 2.486416162852988e-05,
+      "loss": 0.8034,
+      "step": 437250
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 2.4863013698630137e-05,
+      "loss": 0.7206,
+      "step": 437260
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 2.4861865768730387e-05,
+      "loss": 0.8451,
+      "step": 437270
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 2.486071783883064e-05,
+      "loss": 0.8489,
+      "step": 437280
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 2.4859569908930893e-05,
+      "loss": 0.7751,
+      "step": 437290
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 2.4858421979031146e-05,
+      "loss": 0.7928,
+      "step": 437300
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 2.4857274049131396e-05,
+      "loss": 0.813,
+      "step": 437310
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 2.4856126119231653e-05,
+      "loss": 0.8025,
+      "step": 437320
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 2.4854978189331903e-05,
+      "loss": 0.7662,
+      "step": 437330
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 2.4853830259432156e-05,
+      "loss": 0.7931,
+      "step": 437340
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 2.4852682329532406e-05,
+      "loss": 0.8421,
+      "step": 437350
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 2.4851534399632662e-05,
+      "loss": 0.843,
+      "step": 437360
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 2.4850386469732912e-05,
+      "loss": 0.7594,
+      "step": 437370
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 2.4849238539833162e-05,
+      "loss": 0.7376,
+      "step": 437380
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 2.484809060993342e-05,
+      "loss": 0.8456,
+      "step": 437390
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 2.484694268003367e-05,
+      "loss": 0.79,
+      "step": 437400
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 2.4845794750133922e-05,
+      "loss": 0.743,
+      "step": 437410
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 2.4844646820234175e-05,
+      "loss": 0.8086,
+      "step": 437420
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 2.484349889033443e-05,
+      "loss": 0.8238,
+      "step": 437430
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 2.4842350960434678e-05,
+      "loss": 0.7639,
+      "step": 437440
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 2.4841203030534935e-05,
+      "loss": 0.817,
+      "step": 437450
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 2.4840055100635185e-05,
+      "loss": 0.7289,
+      "step": 437460
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 2.4838907170735438e-05,
+      "loss": 0.836,
+      "step": 437470
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 2.483775924083569e-05,
+      "loss": 0.7753,
+      "step": 437480
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 2.4836611310935944e-05,
+      "loss": 0.7694,
+      "step": 437490
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 2.4835463381036194e-05,
+      "loss": 0.7747,
+      "step": 437500
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 2.483431545113645e-05,
+      "loss": 0.7632,
+      "step": 437510
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 2.48331675212367e-05,
+      "loss": 0.8308,
+      "step": 437520
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 2.483201959133695e-05,
+      "loss": 0.8076,
+      "step": 437530
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 2.4830871661437207e-05,
+      "loss": 0.8324,
+      "step": 437540
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 2.4829723731537457e-05,
+      "loss": 0.8074,
+      "step": 437550
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 2.482857580163771e-05,
+      "loss": 0.8195,
+      "step": 437560
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 2.4827427871737964e-05,
+      "loss": 0.7234,
+      "step": 437570
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 2.4826279941838217e-05,
+      "loss": 0.7287,
+      "step": 437580
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 2.4825132011938467e-05,
+      "loss": 0.8554,
+      "step": 437590
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 2.4823984082038723e-05,
+      "loss": 0.8456,
+      "step": 437600
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 2.4822836152138973e-05,
+      "loss": 0.8113,
+      "step": 437610
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 2.4821688222239226e-05,
+      "loss": 0.8286,
+      "step": 437620
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 2.482054029233948e-05,
+      "loss": 0.797,
+      "step": 437630
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 2.4819392362439733e-05,
+      "loss": 0.8526,
+      "step": 437640
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 2.4818244432539983e-05,
+      "loss": 0.7304,
+      "step": 437650
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 2.481709650264024e-05,
+      "loss": 0.7545,
+      "step": 437660
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 2.481594857274049e-05,
+      "loss": 0.7403,
+      "step": 437670
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 2.481480064284074e-05,
+      "loss": 0.8314,
+      "step": 437680
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 2.4813652712940996e-05,
+      "loss": 0.8115,
+      "step": 437690
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 2.4812504783041246e-05,
+      "loss": 0.8466,
+      "step": 437700
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 2.48113568531415e-05,
+      "loss": 0.7468,
+      "step": 437710
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 2.4810208923241752e-05,
+      "loss": 0.8072,
+      "step": 437720
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 2.4809060993342005e-05,
+      "loss": 0.7458,
+      "step": 437730
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 2.4807913063442255e-05,
+      "loss": 0.9014,
+      "step": 437740
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 2.4806765133542512e-05,
+      "loss": 0.8121,
+      "step": 437750
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 2.480561720364276e-05,
+      "loss": 0.7618,
+      "step": 437760
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 2.4804469273743015e-05,
+      "loss": 0.8372,
+      "step": 437770
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 2.4803321343843268e-05,
+      "loss": 0.6755,
+      "step": 437780
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 2.480217341394352e-05,
+      "loss": 0.8585,
+      "step": 437790
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 2.480102548404377e-05,
+      "loss": 0.8978,
+      "step": 437800
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 2.4799877554144028e-05,
+      "loss": 0.7429,
+      "step": 437810
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 2.4798729624244278e-05,
+      "loss": 0.8138,
+      "step": 437820
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 2.4797581694344528e-05,
+      "loss": 0.7181,
+      "step": 437830
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 2.4796433764444784e-05,
+      "loss": 0.8271,
+      "step": 437840
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 2.4795285834545034e-05,
+      "loss": 0.7234,
+      "step": 437850
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 2.4794137904645287e-05,
+      "loss": 0.7657,
+      "step": 437860
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 2.479298997474554e-05,
+      "loss": 0.8016,
+      "step": 437870
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 2.4791842044845794e-05,
+      "loss": 0.7521,
+      "step": 437880
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 2.4790694114946044e-05,
+      "loss": 0.7928,
+      "step": 437890
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 2.47895461850463e-05,
+      "loss": 0.8201,
+      "step": 437900
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 2.478839825514655e-05,
+      "loss": 0.8074,
+      "step": 437910
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 2.4787250325246803e-05,
+      "loss": 0.8418,
+      "step": 437920
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 2.4786102395347057e-05,
+      "loss": 0.7793,
+      "step": 437930
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 2.478495446544731e-05,
+      "loss": 0.7278,
+      "step": 437940
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 2.478380653554756e-05,
+      "loss": 0.812,
+      "step": 437950
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 2.4782658605647816e-05,
+      "loss": 0.7599,
+      "step": 437960
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 2.4781510675748066e-05,
+      "loss": 0.8091,
+      "step": 437970
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 2.4780362745848316e-05,
+      "loss": 0.8167,
+      "step": 437980
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 2.4779214815948573e-05,
+      "loss": 0.6718,
+      "step": 437990
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 2.4778066886048822e-05,
+      "loss": 0.7256,
+      "step": 438000
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 2.4776918956149076e-05,
+      "loss": 0.8293,
+      "step": 438010
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 2.4775771026249326e-05,
+      "loss": 0.775,
+      "step": 438020
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 2.4774623096349582e-05,
+      "loss": 0.6956,
+      "step": 438030
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 2.4773475166449832e-05,
+      "loss": 0.7304,
+      "step": 438040
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 2.4772327236550085e-05,
+      "loss": 0.8067,
+      "step": 438050
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 2.477117930665034e-05,
+      "loss": 0.7817,
+      "step": 438060
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 2.4770031376750592e-05,
+      "loss": 0.79,
+      "step": 438070
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 2.476888344685084e-05,
+      "loss": 0.8099,
+      "step": 438080
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 2.4767735516951098e-05,
+      "loss": 0.8921,
+      "step": 438090
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 2.4766587587051348e-05,
+      "loss": 0.7629,
+      "step": 438100
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 2.4765439657151598e-05,
+      "loss": 0.8638,
+      "step": 438110
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 2.4764291727251855e-05,
+      "loss": 0.8814,
+      "step": 438120
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 2.4763143797352104e-05,
+      "loss": 0.8547,
+      "step": 438130
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 2.4761995867452358e-05,
+      "loss": 0.6967,
+      "step": 438140
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 2.476084793755261e-05,
+      "loss": 0.9242,
+      "step": 438150
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 2.4759700007652864e-05,
+      "loss": 0.7902,
+      "step": 438160
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 2.4758552077753114e-05,
+      "loss": 0.8758,
+      "step": 438170
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 2.475740414785337e-05,
+      "loss": 0.8008,
+      "step": 438180
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 2.475625621795362e-05,
+      "loss": 0.7858,
+      "step": 438190
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 2.4755108288053874e-05,
+      "loss": 0.8182,
+      "step": 438200
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 2.4753960358154127e-05,
+      "loss": 0.8107,
+      "step": 438210
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 2.475281242825438e-05,
+      "loss": 0.851,
+      "step": 438220
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 2.475166449835463e-05,
+      "loss": 0.7339,
+      "step": 438230
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 2.4750516568454887e-05,
+      "loss": 0.843,
+      "step": 438240
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 2.4749368638555137e-05,
+      "loss": 0.7826,
+      "step": 438250
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 2.4748220708655386e-05,
+      "loss": 0.7914,
+      "step": 438260
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 2.4747072778755643e-05,
+      "loss": 0.7537,
+      "step": 438270
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 2.4745924848855893e-05,
+      "loss": 0.723,
+      "step": 438280
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 2.4744776918956146e-05,
+      "loss": 0.7709,
+      "step": 438290
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 2.47436289890564e-05,
+      "loss": 0.8631,
+      "step": 438300
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 2.4742481059156653e-05,
+      "loss": 0.8322,
+      "step": 438310
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 2.4741333129256903e-05,
+      "loss": 0.8295,
+      "step": 438320
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 2.474018519935716e-05,
+      "loss": 0.8618,
+      "step": 438330
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 2.473903726945741e-05,
+      "loss": 0.8675,
+      "step": 438340
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 2.4737889339557662e-05,
+      "loss": 0.8861,
+      "step": 438350
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 2.4736741409657915e-05,
+      "loss": 0.7803,
+      "step": 438360
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 2.473559347975817e-05,
+      "loss": 0.7614,
+      "step": 438370
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 2.473444554985842e-05,
+      "loss": 0.7789,
+      "step": 438380
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 2.4733297619958675e-05,
+      "loss": 0.7826,
+      "step": 438390
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 2.4732149690058925e-05,
+      "loss": 0.8457,
+      "step": 438400
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 2.4731001760159175e-05,
+      "loss": 0.7171,
+      "step": 438410
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 2.472985383025943e-05,
+      "loss": 0.7858,
+      "step": 438420
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 2.472870590035968e-05,
+      "loss": 0.8705,
+      "step": 438430
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 2.4727557970459935e-05,
+      "loss": 0.8131,
+      "step": 438440
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 2.4726410040560188e-05,
+      "loss": 0.7529,
+      "step": 438450
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 2.472526211066044e-05,
+      "loss": 0.9401,
+      "step": 438460
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 2.472411418076069e-05,
+      "loss": 0.7326,
+      "step": 438470
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 2.4722966250860948e-05,
+      "loss": 0.7608,
+      "step": 438480
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 2.4721818320961197e-05,
+      "loss": 0.7499,
+      "step": 438490
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 2.472067039106145e-05,
+      "loss": 0.7766,
+      "step": 438500
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 2.4719522461161704e-05,
+      "loss": 0.8576,
+      "step": 438510
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 2.4718374531261957e-05,
+      "loss": 0.874,
+      "step": 438520
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 2.4717226601362207e-05,
+      "loss": 0.8139,
+      "step": 438530
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 2.4716078671462464e-05,
+      "loss": 0.9012,
+      "step": 438540
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 2.4714930741562714e-05,
+      "loss": 0.7992,
+      "step": 438550
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 2.4713782811662963e-05,
+      "loss": 0.81,
+      "step": 438560
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 2.471263488176322e-05,
+      "loss": 0.7709,
+      "step": 438570
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 2.471148695186347e-05,
+      "loss": 0.9288,
+      "step": 438580
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 2.4710339021963723e-05,
+      "loss": 0.8295,
+      "step": 438590
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 2.4709191092063976e-05,
+      "loss": 0.8182,
+      "step": 438600
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 2.470804316216423e-05,
+      "loss": 0.843,
+      "step": 438610
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 2.470689523226448e-05,
+      "loss": 0.7714,
+      "step": 438620
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 2.4705747302364736e-05,
+      "loss": 0.9133,
+      "step": 438630
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 2.4704599372464986e-05,
+      "loss": 0.8065,
+      "step": 438640
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 2.470345144256524e-05,
+      "loss": 0.8216,
+      "step": 438650
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 2.4702303512665492e-05,
+      "loss": 0.8275,
+      "step": 438660
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 2.4701155582765746e-05,
+      "loss": 0.7756,
+      "step": 438670
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 2.4700007652865996e-05,
+      "loss": 0.7918,
+      "step": 438680
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 2.4698859722966252e-05,
+      "loss": 0.8051,
+      "step": 438690
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 2.4697711793066502e-05,
+      "loss": 0.7725,
+      "step": 438700
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 2.4696563863166752e-05,
+      "loss": 0.8078,
+      "step": 438710
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 2.4695415933267005e-05,
+      "loss": 0.7786,
+      "step": 438720
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 2.469426800336726e-05,
+      "loss": 0.8091,
+      "step": 438730
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 2.469312007346751e-05,
+      "loss": 0.8283,
+      "step": 438740
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 2.469197214356776e-05,
+      "loss": 0.8067,
+      "step": 438750
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 2.4690824213668018e-05,
+      "loss": 0.7865,
+      "step": 438760
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 2.4689676283768268e-05,
+      "loss": 0.6921,
+      "step": 438770
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 2.468852835386852e-05,
+      "loss": 0.8079,
+      "step": 438780
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 2.4687380423968774e-05,
+      "loss": 0.8002,
+      "step": 438790
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 2.4686232494069028e-05,
+      "loss": 0.7506,
+      "step": 438800
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 2.4685084564169278e-05,
+      "loss": 0.7912,
+      "step": 438810
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 2.4683936634269534e-05,
+      "loss": 0.8051,
+      "step": 438820
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 2.4682788704369784e-05,
+      "loss": 0.9071,
+      "step": 438830
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 2.4681640774470034e-05,
+      "loss": 0.7934,
+      "step": 438840
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 2.468049284457029e-05,
+      "loss": 0.8185,
+      "step": 438850
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 2.467934491467054e-05,
+      "loss": 0.8319,
+      "step": 438860
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 2.4678196984770794e-05,
+      "loss": 0.86,
+      "step": 438870
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 2.4677049054871047e-05,
+      "loss": 0.8239,
+      "step": 438880
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 2.46759011249713e-05,
+      "loss": 0.7112,
+      "step": 438890
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 2.467475319507155e-05,
+      "loss": 0.7031,
+      "step": 438900
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 2.4673605265171807e-05,
+      "loss": 0.8482,
+      "step": 438910
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 2.4672457335272056e-05,
+      "loss": 0.7861,
+      "step": 438920
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 2.467130940537231e-05,
+      "loss": 0.8338,
+      "step": 438930
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 2.4670161475472563e-05,
+      "loss": 0.7836,
+      "step": 438940
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 2.4669013545572816e-05,
+      "loss": 0.7325,
+      "step": 438950
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 2.4667865615673066e-05,
+      "loss": 0.8005,
+      "step": 438960
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 2.4666717685773323e-05,
+      "loss": 0.7752,
+      "step": 438970
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 2.4665569755873572e-05,
+      "loss": 0.7801,
+      "step": 438980
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 2.4664421825973822e-05,
+      "loss": 0.7841,
+      "step": 438990
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 2.466327389607408e-05,
+      "loss": 0.8461,
+      "step": 439000
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 2.466212596617433e-05,
+      "loss": 0.7516,
+      "step": 439010
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 2.4660978036274582e-05,
+      "loss": 0.8664,
+      "step": 439020
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 2.4659830106374835e-05,
+      "loss": 0.7459,
+      "step": 439030
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 2.465868217647509e-05,
+      "loss": 0.7859,
+      "step": 439040
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 2.465753424657534e-05,
+      "loss": 0.8252,
+      "step": 439050
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 2.4656386316675595e-05,
+      "loss": 0.7433,
+      "step": 439060
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 2.4655238386775845e-05,
+      "loss": 0.7613,
+      "step": 439070
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 2.4654090456876098e-05,
+      "loss": 0.8463,
+      "step": 439080
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 2.465294252697635e-05,
+      "loss": 0.7397,
+      "step": 439090
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 2.4651794597076605e-05,
+      "loss": 0.7049,
+      "step": 439100
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 2.4650646667176854e-05,
+      "loss": 0.8575,
+      "step": 439110
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 2.464949873727711e-05,
+      "loss": 0.839,
+      "step": 439120
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 2.464835080737736e-05,
+      "loss": 0.8184,
+      "step": 439130
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 2.464720287747761e-05,
+      "loss": 0.8084,
+      "step": 439140
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 2.4646054947577867e-05,
+      "loss": 0.7825,
+      "step": 439150
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 2.4644907017678117e-05,
+      "loss": 0.8846,
+      "step": 439160
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 2.464375908777837e-05,
+      "loss": 0.8698,
+      "step": 439170
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 2.4642611157878624e-05,
+      "loss": 0.7482,
+      "step": 439180
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 2.4641463227978877e-05,
+      "loss": 0.7983,
+      "step": 439190
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 2.4640315298079127e-05,
+      "loss": 0.7753,
+      "step": 439200
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 2.4639167368179384e-05,
+      "loss": 0.8101,
+      "step": 439210
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 2.4638019438279633e-05,
+      "loss": 0.8724,
+      "step": 439220
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 2.4636871508379887e-05,
+      "loss": 0.7796,
+      "step": 439230
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 2.463572357848014e-05,
+      "loss": 0.756,
+      "step": 439240
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 2.4634575648580393e-05,
+      "loss": 0.7644,
+      "step": 439250
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 2.4633427718680643e-05,
+      "loss": 0.7925,
+      "step": 439260
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 2.46322797887809e-05,
+      "loss": 0.763,
+      "step": 439270
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 2.463113185888115e-05,
+      "loss": 0.845,
+      "step": 439280
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 2.46299839289814e-05,
+      "loss": 0.7944,
+      "step": 439290
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 2.4628835999081656e-05,
+      "loss": 0.8481,
+      "step": 439300
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 2.4627688069181906e-05,
+      "loss": 0.9129,
+      "step": 439310
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 2.462654013928216e-05,
+      "loss": 0.7864,
+      "step": 439320
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 2.4625392209382412e-05,
+      "loss": 0.741,
+      "step": 439330
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 2.4624244279482665e-05,
+      "loss": 0.8988,
+      "step": 439340
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 2.4623096349582915e-05,
+      "loss": 0.7891,
+      "step": 439350
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 2.4621948419683172e-05,
+      "loss": 0.8194,
+      "step": 439360
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 2.4620800489783422e-05,
+      "loss": 0.7767,
+      "step": 439370
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 2.4619652559883675e-05,
+      "loss": 0.7305,
+      "step": 439380
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 2.4618504629983925e-05,
+      "loss": 0.8216,
+      "step": 439390
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 2.461735670008418e-05,
+      "loss": 0.7097,
+      "step": 439400
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 2.461620877018443e-05,
+      "loss": 0.7468,
+      "step": 439410
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 2.461506084028468e-05,
+      "loss": 0.7554,
+      "step": 439420
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 2.4613912910384938e-05,
+      "loss": 0.8038,
+      "step": 439430
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 2.4612764980485188e-05,
+      "loss": 0.8566,
+      "step": 439440
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 2.461161705058544e-05,
+      "loss": 0.8229,
+      "step": 439450
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 2.4610469120685694e-05,
+      "loss": 0.8263,
+      "step": 439460
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 2.4609321190785947e-05,
+      "loss": 0.7601,
+      "step": 439470
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 2.4608173260886197e-05,
+      "loss": 0.8335,
+      "step": 439480
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 2.4607025330986454e-05,
+      "loss": 0.8192,
+      "step": 439490
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 2.4605877401086704e-05,
+      "loss": 0.8219,
+      "step": 439500
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 2.4604729471186957e-05,
+      "loss": 0.8566,
+      "step": 439510
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 2.460358154128721e-05,
+      "loss": 0.7656,
+      "step": 439520
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 2.4602433611387464e-05,
+      "loss": 0.8076,
+      "step": 439530
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 2.4601285681487713e-05,
+      "loss": 0.7946,
+      "step": 439540
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 2.460013775158797e-05,
+      "loss": 0.786,
+      "step": 439550
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 2.459898982168822e-05,
+      "loss": 0.8204,
+      "step": 439560
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 2.459784189178847e-05,
+      "loss": 0.8216,
+      "step": 439570
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 2.4596693961888726e-05,
+      "loss": 0.7973,
+      "step": 439580
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 2.4595546031988976e-05,
+      "loss": 0.768,
+      "step": 439590
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 2.459439810208923e-05,
+      "loss": 0.7938,
+      "step": 439600
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 2.4593250172189483e-05,
+      "loss": 0.7811,
+      "step": 439610
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 2.4592102242289736e-05,
+      "loss": 0.7755,
+      "step": 439620
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 2.4590954312389986e-05,
+      "loss": 0.7949,
+      "step": 439630
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 2.4589806382490242e-05,
+      "loss": 0.764,
+      "step": 439640
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 2.4588658452590492e-05,
+      "loss": 0.7717,
+      "step": 439650
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 2.4587510522690746e-05,
+      "loss": 0.7565,
+      "step": 439660
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 2.4586362592791e-05,
+      "loss": 0.821,
+      "step": 439670
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 2.4585214662891252e-05,
+      "loss": 0.7404,
+      "step": 439680
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 2.4584066732991502e-05,
+      "loss": 0.7457,
+      "step": 439690
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 2.458291880309176e-05,
+      "loss": 0.7438,
+      "step": 439700
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 2.458177087319201e-05,
+      "loss": 0.7579,
+      "step": 439710
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 2.4580622943292258e-05,
+      "loss": 0.7769,
+      "step": 439720
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 2.4579475013392515e-05,
+      "loss": 0.8482,
+      "step": 439730
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 2.4578327083492765e-05,
+      "loss": 0.7442,
+      "step": 439740
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 2.4577179153593018e-05,
+      "loss": 0.783,
+      "step": 439750
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 2.457603122369327e-05,
+      "loss": 0.7962,
+      "step": 439760
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 2.4574883293793524e-05,
+      "loss": 0.7954,
+      "step": 439770
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 2.4573735363893774e-05,
+      "loss": 0.7566,
+      "step": 439780
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 2.457258743399403e-05,
+      "loss": 0.7408,
+      "step": 439790
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 2.457143950409428e-05,
+      "loss": 0.8219,
+      "step": 439800
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 2.4570291574194534e-05,
+      "loss": 0.8455,
+      "step": 439810
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 2.4569143644294787e-05,
+      "loss": 0.8159,
+      "step": 439820
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 2.456799571439504e-05,
+      "loss": 0.7445,
+      "step": 439830
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 2.456684778449529e-05,
+      "loss": 0.729,
+      "step": 439840
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 2.4565699854595547e-05,
+      "loss": 0.811,
+      "step": 439850
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 2.4564551924695797e-05,
+      "loss": 0.8216,
+      "step": 439860
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 2.4563403994796047e-05,
+      "loss": 0.8007,
+      "step": 439870
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 2.4562256064896303e-05,
+      "loss": 0.8348,
+      "step": 439880
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 2.4561108134996553e-05,
+      "loss": 0.8593,
+      "step": 439890
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 2.4559960205096806e-05,
+      "loss": 0.7581,
+      "step": 439900
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 2.455881227519706e-05,
+      "loss": 0.7798,
+      "step": 439910
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 2.4557664345297313e-05,
+      "loss": 0.8421,
+      "step": 439920
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 2.4556516415397563e-05,
+      "loss": 0.7769,
+      "step": 439930
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 2.455536848549782e-05,
+      "loss": 0.7936,
+      "step": 439940
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 2.455422055559807e-05,
+      "loss": 0.7557,
+      "step": 439950
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 2.4553072625698322e-05,
+      "loss": 0.7601,
+      "step": 439960
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 2.4551924695798576e-05,
+      "loss": 0.8412,
+      "step": 439970
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 2.455077676589883e-05,
+      "loss": 0.8315,
+      "step": 439980
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 2.454962883599908e-05,
+      "loss": 0.7916,
+      "step": 439990
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 2.4548480906099335e-05,
+      "loss": 0.8715,
+      "step": 440000
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 2.4547332976199585e-05,
+      "loss": 0.8336,
+      "step": 440010
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 2.4546185046299835e-05,
+      "loss": 0.8618,
+      "step": 440020
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 2.4545037116400092e-05,
+      "loss": 0.8019,
+      "step": 440030
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 2.454388918650034e-05,
+      "loss": 0.8294,
+      "step": 440040
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 2.4542741256600595e-05,
+      "loss": 0.8214,
+      "step": 440050
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 2.4541593326700848e-05,
+      "loss": 0.7986,
+      "step": 440060
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 2.45404453968011e-05,
+      "loss": 0.8347,
+      "step": 440070
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 2.453929746690135e-05,
+      "loss": 0.8799,
+      "step": 440080
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 2.4538149537001604e-05,
+      "loss": 0.7763,
+      "step": 440090
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 2.4537001607101858e-05,
+      "loss": 0.8322,
+      "step": 440100
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 2.453585367720211e-05,
+      "loss": 0.918,
+      "step": 440110
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 2.453470574730236e-05,
+      "loss": 0.7428,
+      "step": 440120
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 2.4533557817402617e-05,
+      "loss": 0.8343,
+      "step": 440130
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 2.4532409887502867e-05,
+      "loss": 0.7842,
+      "step": 440140
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 2.4531261957603117e-05,
+      "loss": 0.7836,
+      "step": 440150
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 2.4530114027703374e-05,
+      "loss": 0.7375,
+      "step": 440160
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 2.4528966097803624e-05,
+      "loss": 0.7275,
+      "step": 440170
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 2.4527818167903877e-05,
+      "loss": 0.72,
+      "step": 440180
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 2.452667023800413e-05,
+      "loss": 0.7879,
+      "step": 440190
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 2.4525522308104383e-05,
+      "loss": 0.8126,
+      "step": 440200
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 2.4524374378204633e-05,
+      "loss": 0.8238,
+      "step": 440210
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 2.452322644830489e-05,
+      "loss": 0.7404,
+      "step": 440220
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 2.452207851840514e-05,
+      "loss": 0.7429,
+      "step": 440230
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 2.4520930588505393e-05,
+      "loss": 0.9392,
+      "step": 440240
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 2.4519782658605646e-05,
+      "loss": 0.7,
+      "step": 440250
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 2.45186347287059e-05,
+      "loss": 0.7697,
+      "step": 440260
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 2.451748679880615e-05,
+      "loss": 0.7725,
+      "step": 440270
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 2.4516338868906406e-05,
+      "loss": 0.7279,
+      "step": 440280
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 2.4515190939006656e-05,
+      "loss": 0.8081,
+      "step": 440290
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 2.4514043009106906e-05,
+      "loss": 0.7423,
+      "step": 440300
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 2.4512895079207162e-05,
+      "loss": 0.8424,
+      "step": 440310
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 2.4511747149307412e-05,
+      "loss": 0.8185,
+      "step": 440320
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 2.4510599219407665e-05,
+      "loss": 0.835,
+      "step": 440330
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 2.450945128950792e-05,
+      "loss": 0.7739,
+      "step": 440340
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 2.4508303359608172e-05,
+      "loss": 0.7115,
+      "step": 440350
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 2.450715542970842e-05,
+      "loss": 0.8336,
+      "step": 440360
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 2.450600749980868e-05,
+      "loss": 0.9152,
+      "step": 440370
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 2.4504859569908928e-05,
+      "loss": 0.797,
+      "step": 440380
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 2.450371164000918e-05,
+      "loss": 0.7891,
+      "step": 440390
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 2.4502563710109435e-05,
+      "loss": 0.7437,
+      "step": 440400
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 2.4501415780209688e-05,
+      "loss": 0.7143,
+      "step": 440410
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 2.4500267850309938e-05,
+      "loss": 0.8165,
+      "step": 440420
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 2.4499119920410194e-05,
+      "loss": 0.7061,
+      "step": 440430
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 2.4497971990510444e-05,
+      "loss": 0.6924,
+      "step": 440440
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 2.4496824060610694e-05,
+      "loss": 0.7952,
+      "step": 440450
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 2.449567613071095e-05,
+      "loss": 0.8334,
+      "step": 440460
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 2.44945282008112e-05,
+      "loss": 0.8364,
+      "step": 440470
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 2.4493380270911454e-05,
+      "loss": 0.7762,
+      "step": 440480
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 2.4492232341011707e-05,
+      "loss": 0.778,
+      "step": 440490
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 2.449108441111196e-05,
+      "loss": 0.8473,
+      "step": 440500
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 2.448993648121221e-05,
+      "loss": 0.7411,
+      "step": 440510
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 2.4488788551312467e-05,
+      "loss": 0.8243,
+      "step": 440520
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 2.4487640621412717e-05,
+      "loss": 0.7691,
+      "step": 440530
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 2.448649269151297e-05,
+      "loss": 0.7984,
+      "step": 440540
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 2.4485344761613223e-05,
+      "loss": 0.8877,
+      "step": 440550
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 2.4484196831713476e-05,
+      "loss": 0.7873,
+      "step": 440560
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 2.4483048901813726e-05,
+      "loss": 0.802,
+      "step": 440570
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 2.4481900971913983e-05,
+      "loss": 0.7724,
+      "step": 440580
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 2.4480753042014233e-05,
+      "loss": 0.8372,
+      "step": 440590
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 2.4479605112114483e-05,
+      "loss": 0.8286,
+      "step": 440600
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 2.447845718221474e-05,
+      "loss": 0.8818,
+      "step": 440610
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 2.447730925231499e-05,
+      "loss": 0.8293,
+      "step": 440620
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 2.4476161322415242e-05,
+      "loss": 0.7905,
+      "step": 440630
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 2.4475013392515496e-05,
+      "loss": 0.8104,
+      "step": 440640
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 2.447386546261575e-05,
+      "loss": 0.7551,
+      "step": 440650
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 2.4472717532716e-05,
+      "loss": 0.7086,
+      "step": 440660
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 2.4471569602816255e-05,
+      "loss": 0.8325,
+      "step": 440670
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 2.4470421672916505e-05,
+      "loss": 0.8404,
+      "step": 440680
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 2.446927374301676e-05,
+      "loss": 0.806,
+      "step": 440690
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 2.446812581311701e-05,
+      "loss": 0.8461,
+      "step": 440700
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 2.4466977883217265e-05,
+      "loss": 0.8281,
+      "step": 440710
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 2.4465829953317515e-05,
+      "loss": 0.806,
+      "step": 440720
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 2.446468202341777e-05,
+      "loss": 0.7975,
+      "step": 440730
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 2.446353409351802e-05,
+      "loss": 0.7926,
+      "step": 440740
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 2.446238616361827e-05,
+      "loss": 0.7774,
+      "step": 440750
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 2.4461238233718524e-05,
+      "loss": 0.8702,
+      "step": 440760
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 2.4460090303818778e-05,
+      "loss": 0.7942,
+      "step": 440770
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 2.445894237391903e-05,
+      "loss": 0.8258,
+      "step": 440780
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 2.445779444401928e-05,
+      "loss": 0.7936,
+      "step": 440790
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 2.4456646514119537e-05,
+      "loss": 0.8028,
+      "step": 440800
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 2.4455498584219787e-05,
+      "loss": 0.8132,
+      "step": 440810
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 2.445435065432004e-05,
+      "loss": 0.8179,
+      "step": 440820
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 2.4453202724420294e-05,
+      "loss": 0.8067,
+      "step": 440830
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 2.4452054794520547e-05,
+      "loss": 0.8275,
+      "step": 440840
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 2.4450906864620797e-05,
+      "loss": 0.8419,
+      "step": 440850
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 2.4449758934721053e-05,
+      "loss": 0.7637,
+      "step": 440860
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 2.4448611004821303e-05,
+      "loss": 0.8162,
+      "step": 440870
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 2.4447463074921553e-05,
+      "loss": 0.8414,
+      "step": 440880
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 2.444631514502181e-05,
+      "loss": 0.8445,
+      "step": 440890
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 2.444516721512206e-05,
+      "loss": 0.798,
+      "step": 440900
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 2.4444019285222313e-05,
+      "loss": 0.8356,
+      "step": 440910
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 2.4442871355322566e-05,
+      "loss": 0.7493,
+      "step": 440920
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 2.444172342542282e-05,
+      "loss": 0.8003,
+      "step": 440930
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 2.444057549552307e-05,
+      "loss": 0.8731,
+      "step": 440940
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 2.4439427565623326e-05,
+      "loss": 0.7285,
+      "step": 440950
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 2.4438279635723576e-05,
+      "loss": 0.7742,
+      "step": 440960
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 2.443713170582383e-05,
+      "loss": 0.7959,
+      "step": 440970
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 2.4435983775924082e-05,
+      "loss": 0.7526,
+      "step": 440980
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 2.4434835846024335e-05,
+      "loss": 0.8001,
+      "step": 440990
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 2.4433687916124585e-05,
+      "loss": 0.8195,
+      "step": 441000
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 2.4432539986224842e-05,
+      "loss": 0.8185,
+      "step": 441010
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 2.443139205632509e-05,
+      "loss": 0.8003,
+      "step": 441020
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 2.443024412642534e-05,
+      "loss": 0.8204,
+      "step": 441030
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 2.4429096196525598e-05,
+      "loss": 0.8379,
+      "step": 441040
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 2.4427948266625848e-05,
+      "loss": 0.8793,
+      "step": 441050
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 2.44268003367261e-05,
+      "loss": 0.7767,
+      "step": 441060
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 2.4425652406826354e-05,
+      "loss": 0.7296,
+      "step": 441070
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 2.4424504476926608e-05,
+      "loss": 0.7801,
+      "step": 441080
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 2.4423356547026858e-05,
+      "loss": 0.8434,
+      "step": 441090
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 2.4422208617127114e-05,
+      "loss": 0.8337,
+      "step": 441100
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 2.4421060687227364e-05,
+      "loss": 0.7225,
+      "step": 441110
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 2.4419912757327617e-05,
+      "loss": 0.7927,
+      "step": 441120
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 2.441876482742787e-05,
+      "loss": 0.76,
+      "step": 441130
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 2.4417616897528124e-05,
+      "loss": 0.8021,
+      "step": 441140
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 2.4416468967628374e-05,
+      "loss": 0.8378,
+      "step": 441150
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 2.441532103772863e-05,
+      "loss": 0.7681,
+      "step": 441160
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 2.441417310782888e-05,
+      "loss": 0.7917,
+      "step": 441170
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 2.441302517792913e-05,
+      "loss": 0.9029,
+      "step": 441180
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 2.4411877248029387e-05,
+      "loss": 0.7568,
+      "step": 441190
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 2.4410729318129636e-05,
+      "loss": 0.783,
+      "step": 441200
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 2.440958138822989e-05,
+      "loss": 0.8335,
+      "step": 441210
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 2.4408433458330143e-05,
+      "loss": 0.8266,
+      "step": 441220
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 2.4407285528430396e-05,
+      "loss": 0.817,
+      "step": 441230
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 2.4406137598530646e-05,
+      "loss": 0.9007,
+      "step": 441240
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 2.4404989668630903e-05,
+      "loss": 0.7357,
+      "step": 441250
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 2.4403841738731153e-05,
+      "loss": 0.806,
+      "step": 441260
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 2.4402693808831406e-05,
+      "loss": 0.7914,
+      "step": 441270
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 2.440154587893166e-05,
+      "loss": 0.8406,
+      "step": 441280
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 2.4400397949031912e-05,
+      "loss": 0.8582,
+      "step": 441290
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 2.4399250019132162e-05,
+      "loss": 0.7789,
+      "step": 441300
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 2.439810208923242e-05,
+      "loss": 0.8613,
+      "step": 441310
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 2.439695415933267e-05,
+      "loss": 0.8341,
+      "step": 441320
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 2.439580622943292e-05,
+      "loss": 0.8712,
+      "step": 441330
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 2.4394658299533175e-05,
+      "loss": 0.7455,
+      "step": 441340
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 2.4393510369633425e-05,
+      "loss": 0.7875,
+      "step": 441350
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 2.4392362439733678e-05,
+      "loss": 0.7733,
+      "step": 441360
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 2.439121450983393e-05,
+      "loss": 0.774,
+      "step": 441370
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 2.4390066579934185e-05,
+      "loss": 0.79,
+      "step": 441380
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 2.4388918650034435e-05,
+      "loss": 0.8079,
+      "step": 441390
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 2.438777072013469e-05,
+      "loss": 0.7875,
+      "step": 441400
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 2.438662279023494e-05,
+      "loss": 0.8029,
+      "step": 441410
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 2.4385474860335194e-05,
+      "loss": 0.8331,
+      "step": 441420
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 2.4384326930435447e-05,
+      "loss": 0.8663,
+      "step": 441430
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 2.43831790005357e-05,
+      "loss": 0.8832,
+      "step": 441440
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 2.438203107063595e-05,
+      "loss": 0.7726,
+      "step": 441450
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 2.43808831407362e-05,
+      "loss": 0.8381,
+      "step": 441460
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 2.4379735210836457e-05,
+      "loss": 0.763,
+      "step": 441470
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 2.4378587280936707e-05,
+      "loss": 0.7462,
+      "step": 441480
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 2.437743935103696e-05,
+      "loss": 0.8602,
+      "step": 441490
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 2.4376291421137213e-05,
+      "loss": 0.8403,
+      "step": 441500
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 2.4375143491237467e-05,
+      "loss": 0.7947,
+      "step": 441510
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 2.4373995561337716e-05,
+      "loss": 0.7756,
+      "step": 441520
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 2.4372847631437973e-05,
+      "loss": 0.7665,
+      "step": 441530
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 2.4371699701538223e-05,
+      "loss": 0.7785,
+      "step": 441540
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 2.4370551771638476e-05,
+      "loss": 0.7683,
+      "step": 441550
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 2.436940384173873e-05,
+      "loss": 0.8565,
+      "step": 441560
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 2.4368255911838983e-05,
+      "loss": 0.7568,
+      "step": 441570
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 2.4367107981939233e-05,
+      "loss": 0.849,
+      "step": 441580
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 2.436596005203949e-05,
+      "loss": 0.9122,
+      "step": 441590
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 2.436481212213974e-05,
+      "loss": 0.7465,
+      "step": 441600
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 2.436366419223999e-05,
+      "loss": 0.812,
+      "step": 441610
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 2.4362516262340246e-05,
+      "loss": 0.7811,
+      "step": 441620
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 2.4361368332440495e-05,
+      "loss": 0.8368,
+      "step": 441630
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 2.436022040254075e-05,
+      "loss": 0.7666,
+      "step": 441640
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 2.4359072472641002e-05,
+      "loss": 0.8516,
+      "step": 441650
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 2.4357924542741255e-05,
+      "loss": 0.7112,
+      "step": 441660
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 2.4356776612841505e-05,
+      "loss": 0.8251,
+      "step": 441670
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 2.435562868294176e-05,
+      "loss": 0.8002,
+      "step": 441680
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 2.435448075304201e-05,
+      "loss": 0.7276,
+      "step": 441690
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 2.4353332823142265e-05,
+      "loss": 0.8252,
+      "step": 441700
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 2.4352184893242518e-05,
+      "loss": 0.8855,
+      "step": 441710
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 2.435103696334277e-05,
+      "loss": 0.8393,
+      "step": 441720
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 2.434988903344302e-05,
+      "loss": 0.869,
+      "step": 441730
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 2.4348741103543278e-05,
+      "loss": 0.7261,
+      "step": 441740
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 2.4347593173643528e-05,
+      "loss": 0.8846,
+      "step": 441750
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 2.4346445243743777e-05,
+      "loss": 0.7825,
+      "step": 441760
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 2.4345297313844034e-05,
+      "loss": 0.7943,
+      "step": 441770
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 2.4344149383944284e-05,
+      "loss": 0.7534,
+      "step": 441780
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 2.4343001454044537e-05,
+      "loss": 0.7072,
+      "step": 441790
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 2.434185352414479e-05,
+      "loss": 0.8084,
+      "step": 441800
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 2.4340705594245044e-05,
+      "loss": 0.8011,
+      "step": 441810
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 2.4339557664345293e-05,
+      "loss": 0.8127,
+      "step": 441820
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 2.433840973444555e-05,
+      "loss": 0.7927,
+      "step": 441830
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 2.43372618045458e-05,
+      "loss": 0.7984,
+      "step": 441840
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 2.4336113874646053e-05,
+      "loss": 0.8137,
+      "step": 441850
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 2.4334965944746306e-05,
+      "loss": 0.7848,
+      "step": 441860
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 2.433381801484656e-05,
+      "loss": 0.7978,
+      "step": 441870
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 2.433267008494681e-05,
+      "loss": 0.8415,
+      "step": 441880
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 2.4331522155047066e-05,
+      "loss": 0.8623,
+      "step": 441890
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 2.4330374225147316e-05,
+      "loss": 0.7964,
+      "step": 441900
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 2.4329226295247566e-05,
+      "loss": 0.7777,
+      "step": 441910
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 2.4328078365347822e-05,
+      "loss": 0.8537,
+      "step": 441920
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 2.4326930435448072e-05,
+      "loss": 0.7578,
+      "step": 441930
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 2.4325782505548326e-05,
+      "loss": 0.8364,
+      "step": 441940
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 2.432463457564858e-05,
+      "loss": 0.8065,
+      "step": 441950
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 2.4323486645748832e-05,
+      "loss": 0.8171,
+      "step": 441960
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 2.4322338715849082e-05,
+      "loss": 0.8339,
+      "step": 441970
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 2.432119078594934e-05,
+      "loss": 0.8427,
+      "step": 441980
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 2.432004285604959e-05,
+      "loss": 0.8272,
+      "step": 441990
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 2.431889492614984e-05,
+      "loss": 0.8085,
+      "step": 442000
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 2.4317746996250095e-05,
+      "loss": 0.7454,
+      "step": 442010
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 2.4316599066350348e-05,
+      "loss": 0.9096,
+      "step": 442020
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 2.4315451136450598e-05,
+      "loss": 0.7843,
+      "step": 442030
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 2.4314303206550855e-05,
+      "loss": 0.8042,
+      "step": 442040
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 2.4313155276651104e-05,
+      "loss": 0.8181,
+      "step": 442050
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 2.4312007346751354e-05,
+      "loss": 0.7785,
+      "step": 442060
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 2.431085941685161e-05,
+      "loss": 0.8894,
+      "step": 442070
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 2.430971148695186e-05,
+      "loss": 0.9195,
+      "step": 442080
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 2.4308563557052114e-05,
+      "loss": 0.7235,
+      "step": 442090
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 2.4307415627152367e-05,
+      "loss": 0.8449,
+      "step": 442100
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 2.430626769725262e-05,
+      "loss": 0.7578,
+      "step": 442110
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 2.430511976735287e-05,
+      "loss": 0.7759,
+      "step": 442120
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 2.4303971837453124e-05,
+      "loss": 0.7846,
+      "step": 442130
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 2.4302823907553377e-05,
+      "loss": 0.8327,
+      "step": 442140
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 2.430167597765363e-05,
+      "loss": 0.8262,
+      "step": 442150
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 2.430052804775388e-05,
+      "loss": 0.7715,
+      "step": 442160
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 2.4299380117854137e-05,
+      "loss": 0.8201,
+      "step": 442170
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 2.4298232187954386e-05,
+      "loss": 0.7407,
+      "step": 442180
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 2.4297084258054636e-05,
+      "loss": 0.8374,
+      "step": 442190
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 2.4295936328154893e-05,
+      "loss": 0.7091,
+      "step": 442200
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 2.4294788398255143e-05,
+      "loss": 0.906,
+      "step": 442210
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 2.4293640468355396e-05,
+      "loss": 0.8634,
+      "step": 442220
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 2.429249253845565e-05,
+      "loss": 0.8421,
+      "step": 442230
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 2.4291344608555903e-05,
+      "loss": 0.7926,
+      "step": 442240
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 2.4290196678656152e-05,
+      "loss": 0.8112,
+      "step": 442250
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 2.428904874875641e-05,
+      "loss": 0.8212,
+      "step": 442260
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 2.428790081885666e-05,
+      "loss": 0.7811,
+      "step": 442270
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 2.4286752888956912e-05,
+      "loss": 0.7708,
+      "step": 442280
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 2.4285604959057165e-05,
+      "loss": 0.7348,
+      "step": 442290
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 2.428445702915742e-05,
+      "loss": 0.7468,
+      "step": 442300
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 2.428330909925767e-05,
+      "loss": 0.8528,
+      "step": 442310
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 2.4282161169357925e-05,
+      "loss": 0.8281,
+      "step": 442320
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 2.4281013239458175e-05,
+      "loss": 0.7432,
+      "step": 442330
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 2.4279865309558425e-05,
+      "loss": 0.8404,
+      "step": 442340
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 2.427871737965868e-05,
+      "loss": 0.7809,
+      "step": 442350
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 2.427756944975893e-05,
+      "loss": 0.7847,
+      "step": 442360
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 2.4276421519859184e-05,
+      "loss": 0.8289,
+      "step": 442370
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 2.4275273589959438e-05,
+      "loss": 0.7172,
+      "step": 442380
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 2.427412566005969e-05,
+      "loss": 0.7726,
+      "step": 442390
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 2.427297773015994e-05,
+      "loss": 0.8613,
+      "step": 442400
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 2.4271829800260197e-05,
+      "loss": 0.7512,
+      "step": 442410
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 2.4270681870360447e-05,
+      "loss": 0.8311,
+      "step": 442420
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 2.42695339404607e-05,
+      "loss": 0.7568,
+      "step": 442430
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 2.4268386010560954e-05,
+      "loss": 0.7794,
+      "step": 442440
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 2.4267238080661207e-05,
+      "loss": 0.8049,
+      "step": 442450
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 2.4266090150761457e-05,
+      "loss": 0.6769,
+      "step": 442460
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 2.4264942220861714e-05,
+      "loss": 0.8058,
+      "step": 442470
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 2.4263794290961963e-05,
+      "loss": 0.833,
+      "step": 442480
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 2.4262646361062213e-05,
+      "loss": 0.8142,
+      "step": 442490
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 2.426149843116247e-05,
+      "loss": 0.7859,
+      "step": 442500
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 2.426035050126272e-05,
+      "loss": 0.7275,
+      "step": 442510
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 2.4259202571362973e-05,
+      "loss": 0.7685,
+      "step": 442520
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 2.4258054641463226e-05,
+      "loss": 0.8107,
+      "step": 442530
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 2.425690671156348e-05,
+      "loss": 0.7492,
+      "step": 442540
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 2.425575878166373e-05,
+      "loss": 0.7389,
+      "step": 442550
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 2.4254610851763986e-05,
+      "loss": 0.8115,
+      "step": 442560
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 2.4253462921864236e-05,
+      "loss": 0.8787,
+      "step": 442570
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 2.425231499196449e-05,
+      "loss": 0.7668,
+      "step": 442580
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 2.4251167062064742e-05,
+      "loss": 0.839,
+      "step": 442590
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 2.4250019132164996e-05,
+      "loss": 0.7465,
+      "step": 442600
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 2.4248871202265245e-05,
+      "loss": 0.79,
+      "step": 442610
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 2.4247723272365502e-05,
+      "loss": 0.8197,
+      "step": 442620
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 2.4246575342465752e-05,
+      "loss": 0.8522,
+      "step": 442630
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 2.4245427412566002e-05,
+      "loss": 0.835,
+      "step": 442640
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 2.424427948266626e-05,
+      "loss": 0.764,
+      "step": 442650
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 2.4243131552766508e-05,
+      "loss": 0.8301,
+      "step": 442660
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 2.424198362286676e-05,
+      "loss": 0.8196,
+      "step": 442670
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 2.4240835692967015e-05,
+      "loss": 0.7351,
+      "step": 442680
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 2.4239687763067268e-05,
+      "loss": 0.7848,
+      "step": 442690
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 2.4238539833167518e-05,
+      "loss": 0.814,
+      "step": 442700
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 2.4237391903267774e-05,
+      "loss": 0.8045,
+      "step": 442710
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 2.4236243973368024e-05,
+      "loss": 0.9392,
+      "step": 442720
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 2.4235096043468278e-05,
+      "loss": 0.7658,
+      "step": 442730
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 2.423394811356853e-05,
+      "loss": 0.8228,
+      "step": 442740
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 2.4232800183668784e-05,
+      "loss": 0.8057,
+      "step": 442750
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 2.4231652253769034e-05,
+      "loss": 0.7659,
+      "step": 442760
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 2.423050432386929e-05,
+      "loss": 0.8422,
+      "step": 442770
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 2.422935639396954e-05,
+      "loss": 0.7551,
+      "step": 442780
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 2.422820846406979e-05,
+      "loss": 0.8368,
+      "step": 442790
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 2.4227060534170047e-05,
+      "loss": 0.8067,
+      "step": 442800
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 2.4225912604270297e-05,
+      "loss": 0.7751,
+      "step": 442810
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 2.422476467437055e-05,
+      "loss": 0.8097,
+      "step": 442820
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 2.42236167444708e-05,
+      "loss": 0.8323,
+      "step": 442830
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 2.4222468814571056e-05,
+      "loss": 0.7945,
+      "step": 442840
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 2.4221320884671306e-05,
+      "loss": 0.8253,
+      "step": 442850
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 2.422017295477156e-05,
+      "loss": 0.7787,
+      "step": 442860
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 2.4219025024871813e-05,
+      "loss": 0.8615,
+      "step": 442870
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 2.4217877094972066e-05,
+      "loss": 0.7167,
+      "step": 442880
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 2.4216729165072316e-05,
+      "loss": 0.7053,
+      "step": 442890
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 2.4215581235172572e-05,
+      "loss": 0.9016,
+      "step": 442900
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 2.4214433305272822e-05,
+      "loss": 0.756,
+      "step": 442910
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 2.4213285375373072e-05,
+      "loss": 0.8154,
+      "step": 442920
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 2.421213744547333e-05,
+      "loss": 0.8661,
+      "step": 442930
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 2.421098951557358e-05,
+      "loss": 0.844,
+      "step": 442940
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 2.4209841585673832e-05,
+      "loss": 0.8496,
+      "step": 442950
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 2.4208693655774085e-05,
+      "loss": 0.8011,
+      "step": 442960
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 2.420754572587434e-05,
+      "loss": 0.7703,
+      "step": 442970
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 2.4206397795974588e-05,
+      "loss": 0.8056,
+      "step": 442980
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 2.4205249866074845e-05,
+      "loss": 0.8039,
+      "step": 442990
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 2.4204101936175095e-05,
+      "loss": 0.8919,
+      "step": 443000
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 2.4202954006275348e-05,
+      "loss": 0.8824,
+      "step": 443010
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 2.42018060763756e-05,
+      "loss": 0.7953,
+      "step": 443020
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 2.4200658146475854e-05,
+      "loss": 0.7501,
+      "step": 443030
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 2.4199510216576104e-05,
+      "loss": 0.7841,
+      "step": 443040
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 2.419836228667636e-05,
+      "loss": 0.8339,
+      "step": 443050
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 2.419721435677661e-05,
+      "loss": 0.7769,
+      "step": 443060
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 2.419606642687686e-05,
+      "loss": 0.7466,
+      "step": 443070
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 2.4194918496977117e-05,
+      "loss": 0.7738,
+      "step": 443080
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 2.4193770567077367e-05,
+      "loss": 0.8604,
+      "step": 443090
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 2.419262263717762e-05,
+      "loss": 0.7742,
+      "step": 443100
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 2.4191474707277874e-05,
+      "loss": 0.8584,
+      "step": 443110
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 2.4190326777378127e-05,
+      "loss": 0.7847,
+      "step": 443120
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 2.4189178847478377e-05,
+      "loss": 0.8279,
+      "step": 443130
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 2.4188030917578633e-05,
+      "loss": 0.793,
+      "step": 443140
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 2.4186882987678883e-05,
+      "loss": 0.8326,
+      "step": 443150
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 2.4185735057779136e-05,
+      "loss": 0.8188,
+      "step": 443160
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 2.418458712787939e-05,
+      "loss": 0.8731,
+      "step": 443170
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 2.4183439197979643e-05,
+      "loss": 0.7946,
+      "step": 443180
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 2.4182291268079893e-05,
+      "loss": 0.7581,
+      "step": 443190
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 2.418114333818015e-05,
+      "loss": 0.8934,
+      "step": 443200
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 2.41799954082804e-05,
+      "loss": 0.7194,
+      "step": 443210
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 2.417884747838065e-05,
+      "loss": 0.7838,
+      "step": 443220
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 2.4177699548480906e-05,
+      "loss": 0.7753,
+      "step": 443230
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 2.4176551618581156e-05,
+      "loss": 0.7879,
+      "step": 443240
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 2.417540368868141e-05,
+      "loss": 0.803,
+      "step": 443250
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 2.4174255758781662e-05,
+      "loss": 0.7907,
+      "step": 443260
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 2.4173107828881915e-05,
+      "loss": 0.8465,
+      "step": 443270
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 2.4171959898982165e-05,
+      "loss": 0.8198,
+      "step": 443280
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 2.4170811969082422e-05,
+      "loss": 0.9604,
+      "step": 443290
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 2.416966403918267e-05,
+      "loss": 0.7843,
+      "step": 443300
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 2.4168516109282925e-05,
+      "loss": 0.7889,
+      "step": 443310
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 2.4167368179383178e-05,
+      "loss": 0.8353,
+      "step": 443320
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 2.416622024948343e-05,
+      "loss": 0.7573,
+      "step": 443330
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 2.416507231958368e-05,
+      "loss": 0.8563,
+      "step": 443340
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 2.4163924389683938e-05,
+      "loss": 0.8711,
+      "step": 443350
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 2.4162776459784188e-05,
+      "loss": 0.8585,
+      "step": 443360
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 2.4161628529884438e-05,
+      "loss": 0.7566,
+      "step": 443370
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 2.4160480599984694e-05,
+      "loss": 0.7961,
+      "step": 443380
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 2.4159332670084944e-05,
+      "loss": 0.7901,
+      "step": 443390
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 2.4158184740185197e-05,
+      "loss": 0.8487,
+      "step": 443400
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 2.415703681028545e-05,
+      "loss": 0.7782,
+      "step": 443410
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 2.4155888880385704e-05,
+      "loss": 0.7968,
+      "step": 443420
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 2.4154740950485954e-05,
+      "loss": 0.7682,
+      "step": 443430
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 2.415359302058621e-05,
+      "loss": 0.8056,
+      "step": 443440
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 2.415244509068646e-05,
+      "loss": 0.8106,
+      "step": 443450
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 2.4151297160786713e-05,
+      "loss": 0.7741,
+      "step": 443460
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 2.4150149230886967e-05,
+      "loss": 0.8139,
+      "step": 443470
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 2.414900130098722e-05,
+      "loss": 0.776,
+      "step": 443480
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 2.414785337108747e-05,
+      "loss": 0.7543,
+      "step": 443490
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 2.4146705441187726e-05,
+      "loss": 0.8509,
+      "step": 443500
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 2.4145557511287976e-05,
+      "loss": 0.8023,
+      "step": 443510
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 2.4144409581388226e-05,
+      "loss": 0.7604,
+      "step": 443520
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 2.414326165148848e-05,
+      "loss": 0.8917,
+      "step": 443530
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 2.4142113721588733e-05,
+      "loss": 0.7767,
+      "step": 443540
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 2.4140965791688986e-05,
+      "loss": 0.8687,
+      "step": 443550
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 2.4139817861789236e-05,
+      "loss": 0.9219,
+      "step": 443560
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 2.4138669931889492e-05,
+      "loss": 0.7926,
+      "step": 443570
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 2.4137522001989742e-05,
+      "loss": 0.8413,
+      "step": 443580
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 2.4136374072089995e-05,
+      "loss": 0.8925,
+      "step": 443590
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 2.413522614219025e-05,
+      "loss": 0.703,
+      "step": 443600
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 2.4134078212290502e-05,
+      "loss": 0.8185,
+      "step": 443610
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 2.4132930282390752e-05,
+      "loss": 0.7872,
+      "step": 443620
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 2.413178235249101e-05,
+      "loss": 0.78,
+      "step": 443630
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 2.4130634422591258e-05,
+      "loss": 0.7946,
+      "step": 443640
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 2.4129486492691508e-05,
+      "loss": 0.8831,
+      "step": 443650
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 2.4128338562791765e-05,
+      "loss": 0.7272,
+      "step": 443660
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 2.4127190632892015e-05,
+      "loss": 0.7581,
+      "step": 443670
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 2.4126042702992268e-05,
+      "loss": 0.8877,
+      "step": 443680
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 2.412489477309252e-05,
+      "loss": 0.7469,
+      "step": 443690
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 2.4123746843192774e-05,
+      "loss": 0.7786,
+      "step": 443700
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 2.4122598913293024e-05,
+      "loss": 0.882,
+      "step": 443710
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 2.412145098339328e-05,
+      "loss": 0.8153,
+      "step": 443720
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 2.412030305349353e-05,
+      "loss": 0.8026,
+      "step": 443730
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 2.4119155123593784e-05,
+      "loss": 0.8004,
+      "step": 443740
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 2.4118007193694037e-05,
+      "loss": 0.7642,
+      "step": 443750
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 2.411685926379429e-05,
+      "loss": 0.8367,
+      "step": 443760
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 2.411571133389454e-05,
+      "loss": 0.7845,
+      "step": 443770
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 2.4114563403994797e-05,
+      "loss": 0.9165,
+      "step": 443780
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 2.4113415474095047e-05,
+      "loss": 0.7983,
+      "step": 443790
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 2.4112267544195297e-05,
+      "loss": 0.7909,
+      "step": 443800
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 2.4111119614295553e-05,
+      "loss": 0.7855,
+      "step": 443810
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 2.4109971684395803e-05,
+      "loss": 0.7652,
+      "step": 443820
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 2.4108823754496056e-05,
+      "loss": 0.7757,
+      "step": 443830
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 2.410767582459631e-05,
+      "loss": 0.8126,
+      "step": 443840
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 2.4106527894696563e-05,
+      "loss": 0.8723,
+      "step": 443850
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 2.4105379964796813e-05,
+      "loss": 0.7992,
+      "step": 443860
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 2.410423203489707e-05,
+      "loss": 0.765,
+      "step": 443870
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 2.410308410499732e-05,
+      "loss": 0.7894,
+      "step": 443880
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 2.4101936175097572e-05,
+      "loss": 0.8063,
+      "step": 443890
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 2.4100788245197826e-05,
+      "loss": 0.8586,
+      "step": 443900
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 2.409964031529808e-05,
+      "loss": 0.8382,
+      "step": 443910
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 2.409849238539833e-05,
+      "loss": 0.8522,
+      "step": 443920
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 2.4097344455498585e-05,
+      "loss": 0.8365,
+      "step": 443930
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 2.4096196525598835e-05,
+      "loss": 0.7705,
+      "step": 443940
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 2.4095048595699085e-05,
+      "loss": 0.7724,
+      "step": 443950
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 2.409390066579934e-05,
+      "loss": 0.9037,
+      "step": 443960
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 2.409275273589959e-05,
+      "loss": 0.8135,
+      "step": 443970
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 2.4091604805999845e-05,
+      "loss": 0.8365,
+      "step": 443980
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 2.4090456876100098e-05,
+      "loss": 0.8163,
+      "step": 443990
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 2.408930894620035e-05,
+      "loss": 0.753,
+      "step": 444000
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 2.40881610163006e-05,
+      "loss": 0.7894,
+      "step": 444010
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 2.4087013086400858e-05,
+      "loss": 0.8518,
+      "step": 444020
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 2.4085865156501108e-05,
+      "loss": 0.7926,
+      "step": 444030
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 2.408471722660136e-05,
+      "loss": 0.7591,
+      "step": 444040
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 2.4083569296701614e-05,
+      "loss": 0.7688,
+      "step": 444050
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 2.4082421366801867e-05,
+      "loss": 0.8283,
+      "step": 444060
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 2.4081273436902117e-05,
+      "loss": 0.8196,
+      "step": 444070
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 2.4080125507002374e-05,
+      "loss": 0.8561,
+      "step": 444080
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 2.4078977577102624e-05,
+      "loss": 0.8955,
+      "step": 444090
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 2.4077829647202873e-05,
+      "loss": 0.8243,
+      "step": 444100
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 2.407668171730313e-05,
+      "loss": 0.8578,
+      "step": 444110
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 2.407553378740338e-05,
+      "loss": 0.7313,
+      "step": 444120
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 2.4074385857503633e-05,
+      "loss": 0.6935,
+      "step": 444130
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 2.4073237927603886e-05,
+      "loss": 0.8328,
+      "step": 444140
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 2.407208999770414e-05,
+      "loss": 0.752,
+      "step": 444150
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 2.407094206780439e-05,
+      "loss": 0.7471,
+      "step": 444160
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 2.4069794137904646e-05,
+      "loss": 0.891,
+      "step": 444170
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 2.4068646208004896e-05,
+      "loss": 0.7234,
+      "step": 444180
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 2.406749827810515e-05,
+      "loss": 0.7057,
+      "step": 444190
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 2.40663503482054e-05,
+      "loss": 0.8103,
+      "step": 444200
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 2.4065202418305656e-05,
+      "loss": 0.7662,
+      "step": 444210
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 2.4064054488405906e-05,
+      "loss": 0.8833,
+      "step": 444220
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 2.4062906558506155e-05,
+      "loss": 0.8374,
+      "step": 444230
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 2.4061758628606412e-05,
+      "loss": 0.8648,
+      "step": 444240
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 2.4060610698706662e-05,
+      "loss": 0.7982,
+      "step": 444250
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 2.4059462768806915e-05,
+      "loss": 0.8289,
+      "step": 444260
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 2.405831483890717e-05,
+      "loss": 0.8941,
+      "step": 444270
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 2.405716690900742e-05,
+      "loss": 0.7685,
+      "step": 444280
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 2.405601897910767e-05,
+      "loss": 0.7832,
+      "step": 444290
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 2.4054871049207928e-05,
+      "loss": 0.7936,
+      "step": 444300
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 2.4053723119308178e-05,
+      "loss": 0.6769,
+      "step": 444310
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 2.4052575189408428e-05,
+      "loss": 0.8488,
+      "step": 444320
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 2.4051427259508684e-05,
+      "loss": 0.8131,
+      "step": 444330
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 2.4050279329608938e-05,
+      "loss": 0.7531,
+      "step": 444340
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 2.4049131399709188e-05,
+      "loss": 0.7938,
+      "step": 444350
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 2.4047983469809444e-05,
+      "loss": 0.8072,
+      "step": 444360
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 2.4046835539909694e-05,
+      "loss": 0.7742,
+      "step": 444370
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 2.4045687610009944e-05,
+      "loss": 0.8748,
+      "step": 444380
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 2.40445396801102e-05,
+      "loss": 0.7646,
+      "step": 444390
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 2.404339175021045e-05,
+      "loss": 0.8611,
+      "step": 444400
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 2.4042243820310704e-05,
+      "loss": 0.7793,
+      "step": 444410
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 2.4041095890410957e-05,
+      "loss": 0.7485,
+      "step": 444420
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 2.403994796051121e-05,
+      "loss": 0.8692,
+      "step": 444430
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 2.403880003061146e-05,
+      "loss": 0.9346,
+      "step": 444440
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 2.4037652100711717e-05,
+      "loss": 0.783,
+      "step": 444450
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 2.4036504170811966e-05,
+      "loss": 0.8719,
+      "step": 444460
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 2.4035356240912216e-05,
+      "loss": 0.804,
+      "step": 444470
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 2.4034208311012473e-05,
+      "loss": 0.8209,
+      "step": 444480
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 2.4033060381112726e-05,
+      "loss": 0.76,
+      "step": 444490
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 2.4031912451212976e-05,
+      "loss": 0.7471,
+      "step": 444500
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 2.4030764521313233e-05,
+      "loss": 0.7463,
+      "step": 444510
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 2.4029616591413483e-05,
+      "loss": 0.7775,
+      "step": 444520
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 2.4028468661513732e-05,
+      "loss": 0.7656,
+      "step": 444530
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 2.402732073161399e-05,
+      "loss": 0.7268,
+      "step": 444540
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 2.402617280171424e-05,
+      "loss": 0.8082,
+      "step": 444550
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 2.4025024871814492e-05,
+      "loss": 0.8196,
+      "step": 444560
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 2.4023876941914745e-05,
+      "loss": 0.8375,
+      "step": 444570
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 2.4022729012015e-05,
+      "loss": 0.8469,
+      "step": 444580
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 2.402158108211525e-05,
+      "loss": 0.8086,
+      "step": 444590
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 2.4020433152215505e-05,
+      "loss": 0.8949,
+      "step": 444600
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 2.4019285222315755e-05,
+      "loss": 0.9069,
+      "step": 444610
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 2.4018137292416005e-05,
+      "loss": 0.7732,
+      "step": 444620
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 2.401698936251626e-05,
+      "loss": 0.7565,
+      "step": 444630
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 2.4015841432616515e-05,
+      "loss": 0.7581,
+      "step": 444640
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 2.4014693502716765e-05,
+      "loss": 0.8092,
+      "step": 444650
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 2.401354557281702e-05,
+      "loss": 0.799,
+      "step": 444660
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 2.401239764291727e-05,
+      "loss": 0.7329,
+      "step": 444670
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 2.401124971301752e-05,
+      "loss": 0.7893,
+      "step": 444680
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 2.4010101783117777e-05,
+      "loss": 0.7551,
+      "step": 444690
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 2.4008953853218027e-05,
+      "loss": 0.8073,
+      "step": 444700
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 2.400780592331828e-05,
+      "loss": 0.7826,
+      "step": 444710
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 2.4006657993418534e-05,
+      "loss": 0.8456,
+      "step": 444720
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 2.4005510063518787e-05,
+      "loss": 0.8678,
+      "step": 444730
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 2.4004362133619037e-05,
+      "loss": 0.7547,
+      "step": 444740
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 2.4003214203719294e-05,
+      "loss": 0.6875,
+      "step": 444750
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 2.4002066273819543e-05,
+      "loss": 0.8072,
+      "step": 444760
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 2.4000918343919793e-05,
+      "loss": 0.8012,
+      "step": 444770
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 2.399977041402005e-05,
+      "loss": 0.7653,
+      "step": 444780
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 2.3998622484120303e-05,
+      "loss": 0.7407,
+      "step": 444790
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 2.3997474554220553e-05,
+      "loss": 0.7219,
+      "step": 444800
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 2.399632662432081e-05,
+      "loss": 0.7894,
+      "step": 444810
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 2.399517869442106e-05,
+      "loss": 0.76,
+      "step": 444820
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 2.399403076452131e-05,
+      "loss": 0.7996,
+      "step": 444830
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 2.3992882834621566e-05,
+      "loss": 0.7328,
+      "step": 444840
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 2.3991734904721816e-05,
+      "loss": 0.7416,
+      "step": 444850
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 2.399058697482207e-05,
+      "loss": 0.7663,
+      "step": 444860
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 2.3989439044922322e-05,
+      "loss": 0.8241,
+      "step": 444870
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 2.3988291115022576e-05,
+      "loss": 0.7599,
+      "step": 444880
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 2.3987143185122825e-05,
+      "loss": 0.8128,
+      "step": 444890
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 2.3985995255223075e-05,
+      "loss": 0.8131,
+      "step": 444900
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 2.3984847325323332e-05,
+      "loss": 0.7695,
+      "step": 444910
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 2.3983699395423582e-05,
+      "loss": 0.8219,
+      "step": 444920
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 2.3982551465523835e-05,
+      "loss": 0.7066,
+      "step": 444930
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 2.398140353562409e-05,
+      "loss": 0.8177,
+      "step": 444940
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 2.398025560572434e-05,
+      "loss": 0.8275,
+      "step": 444950
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 2.397910767582459e-05,
+      "loss": 0.8273,
+      "step": 444960
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 2.3977959745924848e-05,
+      "loss": 0.8411,
+      "step": 444970
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 2.3976811816025098e-05,
+      "loss": 0.8089,
+      "step": 444980
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 2.397566388612535e-05,
+      "loss": 0.769,
+      "step": 444990
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 2.3974515956225604e-05,
+      "loss": 0.771,
+      "step": 445000
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 2.3973368026325858e-05,
+      "loss": 0.8191,
+      "step": 445010
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 2.3972220096426107e-05,
+      "loss": 0.7664,
+      "step": 445020
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 2.3971072166526364e-05,
+      "loss": 0.8438,
+      "step": 445030
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 2.3969924236626614e-05,
+      "loss": 0.8289,
+      "step": 445040
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 2.3968776306726864e-05,
+      "loss": 0.8533,
+      "step": 445050
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 2.396762837682712e-05,
+      "loss": 0.7476,
+      "step": 445060
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 2.396648044692737e-05,
+      "loss": 0.7713,
+      "step": 445070
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 2.3965332517027623e-05,
+      "loss": 0.8594,
+      "step": 445080
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 2.396418458712788e-05,
+      "loss": 0.8884,
+      "step": 445090
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 2.396303665722813e-05,
+      "loss": 0.8457,
+      "step": 445100
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 2.396188872732838e-05,
+      "loss": 0.804,
+      "step": 445110
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 2.3960740797428636e-05,
+      "loss": 0.8391,
+      "step": 445120
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 2.3959592867528886e-05,
+      "loss": 0.8486,
+      "step": 445130
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 2.395844493762914e-05,
+      "loss": 0.8681,
+      "step": 445140
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 2.3957297007729393e-05,
+      "loss": 0.7676,
+      "step": 445150
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 2.3956149077829646e-05,
+      "loss": 0.884,
+      "step": 445160
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 2.3955001147929896e-05,
+      "loss": 0.801,
+      "step": 445170
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 2.3953853218030152e-05,
+      "loss": 0.7936,
+      "step": 445180
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 2.3952705288130402e-05,
+      "loss": 0.8768,
+      "step": 445190
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 2.3951557358230652e-05,
+      "loss": 0.8102,
+      "step": 445200
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 2.395040942833091e-05,
+      "loss": 0.8596,
+      "step": 445210
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 2.394926149843116e-05,
+      "loss": 0.859,
+      "step": 445220
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 2.3948113568531412e-05,
+      "loss": 0.8077,
+      "step": 445230
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 2.394696563863167e-05,
+      "loss": 0.7425,
+      "step": 445240
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 2.394581770873192e-05,
+      "loss": 0.8575,
+      "step": 445250
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 2.3944669778832168e-05,
+      "loss": 0.8731,
+      "step": 445260
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 2.3943521848932425e-05,
+      "loss": 0.8819,
+      "step": 445270
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 2.3942373919032675e-05,
+      "loss": 0.7529,
+      "step": 445280
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 2.3941225989132928e-05,
+      "loss": 0.8609,
+      "step": 445290
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 2.394007805923318e-05,
+      "loss": 0.8098,
+      "step": 445300
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 2.3938930129333434e-05,
+      "loss": 0.8019,
+      "step": 445310
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 2.3937782199433684e-05,
+      "loss": 0.7944,
+      "step": 445320
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 2.393663426953394e-05,
+      "loss": 0.873,
+      "step": 445330
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 2.393548633963419e-05,
+      "loss": 0.7982,
+      "step": 445340
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 2.393433840973444e-05,
+      "loss": 0.6937,
+      "step": 445350
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 2.3933190479834697e-05,
+      "loss": 0.7866,
+      "step": 445360
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 2.3932042549934947e-05,
+      "loss": 0.809,
+      "step": 445370
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 2.39308946200352e-05,
+      "loss": 0.7022,
+      "step": 445380
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 2.3929746690135457e-05,
+      "loss": 0.735,
+      "step": 445390
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 2.3928598760235707e-05,
+      "loss": 0.7577,
+      "step": 445400
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 2.3927450830335957e-05,
+      "loss": 0.8545,
+      "step": 445410
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 2.3926302900436213e-05,
+      "loss": 0.7577,
+      "step": 445420
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 2.3925154970536463e-05,
+      "loss": 0.8034,
+      "step": 445430
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 2.3924007040636716e-05,
+      "loss": 0.7473,
+      "step": 445440
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 2.392285911073697e-05,
+      "loss": 0.8373,
+      "step": 445450
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 2.3921711180837223e-05,
+      "loss": 0.8205,
+      "step": 445460
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 2.3920563250937473e-05,
+      "loss": 0.88,
+      "step": 445470
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 2.391941532103773e-05,
+      "loss": 0.8158,
+      "step": 445480
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 2.391826739113798e-05,
+      "loss": 0.8604,
+      "step": 445490
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 2.391711946123823e-05,
+      "loss": 0.8397,
+      "step": 445500
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 2.3915971531338486e-05,
+      "loss": 0.8597,
+      "step": 445510
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 2.391482360143874e-05,
+      "loss": 0.8066,
+      "step": 445520
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 2.391367567153899e-05,
+      "loss": 0.8276,
+      "step": 445530
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 2.3912527741639246e-05,
+      "loss": 0.8008,
+      "step": 445540
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 2.3911379811739495e-05,
+      "loss": 0.7564,
+      "step": 445550
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 2.3910231881839745e-05,
+      "loss": 0.8057,
+      "step": 445560
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 2.390908395194e-05,
+      "loss": 0.8151,
+      "step": 445570
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 2.3907936022040252e-05,
+      "loss": 0.733,
+      "step": 445580
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 2.3906788092140505e-05,
+      "loss": 0.7605,
+      "step": 445590
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 2.3905640162240755e-05,
+      "loss": 0.7981,
+      "step": 445600
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 2.390449223234101e-05,
+      "loss": 0.7773,
+      "step": 445610
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 2.390334430244126e-05,
+      "loss": 0.8224,
+      "step": 445620
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 2.390219637254151e-05,
+      "loss": 0.8525,
+      "step": 445630
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 2.3901048442641768e-05,
+      "loss": 0.8939,
+      "step": 445640
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 2.3899900512742018e-05,
+      "loss": 0.8061,
+      "step": 445650
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 2.389875258284227e-05,
+      "loss": 0.9068,
+      "step": 445660
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 2.3897604652942527e-05,
+      "loss": 0.7497,
+      "step": 445670
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 2.3896456723042777e-05,
+      "loss": 0.8407,
+      "step": 445680
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 2.3895308793143027e-05,
+      "loss": 0.7702,
+      "step": 445690
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 2.3894160863243284e-05,
+      "loss": 0.8351,
+      "step": 445700
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 2.3893012933343534e-05,
+      "loss": 0.7895,
+      "step": 445710
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 2.3891865003443787e-05,
+      "loss": 0.872,
+      "step": 445720
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 2.389071707354404e-05,
+      "loss": 0.6958,
+      "step": 445730
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 2.3889569143644293e-05,
+      "loss": 0.7977,
+      "step": 445740
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 2.3888421213744543e-05,
+      "loss": 0.8498,
+      "step": 445750
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 2.38872732838448e-05,
+      "loss": 0.788,
+      "step": 445760
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 2.388612535394505e-05,
+      "loss": 0.7829,
+      "step": 445770
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 2.38849774240453e-05,
+      "loss": 0.8177,
+      "step": 445780
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 2.3883829494145556e-05,
+      "loss": 0.7718,
+      "step": 445790
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 2.3882681564245806e-05,
+      "loss": 0.8174,
+      "step": 445800
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 2.388153363434606e-05,
+      "loss": 0.7996,
+      "step": 445810
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 2.3880385704446316e-05,
+      "loss": 0.8025,
+      "step": 445820
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 2.3879237774546566e-05,
+      "loss": 0.9079,
+      "step": 445830
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 2.3878089844646816e-05,
+      "loss": 0.8242,
+      "step": 445840
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 2.3876941914747072e-05,
+      "loss": 0.756,
+      "step": 445850
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 2.3875793984847322e-05,
+      "loss": 0.8432,
+      "step": 445860
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 2.3874646054947575e-05,
+      "loss": 0.8095,
+      "step": 445870
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 2.387349812504783e-05,
+      "loss": 0.8147,
+      "step": 445880
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 2.3872350195148082e-05,
+      "loss": 0.7534,
+      "step": 445890
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 2.3871202265248332e-05,
+      "loss": 0.8417,
+      "step": 445900
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 2.387005433534859e-05,
+      "loss": 0.7632,
+      "step": 445910
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 2.3868906405448838e-05,
+      "loss": 0.8104,
+      "step": 445920
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 2.3867758475549088e-05,
+      "loss": 0.7379,
+      "step": 445930
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 2.3866610545649345e-05,
+      "loss": 0.8914,
+      "step": 445940
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 2.3865462615749595e-05,
+      "loss": 0.8157,
+      "step": 445950
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 2.3864314685849848e-05,
+      "loss": 0.7769,
+      "step": 445960
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 2.3863166755950104e-05,
+      "loss": 0.8126,
+      "step": 445970
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 2.3862018826050354e-05,
+      "loss": 0.757,
+      "step": 445980
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 2.3860870896150604e-05,
+      "loss": 0.7464,
+      "step": 445990
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 2.385972296625086e-05,
+      "loss": 0.8297,
+      "step": 446000
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 2.385857503635111e-05,
+      "loss": 0.8058,
+      "step": 446010
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 2.3857427106451364e-05,
+      "loss": 0.7557,
+      "step": 446020
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 2.3856279176551617e-05,
+      "loss": 0.7726,
+      "step": 446030
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 2.385513124665187e-05,
+      "loss": 0.7623,
+      "step": 446040
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 2.385398331675212e-05,
+      "loss": 0.7927,
+      "step": 446050
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 2.3852835386852377e-05,
+      "loss": 0.7912,
+      "step": 446060
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 2.3851687456952627e-05,
+      "loss": 0.7957,
+      "step": 446070
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 2.3850539527052877e-05,
+      "loss": 0.8414,
+      "step": 446080
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 2.3849391597153133e-05,
+      "loss": 0.8053,
+      "step": 446090
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 2.3848243667253383e-05,
+      "loss": 0.8163,
+      "step": 446100
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 2.3847095737353636e-05,
+      "loss": 0.7567,
+      "step": 446110
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 2.3845947807453893e-05,
+      "loss": 0.8251,
+      "step": 446120
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 2.3844799877554143e-05,
+      "loss": 0.714,
+      "step": 446130
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 2.3843651947654393e-05,
+      "loss": 0.6973,
+      "step": 446140
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 2.384250401775465e-05,
+      "loss": 0.7436,
+      "step": 446150
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 2.38413560878549e-05,
+      "loss": 0.7913,
+      "step": 446160
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 2.3840208157955152e-05,
+      "loss": 0.7922,
+      "step": 446170
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 2.3839060228055406e-05,
+      "loss": 0.8139,
+      "step": 446180
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 2.383791229815566e-05,
+      "loss": 0.8662,
+      "step": 446190
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 2.383676436825591e-05,
+      "loss": 0.7783,
+      "step": 446200
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 2.3835616438356165e-05,
+      "loss": 0.7283,
+      "step": 446210
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 2.3834468508456415e-05,
+      "loss": 0.8246,
+      "step": 446220
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 2.3833320578556665e-05,
+      "loss": 0.7519,
+      "step": 446230
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 2.383217264865692e-05,
+      "loss": 0.7641,
+      "step": 446240
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 2.383102471875717e-05,
+      "loss": 0.7706,
+      "step": 446250
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 2.3829876788857425e-05,
+      "loss": 0.8756,
+      "step": 446260
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 2.3828728858957675e-05,
+      "loss": 0.8259,
+      "step": 446270
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 2.382758092905793e-05,
+      "loss": 0.7509,
+      "step": 446280
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 2.382643299915818e-05,
+      "loss": 0.7792,
+      "step": 446290
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 2.3825285069258434e-05,
+      "loss": 0.8269,
+      "step": 446300
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 2.3824137139358688e-05,
+      "loss": 0.7548,
+      "step": 446310
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 2.382298920945894e-05,
+      "loss": 0.7679,
+      "step": 446320
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 2.382184127955919e-05,
+      "loss": 0.8714,
+      "step": 446330
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 2.3820693349659447e-05,
+      "loss": 0.752,
+      "step": 446340
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 2.3819545419759697e-05,
+      "loss": 0.8248,
+      "step": 446350
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 2.3818397489859947e-05,
+      "loss": 0.7353,
+      "step": 446360
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 2.3817249559960204e-05,
+      "loss": 0.7417,
+      "step": 446370
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 2.3816101630060453e-05,
+      "loss": 0.805,
+      "step": 446380
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 2.3814953700160707e-05,
+      "loss": 0.8003,
+      "step": 446390
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 2.381380577026096e-05,
+      "loss": 0.906,
+      "step": 446400
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 2.3812657840361213e-05,
+      "loss": 0.7851,
+      "step": 446410
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 2.3811509910461463e-05,
+      "loss": 0.7464,
+      "step": 446420
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 2.381036198056172e-05,
+      "loss": 0.8714,
+      "step": 446430
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 2.380921405066197e-05,
+      "loss": 0.8737,
+      "step": 446440
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 2.3808066120762223e-05,
+      "loss": 0.7866,
+      "step": 446450
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 2.3806918190862476e-05,
+      "loss": 0.8072,
+      "step": 446460
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 2.380577026096273e-05,
+      "loss": 0.79,
+      "step": 446470
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 2.380462233106298e-05,
+      "loss": 0.7922,
+      "step": 446480
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 2.3803474401163236e-05,
+      "loss": 0.7511,
+      "step": 446490
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 2.3802326471263486e-05,
+      "loss": 0.7738,
+      "step": 446500
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 2.3801178541363735e-05,
+      "loss": 0.8539,
+      "step": 446510
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 2.3800030611463992e-05,
+      "loss": 0.7717,
+      "step": 446520
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 2.3798882681564242e-05,
+      "loss": 0.8931,
+      "step": 446530
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 2.3797734751664495e-05,
+      "loss": 0.7694,
+      "step": 446540
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 2.379658682176475e-05,
+      "loss": 0.7178,
+      "step": 446550
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 2.3795438891865e-05,
+      "loss": 0.7692,
+      "step": 446560
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 2.379429096196525e-05,
+      "loss": 0.7797,
+      "step": 446570
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 2.3793143032065508e-05,
+      "loss": 0.8389,
+      "step": 446580
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 2.3791995102165758e-05,
+      "loss": 0.8401,
+      "step": 446590
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 2.379084717226601e-05,
+      "loss": 0.8065,
+      "step": 446600
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 2.3789699242366265e-05,
+      "loss": 0.8745,
+      "step": 446610
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 2.3788551312466518e-05,
+      "loss": 0.7782,
+      "step": 446620
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 2.3787403382566768e-05,
+      "loss": 0.7736,
+      "step": 446630
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 2.3786255452667024e-05,
+      "loss": 0.7881,
+      "step": 446640
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 2.3785107522767274e-05,
+      "loss": 0.7519,
+      "step": 446650
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 2.3783959592867524e-05,
+      "loss": 0.7697,
+      "step": 446660
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 2.378281166296778e-05,
+      "loss": 0.8055,
+      "step": 446670
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 2.378166373306803e-05,
+      "loss": 0.8081,
+      "step": 446680
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 2.3780515803168284e-05,
+      "loss": 0.7535,
+      "step": 446690
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 2.3779367873268537e-05,
+      "loss": 0.7728,
+      "step": 446700
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 2.377821994336879e-05,
+      "loss": 0.8149,
+      "step": 446710
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 2.377707201346904e-05,
+      "loss": 0.8567,
+      "step": 446720
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 2.3775924083569297e-05,
+      "loss": 0.8732,
+      "step": 446730
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 2.3774776153669547e-05,
+      "loss": 0.9035,
+      "step": 446740
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 2.37736282237698e-05,
+      "loss": 0.8061,
+      "step": 446750
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 2.3772480293870053e-05,
+      "loss": 0.8075,
+      "step": 446760
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 2.3771332363970306e-05,
+      "loss": 0.8202,
+      "step": 446770
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 2.3770184434070556e-05,
+      "loss": 0.7314,
+      "step": 446780
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 2.3769036504170813e-05,
+      "loss": 0.8277,
+      "step": 446790
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 2.3767888574271063e-05,
+      "loss": 0.7802,
+      "step": 446800
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 2.3766740644371312e-05,
+      "loss": 0.8685,
+      "step": 446810
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 2.376559271447157e-05,
+      "loss": 0.7646,
+      "step": 446820
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 2.376444478457182e-05,
+      "loss": 0.7978,
+      "step": 446830
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 2.3763296854672072e-05,
+      "loss": 0.8604,
+      "step": 446840
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 2.3762148924772325e-05,
+      "loss": 0.8872,
+      "step": 446850
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 2.376100099487258e-05,
+      "loss": 0.8177,
+      "step": 446860
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 2.375985306497283e-05,
+      "loss": 0.8241,
+      "step": 446870
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 2.3758705135073085e-05,
+      "loss": 0.7056,
+      "step": 446880
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 2.3757557205173335e-05,
+      "loss": 0.7702,
+      "step": 446890
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 2.3756409275273588e-05,
+      "loss": 0.8629,
+      "step": 446900
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 2.375526134537384e-05,
+      "loss": 0.846,
+      "step": 446910
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 2.3754113415474095e-05,
+      "loss": 0.7556,
+      "step": 446920
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 2.3752965485574345e-05,
+      "loss": 0.7574,
+      "step": 446930
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 2.3751817555674594e-05,
+      "loss": 0.7913,
+      "step": 446940
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 2.375066962577485e-05,
+      "loss": 0.8029,
+      "step": 446950
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 2.37495216958751e-05,
+      "loss": 0.7842,
+      "step": 446960
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 2.3748373765975354e-05,
+      "loss": 0.8218,
+      "step": 446970
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 2.3747225836075607e-05,
+      "loss": 0.8523,
+      "step": 446980
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 2.374607790617586e-05,
+      "loss": 0.7215,
+      "step": 446990
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 2.374492997627611e-05,
+      "loss": 0.8601,
+      "step": 447000
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 2.3743782046376367e-05,
+      "loss": 0.7979,
+      "step": 447010
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 2.3742634116476617e-05,
+      "loss": 0.7243,
+      "step": 447020
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 2.374148618657687e-05,
+      "loss": 0.8692,
+      "step": 447030
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 2.3740338256677123e-05,
+      "loss": 0.8262,
+      "step": 447040
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 2.3739190326777377e-05,
+      "loss": 0.7759,
+      "step": 447050
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 2.3738042396877627e-05,
+      "loss": 0.7003,
+      "step": 447060
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 2.3736894466977883e-05,
+      "loss": 0.817,
+      "step": 447070
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 2.3735746537078133e-05,
+      "loss": 0.8103,
+      "step": 447080
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 2.3734598607178383e-05,
+      "loss": 0.7959,
+      "step": 447090
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 2.373345067727864e-05,
+      "loss": 0.7927,
+      "step": 447100
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 2.373230274737889e-05,
+      "loss": 0.8392,
+      "step": 447110
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 2.3731154817479143e-05,
+      "loss": 0.8022,
+      "step": 447120
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 2.3730006887579396e-05,
+      "loss": 0.7819,
+      "step": 447130
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 2.372885895767965e-05,
+      "loss": 0.8338,
+      "step": 447140
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 2.37277110277799e-05,
+      "loss": 0.8653,
+      "step": 447150
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 2.3726563097880156e-05,
+      "loss": 0.716,
+      "step": 447160
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 2.3725415167980405e-05,
+      "loss": 0.7561,
+      "step": 447170
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 2.372426723808066e-05,
+      "loss": 0.759,
+      "step": 447180
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 2.3723119308180912e-05,
+      "loss": 0.8867,
+      "step": 447190
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 2.3721971378281165e-05,
+      "loss": 0.7843,
+      "step": 447200
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 2.3720823448381415e-05,
+      "loss": 0.7651,
+      "step": 447210
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 2.371967551848167e-05,
+      "loss": 0.7206,
+      "step": 447220
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 2.371852758858192e-05,
+      "loss": 0.8423,
+      "step": 447230
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 2.371737965868217e-05,
+      "loss": 0.8366,
+      "step": 447240
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 2.3716231728782428e-05,
+      "loss": 0.8624,
+      "step": 447250
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 2.3715083798882678e-05,
+      "loss": 0.7103,
+      "step": 447260
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 2.371393586898293e-05,
+      "loss": 0.7456,
+      "step": 447270
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 2.3712787939083184e-05,
+      "loss": 0.7754,
+      "step": 447280
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 2.3711640009183438e-05,
+      "loss": 0.8593,
+      "step": 447290
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 2.3710492079283687e-05,
+      "loss": 0.8025,
+      "step": 447300
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 2.3709344149383944e-05,
+      "loss": 0.8633,
+      "step": 447310
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 2.3708196219484194e-05,
+      "loss": 0.8167,
+      "step": 447320
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 2.3707048289584447e-05,
+      "loss": 0.845,
+      "step": 447330
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 2.37059003596847e-05,
+      "loss": 0.7947,
+      "step": 447340
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 2.3704752429784954e-05,
+      "loss": 0.7461,
+      "step": 447350
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 2.3703604499885203e-05,
+      "loss": 0.7971,
+      "step": 447360
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 2.370245656998546e-05,
+      "loss": 0.7732,
+      "step": 447370
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 2.370130864008571e-05,
+      "loss": 0.8473,
+      "step": 447380
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 2.370016071018596e-05,
+      "loss": 0.7687,
+      "step": 447390
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 2.3699012780286216e-05,
+      "loss": 0.8091,
+      "step": 447400
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 2.3697864850386466e-05,
+      "loss": 0.8753,
+      "step": 447410
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 2.369671692048672e-05,
+      "loss": 0.7632,
+      "step": 447420
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 2.3695568990586973e-05,
+      "loss": 0.8708,
+      "step": 447430
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 2.3694421060687226e-05,
+      "loss": 0.7907,
+      "step": 447440
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 2.3693273130787476e-05,
+      "loss": 0.7949,
+      "step": 447450
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 2.3692125200887733e-05,
+      "loss": 0.8114,
+      "step": 447460
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 2.3690977270987982e-05,
+      "loss": 0.7969,
+      "step": 447470
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 2.3689829341088236e-05,
+      "loss": 0.8774,
+      "step": 447480
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 2.368868141118849e-05,
+      "loss": 0.9279,
+      "step": 447490
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 2.3687533481288742e-05,
+      "loss": 0.8116,
+      "step": 447500
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 2.3686385551388992e-05,
+      "loss": 0.7499,
+      "step": 447510
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 2.368523762148925e-05,
+      "loss": 0.8009,
+      "step": 447520
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 2.36840896915895e-05,
+      "loss": 0.8542,
+      "step": 447530
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 2.3682941761689748e-05,
+      "loss": 0.8694,
+      "step": 447540
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 2.3681793831790005e-05,
+      "loss": 0.8372,
+      "step": 447550
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 2.3680645901890255e-05,
+      "loss": 0.7968,
+      "step": 447560
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 2.3679497971990508e-05,
+      "loss": 0.7691,
+      "step": 447570
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 2.367835004209076e-05,
+      "loss": 0.8718,
+      "step": 447580
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 2.3677202112191015e-05,
+      "loss": 0.8645,
+      "step": 447590
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 2.3676054182291264e-05,
+      "loss": 0.8309,
+      "step": 447600
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 2.367490625239152e-05,
+      "loss": 0.8407,
+      "step": 447610
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 2.367375832249177e-05,
+      "loss": 0.8852,
+      "step": 447620
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 2.3672610392592024e-05,
+      "loss": 0.7268,
+      "step": 447630
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 2.3671462462692274e-05,
+      "loss": 0.8839,
+      "step": 447640
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 2.367031453279253e-05,
+      "loss": 0.7451,
+      "step": 447650
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 2.366916660289278e-05,
+      "loss": 0.8695,
+      "step": 447660
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 2.366801867299303e-05,
+      "loss": 0.7365,
+      "step": 447670
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 2.3666870743093287e-05,
+      "loss": 0.8368,
+      "step": 447680
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 2.3665722813193537e-05,
+      "loss": 0.9212,
+      "step": 447690
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 2.366457488329379e-05,
+      "loss": 0.8021,
+      "step": 447700
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 2.3663426953394043e-05,
+      "loss": 0.7962,
+      "step": 447710
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 2.3662279023494296e-05,
+      "loss": 0.8205,
+      "step": 447720
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 2.3661131093594546e-05,
+      "loss": 0.766,
+      "step": 447730
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 2.3659983163694803e-05,
+      "loss": 0.9035,
+      "step": 447740
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 2.3658835233795053e-05,
+      "loss": 0.842,
+      "step": 447750
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 2.3657687303895306e-05,
+      "loss": 0.8141,
+      "step": 447760
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 2.365653937399556e-05,
+      "loss": 0.8196,
+      "step": 447770
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 2.3655391444095813e-05,
+      "loss": 0.8512,
+      "step": 447780
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 2.3654243514196062e-05,
+      "loss": 0.8565,
+      "step": 447790
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 2.365309558429632e-05,
+      "loss": 0.8473,
+      "step": 447800
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 2.365194765439657e-05,
+      "loss": 0.8078,
+      "step": 447810
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 2.365079972449682e-05,
+      "loss": 0.7606,
+      "step": 447820
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 2.3649651794597075e-05,
+      "loss": 0.7797,
+      "step": 447830
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 2.3648503864697325e-05,
+      "loss": 0.8152,
+      "step": 447840
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 2.364735593479758e-05,
+      "loss": 0.8196,
+      "step": 447850
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 2.3646208004897832e-05,
+      "loss": 0.8092,
+      "step": 447860
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 2.3645060074998085e-05,
+      "loss": 0.7201,
+      "step": 447870
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 2.3643912145098335e-05,
+      "loss": 0.7862,
+      "step": 447880
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 2.364276421519859e-05,
+      "loss": 0.8023,
+      "step": 447890
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 2.364161628529884e-05,
+      "loss": 0.7416,
+      "step": 447900
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 2.3640468355399095e-05,
+      "loss": 0.8475,
+      "step": 447910
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 2.3639320425499348e-05,
+      "loss": 0.7356,
+      "step": 447920
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 2.36381724955996e-05,
+      "loss": 0.7397,
+      "step": 447930
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 2.363702456569985e-05,
+      "loss": 0.8217,
+      "step": 447940
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 2.3635876635800108e-05,
+      "loss": 0.7608,
+      "step": 447950
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 2.3634728705900357e-05,
+      "loss": 0.8126,
+      "step": 447960
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 2.3633580776000607e-05,
+      "loss": 0.7964,
+      "step": 447970
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 2.3632432846100864e-05,
+      "loss": 0.8362,
+      "step": 447980
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 2.3631284916201114e-05,
+      "loss": 0.8107,
+      "step": 447990
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 2.3630136986301367e-05,
+      "loss": 0.8609,
+      "step": 448000
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 2.362898905640162e-05,
+      "loss": 0.7743,
+      "step": 448010
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 2.3627841126501873e-05,
+      "loss": 0.7995,
+      "step": 448020
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 2.3626693196602123e-05,
+      "loss": 0.7266,
+      "step": 448030
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 2.362554526670238e-05,
+      "loss": 0.7061,
+      "step": 448040
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 2.362439733680263e-05,
+      "loss": 0.7644,
+      "step": 448050
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 2.3623249406902883e-05,
+      "loss": 0.7703,
+      "step": 448060
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 2.3622101477003136e-05,
+      "loss": 0.8653,
+      "step": 448070
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 2.362095354710339e-05,
+      "loss": 0.7385,
+      "step": 448080
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 2.361980561720364e-05,
+      "loss": 0.8072,
+      "step": 448090
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 2.3618657687303896e-05,
+      "loss": 0.8328,
+      "step": 448100
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 2.3617509757404146e-05,
+      "loss": 0.8135,
+      "step": 448110
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 2.3616361827504396e-05,
+      "loss": 0.7669,
+      "step": 448120
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 2.3615213897604652e-05,
+      "loss": 0.8472,
+      "step": 448130
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 2.3614065967704902e-05,
+      "loss": 0.7619,
+      "step": 448140
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 2.3612918037805155e-05,
+      "loss": 0.7689,
+      "step": 448150
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 2.361177010790541e-05,
+      "loss": 0.7272,
+      "step": 448160
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 2.3610622178005662e-05,
+      "loss": 0.8097,
+      "step": 448170
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 2.3609474248105912e-05,
+      "loss": 0.8371,
+      "step": 448180
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 2.360832631820617e-05,
+      "loss": 0.8619,
+      "step": 448190
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 2.3607178388306418e-05,
+      "loss": 0.84,
+      "step": 448200
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 2.360603045840667e-05,
+      "loss": 0.8126,
+      "step": 448210
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 2.3604882528506925e-05,
+      "loss": 0.8861,
+      "step": 448220
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 2.3603734598607178e-05,
+      "loss": 0.7419,
+      "step": 448230
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 2.3602586668707428e-05,
+      "loss": 0.7636,
+      "step": 448240
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 2.3601438738807684e-05,
+      "loss": 0.7478,
+      "step": 448250
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 2.3600290808907934e-05,
+      "loss": 0.7856,
+      "step": 448260
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 2.3599142879008184e-05,
+      "loss": 0.7653,
+      "step": 448270
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 2.359799494910844e-05,
+      "loss": 0.8722,
+      "step": 448280
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 2.359684701920869e-05,
+      "loss": 0.7907,
+      "step": 448290
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 2.3595699089308944e-05,
+      "loss": 0.8042,
+      "step": 448300
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 2.3594551159409194e-05,
+      "loss": 0.7902,
+      "step": 448310
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 2.359340322950945e-05,
+      "loss": 0.8207,
+      "step": 448320
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 2.35922552996097e-05,
+      "loss": 0.7999,
+      "step": 448330
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 2.3591107369709953e-05,
+      "loss": 0.7824,
+      "step": 448340
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 2.3589959439810207e-05,
+      "loss": 0.7789,
+      "step": 448350
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 2.358881150991046e-05,
+      "loss": 0.8071,
+      "step": 448360
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 2.358766358001071e-05,
+      "loss": 0.7125,
+      "step": 448370
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 2.3586515650110966e-05,
+      "loss": 0.7644,
+      "step": 448380
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 2.3585367720211216e-05,
+      "loss": 0.8172,
+      "step": 448390
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 2.3584219790311466e-05,
+      "loss": 0.8229,
+      "step": 448400
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 2.3583071860411723e-05,
+      "loss": 0.7631,
+      "step": 448410
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 2.3581923930511973e-05,
+      "loss": 0.7186,
+      "step": 448420
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 2.3580776000612226e-05,
+      "loss": 0.8251,
+      "step": 448430
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 2.357962807071248e-05,
+      "loss": 0.8418,
+      "step": 448440
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 2.3578480140812732e-05,
+      "loss": 0.8934,
+      "step": 448450
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 2.3577332210912982e-05,
+      "loss": 0.8338,
+      "step": 448460
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 2.357618428101324e-05,
+      "loss": 0.7698,
+      "step": 448470
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 2.357503635111349e-05,
+      "loss": 0.8226,
+      "step": 448480
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 2.3573888421213742e-05,
+      "loss": 0.8216,
+      "step": 448490
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 2.3572740491313995e-05,
+      "loss": 0.7827,
+      "step": 448500
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 2.357159256141425e-05,
+      "loss": 0.7358,
+      "step": 448510
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 2.3570444631514498e-05,
+      "loss": 0.7641,
+      "step": 448520
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 2.3569296701614755e-05,
+      "loss": 0.8679,
+      "step": 448530
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 2.3568148771715005e-05,
+      "loss": 0.7971,
+      "step": 448540
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 2.3567000841815255e-05,
+      "loss": 0.7977,
+      "step": 448550
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 2.356585291191551e-05,
+      "loss": 0.8036,
+      "step": 448560
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 2.356470498201576e-05,
+      "loss": 0.7574,
+      "step": 448570
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 2.3563557052116014e-05,
+      "loss": 0.764,
+      "step": 448580
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 2.3562409122216268e-05,
+      "loss": 0.7335,
+      "step": 448590
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 2.356126119231652e-05,
+      "loss": 0.7995,
+      "step": 448600
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 2.356011326241677e-05,
+      "loss": 0.7523,
+      "step": 448610
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 2.3558965332517027e-05,
+      "loss": 0.8838,
+      "step": 448620
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 2.3557817402617277e-05,
+      "loss": 0.8646,
+      "step": 448630
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 2.355666947271753e-05,
+      "loss": 0.7072,
+      "step": 448640
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 2.3555521542817784e-05,
+      "loss": 0.8252,
+      "step": 448650
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 2.3554373612918037e-05,
+      "loss": 0.8689,
+      "step": 448660
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 2.3553225683018287e-05,
+      "loss": 0.8356,
+      "step": 448670
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 2.3552077753118543e-05,
+      "loss": 0.9173,
+      "step": 448680
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 2.3550929823218793e-05,
+      "loss": 0.819,
+      "step": 448690
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 2.3549781893319043e-05,
+      "loss": 0.7382,
+      "step": 448700
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 2.35486339634193e-05,
+      "loss": 0.7592,
+      "step": 448710
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 2.354748603351955e-05,
+      "loss": 0.799,
+      "step": 448720
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 2.3546338103619803e-05,
+      "loss": 0.8124,
+      "step": 448730
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 2.3545190173720056e-05,
+      "loss": 0.8611,
+      "step": 448740
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 2.354404224382031e-05,
+      "loss": 0.8459,
+      "step": 448750
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 2.354289431392056e-05,
+      "loss": 0.8216,
+      "step": 448760
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 2.3541746384020816e-05,
+      "loss": 0.7571,
+      "step": 448770
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 2.3540598454121066e-05,
+      "loss": 0.7963,
+      "step": 448780
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 2.353945052422132e-05,
+      "loss": 0.7455,
+      "step": 448790
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 2.3538302594321572e-05,
+      "loss": 0.7865,
+      "step": 448800
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 2.3537154664421825e-05,
+      "loss": 0.7888,
+      "step": 448810
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 2.3536006734522075e-05,
+      "loss": 0.727,
+      "step": 448820
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 2.3534858804622332e-05,
+      "loss": 0.8292,
+      "step": 448830
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 2.3533710874722582e-05,
+      "loss": 0.7788,
+      "step": 448840
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 2.353256294482283e-05,
+      "loss": 0.7806,
+      "step": 448850
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 2.3531415014923088e-05,
+      "loss": 0.9265,
+      "step": 448860
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 2.3530267085023338e-05,
+      "loss": 0.8,
+      "step": 448870
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 2.352911915512359e-05,
+      "loss": 0.831,
+      "step": 448880
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 2.3527971225223845e-05,
+      "loss": 0.8077,
+      "step": 448890
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 2.3526823295324098e-05,
+      "loss": 0.8275,
+      "step": 448900
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 2.3525675365424348e-05,
+      "loss": 0.7998,
+      "step": 448910
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 2.3524527435524604e-05,
+      "loss": 0.8765,
+      "step": 448920
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 2.3523379505624854e-05,
+      "loss": 0.7729,
+      "step": 448930
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 2.3522231575725107e-05,
+      "loss": 0.7549,
+      "step": 448940
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 2.352108364582536e-05,
+      "loss": 0.7157,
+      "step": 448950
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 2.3519935715925614e-05,
+      "loss": 0.8606,
+      "step": 448960
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 2.3518787786025864e-05,
+      "loss": 0.8445,
+      "step": 448970
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 2.351763985612612e-05,
+      "loss": 0.775,
+      "step": 448980
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 2.351649192622637e-05,
+      "loss": 0.8546,
+      "step": 448990
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 2.351534399632662e-05,
+      "loss": 0.7717,
+      "step": 449000
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 2.3514196066426873e-05,
+      "loss": 0.9112,
+      "step": 449010
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 2.3513048136527127e-05,
+      "loss": 0.7146,
+      "step": 449020
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 2.351190020662738e-05,
+      "loss": 0.8497,
+      "step": 449030
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 2.351075227672763e-05,
+      "loss": 0.7856,
+      "step": 449040
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 2.3509604346827886e-05,
+      "loss": 0.7962,
+      "step": 449050
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 2.3508456416928136e-05,
+      "loss": 0.8371,
+      "step": 449060
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 2.350730848702839e-05,
+      "loss": 0.7755,
+      "step": 449070
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 2.3506160557128643e-05,
+      "loss": 0.7764,
+      "step": 449080
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 2.3505012627228896e-05,
+      "loss": 0.8185,
+      "step": 449090
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 2.3503864697329146e-05,
+      "loss": 0.7571,
+      "step": 449100
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 2.3502716767429402e-05,
+      "loss": 0.835,
+      "step": 449110
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 2.3501568837529652e-05,
+      "loss": 0.7811,
+      "step": 449120
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 2.3500420907629902e-05,
+      "loss": 0.9499,
+      "step": 449130
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 2.349927297773016e-05,
+      "loss": 0.7593,
+      "step": 449140
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 2.349812504783041e-05,
+      "loss": 0.754,
+      "step": 449150
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 2.3496977117930662e-05,
+      "loss": 0.833,
+      "step": 449160
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 2.3495829188030915e-05,
+      "loss": 0.8857,
+      "step": 449170
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 2.3494681258131168e-05,
+      "loss": 0.8418,
+      "step": 449180
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 2.3493533328231418e-05,
+      "loss": 0.853,
+      "step": 449190
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 2.3492385398331675e-05,
+      "loss": 0.7047,
+      "step": 449200
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 2.3491237468431925e-05,
+      "loss": 0.749,
+      "step": 449210
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 2.3490089538532178e-05,
+      "loss": 0.9117,
+      "step": 449220
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 2.348894160863243e-05,
+      "loss": 0.7638,
+      "step": 449230
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 2.3487793678732684e-05,
+      "loss": 0.7792,
+      "step": 449240
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 2.3486645748832934e-05,
+      "loss": 0.8581,
+      "step": 449250
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 2.348549781893319e-05,
+      "loss": 0.7832,
+      "step": 449260
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 2.348434988903344e-05,
+      "loss": 0.7582,
+      "step": 449270
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 2.348320195913369e-05,
+      "loss": 0.8029,
+      "step": 449280
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 2.3482054029233947e-05,
+      "loss": 0.7793,
+      "step": 449290
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 2.3480906099334197e-05,
+      "loss": 0.9097,
+      "step": 449300
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 2.347975816943445e-05,
+      "loss": 0.7491,
+      "step": 449310
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 2.3478610239534703e-05,
+      "loss": 0.7971,
+      "step": 449320
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 2.3477462309634957e-05,
+      "loss": 0.78,
+      "step": 449330
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 2.3476314379735207e-05,
+      "loss": 0.8191,
+      "step": 449340
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 2.3475166449835463e-05,
+      "loss": 0.7844,
+      "step": 449350
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 2.3474018519935713e-05,
+      "loss": 0.8257,
+      "step": 449360
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 2.3472870590035966e-05,
+      "loss": 0.8744,
+      "step": 449370
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 2.347172266013622e-05,
+      "loss": 0.7421,
+      "step": 449380
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 2.3470574730236473e-05,
+      "loss": 0.7927,
+      "step": 449390
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 2.3469426800336723e-05,
+      "loss": 0.8244,
+      "step": 449400
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 2.346827887043698e-05,
+      "loss": 0.7725,
+      "step": 449410
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 2.346713094053723e-05,
+      "loss": 0.7781,
+      "step": 449420
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 2.346598301063748e-05,
+      "loss": 0.8451,
+      "step": 449430
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 2.3464835080737736e-05,
+      "loss": 0.7962,
+      "step": 449440
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 2.3463687150837985e-05,
+      "loss": 0.7856,
+      "step": 449450
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 2.346253922093824e-05,
+      "loss": 0.8498,
+      "step": 449460
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 2.3461391291038492e-05,
+      "loss": 0.883,
+      "step": 449470
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 2.3460243361138745e-05,
+      "loss": 0.8081,
+      "step": 449480
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 2.3459095431238995e-05,
+      "loss": 0.7889,
+      "step": 449490
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 2.345794750133925e-05,
+      "loss": 0.901,
+      "step": 449500
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 2.34567995714395e-05,
+      "loss": 0.7991,
+      "step": 449510
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 2.3455651641539755e-05,
+      "loss": 0.829,
+      "step": 449520
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 2.3454503711640008e-05,
+      "loss": 0.7424,
+      "step": 449530
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 2.345335578174026e-05,
+      "loss": 0.8339,
+      "step": 449540
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 2.345220785184051e-05,
+      "loss": 0.7791,
+      "step": 449550
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 2.3451059921940768e-05,
+      "loss": 0.8462,
+      "step": 449560
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 2.3449911992041018e-05,
+      "loss": 0.8281,
+      "step": 449570
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 2.3448764062141267e-05,
+      "loss": 0.7438,
+      "step": 449580
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 2.3447616132241524e-05,
+      "loss": 0.8235,
+      "step": 449590
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 2.3446468202341774e-05,
+      "loss": 0.8432,
+      "step": 449600
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 2.3445320272442027e-05,
+      "loss": 0.8051,
+      "step": 449610
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 2.344417234254228e-05,
+      "loss": 0.7313,
+      "step": 449620
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 2.3443024412642534e-05,
+      "loss": 0.778,
+      "step": 449630
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 2.3441876482742784e-05,
+      "loss": 0.7516,
+      "step": 449640
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 2.344072855284304e-05,
+      "loss": 0.7579,
+      "step": 449650
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 2.343958062294329e-05,
+      "loss": 0.8119,
+      "step": 449660
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 2.3438432693043543e-05,
+      "loss": 0.7708,
+      "step": 449670
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 2.3437284763143793e-05,
+      "loss": 0.832,
+      "step": 449680
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 2.343613683324405e-05,
+      "loss": 0.8451,
+      "step": 449690
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 2.34349889033443e-05,
+      "loss": 0.8032,
+      "step": 449700
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 2.343384097344455e-05,
+      "loss": 0.8876,
+      "step": 449710
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 2.3432693043544806e-05,
+      "loss": 0.7825,
+      "step": 449720
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 2.3431545113645056e-05,
+      "loss": 0.8222,
+      "step": 449730
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 2.343039718374531e-05,
+      "loss": 0.7456,
+      "step": 449740
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 2.3429249253845562e-05,
+      "loss": 0.8798,
+      "step": 449750
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 2.3428101323945816e-05,
+      "loss": 0.8064,
+      "step": 449760
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 2.3426953394046066e-05,
+      "loss": 0.7732,
+      "step": 449770
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 2.3425805464146322e-05,
+      "loss": 0.8785,
+      "step": 449780
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 2.3424657534246572e-05,
+      "loss": 0.812,
+      "step": 449790
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 2.3423509604346825e-05,
+      "loss": 0.9026,
+      "step": 449800
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 2.342236167444708e-05,
+      "loss": 0.8046,
+      "step": 449810
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 2.3421213744547332e-05,
+      "loss": 0.776,
+      "step": 449820
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 2.342006581464758e-05,
+      "loss": 0.7134,
+      "step": 449830
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 2.3418917884747838e-05,
+      "loss": 0.7824,
+      "step": 449840
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 2.3417769954848088e-05,
+      "loss": 0.798,
+      "step": 449850
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 2.3416622024948338e-05,
+      "loss": 0.8349,
+      "step": 449860
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 2.3415474095048595e-05,
+      "loss": 0.7593,
+      "step": 449870
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 2.3414326165148844e-05,
+      "loss": 0.8028,
+      "step": 449880
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 2.3413178235249098e-05,
+      "loss": 0.7583,
+      "step": 449890
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 2.341203030534935e-05,
+      "loss": 0.8023,
+      "step": 449900
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 2.3410882375449604e-05,
+      "loss": 0.8499,
+      "step": 449910
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 2.3409734445549854e-05,
+      "loss": 0.8236,
+      "step": 449920
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 2.340858651565011e-05,
+      "loss": 0.7978,
+      "step": 449930
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 2.340743858575036e-05,
+      "loss": 0.7523,
+      "step": 449940
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 2.3406290655850614e-05,
+      "loss": 0.8699,
+      "step": 449950
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 2.3405142725950867e-05,
+      "loss": 0.8436,
+      "step": 449960
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 2.340399479605112e-05,
+      "loss": 0.8293,
+      "step": 449970
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 2.340284686615137e-05,
+      "loss": 0.8596,
+      "step": 449980
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 2.3401698936251627e-05,
+      "loss": 0.8384,
+      "step": 449990
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 2.3400551006351877e-05,
+      "loss": 0.7525,
+      "step": 450000
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 2.3399403076452126e-05,
+      "loss": 0.7004,
+      "step": 450010
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 2.3398255146552383e-05,
+      "loss": 0.7409,
+      "step": 450020
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 2.3397107216652633e-05,
+      "loss": 0.7977,
+      "step": 450030
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 2.3395959286752886e-05,
+      "loss": 0.7642,
+      "step": 450040
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 2.339481135685314e-05,
+      "loss": 0.8205,
+      "step": 450050
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 2.3393663426953393e-05,
+      "loss": 0.734,
+      "step": 450060
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 2.3392515497053642e-05,
+      "loss": 0.8474,
+      "step": 450070
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 2.33913675671539e-05,
+      "loss": 0.8722,
+      "step": 450080
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 2.339021963725415e-05,
+      "loss": 0.8712,
+      "step": 450090
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 2.3389071707354402e-05,
+      "loss": 0.8419,
+      "step": 450100
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 2.3387923777454655e-05,
+      "loss": 0.8962,
+      "step": 450110
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 2.338677584755491e-05,
+      "loss": 0.8431,
+      "step": 450120
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 2.338562791765516e-05,
+      "loss": 0.8384,
+      "step": 450130
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 2.3384479987755415e-05,
+      "loss": 0.8296,
+      "step": 450140
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 2.3383332057855665e-05,
+      "loss": 0.8365,
+      "step": 450150
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 2.3382184127955915e-05,
+      "loss": 0.8466,
+      "step": 450160
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 2.338103619805617e-05,
+      "loss": 0.8514,
+      "step": 450170
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 2.337988826815642e-05,
+      "loss": 0.8669,
+      "step": 450180
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 2.3378740338256675e-05,
+      "loss": 0.8118,
+      "step": 450190
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 2.3377592408356928e-05,
+      "loss": 0.8229,
+      "step": 450200
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 2.337644447845718e-05,
+      "loss": 0.8125,
+      "step": 450210
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 2.337529654855743e-05,
+      "loss": 0.8283,
+      "step": 450220
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 2.3374148618657688e-05,
+      "loss": 0.8741,
+      "step": 450230
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 2.3373000688757937e-05,
+      "loss": 0.7668,
+      "step": 450240
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 2.337185275885819e-05,
+      "loss": 0.7999,
+      "step": 450250
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 2.3370704828958444e-05,
+      "loss": 0.8677,
+      "step": 450260
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 2.3369556899058697e-05,
+      "loss": 0.8603,
+      "step": 450270
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 2.3368408969158947e-05,
+      "loss": 0.8978,
+      "step": 450280
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 2.3367261039259204e-05,
+      "loss": 0.8109,
+      "step": 450290
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 2.3366113109359453e-05,
+      "loss": 0.8103,
+      "step": 450300
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 2.3364965179459703e-05,
+      "loss": 0.8463,
+      "step": 450310
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 2.336381724955996e-05,
+      "loss": 0.8855,
+      "step": 450320
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 2.336266931966021e-05,
+      "loss": 0.862,
+      "step": 450330
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 2.3361521389760463e-05,
+      "loss": 0.7171,
+      "step": 450340
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 2.3360373459860716e-05,
+      "loss": 0.8196,
+      "step": 450350
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 2.335922552996097e-05,
+      "loss": 0.7618,
+      "step": 450360
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 2.335807760006122e-05,
+      "loss": 0.8096,
+      "step": 450370
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 2.3356929670161473e-05,
+      "loss": 0.7772,
+      "step": 450380
+    },
+    {
+      "epoch": 20.66,
+      "learning_rate": 2.3355781740261726e-05,
+      "loss": 0.8826,
+      "step": 450390
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 2.335463381036198e-05,
+      "loss": 0.8056,
+      "step": 450400
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 2.335348588046223e-05,
+      "loss": 0.8473,
+      "step": 450410
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 2.3352337950562486e-05,
+      "loss": 0.7955,
+      "step": 450420
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 2.3351190020662735e-05,
+      "loss": 0.8254,
+      "step": 450430
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 2.3350042090762985e-05,
+      "loss": 0.7836,
+      "step": 450440
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 2.3348894160863242e-05,
+      "loss": 0.7481,
+      "step": 450450
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 2.3347746230963492e-05,
+      "loss": 0.8763,
+      "step": 450460
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 2.3346598301063745e-05,
+      "loss": 0.7661,
+      "step": 450470
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 2.3345450371163998e-05,
+      "loss": 0.8323,
+      "step": 450480
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 2.334430244126425e-05,
+      "loss": 0.7658,
+      "step": 450490
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 2.33431545113645e-05,
+      "loss": 0.8379,
+      "step": 450500
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 2.3342006581464758e-05,
+      "loss": 0.7657,
+      "step": 450510
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 2.3340858651565008e-05,
+      "loss": 0.7815,
+      "step": 450520
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 2.333971072166526e-05,
+      "loss": 0.7153,
+      "step": 450530
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 2.3338562791765514e-05,
+      "loss": 0.7996,
+      "step": 450540
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 2.3337414861865768e-05,
+      "loss": 0.8403,
+      "step": 450550
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 2.3336266931966017e-05,
+      "loss": 0.7642,
+      "step": 450560
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 2.3335119002066274e-05,
+      "loss": 0.8733,
+      "step": 450570
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 2.3333971072166524e-05,
+      "loss": 0.7493,
+      "step": 450580
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 2.3332823142266774e-05,
+      "loss": 0.8847,
+      "step": 450590
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 2.333167521236703e-05,
+      "loss": 0.8351,
+      "step": 450600
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 2.333052728246728e-05,
+      "loss": 0.8664,
+      "step": 450610
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 2.3329379352567534e-05,
+      "loss": 0.8069,
+      "step": 450620
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 2.3328231422667787e-05,
+      "loss": 0.7579,
+      "step": 450630
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 2.332708349276804e-05,
+      "loss": 0.8122,
+      "step": 450640
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 2.332593556286829e-05,
+      "loss": 0.8013,
+      "step": 450650
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 2.3324787632968546e-05,
+      "loss": 0.8339,
+      "step": 450660
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 2.3323639703068796e-05,
+      "loss": 0.7823,
+      "step": 450670
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 2.332249177316905e-05,
+      "loss": 0.8588,
+      "step": 450680
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 2.3321343843269303e-05,
+      "loss": 0.7602,
+      "step": 450690
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 2.3320195913369556e-05,
+      "loss": 0.7672,
+      "step": 450700
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 2.3319047983469806e-05,
+      "loss": 0.7917,
+      "step": 450710
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 2.3317900053570063e-05,
+      "loss": 0.8691,
+      "step": 450720
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 2.3316752123670312e-05,
+      "loss": 0.8478,
+      "step": 450730
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 2.3315604193770562e-05,
+      "loss": 0.8526,
+      "step": 450740
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 2.331445626387082e-05,
+      "loss": 0.7779,
+      "step": 450750
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 2.331330833397107e-05,
+      "loss": 0.8797,
+      "step": 450760
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 2.3312160404071322e-05,
+      "loss": 0.8693,
+      "step": 450770
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 2.3311012474171575e-05,
+      "loss": 0.7426,
+      "step": 450780
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 2.330986454427183e-05,
+      "loss": 0.8376,
+      "step": 450790
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 2.330871661437208e-05,
+      "loss": 0.7162,
+      "step": 450800
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 2.3307568684472335e-05,
+      "loss": 0.791,
+      "step": 450810
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 2.3306420754572585e-05,
+      "loss": 0.8331,
+      "step": 450820
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 2.3305272824672838e-05,
+      "loss": 0.8387,
+      "step": 450830
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 2.330412489477309e-05,
+      "loss": 0.9128,
+      "step": 450840
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 2.3302976964873345e-05,
+      "loss": 0.8886,
+      "step": 450850
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 2.3301829034973594e-05,
+      "loss": 0.8513,
+      "step": 450860
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 2.330068110507385e-05,
+      "loss": 0.7861,
+      "step": 450870
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 2.32995331751741e-05,
+      "loss": 0.7774,
+      "step": 450880
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 2.329838524527435e-05,
+      "loss": 0.8224,
+      "step": 450890
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 2.3297237315374607e-05,
+      "loss": 0.769,
+      "step": 450900
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 2.3296089385474857e-05,
+      "loss": 0.7551,
+      "step": 450910
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 2.329494145557511e-05,
+      "loss": 0.8206,
+      "step": 450920
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 2.3293793525675364e-05,
+      "loss": 0.8257,
+      "step": 450930
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 2.3292645595775617e-05,
+      "loss": 0.925,
+      "step": 450940
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 2.3291497665875867e-05,
+      "loss": 0.8068,
+      "step": 450950
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 2.3290349735976123e-05,
+      "loss": 0.8426,
+      "step": 450960
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 2.3289201806076373e-05,
+      "loss": 0.789,
+      "step": 450970
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 2.3288053876176627e-05,
+      "loss": 0.8569,
+      "step": 450980
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 2.328690594627688e-05,
+      "loss": 0.8233,
+      "step": 450990
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 2.3285758016377133e-05,
+      "loss": 0.799,
+      "step": 451000
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 2.3284610086477383e-05,
+      "loss": 0.7854,
+      "step": 451010
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 2.328346215657764e-05,
+      "loss": 0.7301,
+      "step": 451020
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 2.328231422667789e-05,
+      "loss": 0.7651,
+      "step": 451030
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 2.328116629677814e-05,
+      "loss": 0.7948,
+      "step": 451040
+    },
+    {
+      "epoch": 20.69,
+      "learning_rate": 2.3280018366878396e-05,
+      "loss": 0.8792,
+      "step": 451050
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 2.3278870436978646e-05,
+      "loss": 0.7588,
+      "step": 451060
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 2.32777225070789e-05,
+      "loss": 0.8398,
+      "step": 451070
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 2.327657457717915e-05,
+      "loss": 0.9005,
+      "step": 451080
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 2.3275426647279405e-05,
+      "loss": 0.8562,
+      "step": 451090
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 2.3274278717379655e-05,
+      "loss": 0.8199,
+      "step": 451100
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 2.327313078747991e-05,
+      "loss": 0.7628,
+      "step": 451110
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 2.3271982857580162e-05,
+      "loss": 0.8161,
+      "step": 451120
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 2.3270834927680415e-05,
+      "loss": 0.7909,
+      "step": 451130
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 2.3269686997780665e-05,
+      "loss": 0.808,
+      "step": 451140
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 2.326853906788092e-05,
+      "loss": 0.7927,
+      "step": 451150
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 2.326739113798117e-05,
+      "loss": 0.906,
+      "step": 451160
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 2.326624320808142e-05,
+      "loss": 0.8523,
+      "step": 451170
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 2.3265095278181678e-05,
+      "loss": 0.7619,
+      "step": 451180
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 2.3263947348281928e-05,
+      "loss": 0.8395,
+      "step": 451190
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 2.326279941838218e-05,
+      "loss": 0.8189,
+      "step": 451200
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 2.3261651488482434e-05,
+      "loss": 0.7422,
+      "step": 451210
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 2.3260503558582687e-05,
+      "loss": 0.8323,
+      "step": 451220
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 2.3259355628682937e-05,
+      "loss": 0.7736,
+      "step": 451230
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 2.3258207698783194e-05,
+      "loss": 0.7887,
+      "step": 451240
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 2.3257059768883444e-05,
+      "loss": 0.8818,
+      "step": 451250
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 2.3255911838983697e-05,
+      "loss": 0.8232,
+      "step": 451260
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 2.325476390908395e-05,
+      "loss": 0.8418,
+      "step": 451270
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 2.3253615979184203e-05,
+      "loss": 0.8588,
+      "step": 451280
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 2.3252468049284453e-05,
+      "loss": 0.7893,
+      "step": 451290
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 2.325132011938471e-05,
+      "loss": 0.7806,
+      "step": 451300
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 2.325017218948496e-05,
+      "loss": 0.7865,
+      "step": 451310
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 2.324902425958521e-05,
+      "loss": 0.7991,
+      "step": 451320
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 2.3247876329685466e-05,
+      "loss": 0.7391,
+      "step": 451330
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 2.3246728399785716e-05,
+      "loss": 0.7373,
+      "step": 451340
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 2.324558046988597e-05,
+      "loss": 0.7682,
+      "step": 451350
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 2.3244432539986223e-05,
+      "loss": 0.7456,
+      "step": 451360
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 2.3243284610086476e-05,
+      "loss": 0.8252,
+      "step": 451370
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 2.3242136680186726e-05,
+      "loss": 0.8469,
+      "step": 451380
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 2.3240988750286982e-05,
+      "loss": 0.8963,
+      "step": 451390
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 2.3239840820387232e-05,
+      "loss": 0.7972,
+      "step": 451400
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 2.3238692890487485e-05,
+      "loss": 0.8544,
+      "step": 451410
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 2.323754496058774e-05,
+      "loss": 0.8507,
+      "step": 451420
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 2.3236397030687992e-05,
+      "loss": 0.7327,
+      "step": 451430
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 2.3235249100788242e-05,
+      "loss": 0.8319,
+      "step": 451440
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 2.32341011708885e-05,
+      "loss": 0.8776,
+      "step": 451450
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 2.3232953240988748e-05,
+      "loss": 0.7633,
+      "step": 451460
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 2.3231805311088998e-05,
+      "loss": 0.7812,
+      "step": 451470
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 2.3230657381189255e-05,
+      "loss": 0.7629,
+      "step": 451480
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 2.3229509451289505e-05,
+      "loss": 0.8634,
+      "step": 451490
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 2.3228361521389758e-05,
+      "loss": 0.7535,
+      "step": 451500
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 2.322721359149001e-05,
+      "loss": 0.8237,
+      "step": 451510
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 2.3226065661590264e-05,
+      "loss": 0.8669,
+      "step": 451520
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 2.3224917731690514e-05,
+      "loss": 0.7838,
+      "step": 451530
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 2.322376980179077e-05,
+      "loss": 0.7394,
+      "step": 451540
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 2.322262187189102e-05,
+      "loss": 0.8281,
+      "step": 451550
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 2.3221473941991274e-05,
+      "loss": 0.815,
+      "step": 451560
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 2.3220326012091527e-05,
+      "loss": 0.7977,
+      "step": 451570
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 2.321917808219178e-05,
+      "loss": 0.7849,
+      "step": 451580
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 2.321803015229203e-05,
+      "loss": 0.8187,
+      "step": 451590
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 2.3216882222392287e-05,
+      "loss": 0.7789,
+      "step": 451600
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 2.3215734292492537e-05,
+      "loss": 0.7493,
+      "step": 451610
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 2.3214586362592787e-05,
+      "loss": 0.8473,
+      "step": 451620
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 2.3213438432693043e-05,
+      "loss": 0.8669,
+      "step": 451630
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 2.3212290502793293e-05,
+      "loss": 0.756,
+      "step": 451640
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 2.3211142572893546e-05,
+      "loss": 0.8297,
+      "step": 451650
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 2.32099946429938e-05,
+      "loss": 0.7775,
+      "step": 451660
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 2.3208846713094053e-05,
+      "loss": 0.738,
+      "step": 451670
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 2.3207698783194303e-05,
+      "loss": 0.8703,
+      "step": 451680
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 2.320655085329456e-05,
+      "loss": 0.834,
+      "step": 451690
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 2.320540292339481e-05,
+      "loss": 0.816,
+      "step": 451700
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 2.3204254993495062e-05,
+      "loss": 0.8332,
+      "step": 451710
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 2.3203107063595316e-05,
+      "loss": 0.7674,
+      "step": 451720
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 2.320195913369557e-05,
+      "loss": 0.8011,
+      "step": 451730
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 2.320081120379582e-05,
+      "loss": 0.7635,
+      "step": 451740
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 2.319966327389607e-05,
+      "loss": 0.7759,
+      "step": 451750
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 2.3198515343996325e-05,
+      "loss": 0.7424,
+      "step": 451760
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 2.3197367414096575e-05,
+      "loss": 0.8723,
+      "step": 451770
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 2.319621948419683e-05,
+      "loss": 0.8318,
+      "step": 451780
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 2.319507155429708e-05,
+      "loss": 0.7864,
+      "step": 451790
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 2.3193923624397335e-05,
+      "loss": 0.7602,
+      "step": 451800
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 2.3192775694497585e-05,
+      "loss": 0.8958,
+      "step": 451810
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 2.319162776459784e-05,
+      "loss": 0.7939,
+      "step": 451820
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 2.319047983469809e-05,
+      "loss": 0.8248,
+      "step": 451830
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 2.3189331904798344e-05,
+      "loss": 0.8705,
+      "step": 451840
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 2.3188183974898598e-05,
+      "loss": 0.7909,
+      "step": 451850
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 2.318703604499885e-05,
+      "loss": 0.7507,
+      "step": 451860
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 2.31858881150991e-05,
+      "loss": 0.8021,
+      "step": 451870
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 2.3184740185199357e-05,
+      "loss": 0.8555,
+      "step": 451880
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 2.3183592255299607e-05,
+      "loss": 0.6612,
+      "step": 451890
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 2.3182444325399857e-05,
+      "loss": 0.8657,
+      "step": 451900
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 2.3181296395500114e-05,
+      "loss": 0.7845,
+      "step": 451910
+    },
+    {
+      "epoch": 20.73,
+      "learning_rate": 2.3180148465600364e-05,
+      "loss": 0.8793,
+      "step": 451920
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 2.3179000535700617e-05,
+      "loss": 0.8963,
+      "step": 451930
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 2.317785260580087e-05,
+      "loss": 0.8092,
+      "step": 451940
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 2.3176704675901123e-05,
+      "loss": 0.7948,
+      "step": 451950
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 2.3175556746001373e-05,
+      "loss": 0.7746,
+      "step": 451960
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 2.317440881610163e-05,
+      "loss": 0.8282,
+      "step": 451970
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 2.317326088620188e-05,
+      "loss": 0.7112,
+      "step": 451980
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 2.3172112956302133e-05,
+      "loss": 0.8595,
+      "step": 451990
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 2.3170965026402386e-05,
+      "loss": 0.8202,
+      "step": 452000
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 2.316981709650264e-05,
+      "loss": 0.9668,
+      "step": 452010
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 2.316866916660289e-05,
+      "loss": 0.8373,
+      "step": 452020
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 2.3167521236703146e-05,
+      "loss": 0.7989,
+      "step": 452030
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 2.3166373306803396e-05,
+      "loss": 0.8714,
+      "step": 452040
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 2.3165225376903646e-05,
+      "loss": 0.7539,
+      "step": 452050
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 2.3164077447003902e-05,
+      "loss": 0.8672,
+      "step": 452060
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 2.3162929517104152e-05,
+      "loss": 0.772,
+      "step": 452070
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 2.3161781587204405e-05,
+      "loss": 0.7463,
+      "step": 452080
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 2.316063365730466e-05,
+      "loss": 0.8669,
+      "step": 452090
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 2.3159485727404912e-05,
+      "loss": 0.7141,
+      "step": 452100
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 2.315833779750516e-05,
+      "loss": 0.8153,
+      "step": 452110
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 2.3157189867605418e-05,
+      "loss": 0.8613,
+      "step": 452120
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 2.3156041937705668e-05,
+      "loss": 0.7935,
+      "step": 452130
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 2.315489400780592e-05,
+      "loss": 0.816,
+      "step": 452140
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 2.3153746077906175e-05,
+      "loss": 0.7995,
+      "step": 452150
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 2.3152598148006428e-05,
+      "loss": 0.6633,
+      "step": 452160
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 2.3151450218106678e-05,
+      "loss": 0.8111,
+      "step": 452170
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 2.3150302288206934e-05,
+      "loss": 0.8268,
+      "step": 452180
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 2.3149154358307184e-05,
+      "loss": 0.8199,
+      "step": 452190
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 2.3148006428407434e-05,
+      "loss": 0.8051,
+      "step": 452200
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 2.314685849850769e-05,
+      "loss": 0.9486,
+      "step": 452210
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 2.314571056860794e-05,
+      "loss": 0.808,
+      "step": 452220
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 2.3144562638708194e-05,
+      "loss": 0.8657,
+      "step": 452230
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 2.3143414708808447e-05,
+      "loss": 0.73,
+      "step": 452240
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 2.31422667789087e-05,
+      "loss": 0.7875,
+      "step": 452250
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 2.314111884900895e-05,
+      "loss": 0.8494,
+      "step": 452260
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 2.3139970919109207e-05,
+      "loss": 0.8113,
+      "step": 452270
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 2.3138822989209457e-05,
+      "loss": 0.8582,
+      "step": 452280
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 2.313767505930971e-05,
+      "loss": 0.758,
+      "step": 452290
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 2.3136527129409963e-05,
+      "loss": 0.7734,
+      "step": 452300
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 2.3135379199510216e-05,
+      "loss": 0.8659,
+      "step": 452310
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 2.3134231269610466e-05,
+      "loss": 0.7113,
+      "step": 452320
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 2.3133083339710723e-05,
+      "loss": 0.8757,
+      "step": 452330
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 2.3131935409810973e-05,
+      "loss": 0.8078,
+      "step": 452340
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 2.3130787479911222e-05,
+      "loss": 0.8364,
+      "step": 452350
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 2.312963955001148e-05,
+      "loss": 0.7217,
+      "step": 452360
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 2.312849162011173e-05,
+      "loss": 0.7597,
+      "step": 452370
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 2.3127343690211982e-05,
+      "loss": 0.7843,
+      "step": 452380
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 2.3126195760312235e-05,
+      "loss": 0.8059,
+      "step": 452390
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 2.312504783041249e-05,
+      "loss": 0.7509,
+      "step": 452400
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 2.312389990051274e-05,
+      "loss": 0.8041,
+      "step": 452410
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 2.3122751970612995e-05,
+      "loss": 0.7639,
+      "step": 452420
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 2.3121604040713245e-05,
+      "loss": 0.815,
+      "step": 452430
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 2.3120456110813498e-05,
+      "loss": 0.8509,
+      "step": 452440
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 2.3119308180913748e-05,
+      "loss": 0.7828,
+      "step": 452450
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 2.3118160251014005e-05,
+      "loss": 0.8037,
+      "step": 452460
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 2.3117012321114255e-05,
+      "loss": 0.8138,
+      "step": 452470
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 2.3115864391214504e-05,
+      "loss": 0.788,
+      "step": 452480
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 2.311471646131476e-05,
+      "loss": 0.7471,
+      "step": 452490
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 2.311356853141501e-05,
+      "loss": 0.7381,
+      "step": 452500
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 2.3112420601515264e-05,
+      "loss": 0.794,
+      "step": 452510
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 2.3111272671615517e-05,
+      "loss": 0.7604,
+      "step": 452520
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 2.311012474171577e-05,
+      "loss": 0.8089,
+      "step": 452530
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 2.310897681181602e-05,
+      "loss": 0.873,
+      "step": 452540
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 2.3107828881916277e-05,
+      "loss": 0.8224,
+      "step": 452550
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 2.3106680952016527e-05,
+      "loss": 0.7994,
+      "step": 452560
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 2.310553302211678e-05,
+      "loss": 0.7557,
+      "step": 452570
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 2.3104385092217034e-05,
+      "loss": 0.7469,
+      "step": 452580
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 2.3103237162317287e-05,
+      "loss": 0.8024,
+      "step": 452590
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 2.3102089232417537e-05,
+      "loss": 0.8322,
+      "step": 452600
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 2.3100941302517793e-05,
+      "loss": 0.7919,
+      "step": 452610
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 2.3099793372618043e-05,
+      "loss": 0.7743,
+      "step": 452620
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 2.3098645442718293e-05,
+      "loss": 0.774,
+      "step": 452630
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 2.309749751281855e-05,
+      "loss": 0.8443,
+      "step": 452640
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 2.30963495829188e-05,
+      "loss": 0.814,
+      "step": 452650
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 2.3095201653019053e-05,
+      "loss": 0.7621,
+      "step": 452660
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 2.3094053723119306e-05,
+      "loss": 0.8717,
+      "step": 452670
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 2.309290579321956e-05,
+      "loss": 0.7763,
+      "step": 452680
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 2.309175786331981e-05,
+      "loss": 0.9791,
+      "step": 452690
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 2.3090609933420066e-05,
+      "loss": 0.8267,
+      "step": 452700
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 2.3089462003520315e-05,
+      "loss": 0.8532,
+      "step": 452710
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 2.308831407362057e-05,
+      "loss": 0.7903,
+      "step": 452720
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 2.3087166143720822e-05,
+      "loss": 0.8417,
+      "step": 452730
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 2.3086018213821075e-05,
+      "loss": 0.7958,
+      "step": 452740
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 2.3084870283921325e-05,
+      "loss": 0.8515,
+      "step": 452750
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 2.3083722354021582e-05,
+      "loss": 0.8168,
+      "step": 452760
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 2.308257442412183e-05,
+      "loss": 0.712,
+      "step": 452770
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 2.308142649422208e-05,
+      "loss": 0.7943,
+      "step": 452780
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 2.3080278564322338e-05,
+      "loss": 0.8468,
+      "step": 452790
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 2.3079130634422588e-05,
+      "loss": 0.8733,
+      "step": 452800
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 2.307798270452284e-05,
+      "loss": 0.7263,
+      "step": 452810
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 2.3076834774623094e-05,
+      "loss": 0.8671,
+      "step": 452820
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 2.3075686844723348e-05,
+      "loss": 0.8164,
+      "step": 452830
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 2.3074538914823597e-05,
+      "loss": 0.8058,
+      "step": 452840
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 2.3073390984923854e-05,
+      "loss": 0.7858,
+      "step": 452850
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 2.3072243055024104e-05,
+      "loss": 0.8586,
+      "step": 452860
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 2.3071095125124357e-05,
+      "loss": 0.7624,
+      "step": 452870
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 2.306994719522461e-05,
+      "loss": 0.8536,
+      "step": 452880
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 2.3068799265324864e-05,
+      "loss": 0.8262,
+      "step": 452890
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 2.3067651335425114e-05,
+      "loss": 0.7913,
+      "step": 452900
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 2.306650340552537e-05,
+      "loss": 0.7103,
+      "step": 452910
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 2.306535547562562e-05,
+      "loss": 0.7755,
+      "step": 452920
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 2.306420754572587e-05,
+      "loss": 0.7426,
+      "step": 452930
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 2.3063059615826127e-05,
+      "loss": 0.7951,
+      "step": 452940
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 2.3061911685926376e-05,
+      "loss": 0.8892,
+      "step": 452950
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 2.306076375602663e-05,
+      "loss": 0.7848,
+      "step": 452960
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 2.3059615826126883e-05,
+      "loss": 0.7479,
+      "step": 452970
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 2.3058467896227136e-05,
+      "loss": 0.864,
+      "step": 452980
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 2.3057319966327386e-05,
+      "loss": 0.7979,
+      "step": 452990
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 2.3056172036427643e-05,
+      "loss": 0.7715,
+      "step": 453000
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 2.3055024106527892e-05,
+      "loss": 0.7606,
+      "step": 453010
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 2.3053876176628146e-05,
+      "loss": 0.8739,
+      "step": 453020
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 2.30527282467284e-05,
+      "loss": 0.8104,
+      "step": 453030
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 2.3051580316828652e-05,
+      "loss": 0.8664,
+      "step": 453040
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 2.3050432386928902e-05,
+      "loss": 0.8394,
+      "step": 453050
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 2.304928445702916e-05,
+      "loss": 0.8357,
+      "step": 453060
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 2.304813652712941e-05,
+      "loss": 0.8048,
+      "step": 453070
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 2.304698859722966e-05,
+      "loss": 0.8322,
+      "step": 453080
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 2.3045840667329915e-05,
+      "loss": 0.8198,
+      "step": 453090
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 2.3044692737430165e-05,
+      "loss": 0.803,
+      "step": 453100
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 2.3043544807530418e-05,
+      "loss": 0.8105,
+      "step": 453110
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 2.3042396877630668e-05,
+      "loss": 0.7787,
+      "step": 453120
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 2.3041248947730925e-05,
+      "loss": 0.7272,
+      "step": 453130
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 2.3040101017831174e-05,
+      "loss": 0.8156,
+      "step": 453140
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 2.3038953087931428e-05,
+      "loss": 0.7666,
+      "step": 453150
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 2.303780515803168e-05,
+      "loss": 0.7261,
+      "step": 453160
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 2.3036657228131934e-05,
+      "loss": 0.7585,
+      "step": 453170
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 2.3035509298232184e-05,
+      "loss": 0.8457,
+      "step": 453180
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 2.303436136833244e-05,
+      "loss": 0.814,
+      "step": 453190
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 2.303321343843269e-05,
+      "loss": 0.8478,
+      "step": 453200
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 2.303206550853294e-05,
+      "loss": 0.8012,
+      "step": 453210
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 2.3030917578633197e-05,
+      "loss": 0.7659,
+      "step": 453220
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 2.3029769648733447e-05,
+      "loss": 0.6913,
+      "step": 453230
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 2.30286217188337e-05,
+      "loss": 0.8558,
+      "step": 453240
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 2.3027473788933953e-05,
+      "loss": 0.7741,
+      "step": 453250
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 2.3026325859034207e-05,
+      "loss": 0.7732,
+      "step": 453260
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 2.3025177929134456e-05,
+      "loss": 0.8403,
+      "step": 453270
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 2.3024029999234713e-05,
+      "loss": 0.8384,
+      "step": 453280
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 2.3022882069334963e-05,
+      "loss": 0.8152,
+      "step": 453290
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 2.3021734139435216e-05,
+      "loss": 0.7888,
+      "step": 453300
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 2.302058620953547e-05,
+      "loss": 0.7985,
+      "step": 453310
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 2.3019438279635723e-05,
+      "loss": 0.872,
+      "step": 453320
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 2.3018290349735972e-05,
+      "loss": 0.7791,
+      "step": 453330
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 2.301714241983623e-05,
+      "loss": 0.7882,
+      "step": 453340
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 2.301599448993648e-05,
+      "loss": 0.7596,
+      "step": 453350
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 2.301484656003673e-05,
+      "loss": 0.8483,
+      "step": 453360
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 2.3013698630136985e-05,
+      "loss": 0.8118,
+      "step": 453370
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 2.3012550700237235e-05,
+      "loss": 0.827,
+      "step": 453380
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 2.301140277033749e-05,
+      "loss": 0.9431,
+      "step": 453390
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 2.3010254840437742e-05,
+      "loss": 0.8108,
+      "step": 453400
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 2.3009106910537995e-05,
+      "loss": 0.803,
+      "step": 453410
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 2.3007958980638245e-05,
+      "loss": 0.7355,
+      "step": 453420
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 2.30068110507385e-05,
+      "loss": 0.8552,
+      "step": 453430
+    },
+    {
+      "epoch": 20.8,
+      "learning_rate": 2.300566312083875e-05,
+      "loss": 0.8287,
+      "step": 453440
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 2.3004515190939005e-05,
+      "loss": 0.8806,
+      "step": 453450
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 2.3003367261039258e-05,
+      "loss": 0.7601,
+      "step": 453460
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 2.300221933113951e-05,
+      "loss": 0.8848,
+      "step": 453470
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 2.300107140123976e-05,
+      "loss": 0.8322,
+      "step": 453480
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 2.2999923471340018e-05,
+      "loss": 0.7903,
+      "step": 453490
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 2.2998775541440267e-05,
+      "loss": 0.7907,
+      "step": 453500
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 2.2997627611540517e-05,
+      "loss": 0.7812,
+      "step": 453510
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 2.2996479681640774e-05,
+      "loss": 0.7756,
+      "step": 453520
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 2.2995331751741024e-05,
+      "loss": 0.7771,
+      "step": 453530
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 2.2994183821841277e-05,
+      "loss": 0.8048,
+      "step": 453540
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 2.299303589194153e-05,
+      "loss": 0.861,
+      "step": 453550
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 2.2991887962041783e-05,
+      "loss": 0.778,
+      "step": 453560
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 2.2990740032142033e-05,
+      "loss": 0.7795,
+      "step": 453570
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 2.298959210224229e-05,
+      "loss": 0.7994,
+      "step": 453580
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 2.298844417234254e-05,
+      "loss": 0.852,
+      "step": 453590
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 2.2987296242442793e-05,
+      "loss": 0.8553,
+      "step": 453600
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 2.2986148312543046e-05,
+      "loss": 0.8803,
+      "step": 453610
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 2.29850003826433e-05,
+      "loss": 0.8241,
+      "step": 453620
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 2.298385245274355e-05,
+      "loss": 0.8004,
+      "step": 453630
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 2.2982704522843806e-05,
+      "loss": 0.7729,
+      "step": 453640
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 2.2981556592944056e-05,
+      "loss": 0.7741,
+      "step": 453650
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 2.2980408663044306e-05,
+      "loss": 0.8138,
+      "step": 453660
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 2.2979260733144562e-05,
+      "loss": 0.8492,
+      "step": 453670
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 2.2978112803244812e-05,
+      "loss": 0.7908,
+      "step": 453680
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 2.2976964873345065e-05,
+      "loss": 0.7703,
+      "step": 453690
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 2.297581694344532e-05,
+      "loss": 0.843,
+      "step": 453700
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 2.2974669013545572e-05,
+      "loss": 0.7657,
+      "step": 453710
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 2.2973521083645822e-05,
+      "loss": 0.8074,
+      "step": 453720
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 2.297237315374608e-05,
+      "loss": 0.7093,
+      "step": 453730
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 2.297122522384633e-05,
+      "loss": 0.8212,
+      "step": 453740
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 2.297007729394658e-05,
+      "loss": 0.88,
+      "step": 453750
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 2.2968929364046835e-05,
+      "loss": 0.8449,
+      "step": 453760
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 2.2967781434147088e-05,
+      "loss": 0.8165,
+      "step": 453770
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 2.2966633504247338e-05,
+      "loss": 0.7849,
+      "step": 453780
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 2.2965485574347595e-05,
+      "loss": 0.7668,
+      "step": 453790
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 2.2964337644447844e-05,
+      "loss": 0.814,
+      "step": 453800
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 2.2963189714548094e-05,
+      "loss": 0.891,
+      "step": 453810
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 2.2962041784648347e-05,
+      "loss": 0.8179,
+      "step": 453820
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 2.29608938547486e-05,
+      "loss": 0.8914,
+      "step": 453830
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 2.2959745924848854e-05,
+      "loss": 0.8428,
+      "step": 453840
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 2.2958597994949104e-05,
+      "loss": 0.8316,
+      "step": 453850
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 2.295745006504936e-05,
+      "loss": 0.8366,
+      "step": 453860
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 2.295630213514961e-05,
+      "loss": 0.7155,
+      "step": 453870
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 2.2955154205249864e-05,
+      "loss": 0.7621,
+      "step": 453880
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 2.2954006275350117e-05,
+      "loss": 0.7801,
+      "step": 453890
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 2.295285834545037e-05,
+      "loss": 0.7787,
+      "step": 453900
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 2.295171041555062e-05,
+      "loss": 0.8032,
+      "step": 453910
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 2.2950562485650877e-05,
+      "loss": 0.8675,
+      "step": 453920
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 2.2949414555751126e-05,
+      "loss": 0.8685,
+      "step": 453930
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 2.2948266625851376e-05,
+      "loss": 0.7955,
+      "step": 453940
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 2.2947118695951633e-05,
+      "loss": 0.7316,
+      "step": 453950
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 2.2945970766051883e-05,
+      "loss": 0.8152,
+      "step": 453960
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 2.2944822836152136e-05,
+      "loss": 0.8087,
+      "step": 453970
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 2.294367490625239e-05,
+      "loss": 0.7464,
+      "step": 453980
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 2.2942526976352642e-05,
+      "loss": 0.7834,
+      "step": 453990
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 2.2941379046452892e-05,
+      "loss": 0.8721,
+      "step": 454000
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 2.294023111655315e-05,
+      "loss": 0.8321,
+      "step": 454010
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 2.29390831866534e-05,
+      "loss": 0.7438,
+      "step": 454020
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 2.2937935256753652e-05,
+      "loss": 0.7775,
+      "step": 454030
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 2.2936787326853905e-05,
+      "loss": 0.7546,
+      "step": 454040
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 2.293563939695416e-05,
+      "loss": 0.7679,
+      "step": 454050
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 2.293449146705441e-05,
+      "loss": 0.887,
+      "step": 454060
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 2.2933343537154665e-05,
+      "loss": 0.7804,
+      "step": 454070
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 2.2932195607254915e-05,
+      "loss": 0.8318,
+      "step": 454080
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 2.2931047677355165e-05,
+      "loss": 0.8557,
+      "step": 454090
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 2.292989974745542e-05,
+      "loss": 0.7718,
+      "step": 454100
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 2.292875181755567e-05,
+      "loss": 0.8122,
+      "step": 454110
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 2.2927603887655924e-05,
+      "loss": 0.7995,
+      "step": 454120
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 2.2926455957756178e-05,
+      "loss": 0.8051,
+      "step": 454130
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 2.292530802785643e-05,
+      "loss": 0.8616,
+      "step": 454140
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 2.292416009795668e-05,
+      "loss": 0.7591,
+      "step": 454150
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 2.2923012168056937e-05,
+      "loss": 0.7818,
+      "step": 454160
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 2.2921864238157187e-05,
+      "loss": 0.8603,
+      "step": 454170
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 2.292071630825744e-05,
+      "loss": 0.8253,
+      "step": 454180
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 2.2919568378357694e-05,
+      "loss": 0.795,
+      "step": 454190
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 2.2918420448457947e-05,
+      "loss": 0.9382,
+      "step": 454200
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 2.2917272518558197e-05,
+      "loss": 0.7876,
+      "step": 454210
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 2.2916124588658453e-05,
+      "loss": 0.8225,
+      "step": 454220
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 2.2914976658758703e-05,
+      "loss": 0.8393,
+      "step": 454230
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 2.2913828728858953e-05,
+      "loss": 0.7788,
+      "step": 454240
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 2.291268079895921e-05,
+      "loss": 0.8331,
+      "step": 454250
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 2.291153286905946e-05,
+      "loss": 0.7752,
+      "step": 454260
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 2.2910384939159713e-05,
+      "loss": 0.8223,
+      "step": 454270
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 2.2909237009259966e-05,
+      "loss": 0.7349,
+      "step": 454280
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 2.290808907936022e-05,
+      "loss": 0.8522,
+      "step": 454290
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 2.290694114946047e-05,
+      "loss": 0.7463,
+      "step": 454300
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 2.2905793219560726e-05,
+      "loss": 0.8536,
+      "step": 454310
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 2.2904645289660976e-05,
+      "loss": 0.8526,
+      "step": 454320
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 2.290349735976123e-05,
+      "loss": 0.7607,
+      "step": 454330
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 2.2902349429861482e-05,
+      "loss": 0.825,
+      "step": 454340
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 2.2901201499961735e-05,
+      "loss": 0.8619,
+      "step": 454350
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 2.2900053570061985e-05,
+      "loss": 0.7745,
+      "step": 454360
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 2.2898905640162242e-05,
+      "loss": 0.7548,
+      "step": 454370
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 2.2897757710262492e-05,
+      "loss": 0.7716,
+      "step": 454380
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 2.289660978036274e-05,
+      "loss": 0.7473,
+      "step": 454390
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 2.2895461850462998e-05,
+      "loss": 0.8892,
+      "step": 454400
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 2.2894313920563248e-05,
+      "loss": 0.7795,
+      "step": 454410
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 2.28931659906635e-05,
+      "loss": 0.79,
+      "step": 454420
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 2.2892018060763755e-05,
+      "loss": 0.7557,
+      "step": 454430
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 2.2890870130864008e-05,
+      "loss": 0.7511,
+      "step": 454440
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 2.2889722200964258e-05,
+      "loss": 0.8339,
+      "step": 454450
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 2.2888574271064514e-05,
+      "loss": 0.7267,
+      "step": 454460
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 2.2887426341164764e-05,
+      "loss": 0.8029,
+      "step": 454470
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 2.2886278411265017e-05,
+      "loss": 0.7979,
+      "step": 454480
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 2.2885130481365267e-05,
+      "loss": 0.8335,
+      "step": 454490
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 2.2883982551465524e-05,
+      "loss": 0.7613,
+      "step": 454500
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 2.2882834621565774e-05,
+      "loss": 0.6901,
+      "step": 454510
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 2.2881686691666024e-05,
+      "loss": 0.7565,
+      "step": 454520
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 2.288053876176628e-05,
+      "loss": 0.7722,
+      "step": 454530
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 2.287939083186653e-05,
+      "loss": 0.8364,
+      "step": 454540
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 2.2878242901966783e-05,
+      "loss": 0.7909,
+      "step": 454550
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 2.2877094972067037e-05,
+      "loss": 0.8186,
+      "step": 454560
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 2.287594704216729e-05,
+      "loss": 0.7889,
+      "step": 454570
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 2.287479911226754e-05,
+      "loss": 0.767,
+      "step": 454580
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 2.2873651182367796e-05,
+      "loss": 0.7857,
+      "step": 454590
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 2.2872503252468046e-05,
+      "loss": 0.8593,
+      "step": 454600
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 2.28713553225683e-05,
+      "loss": 0.7652,
+      "step": 454610
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 2.2870207392668553e-05,
+      "loss": 0.7645,
+      "step": 454620
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 2.2869059462768806e-05,
+      "loss": 0.8194,
+      "step": 454630
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 2.2867911532869056e-05,
+      "loss": 0.8155,
+      "step": 454640
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 2.2866763602969312e-05,
+      "loss": 0.7182,
+      "step": 454650
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 2.2865615673069562e-05,
+      "loss": 0.8214,
+      "step": 454660
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 2.2864467743169812e-05,
+      "loss": 0.8428,
+      "step": 454670
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 2.286331981327007e-05,
+      "loss": 0.8824,
+      "step": 454680
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 2.286217188337032e-05,
+      "loss": 0.8463,
+      "step": 454690
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 2.2861023953470572e-05,
+      "loss": 0.8422,
+      "step": 454700
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 2.2859876023570825e-05,
+      "loss": 0.7654,
+      "step": 454710
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 2.2858728093671078e-05,
+      "loss": 0.8702,
+      "step": 454720
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 2.2857580163771328e-05,
+      "loss": 0.8838,
+      "step": 454730
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 2.2856432233871585e-05,
+      "loss": 0.7935,
+      "step": 454740
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 2.2855284303971835e-05,
+      "loss": 0.7624,
+      "step": 454750
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 2.2854136374072088e-05,
+      "loss": 0.7997,
+      "step": 454760
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 2.285298844417234e-05,
+      "loss": 0.8789,
+      "step": 454770
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 2.2851840514272594e-05,
+      "loss": 0.7395,
+      "step": 454780
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 2.2850692584372844e-05,
+      "loss": 0.8119,
+      "step": 454790
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 2.28495446544731e-05,
+      "loss": 0.8586,
+      "step": 454800
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 2.284839672457335e-05,
+      "loss": 0.8463,
+      "step": 454810
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 2.28472487946736e-05,
+      "loss": 0.8529,
+      "step": 454820
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 2.2846100864773857e-05,
+      "loss": 0.7777,
+      "step": 454830
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 2.2844952934874107e-05,
+      "loss": 0.8098,
+      "step": 454840
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 2.284380500497436e-05,
+      "loss": 0.72,
+      "step": 454850
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 2.2842657075074614e-05,
+      "loss": 0.8475,
+      "step": 454860
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 2.2841509145174867e-05,
+      "loss": 0.7769,
+      "step": 454870
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 2.2840361215275117e-05,
+      "loss": 0.7648,
+      "step": 454880
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 2.2839213285375373e-05,
+      "loss": 0.8589,
+      "step": 454890
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 2.2838065355475623e-05,
+      "loss": 0.907,
+      "step": 454900
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 2.2836917425575876e-05,
+      "loss": 0.8349,
+      "step": 454910
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 2.283576949567613e-05,
+      "loss": 0.811,
+      "step": 454920
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 2.2834621565776383e-05,
+      "loss": 0.7261,
+      "step": 454930
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 2.2833473635876633e-05,
+      "loss": 0.8102,
+      "step": 454940
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 2.283232570597689e-05,
+      "loss": 0.8155,
+      "step": 454950
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 2.283117777607714e-05,
+      "loss": 0.8208,
+      "step": 454960
+    },
+    {
+      "epoch": 20.87,
+      "learning_rate": 2.283002984617739e-05,
+      "loss": 0.8471,
+      "step": 454970
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 2.2828881916277646e-05,
+      "loss": 0.9682,
+      "step": 454980
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 2.2827733986377896e-05,
+      "loss": 0.9225,
+      "step": 454990
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 2.282658605647815e-05,
+      "loss": 0.7889,
+      "step": 455000
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 2.2825438126578402e-05,
+      "loss": 0.7628,
+      "step": 455010
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 2.2824290196678655e-05,
+      "loss": 0.868,
+      "step": 455020
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 2.2823142266778905e-05,
+      "loss": 0.831,
+      "step": 455030
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 2.2821994336879162e-05,
+      "loss": 0.8322,
+      "step": 455040
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 2.282084640697941e-05,
+      "loss": 0.8047,
+      "step": 455050
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 2.2819698477079665e-05,
+      "loss": 0.8534,
+      "step": 455060
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 2.2818550547179918e-05,
+      "loss": 0.6789,
+      "step": 455070
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 2.281740261728017e-05,
+      "loss": 0.8034,
+      "step": 455080
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 2.281625468738042e-05,
+      "loss": 0.7893,
+      "step": 455090
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 2.2815106757480678e-05,
+      "loss": 0.79,
+      "step": 455100
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 2.2813958827580928e-05,
+      "loss": 0.7616,
+      "step": 455110
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 2.2812810897681178e-05,
+      "loss": 0.8414,
+      "step": 455120
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 2.2811662967781434e-05,
+      "loss": 0.7916,
+      "step": 455130
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 2.2810515037881684e-05,
+      "loss": 0.7669,
+      "step": 455140
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 2.2809367107981937e-05,
+      "loss": 0.8434,
+      "step": 455150
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 2.280821917808219e-05,
+      "loss": 0.7752,
+      "step": 455160
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 2.2807071248182444e-05,
+      "loss": 0.8302,
+      "step": 455170
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 2.2805923318282694e-05,
+      "loss": 0.8352,
+      "step": 455180
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 2.2804775388382947e-05,
+      "loss": 0.7761,
+      "step": 455190
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 2.28036274584832e-05,
+      "loss": 0.7591,
+      "step": 455200
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 2.2802479528583453e-05,
+      "loss": 0.7241,
+      "step": 455210
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 2.2801331598683703e-05,
+      "loss": 0.7794,
+      "step": 455220
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 2.280018366878396e-05,
+      "loss": 0.8718,
+      "step": 455230
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 2.279903573888421e-05,
+      "loss": 0.7864,
+      "step": 455240
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 2.279788780898446e-05,
+      "loss": 0.8207,
+      "step": 455250
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 2.2796739879084716e-05,
+      "loss": 0.7999,
+      "step": 455260
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 2.2795591949184966e-05,
+      "loss": 0.7965,
+      "step": 455270
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 2.279444401928522e-05,
+      "loss": 0.9087,
+      "step": 455280
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 2.2793296089385472e-05,
+      "loss": 0.9215,
+      "step": 455290
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 2.2792148159485726e-05,
+      "loss": 0.8326,
+      "step": 455300
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 2.2791000229585976e-05,
+      "loss": 0.8468,
+      "step": 455310
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 2.2789852299686232e-05,
+      "loss": 0.86,
+      "step": 455320
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 2.2788704369786482e-05,
+      "loss": 0.7985,
+      "step": 455330
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 2.2787556439886735e-05,
+      "loss": 0.8561,
+      "step": 455340
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 2.278640850998699e-05,
+      "loss": 0.8538,
+      "step": 455350
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 2.2785260580087242e-05,
+      "loss": 0.7927,
+      "step": 455360
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 2.278411265018749e-05,
+      "loss": 0.8425,
+      "step": 455370
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 2.2782964720287748e-05,
+      "loss": 0.8154,
+      "step": 455380
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 2.2781816790387998e-05,
+      "loss": 0.873,
+      "step": 455390
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 2.2780668860488248e-05,
+      "loss": 0.8159,
+      "step": 455400
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 2.2779520930588505e-05,
+      "loss": 0.8422,
+      "step": 455410
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 2.2778373000688754e-05,
+      "loss": 0.8019,
+      "step": 455420
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 2.2777225070789008e-05,
+      "loss": 0.8458,
+      "step": 455430
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 2.277607714088926e-05,
+      "loss": 0.8177,
+      "step": 455440
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 2.2774929210989514e-05,
+      "loss": 0.7776,
+      "step": 455450
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 2.2773781281089764e-05,
+      "loss": 0.7828,
+      "step": 455460
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 2.277263335119002e-05,
+      "loss": 0.7992,
+      "step": 455470
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 2.277148542129027e-05,
+      "loss": 0.8728,
+      "step": 455480
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 2.2770337491390524e-05,
+      "loss": 0.7995,
+      "step": 455490
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 2.2769189561490777e-05,
+      "loss": 0.897,
+      "step": 455500
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 2.276804163159103e-05,
+      "loss": 0.8052,
+      "step": 455510
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 2.276689370169128e-05,
+      "loss": 0.8347,
+      "step": 455520
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 2.2765745771791537e-05,
+      "loss": 0.8383,
+      "step": 455530
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 2.2764597841891787e-05,
+      "loss": 0.8334,
+      "step": 455540
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 2.2763449911992036e-05,
+      "loss": 0.8219,
+      "step": 455550
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 2.2762301982092293e-05,
+      "loss": 0.766,
+      "step": 455560
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 2.2761154052192543e-05,
+      "loss": 0.8477,
+      "step": 455570
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 2.2760006122292796e-05,
+      "loss": 0.7756,
+      "step": 455580
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 2.275885819239305e-05,
+      "loss": 0.793,
+      "step": 455590
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 2.2757710262493303e-05,
+      "loss": 0.7332,
+      "step": 455600
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 2.2756562332593553e-05,
+      "loss": 0.8914,
+      "step": 455610
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 2.275541440269381e-05,
+      "loss": 0.7541,
+      "step": 455620
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 2.275426647279406e-05,
+      "loss": 0.7244,
+      "step": 455630
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 2.2753118542894312e-05,
+      "loss": 0.8837,
+      "step": 455640
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 2.2751970612994565e-05,
+      "loss": 0.7772,
+      "step": 455650
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 2.275082268309482e-05,
+      "loss": 0.9267,
+      "step": 455660
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 2.274967475319507e-05,
+      "loss": 0.7614,
+      "step": 455670
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 2.2748526823295325e-05,
+      "loss": 0.8551,
+      "step": 455680
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 2.2747378893395575e-05,
+      "loss": 0.8177,
+      "step": 455690
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 2.2746230963495825e-05,
+      "loss": 0.8153,
+      "step": 455700
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 2.274508303359608e-05,
+      "loss": 0.8205,
+      "step": 455710
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 2.274393510369633e-05,
+      "loss": 0.847,
+      "step": 455720
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 2.2742787173796585e-05,
+      "loss": 0.7829,
+      "step": 455730
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 2.2741639243896838e-05,
+      "loss": 0.7906,
+      "step": 455740
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 2.274049131399709e-05,
+      "loss": 0.7699,
+      "step": 455750
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 2.273934338409734e-05,
+      "loss": 0.8391,
+      "step": 455760
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 2.2738195454197598e-05,
+      "loss": 0.7848,
+      "step": 455770
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 2.2737047524297847e-05,
+      "loss": 0.8901,
+      "step": 455780
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 2.27358995943981e-05,
+      "loss": 0.8212,
+      "step": 455790
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 2.2734751664498354e-05,
+      "loss": 0.7811,
+      "step": 455800
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 2.2733603734598607e-05,
+      "loss": 0.741,
+      "step": 455810
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 2.2732455804698857e-05,
+      "loss": 0.8159,
+      "step": 455820
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 2.2731307874799114e-05,
+      "loss": 0.8285,
+      "step": 455830
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 2.2730159944899364e-05,
+      "loss": 0.8968,
+      "step": 455840
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 2.2729012014999613e-05,
+      "loss": 0.8857,
+      "step": 455850
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 2.2727864085099867e-05,
+      "loss": 0.7322,
+      "step": 455860
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 2.272671615520012e-05,
+      "loss": 0.8604,
+      "step": 455870
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 2.2725568225300373e-05,
+      "loss": 0.8606,
+      "step": 455880
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 2.2724420295400623e-05,
+      "loss": 0.8355,
+      "step": 455890
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 2.272327236550088e-05,
+      "loss": 0.7784,
+      "step": 455900
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 2.272212443560113e-05,
+      "loss": 0.7478,
+      "step": 455910
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 2.2720976505701383e-05,
+      "loss": 0.7435,
+      "step": 455920
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 2.2719828575801636e-05,
+      "loss": 0.8191,
+      "step": 455930
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 2.271868064590189e-05,
+      "loss": 0.7747,
+      "step": 455940
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 2.271753271600214e-05,
+      "loss": 0.8289,
+      "step": 455950
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 2.2716384786102396e-05,
+      "loss": 0.786,
+      "step": 455960
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 2.2715236856202646e-05,
+      "loss": 0.742,
+      "step": 455970
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 2.2714088926302895e-05,
+      "loss": 0.8406,
+      "step": 455980
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 2.2712940996403152e-05,
+      "loss": 0.766,
+      "step": 455990
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 2.2711793066503402e-05,
+      "loss": 0.8373,
+      "step": 456000
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 2.2710645136603655e-05,
+      "loss": 0.741,
+      "step": 456010
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 2.270949720670391e-05,
+      "loss": 0.8747,
+      "step": 456020
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 2.270834927680416e-05,
+      "loss": 0.7212,
+      "step": 456030
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 2.270720134690441e-05,
+      "loss": 0.8739,
+      "step": 456040
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 2.2706053417004668e-05,
+      "loss": 0.726,
+      "step": 456050
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 2.2704905487104918e-05,
+      "loss": 0.8826,
+      "step": 456060
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 2.270375755720517e-05,
+      "loss": 0.8124,
+      "step": 456070
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 2.2702609627305424e-05,
+      "loss": 0.7454,
+      "step": 456080
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 2.2701461697405678e-05,
+      "loss": 0.8373,
+      "step": 456090
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 2.2700313767505928e-05,
+      "loss": 0.8727,
+      "step": 456100
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 2.2699165837606184e-05,
+      "loss": 0.8514,
+      "step": 456110
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 2.2698017907706434e-05,
+      "loss": 0.7924,
+      "step": 456120
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 2.2696869977806684e-05,
+      "loss": 0.8652,
+      "step": 456130
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 2.269572204790694e-05,
+      "loss": 0.8905,
+      "step": 456140
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 2.269457411800719e-05,
+      "loss": 0.742,
+      "step": 456150
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 2.2693426188107444e-05,
+      "loss": 0.8246,
+      "step": 456160
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 2.2692278258207697e-05,
+      "loss": 0.7576,
+      "step": 456170
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 2.269113032830795e-05,
+      "loss": 0.8443,
+      "step": 456180
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 2.26899823984082e-05,
+      "loss": 0.7893,
+      "step": 456190
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 2.2688834468508457e-05,
+      "loss": 0.8004,
+      "step": 456200
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 2.2687686538608706e-05,
+      "loss": 0.7906,
+      "step": 456210
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 2.268653860870896e-05,
+      "loss": 0.8233,
+      "step": 456220
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 2.2685390678809213e-05,
+      "loss": 0.7818,
+      "step": 456230
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 2.2684242748909466e-05,
+      "loss": 0.8631,
+      "step": 456240
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 2.2683094819009716e-05,
+      "loss": 0.8243,
+      "step": 456250
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 2.2681946889109973e-05,
+      "loss": 0.877,
+      "step": 456260
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 2.2680798959210222e-05,
+      "loss": 0.8625,
+      "step": 456270
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 2.2679651029310472e-05,
+      "loss": 0.8554,
+      "step": 456280
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 2.267850309941073e-05,
+      "loss": 0.7822,
+      "step": 456290
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 2.267735516951098e-05,
+      "loss": 0.8267,
+      "step": 456300
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 2.2676207239611232e-05,
+      "loss": 0.846,
+      "step": 456310
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 2.2675059309711485e-05,
+      "loss": 0.7657,
+      "step": 456320
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 2.267391137981174e-05,
+      "loss": 0.8537,
+      "step": 456330
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 2.267276344991199e-05,
+      "loss": 0.7213,
+      "step": 456340
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 2.2671615520012245e-05,
+      "loss": 0.7622,
+      "step": 456350
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 2.2670467590112495e-05,
+      "loss": 0.7899,
+      "step": 456360
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 2.2669319660212748e-05,
+      "loss": 0.8421,
+      "step": 456370
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 2.2668171730313e-05,
+      "loss": 0.8588,
+      "step": 456380
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 2.2667023800413255e-05,
+      "loss": 0.8749,
+      "step": 456390
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 2.2665875870513504e-05,
+      "loss": 0.8729,
+      "step": 456400
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 2.266472794061376e-05,
+      "loss": 0.8363,
+      "step": 456410
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 2.266358001071401e-05,
+      "loss": 0.8202,
+      "step": 456420
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 2.266243208081426e-05,
+      "loss": 0.8345,
+      "step": 456430
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 2.2661284150914517e-05,
+      "loss": 0.7932,
+      "step": 456440
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 2.2660136221014767e-05,
+      "loss": 0.8977,
+      "step": 456450
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 2.265898829111502e-05,
+      "loss": 0.793,
+      "step": 456460
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 2.2657840361215274e-05,
+      "loss": 0.9,
+      "step": 456470
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 2.2656692431315527e-05,
+      "loss": 0.9995,
+      "step": 456480
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 2.2655544501415777e-05,
+      "loss": 0.7148,
+      "step": 456490
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 2.2654396571516033e-05,
+      "loss": 0.7652,
+      "step": 456500
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 2.2653248641616283e-05,
+      "loss": 0.7967,
+      "step": 456510
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 2.2652100711716537e-05,
+      "loss": 0.8516,
+      "step": 456520
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 2.265095278181679e-05,
+      "loss": 0.8102,
+      "step": 456530
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 2.2649804851917043e-05,
+      "loss": 0.83,
+      "step": 456540
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 2.2648656922017293e-05,
+      "loss": 0.8467,
+      "step": 456550
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 2.2647508992117543e-05,
+      "loss": 0.8192,
+      "step": 456560
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 2.26463610622178e-05,
+      "loss": 0.8151,
+      "step": 456570
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 2.264521313231805e-05,
+      "loss": 0.8267,
+      "step": 456580
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 2.2644065202418303e-05,
+      "loss": 0.7471,
+      "step": 456590
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 2.2642917272518556e-05,
+      "loss": 0.8671,
+      "step": 456600
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 2.264176934261881e-05,
+      "loss": 0.8138,
+      "step": 456610
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 2.264062141271906e-05,
+      "loss": 0.8656,
+      "step": 456620
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 2.2639473482819315e-05,
+      "loss": 0.7947,
+      "step": 456630
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 2.2638325552919565e-05,
+      "loss": 0.7822,
+      "step": 456640
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 2.263717762301982e-05,
+      "loss": 0.8226,
+      "step": 456650
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 2.2636029693120072e-05,
+      "loss": 0.6797,
+      "step": 456660
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 2.2634881763220325e-05,
+      "loss": 0.7977,
+      "step": 456670
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 2.2633733833320575e-05,
+      "loss": 0.7707,
+      "step": 456680
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 2.263258590342083e-05,
+      "loss": 0.7946,
+      "step": 456690
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 2.263143797352108e-05,
+      "loss": 0.7961,
+      "step": 456700
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 2.263029004362133e-05,
+      "loss": 0.78,
+      "step": 456710
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 2.2629142113721588e-05,
+      "loss": 0.8904,
+      "step": 456720
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 2.2627994183821838e-05,
+      "loss": 0.8001,
+      "step": 456730
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 2.262684625392209e-05,
+      "loss": 0.7286,
+      "step": 456740
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 2.2625698324022344e-05,
+      "loss": 0.8532,
+      "step": 456750
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 2.2624550394122597e-05,
+      "loss": 0.7404,
+      "step": 456760
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 2.2623402464222847e-05,
+      "loss": 0.904,
+      "step": 456770
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 2.2622254534323104e-05,
+      "loss": 0.8022,
+      "step": 456780
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 2.2621106604423354e-05,
+      "loss": 0.7986,
+      "step": 456790
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 2.2619958674523607e-05,
+      "loss": 0.874,
+      "step": 456800
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 2.261881074462386e-05,
+      "loss": 0.7958,
+      "step": 456810
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 2.2617662814724114e-05,
+      "loss": 0.8517,
+      "step": 456820
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 2.2616514884824363e-05,
+      "loss": 0.798,
+      "step": 456830
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 2.261536695492462e-05,
+      "loss": 0.8531,
+      "step": 456840
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 2.261421902502487e-05,
+      "loss": 0.8487,
+      "step": 456850
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 2.261307109512512e-05,
+      "loss": 0.7863,
+      "step": 456860
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 2.2611923165225376e-05,
+      "loss": 0.8609,
+      "step": 456870
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 2.2610775235325626e-05,
+      "loss": 0.7743,
+      "step": 456880
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 2.260962730542588e-05,
+      "loss": 0.866,
+      "step": 456890
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 2.2608479375526133e-05,
+      "loss": 0.8722,
+      "step": 456900
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 2.2607331445626386e-05,
+      "loss": 0.8689,
+      "step": 456910
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 2.2606183515726636e-05,
+      "loss": 0.787,
+      "step": 456920
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 2.2605035585826892e-05,
+      "loss": 0.7912,
+      "step": 456930
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 2.2603887655927142e-05,
+      "loss": 0.764,
+      "step": 456940
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 2.2602739726027396e-05,
+      "loss": 0.8877,
+      "step": 456950
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 2.260159179612765e-05,
+      "loss": 0.941,
+      "step": 456960
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 2.2600443866227902e-05,
+      "loss": 0.7123,
+      "step": 456970
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 2.2599295936328152e-05,
+      "loss": 0.7594,
+      "step": 456980
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 2.259814800642841e-05,
+      "loss": 0.7275,
+      "step": 456990
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 2.259700007652866e-05,
+      "loss": 0.8283,
+      "step": 457000
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 2.2595852146628908e-05,
+      "loss": 0.8074,
+      "step": 457010
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 2.2594704216729165e-05,
+      "loss": 0.8191,
+      "step": 457020
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 2.2593556286829415e-05,
+      "loss": 0.8255,
+      "step": 457030
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 2.2592408356929668e-05,
+      "loss": 0.8202,
+      "step": 457040
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 2.259126042702992e-05,
+      "loss": 0.8135,
+      "step": 457050
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 2.2590112497130174e-05,
+      "loss": 0.8634,
+      "step": 457060
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 2.2588964567230424e-05,
+      "loss": 0.7456,
+      "step": 457070
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 2.258781663733068e-05,
+      "loss": 0.828,
+      "step": 457080
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 2.258666870743093e-05,
+      "loss": 0.7953,
+      "step": 457090
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 2.2585520777531184e-05,
+      "loss": 0.8006,
+      "step": 457100
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 2.2584372847631437e-05,
+      "loss": 0.6588,
+      "step": 457110
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 2.258322491773169e-05,
+      "loss": 0.8732,
+      "step": 457120
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 2.258207698783194e-05,
+      "loss": 0.8739,
+      "step": 457130
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 2.2580929057932197e-05,
+      "loss": 0.8107,
+      "step": 457140
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 2.2579781128032447e-05,
+      "loss": 0.8125,
+      "step": 457150
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 2.2578633198132697e-05,
+      "loss": 0.7887,
+      "step": 457160
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 2.2577485268232953e-05,
+      "loss": 0.7828,
+      "step": 457170
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 2.2576337338333203e-05,
+      "loss": 0.7584,
+      "step": 457180
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 2.2575189408433456e-05,
+      "loss": 0.845,
+      "step": 457190
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 2.257404147853371e-05,
+      "loss": 0.857,
+      "step": 457200
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 2.2572893548633963e-05,
+      "loss": 0.8268,
+      "step": 457210
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 2.2571745618734213e-05,
+      "loss": 0.7172,
+      "step": 457220
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 2.2570597688834466e-05,
+      "loss": 0.8225,
+      "step": 457230
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 2.256944975893472e-05,
+      "loss": 0.793,
+      "step": 457240
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 2.2568301829034972e-05,
+      "loss": 0.8378,
+      "step": 457250
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 2.2567153899135222e-05,
+      "loss": 0.7855,
+      "step": 457260
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 2.256600596923548e-05,
+      "loss": 0.8404,
+      "step": 457270
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 2.256485803933573e-05,
+      "loss": 0.8593,
+      "step": 457280
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 2.256371010943598e-05,
+      "loss": 0.8196,
+      "step": 457290
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 2.2562562179536235e-05,
+      "loss": 0.8434,
+      "step": 457300
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 2.2561414249636485e-05,
+      "loss": 0.8219,
+      "step": 457310
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 2.256026631973674e-05,
+      "loss": 0.7552,
+      "step": 457320
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 2.255911838983699e-05,
+      "loss": 0.7857,
+      "step": 457330
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 2.2557970459937245e-05,
+      "loss": 0.7776,
+      "step": 457340
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 2.2556822530037495e-05,
+      "loss": 0.7474,
+      "step": 457350
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 2.255567460013775e-05,
+      "loss": 0.8779,
+      "step": 457360
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 2.2554526670238e-05,
+      "loss": 0.8552,
+      "step": 457370
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 2.2553378740338254e-05,
+      "loss": 0.9072,
+      "step": 457380
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 2.2552230810438508e-05,
+      "loss": 0.8304,
+      "step": 457390
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 2.255108288053876e-05,
+      "loss": 0.7992,
+      "step": 457400
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 2.254993495063901e-05,
+      "loss": 0.8725,
+      "step": 457410
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 2.2548787020739267e-05,
+      "loss": 0.7985,
+      "step": 457420
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 2.2547639090839517e-05,
+      "loss": 0.787,
+      "step": 457430
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 2.2546491160939767e-05,
+      "loss": 0.7957,
+      "step": 457440
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 2.2545343231040024e-05,
+      "loss": 0.7158,
+      "step": 457450
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 2.2544195301140274e-05,
+      "loss": 0.8201,
+      "step": 457460
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 2.2543047371240527e-05,
+      "loss": 0.7852,
+      "step": 457470
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 2.254189944134078e-05,
+      "loss": 0.8651,
+      "step": 457480
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 2.2540751511441033e-05,
+      "loss": 0.8032,
+      "step": 457490
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 2.2539603581541283e-05,
+      "loss": 0.7727,
+      "step": 457500
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 2.253845565164154e-05,
+      "loss": 0.8058,
+      "step": 457510
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 2.253730772174179e-05,
+      "loss": 0.8107,
+      "step": 457520
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 2.2536159791842043e-05,
+      "loss": 0.8072,
+      "step": 457530
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 2.2535011861942296e-05,
+      "loss": 0.7761,
+      "step": 457540
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 2.253386393204255e-05,
+      "loss": 0.7684,
+      "step": 457550
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 2.25327160021428e-05,
+      "loss": 0.8115,
+      "step": 457560
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 2.2531568072243056e-05,
+      "loss": 0.8533,
+      "step": 457570
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 2.2530420142343306e-05,
+      "loss": 0.8076,
+      "step": 457580
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 2.2529272212443556e-05,
+      "loss": 0.7109,
+      "step": 457590
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 2.2528124282543812e-05,
+      "loss": 0.9426,
+      "step": 457600
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 2.2526976352644062e-05,
+      "loss": 0.9082,
+      "step": 457610
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 2.2525828422744315e-05,
+      "loss": 0.8785,
+      "step": 457620
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 2.252468049284457e-05,
+      "loss": 0.7535,
+      "step": 457630
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 2.2523532562944822e-05,
+      "loss": 0.8056,
+      "step": 457640
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 2.252238463304507e-05,
+      "loss": 0.8885,
+      "step": 457650
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 2.2521236703145328e-05,
+      "loss": 0.7356,
+      "step": 457660
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 2.2520088773245578e-05,
+      "loss": 0.7866,
+      "step": 457670
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 2.251894084334583e-05,
+      "loss": 0.7023,
+      "step": 457680
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 2.2517792913446085e-05,
+      "loss": 0.744,
+      "step": 457690
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 2.2516644983546338e-05,
+      "loss": 0.771,
+      "step": 457700
+    },
+    {
+      "epoch": 21.0,
+      "eval_cer": 0.6688024149428105,
+      "eval_em": 0.007357212003872217,
+      "eval_f1": 0.007357212003872217,
+      "eval_loss": 1.3091018199920654,
+      "eval_runtime": 1588.2241,
+      "eval_samples_per_second": 6.504,
+      "eval_steps_per_second": 0.813,
+      "eval_wer": 0.9926427879961278,
+      "step": 457700
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 2.2515497053646588e-05,
+      "loss": 0.8266,
+      "step": 457710
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 2.2514349123746844e-05,
+      "loss": 0.6792,
+      "step": 457720
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 2.2513201193847094e-05,
+      "loss": 0.8145,
+      "step": 457730
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 2.2512053263947344e-05,
+      "loss": 0.7785,
+      "step": 457740
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 2.25109053340476e-05,
+      "loss": 0.846,
+      "step": 457750
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 2.250975740414785e-05,
+      "loss": 0.7789,
+      "step": 457760
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 2.2508609474248104e-05,
+      "loss": 0.7227,
+      "step": 457770
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 2.2507461544348357e-05,
+      "loss": 0.7856,
+      "step": 457780
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 2.250631361444861e-05,
+      "loss": 0.8179,
+      "step": 457790
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 2.250516568454886e-05,
+      "loss": 0.7723,
+      "step": 457800
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 2.2504017754649117e-05,
+      "loss": 0.756,
+      "step": 457810
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 2.2502869824749367e-05,
+      "loss": 0.7561,
+      "step": 457820
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 2.250172189484962e-05,
+      "loss": 0.7797,
+      "step": 457830
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 2.2500573964949873e-05,
+      "loss": 0.7632,
+      "step": 457840
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 2.2499426035050126e-05,
+      "loss": 0.7595,
+      "step": 457850
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 2.2498278105150376e-05,
+      "loss": 0.8956,
+      "step": 457860
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 2.2497130175250633e-05,
+      "loss": 0.7357,
+      "step": 457870
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 2.2495982245350883e-05,
+      "loss": 0.7889,
+      "step": 457880
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 2.2494834315451133e-05,
+      "loss": 0.7648,
+      "step": 457890
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 2.249368638555139e-05,
+      "loss": 0.7871,
+      "step": 457900
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 2.249253845565164e-05,
+      "loss": 0.7213,
+      "step": 457910
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 2.2491390525751892e-05,
+      "loss": 0.7996,
+      "step": 457920
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 2.2490242595852142e-05,
+      "loss": 0.7706,
+      "step": 457930
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 2.24890946659524e-05,
+      "loss": 0.7732,
+      "step": 457940
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 2.248794673605265e-05,
+      "loss": 0.726,
+      "step": 457950
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 2.2486798806152902e-05,
+      "loss": 0.7882,
+      "step": 457960
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 2.2485650876253155e-05,
+      "loss": 0.8088,
+      "step": 457970
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 2.248450294635341e-05,
+      "loss": 0.7618,
+      "step": 457980
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 2.2483355016453658e-05,
+      "loss": 0.7896,
+      "step": 457990
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 2.2482207086553915e-05,
+      "loss": 0.7226,
+      "step": 458000
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 2.2481059156654165e-05,
+      "loss": 0.7033,
+      "step": 458010
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 2.2479911226754415e-05,
+      "loss": 0.7831,
+      "step": 458020
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 2.247876329685467e-05,
+      "loss": 0.7812,
+      "step": 458030
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 2.247761536695492e-05,
+      "loss": 0.7254,
+      "step": 458040
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 2.2476467437055174e-05,
+      "loss": 0.7632,
+      "step": 458050
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 2.2475319507155427e-05,
+      "loss": 0.7768,
+      "step": 458060
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 2.247417157725568e-05,
+      "loss": 0.7815,
+      "step": 458070
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 2.247302364735593e-05,
+      "loss": 0.6735,
+      "step": 458080
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 2.2471875717456187e-05,
+      "loss": 0.8947,
+      "step": 458090
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 2.2470727787556437e-05,
+      "loss": 0.7589,
+      "step": 458100
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 2.246957985765669e-05,
+      "loss": 0.7532,
+      "step": 458110
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 2.2468431927756944e-05,
+      "loss": 0.7382,
+      "step": 458120
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 2.2467283997857197e-05,
+      "loss": 0.8166,
+      "step": 458130
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 2.2466136067957447e-05,
+      "loss": 0.6918,
+      "step": 458140
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 2.2464988138057703e-05,
+      "loss": 0.786,
+      "step": 458150
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 2.2463840208157953e-05,
+      "loss": 0.7943,
+      "step": 458160
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 2.2462692278258203e-05,
+      "loss": 0.6973,
+      "step": 458170
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 2.246154434835846e-05,
+      "loss": 0.806,
+      "step": 458180
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 2.246039641845871e-05,
+      "loss": 0.8326,
+      "step": 458190
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 2.2459248488558963e-05,
+      "loss": 0.8015,
+      "step": 458200
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 2.2458100558659216e-05,
+      "loss": 0.7437,
+      "step": 458210
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 2.245695262875947e-05,
+      "loss": 0.73,
+      "step": 458220
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 2.245580469885972e-05,
+      "loss": 0.7543,
+      "step": 458230
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 2.2454656768959976e-05,
+      "loss": 0.802,
+      "step": 458240
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 2.2453508839060226e-05,
+      "loss": 0.7282,
+      "step": 458250
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 2.245236090916048e-05,
+      "loss": 0.8016,
+      "step": 458260
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 2.2451212979260732e-05,
+      "loss": 0.8183,
+      "step": 458270
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 2.2450065049360985e-05,
+      "loss": 0.7851,
+      "step": 458280
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 2.2448917119461235e-05,
+      "loss": 0.7744,
+      "step": 458290
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 2.2447769189561492e-05,
+      "loss": 0.7222,
+      "step": 458300
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 2.244662125966174e-05,
+      "loss": 0.6676,
+      "step": 458310
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 2.244547332976199e-05,
+      "loss": 0.7147,
+      "step": 458320
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 2.2444325399862248e-05,
+      "loss": 0.7705,
+      "step": 458330
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 2.2443177469962498e-05,
+      "loss": 0.7824,
+      "step": 458340
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 2.244202954006275e-05,
+      "loss": 0.7397,
+      "step": 458350
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 2.2440881610163004e-05,
+      "loss": 0.7386,
+      "step": 458360
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 2.2439733680263258e-05,
+      "loss": 0.761,
+      "step": 458370
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 2.2438585750363508e-05,
+      "loss": 0.7591,
+      "step": 458380
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 2.2437437820463764e-05,
+      "loss": 0.8544,
+      "step": 458390
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 2.2436289890564014e-05,
+      "loss": 0.7775,
+      "step": 458400
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 2.2435141960664267e-05,
+      "loss": 0.7955,
+      "step": 458410
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 2.243399403076452e-05,
+      "loss": 0.8299,
+      "step": 458420
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 2.2432846100864774e-05,
+      "loss": 0.8339,
+      "step": 458430
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 2.2431698170965024e-05,
+      "loss": 0.7329,
+      "step": 458440
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 2.243055024106528e-05,
+      "loss": 0.751,
+      "step": 458450
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 2.242940231116553e-05,
+      "loss": 0.7433,
+      "step": 458460
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 2.242825438126578e-05,
+      "loss": 0.8,
+      "step": 458470
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 2.2427106451366037e-05,
+      "loss": 0.7441,
+      "step": 458480
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 2.2425958521466286e-05,
+      "loss": 0.7473,
+      "step": 458490
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 2.242481059156654e-05,
+      "loss": 0.8331,
+      "step": 458500
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 2.2423662661666793e-05,
+      "loss": 0.8142,
+      "step": 458510
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 2.2422514731767046e-05,
+      "loss": 0.7498,
+      "step": 458520
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 2.2421366801867296e-05,
+      "loss": 0.8241,
+      "step": 458530
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 2.2420218871967553e-05,
+      "loss": 0.7911,
+      "step": 458540
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 2.2419070942067802e-05,
+      "loss": 0.8106,
+      "step": 458550
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 2.2417923012168056e-05,
+      "loss": 0.7351,
+      "step": 458560
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 2.241677508226831e-05,
+      "loss": 0.8126,
+      "step": 458570
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 2.2415627152368562e-05,
+      "loss": 0.8319,
+      "step": 458580
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 2.2414479222468812e-05,
+      "loss": 0.7781,
+      "step": 458590
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 2.2413331292569062e-05,
+      "loss": 0.801,
+      "step": 458600
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 2.241218336266932e-05,
+      "loss": 0.8008,
+      "step": 458610
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 2.241103543276957e-05,
+      "loss": 0.7553,
+      "step": 458620
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 2.240988750286982e-05,
+      "loss": 0.7327,
+      "step": 458630
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 2.2408739572970075e-05,
+      "loss": 0.7671,
+      "step": 458640
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 2.2407591643070328e-05,
+      "loss": 0.7771,
+      "step": 458650
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 2.2406443713170578e-05,
+      "loss": 0.7564,
+      "step": 458660
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 2.2405295783270835e-05,
+      "loss": 0.7958,
+      "step": 458670
+    },
+    {
+      "epoch": 21.04,
+      "learning_rate": 2.2404147853371084e-05,
+      "loss": 0.8209,
+      "step": 458680
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 2.2402999923471338e-05,
+      "loss": 0.8383,
+      "step": 458690
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 2.240185199357159e-05,
+      "loss": 0.7862,
+      "step": 458700
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 2.2400704063671844e-05,
+      "loss": 0.7257,
+      "step": 458710
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 2.2399556133772094e-05,
+      "loss": 0.7781,
+      "step": 458720
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 2.239840820387235e-05,
+      "loss": 0.7233,
+      "step": 458730
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 2.23972602739726e-05,
+      "loss": 0.8082,
+      "step": 458740
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 2.239611234407285e-05,
+      "loss": 0.7384,
+      "step": 458750
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 2.2394964414173107e-05,
+      "loss": 0.7549,
+      "step": 458760
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 2.2393816484273357e-05,
+      "loss": 0.7664,
+      "step": 458770
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 2.239266855437361e-05,
+      "loss": 0.8608,
+      "step": 458780
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 2.2391520624473863e-05,
+      "loss": 0.8,
+      "step": 458790
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 2.2390372694574117e-05,
+      "loss": 0.8719,
+      "step": 458800
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 2.2389224764674366e-05,
+      "loss": 0.7301,
+      "step": 458810
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 2.2388076834774623e-05,
+      "loss": 0.7465,
+      "step": 458820
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 2.2386928904874873e-05,
+      "loss": 0.732,
+      "step": 458830
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 2.2385780974975126e-05,
+      "loss": 0.8022,
+      "step": 458840
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 2.238463304507538e-05,
+      "loss": 0.7086,
+      "step": 458850
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 2.2383485115175633e-05,
+      "loss": 0.7558,
+      "step": 458860
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 2.2382337185275883e-05,
+      "loss": 0.7601,
+      "step": 458870
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 2.238118925537614e-05,
+      "loss": 0.8224,
+      "step": 458880
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 2.238004132547639e-05,
+      "loss": 0.8296,
+      "step": 458890
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 2.237889339557664e-05,
+      "loss": 0.709,
+      "step": 458900
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 2.2377745465676896e-05,
+      "loss": 0.8224,
+      "step": 458910
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 2.2376597535777145e-05,
+      "loss": 0.7682,
+      "step": 458920
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 2.23754496058774e-05,
+      "loss": 0.7926,
+      "step": 458930
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 2.2374301675977652e-05,
+      "loss": 0.7347,
+      "step": 458940
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 2.2373153746077905e-05,
+      "loss": 0.7051,
+      "step": 458950
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 2.2372005816178155e-05,
+      "loss": 0.7976,
+      "step": 458960
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 2.237085788627841e-05,
+      "loss": 0.7617,
+      "step": 458970
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 2.236970995637866e-05,
+      "loss": 0.751,
+      "step": 458980
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 2.2368562026478915e-05,
+      "loss": 0.7923,
+      "step": 458990
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 2.2367414096579168e-05,
+      "loss": 0.7445,
+      "step": 459000
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 2.236626616667942e-05,
+      "loss": 0.6955,
+      "step": 459010
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 2.236511823677967e-05,
+      "loss": 0.7583,
+      "step": 459020
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 2.2363970306879928e-05,
+      "loss": 0.7379,
+      "step": 459030
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 2.2362822376980177e-05,
+      "loss": 0.8158,
+      "step": 459040
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 2.2361674447080427e-05,
+      "loss": 0.8026,
+      "step": 459050
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 2.2360526517180684e-05,
+      "loss": 0.8286,
+      "step": 459060
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 2.2359378587280934e-05,
+      "loss": 0.8243,
+      "step": 459070
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 2.2358230657381187e-05,
+      "loss": 0.8169,
+      "step": 459080
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 2.235708272748144e-05,
+      "loss": 0.7795,
+      "step": 459090
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 2.2355934797581694e-05,
+      "loss": 0.8431,
+      "step": 459100
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 2.2354786867681943e-05,
+      "loss": 0.8853,
+      "step": 459110
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 2.23536389377822e-05,
+      "loss": 0.7254,
+      "step": 459120
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 2.235249100788245e-05,
+      "loss": 0.7285,
+      "step": 459130
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 2.2351343077982703e-05,
+      "loss": 0.8778,
+      "step": 459140
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 2.2350195148082956e-05,
+      "loss": 0.8362,
+      "step": 459150
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 2.234904721818321e-05,
+      "loss": 0.7838,
+      "step": 459160
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 2.234789928828346e-05,
+      "loss": 0.8419,
+      "step": 459170
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 2.2346751358383716e-05,
+      "loss": 0.8413,
+      "step": 459180
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 2.2345603428483966e-05,
+      "loss": 0.6963,
+      "step": 459190
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 2.2344455498584216e-05,
+      "loss": 0.8104,
+      "step": 459200
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 2.2343307568684472e-05,
+      "loss": 0.7084,
+      "step": 459210
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 2.2342159638784722e-05,
+      "loss": 0.7658,
+      "step": 459220
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 2.2341011708884976e-05,
+      "loss": 0.7253,
+      "step": 459230
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 2.233986377898523e-05,
+      "loss": 0.7611,
+      "step": 459240
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 2.2338715849085482e-05,
+      "loss": 0.7898,
+      "step": 459250
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 2.2337567919185732e-05,
+      "loss": 0.7679,
+      "step": 459260
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 2.233641998928599e-05,
+      "loss": 0.7852,
+      "step": 459270
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 2.233527205938624e-05,
+      "loss": 0.8216,
+      "step": 459280
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 2.233412412948649e-05,
+      "loss": 0.7078,
+      "step": 459290
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 2.233297619958674e-05,
+      "loss": 0.7143,
+      "step": 459300
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 2.2331828269686998e-05,
+      "loss": 0.8001,
+      "step": 459310
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 2.2330680339787248e-05,
+      "loss": 0.7695,
+      "step": 459320
+    },
+    {
+      "epoch": 21.07,
+      "learning_rate": 2.2329532409887498e-05,
+      "loss": 0.7688,
+      "step": 459330
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 2.2328384479987754e-05,
+      "loss": 0.7977,
+      "step": 459340
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 2.2327236550088004e-05,
+      "loss": 0.7842,
+      "step": 459350
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 2.2326088620188258e-05,
+      "loss": 0.748,
+      "step": 459360
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 2.232494069028851e-05,
+      "loss": 0.8388,
+      "step": 459370
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 2.2323792760388764e-05,
+      "loss": 0.761,
+      "step": 459380
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 2.2322644830489014e-05,
+      "loss": 0.8261,
+      "step": 459390
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 2.232149690058927e-05,
+      "loss": 0.7651,
+      "step": 459400
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 2.232034897068952e-05,
+      "loss": 0.741,
+      "step": 459410
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 2.2319201040789774e-05,
+      "loss": 0.7731,
+      "step": 459420
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 2.2318053110890027e-05,
+      "loss": 0.7731,
+      "step": 459430
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 2.231690518099028e-05,
+      "loss": 0.7187,
+      "step": 459440
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 2.231575725109053e-05,
+      "loss": 0.7501,
+      "step": 459450
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 2.2314609321190787e-05,
+      "loss": 0.8176,
+      "step": 459460
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 2.2313461391291036e-05,
+      "loss": 0.7771,
+      "step": 459470
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 2.2312313461391286e-05,
+      "loss": 0.786,
+      "step": 459480
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 2.2311165531491543e-05,
+      "loss": 0.6862,
+      "step": 459490
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 2.2310017601591793e-05,
+      "loss": 0.8194,
+      "step": 459500
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 2.2308869671692046e-05,
+      "loss": 0.7954,
+      "step": 459510
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 2.23077217417923e-05,
+      "loss": 0.8076,
+      "step": 459520
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 2.2306573811892552e-05,
+      "loss": 0.7815,
+      "step": 459530
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 2.2305425881992802e-05,
+      "loss": 0.6919,
+      "step": 459540
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 2.230427795209306e-05,
+      "loss": 0.7549,
+      "step": 459550
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 2.230313002219331e-05,
+      "loss": 0.8005,
+      "step": 459560
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 2.2301982092293562e-05,
+      "loss": 0.8301,
+      "step": 459570
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 2.2300834162393815e-05,
+      "loss": 0.743,
+      "step": 459580
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 2.229968623249407e-05,
+      "loss": 0.7708,
+      "step": 459590
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 2.229853830259432e-05,
+      "loss": 0.7748,
+      "step": 459600
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 2.2297390372694575e-05,
+      "loss": 0.7659,
+      "step": 459610
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 2.2296242442794825e-05,
+      "loss": 0.7933,
+      "step": 459620
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 2.2295094512895075e-05,
+      "loss": 0.7249,
+      "step": 459630
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 2.229394658299533e-05,
+      "loss": 0.8531,
+      "step": 459640
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 2.229279865309558e-05,
+      "loss": 0.8488,
+      "step": 459650
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 2.2291650723195834e-05,
+      "loss": 0.7214,
+      "step": 459660
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 2.2290502793296088e-05,
+      "loss": 0.7203,
+      "step": 459670
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 2.228935486339634e-05,
+      "loss": 0.7285,
+      "step": 459680
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 2.228820693349659e-05,
+      "loss": 0.8146,
+      "step": 459690
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 2.2287059003596847e-05,
+      "loss": 0.7977,
+      "step": 459700
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 2.2285911073697097e-05,
+      "loss": 0.7874,
+      "step": 459710
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 2.228476314379735e-05,
+      "loss": 0.7935,
+      "step": 459720
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 2.2283615213897604e-05,
+      "loss": 0.7654,
+      "step": 459730
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 2.2282467283997857e-05,
+      "loss": 0.8117,
+      "step": 459740
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 2.2281319354098107e-05,
+      "loss": 0.7501,
+      "step": 459750
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 2.2280171424198364e-05,
+      "loss": 0.8217,
+      "step": 459760
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 2.2279023494298613e-05,
+      "loss": 0.8656,
+      "step": 459770
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 2.2277875564398863e-05,
+      "loss": 0.6855,
+      "step": 459780
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 2.227672763449912e-05,
+      "loss": 0.8079,
+      "step": 459790
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 2.227557970459937e-05,
+      "loss": 0.8085,
+      "step": 459800
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 2.2274431774699623e-05,
+      "loss": 0.7796,
+      "step": 459810
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 2.2273283844799876e-05,
+      "loss": 0.8177,
+      "step": 459820
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 2.227213591490013e-05,
+      "loss": 0.7538,
+      "step": 459830
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 2.227098798500038e-05,
+      "loss": 0.6964,
+      "step": 459840
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 2.2269840055100636e-05,
+      "loss": 0.7413,
+      "step": 459850
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 2.2268692125200886e-05,
+      "loss": 0.7799,
+      "step": 459860
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 2.226754419530114e-05,
+      "loss": 0.841,
+      "step": 459870
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 2.2266396265401392e-05,
+      "loss": 0.8246,
+      "step": 459880
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 2.2265248335501645e-05,
+      "loss": 0.7979,
+      "step": 459890
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 2.2264100405601895e-05,
+      "loss": 0.8545,
+      "step": 459900
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 2.2262952475702152e-05,
+      "loss": 0.7486,
+      "step": 459910
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 2.2261804545802402e-05,
+      "loss": 0.7889,
+      "step": 459920
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 2.226065661590265e-05,
+      "loss": 0.6515,
+      "step": 459930
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 2.225950868600291e-05,
+      "loss": 0.7874,
+      "step": 459940
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 2.2258360756103158e-05,
+      "loss": 0.7784,
+      "step": 459950
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 2.225721282620341e-05,
+      "loss": 0.7875,
+      "step": 459960
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 2.2256064896303665e-05,
+      "loss": 0.7532,
+      "step": 459970
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 2.2254916966403918e-05,
+      "loss": 0.8867,
+      "step": 459980
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 2.2253769036504168e-05,
+      "loss": 0.7969,
+      "step": 459990
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 2.225262110660442e-05,
+      "loss": 0.7981,
+      "step": 460000
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 2.2251473176704674e-05,
+      "loss": 0.7119,
+      "step": 460010
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 2.2250325246804927e-05,
+      "loss": 0.7484,
+      "step": 460020
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 2.2249177316905177e-05,
+      "loss": 0.7133,
+      "step": 460030
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 2.2248029387005434e-05,
+      "loss": 0.8336,
+      "step": 460040
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 2.2246881457105684e-05,
+      "loss": 0.7301,
+      "step": 460050
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 2.2245733527205934e-05,
+      "loss": 0.7576,
+      "step": 460060
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 2.224458559730619e-05,
+      "loss": 0.9276,
+      "step": 460070
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 2.224343766740644e-05,
+      "loss": 0.8137,
+      "step": 460080
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 2.2242289737506693e-05,
+      "loss": 0.8253,
+      "step": 460090
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 2.2241141807606947e-05,
+      "loss": 0.7589,
+      "step": 460100
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 2.22399938777072e-05,
+      "loss": 0.7472,
+      "step": 460110
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 2.223884594780745e-05,
+      "loss": 0.7699,
+      "step": 460120
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 2.2237698017907706e-05,
+      "loss": 0.8102,
+      "step": 460130
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 2.2236550088007956e-05,
+      "loss": 0.7308,
+      "step": 460140
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 2.223540215810821e-05,
+      "loss": 0.7904,
+      "step": 460150
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 2.2234254228208463e-05,
+      "loss": 0.7869,
+      "step": 460160
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 2.2233106298308716e-05,
+      "loss": 0.6826,
+      "step": 460170
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 2.2231958368408966e-05,
+      "loss": 0.8551,
+      "step": 460180
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 2.2230810438509222e-05,
+      "loss": 0.783,
+      "step": 460190
+    },
+    {
+      "epoch": 21.11,
+      "learning_rate": 2.2229662508609472e-05,
+      "loss": 0.8602,
+      "step": 460200
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 2.2228514578709722e-05,
+      "loss": 0.8266,
+      "step": 460210
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 2.222736664880998e-05,
+      "loss": 0.7919,
+      "step": 460220
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 2.222621871891023e-05,
+      "loss": 0.8522,
+      "step": 460230
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 2.2225070789010482e-05,
+      "loss": 0.785,
+      "step": 460240
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 2.2223922859110735e-05,
+      "loss": 0.7623,
+      "step": 460250
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 2.222277492921099e-05,
+      "loss": 0.8318,
+      "step": 460260
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 2.2221626999311238e-05,
+      "loss": 0.665,
+      "step": 460270
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 2.2220479069411495e-05,
+      "loss": 0.7845,
+      "step": 460280
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 2.2219331139511745e-05,
+      "loss": 0.7925,
+      "step": 460290
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 2.2218183209611998e-05,
+      "loss": 0.8073,
+      "step": 460300
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 2.221703527971225e-05,
+      "loss": 0.7353,
+      "step": 460310
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 2.2215887349812504e-05,
+      "loss": 0.7645,
+      "step": 460320
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 2.2214739419912754e-05,
+      "loss": 0.7777,
+      "step": 460330
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 2.221359149001301e-05,
+      "loss": 0.7687,
+      "step": 460340
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 2.221244356011326e-05,
+      "loss": 0.7079,
+      "step": 460350
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 2.221129563021351e-05,
+      "loss": 0.7812,
+      "step": 460360
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 2.2210147700313767e-05,
+      "loss": 0.7873,
+      "step": 460370
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 2.2208999770414017e-05,
+      "loss": 0.7994,
+      "step": 460380
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 2.220785184051427e-05,
+      "loss": 0.7385,
+      "step": 460390
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 2.2206703910614524e-05,
+      "loss": 0.8232,
+      "step": 460400
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 2.2205555980714777e-05,
+      "loss": 0.747,
+      "step": 460410
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 2.2204408050815027e-05,
+      "loss": 0.7347,
+      "step": 460420
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 2.2203260120915283e-05,
+      "loss": 0.7354,
+      "step": 460430
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 2.2202112191015533e-05,
+      "loss": 0.8169,
+      "step": 460440
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 2.2200964261115786e-05,
+      "loss": 0.8407,
+      "step": 460450
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 2.219981633121604e-05,
+      "loss": 0.7727,
+      "step": 460460
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 2.2198668401316293e-05,
+      "loss": 0.7341,
+      "step": 460470
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 2.2197520471416543e-05,
+      "loss": 0.8415,
+      "step": 460480
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 2.21963725415168e-05,
+      "loss": 0.7674,
+      "step": 460490
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 2.219522461161705e-05,
+      "loss": 0.7718,
+      "step": 460500
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 2.21940766817173e-05,
+      "loss": 0.7331,
+      "step": 460510
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 2.2192928751817556e-05,
+      "loss": 0.7888,
+      "step": 460520
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 2.2191780821917806e-05,
+      "loss": 0.721,
+      "step": 460530
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 2.219063289201806e-05,
+      "loss": 0.8083,
+      "step": 460540
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 2.2189484962118312e-05,
+      "loss": 0.7511,
+      "step": 460550
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 2.2188337032218565e-05,
+      "loss": 0.7964,
+      "step": 460560
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 2.2187189102318815e-05,
+      "loss": 0.7448,
+      "step": 460570
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 2.2186041172419072e-05,
+      "loss": 0.6672,
+      "step": 460580
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 2.218489324251932e-05,
+      "loss": 0.8163,
+      "step": 460590
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 2.2183745312619575e-05,
+      "loss": 0.7609,
+      "step": 460600
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 2.2182597382719828e-05,
+      "loss": 0.7786,
+      "step": 460610
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 2.218144945282008e-05,
+      "loss": 0.7703,
+      "step": 460620
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 2.218030152292033e-05,
+      "loss": 0.7823,
+      "step": 460630
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 2.2179153593020588e-05,
+      "loss": 0.7399,
+      "step": 460640
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 2.2178005663120838e-05,
+      "loss": 0.7508,
+      "step": 460650
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 2.2176857733221088e-05,
+      "loss": 0.7591,
+      "step": 460660
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 2.217570980332134e-05,
+      "loss": 0.8137,
+      "step": 460670
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 2.2174561873421594e-05,
+      "loss": 0.6987,
+      "step": 460680
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 2.2173413943521847e-05,
+      "loss": 0.7224,
+      "step": 460690
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 2.2172266013622097e-05,
+      "loss": 0.7837,
+      "step": 460700
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 2.2171118083722354e-05,
+      "loss": 0.7155,
+      "step": 460710
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 2.2169970153822604e-05,
+      "loss": 0.8668,
+      "step": 460720
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 2.2168822223922857e-05,
+      "loss": 0.7435,
+      "step": 460730
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 2.216767429402311e-05,
+      "loss": 0.705,
+      "step": 460740
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 2.2166526364123363e-05,
+      "loss": 0.6232,
+      "step": 460750
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 2.2165378434223613e-05,
+      "loss": 0.6753,
+      "step": 460760
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 2.216423050432387e-05,
+      "loss": 0.7004,
+      "step": 460770
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 2.216308257442412e-05,
+      "loss": 0.7695,
+      "step": 460780
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 2.216193464452437e-05,
+      "loss": 0.8701,
+      "step": 460790
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 2.2160786714624626e-05,
+      "loss": 0.7383,
+      "step": 460800
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 2.2159638784724876e-05,
+      "loss": 0.7087,
+      "step": 460810
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 2.215849085482513e-05,
+      "loss": 0.8134,
+      "step": 460820
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 2.2157342924925383e-05,
+      "loss": 0.7322,
+      "step": 460830
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 2.2156194995025636e-05,
+      "loss": 0.7959,
+      "step": 460840
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 2.2155047065125886e-05,
+      "loss": 0.7385,
+      "step": 460850
+    },
+    {
+      "epoch": 21.14,
+      "learning_rate": 2.2153899135226142e-05,
+      "loss": 0.7649,
+      "step": 460860
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 2.2152751205326392e-05,
+      "loss": 0.6772,
+      "step": 460870
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 2.2151603275426645e-05,
+      "loss": 0.7686,
+      "step": 460880
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 2.21504553455269e-05,
+      "loss": 0.8022,
+      "step": 460890
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 2.2149307415627152e-05,
+      "loss": 0.7788,
+      "step": 460900
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 2.21481594857274e-05,
+      "loss": 0.7262,
+      "step": 460910
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 2.214701155582766e-05,
+      "loss": 0.7754,
+      "step": 460920
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 2.2145863625927908e-05,
+      "loss": 0.757,
+      "step": 460930
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 2.2144715696028158e-05,
+      "loss": 0.776,
+      "step": 460940
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 2.2143567766128415e-05,
+      "loss": 0.8421,
+      "step": 460950
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 2.2142419836228665e-05,
+      "loss": 0.7449,
+      "step": 460960
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 2.2141271906328918e-05,
+      "loss": 0.7316,
+      "step": 460970
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 2.214012397642917e-05,
+      "loss": 0.7982,
+      "step": 460980
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 2.2138976046529424e-05,
+      "loss": 0.7705,
+      "step": 460990
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 2.2137828116629674e-05,
+      "loss": 0.7752,
+      "step": 461000
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 2.213668018672993e-05,
+      "loss": 0.7851,
+      "step": 461010
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 2.213553225683018e-05,
+      "loss": 0.787,
+      "step": 461020
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 2.2134384326930434e-05,
+      "loss": 0.7966,
+      "step": 461030
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 2.2133236397030687e-05,
+      "loss": 0.7247,
+      "step": 461040
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 2.213208846713094e-05,
+      "loss": 0.7785,
+      "step": 461050
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 2.213094053723119e-05,
+      "loss": 0.7371,
+      "step": 461060
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 2.2129792607331447e-05,
+      "loss": 0.7425,
+      "step": 461070
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 2.2128644677431697e-05,
+      "loss": 0.8169,
+      "step": 461080
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 2.2127496747531946e-05,
+      "loss": 0.7555,
+      "step": 461090
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 2.2126348817632203e-05,
+      "loss": 0.691,
+      "step": 461100
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 2.2125200887732453e-05,
+      "loss": 0.7464,
+      "step": 461110
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 2.2124052957832706e-05,
+      "loss": 0.8433,
+      "step": 461120
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 2.212290502793296e-05,
+      "loss": 0.7825,
+      "step": 461130
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 2.2121757098033213e-05,
+      "loss": 0.6972,
+      "step": 461140
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 2.2120609168133463e-05,
+      "loss": 0.7885,
+      "step": 461150
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 2.211946123823372e-05,
+      "loss": 0.7497,
+      "step": 461160
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 2.211831330833397e-05,
+      "loss": 0.7821,
+      "step": 461170
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 2.2117165378434222e-05,
+      "loss": 0.7652,
+      "step": 461180
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 2.2116017448534476e-05,
+      "loss": 0.797,
+      "step": 461190
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 2.211486951863473e-05,
+      "loss": 0.7977,
+      "step": 461200
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 2.211372158873498e-05,
+      "loss": 0.7059,
+      "step": 461210
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 2.2112573658835235e-05,
+      "loss": 0.7734,
+      "step": 461220
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 2.2111425728935485e-05,
+      "loss": 0.8302,
+      "step": 461230
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 2.2110277799035735e-05,
+      "loss": 0.7496,
+      "step": 461240
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 2.210912986913599e-05,
+      "loss": 0.7357,
+      "step": 461250
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 2.210798193923624e-05,
+      "loss": 0.7978,
+      "step": 461260
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 2.2106834009336495e-05,
+      "loss": 0.795,
+      "step": 461270
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 2.2105686079436748e-05,
+      "loss": 0.7206,
+      "step": 461280
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 2.2104538149537e-05,
+      "loss": 0.8009,
+      "step": 461290
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 2.210339021963725e-05,
+      "loss": 0.7044,
+      "step": 461300
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 2.2102242289737508e-05,
+      "loss": 0.7854,
+      "step": 461310
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 2.2101094359837758e-05,
+      "loss": 0.7704,
+      "step": 461320
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 2.209994642993801e-05,
+      "loss": 0.8031,
+      "step": 461330
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 2.2098798500038264e-05,
+      "loss": 0.7311,
+      "step": 461340
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 2.2097650570138517e-05,
+      "loss": 0.7912,
+      "step": 461350
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 2.2096502640238767e-05,
+      "loss": 0.7579,
+      "step": 461360
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 2.2095354710339017e-05,
+      "loss": 0.7432,
+      "step": 461370
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 2.2094206780439274e-05,
+      "loss": 0.8556,
+      "step": 461380
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 2.2093058850539523e-05,
+      "loss": 0.8293,
+      "step": 461390
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 2.2091910920639777e-05,
+      "loss": 0.753,
+      "step": 461400
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 2.209076299074003e-05,
+      "loss": 0.7531,
+      "step": 461410
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 2.2089615060840283e-05,
+      "loss": 0.8001,
+      "step": 461420
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 2.2088467130940533e-05,
+      "loss": 0.8183,
+      "step": 461430
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 2.208731920104079e-05,
+      "loss": 0.7232,
+      "step": 461440
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 2.208617127114104e-05,
+      "loss": 0.7288,
+      "step": 461450
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 2.2085023341241293e-05,
+      "loss": 0.7634,
+      "step": 461460
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 2.2083875411341546e-05,
+      "loss": 0.8151,
+      "step": 461470
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 2.20827274814418e-05,
+      "loss": 0.8125,
+      "step": 461480
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 2.208157955154205e-05,
+      "loss": 0.7182,
+      "step": 461490
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 2.2080431621642306e-05,
+      "loss": 0.8489,
+      "step": 461500
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 2.2079283691742556e-05,
+      "loss": 0.8207,
+      "step": 461510
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 2.2078135761842805e-05,
+      "loss": 0.8033,
+      "step": 461520
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 2.2076987831943062e-05,
+      "loss": 0.7407,
+      "step": 461530
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 2.2075839902043312e-05,
+      "loss": 0.8064,
+      "step": 461540
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 2.2074691972143565e-05,
+      "loss": 0.7491,
+      "step": 461550
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 2.207354404224382e-05,
+      "loss": 0.7655,
+      "step": 461560
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 2.207239611234407e-05,
+      "loss": 0.7757,
+      "step": 461570
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 2.207124818244432e-05,
+      "loss": 0.7839,
+      "step": 461580
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 2.2070100252544578e-05,
+      "loss": 0.789,
+      "step": 461590
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 2.2068952322644828e-05,
+      "loss": 0.7406,
+      "step": 461600
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 2.206780439274508e-05,
+      "loss": 0.8182,
+      "step": 461610
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 2.2066656462845334e-05,
+      "loss": 0.8047,
+      "step": 461620
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 2.2065508532945588e-05,
+      "loss": 0.7138,
+      "step": 461630
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 2.2064360603045838e-05,
+      "loss": 0.7647,
+      "step": 461640
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 2.2063212673146094e-05,
+      "loss": 0.7607,
+      "step": 461650
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 2.2062064743246344e-05,
+      "loss": 0.8114,
+      "step": 461660
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 2.2060916813346594e-05,
+      "loss": 0.7915,
+      "step": 461670
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 2.205976888344685e-05,
+      "loss": 0.6774,
+      "step": 461680
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 2.20586209535471e-05,
+      "loss": 0.7759,
+      "step": 461690
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 2.2057473023647354e-05,
+      "loss": 0.7783,
+      "step": 461700
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 2.2056325093747607e-05,
+      "loss": 0.821,
+      "step": 461710
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 2.205517716384786e-05,
+      "loss": 0.8137,
+      "step": 461720
+    },
+    {
+      "epoch": 21.18,
+      "learning_rate": 2.205402923394811e-05,
+      "loss": 0.736,
+      "step": 461730
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 2.2052881304048367e-05,
+      "loss": 0.7731,
+      "step": 461740
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 2.2051733374148616e-05,
+      "loss": 0.8078,
+      "step": 461750
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 2.205058544424887e-05,
+      "loss": 0.7182,
+      "step": 461760
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 2.2049437514349123e-05,
+      "loss": 0.7886,
+      "step": 461770
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 2.2048289584449376e-05,
+      "loss": 0.8154,
+      "step": 461780
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 2.2047141654549626e-05,
+      "loss": 0.8068,
+      "step": 461790
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 2.2045993724649883e-05,
+      "loss": 0.8121,
+      "step": 461800
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 2.2044845794750133e-05,
+      "loss": 0.762,
+      "step": 461810
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 2.2043697864850382e-05,
+      "loss": 0.7688,
+      "step": 461820
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 2.204254993495064e-05,
+      "loss": 0.8301,
+      "step": 461830
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 2.204140200505089e-05,
+      "loss": 0.8308,
+      "step": 461840
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 2.2040254075151142e-05,
+      "loss": 0.7352,
+      "step": 461850
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 2.2039106145251395e-05,
+      "loss": 0.7704,
+      "step": 461860
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 2.203795821535165e-05,
+      "loss": 0.7909,
+      "step": 461870
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 2.20368102854519e-05,
+      "loss": 0.7757,
+      "step": 461880
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 2.2035662355552155e-05,
+      "loss": 0.7426,
+      "step": 461890
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 2.2034514425652405e-05,
+      "loss": 0.8174,
+      "step": 461900
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 2.2033366495752658e-05,
+      "loss": 0.793,
+      "step": 461910
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 2.203221856585291e-05,
+      "loss": 0.7797,
+      "step": 461920
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 2.2031070635953165e-05,
+      "loss": 0.8378,
+      "step": 461930
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 2.2029922706053415e-05,
+      "loss": 0.8083,
+      "step": 461940
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 2.202877477615367e-05,
+      "loss": 0.7796,
+      "step": 461950
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 2.202762684625392e-05,
+      "loss": 0.742,
+      "step": 461960
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 2.202647891635417e-05,
+      "loss": 0.6927,
+      "step": 461970
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 2.2025330986454427e-05,
+      "loss": 0.8367,
+      "step": 461980
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 2.2024183056554677e-05,
+      "loss": 0.7676,
+      "step": 461990
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 2.202303512665493e-05,
+      "loss": 0.7634,
+      "step": 462000
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 2.2021887196755184e-05,
+      "loss": 0.8247,
+      "step": 462010
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 2.2020739266855437e-05,
+      "loss": 0.7462,
+      "step": 462020
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 2.2019591336955687e-05,
+      "loss": 0.7255,
+      "step": 462030
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 2.2018443407055937e-05,
+      "loss": 0.8004,
+      "step": 462040
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 2.2017295477156193e-05,
+      "loss": 0.7214,
+      "step": 462050
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 2.2016147547256447e-05,
+      "loss": 0.8034,
+      "step": 462060
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 2.2014999617356696e-05,
+      "loss": 0.8441,
+      "step": 462070
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 2.2013851687456953e-05,
+      "loss": 0.7366,
+      "step": 462080
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 2.2012703757557203e-05,
+      "loss": 0.7255,
+      "step": 462090
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 2.2011555827657453e-05,
+      "loss": 0.7793,
+      "step": 462100
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 2.201040789775771e-05,
+      "loss": 0.7812,
+      "step": 462110
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 2.200925996785796e-05,
+      "loss": 0.7505,
+      "step": 462120
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 2.2008112037958213e-05,
+      "loss": 0.7668,
+      "step": 462130
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 2.2006964108058466e-05,
+      "loss": 0.7466,
+      "step": 462140
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 2.200581617815872e-05,
+      "loss": 0.7555,
+      "step": 462150
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 2.200466824825897e-05,
+      "loss": 0.7604,
+      "step": 462160
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 2.2003520318359226e-05,
+      "loss": 0.7774,
+      "step": 462170
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 2.2002372388459475e-05,
+      "loss": 0.7833,
+      "step": 462180
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 2.2001224458559725e-05,
+      "loss": 0.826,
+      "step": 462190
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 2.2000076528659982e-05,
+      "loss": 0.747,
+      "step": 462200
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 2.1998928598760235e-05,
+      "loss": 0.8625,
+      "step": 462210
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 2.1997780668860485e-05,
+      "loss": 0.8005,
+      "step": 462220
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 2.199663273896074e-05,
+      "loss": 0.7729,
+      "step": 462230
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 2.199548480906099e-05,
+      "loss": 0.8278,
+      "step": 462240
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 2.199433687916124e-05,
+      "loss": 0.7969,
+      "step": 462250
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 2.1993188949261498e-05,
+      "loss": 0.711,
+      "step": 462260
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 2.1992041019361748e-05,
+      "loss": 0.713,
+      "step": 462270
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 2.1990893089462e-05,
+      "loss": 0.8266,
+      "step": 462280
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 2.1989745159562254e-05,
+      "loss": 0.8182,
+      "step": 462290
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 2.1988597229662508e-05,
+      "loss": 0.7233,
+      "step": 462300
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 2.1987449299762757e-05,
+      "loss": 0.7101,
+      "step": 462310
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 2.1986301369863014e-05,
+      "loss": 0.8222,
+      "step": 462320
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 2.1985153439963264e-05,
+      "loss": 0.7669,
+      "step": 462330
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 2.1984005510063517e-05,
+      "loss": 0.7601,
+      "step": 462340
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 2.198285758016377e-05,
+      "loss": 0.7769,
+      "step": 462350
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 2.1981709650264024e-05,
+      "loss": 0.7878,
+      "step": 462360
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 2.1980561720364273e-05,
+      "loss": 0.7821,
+      "step": 462370
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 2.197941379046453e-05,
+      "loss": 0.8167,
+      "step": 462380
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 2.197826586056478e-05,
+      "loss": 0.7635,
+      "step": 462390
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 2.197711793066503e-05,
+      "loss": 0.75,
+      "step": 462400
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 2.1975970000765286e-05,
+      "loss": 0.7298,
+      "step": 462410
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 2.1974822070865536e-05,
+      "loss": 0.8204,
+      "step": 462420
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 2.197367414096579e-05,
+      "loss": 0.7056,
+      "step": 462430
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 2.1972526211066043e-05,
+      "loss": 0.777,
+      "step": 462440
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 2.1971378281166296e-05,
+      "loss": 0.7667,
+      "step": 462450
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 2.1970230351266546e-05,
+      "loss": 0.8062,
+      "step": 462460
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 2.1969082421366802e-05,
+      "loss": 0.7728,
+      "step": 462470
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 2.1967934491467052e-05,
+      "loss": 0.7572,
+      "step": 462480
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 2.1966786561567306e-05,
+      "loss": 0.7476,
+      "step": 462490
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 2.196563863166756e-05,
+      "loss": 0.8189,
+      "step": 462500
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 2.1964490701767812e-05,
+      "loss": 0.8581,
+      "step": 462510
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 2.1963342771868062e-05,
+      "loss": 0.773,
+      "step": 462520
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 2.196219484196832e-05,
+      "loss": 0.7175,
+      "step": 462530
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 2.196104691206857e-05,
+      "loss": 0.7452,
+      "step": 462540
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 2.1959898982168818e-05,
+      "loss": 0.7865,
+      "step": 462550
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 2.1958751052269075e-05,
+      "loss": 0.758,
+      "step": 462560
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 2.1957603122369325e-05,
+      "loss": 0.7999,
+      "step": 462570
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 2.1956455192469578e-05,
+      "loss": 0.7116,
+      "step": 462580
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 2.195530726256983e-05,
+      "loss": 0.8294,
+      "step": 462590
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 2.1954159332670084e-05,
+      "loss": 0.7538,
+      "step": 462600
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 2.1953011402770334e-05,
+      "loss": 0.7573,
+      "step": 462610
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 2.195186347287059e-05,
+      "loss": 0.7519,
+      "step": 462620
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 2.195071554297084e-05,
+      "loss": 0.7717,
+      "step": 462630
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 2.1949567613071094e-05,
+      "loss": 0.8015,
+      "step": 462640
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 2.1948419683171347e-05,
+      "loss": 0.7328,
+      "step": 462650
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 2.19472717532716e-05,
+      "loss": 0.805,
+      "step": 462660
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 2.194612382337185e-05,
+      "loss": 0.8005,
+      "step": 462670
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 2.1944975893472107e-05,
+      "loss": 0.8285,
+      "step": 462680
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 2.1943827963572357e-05,
+      "loss": 0.7653,
+      "step": 462690
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 2.1942680033672607e-05,
+      "loss": 0.8158,
+      "step": 462700
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 2.1941532103772863e-05,
+      "loss": 0.8306,
+      "step": 462710
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 2.1940384173873113e-05,
+      "loss": 0.7336,
+      "step": 462720
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 2.1939236243973366e-05,
+      "loss": 0.8383,
+      "step": 462730
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 2.1938088314073616e-05,
+      "loss": 0.7896,
+      "step": 462740
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 2.1936940384173873e-05,
+      "loss": 0.828,
+      "step": 462750
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 2.1935792454274123e-05,
+      "loss": 0.7833,
+      "step": 462760
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 2.1934644524374373e-05,
+      "loss": 0.7863,
+      "step": 462770
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 2.193349659447463e-05,
+      "loss": 0.7428,
+      "step": 462780
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 2.1932348664574883e-05,
+      "loss": 0.7391,
+      "step": 462790
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 2.1931200734675132e-05,
+      "loss": 0.8248,
+      "step": 462800
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 2.193005280477539e-05,
+      "loss": 0.8052,
+      "step": 462810
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 2.192890487487564e-05,
+      "loss": 0.7986,
+      "step": 462820
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 2.192775694497589e-05,
+      "loss": 0.8151,
+      "step": 462830
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 2.1926609015076145e-05,
+      "loss": 0.7816,
+      "step": 462840
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 2.1925461085176395e-05,
+      "loss": 0.7949,
+      "step": 462850
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 2.192431315527665e-05,
+      "loss": 0.8041,
+      "step": 462860
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 2.19231652253769e-05,
+      "loss": 0.8251,
+      "step": 462870
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 2.1922017295477155e-05,
+      "loss": 0.8625,
+      "step": 462880
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 2.1920869365577405e-05,
+      "loss": 0.8056,
+      "step": 462890
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 2.191972143567766e-05,
+      "loss": 0.7454,
+      "step": 462900
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 2.191857350577791e-05,
+      "loss": 0.7414,
+      "step": 462910
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 2.191742557587816e-05,
+      "loss": 0.7591,
+      "step": 462920
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 2.1916277645978418e-05,
+      "loss": 0.7518,
+      "step": 462930
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 2.191512971607867e-05,
+      "loss": 0.8133,
+      "step": 462940
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 2.191398178617892e-05,
+      "loss": 0.7772,
+      "step": 462950
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 2.1912833856279177e-05,
+      "loss": 0.8057,
+      "step": 462960
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 2.1911685926379427e-05,
+      "loss": 0.7754,
+      "step": 462970
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 2.1910537996479677e-05,
+      "loss": 0.7417,
+      "step": 462980
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 2.1909390066579934e-05,
+      "loss": 0.829,
+      "step": 462990
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 2.1908242136680184e-05,
+      "loss": 0.8456,
+      "step": 463000
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 2.1907094206780437e-05,
+      "loss": 0.7721,
+      "step": 463010
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 2.190594627688069e-05,
+      "loss": 0.7972,
+      "step": 463020
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 2.1904798346980943e-05,
+      "loss": 0.8384,
+      "step": 463030
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 2.1903650417081193e-05,
+      "loss": 0.7655,
+      "step": 463040
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 2.190250248718145e-05,
+      "loss": 0.7839,
+      "step": 463050
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 2.19013545572817e-05,
+      "loss": 0.6335,
+      "step": 463060
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 2.190020662738195e-05,
+      "loss": 0.8133,
+      "step": 463070
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 2.1899058697482206e-05,
+      "loss": 0.8469,
+      "step": 463080
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 2.189791076758246e-05,
+      "loss": 0.7519,
+      "step": 463090
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 2.189676283768271e-05,
+      "loss": 0.8123,
+      "step": 463100
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 2.1895614907782966e-05,
+      "loss": 0.8822,
+      "step": 463110
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 2.1894466977883216e-05,
+      "loss": 0.8124,
+      "step": 463120
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 2.1893319047983466e-05,
+      "loss": 0.826,
+      "step": 463130
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 2.1892171118083722e-05,
+      "loss": 0.7635,
+      "step": 463140
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 2.1891023188183972e-05,
+      "loss": 0.8737,
+      "step": 463150
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 2.1889875258284225e-05,
+      "loss": 0.7061,
+      "step": 463160
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 2.188872732838448e-05,
+      "loss": 0.8146,
+      "step": 463170
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 2.1887579398484732e-05,
+      "loss": 0.7845,
+      "step": 463180
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 2.1886431468584982e-05,
+      "loss": 0.7586,
+      "step": 463190
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 2.188528353868524e-05,
+      "loss": 0.9116,
+      "step": 463200
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 2.1884135608785488e-05,
+      "loss": 0.7731,
+      "step": 463210
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 2.1882987678885738e-05,
+      "loss": 0.7974,
+      "step": 463220
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 2.1881839748985995e-05,
+      "loss": 0.7589,
+      "step": 463230
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 2.1880691819086248e-05,
+      "loss": 0.7093,
+      "step": 463240
+    },
+    {
+      "epoch": 21.25,
+      "learning_rate": 2.1879543889186498e-05,
+      "loss": 0.8157,
+      "step": 463250
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 2.1878395959286754e-05,
+      "loss": 0.7592,
+      "step": 463260
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 2.1877248029387004e-05,
+      "loss": 0.7773,
+      "step": 463270
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 2.1876100099487254e-05,
+      "loss": 0.7841,
+      "step": 463280
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 2.187495216958751e-05,
+      "loss": 0.7946,
+      "step": 463290
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 2.187380423968776e-05,
+      "loss": 0.8215,
+      "step": 463300
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 2.1872656309788014e-05,
+      "loss": 0.7686,
+      "step": 463310
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 2.1871508379888267e-05,
+      "loss": 0.7731,
+      "step": 463320
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 2.187036044998852e-05,
+      "loss": 0.8076,
+      "step": 463330
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 2.186921252008877e-05,
+      "loss": 0.7865,
+      "step": 463340
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 2.1868064590189027e-05,
+      "loss": 0.7136,
+      "step": 463350
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 2.1866916660289277e-05,
+      "loss": 0.7473,
+      "step": 463360
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 2.1865768730389527e-05,
+      "loss": 0.7565,
+      "step": 463370
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 2.1864620800489783e-05,
+      "loss": 0.7493,
+      "step": 463380
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 2.1863472870590036e-05,
+      "loss": 0.7814,
+      "step": 463390
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 2.1862324940690286e-05,
+      "loss": 0.8161,
+      "step": 463400
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 2.1861177010790536e-05,
+      "loss": 0.8013,
+      "step": 463410
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 2.1860029080890793e-05,
+      "loss": 0.7925,
+      "step": 463420
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 2.1858881150991043e-05,
+      "loss": 0.8594,
+      "step": 463430
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 2.1857733221091296e-05,
+      "loss": 0.8885,
+      "step": 463440
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 2.185658529119155e-05,
+      "loss": 0.8411,
+      "step": 463450
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 2.1855437361291802e-05,
+      "loss": 0.7007,
+      "step": 463460
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 2.1854289431392052e-05,
+      "loss": 0.8594,
+      "step": 463470
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 2.185314150149231e-05,
+      "loss": 0.732,
+      "step": 463480
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 2.185199357159256e-05,
+      "loss": 0.7819,
+      "step": 463490
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 2.185084564169281e-05,
+      "loss": 0.8009,
+      "step": 463500
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 2.1849697711793065e-05,
+      "loss": 0.8017,
+      "step": 463510
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 2.1848549781893315e-05,
+      "loss": 0.7925,
+      "step": 463520
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 2.1847401851993568e-05,
+      "loss": 0.7371,
+      "step": 463530
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 2.1846253922093825e-05,
+      "loss": 0.7925,
+      "step": 463540
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 2.1845105992194075e-05,
+      "loss": 0.8115,
+      "step": 463550
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 2.1843958062294325e-05,
+      "loss": 0.6594,
+      "step": 463560
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 2.184281013239458e-05,
+      "loss": 0.8475,
+      "step": 463570
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 2.184166220249483e-05,
+      "loss": 0.7807,
+      "step": 463580
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 2.1840514272595084e-05,
+      "loss": 0.856,
+      "step": 463590
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 2.1839366342695338e-05,
+      "loss": 0.7729,
+      "step": 463600
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 2.183821841279559e-05,
+      "loss": 0.7571,
+      "step": 463610
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 2.183707048289584e-05,
+      "loss": 0.7083,
+      "step": 463620
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 2.1835922552996097e-05,
+      "loss": 0.6852,
+      "step": 463630
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 2.1834774623096347e-05,
+      "loss": 0.7285,
+      "step": 463640
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 2.1833626693196597e-05,
+      "loss": 0.7317,
+      "step": 463650
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 2.1832478763296854e-05,
+      "loss": 0.7136,
+      "step": 463660
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 2.1831330833397103e-05,
+      "loss": 0.7683,
+      "step": 463670
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 2.1830182903497357e-05,
+      "loss": 0.855,
+      "step": 463680
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 2.1829034973597613e-05,
+      "loss": 0.8285,
+      "step": 463690
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 2.1827887043697863e-05,
+      "loss": 0.8428,
+      "step": 463700
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 2.1826739113798113e-05,
+      "loss": 0.7721,
+      "step": 463710
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 2.182559118389837e-05,
+      "loss": 0.7465,
+      "step": 463720
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 2.182444325399862e-05,
+      "loss": 0.8142,
+      "step": 463730
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 2.1823295324098873e-05,
+      "loss": 0.7805,
+      "step": 463740
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 2.1822147394199126e-05,
+      "loss": 0.7351,
+      "step": 463750
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 2.182099946429938e-05,
+      "loss": 0.7804,
+      "step": 463760
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 2.181985153439963e-05,
+      "loss": 0.8007,
+      "step": 463770
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 2.1818703604499886e-05,
+      "loss": 0.7157,
+      "step": 463780
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 2.1817555674600136e-05,
+      "loss": 0.7712,
+      "step": 463790
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 2.1816407744700385e-05,
+      "loss": 0.7324,
+      "step": 463800
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 2.1815259814800642e-05,
+      "loss": 0.7037,
+      "step": 463810
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 2.1814111884900892e-05,
+      "loss": 0.8411,
+      "step": 463820
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 2.1812963955001145e-05,
+      "loss": 0.8175,
+      "step": 463830
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 2.1811816025101402e-05,
+      "loss": 0.7354,
+      "step": 463840
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 2.181066809520165e-05,
+      "loss": 0.7843,
+      "step": 463850
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 2.18095201653019e-05,
+      "loss": 0.7602,
+      "step": 463860
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 2.1808372235402158e-05,
+      "loss": 0.8578,
+      "step": 463870
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 2.1807224305502408e-05,
+      "loss": 0.7561,
+      "step": 463880
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 2.180607637560266e-05,
+      "loss": 0.8231,
+      "step": 463890
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 2.1804928445702914e-05,
+      "loss": 0.7897,
+      "step": 463900
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 2.1803780515803168e-05,
+      "loss": 0.7064,
+      "step": 463910
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 2.1802632585903418e-05,
+      "loss": 0.8254,
+      "step": 463920
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 2.1801484656003674e-05,
+      "loss": 0.8208,
+      "step": 463930
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 2.1800336726103924e-05,
+      "loss": 0.7648,
+      "step": 463940
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 2.1799188796204174e-05,
+      "loss": 0.7709,
+      "step": 463950
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 2.179804086630443e-05,
+      "loss": 0.7183,
+      "step": 463960
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 2.179689293640468e-05,
+      "loss": 0.7999,
+      "step": 463970
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 2.1795745006504934e-05,
+      "loss": 0.8014,
+      "step": 463980
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 2.179459707660519e-05,
+      "loss": 0.8621,
+      "step": 463990
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 2.179344914670544e-05,
+      "loss": 0.7312,
+      "step": 464000
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 2.179230121680569e-05,
+      "loss": 0.793,
+      "step": 464010
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 2.1791153286905947e-05,
+      "loss": 0.7126,
+      "step": 464020
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 2.1790005357006196e-05,
+      "loss": 0.7467,
+      "step": 464030
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 2.178885742710645e-05,
+      "loss": 0.7901,
+      "step": 464040
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 2.1787709497206703e-05,
+      "loss": 0.8575,
+      "step": 464050
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 2.1786561567306956e-05,
+      "loss": 0.7539,
+      "step": 464060
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 2.1785413637407206e-05,
+      "loss": 0.7476,
+      "step": 464070
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 2.1784265707507463e-05,
+      "loss": 0.832,
+      "step": 464080
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 2.1783117777607713e-05,
+      "loss": 0.8015,
+      "step": 464090
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 2.1781969847707962e-05,
+      "loss": 0.7966,
+      "step": 464100
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 2.1780821917808216e-05,
+      "loss": 0.7879,
+      "step": 464110
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 2.177967398790847e-05,
+      "loss": 0.7583,
+      "step": 464120
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 2.1778526058008722e-05,
+      "loss": 0.7755,
+      "step": 464130
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 2.1777378128108972e-05,
+      "loss": 0.8171,
+      "step": 464140
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 2.177623019820923e-05,
+      "loss": 0.7744,
+      "step": 464150
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 2.177508226830948e-05,
+      "loss": 0.7106,
+      "step": 464160
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 2.1773934338409732e-05,
+      "loss": 0.819,
+      "step": 464170
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 2.1772786408509985e-05,
+      "loss": 0.741,
+      "step": 464180
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 2.1771638478610238e-05,
+      "loss": 0.8215,
+      "step": 464190
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 2.1770490548710488e-05,
+      "loss": 0.7896,
+      "step": 464200
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 2.1769342618810745e-05,
+      "loss": 0.7444,
+      "step": 464210
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 2.1768194688910995e-05,
+      "loss": 0.8741,
+      "step": 464220
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 2.1767046759011244e-05,
+      "loss": 0.7642,
+      "step": 464230
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 2.17658988291115e-05,
+      "loss": 0.8191,
+      "step": 464240
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 2.176475089921175e-05,
+      "loss": 0.7299,
+      "step": 464250
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 2.1763602969312004e-05,
+      "loss": 0.8614,
+      "step": 464260
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 2.1762455039412257e-05,
+      "loss": 0.7503,
+      "step": 464270
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 2.176130710951251e-05,
+      "loss": 0.7588,
+      "step": 464280
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 2.176015917961276e-05,
+      "loss": 0.7789,
+      "step": 464290
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 2.1759011249713017e-05,
+      "loss": 0.7911,
+      "step": 464300
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 2.1757863319813267e-05,
+      "loss": 0.7997,
+      "step": 464310
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 2.175671538991352e-05,
+      "loss": 0.7532,
+      "step": 464320
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 2.1755567460013773e-05,
+      "loss": 0.6849,
+      "step": 464330
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 2.1754419530114027e-05,
+      "loss": 0.7865,
+      "step": 464340
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 2.1753271600214277e-05,
+      "loss": 0.9304,
+      "step": 464350
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 2.1752123670314533e-05,
+      "loss": 0.7503,
+      "step": 464360
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 2.1750975740414783e-05,
+      "loss": 0.7963,
+      "step": 464370
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 2.1749827810515033e-05,
+      "loss": 0.8423,
+      "step": 464380
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 2.174867988061529e-05,
+      "loss": 0.7194,
+      "step": 464390
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 2.174753195071554e-05,
+      "loss": 0.8005,
+      "step": 464400
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 2.1746384020815793e-05,
+      "loss": 0.8026,
+      "step": 464410
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 2.1745236090916046e-05,
+      "loss": 0.8089,
+      "step": 464420
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 2.17440881610163e-05,
+      "loss": 0.7584,
+      "step": 464430
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 2.174294023111655e-05,
+      "loss": 0.827,
+      "step": 464440
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 2.1741792301216806e-05,
+      "loss": 0.7383,
+      "step": 464450
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 2.1740644371317055e-05,
+      "loss": 0.7842,
+      "step": 464460
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 2.173949644141731e-05,
+      "loss": 0.7944,
+      "step": 464470
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 2.1738348511517562e-05,
+      "loss": 0.737,
+      "step": 464480
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 2.1737200581617815e-05,
+      "loss": 0.8216,
+      "step": 464490
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 2.1736052651718065e-05,
+      "loss": 0.7313,
+      "step": 464500
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 2.173490472181832e-05,
+      "loss": 0.7862,
+      "step": 464510
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 2.173375679191857e-05,
+      "loss": 0.8035,
+      "step": 464520
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 2.173260886201882e-05,
+      "loss": 0.7918,
+      "step": 464530
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 2.1731460932119078e-05,
+      "loss": 0.713,
+      "step": 464540
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 2.1730313002219328e-05,
+      "loss": 0.8123,
+      "step": 464550
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 2.172916507231958e-05,
+      "loss": 0.7695,
+      "step": 464560
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 2.1728017142419834e-05,
+      "loss": 0.7842,
+      "step": 464570
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 2.1726869212520088e-05,
+      "loss": 0.672,
+      "step": 464580
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 2.1725721282620337e-05,
+      "loss": 0.7652,
+      "step": 464590
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 2.1724573352720594e-05,
+      "loss": 0.7958,
+      "step": 464600
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 2.1723425422820844e-05,
+      "loss": 0.7705,
+      "step": 464610
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 2.1722277492921097e-05,
+      "loss": 0.7865,
+      "step": 464620
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 2.172112956302135e-05,
+      "loss": 0.8365,
+      "step": 464630
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 2.1719981633121604e-05,
+      "loss": 0.6994,
+      "step": 464640
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 2.1718833703221853e-05,
+      "loss": 0.7729,
+      "step": 464650
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 2.171768577332211e-05,
+      "loss": 0.8745,
+      "step": 464660
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 2.171653784342236e-05,
+      "loss": 0.7916,
+      "step": 464670
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 2.171538991352261e-05,
+      "loss": 0.784,
+      "step": 464680
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 2.1714241983622866e-05,
+      "loss": 0.8201,
+      "step": 464690
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 2.1713094053723116e-05,
+      "loss": 0.7864,
+      "step": 464700
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 2.171194612382337e-05,
+      "loss": 0.7445,
+      "step": 464710
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 2.1710798193923623e-05,
+      "loss": 0.7393,
+      "step": 464720
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 2.1709650264023876e-05,
+      "loss": 0.8002,
+      "step": 464730
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 2.1708502334124126e-05,
+      "loss": 0.7461,
+      "step": 464740
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 2.1707354404224383e-05,
+      "loss": 0.8407,
+      "step": 464750
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 2.1706206474324632e-05,
+      "loss": 0.7669,
+      "step": 464760
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 2.1705058544424886e-05,
+      "loss": 0.7192,
+      "step": 464770
+    },
+    {
+      "epoch": 21.32,
+      "learning_rate": 2.1703910614525135e-05,
+      "loss": 0.7773,
+      "step": 464780
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 2.1702762684625392e-05,
+      "loss": 0.7526,
+      "step": 464790
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 2.1701614754725642e-05,
+      "loss": 0.7748,
+      "step": 464800
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 2.1700466824825892e-05,
+      "loss": 0.774,
+      "step": 464810
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 2.169931889492615e-05,
+      "loss": 0.768,
+      "step": 464820
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 2.1698170965026398e-05,
+      "loss": 0.8145,
+      "step": 464830
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 2.169702303512665e-05,
+      "loss": 0.8056,
+      "step": 464840
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 2.1695875105226905e-05,
+      "loss": 0.7937,
+      "step": 464850
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 2.1694727175327158e-05,
+      "loss": 0.7881,
+      "step": 464860
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 2.1693579245427408e-05,
+      "loss": 0.8893,
+      "step": 464870
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 2.1692431315527664e-05,
+      "loss": 0.7558,
+      "step": 464880
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 2.1691283385627914e-05,
+      "loss": 0.7904,
+      "step": 464890
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 2.1690135455728168e-05,
+      "loss": 0.8372,
+      "step": 464900
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 2.168898752582842e-05,
+      "loss": 0.8029,
+      "step": 464910
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 2.1687839595928674e-05,
+      "loss": 0.7567,
+      "step": 464920
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 2.1686691666028924e-05,
+      "loss": 0.7728,
+      "step": 464930
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 2.168554373612918e-05,
+      "loss": 0.8208,
+      "step": 464940
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 2.168439580622943e-05,
+      "loss": 0.8026,
+      "step": 464950
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 2.168324787632968e-05,
+      "loss": 0.7531,
+      "step": 464960
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 2.1682099946429937e-05,
+      "loss": 0.8417,
+      "step": 464970
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 2.1680952016530187e-05,
+      "loss": 0.6801,
+      "step": 464980
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 2.167980408663044e-05,
+      "loss": 0.7711,
+      "step": 464990
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 2.1678656156730693e-05,
+      "loss": 0.8357,
+      "step": 465000
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 2.1677508226830946e-05,
+      "loss": 0.7738,
+      "step": 465010
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 2.1676360296931196e-05,
+      "loss": 0.7489,
+      "step": 465020
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 2.1675212367031453e-05,
+      "loss": 0.7562,
+      "step": 465030
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 2.1674064437131703e-05,
+      "loss": 0.8021,
+      "step": 465040
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 2.1672916507231956e-05,
+      "loss": 0.8724,
+      "step": 465050
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 2.167176857733221e-05,
+      "loss": 0.7214,
+      "step": 465060
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 2.1670620647432463e-05,
+      "loss": 0.7849,
+      "step": 465070
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 2.1669472717532712e-05,
+      "loss": 0.7091,
+      "step": 465080
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 2.166832478763297e-05,
+      "loss": 0.746,
+      "step": 465090
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 2.166717685773322e-05,
+      "loss": 0.7622,
+      "step": 465100
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 2.166602892783347e-05,
+      "loss": 0.7926,
+      "step": 465110
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 2.1664880997933725e-05,
+      "loss": 0.695,
+      "step": 465120
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 2.1663733068033975e-05,
+      "loss": 0.788,
+      "step": 465130
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 2.166258513813423e-05,
+      "loss": 0.796,
+      "step": 465140
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 2.1661437208234482e-05,
+      "loss": 0.7072,
+      "step": 465150
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 2.1660289278334735e-05,
+      "loss": 0.8782,
+      "step": 465160
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 2.1659141348434985e-05,
+      "loss": 0.7642,
+      "step": 465170
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 2.165799341853524e-05,
+      "loss": 0.7769,
+      "step": 465180
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 2.165684548863549e-05,
+      "loss": 0.7996,
+      "step": 465190
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 2.1655697558735745e-05,
+      "loss": 0.742,
+      "step": 465200
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 2.1654549628835998e-05,
+      "loss": 0.8521,
+      "step": 465210
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 2.165340169893625e-05,
+      "loss": 0.857,
+      "step": 465220
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 2.16522537690365e-05,
+      "loss": 0.8905,
+      "step": 465230
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 2.1651105839136758e-05,
+      "loss": 0.8477,
+      "step": 465240
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 2.1649957909237007e-05,
+      "loss": 0.7852,
+      "step": 465250
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 2.1648809979337257e-05,
+      "loss": 0.7482,
+      "step": 465260
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 2.1647662049437514e-05,
+      "loss": 0.7889,
+      "step": 465270
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 2.1646514119537764e-05,
+      "loss": 0.7668,
+      "step": 465280
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 2.1645366189638017e-05,
+      "loss": 0.7105,
+      "step": 465290
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 2.164421825973827e-05,
+      "loss": 0.8658,
+      "step": 465300
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 2.1643070329838523e-05,
+      "loss": 0.6952,
+      "step": 465310
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 2.1641922399938773e-05,
+      "loss": 0.8366,
+      "step": 465320
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 2.164077447003903e-05,
+      "loss": 0.7059,
+      "step": 465330
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 2.163962654013928e-05,
+      "loss": 0.8352,
+      "step": 465340
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 2.1638478610239533e-05,
+      "loss": 0.6802,
+      "step": 465350
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 2.1637330680339786e-05,
+      "loss": 0.7523,
+      "step": 465360
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 2.163618275044004e-05,
+      "loss": 0.7365,
+      "step": 465370
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 2.163503482054029e-05,
+      "loss": 0.7668,
+      "step": 465380
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 2.1633886890640546e-05,
+      "loss": 0.7932,
+      "step": 465390
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 2.1632738960740796e-05,
+      "loss": 0.729,
+      "step": 465400
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 2.1631591030841046e-05,
+      "loss": 0.7962,
+      "step": 465410
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 2.1630443100941302e-05,
+      "loss": 0.7285,
+      "step": 465420
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 2.1629295171041552e-05,
+      "loss": 0.8132,
+      "step": 465430
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 2.1628147241141805e-05,
+      "loss": 0.8373,
+      "step": 465440
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 2.162699931124206e-05,
+      "loss": 0.7909,
+      "step": 465450
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 2.1625851381342312e-05,
+      "loss": 0.8887,
+      "step": 465460
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 2.1624703451442562e-05,
+      "loss": 0.8247,
+      "step": 465470
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 2.1623555521542815e-05,
+      "loss": 0.7677,
+      "step": 465480
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 2.1622407591643068e-05,
+      "loss": 0.8113,
+      "step": 465490
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 2.162125966174332e-05,
+      "loss": 0.7273,
+      "step": 465500
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 2.162011173184357e-05,
+      "loss": 0.7653,
+      "step": 465510
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 2.1618963801943828e-05,
+      "loss": 0.778,
+      "step": 465520
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 2.1617815872044078e-05,
+      "loss": 0.775,
+      "step": 465530
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 2.1616667942144328e-05,
+      "loss": 0.7891,
+      "step": 465540
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 2.1615520012244584e-05,
+      "loss": 0.7953,
+      "step": 465550
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 2.1614372082344834e-05,
+      "loss": 0.7427,
+      "step": 465560
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 2.1613224152445087e-05,
+      "loss": 0.8904,
+      "step": 465570
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 2.161207622254534e-05,
+      "loss": 0.8123,
+      "step": 465580
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 2.1610928292645594e-05,
+      "loss": 0.7893,
+      "step": 465590
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 2.1609780362745844e-05,
+      "loss": 0.9204,
+      "step": 465600
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 2.16086324328461e-05,
+      "loss": 0.7441,
+      "step": 465610
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 2.160748450294635e-05,
+      "loss": 0.8066,
+      "step": 465620
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 2.1606336573046603e-05,
+      "loss": 0.7559,
+      "step": 465630
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 2.1605188643146857e-05,
+      "loss": 0.743,
+      "step": 465640
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 2.160404071324711e-05,
+      "loss": 0.7717,
+      "step": 465650
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 2.160289278334736e-05,
+      "loss": 0.7564,
+      "step": 465660
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 2.1601744853447616e-05,
+      "loss": 0.7184,
+      "step": 465670
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 2.1600596923547866e-05,
+      "loss": 0.7005,
+      "step": 465680
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 2.1599448993648116e-05,
+      "loss": 0.7084,
+      "step": 465690
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 2.1598301063748373e-05,
+      "loss": 0.8186,
+      "step": 465700
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 2.1597153133848623e-05,
+      "loss": 0.8113,
+      "step": 465710
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 2.1596005203948876e-05,
+      "loss": 0.7682,
+      "step": 465720
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 2.159485727404913e-05,
+      "loss": 0.7685,
+      "step": 465730
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 2.1593709344149382e-05,
+      "loss": 0.7912,
+      "step": 465740
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 2.1592561414249632e-05,
+      "loss": 0.7764,
+      "step": 465750
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 2.159141348434989e-05,
+      "loss": 0.7984,
+      "step": 465760
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 2.159026555445014e-05,
+      "loss": 0.7352,
+      "step": 465770
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 2.1589117624550392e-05,
+      "loss": 0.7136,
+      "step": 465780
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 2.1587969694650645e-05,
+      "loss": 0.699,
+      "step": 465790
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 2.15868217647509e-05,
+      "loss": 0.8619,
+      "step": 465800
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 2.1585673834851148e-05,
+      "loss": 0.8061,
+      "step": 465810
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 2.1584525904951405e-05,
+      "loss": 0.7583,
+      "step": 465820
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 2.1583377975051655e-05,
+      "loss": 0.8053,
+      "step": 465830
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 2.1582230045151905e-05,
+      "loss": 0.7806,
+      "step": 465840
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 2.158108211525216e-05,
+      "loss": 0.7588,
+      "step": 465850
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 2.157993418535241e-05,
+      "loss": 0.7079,
+      "step": 465860
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 2.1578786255452664e-05,
+      "loss": 0.7511,
+      "step": 465870
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 2.1577638325552918e-05,
+      "loss": 0.7894,
+      "step": 465880
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 2.157649039565317e-05,
+      "loss": 0.8697,
+      "step": 465890
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 2.157534246575342e-05,
+      "loss": 0.7784,
+      "step": 465900
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 2.1574194535853677e-05,
+      "loss": 0.7231,
+      "step": 465910
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 2.1573046605953927e-05,
+      "loss": 0.8852,
+      "step": 465920
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 2.157189867605418e-05,
+      "loss": 0.7209,
+      "step": 465930
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 2.1570750746154434e-05,
+      "loss": 0.8135,
+      "step": 465940
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 2.1569602816254687e-05,
+      "loss": 0.7676,
+      "step": 465950
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 2.1568454886354937e-05,
+      "loss": 0.7575,
+      "step": 465960
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 2.1567306956455193e-05,
+      "loss": 0.7843,
+      "step": 465970
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 2.1566159026555443e-05,
+      "loss": 0.871,
+      "step": 465980
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 2.1565011096655693e-05,
+      "loss": 0.8105,
+      "step": 465990
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 2.156386316675595e-05,
+      "loss": 0.8459,
+      "step": 466000
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 2.15627152368562e-05,
+      "loss": 0.7866,
+      "step": 466010
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 2.1561567306956453e-05,
+      "loss": 0.8148,
+      "step": 466020
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 2.1560419377056706e-05,
+      "loss": 0.822,
+      "step": 466030
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 2.155927144715696e-05,
+      "loss": 0.7999,
+      "step": 466040
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 2.155812351725721e-05,
+      "loss": 0.7738,
+      "step": 466050
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 2.1556975587357466e-05,
+      "loss": 0.7976,
+      "step": 466060
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 2.1555827657457716e-05,
+      "loss": 0.7912,
+      "step": 466070
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 2.155467972755797e-05,
+      "loss": 0.7819,
+      "step": 466080
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 2.1553531797658222e-05,
+      "loss": 0.8481,
+      "step": 466090
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 2.1552383867758475e-05,
+      "loss": 0.7539,
+      "step": 466100
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 2.1551235937858725e-05,
+      "loss": 0.7885,
+      "step": 466110
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 2.1550088007958982e-05,
+      "loss": 0.7851,
+      "step": 466120
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 2.1548940078059232e-05,
+      "loss": 0.945,
+      "step": 466130
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 2.154779214815948e-05,
+      "loss": 0.7758,
+      "step": 466140
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 2.1546644218259735e-05,
+      "loss": 0.7552,
+      "step": 466150
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 2.1545496288359988e-05,
+      "loss": 0.7983,
+      "step": 466160
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 2.154434835846024e-05,
+      "loss": 0.7664,
+      "step": 466170
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 2.154320042856049e-05,
+      "loss": 0.7765,
+      "step": 466180
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 2.1542052498660748e-05,
+      "loss": 0.8265,
+      "step": 466190
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 2.1540904568760998e-05,
+      "loss": 0.8428,
+      "step": 466200
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 2.153975663886125e-05,
+      "loss": 0.7963,
+      "step": 466210
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 2.1538608708961504e-05,
+      "loss": 0.7025,
+      "step": 466220
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 2.1537460779061757e-05,
+      "loss": 0.7303,
+      "step": 466230
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 2.1536312849162007e-05,
+      "loss": 0.8576,
+      "step": 466240
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 2.1535164919262264e-05,
+      "loss": 0.6635,
+      "step": 466250
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 2.1534016989362514e-05,
+      "loss": 0.7337,
+      "step": 466260
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 2.1532869059462764e-05,
+      "loss": 0.8141,
+      "step": 466270
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 2.153172112956302e-05,
+      "loss": 0.7686,
+      "step": 466280
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 2.153057319966327e-05,
+      "loss": 0.7408,
+      "step": 466290
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 2.1529425269763523e-05,
+      "loss": 0.7396,
+      "step": 466300
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 2.1528277339863777e-05,
+      "loss": 0.7632,
+      "step": 466310
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 2.152712940996403e-05,
+      "loss": 0.8519,
+      "step": 466320
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 2.152598148006428e-05,
+      "loss": 0.7525,
+      "step": 466330
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 2.1524833550164536e-05,
+      "loss": 0.7963,
+      "step": 466340
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 2.1523685620264786e-05,
+      "loss": 0.7833,
+      "step": 466350
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 2.152253769036504e-05,
+      "loss": 0.7911,
+      "step": 466360
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 2.1521389760465293e-05,
+      "loss": 0.8275,
+      "step": 466370
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 2.1520241830565546e-05,
+      "loss": 0.8387,
+      "step": 466380
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 2.1519093900665796e-05,
+      "loss": 0.7817,
+      "step": 466390
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 2.1517945970766052e-05,
+      "loss": 0.7468,
+      "step": 466400
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 2.1516798040866302e-05,
+      "loss": 0.7795,
+      "step": 466410
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 2.1515650110966552e-05,
+      "loss": 0.7154,
+      "step": 466420
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 2.151450218106681e-05,
+      "loss": 0.8066,
+      "step": 466430
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 2.151335425116706e-05,
+      "loss": 0.8564,
+      "step": 466440
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 2.1512206321267312e-05,
+      "loss": 0.6678,
+      "step": 466450
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 2.1511058391367565e-05,
+      "loss": 0.8263,
+      "step": 466460
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 2.1509910461467818e-05,
+      "loss": 0.8088,
+      "step": 466470
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 2.1508762531568068e-05,
+      "loss": 0.7357,
+      "step": 466480
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 2.1507614601668325e-05,
+      "loss": 0.8643,
+      "step": 466490
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 2.1506466671768575e-05,
+      "loss": 0.6944,
+      "step": 466500
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 2.1505318741868828e-05,
+      "loss": 0.8633,
+      "step": 466510
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 2.150417081196908e-05,
+      "loss": 0.7416,
+      "step": 466520
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 2.1503022882069334e-05,
+      "loss": 0.7156,
+      "step": 466530
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 2.1501874952169584e-05,
+      "loss": 0.7476,
+      "step": 466540
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 2.150072702226984e-05,
+      "loss": 0.7076,
+      "step": 466550
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 2.149957909237009e-05,
+      "loss": 0.8767,
+      "step": 466560
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 2.149843116247034e-05,
+      "loss": 0.852,
+      "step": 466570
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 2.1497283232570597e-05,
+      "loss": 0.7758,
+      "step": 466580
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 2.1496135302670847e-05,
+      "loss": 0.7708,
+      "step": 466590
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 2.14949873727711e-05,
+      "loss": 0.8172,
+      "step": 466600
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 2.1493839442871353e-05,
+      "loss": 0.7918,
+      "step": 466610
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 2.1492691512971607e-05,
+      "loss": 0.768,
+      "step": 466620
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 2.1491543583071857e-05,
+      "loss": 0.8926,
+      "step": 466630
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 2.1490395653172113e-05,
+      "loss": 0.7831,
+      "step": 466640
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 2.1489247723272363e-05,
+      "loss": 0.791,
+      "step": 466650
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 2.1488099793372616e-05,
+      "loss": 0.8841,
+      "step": 466660
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 2.148695186347287e-05,
+      "loss": 0.7624,
+      "step": 466670
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 2.1485803933573123e-05,
+      "loss": 0.8765,
+      "step": 466680
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 2.1484656003673373e-05,
+      "loss": 0.7951,
+      "step": 466690
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 2.148350807377363e-05,
+      "loss": 0.7677,
+      "step": 466700
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 2.148236014387388e-05,
+      "loss": 0.7782,
+      "step": 466710
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 2.148121221397413e-05,
+      "loss": 0.7088,
+      "step": 466720
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 2.1480064284074386e-05,
+      "loss": 0.7137,
+      "step": 466730
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 2.1478916354174635e-05,
+      "loss": 0.8102,
+      "step": 466740
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 2.147776842427489e-05,
+      "loss": 0.8613,
+      "step": 466750
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 2.1476620494375142e-05,
+      "loss": 0.8456,
+      "step": 466760
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 2.1475472564475395e-05,
+      "loss": 0.7127,
+      "step": 466770
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 2.1474324634575645e-05,
+      "loss": 0.7358,
+      "step": 466780
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 2.14731767046759e-05,
+      "loss": 0.7486,
+      "step": 466790
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 2.147202877477615e-05,
+      "loss": 0.7737,
+      "step": 466800
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 2.1470880844876405e-05,
+      "loss": 0.7257,
+      "step": 466810
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 2.1469732914976658e-05,
+      "loss": 0.8005,
+      "step": 466820
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 2.146858498507691e-05,
+      "loss": 0.8268,
+      "step": 466830
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 2.146743705517716e-05,
+      "loss": 0.7309,
+      "step": 466840
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 2.146628912527741e-05,
+      "loss": 0.7721,
+      "step": 466850
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 2.1465141195377668e-05,
+      "loss": 0.8187,
+      "step": 466860
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 2.1463993265477917e-05,
+      "loss": 0.7684,
+      "step": 466870
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 2.146284533557817e-05,
+      "loss": 0.7361,
+      "step": 466880
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 2.1461697405678424e-05,
+      "loss": 0.7714,
+      "step": 466890
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 2.1460549475778677e-05,
+      "loss": 0.7598,
+      "step": 466900
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 2.1459401545878927e-05,
+      "loss": 0.7055,
+      "step": 466910
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 2.1458253615979184e-05,
+      "loss": 0.7775,
+      "step": 466920
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 2.1457105686079433e-05,
+      "loss": 0.8212,
+      "step": 466930
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 2.1455957756179687e-05,
+      "loss": 0.859,
+      "step": 466940
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 2.145480982627994e-05,
+      "loss": 0.8131,
+      "step": 466950
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 2.1453661896380193e-05,
+      "loss": 0.7561,
+      "step": 466960
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 2.1452513966480443e-05,
+      "loss": 0.8414,
+      "step": 466970
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 2.14513660365807e-05,
+      "loss": 0.7925,
+      "step": 466980
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 2.145021810668095e-05,
+      "loss": 0.8777,
+      "step": 466990
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 2.14490701767812e-05,
+      "loss": 0.6964,
+      "step": 467000
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 2.1447922246881456e-05,
+      "loss": 0.8219,
+      "step": 467010
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 2.1446774316981706e-05,
+      "loss": 0.7999,
+      "step": 467020
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 2.144562638708196e-05,
+      "loss": 0.8012,
+      "step": 467030
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 2.1444478457182212e-05,
+      "loss": 0.771,
+      "step": 467040
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 2.1443330527282466e-05,
+      "loss": 0.7874,
+      "step": 467050
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 2.1442182597382715e-05,
+      "loss": 0.7788,
+      "step": 467060
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 2.1441034667482972e-05,
+      "loss": 0.7905,
+      "step": 467070
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 2.1439886737583222e-05,
+      "loss": 0.8666,
+      "step": 467080
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 2.1438738807683475e-05,
+      "loss": 0.7789,
+      "step": 467090
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 2.143759087778373e-05,
+      "loss": 0.7708,
+      "step": 467100
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 2.1436442947883982e-05,
+      "loss": 0.7548,
+      "step": 467110
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 2.143529501798423e-05,
+      "loss": 0.7231,
+      "step": 467120
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 2.1434147088084488e-05,
+      "loss": 0.7375,
+      "step": 467130
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 2.1432999158184738e-05,
+      "loss": 0.7726,
+      "step": 467140
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 2.1431851228284988e-05,
+      "loss": 0.683,
+      "step": 467150
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 2.1430703298385245e-05,
+      "loss": 0.7856,
+      "step": 467160
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 2.1429555368485494e-05,
+      "loss": 0.7766,
+      "step": 467170
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 2.1428407438585748e-05,
+      "loss": 0.7135,
+      "step": 467180
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 2.1427259508686e-05,
+      "loss": 0.8859,
+      "step": 467190
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 2.1426111578786254e-05,
+      "loss": 0.8201,
+      "step": 467200
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 2.1424963648886504e-05,
+      "loss": 0.8294,
+      "step": 467210
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 2.142381571898676e-05,
+      "loss": 0.7044,
+      "step": 467220
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 2.142266778908701e-05,
+      "loss": 0.7664,
+      "step": 467230
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 2.1421519859187264e-05,
+      "loss": 0.7935,
+      "step": 467240
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 2.1420371929287517e-05,
+      "loss": 0.7595,
+      "step": 467250
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 2.141922399938777e-05,
+      "loss": 0.7845,
+      "step": 467260
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 2.141807606948802e-05,
+      "loss": 0.7898,
+      "step": 467270
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 2.1416928139588277e-05,
+      "loss": 0.8247,
+      "step": 467280
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 2.1415780209688527e-05,
+      "loss": 0.8194,
+      "step": 467290
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 2.1414632279788776e-05,
+      "loss": 0.7756,
+      "step": 467300
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 2.1413484349889033e-05,
+      "loss": 0.7766,
+      "step": 467310
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 2.1412336419989283e-05,
+      "loss": 0.8427,
+      "step": 467320
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 2.1411188490089536e-05,
+      "loss": 0.8208,
+      "step": 467330
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 2.141004056018979e-05,
+      "loss": 0.7684,
+      "step": 467340
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 2.1408892630290043e-05,
+      "loss": 0.8083,
+      "step": 467350
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 2.1407744700390292e-05,
+      "loss": 0.7988,
+      "step": 467360
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 2.140659677049055e-05,
+      "loss": 0.7416,
+      "step": 467370
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 2.14054488405908e-05,
+      "loss": 0.861,
+      "step": 467380
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 2.1404300910691052e-05,
+      "loss": 0.7858,
+      "step": 467390
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 2.1403152980791305e-05,
+      "loss": 0.7866,
+      "step": 467400
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 2.140200505089156e-05,
+      "loss": 0.8122,
+      "step": 467410
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 2.140085712099181e-05,
+      "loss": 0.7941,
+      "step": 467420
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 2.1399709191092065e-05,
+      "loss": 0.829,
+      "step": 467430
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 2.1398561261192315e-05,
+      "loss": 0.8598,
+      "step": 467440
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 2.1397413331292565e-05,
+      "loss": 0.7715,
+      "step": 467450
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 2.139626540139282e-05,
+      "loss": 0.7027,
+      "step": 467460
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 2.139511747149307e-05,
+      "loss": 0.762,
+      "step": 467470
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 2.1393969541593325e-05,
+      "loss": 0.8119,
+      "step": 467480
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 2.1392821611693578e-05,
+      "loss": 0.699,
+      "step": 467490
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 2.139167368179383e-05,
+      "loss": 0.7716,
+      "step": 467500
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 2.139052575189408e-05,
+      "loss": 0.7416,
+      "step": 467510
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 2.1389377821994338e-05,
+      "loss": 0.7438,
+      "step": 467520
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 2.1388229892094587e-05,
+      "loss": 0.7733,
+      "step": 467530
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 2.138708196219484e-05,
+      "loss": 0.7049,
+      "step": 467540
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 2.138593403229509e-05,
+      "loss": 0.8234,
+      "step": 467550
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 2.1384786102395347e-05,
+      "loss": 0.7756,
+      "step": 467560
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 2.1383638172495597e-05,
+      "loss": 0.8609,
+      "step": 467570
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 2.1382490242595847e-05,
+      "loss": 0.8881,
+      "step": 467580
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 2.1381342312696103e-05,
+      "loss": 0.7716,
+      "step": 467590
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 2.1380194382796353e-05,
+      "loss": 0.6897,
+      "step": 467600
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 2.1379046452896607e-05,
+      "loss": 0.7325,
+      "step": 467610
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 2.137789852299686e-05,
+      "loss": 0.7594,
+      "step": 467620
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 2.1376750593097113e-05,
+      "loss": 0.8994,
+      "step": 467630
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 2.1375602663197363e-05,
+      "loss": 0.803,
+      "step": 467640
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 2.137445473329762e-05,
+      "loss": 0.7994,
+      "step": 467650
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 2.137330680339787e-05,
+      "loss": 0.7925,
+      "step": 467660
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 2.1372158873498123e-05,
+      "loss": 0.8065,
+      "step": 467670
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 2.1371010943598376e-05,
+      "loss": 0.8183,
+      "step": 467680
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 2.136986301369863e-05,
+      "loss": 0.7277,
+      "step": 467690
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 2.136871508379888e-05,
+      "loss": 0.783,
+      "step": 467700
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 2.1367567153899136e-05,
+      "loss": 0.7671,
+      "step": 467710
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 2.1366419223999385e-05,
+      "loss": 0.8073,
+      "step": 467720
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 2.1365271294099635e-05,
+      "loss": 0.7857,
+      "step": 467730
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 2.1364123364199892e-05,
+      "loss": 0.7596,
+      "step": 467740
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 2.1362975434300142e-05,
+      "loss": 0.8288,
+      "step": 467750
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 2.1361827504400395e-05,
+      "loss": 0.7905,
+      "step": 467760
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 2.1360679574500648e-05,
+      "loss": 0.8175,
+      "step": 467770
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 2.13595316446009e-05,
+      "loss": 0.8521,
+      "step": 467780
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 2.135838371470115e-05,
+      "loss": 0.8133,
+      "step": 467790
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 2.1357235784801408e-05,
+      "loss": 0.7218,
+      "step": 467800
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 2.1356087854901658e-05,
+      "loss": 0.853,
+      "step": 467810
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 2.135493992500191e-05,
+      "loss": 0.7234,
+      "step": 467820
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 2.1353791995102164e-05,
+      "loss": 0.7665,
+      "step": 467830
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 2.1352644065202418e-05,
+      "loss": 0.813,
+      "step": 467840
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 2.1351496135302667e-05,
+      "loss": 0.8406,
+      "step": 467850
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 2.1350348205402924e-05,
+      "loss": 0.7628,
+      "step": 467860
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 2.1349200275503174e-05,
+      "loss": 0.8332,
+      "step": 467870
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 2.1348052345603424e-05,
+      "loss": 0.7901,
+      "step": 467880
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 2.134690441570368e-05,
+      "loss": 0.7683,
+      "step": 467890
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 2.134575648580393e-05,
+      "loss": 0.7654,
+      "step": 467900
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 2.1344608555904183e-05,
+      "loss": 0.7781,
+      "step": 467910
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 2.1343460626004437e-05,
+      "loss": 0.7538,
+      "step": 467920
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 2.134231269610469e-05,
+      "loss": 0.7968,
+      "step": 467930
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 2.134116476620494e-05,
+      "loss": 0.8039,
+      "step": 467940
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 2.1340016836305196e-05,
+      "loss": 0.7916,
+      "step": 467950
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 2.1338868906405446e-05,
+      "loss": 0.8295,
+      "step": 467960
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 2.13377209765057e-05,
+      "loss": 0.7512,
+      "step": 467970
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 2.1336573046605953e-05,
+      "loss": 0.8104,
+      "step": 467980
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 2.1335425116706206e-05,
+      "loss": 0.8192,
+      "step": 467990
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 2.1334277186806456e-05,
+      "loss": 0.8529,
+      "step": 468000
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 2.1333129256906713e-05,
+      "loss": 0.8803,
+      "step": 468010
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 2.1331981327006962e-05,
+      "loss": 0.8081,
+      "step": 468020
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 2.1330833397107212e-05,
+      "loss": 0.7848,
+      "step": 468030
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 2.132968546720747e-05,
+      "loss": 0.779,
+      "step": 468040
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 2.132853753730772e-05,
+      "loss": 0.8045,
+      "step": 468050
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 2.1327389607407972e-05,
+      "loss": 0.7255,
+      "step": 468060
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 2.1326241677508225e-05,
+      "loss": 0.7731,
+      "step": 468070
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 2.132509374760848e-05,
+      "loss": 0.7599,
+      "step": 468080
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 2.1323945817708728e-05,
+      "loss": 0.7146,
+      "step": 468090
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 2.1322797887808985e-05,
+      "loss": 0.7176,
+      "step": 468100
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 2.1321649957909235e-05,
+      "loss": 0.8017,
+      "step": 468110
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 2.1320502028009488e-05,
+      "loss": 0.7126,
+      "step": 468120
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 2.131935409810974e-05,
+      "loss": 0.7983,
+      "step": 468130
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 2.1318206168209995e-05,
+      "loss": 0.7719,
+      "step": 468140
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 2.1317058238310244e-05,
+      "loss": 0.7352,
+      "step": 468150
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 2.13159103084105e-05,
+      "loss": 0.8452,
+      "step": 468160
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 2.131476237851075e-05,
+      "loss": 0.8662,
+      "step": 468170
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 2.1313614448611e-05,
+      "loss": 0.8013,
+      "step": 468180
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 2.1312466518711257e-05,
+      "loss": 0.7332,
+      "step": 468190
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 2.1311318588811507e-05,
+      "loss": 0.8442,
+      "step": 468200
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 2.131017065891176e-05,
+      "loss": 0.8295,
+      "step": 468210
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 2.130902272901201e-05,
+      "loss": 0.8667,
+      "step": 468220
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 2.1307874799112267e-05,
+      "loss": 0.7498,
+      "step": 468230
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 2.1306726869212517e-05,
+      "loss": 0.855,
+      "step": 468240
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 2.130557893931277e-05,
+      "loss": 0.7073,
+      "step": 468250
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 2.1304431009413023e-05,
+      "loss": 0.7863,
+      "step": 468260
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 2.1303283079513277e-05,
+      "loss": 0.8121,
+      "step": 468270
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 2.1302135149613526e-05,
+      "loss": 0.7449,
+      "step": 468280
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 2.1300987219713783e-05,
+      "loss": 0.7989,
+      "step": 468290
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 2.1299839289814033e-05,
+      "loss": 0.8601,
+      "step": 468300
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 2.1298691359914283e-05,
+      "loss": 0.7494,
+      "step": 468310
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 2.129754343001454e-05,
+      "loss": 0.9043,
+      "step": 468320
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 2.129639550011479e-05,
+      "loss": 0.8418,
+      "step": 468330
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 2.1295247570215042e-05,
+      "loss": 0.7809,
+      "step": 468340
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 2.1294099640315296e-05,
+      "loss": 0.885,
+      "step": 468350
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 2.129295171041555e-05,
+      "loss": 0.8368,
+      "step": 468360
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 2.12918037805158e-05,
+      "loss": 0.7527,
+      "step": 468370
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 2.1290655850616055e-05,
+      "loss": 0.798,
+      "step": 468380
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 2.1289507920716305e-05,
+      "loss": 0.8363,
+      "step": 468390
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 2.128835999081656e-05,
+      "loss": 0.8282,
+      "step": 468400
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 2.1287212060916812e-05,
+      "loss": 0.8357,
+      "step": 468410
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 2.1286064131017065e-05,
+      "loss": 0.8969,
+      "step": 468420
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 2.1284916201117315e-05,
+      "loss": 0.7055,
+      "step": 468430
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 2.128376827121757e-05,
+      "loss": 0.7988,
+      "step": 468440
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 2.128262034131782e-05,
+      "loss": 0.8069,
+      "step": 468450
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 2.128147241141807e-05,
+      "loss": 0.7001,
+      "step": 468460
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 2.1280324481518328e-05,
+      "loss": 0.7921,
+      "step": 468470
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 2.1279176551618578e-05,
+      "loss": 0.7782,
+      "step": 468480
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 2.127802862171883e-05,
+      "loss": 0.7316,
+      "step": 468490
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 2.1276880691819084e-05,
+      "loss": 0.7816,
+      "step": 468500
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 2.1275732761919337e-05,
+      "loss": 0.8411,
+      "step": 468510
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 2.1274584832019587e-05,
+      "loss": 0.7378,
+      "step": 468520
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 2.1273436902119844e-05,
+      "loss": 0.8398,
+      "step": 468530
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 2.1272288972220094e-05,
+      "loss": 0.7911,
+      "step": 468540
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 2.1271141042320347e-05,
+      "loss": 0.8514,
+      "step": 468550
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 2.12699931124206e-05,
+      "loss": 0.8075,
+      "step": 468560
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 2.1268845182520853e-05,
+      "loss": 0.7555,
+      "step": 468570
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 2.1267697252621103e-05,
+      "loss": 0.8295,
+      "step": 468580
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 2.126654932272136e-05,
+      "loss": 0.7668,
+      "step": 468590
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 2.126540139282161e-05,
+      "loss": 0.7585,
+      "step": 468600
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 2.126425346292186e-05,
+      "loss": 0.8192,
+      "step": 468610
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 2.1263105533022116e-05,
+      "loss": 0.8568,
+      "step": 468620
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 2.1261957603122366e-05,
+      "loss": 0.7786,
+      "step": 468630
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 2.126080967322262e-05,
+      "loss": 0.7837,
+      "step": 468640
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 2.1259661743322873e-05,
+      "loss": 0.8555,
+      "step": 468650
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 2.1258513813423126e-05,
+      "loss": 0.7953,
+      "step": 468660
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 2.1257365883523376e-05,
+      "loss": 0.8128,
+      "step": 468670
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 2.1256217953623632e-05,
+      "loss": 0.8129,
+      "step": 468680
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 2.1255070023723882e-05,
+      "loss": 0.8095,
+      "step": 468690
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 2.1253922093824135e-05,
+      "loss": 0.8009,
+      "step": 468700
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 2.125277416392439e-05,
+      "loss": 0.8598,
+      "step": 468710
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 2.1251626234024642e-05,
+      "loss": 0.8122,
+      "step": 468720
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 2.1250478304124892e-05,
+      "loss": 0.784,
+      "step": 468730
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 2.124933037422515e-05,
+      "loss": 0.7617,
+      "step": 468740
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 2.1248182444325398e-05,
+      "loss": 0.7958,
+      "step": 468750
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 2.1247034514425648e-05,
+      "loss": 0.8068,
+      "step": 468760
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 2.1245886584525905e-05,
+      "loss": 0.7204,
+      "step": 468770
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 2.1244738654626155e-05,
+      "loss": 0.843,
+      "step": 468780
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 2.1243590724726408e-05,
+      "loss": 0.8204,
+      "step": 468790
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 2.124244279482666e-05,
+      "loss": 0.8002,
+      "step": 468800
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 2.1241294864926914e-05,
+      "loss": 0.8883,
+      "step": 468810
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 2.1240146935027164e-05,
+      "loss": 0.8195,
+      "step": 468820
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 2.123899900512742e-05,
+      "loss": 0.7971,
+      "step": 468830
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 2.123785107522767e-05,
+      "loss": 0.7653,
+      "step": 468840
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 2.1236703145327924e-05,
+      "loss": 0.7779,
+      "step": 468850
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 2.1235555215428177e-05,
+      "loss": 0.8622,
+      "step": 468860
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 2.123440728552843e-05,
+      "loss": 0.8563,
+      "step": 468870
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 2.123325935562868e-05,
+      "loss": 0.8345,
+      "step": 468880
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 2.1232111425728937e-05,
+      "loss": 0.7314,
+      "step": 468890
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 2.1230963495829187e-05,
+      "loss": 0.7828,
+      "step": 468900
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 2.1229815565929437e-05,
+      "loss": 0.7941,
+      "step": 468910
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 2.122866763602969e-05,
+      "loss": 0.7825,
+      "step": 468920
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 2.1227519706129943e-05,
+      "loss": 0.8328,
+      "step": 468930
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 2.1226371776230196e-05,
+      "loss": 0.8012,
+      "step": 468940
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 2.1225223846330446e-05,
+      "loss": 0.7194,
+      "step": 468950
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 2.1224075916430703e-05,
+      "loss": 0.7761,
+      "step": 468960
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 2.1222927986530953e-05,
+      "loss": 0.9024,
+      "step": 468970
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 2.1221780056631206e-05,
+      "loss": 0.757,
+      "step": 468980
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 2.122063212673146e-05,
+      "loss": 0.8505,
+      "step": 468990
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 2.1219484196831712e-05,
+      "loss": 0.7571,
+      "step": 469000
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 2.1218336266931962e-05,
+      "loss": 0.8314,
+      "step": 469010
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 2.121718833703222e-05,
+      "loss": 0.7667,
+      "step": 469020
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 2.121604040713247e-05,
+      "loss": 0.721,
+      "step": 469030
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 2.121489247723272e-05,
+      "loss": 0.7872,
+      "step": 469040
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 2.1213744547332975e-05,
+      "loss": 0.8195,
+      "step": 469050
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 2.1212596617433225e-05,
+      "loss": 0.8532,
+      "step": 469060
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 2.1211448687533478e-05,
+      "loss": 0.8931,
+      "step": 469070
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 2.121030075763373e-05,
+      "loss": 0.8206,
+      "step": 469080
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 2.1209152827733985e-05,
+      "loss": 0.8005,
+      "step": 469090
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 2.1208004897834235e-05,
+      "loss": 0.7389,
+      "step": 469100
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 2.120685696793449e-05,
+      "loss": 0.78,
+      "step": 469110
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 2.120570903803474e-05,
+      "loss": 0.765,
+      "step": 469120
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 2.1204561108134994e-05,
+      "loss": 0.8372,
+      "step": 469130
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 2.1203413178235248e-05,
+      "loss": 0.7659,
+      "step": 469140
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 2.12022652483355e-05,
+      "loss": 0.8064,
+      "step": 469150
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 2.120111731843575e-05,
+      "loss": 0.7087,
+      "step": 469160
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 2.1199969388536007e-05,
+      "loss": 0.7555,
+      "step": 469170
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 2.1198821458636257e-05,
+      "loss": 0.7788,
+      "step": 469180
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 2.1197673528736507e-05,
+      "loss": 0.8194,
+      "step": 469190
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 2.1196525598836764e-05,
+      "loss": 0.7393,
+      "step": 469200
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 2.1195377668937014e-05,
+      "loss": 0.7683,
+      "step": 469210
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 2.1194229739037267e-05,
+      "loss": 0.7984,
+      "step": 469220
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 2.119308180913752e-05,
+      "loss": 0.7101,
+      "step": 469230
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 2.1191933879237773e-05,
+      "loss": 0.8292,
+      "step": 469240
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 2.1190785949338023e-05,
+      "loss": 0.886,
+      "step": 469250
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 2.118963801943828e-05,
+      "loss": 0.8272,
+      "step": 469260
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 2.118849008953853e-05,
+      "loss": 0.7957,
+      "step": 469270
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 2.1187342159638783e-05,
+      "loss": 0.837,
+      "step": 469280
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 2.1186194229739036e-05,
+      "loss": 0.7355,
+      "step": 469290
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 2.118504629983929e-05,
+      "loss": 0.8349,
+      "step": 469300
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 2.118389836993954e-05,
+      "loss": 0.7197,
+      "step": 469310
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 2.1182750440039796e-05,
+      "loss": 0.8371,
+      "step": 469320
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 2.1181602510140046e-05,
+      "loss": 0.8028,
+      "step": 469330
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 2.1180454580240296e-05,
+      "loss": 0.8333,
+      "step": 469340
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 2.1179306650340552e-05,
+      "loss": 0.776,
+      "step": 469350
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 2.1178158720440802e-05,
+      "loss": 0.7275,
+      "step": 469360
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 2.1177010790541055e-05,
+      "loss": 0.8317,
+      "step": 469370
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 2.117586286064131e-05,
+      "loss": 0.7393,
+      "step": 469380
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 2.1174714930741562e-05,
+      "loss": 0.803,
+      "step": 469390
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 2.117356700084181e-05,
+      "loss": 0.8168,
+      "step": 469400
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 2.1172419070942068e-05,
+      "loss": 0.8409,
+      "step": 469410
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 2.1171271141042318e-05,
+      "loss": 0.7259,
+      "step": 469420
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 2.117012321114257e-05,
+      "loss": 0.8732,
+      "step": 469430
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 2.1168975281242825e-05,
+      "loss": 0.7635,
+      "step": 469440
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 2.1167827351343078e-05,
+      "loss": 0.7584,
+      "step": 469450
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 2.1166679421443328e-05,
+      "loss": 0.8129,
+      "step": 469460
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 2.1165531491543584e-05,
+      "loss": 0.7929,
+      "step": 469470
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 2.1164383561643834e-05,
+      "loss": 0.7245,
+      "step": 469480
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 2.1163235631744084e-05,
+      "loss": 0.7931,
+      "step": 469490
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 2.116208770184434e-05,
+      "loss": 0.8092,
+      "step": 469500
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 2.116093977194459e-05,
+      "loss": 0.7747,
+      "step": 469510
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 2.1159791842044844e-05,
+      "loss": 0.7533,
+      "step": 469520
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 2.1158643912145097e-05,
+      "loss": 0.7437,
+      "step": 469530
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 2.115749598224535e-05,
+      "loss": 0.7598,
+      "step": 469540
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 2.11563480523456e-05,
+      "loss": 0.7824,
+      "step": 469550
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 2.1155200122445857e-05,
+      "loss": 0.7116,
+      "step": 469560
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 2.1154052192546107e-05,
+      "loss": 0.7498,
+      "step": 469570
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 2.115290426264636e-05,
+      "loss": 0.7696,
+      "step": 469580
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 2.115175633274661e-05,
+      "loss": 0.8104,
+      "step": 469590
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 2.1150608402846866e-05,
+      "loss": 0.7683,
+      "step": 469600
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 2.1149460472947116e-05,
+      "loss": 0.7624,
+      "step": 469610
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 2.1148312543047366e-05,
+      "loss": 0.7738,
+      "step": 469620
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 2.1147164613147623e-05,
+      "loss": 0.7344,
+      "step": 469630
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 2.1146016683247872e-05,
+      "loss": 0.8039,
+      "step": 469640
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 2.1144868753348126e-05,
+      "loss": 0.7585,
+      "step": 469650
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 2.114372082344838e-05,
+      "loss": 0.7546,
+      "step": 469660
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 2.1142572893548632e-05,
+      "loss": 0.7716,
+      "step": 469670
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 2.1141424963648882e-05,
+      "loss": 0.8235,
+      "step": 469680
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 2.114027703374914e-05,
+      "loss": 0.8417,
+      "step": 469690
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 2.113912910384939e-05,
+      "loss": 0.8211,
+      "step": 469700
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 2.1137981173949642e-05,
+      "loss": 0.813,
+      "step": 469710
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 2.1136833244049895e-05,
+      "loss": 0.8037,
+      "step": 469720
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 2.1135685314150148e-05,
+      "loss": 0.7877,
+      "step": 469730
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 2.1134537384250398e-05,
+      "loss": 0.6862,
+      "step": 469740
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 2.1133389454350655e-05,
+      "loss": 0.7556,
+      "step": 469750
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 2.1132241524450905e-05,
+      "loss": 0.7976,
+      "step": 469760
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 2.1131093594551154e-05,
+      "loss": 0.7636,
+      "step": 469770
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 2.112994566465141e-05,
+      "loss": 0.7508,
+      "step": 469780
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 2.112879773475166e-05,
+      "loss": 0.8256,
+      "step": 469790
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 2.1127649804851914e-05,
+      "loss": 0.7584,
+      "step": 469800
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 2.1126501874952167e-05,
+      "loss": 0.8127,
+      "step": 469810
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 2.112535394505242e-05,
+      "loss": 0.7064,
+      "step": 469820
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 2.112420601515267e-05,
+      "loss": 0.8456,
+      "step": 469830
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 2.1123058085252927e-05,
+      "loss": 0.6862,
+      "step": 469840
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 2.1121910155353177e-05,
+      "loss": 0.7924,
+      "step": 469850
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 2.112076222545343e-05,
+      "loss": 0.8139,
+      "step": 469860
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 2.1119614295553683e-05,
+      "loss": 0.7639,
+      "step": 469870
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 2.1118466365653937e-05,
+      "loss": 0.783,
+      "step": 469880
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 2.1117318435754187e-05,
+      "loss": 0.7796,
+      "step": 469890
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 2.1116170505854443e-05,
+      "loss": 0.823,
+      "step": 469900
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 2.1115022575954693e-05,
+      "loss": 0.7686,
+      "step": 469910
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 2.1113874646054943e-05,
+      "loss": 0.7417,
+      "step": 469920
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 2.11127267161552e-05,
+      "loss": 0.8302,
+      "step": 469930
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 2.111157878625545e-05,
+      "loss": 0.8438,
+      "step": 469940
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 2.1110430856355703e-05,
+      "loss": 0.7936,
+      "step": 469950
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 2.1109282926455956e-05,
+      "loss": 0.8861,
+      "step": 469960
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 2.110813499655621e-05,
+      "loss": 0.7709,
+      "step": 469970
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 2.110698706665646e-05,
+      "loss": 0.7558,
+      "step": 469980
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 2.1105839136756716e-05,
+      "loss": 0.776,
+      "step": 469990
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 2.1104691206856965e-05,
+      "loss": 0.8756,
+      "step": 470000
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 2.110354327695722e-05,
+      "loss": 0.8318,
+      "step": 470010
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 2.1102395347057472e-05,
+      "loss": 0.8369,
+      "step": 470020
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 2.1101247417157725e-05,
+      "loss": 0.7788,
+      "step": 470030
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 2.1100099487257975e-05,
+      "loss": 0.8337,
+      "step": 470040
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 2.109895155735823e-05,
+      "loss": 0.8374,
+      "step": 470050
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 2.109780362745848e-05,
+      "loss": 0.8403,
+      "step": 470060
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 2.109665569755873e-05,
+      "loss": 0.8191,
+      "step": 470070
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 2.1095507767658988e-05,
+      "loss": 0.6819,
+      "step": 470080
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 2.1094359837759238e-05,
+      "loss": 0.8748,
+      "step": 470090
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 2.109321190785949e-05,
+      "loss": 0.829,
+      "step": 470100
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 2.1092063977959744e-05,
+      "loss": 0.7751,
+      "step": 470110
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 2.1090916048059998e-05,
+      "loss": 0.7595,
+      "step": 470120
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 2.1089768118160247e-05,
+      "loss": 0.7305,
+      "step": 470130
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 2.1088620188260504e-05,
+      "loss": 0.7545,
+      "step": 470140
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 2.1087472258360754e-05,
+      "loss": 0.787,
+      "step": 470150
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 2.1086324328461007e-05,
+      "loss": 0.7795,
+      "step": 470160
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 2.108517639856126e-05,
+      "loss": 0.8058,
+      "step": 470170
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 2.1084028468661514e-05,
+      "loss": 0.7895,
+      "step": 470180
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 2.1082880538761764e-05,
+      "loss": 0.7619,
+      "step": 470190
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 2.108173260886202e-05,
+      "loss": 0.7938,
+      "step": 470200
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 2.108058467896227e-05,
+      "loss": 0.7544,
+      "step": 470210
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 2.107943674906252e-05,
+      "loss": 0.8114,
+      "step": 470220
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 2.1078288819162776e-05,
+      "loss": 0.7247,
+      "step": 470230
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 2.1077140889263026e-05,
+      "loss": 0.8385,
+      "step": 470240
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 2.107599295936328e-05,
+      "loss": 0.8571,
+      "step": 470250
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 2.1074845029463533e-05,
+      "loss": 0.7993,
+      "step": 470260
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 2.1073697099563786e-05,
+      "loss": 0.7918,
+      "step": 470270
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 2.1072549169664036e-05,
+      "loss": 0.7996,
+      "step": 470280
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 2.107140123976429e-05,
+      "loss": 0.8528,
+      "step": 470290
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 2.1070253309864542e-05,
+      "loss": 0.6925,
+      "step": 470300
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 2.1069105379964796e-05,
+      "loss": 0.8557,
+      "step": 470310
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 2.1067957450065046e-05,
+      "loss": 0.8054,
+      "step": 470320
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 2.1066809520165302e-05,
+      "loss": 0.7769,
+      "step": 470330
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 2.1065661590265552e-05,
+      "loss": 0.72,
+      "step": 470340
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 2.1064513660365802e-05,
+      "loss": 0.7562,
+      "step": 470350
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 2.106336573046606e-05,
+      "loss": 0.7934,
+      "step": 470360
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 2.106221780056631e-05,
+      "loss": 0.8141,
+      "step": 470370
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 2.106106987066656e-05,
+      "loss": 0.7266,
+      "step": 470380
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 2.1059921940766815e-05,
+      "loss": 0.6165,
+      "step": 470390
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 2.1058774010867068e-05,
+      "loss": 0.7721,
+      "step": 470400
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 2.1057626080967318e-05,
+      "loss": 0.803,
+      "step": 470410
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 2.1056478151067575e-05,
+      "loss": 0.8916,
+      "step": 470420
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 2.1055330221167824e-05,
+      "loss": 0.8602,
+      "step": 470430
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 2.1054182291268078e-05,
+      "loss": 0.8174,
+      "step": 470440
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 2.105303436136833e-05,
+      "loss": 0.8128,
+      "step": 470450
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 2.1051886431468584e-05,
+      "loss": 0.7499,
+      "step": 470460
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 2.1050738501568834e-05,
+      "loss": 0.8665,
+      "step": 470470
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 2.104959057166909e-05,
+      "loss": 0.7596,
+      "step": 470480
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 2.104844264176934e-05,
+      "loss": 0.8162,
+      "step": 470490
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 2.104729471186959e-05,
+      "loss": 0.7239,
+      "step": 470500
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 2.1046146781969847e-05,
+      "loss": 0.786,
+      "step": 470510
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 2.1044998852070097e-05,
+      "loss": 0.779,
+      "step": 470520
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 2.104385092217035e-05,
+      "loss": 0.746,
+      "step": 470530
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 2.1042702992270603e-05,
+      "loss": 0.7688,
+      "step": 470540
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 2.1041555062370857e-05,
+      "loss": 0.7956,
+      "step": 470550
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 2.1040407132471106e-05,
+      "loss": 0.7847,
+      "step": 470560
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 2.1039259202571363e-05,
+      "loss": 0.6858,
+      "step": 470570
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 2.1038111272671613e-05,
+      "loss": 0.8782,
+      "step": 470580
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 2.1036963342771866e-05,
+      "loss": 0.8078,
+      "step": 470590
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 2.103581541287212e-05,
+      "loss": 0.7724,
+      "step": 470600
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 2.1034667482972373e-05,
+      "loss": 0.809,
+      "step": 470610
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 2.1033519553072622e-05,
+      "loss": 0.7802,
+      "step": 470620
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 2.103237162317288e-05,
+      "loss": 0.8082,
+      "step": 470630
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 2.103122369327313e-05,
+      "loss": 0.82,
+      "step": 470640
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 2.103007576337338e-05,
+      "loss": 0.7629,
+      "step": 470650
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 2.1028927833473635e-05,
+      "loss": 0.7768,
+      "step": 470660
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 2.1027779903573885e-05,
+      "loss": 0.736,
+      "step": 470670
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 2.102663197367414e-05,
+      "loss": 0.7239,
+      "step": 470680
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 2.1025484043774392e-05,
+      "loss": 0.8123,
+      "step": 470690
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 2.1024336113874645e-05,
+      "loss": 0.8236,
+      "step": 470700
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 2.1023188183974895e-05,
+      "loss": 0.7377,
+      "step": 470710
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 2.102204025407515e-05,
+      "loss": 0.7352,
+      "step": 470720
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 2.10208923241754e-05,
+      "loss": 0.7475,
+      "step": 470730
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 2.1019744394275655e-05,
+      "loss": 0.7572,
+      "step": 470740
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 2.1018596464375908e-05,
+      "loss": 0.8027,
+      "step": 470750
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 2.101744853447616e-05,
+      "loss": 0.753,
+      "step": 470760
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 2.101630060457641e-05,
+      "loss": 0.7796,
+      "step": 470770
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 2.1015152674676668e-05,
+      "loss": 0.8228,
+      "step": 470780
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 2.1014004744776917e-05,
+      "loss": 0.7151,
+      "step": 470790
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 2.1012856814877167e-05,
+      "loss": 0.7805,
+      "step": 470800
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 2.1011708884977424e-05,
+      "loss": 0.8378,
+      "step": 470810
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 2.1010560955077674e-05,
+      "loss": 0.7807,
+      "step": 470820
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 2.1009413025177927e-05,
+      "loss": 0.8101,
+      "step": 470830
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 2.100826509527818e-05,
+      "loss": 0.8218,
+      "step": 470840
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 2.1007117165378433e-05,
+      "loss": 0.7098,
+      "step": 470850
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 2.1005969235478683e-05,
+      "loss": 0.7675,
+      "step": 470860
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 2.100482130557894e-05,
+      "loss": 0.7452,
+      "step": 470870
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 2.100367337567919e-05,
+      "loss": 0.7381,
+      "step": 470880
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 2.1002525445779443e-05,
+      "loss": 0.7488,
+      "step": 470890
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 2.1001377515879696e-05,
+      "loss": 0.8025,
+      "step": 470900
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 2.100022958597995e-05,
+      "loss": 0.7396,
+      "step": 470910
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 2.09990816560802e-05,
+      "loss": 0.7412,
+      "step": 470920
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 2.0997933726180456e-05,
+      "loss": 0.8232,
+      "step": 470930
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 2.0996785796280706e-05,
+      "loss": 0.8133,
+      "step": 470940
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 2.0995637866380956e-05,
+      "loss": 0.8014,
+      "step": 470950
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 2.099448993648121e-05,
+      "loss": 0.8303,
+      "step": 470960
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 2.0993342006581462e-05,
+      "loss": 0.7073,
+      "step": 470970
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 2.0992194076681715e-05,
+      "loss": 0.7601,
+      "step": 470980
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 2.0991046146781965e-05,
+      "loss": 0.793,
+      "step": 470990
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 2.0989898216882222e-05,
+      "loss": 0.8336,
+      "step": 471000
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 2.0988750286982472e-05,
+      "loss": 0.7498,
+      "step": 471010
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 2.0987602357082725e-05,
+      "loss": 0.8155,
+      "step": 471020
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 2.0986454427182978e-05,
+      "loss": 0.7765,
+      "step": 471030
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 2.098530649728323e-05,
+      "loss": 0.8131,
+      "step": 471040
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 2.098415856738348e-05,
+      "loss": 0.7692,
+      "step": 471050
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 2.0983010637483738e-05,
+      "loss": 0.8059,
+      "step": 471060
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 2.0981862707583988e-05,
+      "loss": 0.7905,
+      "step": 471070
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 2.0980714777684238e-05,
+      "loss": 0.7342,
+      "step": 471080
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 2.0979566847784494e-05,
+      "loss": 0.6992,
+      "step": 471090
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 2.0978418917884744e-05,
+      "loss": 0.7432,
+      "step": 471100
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 2.0977270987984997e-05,
+      "loss": 0.6976,
+      "step": 471110
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 2.097612305808525e-05,
+      "loss": 0.7308,
+      "step": 471120
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 2.0974975128185504e-05,
+      "loss": 0.8423,
+      "step": 471130
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 2.0973827198285754e-05,
+      "loss": 0.7906,
+      "step": 471140
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 2.097267926838601e-05,
+      "loss": 0.8006,
+      "step": 471150
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 2.097153133848626e-05,
+      "loss": 0.8185,
+      "step": 471160
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 2.0970383408586514e-05,
+      "loss": 0.784,
+      "step": 471170
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 2.0969235478686767e-05,
+      "loss": 0.7912,
+      "step": 471180
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 2.096808754878702e-05,
+      "loss": 0.7813,
+      "step": 471190
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 2.096693961888727e-05,
+      "loss": 0.7183,
+      "step": 471200
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 2.0965791688987526e-05,
+      "loss": 0.7235,
+      "step": 471210
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 2.0964643759087776e-05,
+      "loss": 0.7592,
+      "step": 471220
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 2.0963495829188026e-05,
+      "loss": 0.7741,
+      "step": 471230
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 2.0962347899288283e-05,
+      "loss": 0.763,
+      "step": 471240
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 2.0961199969388533e-05,
+      "loss": 0.6796,
+      "step": 471250
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 2.0960052039488786e-05,
+      "loss": 0.6922,
+      "step": 471260
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 2.095890410958904e-05,
+      "loss": 0.8596,
+      "step": 471270
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 2.0957756179689292e-05,
+      "loss": 0.6986,
+      "step": 471280
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 2.0956608249789542e-05,
+      "loss": 0.7869,
+      "step": 471290
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 2.09554603198898e-05,
+      "loss": 0.7506,
+      "step": 471300
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 2.095431238999005e-05,
+      "loss": 0.7365,
+      "step": 471310
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 2.0953164460090302e-05,
+      "loss": 0.7646,
+      "step": 471320
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 2.0952016530190555e-05,
+      "loss": 0.8447,
+      "step": 471330
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 2.095086860029081e-05,
+      "loss": 0.8449,
+      "step": 471340
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 2.094972067039106e-05,
+      "loss": 0.7847,
+      "step": 471350
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 2.0948572740491315e-05,
+      "loss": 0.8072,
+      "step": 471360
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 2.0947424810591565e-05,
+      "loss": 0.7846,
+      "step": 471370
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 2.0946276880691815e-05,
+      "loss": 0.8139,
+      "step": 471380
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 2.094512895079207e-05,
+      "loss": 0.7859,
+      "step": 471390
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 2.094398102089232e-05,
+      "loss": 0.8337,
+      "step": 471400
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 2.0942833090992574e-05,
+      "loss": 0.7641,
+      "step": 471410
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 2.0941685161092828e-05,
+      "loss": 0.8529,
+      "step": 471420
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 2.094053723119308e-05,
+      "loss": 0.8773,
+      "step": 471430
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 2.093938930129333e-05,
+      "loss": 0.8096,
+      "step": 471440
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 2.0938241371393587e-05,
+      "loss": 0.8003,
+      "step": 471450
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 2.0937093441493837e-05,
+      "loss": 0.8232,
+      "step": 471460
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 2.093594551159409e-05,
+      "loss": 0.758,
+      "step": 471470
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 2.0934797581694344e-05,
+      "loss": 0.7939,
+      "step": 471480
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 2.0933649651794597e-05,
+      "loss": 0.83,
+      "step": 471490
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 2.0932501721894847e-05,
+      "loss": 0.8062,
+      "step": 471500
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 2.0931353791995103e-05,
+      "loss": 0.786,
+      "step": 471510
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 2.0930205862095353e-05,
+      "loss": 0.8192,
+      "step": 471520
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 2.0929057932195603e-05,
+      "loss": 0.8164,
+      "step": 471530
+    },
+    {
+      "epoch": 21.63,
+      "learning_rate": 2.092791000229586e-05,
+      "loss": 0.7791,
+      "step": 471540
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 2.092676207239611e-05,
+      "loss": 0.8333,
+      "step": 471550
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 2.0925614142496363e-05,
+      "loss": 0.8242,
+      "step": 471560
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 2.0924466212596616e-05,
+      "loss": 0.8432,
+      "step": 471570
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 2.092331828269687e-05,
+      "loss": 0.7607,
+      "step": 471580
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 2.092217035279712e-05,
+      "loss": 0.7617,
+      "step": 471590
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 2.0921022422897376e-05,
+      "loss": 0.7691,
+      "step": 471600
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 2.0919874492997626e-05,
+      "loss": 0.8294,
+      "step": 471610
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 2.091872656309788e-05,
+      "loss": 0.7346,
+      "step": 471620
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 2.0917578633198132e-05,
+      "loss": 0.8315,
+      "step": 471630
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 2.0916430703298385e-05,
+      "loss": 0.8032,
+      "step": 471640
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 2.0915282773398635e-05,
+      "loss": 0.8214,
+      "step": 471650
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 2.0914134843498885e-05,
+      "loss": 0.7749,
+      "step": 471660
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 2.0912986913599142e-05,
+      "loss": 0.8246,
+      "step": 471670
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 2.091183898369939e-05,
+      "loss": 0.8356,
+      "step": 471680
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 2.0910691053799645e-05,
+      "loss": 0.768,
+      "step": 471690
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 2.0909543123899898e-05,
+      "loss": 0.8433,
+      "step": 471700
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 2.090839519400015e-05,
+      "loss": 0.7137,
+      "step": 471710
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 2.09072472641004e-05,
+      "loss": 0.8121,
+      "step": 471720
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 2.0906099334200658e-05,
+      "loss": 0.7846,
+      "step": 471730
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 2.0904951404300908e-05,
+      "loss": 0.7108,
+      "step": 471740
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 2.090380347440116e-05,
+      "loss": 0.7165,
+      "step": 471750
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 2.0902655544501414e-05,
+      "loss": 0.7562,
+      "step": 471760
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 2.0901507614601667e-05,
+      "loss": 0.781,
+      "step": 471770
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 2.0900359684701917e-05,
+      "loss": 0.8569,
+      "step": 471780
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 2.0899211754802174e-05,
+      "loss": 0.8678,
+      "step": 471790
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 2.0898063824902424e-05,
+      "loss": 0.7988,
+      "step": 471800
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 2.0896915895002674e-05,
+      "loss": 0.732,
+      "step": 471810
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 2.089576796510293e-05,
+      "loss": 0.7725,
+      "step": 471820
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 2.089462003520318e-05,
+      "loss": 0.8362,
+      "step": 471830
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 2.0893472105303433e-05,
+      "loss": 0.8077,
+      "step": 471840
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 2.0892324175403687e-05,
+      "loss": 0.8205,
+      "step": 471850
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 2.089117624550394e-05,
+      "loss": 0.8149,
+      "step": 471860
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 2.089002831560419e-05,
+      "loss": 0.9014,
+      "step": 471870
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 2.0888880385704446e-05,
+      "loss": 0.8593,
+      "step": 471880
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 2.0887732455804696e-05,
+      "loss": 0.7057,
+      "step": 471890
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 2.088658452590495e-05,
+      "loss": 0.7283,
+      "step": 471900
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 2.0885436596005203e-05,
+      "loss": 0.733,
+      "step": 471910
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 2.0884288666105456e-05,
+      "loss": 0.8032,
+      "step": 471920
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 2.0883140736205706e-05,
+      "loss": 0.7989,
+      "step": 471930
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 2.0881992806305962e-05,
+      "loss": 0.7685,
+      "step": 471940
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 2.0880844876406212e-05,
+      "loss": 0.6942,
+      "step": 471950
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 2.0879696946506462e-05,
+      "loss": 0.7373,
+      "step": 471960
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 2.087854901660672e-05,
+      "loss": 0.8044,
+      "step": 471970
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 2.087740108670697e-05,
+      "loss": 0.8051,
+      "step": 471980
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 2.0876253156807222e-05,
+      "loss": 0.7831,
+      "step": 471990
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 2.0875105226907475e-05,
+      "loss": 0.7533,
+      "step": 472000
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 2.0873957297007728e-05,
+      "loss": 0.8264,
+      "step": 472010
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 2.0872809367107978e-05,
+      "loss": 0.8668,
+      "step": 472020
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 2.0871661437208235e-05,
+      "loss": 0.7323,
+      "step": 472030
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 2.0870513507308485e-05,
+      "loss": 0.768,
+      "step": 472040
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 2.0869365577408738e-05,
+      "loss": 0.7435,
+      "step": 472050
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 2.086821764750899e-05,
+      "loss": 0.7997,
+      "step": 472060
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 2.0867069717609244e-05,
+      "loss": 0.8282,
+      "step": 472070
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 2.0865921787709494e-05,
+      "loss": 0.8247,
+      "step": 472080
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 2.086477385780975e-05,
+      "loss": 0.8513,
+      "step": 472090
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 2.086362592791e-05,
+      "loss": 0.773,
+      "step": 472100
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 2.086247799801025e-05,
+      "loss": 0.7621,
+      "step": 472110
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 2.0861330068110507e-05,
+      "loss": 0.7587,
+      "step": 472120
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 2.0860182138210757e-05,
+      "loss": 0.8002,
+      "step": 472130
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 2.085903420831101e-05,
+      "loss": 0.7806,
+      "step": 472140
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 2.0857886278411264e-05,
+      "loss": 0.866,
+      "step": 472150
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 2.0856738348511517e-05,
+      "loss": 0.6994,
+      "step": 472160
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 2.0855590418611767e-05,
+      "loss": 0.8466,
+      "step": 472170
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 2.0854442488712023e-05,
+      "loss": 0.7526,
+      "step": 472180
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 2.0853294558812273e-05,
+      "loss": 0.7998,
+      "step": 472190
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 2.0852146628912526e-05,
+      "loss": 0.7847,
+      "step": 472200
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 2.085099869901278e-05,
+      "loss": 0.6888,
+      "step": 472210
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 2.0849850769113033e-05,
+      "loss": 0.7288,
+      "step": 472220
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 2.0848702839213283e-05,
+      "loss": 0.7978,
+      "step": 472230
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 2.084755490931354e-05,
+      "loss": 0.849,
+      "step": 472240
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 2.084640697941379e-05,
+      "loss": 0.9009,
+      "step": 472250
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 2.084525904951404e-05,
+      "loss": 0.8339,
+      "step": 472260
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 2.0844111119614296e-05,
+      "loss": 0.7584,
+      "step": 472270
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 2.0842963189714546e-05,
+      "loss": 0.851,
+      "step": 472280
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 2.08418152598148e-05,
+      "loss": 0.8857,
+      "step": 472290
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 2.0840667329915052e-05,
+      "loss": 0.7879,
+      "step": 472300
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 2.0839519400015305e-05,
+      "loss": 0.8109,
+      "step": 472310
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 2.0838371470115555e-05,
+      "loss": 0.8112,
+      "step": 472320
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 2.083722354021581e-05,
+      "loss": 0.7946,
+      "step": 472330
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 2.083607561031606e-05,
+      "loss": 0.7469,
+      "step": 472340
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 2.0834927680416315e-05,
+      "loss": 0.8086,
+      "step": 472350
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 2.0833779750516565e-05,
+      "loss": 0.9354,
+      "step": 472360
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 2.083263182061682e-05,
+      "loss": 0.752,
+      "step": 472370
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 2.083148389071707e-05,
+      "loss": 0.8064,
+      "step": 472380
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 2.083033596081732e-05,
+      "loss": 0.7974,
+      "step": 472390
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 2.0829188030917578e-05,
+      "loss": 0.7376,
+      "step": 472400
+    },
+    {
+      "epoch": 21.67,
+      "learning_rate": 2.0828040101017827e-05,
+      "loss": 0.7229,
+      "step": 472410
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 2.082689217111808e-05,
+      "loss": 0.8242,
+      "step": 472420
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 2.0825744241218334e-05,
+      "loss": 0.8421,
+      "step": 472430
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 2.0824596311318587e-05,
+      "loss": 0.723,
+      "step": 472440
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 2.0823448381418837e-05,
+      "loss": 0.8385,
+      "step": 472450
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 2.0822300451519094e-05,
+      "loss": 0.8551,
+      "step": 472460
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 2.0821152521619344e-05,
+      "loss": 0.8114,
+      "step": 472470
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 2.0820004591719597e-05,
+      "loss": 0.8084,
+      "step": 472480
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 2.081885666181985e-05,
+      "loss": 0.7869,
+      "step": 472490
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 2.0817708731920103e-05,
+      "loss": 0.8025,
+      "step": 472500
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 2.0816560802020353e-05,
+      "loss": 0.8693,
+      "step": 472510
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 2.081541287212061e-05,
+      "loss": 0.8006,
+      "step": 472520
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 2.081426494222086e-05,
+      "loss": 0.8167,
+      "step": 472530
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 2.081311701232111e-05,
+      "loss": 0.8388,
+      "step": 472540
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 2.0811969082421366e-05,
+      "loss": 0.7845,
+      "step": 472550
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 2.0810821152521616e-05,
+      "loss": 0.8054,
+      "step": 472560
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 2.080967322262187e-05,
+      "loss": 0.819,
+      "step": 472570
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 2.0808525292722122e-05,
+      "loss": 0.7531,
+      "step": 472580
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 2.0807377362822376e-05,
+      "loss": 0.7137,
+      "step": 472590
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 2.0806229432922626e-05,
+      "loss": 0.7906,
+      "step": 472600
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 2.0805081503022882e-05,
+      "loss": 0.7864,
+      "step": 472610
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 2.0803933573123132e-05,
+      "loss": 0.8332,
+      "step": 472620
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 2.0802785643223385e-05,
+      "loss": 0.8156,
+      "step": 472630
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 2.080163771332364e-05,
+      "loss": 0.8657,
+      "step": 472640
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 2.0800489783423892e-05,
+      "loss": 0.7764,
+      "step": 472650
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 2.079934185352414e-05,
+      "loss": 0.8406,
+      "step": 472660
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 2.0798193923624398e-05,
+      "loss": 0.7529,
+      "step": 472670
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 2.0797045993724648e-05,
+      "loss": 0.829,
+      "step": 472680
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 2.0795898063824898e-05,
+      "loss": 0.8314,
+      "step": 472690
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 2.0794750133925155e-05,
+      "loss": 0.774,
+      "step": 472700
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 2.0793602204025404e-05,
+      "loss": 0.7912,
+      "step": 472710
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 2.0792454274125658e-05,
+      "loss": 0.7943,
+      "step": 472720
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 2.079130634422591e-05,
+      "loss": 0.7913,
+      "step": 472730
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 2.0790158414326164e-05,
+      "loss": 0.7887,
+      "step": 472740
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 2.0789010484426414e-05,
+      "loss": 0.7547,
+      "step": 472750
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 2.078786255452667e-05,
+      "loss": 0.7844,
+      "step": 472760
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 2.078671462462692e-05,
+      "loss": 0.7539,
+      "step": 472770
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 2.0785566694727174e-05,
+      "loss": 0.803,
+      "step": 472780
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 2.0784418764827427e-05,
+      "loss": 0.8003,
+      "step": 472790
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 2.078327083492768e-05,
+      "loss": 0.7881,
+      "step": 472800
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 2.078212290502793e-05,
+      "loss": 0.8179,
+      "step": 472810
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 2.0780974975128187e-05,
+      "loss": 0.8489,
+      "step": 472820
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 2.0779827045228437e-05,
+      "loss": 0.8605,
+      "step": 472830
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 2.0778679115328686e-05,
+      "loss": 0.8752,
+      "step": 472840
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 2.0777531185428943e-05,
+      "loss": 0.847,
+      "step": 472850
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 2.0776383255529193e-05,
+      "loss": 0.8018,
+      "step": 472860
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 2.0775235325629446e-05,
+      "loss": 0.7803,
+      "step": 472870
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 2.07740873957297e-05,
+      "loss": 0.8009,
+      "step": 472880
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 2.0772939465829953e-05,
+      "loss": 0.773,
+      "step": 472890
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 2.0771791535930202e-05,
+      "loss": 0.8027,
+      "step": 472900
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 2.077064360603046e-05,
+      "loss": 0.8724,
+      "step": 472910
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 2.076949567613071e-05,
+      "loss": 0.7355,
+      "step": 472920
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 2.0768347746230962e-05,
+      "loss": 0.7874,
+      "step": 472930
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 2.0767199816331215e-05,
+      "loss": 0.7654,
+      "step": 472940
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 2.076605188643147e-05,
+      "loss": 0.7474,
+      "step": 472950
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 2.076490395653172e-05,
+      "loss": 0.8106,
+      "step": 472960
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 2.0763756026631975e-05,
+      "loss": 0.819,
+      "step": 472970
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 2.0762608096732225e-05,
+      "loss": 0.7775,
+      "step": 472980
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 2.0761460166832475e-05,
+      "loss": 0.8004,
+      "step": 472990
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 2.076031223693273e-05,
+      "loss": 0.8025,
+      "step": 473000
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 2.075916430703298e-05,
+      "loss": 0.8189,
+      "step": 473010
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 2.0758016377133235e-05,
+      "loss": 0.7726,
+      "step": 473020
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 2.0756868447233484e-05,
+      "loss": 0.7935,
+      "step": 473030
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 2.075572051733374e-05,
+      "loss": 0.7954,
+      "step": 473040
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 2.075457258743399e-05,
+      "loss": 0.8196,
+      "step": 473050
+    },
+    {
+      "epoch": 21.7,
+      "learning_rate": 2.0753424657534244e-05,
+      "loss": 0.8508,
+      "step": 473060
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 2.0752276727634497e-05,
+      "loss": 0.8741,
+      "step": 473070
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 2.075112879773475e-05,
+      "loss": 0.7599,
+      "step": 473080
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 2.0749980867835e-05,
+      "loss": 0.7983,
+      "step": 473090
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 2.0748832937935257e-05,
+      "loss": 0.732,
+      "step": 473100
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 2.0747685008035507e-05,
+      "loss": 0.7384,
+      "step": 473110
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 2.0746537078135757e-05,
+      "loss": 0.8192,
+      "step": 473120
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 2.0745389148236014e-05,
+      "loss": 0.8493,
+      "step": 473130
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 2.0744241218336263e-05,
+      "loss": 0.7967,
+      "step": 473140
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 2.0743093288436517e-05,
+      "loss": 0.7837,
+      "step": 473150
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 2.074194535853677e-05,
+      "loss": 0.7945,
+      "step": 473160
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 2.0740797428637023e-05,
+      "loss": 0.7923,
+      "step": 473170
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 2.0739649498737273e-05,
+      "loss": 0.7845,
+      "step": 473180
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 2.073850156883753e-05,
+      "loss": 0.8627,
+      "step": 473190
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 2.073735363893778e-05,
+      "loss": 0.7424,
+      "step": 473200
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 2.0736205709038033e-05,
+      "loss": 0.8225,
+      "step": 473210
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 2.0735057779138286e-05,
+      "loss": 0.864,
+      "step": 473220
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 2.073390984923854e-05,
+      "loss": 0.7924,
+      "step": 473230
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 2.073276191933879e-05,
+      "loss": 0.7761,
+      "step": 473240
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 2.0731613989439046e-05,
+      "loss": 0.7699,
+      "step": 473250
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 2.0730466059539295e-05,
+      "loss": 0.7514,
+      "step": 473260
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 2.0729318129639545e-05,
+      "loss": 0.8712,
+      "step": 473270
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 2.0728170199739802e-05,
+      "loss": 0.7884,
+      "step": 473280
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 2.0727022269840052e-05,
+      "loss": 0.7757,
+      "step": 473290
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 2.0725874339940305e-05,
+      "loss": 0.8566,
+      "step": 473300
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 2.072472641004056e-05,
+      "loss": 0.7628,
+      "step": 473310
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 2.072357848014081e-05,
+      "loss": 0.7688,
+      "step": 473320
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 2.072243055024106e-05,
+      "loss": 0.764,
+      "step": 473330
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 2.0721282620341318e-05,
+      "loss": 0.8137,
+      "step": 473340
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 2.0720134690441568e-05,
+      "loss": 0.7404,
+      "step": 473350
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 2.071898676054182e-05,
+      "loss": 0.7757,
+      "step": 473360
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 2.0717838830642074e-05,
+      "loss": 0.9063,
+      "step": 473370
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 2.0716690900742328e-05,
+      "loss": 0.6902,
+      "step": 473380
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 2.0715542970842577e-05,
+      "loss": 0.781,
+      "step": 473390
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 2.0714395040942834e-05,
+      "loss": 0.7743,
+      "step": 473400
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 2.0713247111043084e-05,
+      "loss": 0.7997,
+      "step": 473410
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 2.0712099181143334e-05,
+      "loss": 0.7278,
+      "step": 473420
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 2.071095125124359e-05,
+      "loss": 0.9101,
+      "step": 473430
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 2.070980332134384e-05,
+      "loss": 0.7559,
+      "step": 473440
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 2.0708655391444094e-05,
+      "loss": 0.7834,
+      "step": 473450
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 2.0707507461544347e-05,
+      "loss": 0.77,
+      "step": 473460
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 2.07063595316446e-05,
+      "loss": 0.7286,
+      "step": 473470
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 2.070521160174485e-05,
+      "loss": 0.8415,
+      "step": 473480
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 2.0704063671845107e-05,
+      "loss": 0.7726,
+      "step": 473490
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 2.0702915741945356e-05,
+      "loss": 0.8306,
+      "step": 473500
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 2.070176781204561e-05,
+      "loss": 0.7874,
+      "step": 473510
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 2.0700619882145863e-05,
+      "loss": 0.8256,
+      "step": 473520
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 2.0699471952246116e-05,
+      "loss": 0.797,
+      "step": 473530
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 2.0698324022346366e-05,
+      "loss": 0.8625,
+      "step": 473540
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 2.0697176092446623e-05,
+      "loss": 0.7388,
+      "step": 473550
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 2.0696028162546872e-05,
+      "loss": 0.7848,
+      "step": 473560
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 2.0694880232647122e-05,
+      "loss": 0.7996,
+      "step": 473570
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 2.069373230274738e-05,
+      "loss": 0.7149,
+      "step": 473580
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 2.069258437284763e-05,
+      "loss": 0.8145,
+      "step": 473590
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 2.0691436442947882e-05,
+      "loss": 0.7684,
+      "step": 473600
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 2.0690288513048135e-05,
+      "loss": 0.7724,
+      "step": 473610
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 2.068914058314839e-05,
+      "loss": 0.6704,
+      "step": 473620
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 2.068799265324864e-05,
+      "loss": 0.7434,
+      "step": 473630
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 2.0686844723348895e-05,
+      "loss": 0.8702,
+      "step": 473640
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 2.0685696793449145e-05,
+      "loss": 0.828,
+      "step": 473650
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 2.0684548863549398e-05,
+      "loss": 0.7952,
+      "step": 473660
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 2.068340093364965e-05,
+      "loss": 0.7947,
+      "step": 473670
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 2.0682253003749905e-05,
+      "loss": 0.8071,
+      "step": 473680
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 2.0681105073850154e-05,
+      "loss": 0.7425,
+      "step": 473690
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 2.0679957143950404e-05,
+      "loss": 0.7932,
+      "step": 473700
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 2.067880921405066e-05,
+      "loss": 0.7628,
+      "step": 473710
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 2.067766128415091e-05,
+      "loss": 0.7936,
+      "step": 473720
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 2.0676513354251164e-05,
+      "loss": 0.8075,
+      "step": 473730
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 2.0675365424351417e-05,
+      "loss": 0.7812,
+      "step": 473740
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 2.067421749445167e-05,
+      "loss": 0.7581,
+      "step": 473750
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 2.067306956455192e-05,
+      "loss": 0.744,
+      "step": 473760
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 2.0671921634652177e-05,
+      "loss": 0.7307,
+      "step": 473770
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 2.0670773704752427e-05,
+      "loss": 0.7857,
+      "step": 473780
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 2.066962577485268e-05,
+      "loss": 0.7247,
+      "step": 473790
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 2.0668477844952933e-05,
+      "loss": 0.8003,
+      "step": 473800
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 2.0667329915053187e-05,
+      "loss": 0.7787,
+      "step": 473810
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 2.0666181985153436e-05,
+      "loss": 0.8715,
+      "step": 473820
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 2.0665034055253693e-05,
+      "loss": 0.873,
+      "step": 473830
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 2.0663886125353943e-05,
+      "loss": 0.733,
+      "step": 473840
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 2.0662738195454193e-05,
+      "loss": 0.7041,
+      "step": 473850
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 2.066159026555445e-05,
+      "loss": 0.7425,
+      "step": 473860
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 2.06604423356547e-05,
+      "loss": 0.8386,
+      "step": 473870
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 2.0659294405754952e-05,
+      "loss": 0.7412,
+      "step": 473880
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 2.0658146475855206e-05,
+      "loss": 0.7754,
+      "step": 473890
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 2.065699854595546e-05,
+      "loss": 0.7925,
+      "step": 473900
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 2.065585061605571e-05,
+      "loss": 0.7429,
+      "step": 473910
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 2.0654702686155965e-05,
+      "loss": 0.7584,
+      "step": 473920
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 2.0653554756256215e-05,
+      "loss": 0.8205,
+      "step": 473930
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 2.065240682635647e-05,
+      "loss": 0.8005,
+      "step": 473940
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 2.0651258896456722e-05,
+      "loss": 0.7998,
+      "step": 473950
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 2.0650110966556975e-05,
+      "loss": 0.7738,
+      "step": 473960
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 2.0648963036657225e-05,
+      "loss": 0.7983,
+      "step": 473970
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 2.064781510675748e-05,
+      "loss": 0.7827,
+      "step": 473980
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 2.064666717685773e-05,
+      "loss": 0.7733,
+      "step": 473990
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 2.064551924695798e-05,
+      "loss": 0.8181,
+      "step": 474000
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 2.0644371317058238e-05,
+      "loss": 0.8204,
+      "step": 474010
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 2.0643223387158488e-05,
+      "loss": 0.7392,
+      "step": 474020
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 2.064207545725874e-05,
+      "loss": 0.7719,
+      "step": 474030
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 2.0640927527358994e-05,
+      "loss": 0.8098,
+      "step": 474040
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 2.0639779597459247e-05,
+      "loss": 0.776,
+      "step": 474050
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 2.0638631667559497e-05,
+      "loss": 0.7472,
+      "step": 474060
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 2.0637483737659754e-05,
+      "loss": 0.8023,
+      "step": 474070
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 2.0636335807760004e-05,
+      "loss": 0.7984,
+      "step": 474080
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 2.0635187877860257e-05,
+      "loss": 0.7427,
+      "step": 474090
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 2.063403994796051e-05,
+      "loss": 0.6969,
+      "step": 474100
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 2.0632892018060764e-05,
+      "loss": 0.7785,
+      "step": 474110
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 2.0631744088161013e-05,
+      "loss": 0.8229,
+      "step": 474120
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 2.063059615826127e-05,
+      "loss": 0.7898,
+      "step": 474130
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 2.062944822836152e-05,
+      "loss": 0.737,
+      "step": 474140
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 2.062830029846177e-05,
+      "loss": 0.7763,
+      "step": 474150
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 2.0627152368562026e-05,
+      "loss": 0.8154,
+      "step": 474160
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 2.0626004438662276e-05,
+      "loss": 0.7801,
+      "step": 474170
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 2.062485650876253e-05,
+      "loss": 0.757,
+      "step": 474180
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 2.0623708578862783e-05,
+      "loss": 0.7473,
+      "step": 474190
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 2.0622560648963036e-05,
+      "loss": 0.8144,
+      "step": 474200
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 2.0621412719063286e-05,
+      "loss": 0.8349,
+      "step": 474210
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 2.0620264789163542e-05,
+      "loss": 0.8078,
+      "step": 474220
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 2.0619116859263792e-05,
+      "loss": 0.7777,
+      "step": 474230
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 2.0617968929364045e-05,
+      "loss": 0.8309,
+      "step": 474240
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 2.06168209994643e-05,
+      "loss": 0.8249,
+      "step": 474250
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 2.0615673069564552e-05,
+      "loss": 0.8048,
+      "step": 474260
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 2.0614525139664802e-05,
+      "loss": 0.8153,
+      "step": 474270
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 2.061337720976506e-05,
+      "loss": 0.8126,
+      "step": 474280
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 2.061222927986531e-05,
+      "loss": 0.7678,
+      "step": 474290
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 2.0611081349965558e-05,
+      "loss": 0.8143,
+      "step": 474300
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 2.0609933420065815e-05,
+      "loss": 0.8428,
+      "step": 474310
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 2.0608785490166065e-05,
+      "loss": 0.7705,
+      "step": 474320
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 2.0607637560266318e-05,
+      "loss": 0.7868,
+      "step": 474330
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 2.060648963036657e-05,
+      "loss": 0.8523,
+      "step": 474340
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 2.0605341700466824e-05,
+      "loss": 0.7814,
+      "step": 474350
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 2.0604193770567074e-05,
+      "loss": 0.8142,
+      "step": 474360
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 2.060304584066733e-05,
+      "loss": 0.7324,
+      "step": 474370
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 2.060189791076758e-05,
+      "loss": 0.731,
+      "step": 474380
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 2.0600749980867834e-05,
+      "loss": 0.8347,
+      "step": 474390
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 2.0599602050968084e-05,
+      "loss": 0.7933,
+      "step": 474400
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 2.059845412106834e-05,
+      "loss": 0.7632,
+      "step": 474410
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 2.059730619116859e-05,
+      "loss": 0.7475,
+      "step": 474420
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 2.059615826126884e-05,
+      "loss": 0.8229,
+      "step": 474430
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 2.0595010331369097e-05,
+      "loss": 0.7629,
+      "step": 474440
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 2.0593862401469347e-05,
+      "loss": 0.6842,
+      "step": 474450
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 2.05927144715696e-05,
+      "loss": 0.7127,
+      "step": 474460
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 2.0591566541669853e-05,
+      "loss": 0.7651,
+      "step": 474470
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 2.0590418611770106e-05,
+      "loss": 0.7456,
+      "step": 474480
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 2.0589270681870356e-05,
+      "loss": 0.7996,
+      "step": 474490
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 2.0588122751970613e-05,
+      "loss": 0.7855,
+      "step": 474500
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 2.0586974822070863e-05,
+      "loss": 0.7059,
+      "step": 474510
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 2.0585826892171116e-05,
+      "loss": 0.7612,
+      "step": 474520
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 2.058467896227137e-05,
+      "loss": 0.8817,
+      "step": 474530
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 2.0583531032371622e-05,
+      "loss": 0.7921,
+      "step": 474540
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 2.0582383102471872e-05,
+      "loss": 0.8298,
+      "step": 474550
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 2.058123517257213e-05,
+      "loss": 0.8576,
+      "step": 474560
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 2.058008724267238e-05,
+      "loss": 0.7554,
+      "step": 474570
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 2.057893931277263e-05,
+      "loss": 0.8124,
+      "step": 474580
+    },
+    {
+      "epoch": 21.77,
+      "learning_rate": 2.0577791382872885e-05,
+      "loss": 0.8331,
+      "step": 474590
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 2.0576643452973135e-05,
+      "loss": 0.7785,
+      "step": 474600
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 2.057549552307339e-05,
+      "loss": 0.8162,
+      "step": 474610
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 2.057434759317364e-05,
+      "loss": 0.8217,
+      "step": 474620
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 2.0573199663273895e-05,
+      "loss": 0.7755,
+      "step": 474630
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 2.0572051733374145e-05,
+      "loss": 0.7481,
+      "step": 474640
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 2.05709038034744e-05,
+      "loss": 0.7779,
+      "step": 474650
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 2.056975587357465e-05,
+      "loss": 0.7992,
+      "step": 474660
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 2.0568607943674904e-05,
+      "loss": 0.8486,
+      "step": 474670
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 2.0567460013775158e-05,
+      "loss": 0.7785,
+      "step": 474680
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 2.056631208387541e-05,
+      "loss": 0.7813,
+      "step": 474690
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 2.056516415397566e-05,
+      "loss": 0.8396,
+      "step": 474700
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 2.0564016224075917e-05,
+      "loss": 0.7775,
+      "step": 474710
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 2.0562868294176167e-05,
+      "loss": 0.7927,
+      "step": 474720
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 2.0561720364276417e-05,
+      "loss": 0.8085,
+      "step": 474730
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 2.0560572434376674e-05,
+      "loss": 0.8298,
+      "step": 474740
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 2.0559424504476924e-05,
+      "loss": 0.8276,
+      "step": 474750
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 2.0558276574577177e-05,
+      "loss": 0.8143,
+      "step": 474760
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 2.055712864467743e-05,
+      "loss": 0.7366,
+      "step": 474770
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 2.0555980714777683e-05,
+      "loss": 0.762,
+      "step": 474780
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 2.0554832784877933e-05,
+      "loss": 0.8076,
+      "step": 474790
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 2.055368485497819e-05,
+      "loss": 0.7718,
+      "step": 474800
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 2.055253692507844e-05,
+      "loss": 0.8301,
+      "step": 474810
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 2.0551388995178693e-05,
+      "loss": 0.8484,
+      "step": 474820
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 2.0550241065278946e-05,
+      "loss": 0.7697,
+      "step": 474830
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 2.05490931353792e-05,
+      "loss": 0.824,
+      "step": 474840
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 2.054794520547945e-05,
+      "loss": 0.8097,
+      "step": 474850
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 2.0546797275579706e-05,
+      "loss": 0.7439,
+      "step": 474860
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 2.0545649345679956e-05,
+      "loss": 0.7447,
+      "step": 474870
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 2.0544501415780206e-05,
+      "loss": 0.8383,
+      "step": 474880
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 2.0543353485880462e-05,
+      "loss": 0.815,
+      "step": 474890
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 2.0542205555980712e-05,
+      "loss": 0.706,
+      "step": 474900
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 2.0541057626080965e-05,
+      "loss": 0.8088,
+      "step": 474910
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 2.053990969618122e-05,
+      "loss": 0.7967,
+      "step": 474920
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 2.0538761766281472e-05,
+      "loss": 0.8764,
+      "step": 474930
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 2.053761383638172e-05,
+      "loss": 0.782,
+      "step": 474940
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 2.0536465906481978e-05,
+      "loss": 0.7489,
+      "step": 474950
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 2.0535317976582228e-05,
+      "loss": 0.8327,
+      "step": 474960
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 2.053417004668248e-05,
+      "loss": 0.8505,
+      "step": 474970
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 2.0533022116782735e-05,
+      "loss": 0.8393,
+      "step": 474980
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 2.0531874186882988e-05,
+      "loss": 0.8002,
+      "step": 474990
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 2.0530726256983238e-05,
+      "loss": 0.8209,
+      "step": 475000
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 2.0529578327083494e-05,
+      "loss": 0.8514,
+      "step": 475010
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 2.0528430397183744e-05,
+      "loss": 0.88,
+      "step": 475020
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 2.0527282467283994e-05,
+      "loss": 0.7551,
+      "step": 475030
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 2.052613453738425e-05,
+      "loss": 0.8293,
+      "step": 475040
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 2.05249866074845e-05,
+      "loss": 0.8006,
+      "step": 475050
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 2.0523838677584754e-05,
+      "loss": 0.7048,
+      "step": 475060
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 2.0522690747685007e-05,
+      "loss": 0.7751,
+      "step": 475070
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 2.052154281778526e-05,
+      "loss": 0.7514,
+      "step": 475080
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 2.052039488788551e-05,
+      "loss": 0.7616,
+      "step": 475090
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 2.0519246957985763e-05,
+      "loss": 0.7828,
+      "step": 475100
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 2.0518099028086017e-05,
+      "loss": 0.8371,
+      "step": 475110
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 2.051695109818627e-05,
+      "loss": 0.6952,
+      "step": 475120
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 2.051580316828652e-05,
+      "loss": 0.7652,
+      "step": 475130
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 2.0514655238386776e-05,
+      "loss": 0.8855,
+      "step": 475140
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 2.0513507308487026e-05,
+      "loss": 0.7769,
+      "step": 475150
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 2.0512359378587276e-05,
+      "loss": 0.8235,
+      "step": 475160
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 2.0511211448687533e-05,
+      "loss": 0.7627,
+      "step": 475170
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 2.0510063518787783e-05,
+      "loss": 0.7831,
+      "step": 475180
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 2.0508915588888036e-05,
+      "loss": 0.8225,
+      "step": 475190
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 2.050776765898829e-05,
+      "loss": 0.8009,
+      "step": 475200
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 2.0506619729088542e-05,
+      "loss": 0.7387,
+      "step": 475210
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 2.0505471799188792e-05,
+      "loss": 0.7667,
+      "step": 475220
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 2.050432386928905e-05,
+      "loss": 0.7981,
+      "step": 475230
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 2.05031759393893e-05,
+      "loss": 0.74,
+      "step": 475240
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 2.0502028009489552e-05,
+      "loss": 0.8777,
+      "step": 475250
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 2.0500880079589805e-05,
+      "loss": 0.7661,
+      "step": 475260
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 2.049973214969006e-05,
+      "loss": 0.8019,
+      "step": 475270
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 2.0498584219790308e-05,
+      "loss": 0.7706,
+      "step": 475280
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 2.0497436289890565e-05,
+      "loss": 0.8272,
+      "step": 475290
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 2.0496288359990815e-05,
+      "loss": 0.8396,
+      "step": 475300
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 2.0495140430091065e-05,
+      "loss": 0.7948,
+      "step": 475310
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 2.049399250019132e-05,
+      "loss": 0.814,
+      "step": 475320
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 2.049284457029157e-05,
+      "loss": 0.8324,
+      "step": 475330
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 2.0491696640391824e-05,
+      "loss": 0.7244,
+      "step": 475340
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 2.0490548710492077e-05,
+      "loss": 0.7179,
+      "step": 475350
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 2.048940078059233e-05,
+      "loss": 0.798,
+      "step": 475360
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 2.048825285069258e-05,
+      "loss": 0.8317,
+      "step": 475370
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 2.0487104920792837e-05,
+      "loss": 0.869,
+      "step": 475380
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 2.0485956990893087e-05,
+      "loss": 0.8237,
+      "step": 475390
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 2.048480906099334e-05,
+      "loss": 0.7754,
+      "step": 475400
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 2.0483661131093594e-05,
+      "loss": 0.8107,
+      "step": 475410
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 2.0482513201193847e-05,
+      "loss": 0.8102,
+      "step": 475420
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 2.0481365271294097e-05,
+      "loss": 0.8278,
+      "step": 475430
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 2.0480217341394353e-05,
+      "loss": 0.8097,
+      "step": 475440
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 2.0479069411494603e-05,
+      "loss": 0.7189,
+      "step": 475450
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 2.0477921481594853e-05,
+      "loss": 0.7549,
+      "step": 475460
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 2.047677355169511e-05,
+      "loss": 0.8352,
+      "step": 475470
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 2.047562562179536e-05,
+      "loss": 0.6926,
+      "step": 475480
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 2.0474477691895613e-05,
+      "loss": 0.9005,
+      "step": 475490
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 2.0473329761995866e-05,
+      "loss": 0.747,
+      "step": 475500
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 2.047218183209612e-05,
+      "loss": 0.7715,
+      "step": 475510
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 2.047103390219637e-05,
+      "loss": 0.7728,
+      "step": 475520
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 2.0469885972296626e-05,
+      "loss": 0.823,
+      "step": 475530
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 2.0468738042396876e-05,
+      "loss": 0.6912,
+      "step": 475540
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 2.046759011249713e-05,
+      "loss": 0.7932,
+      "step": 475550
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 2.0466442182597382e-05,
+      "loss": 0.8574,
+      "step": 475560
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 2.0465294252697635e-05,
+      "loss": 0.8438,
+      "step": 475570
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 2.0464146322797885e-05,
+      "loss": 0.7593,
+      "step": 475580
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 2.0462998392898142e-05,
+      "loss": 0.8011,
+      "step": 475590
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 2.046185046299839e-05,
+      "loss": 0.7933,
+      "step": 475600
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 2.046070253309864e-05,
+      "loss": 0.841,
+      "step": 475610
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 2.0459554603198898e-05,
+      "loss": 0.7362,
+      "step": 475620
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 2.0458406673299148e-05,
+      "loss": 0.7171,
+      "step": 475630
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 2.04572587433994e-05,
+      "loss": 0.829,
+      "step": 475640
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 2.0456110813499654e-05,
+      "loss": 0.7984,
+      "step": 475650
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 2.0454962883599908e-05,
+      "loss": 0.8364,
+      "step": 475660
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 2.0453814953700158e-05,
+      "loss": 0.7469,
+      "step": 475670
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 2.0452667023800414e-05,
+      "loss": 0.7822,
+      "step": 475680
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 2.0451519093900664e-05,
+      "loss": 0.7953,
+      "step": 475690
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 2.0450371164000917e-05,
+      "loss": 0.767,
+      "step": 475700
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 2.044922323410117e-05,
+      "loss": 0.7824,
+      "step": 475710
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 2.0448075304201424e-05,
+      "loss": 0.837,
+      "step": 475720
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 2.0446927374301674e-05,
+      "loss": 0.8319,
+      "step": 475730
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 2.044577944440193e-05,
+      "loss": 0.8219,
+      "step": 475740
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 2.044463151450218e-05,
+      "loss": 0.8154,
+      "step": 475750
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 2.044348358460243e-05,
+      "loss": 0.8044,
+      "step": 475760
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 2.0442335654702683e-05,
+      "loss": 0.7345,
+      "step": 475770
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 2.0441187724802936e-05,
+      "loss": 0.8083,
+      "step": 475780
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 2.044003979490319e-05,
+      "loss": 0.7744,
+      "step": 475790
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 2.043889186500344e-05,
+      "loss": 0.645,
+      "step": 475800
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 2.0437743935103696e-05,
+      "loss": 0.8686,
+      "step": 475810
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 2.0436596005203946e-05,
+      "loss": 0.7957,
+      "step": 475820
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 2.04354480753042e-05,
+      "loss": 0.833,
+      "step": 475830
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 2.0434300145404452e-05,
+      "loss": 0.7951,
+      "step": 475840
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 2.0433152215504706e-05,
+      "loss": 0.7565,
+      "step": 475850
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 2.0432004285604956e-05,
+      "loss": 0.8821,
+      "step": 475860
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 2.0430856355705212e-05,
+      "loss": 0.7746,
+      "step": 475870
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 2.0429708425805462e-05,
+      "loss": 0.8284,
+      "step": 475880
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 2.0428560495905712e-05,
+      "loss": 0.7572,
+      "step": 475890
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 2.042741256600597e-05,
+      "loss": 0.8337,
+      "step": 475900
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 2.042626463610622e-05,
+      "loss": 0.8347,
+      "step": 475910
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 2.042511670620647e-05,
+      "loss": 0.8563,
+      "step": 475920
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 2.0423968776306725e-05,
+      "loss": 0.8609,
+      "step": 475930
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 2.0422820846406978e-05,
+      "loss": 0.7978,
+      "step": 475940
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 2.0421672916507228e-05,
+      "loss": 0.8096,
+      "step": 475950
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 2.0420524986607485e-05,
+      "loss": 0.7845,
+      "step": 475960
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 2.0419377056707734e-05,
+      "loss": 0.7911,
+      "step": 475970
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 2.0418229126807988e-05,
+      "loss": 0.7651,
+      "step": 475980
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 2.041708119690824e-05,
+      "loss": 0.7761,
+      "step": 475990
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 2.0415933267008494e-05,
+      "loss": 0.7903,
+      "step": 476000
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 2.0414785337108744e-05,
+      "loss": 0.7401,
+      "step": 476010
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 2.0413637407209e-05,
+      "loss": 0.7721,
+      "step": 476020
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 2.041248947730925e-05,
+      "loss": 0.8035,
+      "step": 476030
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 2.04113415474095e-05,
+      "loss": 0.7656,
+      "step": 476040
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 2.0410193617509757e-05,
+      "loss": 0.742,
+      "step": 476050
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 2.0409045687610007e-05,
+      "loss": 0.7519,
+      "step": 476060
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 2.040789775771026e-05,
+      "loss": 0.8592,
+      "step": 476070
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 2.0406749827810513e-05,
+      "loss": 0.884,
+      "step": 476080
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 2.0405601897910767e-05,
+      "loss": 0.8046,
+      "step": 476090
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 2.0404453968011016e-05,
+      "loss": 0.8094,
+      "step": 476100
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 2.0403306038111273e-05,
+      "loss": 0.7462,
+      "step": 476110
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 2.0402158108211523e-05,
+      "loss": 0.7756,
+      "step": 476120
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 2.0401010178311776e-05,
+      "loss": 0.7417,
+      "step": 476130
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 2.039986224841203e-05,
+      "loss": 0.776,
+      "step": 476140
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 2.0398714318512283e-05,
+      "loss": 0.837,
+      "step": 476150
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 2.0397566388612533e-05,
+      "loss": 0.6678,
+      "step": 476160
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 2.039641845871279e-05,
+      "loss": 0.8832,
+      "step": 476170
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 2.039527052881304e-05,
+      "loss": 0.7843,
+      "step": 476180
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 2.039412259891329e-05,
+      "loss": 0.7572,
+      "step": 476190
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 2.0392974669013545e-05,
+      "loss": 0.7561,
+      "step": 476200
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 2.0391826739113795e-05,
+      "loss": 0.8077,
+      "step": 476210
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 2.039067880921405e-05,
+      "loss": 0.7616,
+      "step": 476220
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 2.0389530879314302e-05,
+      "loss": 0.7741,
+      "step": 476230
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 2.0388382949414555e-05,
+      "loss": 0.7882,
+      "step": 476240
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 2.0387235019514805e-05,
+      "loss": 0.7629,
+      "step": 476250
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 2.038608708961506e-05,
+      "loss": 0.8891,
+      "step": 476260
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 2.038493915971531e-05,
+      "loss": 0.7365,
+      "step": 476270
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 2.0383791229815565e-05,
+      "loss": 0.7585,
+      "step": 476280
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 2.0382643299915818e-05,
+      "loss": 0.9172,
+      "step": 476290
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 2.038149537001607e-05,
+      "loss": 0.8231,
+      "step": 476300
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 2.038034744011632e-05,
+      "loss": 0.7615,
+      "step": 476310
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 2.0379199510216578e-05,
+      "loss": 0.7455,
+      "step": 476320
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 2.0378051580316827e-05,
+      "loss": 0.7327,
+      "step": 476330
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 2.0376903650417077e-05,
+      "loss": 0.9225,
+      "step": 476340
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 2.0375755720517334e-05,
+      "loss": 0.7802,
+      "step": 476350
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 2.0374607790617584e-05,
+      "loss": 0.7432,
+      "step": 476360
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 2.0373459860717837e-05,
+      "loss": 0.7636,
+      "step": 476370
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 2.037231193081809e-05,
+      "loss": 0.7977,
+      "step": 476380
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 2.0371164000918344e-05,
+      "loss": 0.8015,
+      "step": 476390
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 2.0370016071018593e-05,
+      "loss": 0.7799,
+      "step": 476400
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 2.036886814111885e-05,
+      "loss": 0.8469,
+      "step": 476410
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 2.03677202112191e-05,
+      "loss": 0.7349,
+      "step": 476420
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 2.0366572281319353e-05,
+      "loss": 0.7843,
+      "step": 476430
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 2.0365424351419606e-05,
+      "loss": 0.795,
+      "step": 476440
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 2.036427642151986e-05,
+      "loss": 0.7977,
+      "step": 476450
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 2.036312849162011e-05,
+      "loss": 0.8321,
+      "step": 476460
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 2.036198056172036e-05,
+      "loss": 0.7775,
+      "step": 476470
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 2.0360832631820616e-05,
+      "loss": 0.7318,
+      "step": 476480
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 2.0359684701920866e-05,
+      "loss": 0.7774,
+      "step": 476490
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 2.035853677202112e-05,
+      "loss": 0.8025,
+      "step": 476500
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 2.0357388842121372e-05,
+      "loss": 0.8588,
+      "step": 476510
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 2.0356240912221626e-05,
+      "loss": 0.8272,
+      "step": 476520
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 2.0355092982321875e-05,
+      "loss": 0.765,
+      "step": 476530
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 2.0353945052422132e-05,
+      "loss": 0.7663,
+      "step": 476540
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 2.0352797122522382e-05,
+      "loss": 0.813,
+      "step": 476550
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 2.0351649192622635e-05,
+      "loss": 0.7575,
+      "step": 476560
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 2.035050126272289e-05,
+      "loss": 0.75,
+      "step": 476570
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 2.034935333282314e-05,
+      "loss": 0.7648,
+      "step": 476580
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 2.034820540292339e-05,
+      "loss": 0.78,
+      "step": 476590
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 2.0347057473023648e-05,
+      "loss": 0.8111,
+      "step": 476600
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 2.0345909543123898e-05,
+      "loss": 0.8117,
+      "step": 476610
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 2.0344761613224148e-05,
+      "loss": 0.803,
+      "step": 476620
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 2.0343613683324404e-05,
+      "loss": 0.843,
+      "step": 476630
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 2.0342465753424654e-05,
+      "loss": 0.7844,
+      "step": 476640
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 2.0341317823524908e-05,
+      "loss": 0.7634,
+      "step": 476650
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 2.034016989362516e-05,
+      "loss": 0.7572,
+      "step": 476660
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 2.0339021963725414e-05,
+      "loss": 0.7579,
+      "step": 476670
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 2.0337874033825664e-05,
+      "loss": 0.8128,
+      "step": 476680
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 2.033672610392592e-05,
+      "loss": 0.7409,
+      "step": 476690
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 2.033557817402617e-05,
+      "loss": 0.7836,
+      "step": 476700
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 2.0334430244126424e-05,
+      "loss": 0.7495,
+      "step": 476710
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 2.0333282314226677e-05,
+      "loss": 0.832,
+      "step": 476720
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 2.033213438432693e-05,
+      "loss": 0.7883,
+      "step": 476730
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 2.033098645442718e-05,
+      "loss": 0.7588,
+      "step": 476740
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 2.0329838524527437e-05,
+      "loss": 0.8297,
+      "step": 476750
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 2.0328690594627686e-05,
+      "loss": 0.731,
+      "step": 476760
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 2.0327542664727936e-05,
+      "loss": 0.765,
+      "step": 476770
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 2.0326394734828193e-05,
+      "loss": 0.8562,
+      "step": 476780
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 2.0325246804928443e-05,
+      "loss": 0.7505,
+      "step": 476790
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 2.0324098875028696e-05,
+      "loss": 0.781,
+      "step": 476800
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 2.032295094512895e-05,
+      "loss": 0.8376,
+      "step": 476810
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 2.0321803015229202e-05,
+      "loss": 0.7618,
+      "step": 476820
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 2.0320655085329452e-05,
+      "loss": 0.8284,
+      "step": 476830
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 2.031950715542971e-05,
+      "loss": 0.7688,
+      "step": 476840
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 2.031835922552996e-05,
+      "loss": 0.8079,
+      "step": 476850
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 2.0317211295630212e-05,
+      "loss": 0.8861,
+      "step": 476860
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 2.0316063365730465e-05,
+      "loss": 0.7689,
+      "step": 476870
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 2.031491543583072e-05,
+      "loss": 0.7325,
+      "step": 476880
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 2.031376750593097e-05,
+      "loss": 0.8239,
+      "step": 476890
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 2.0312619576031225e-05,
+      "loss": 0.7724,
+      "step": 476900
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 2.0311471646131475e-05,
+      "loss": 0.7795,
+      "step": 476910
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 2.0310323716231725e-05,
+      "loss": 0.8473,
+      "step": 476920
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 2.030917578633198e-05,
+      "loss": 0.9015,
+      "step": 476930
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 2.030802785643223e-05,
+      "loss": 0.814,
+      "step": 476940
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 2.0306879926532484e-05,
+      "loss": 0.8399,
+      "step": 476950
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 2.0305731996632738e-05,
+      "loss": 0.7841,
+      "step": 476960
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 2.030458406673299e-05,
+      "loss": 0.7677,
+      "step": 476970
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 2.030343613683324e-05,
+      "loss": 0.8073,
+      "step": 476980
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 2.0302288206933497e-05,
+      "loss": 0.8014,
+      "step": 476990
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 2.0301140277033747e-05,
+      "loss": 0.7565,
+      "step": 477000
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 2.0299992347134e-05,
+      "loss": 0.7348,
+      "step": 477010
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 2.0298844417234254e-05,
+      "loss": 0.8592,
+      "step": 477020
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 2.0297696487334507e-05,
+      "loss": 0.7791,
+      "step": 477030
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 2.0296548557434757e-05,
+      "loss": 0.7739,
+      "step": 477040
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 2.0295400627535013e-05,
+      "loss": 0.7702,
+      "step": 477050
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 2.0294252697635263e-05,
+      "loss": 0.7374,
+      "step": 477060
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 2.0293104767735513e-05,
+      "loss": 0.8645,
+      "step": 477070
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 2.029195683783577e-05,
+      "loss": 0.7375,
+      "step": 477080
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 2.029080890793602e-05,
+      "loss": 0.7639,
+      "step": 477090
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 2.0289660978036273e-05,
+      "loss": 0.7725,
+      "step": 477100
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 2.0288513048136526e-05,
+      "loss": 0.7607,
+      "step": 477110
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 2.028736511823678e-05,
+      "loss": 0.7812,
+      "step": 477120
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 2.028621718833703e-05,
+      "loss": 0.7792,
+      "step": 477130
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 2.0285069258437283e-05,
+      "loss": 0.7953,
+      "step": 477140
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 2.0283921328537536e-05,
+      "loss": 0.8072,
+      "step": 477150
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 2.028277339863779e-05,
+      "loss": 0.8555,
+      "step": 477160
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 2.028162546873804e-05,
+      "loss": 0.815,
+      "step": 477170
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 2.0280477538838295e-05,
+      "loss": 0.8502,
+      "step": 477180
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 2.0279329608938545e-05,
+      "loss": 0.8361,
+      "step": 477190
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 2.0278181679038795e-05,
+      "loss": 0.8411,
+      "step": 477200
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 2.0277033749139052e-05,
+      "loss": 0.809,
+      "step": 477210
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 2.02758858192393e-05,
+      "loss": 0.7345,
+      "step": 477220
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 2.0274737889339555e-05,
+      "loss": 0.795,
+      "step": 477230
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 2.0273589959439808e-05,
+      "loss": 0.8374,
+      "step": 477240
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 2.027244202954006e-05,
+      "loss": 0.7157,
+      "step": 477250
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 2.027129409964031e-05,
+      "loss": 0.7636,
+      "step": 477260
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 2.0270146169740568e-05,
+      "loss": 0.8623,
+      "step": 477270
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 2.0268998239840818e-05,
+      "loss": 0.8263,
+      "step": 477280
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 2.026785030994107e-05,
+      "loss": 0.8801,
+      "step": 477290
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 2.0266702380041324e-05,
+      "loss": 0.8124,
+      "step": 477300
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 2.0265554450141577e-05,
+      "loss": 0.7504,
+      "step": 477310
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 2.0264406520241827e-05,
+      "loss": 0.7762,
+      "step": 477320
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 2.0263258590342084e-05,
+      "loss": 0.8124,
+      "step": 477330
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 2.0262110660442334e-05,
+      "loss": 0.7815,
+      "step": 477340
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 2.0260962730542584e-05,
+      "loss": 0.7946,
+      "step": 477350
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 2.025981480064284e-05,
+      "loss": 0.7786,
+      "step": 477360
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 2.025866687074309e-05,
+      "loss": 0.7764,
+      "step": 477370
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 2.0257518940843343e-05,
+      "loss": 0.7622,
+      "step": 477380
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 2.0256371010943597e-05,
+      "loss": 0.713,
+      "step": 477390
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 2.025522308104385e-05,
+      "loss": 0.8642,
+      "step": 477400
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 2.02540751511441e-05,
+      "loss": 0.7844,
+      "step": 477410
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 2.0252927221244356e-05,
+      "loss": 0.756,
+      "step": 477420
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 2.0251779291344606e-05,
+      "loss": 0.7781,
+      "step": 477430
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 2.025063136144486e-05,
+      "loss": 0.798,
+      "step": 477440
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 2.0249483431545113e-05,
+      "loss": 0.788,
+      "step": 477450
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 2.0248335501645366e-05,
+      "loss": 0.7967,
+      "step": 477460
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 2.0247187571745616e-05,
+      "loss": 0.8669,
+      "step": 477470
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 2.0246039641845872e-05,
+      "loss": 0.7305,
+      "step": 477480
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 2.0244891711946122e-05,
+      "loss": 0.789,
+      "step": 477490
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 2.0243743782046372e-05,
+      "loss": 0.7481,
+      "step": 477500
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 2.024259585214663e-05,
+      "loss": 0.7651,
+      "step": 477510
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 2.024144792224688e-05,
+      "loss": 0.806,
+      "step": 477520
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 2.0240299992347132e-05,
+      "loss": 0.7035,
+      "step": 477530
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 2.0239152062447385e-05,
+      "loss": 0.7918,
+      "step": 477540
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 2.023800413254764e-05,
+      "loss": 0.7654,
+      "step": 477550
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 2.0236856202647888e-05,
+      "loss": 0.8172,
+      "step": 477560
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 2.0235708272748145e-05,
+      "loss": 0.7414,
+      "step": 477570
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 2.0234560342848395e-05,
+      "loss": 0.8193,
+      "step": 477580
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 2.0233412412948648e-05,
+      "loss": 0.8132,
+      "step": 477590
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 2.02322644830489e-05,
+      "loss": 0.7456,
+      "step": 477600
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 2.0231116553149154e-05,
+      "loss": 0.8847,
+      "step": 477610
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 2.0229968623249404e-05,
+      "loss": 0.8451,
+      "step": 477620
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 2.022882069334966e-05,
+      "loss": 0.7913,
+      "step": 477630
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 2.022767276344991e-05,
+      "loss": 0.8133,
+      "step": 477640
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 2.022652483355016e-05,
+      "loss": 0.8178,
+      "step": 477650
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 2.0225376903650417e-05,
+      "loss": 0.8292,
+      "step": 477660
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 2.0224228973750667e-05,
+      "loss": 0.8652,
+      "step": 477670
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 2.022308104385092e-05,
+      "loss": 0.8112,
+      "step": 477680
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 2.0221933113951174e-05,
+      "loss": 0.8049,
+      "step": 477690
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 2.0220785184051427e-05,
+      "loss": 0.8405,
+      "step": 477700
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 2.0219637254151677e-05,
+      "loss": 0.7441,
+      "step": 477710
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 2.0218489324251933e-05,
+      "loss": 0.7823,
+      "step": 477720
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 2.0217341394352183e-05,
+      "loss": 0.7181,
+      "step": 477730
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 2.0216193464452436e-05,
+      "loss": 0.7565,
+      "step": 477740
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 2.021504553455269e-05,
+      "loss": 0.6971,
+      "step": 477750
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 2.0213897604652943e-05,
+      "loss": 0.788,
+      "step": 477760
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 2.0212749674753193e-05,
+      "loss": 0.844,
+      "step": 477770
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 2.021160174485345e-05,
+      "loss": 0.814,
+      "step": 477780
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 2.02104538149537e-05,
+      "loss": 0.7888,
+      "step": 477790
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 2.020930588505395e-05,
+      "loss": 0.7567,
+      "step": 477800
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 2.0208157955154206e-05,
+      "loss": 0.7541,
+      "step": 477810
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 2.0207010025254456e-05,
+      "loss": 0.7863,
+      "step": 477820
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 2.020586209535471e-05,
+      "loss": 0.7339,
+      "step": 477830
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 2.020471416545496e-05,
+      "loss": 0.7848,
+      "step": 477840
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 2.0203566235555215e-05,
+      "loss": 0.8236,
+      "step": 477850
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 2.0202418305655465e-05,
+      "loss": 0.8064,
+      "step": 477860
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 2.020127037575572e-05,
+      "loss": 0.7326,
+      "step": 477870
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 2.020012244585597e-05,
+      "loss": 0.7605,
+      "step": 477880
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 2.0198974515956225e-05,
+      "loss": 0.7253,
+      "step": 477890
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 2.0197826586056475e-05,
+      "loss": 0.794,
+      "step": 477900
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 2.019667865615673e-05,
+      "loss": 0.7192,
+      "step": 477910
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 2.019553072625698e-05,
+      "loss": 0.7635,
+      "step": 477920
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 2.019438279635723e-05,
+      "loss": 0.7464,
+      "step": 477930
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 2.0193234866457488e-05,
+      "loss": 0.8402,
+      "step": 477940
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 2.0192086936557738e-05,
+      "loss": 0.7929,
+      "step": 477950
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 2.019093900665799e-05,
+      "loss": 0.7244,
+      "step": 477960
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 2.0189791076758244e-05,
+      "loss": 0.7778,
+      "step": 477970
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 2.0188643146858497e-05,
+      "loss": 0.7819,
+      "step": 477980
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 2.0187495216958747e-05,
+      "loss": 0.8184,
+      "step": 477990
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 2.0186347287059004e-05,
+      "loss": 0.8468,
+      "step": 478000
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 2.0185199357159254e-05,
+      "loss": 0.7558,
+      "step": 478010
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 2.0184051427259507e-05,
+      "loss": 0.7969,
+      "step": 478020
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 2.018290349735976e-05,
+      "loss": 0.8601,
+      "step": 478030
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 2.0181755567460013e-05,
+      "loss": 0.8708,
+      "step": 478040
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 2.0180607637560263e-05,
+      "loss": 0.7685,
+      "step": 478050
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 2.017945970766052e-05,
+      "loss": 0.8094,
+      "step": 478060
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 2.017831177776077e-05,
+      "loss": 0.7884,
+      "step": 478070
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 2.017716384786102e-05,
+      "loss": 0.781,
+      "step": 478080
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 2.0176015917961276e-05,
+      "loss": 0.7148,
+      "step": 478090
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 2.0174867988061526e-05,
+      "loss": 0.7535,
+      "step": 478100
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 2.017372005816178e-05,
+      "loss": 0.7764,
+      "step": 478110
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 2.0172572128262033e-05,
+      "loss": 0.7359,
+      "step": 478120
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 2.0171424198362286e-05,
+      "loss": 0.7965,
+      "step": 478130
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 2.0170276268462536e-05,
+      "loss": 0.8043,
+      "step": 478140
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 2.0169128338562792e-05,
+      "loss": 0.7669,
+      "step": 478150
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 2.0167980408663042e-05,
+      "loss": 0.8583,
+      "step": 478160
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 2.0166832478763295e-05,
+      "loss": 0.7674,
+      "step": 478170
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 2.016568454886355e-05,
+      "loss": 0.8258,
+      "step": 478180
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 2.0164536618963802e-05,
+      "loss": 0.8632,
+      "step": 478190
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 2.016338868906405e-05,
+      "loss": 0.7269,
+      "step": 478200
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 2.0162240759164308e-05,
+      "loss": 0.7984,
+      "step": 478210
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 2.0161092829264558e-05,
+      "loss": 0.8158,
+      "step": 478220
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 2.0159944899364808e-05,
+      "loss": 0.775,
+      "step": 478230
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 2.0158796969465065e-05,
+      "loss": 0.8016,
+      "step": 478240
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 2.0157649039565314e-05,
+      "loss": 0.7545,
+      "step": 478250
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 2.0156501109665568e-05,
+      "loss": 0.7909,
+      "step": 478260
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 2.015535317976582e-05,
+      "loss": 0.7556,
+      "step": 478270
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 2.0154205249866074e-05,
+      "loss": 0.8226,
+      "step": 478280
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 2.0153057319966324e-05,
+      "loss": 0.7789,
+      "step": 478290
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 2.015190939006658e-05,
+      "loss": 0.806,
+      "step": 478300
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 2.015076146016683e-05,
+      "loss": 0.8206,
+      "step": 478310
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 2.0149613530267084e-05,
+      "loss": 0.8054,
+      "step": 478320
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 2.0148465600367337e-05,
+      "loss": 0.8105,
+      "step": 478330
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 2.014731767046759e-05,
+      "loss": 0.842,
+      "step": 478340
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 2.014616974056784e-05,
+      "loss": 0.8285,
+      "step": 478350
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 2.0145021810668097e-05,
+      "loss": 0.7677,
+      "step": 478360
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 2.0143873880768347e-05,
+      "loss": 0.8036,
+      "step": 478370
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 2.0142725950868596e-05,
+      "loss": 0.7723,
+      "step": 478380
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 2.0141578020968853e-05,
+      "loss": 0.8019,
+      "step": 478390
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 2.0140430091069103e-05,
+      "loss": 0.7995,
+      "step": 478400
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 2.0139282161169356e-05,
+      "loss": 0.7414,
+      "step": 478410
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 2.013813423126961e-05,
+      "loss": 0.8108,
+      "step": 478420
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 2.0136986301369863e-05,
+      "loss": 0.7458,
+      "step": 478430
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 2.0135838371470113e-05,
+      "loss": 0.7702,
+      "step": 478440
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 2.013469044157037e-05,
+      "loss": 0.8065,
+      "step": 478450
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 2.013354251167062e-05,
+      "loss": 0.7618,
+      "step": 478460
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 2.0132394581770872e-05,
+      "loss": 0.7806,
+      "step": 478470
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 2.0131246651871126e-05,
+      "loss": 0.7186,
+      "step": 478480
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 2.013009872197138e-05,
+      "loss": 0.8116,
+      "step": 478490
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 2.012895079207163e-05,
+      "loss": 0.7268,
+      "step": 478500
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 2.012780286217188e-05,
+      "loss": 0.7792,
+      "step": 478510
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 2.0126654932272135e-05,
+      "loss": 0.7674,
+      "step": 478520
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 2.0125507002372385e-05,
+      "loss": 0.837,
+      "step": 478530
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 2.0124359072472638e-05,
+      "loss": 0.7246,
+      "step": 478540
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 2.012321114257289e-05,
+      "loss": 0.7857,
+      "step": 478550
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 2.0122063212673145e-05,
+      "loss": 0.7903,
+      "step": 478560
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 2.0120915282773395e-05,
+      "loss": 0.8133,
+      "step": 478570
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 2.011976735287365e-05,
+      "loss": 0.7667,
+      "step": 478580
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 2.01186194229739e-05,
+      "loss": 0.7391,
+      "step": 478590
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 2.0117471493074154e-05,
+      "loss": 0.8152,
+      "step": 478600
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 2.0116323563174407e-05,
+      "loss": 0.8003,
+      "step": 478610
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 2.011517563327466e-05,
+      "loss": 0.8596,
+      "step": 478620
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 2.011402770337491e-05,
+      "loss": 0.7349,
+      "step": 478630
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 2.0112879773475167e-05,
+      "loss": 0.7831,
+      "step": 478640
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 2.0111731843575417e-05,
+      "loss": 0.7738,
+      "step": 478650
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 2.0110583913675667e-05,
+      "loss": 0.8376,
+      "step": 478660
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 2.0109435983775924e-05,
+      "loss": 0.9022,
+      "step": 478670
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 2.0108288053876173e-05,
+      "loss": 0.8469,
+      "step": 478680
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 2.0107140123976427e-05,
+      "loss": 0.7469,
+      "step": 478690
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 2.010599219407668e-05,
+      "loss": 0.7428,
+      "step": 478700
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 2.0104844264176933e-05,
+      "loss": 0.7481,
+      "step": 478710
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 2.0103696334277183e-05,
+      "loss": 0.8124,
+      "step": 478720
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 2.010254840437744e-05,
+      "loss": 0.8725,
+      "step": 478730
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 2.010140047447769e-05,
+      "loss": 0.7956,
+      "step": 478740
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 2.0100252544577943e-05,
+      "loss": 0.886,
+      "step": 478750
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 2.0099104614678196e-05,
+      "loss": 0.8086,
+      "step": 478760
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 2.009795668477845e-05,
+      "loss": 0.7938,
+      "step": 478770
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 2.00968087548787e-05,
+      "loss": 0.8258,
+      "step": 478780
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 2.0095660824978956e-05,
+      "loss": 0.7975,
+      "step": 478790
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 2.0094512895079206e-05,
+      "loss": 0.7405,
+      "step": 478800
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 2.0093364965179455e-05,
+      "loss": 0.7897,
+      "step": 478810
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 2.0092217035279712e-05,
+      "loss": 0.8655,
+      "step": 478820
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 2.0091069105379962e-05,
+      "loss": 0.8762,
+      "step": 478830
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 2.0089921175480215e-05,
+      "loss": 0.8086,
+      "step": 478840
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 2.008877324558047e-05,
+      "loss": 0.8044,
+      "step": 478850
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 2.008762531568072e-05,
+      "loss": 0.8186,
+      "step": 478860
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 2.008647738578097e-05,
+      "loss": 0.8043,
+      "step": 478870
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 2.0085329455881228e-05,
+      "loss": 0.7815,
+      "step": 478880
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 2.0084181525981478e-05,
+      "loss": 0.7344,
+      "step": 478890
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 2.008303359608173e-05,
+      "loss": 0.8187,
+      "step": 478900
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 2.0081885666181984e-05,
+      "loss": 0.7582,
+      "step": 478910
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 2.0080737736282238e-05,
+      "loss": 0.8115,
+      "step": 478920
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 2.0079589806382488e-05,
+      "loss": 0.7198,
+      "step": 478930
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 2.0078441876482744e-05,
+      "loss": 0.8205,
+      "step": 478940
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 2.0077293946582994e-05,
+      "loss": 0.8307,
+      "step": 478950
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 2.0076146016683244e-05,
+      "loss": 0.8525,
+      "step": 478960
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 2.00749980867835e-05,
+      "loss": 0.7819,
+      "step": 478970
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 2.007385015688375e-05,
+      "loss": 0.8206,
+      "step": 478980
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 2.0072702226984004e-05,
+      "loss": 0.8219,
+      "step": 478990
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 2.0071554297084257e-05,
+      "loss": 0.7841,
+      "step": 479000
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 2.007040636718451e-05,
+      "loss": 0.825,
+      "step": 479010
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 2.006925843728476e-05,
+      "loss": 0.7649,
+      "step": 479020
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 2.0068110507385017e-05,
+      "loss": 0.8123,
+      "step": 479030
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 2.0066962577485266e-05,
+      "loss": 0.8518,
+      "step": 479040
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 2.006581464758552e-05,
+      "loss": 0.7637,
+      "step": 479050
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 2.0064666717685773e-05,
+      "loss": 0.8296,
+      "step": 479060
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 2.0063518787786026e-05,
+      "loss": 0.7795,
+      "step": 479070
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 2.0062370857886276e-05,
+      "loss": 0.8496,
+      "step": 479080
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 2.0061222927986533e-05,
+      "loss": 0.8729,
+      "step": 479090
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 2.0060074998086782e-05,
+      "loss": 0.7585,
+      "step": 479100
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 2.0058927068187032e-05,
+      "loss": 0.8907,
+      "step": 479110
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 2.005777913828729e-05,
+      "loss": 0.7526,
+      "step": 479120
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 2.005663120838754e-05,
+      "loss": 0.7963,
+      "step": 479130
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 2.0055483278487792e-05,
+      "loss": 0.843,
+      "step": 479140
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 2.0054335348588045e-05,
+      "loss": 0.88,
+      "step": 479150
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 2.00531874186883e-05,
+      "loss": 0.7221,
+      "step": 479160
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 2.005203948878855e-05,
+      "loss": 0.722,
+      "step": 479170
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 2.0050891558888805e-05,
+      "loss": 0.7668,
+      "step": 479180
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 2.0049743628989055e-05,
+      "loss": 0.777,
+      "step": 479190
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 2.0048595699089308e-05,
+      "loss": 0.7996,
+      "step": 479200
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 2.0047447769189558e-05,
+      "loss": 0.7417,
+      "step": 479210
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 2.0046299839289815e-05,
+      "loss": 0.7704,
+      "step": 479220
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 2.0045151909390064e-05,
+      "loss": 0.8018,
+      "step": 479230
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 2.0044003979490314e-05,
+      "loss": 0.8375,
+      "step": 479240
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 2.004285604959057e-05,
+      "loss": 0.7842,
+      "step": 479250
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 2.004170811969082e-05,
+      "loss": 0.7424,
+      "step": 479260
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 2.0040560189791074e-05,
+      "loss": 0.7747,
+      "step": 479270
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 2.0039412259891327e-05,
+      "loss": 0.8034,
+      "step": 479280
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 2.003826432999158e-05,
+      "loss": 0.7455,
+      "step": 479290
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 2.003711640009183e-05,
+      "loss": 0.8416,
+      "step": 479300
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 2.0035968470192087e-05,
+      "loss": 0.7119,
+      "step": 479310
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 2.0034820540292337e-05,
+      "loss": 0.8634,
+      "step": 479320
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 2.003367261039259e-05,
+      "loss": 0.8256,
+      "step": 479330
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 2.0032524680492843e-05,
+      "loss": 0.7954,
+      "step": 479340
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 2.0031376750593097e-05,
+      "loss": 0.7592,
+      "step": 479350
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 2.0030228820693346e-05,
+      "loss": 0.8589,
+      "step": 479360
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 2.0029080890793603e-05,
+      "loss": 0.7763,
+      "step": 479370
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 2.0027932960893853e-05,
+      "loss": 0.7441,
+      "step": 479380
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 2.0026785030994103e-05,
+      "loss": 0.7766,
+      "step": 479390
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 2.002563710109436e-05,
+      "loss": 0.8068,
+      "step": 479400
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 2.002448917119461e-05,
+      "loss": 0.7313,
+      "step": 479410
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 2.0023341241294863e-05,
+      "loss": 0.7961,
+      "step": 479420
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 2.0022193311395116e-05,
+      "loss": 0.7373,
+      "step": 479430
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 2.002104538149537e-05,
+      "loss": 0.7737,
+      "step": 479440
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 2.001989745159562e-05,
+      "loss": 0.8252,
+      "step": 479450
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 2.0018749521695876e-05,
+      "loss": 0.9059,
+      "step": 479460
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 2.0017601591796125e-05,
+      "loss": 0.7502,
+      "step": 479470
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 2.001645366189638e-05,
+      "loss": 0.6759,
+      "step": 479480
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 2.0015305731996632e-05,
+      "loss": 0.7893,
+      "step": 479490
+    },
+    {
+      "epoch": 22.0,
+      "eval_cer": 0.6685880348215238,
+      "eval_em": 0.007357212003872217,
+      "eval_f1": 0.007357212003872217,
+      "eval_loss": 1.3177778720855713,
+      "eval_runtime": 1518.618,
+      "eval_samples_per_second": 6.802,
+      "eval_steps_per_second": 0.851,
+      "eval_wer": 0.9926427879961278,
+      "step": 479495
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 2.0014157802096885e-05,
+      "loss": 0.7196,
+      "step": 479500
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 2.0013009872197135e-05,
+      "loss": 0.7634,
+      "step": 479510
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 2.001186194229739e-05,
+      "loss": 0.7533,
+      "step": 479520
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 2.001071401239764e-05,
+      "loss": 0.768,
+      "step": 479530
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 2.000956608249789e-05,
+      "loss": 0.8082,
+      "step": 479540
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 2.0008418152598148e-05,
+      "loss": 0.7461,
+      "step": 479550
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 2.0007270222698398e-05,
+      "loss": 0.7096,
+      "step": 479560
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 2.000612229279865e-05,
+      "loss": 0.7405,
+      "step": 479570
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 2.0004974362898904e-05,
+      "loss": 0.797,
+      "step": 479580
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 2.0003826432999157e-05,
+      "loss": 0.7919,
+      "step": 479590
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 2.0002678503099407e-05,
+      "loss": 0.6543,
+      "step": 479600
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 2.0001530573199664e-05,
+      "loss": 0.78,
+      "step": 479610
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 2.0000382643299914e-05,
+      "loss": 0.8099,
+      "step": 479620
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 1.9999234713400167e-05,
+      "loss": 0.7576,
+      "step": 479630
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 1.999808678350042e-05,
+      "loss": 0.7518,
+      "step": 479640
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 1.9996938853600674e-05,
+      "loss": 0.7935,
+      "step": 479650
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 1.9995790923700923e-05,
+      "loss": 0.7286,
+      "step": 479660
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 1.999464299380118e-05,
+      "loss": 0.6921,
+      "step": 479670
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 1.999349506390143e-05,
+      "loss": 0.8096,
+      "step": 479680
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 1.999234713400168e-05,
+      "loss": 0.7245,
+      "step": 479690
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 1.9991199204101936e-05,
+      "loss": 0.7678,
+      "step": 479700
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 1.9990051274202186e-05,
+      "loss": 0.7085,
+      "step": 479710
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 1.998890334430244e-05,
+      "loss": 0.7685,
+      "step": 479720
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 1.9987755414402693e-05,
+      "loss": 0.7231,
+      "step": 479730
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 1.9986607484502946e-05,
+      "loss": 0.7367,
+      "step": 479740
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 1.9985459554603196e-05,
+      "loss": 0.7919,
+      "step": 479750
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 1.9984311624703452e-05,
+      "loss": 0.6935,
+      "step": 479760
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 1.9983163694803702e-05,
+      "loss": 0.746,
+      "step": 479770
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 1.9982015764903956e-05,
+      "loss": 0.7161,
+      "step": 479780
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 1.998086783500421e-05,
+      "loss": 0.702,
+      "step": 479790
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 1.9979719905104462e-05,
+      "loss": 0.7886,
+      "step": 479800
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 1.9978571975204712e-05,
+      "loss": 0.6964,
+      "step": 479810
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 1.997742404530497e-05,
+      "loss": 0.7285,
+      "step": 479820
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 1.997627611540522e-05,
+      "loss": 0.8353,
+      "step": 479830
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 1.9975128185505468e-05,
+      "loss": 0.7294,
+      "step": 479840
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 1.9973980255605725e-05,
+      "loss": 0.7568,
+      "step": 479850
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 1.9972832325705975e-05,
+      "loss": 0.8491,
+      "step": 479860
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 1.9971684395806228e-05,
+      "loss": 0.7468,
+      "step": 479870
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 1.9970536465906478e-05,
+      "loss": 0.7312,
+      "step": 479880
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 1.9969388536006734e-05,
+      "loss": 0.7688,
+      "step": 479890
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 1.9968240606106984e-05,
+      "loss": 0.7574,
+      "step": 479900
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 1.9967092676207238e-05,
+      "loss": 0.8443,
+      "step": 479910
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 1.996594474630749e-05,
+      "loss": 0.7548,
+      "step": 479920
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 1.9964796816407744e-05,
+      "loss": 0.7153,
+      "step": 479930
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 1.9963648886507994e-05,
+      "loss": 0.8046,
+      "step": 479940
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 1.996250095660825e-05,
+      "loss": 0.724,
+      "step": 479950
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 1.99613530267085e-05,
+      "loss": 0.7262,
+      "step": 479960
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 1.996020509680875e-05,
+      "loss": 0.8093,
+      "step": 479970
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 1.9959057166909007e-05,
+      "loss": 0.776,
+      "step": 479980
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 1.9957909237009257e-05,
+      "loss": 0.7686,
+      "step": 479990
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 1.995676130710951e-05,
+      "loss": 0.7393,
+      "step": 480000
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 1.9955613377209763e-05,
+      "loss": 0.7194,
+      "step": 480010
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 1.9954465447310016e-05,
+      "loss": 0.8019,
+      "step": 480020
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 1.9953317517410266e-05,
+      "loss": 0.8353,
+      "step": 480030
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 1.9952169587510523e-05,
+      "loss": 0.6778,
+      "step": 480040
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 1.9951021657610773e-05,
+      "loss": 0.7546,
+      "step": 480050
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 1.9949873727711026e-05,
+      "loss": 0.717,
+      "step": 480060
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 1.994872579781128e-05,
+      "loss": 0.8254,
+      "step": 480070
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 1.9947577867911532e-05,
+      "loss": 0.7475,
+      "step": 480080
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 1.9946429938011782e-05,
+      "loss": 0.7329,
+      "step": 480090
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 1.994528200811204e-05,
+      "loss": 0.7532,
+      "step": 480100
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 1.994413407821229e-05,
+      "loss": 0.8168,
+      "step": 480110
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 1.994298614831254e-05,
+      "loss": 0.7173,
+      "step": 480120
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 1.9941838218412795e-05,
+      "loss": 0.7649,
+      "step": 480130
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 1.9940690288513045e-05,
+      "loss": 0.9059,
+      "step": 480140
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 1.99395423586133e-05,
+      "loss": 0.7733,
+      "step": 480150
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 1.993839442871355e-05,
+      "loss": 0.7215,
+      "step": 480160
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 1.9937246498813805e-05,
+      "loss": 0.6809,
+      "step": 480170
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 1.9936098568914055e-05,
+      "loss": 0.7655,
+      "step": 480180
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 1.993495063901431e-05,
+      "loss": 0.7129,
+      "step": 480190
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 1.993380270911456e-05,
+      "loss": 0.827,
+      "step": 480200
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 1.9932654779214814e-05,
+      "loss": 0.8004,
+      "step": 480210
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 1.9931506849315068e-05,
+      "loss": 0.7138,
+      "step": 480220
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 1.993035891941532e-05,
+      "loss": 0.7115,
+      "step": 480230
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 1.992921098951557e-05,
+      "loss": 0.7201,
+      "step": 480240
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 1.9928063059615827e-05,
+      "loss": 0.73,
+      "step": 480250
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 1.9926915129716077e-05,
+      "loss": 0.7776,
+      "step": 480260
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 1.9925767199816327e-05,
+      "loss": 0.7321,
+      "step": 480270
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 1.9924619269916584e-05,
+      "loss": 0.7661,
+      "step": 480280
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 1.9923471340016834e-05,
+      "loss": 0.7199,
+      "step": 480290
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 1.9922323410117087e-05,
+      "loss": 0.829,
+      "step": 480300
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 1.992117548021734e-05,
+      "loss": 0.7726,
+      "step": 480310
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 1.9920027550317593e-05,
+      "loss": 0.7501,
+      "step": 480320
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 1.9918879620417843e-05,
+      "loss": 0.8075,
+      "step": 480330
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 1.99177316905181e-05,
+      "loss": 0.8156,
+      "step": 480340
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 1.991658376061835e-05,
+      "loss": 0.8102,
+      "step": 480350
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 1.9915435830718603e-05,
+      "loss": 0.7815,
+      "step": 480360
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 1.9914287900818856e-05,
+      "loss": 0.7354,
+      "step": 480370
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 1.991313997091911e-05,
+      "loss": 0.7437,
+      "step": 480380
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 1.991199204101936e-05,
+      "loss": 0.7839,
+      "step": 480390
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 1.9910844111119616e-05,
+      "loss": 0.7945,
+      "step": 480400
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 1.9909696181219866e-05,
+      "loss": 0.7246,
+      "step": 480410
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 1.9908548251320116e-05,
+      "loss": 0.7869,
+      "step": 480420
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 1.9907400321420372e-05,
+      "loss": 0.8058,
+      "step": 480430
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 1.9906252391520622e-05,
+      "loss": 0.7224,
+      "step": 480440
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 1.9905104461620875e-05,
+      "loss": 0.8158,
+      "step": 480450
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 1.990395653172113e-05,
+      "loss": 0.7064,
+      "step": 480460
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 1.9902808601821382e-05,
+      "loss": 0.8368,
+      "step": 480470
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 1.9901660671921632e-05,
+      "loss": 0.8274,
+      "step": 480480
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 1.990051274202189e-05,
+      "loss": 0.8079,
+      "step": 480490
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 1.9899364812122138e-05,
+      "loss": 0.6866,
+      "step": 480500
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 1.989821688222239e-05,
+      "loss": 0.713,
+      "step": 480510
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 1.9897068952322645e-05,
+      "loss": 0.7411,
+      "step": 480520
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 1.9895921022422898e-05,
+      "loss": 0.7841,
+      "step": 480530
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 1.9894773092523148e-05,
+      "loss": 0.7873,
+      "step": 480540
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 1.9893625162623404e-05,
+      "loss": 0.8154,
+      "step": 480550
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 1.9892477232723654e-05,
+      "loss": 0.6779,
+      "step": 480560
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 1.9891329302823904e-05,
+      "loss": 0.7298,
+      "step": 480570
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 1.9890181372924157e-05,
+      "loss": 0.7615,
+      "step": 480580
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 1.988903344302441e-05,
+      "loss": 0.7608,
+      "step": 480590
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 1.9887885513124664e-05,
+      "loss": 0.7751,
+      "step": 480600
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 1.9886737583224914e-05,
+      "loss": 0.7522,
+      "step": 480610
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 1.988558965332517e-05,
+      "loss": 0.759,
+      "step": 480620
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 1.988444172342542e-05,
+      "loss": 0.8182,
+      "step": 480630
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 1.988329379352567e-05,
+      "loss": 0.7993,
+      "step": 480640
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 1.9882145863625927e-05,
+      "loss": 0.7509,
+      "step": 480650
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 1.988099793372618e-05,
+      "loss": 0.8068,
+      "step": 480660
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 1.987985000382643e-05,
+      "loss": 0.7182,
+      "step": 480670
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 1.9878702073926686e-05,
+      "loss": 0.7718,
+      "step": 480680
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 1.9877554144026936e-05,
+      "loss": 0.8202,
+      "step": 480690
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 1.9876406214127186e-05,
+      "loss": 0.8226,
+      "step": 480700
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 1.9875258284227443e-05,
+      "loss": 0.795,
+      "step": 480710
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 1.9874110354327693e-05,
+      "loss": 0.7735,
+      "step": 480720
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 1.9872962424427946e-05,
+      "loss": 0.7918,
+      "step": 480730
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 1.98718144945282e-05,
+      "loss": 0.7722,
+      "step": 480740
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 1.9870666564628452e-05,
+      "loss": 0.7813,
+      "step": 480750
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 1.9869518634728702e-05,
+      "loss": 0.6811,
+      "step": 480760
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 1.986837070482896e-05,
+      "loss": 0.7705,
+      "step": 480770
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 1.986722277492921e-05,
+      "loss": 0.7362,
+      "step": 480780
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 1.986607484502946e-05,
+      "loss": 0.7938,
+      "step": 480790
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 1.9864926915129715e-05,
+      "loss": 0.6775,
+      "step": 480800
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 1.986377898522997e-05,
+      "loss": 0.6968,
+      "step": 480810
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 1.9862631055330218e-05,
+      "loss": 0.7432,
+      "step": 480820
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 1.9861483125430475e-05,
+      "loss": 0.8364,
+      "step": 480830
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 1.9860335195530725e-05,
+      "loss": 0.6879,
+      "step": 480840
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 1.9859187265630975e-05,
+      "loss": 0.7909,
+      "step": 480850
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 1.985803933573123e-05,
+      "loss": 0.6729,
+      "step": 480860
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 1.985689140583148e-05,
+      "loss": 0.7155,
+      "step": 480870
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 1.9855743475931734e-05,
+      "loss": 0.755,
+      "step": 480880
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 1.9854595546031988e-05,
+      "loss": 0.7195,
+      "step": 480890
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 1.985344761613224e-05,
+      "loss": 0.777,
+      "step": 480900
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 1.985229968623249e-05,
+      "loss": 0.7735,
+      "step": 480910
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 1.9851151756332747e-05,
+      "loss": 0.8178,
+      "step": 480920
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 1.9850003826432997e-05,
+      "loss": 0.7966,
+      "step": 480930
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 1.9848855896533247e-05,
+      "loss": 0.7769,
+      "step": 480940
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 1.9847707966633504e-05,
+      "loss": 0.8443,
+      "step": 480950
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 1.9846560036733757e-05,
+      "loss": 0.7602,
+      "step": 480960
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 1.9845412106834007e-05,
+      "loss": 0.8082,
+      "step": 480970
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 1.9844264176934263e-05,
+      "loss": 0.7315,
+      "step": 480980
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 1.9843116247034513e-05,
+      "loss": 0.7049,
+      "step": 480990
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 1.9841968317134763e-05,
+      "loss": 0.7762,
+      "step": 481000
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 1.984082038723502e-05,
+      "loss": 0.7344,
+      "step": 481010
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 1.983967245733527e-05,
+      "loss": 0.7541,
+      "step": 481020
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 1.9838524527435523e-05,
+      "loss": 0.7559,
+      "step": 481030
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 1.9837376597535776e-05,
+      "loss": 0.7737,
+      "step": 481040
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 1.983622866763603e-05,
+      "loss": 0.7877,
+      "step": 481050
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 1.983508073773628e-05,
+      "loss": 0.7471,
+      "step": 481060
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 1.9833932807836536e-05,
+      "loss": 0.8264,
+      "step": 481070
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 1.9832784877936786e-05,
+      "loss": 0.7617,
+      "step": 481080
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 1.9831636948037035e-05,
+      "loss": 0.784,
+      "step": 481090
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 1.9830489018137292e-05,
+      "loss": 0.7756,
+      "step": 481100
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 1.9829341088237545e-05,
+      "loss": 0.7026,
+      "step": 481110
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 1.9828193158337795e-05,
+      "loss": 0.765,
+      "step": 481120
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 1.9827045228438052e-05,
+      "loss": 0.7783,
+      "step": 481130
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 1.98258972985383e-05,
+      "loss": 0.7703,
+      "step": 481140
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 1.982474936863855e-05,
+      "loss": 0.7211,
+      "step": 481150
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 1.9823601438738808e-05,
+      "loss": 0.7858,
+      "step": 481160
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 1.9822453508839058e-05,
+      "loss": 0.7962,
+      "step": 481170
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 1.982130557893931e-05,
+      "loss": 0.6769,
+      "step": 481180
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 1.9820157649039564e-05,
+      "loss": 0.7957,
+      "step": 481190
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 1.9819009719139818e-05,
+      "loss": 0.7752,
+      "step": 481200
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 1.9817861789240068e-05,
+      "loss": 0.7568,
+      "step": 481210
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 1.9816713859340324e-05,
+      "loss": 0.7709,
+      "step": 481220
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 1.9815565929440574e-05,
+      "loss": 0.6598,
+      "step": 481230
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 1.9814417999540824e-05,
+      "loss": 0.8197,
+      "step": 481240
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 1.9813270069641077e-05,
+      "loss": 0.7733,
+      "step": 481250
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 1.9812122139741334e-05,
+      "loss": 0.6849,
+      "step": 481260
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 1.9810974209841584e-05,
+      "loss": 0.8202,
+      "step": 481270
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 1.9809826279941833e-05,
+      "loss": 0.8285,
+      "step": 481280
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 1.980867835004209e-05,
+      "loss": 0.7838,
+      "step": 481290
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 1.980753042014234e-05,
+      "loss": 0.7791,
+      "step": 481300
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 1.9806382490242593e-05,
+      "loss": 0.6976,
+      "step": 481310
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 1.9805234560342846e-05,
+      "loss": 0.6933,
+      "step": 481320
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 1.98040866304431e-05,
+      "loss": 0.8342,
+      "step": 481330
+    },
+    {
+      "epoch": 22.08,
+      "learning_rate": 1.980293870054335e-05,
+      "loss": 0.739,
+      "step": 481340
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 1.9801790770643606e-05,
+      "loss": 0.7571,
+      "step": 481350
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 1.9800642840743856e-05,
+      "loss": 0.7771,
+      "step": 481360
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 1.9799494910844106e-05,
+      "loss": 0.8602,
+      "step": 481370
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 1.9798346980944363e-05,
+      "loss": 0.6908,
+      "step": 481380
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 1.9797199051044612e-05,
+      "loss": 0.7168,
+      "step": 481390
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 1.9796051121144866e-05,
+      "loss": 0.7492,
+      "step": 481400
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 1.9794903191245122e-05,
+      "loss": 0.7462,
+      "step": 481410
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 1.9793755261345372e-05,
+      "loss": 0.7788,
+      "step": 481420
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 1.9792607331445622e-05,
+      "loss": 0.7927,
+      "step": 481430
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 1.979145940154588e-05,
+      "loss": 0.6661,
+      "step": 481440
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 1.979031147164613e-05,
+      "loss": 0.7511,
+      "step": 481450
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 1.978916354174638e-05,
+      "loss": 0.7535,
+      "step": 481460
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 1.9788015611846635e-05,
+      "loss": 0.7738,
+      "step": 481470
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 1.9786867681946888e-05,
+      "loss": 0.7544,
+      "step": 481480
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 1.9785719752047138e-05,
+      "loss": 0.7439,
+      "step": 481490
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 1.9784571822147395e-05,
+      "loss": 0.818,
+      "step": 481500
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 1.9783423892247645e-05,
+      "loss": 0.7429,
+      "step": 481510
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 1.9782275962347894e-05,
+      "loss": 0.6671,
+      "step": 481520
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 1.978112803244815e-05,
+      "loss": 0.9208,
+      "step": 481530
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 1.97799801025484e-05,
+      "loss": 0.7723,
+      "step": 481540
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 1.9778832172648654e-05,
+      "loss": 0.751,
+      "step": 481550
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 1.977768424274891e-05,
+      "loss": 0.7805,
+      "step": 481560
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 1.977653631284916e-05,
+      "loss": 0.7932,
+      "step": 481570
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 1.977538838294941e-05,
+      "loss": 0.8444,
+      "step": 481580
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 1.9774240453049667e-05,
+      "loss": 0.7643,
+      "step": 481590
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 1.9773092523149917e-05,
+      "loss": 0.7726,
+      "step": 481600
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 1.977194459325017e-05,
+      "loss": 0.6601,
+      "step": 481610
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 1.9770796663350423e-05,
+      "loss": 0.7109,
+      "step": 481620
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 1.9769648733450677e-05,
+      "loss": 0.7749,
+      "step": 481630
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 1.9768500803550927e-05,
+      "loss": 0.8077,
+      "step": 481640
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 1.9767352873651183e-05,
+      "loss": 0.7387,
+      "step": 481650
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 1.9766204943751433e-05,
+      "loss": 0.8894,
+      "step": 481660
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 1.9765057013851683e-05,
+      "loss": 0.6901,
+      "step": 481670
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 1.976390908395194e-05,
+      "loss": 0.6809,
+      "step": 481680
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 1.976276115405219e-05,
+      "loss": 0.7543,
+      "step": 481690
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 1.9761613224152443e-05,
+      "loss": 0.8245,
+      "step": 481700
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 1.97604652942527e-05,
+      "loss": 0.6592,
+      "step": 481710
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 1.975931736435295e-05,
+      "loss": 0.7179,
+      "step": 481720
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 1.97581694344532e-05,
+      "loss": 0.7294,
+      "step": 481730
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 1.9757021504553456e-05,
+      "loss": 0.8126,
+      "step": 481740
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 1.9755873574653705e-05,
+      "loss": 0.7142,
+      "step": 481750
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 1.975472564475396e-05,
+      "loss": 0.6439,
+      "step": 481760
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 1.9753577714854212e-05,
+      "loss": 0.7432,
+      "step": 481770
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 1.9752429784954465e-05,
+      "loss": 0.8258,
+      "step": 481780
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 1.9751281855054715e-05,
+      "loss": 0.7526,
+      "step": 481790
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 1.975013392515497e-05,
+      "loss": 0.712,
+      "step": 481800
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 1.974898599525522e-05,
+      "loss": 0.723,
+      "step": 481810
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 1.974783806535547e-05,
+      "loss": 0.7578,
+      "step": 481820
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 1.9746690135455728e-05,
+      "loss": 0.7377,
+      "step": 481830
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 1.9745542205555978e-05,
+      "loss": 0.8231,
+      "step": 481840
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 1.974439427565623e-05,
+      "loss": 0.7145,
+      "step": 481850
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 1.9743246345756488e-05,
+      "loss": 0.8031,
+      "step": 481860
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 1.9742098415856738e-05,
+      "loss": 0.7353,
+      "step": 481870
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 1.9740950485956987e-05,
+      "loss": 0.7308,
+      "step": 481880
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 1.9739802556057244e-05,
+      "loss": 0.7573,
+      "step": 481890
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 1.9738654626157494e-05,
+      "loss": 0.7511,
+      "step": 481900
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 1.9737506696257747e-05,
+      "loss": 0.8236,
+      "step": 481910
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 1.9736358766358e-05,
+      "loss": 0.766,
+      "step": 481920
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 1.9735210836458254e-05,
+      "loss": 0.6797,
+      "step": 481930
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 1.9734062906558503e-05,
+      "loss": 0.8,
+      "step": 481940
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 1.9732914976658753e-05,
+      "loss": 0.7295,
+      "step": 481950
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 1.973176704675901e-05,
+      "loss": 0.7776,
+      "step": 481960
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 1.973061911685926e-05,
+      "loss": 0.7534,
+      "step": 481970
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 1.9729471186959513e-05,
+      "loss": 0.804,
+      "step": 481980
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 1.9728323257059766e-05,
+      "loss": 0.7236,
+      "step": 481990
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 1.972717532716002e-05,
+      "loss": 0.7548,
+      "step": 482000
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 1.972602739726027e-05,
+      "loss": 0.7169,
+      "step": 482010
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 1.9724879467360526e-05,
+      "loss": 0.7626,
+      "step": 482020
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 1.9723731537460776e-05,
+      "loss": 0.7587,
+      "step": 482030
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 1.972258360756103e-05,
+      "loss": 0.7272,
+      "step": 482040
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 1.9721435677661282e-05,
+      "loss": 0.7665,
+      "step": 482050
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 1.9720287747761536e-05,
+      "loss": 0.7558,
+      "step": 482060
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 1.9719139817861785e-05,
+      "loss": 0.7562,
+      "step": 482070
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 1.9717991887962042e-05,
+      "loss": 0.7345,
+      "step": 482080
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 1.9716843958062292e-05,
+      "loss": 0.7427,
+      "step": 482090
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 1.9715696028162542e-05,
+      "loss": 0.7102,
+      "step": 482100
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 1.97145480982628e-05,
+      "loss": 0.8171,
+      "step": 482110
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 1.9713400168363048e-05,
+      "loss": 0.746,
+      "step": 482120
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 1.97122522384633e-05,
+      "loss": 0.6455,
+      "step": 482130
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 1.9711104308563555e-05,
+      "loss": 0.8084,
+      "step": 482140
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 1.9709956378663808e-05,
+      "loss": 0.8165,
+      "step": 482150
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 1.9708808448764058e-05,
+      "loss": 0.7337,
+      "step": 482160
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 1.9707660518864314e-05,
+      "loss": 0.7245,
+      "step": 482170
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 1.9706512588964564e-05,
+      "loss": 0.6349,
+      "step": 482180
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 1.9705364659064818e-05,
+      "loss": 0.7842,
+      "step": 482190
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 1.970421672916507e-05,
+      "loss": 0.7613,
+      "step": 482200
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 1.9703068799265324e-05,
+      "loss": 0.8402,
+      "step": 482210
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 1.9701920869365574e-05,
+      "loss": 0.8371,
+      "step": 482220
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 1.970077293946583e-05,
+      "loss": 0.7503,
+      "step": 482230
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 1.969962500956608e-05,
+      "loss": 0.8245,
+      "step": 482240
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 1.969847707966633e-05,
+      "loss": 0.8148,
+      "step": 482250
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 1.9697329149766587e-05,
+      "loss": 0.7958,
+      "step": 482260
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 1.9696181219866837e-05,
+      "loss": 0.8343,
+      "step": 482270
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 1.969503328996709e-05,
+      "loss": 0.7646,
+      "step": 482280
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 1.9693885360067347e-05,
+      "loss": 0.6848,
+      "step": 482290
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 1.9692737430167596e-05,
+      "loss": 0.7474,
+      "step": 482300
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 1.9691589500267846e-05,
+      "loss": 0.7236,
+      "step": 482310
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 1.9690441570368103e-05,
+      "loss": 0.8053,
+      "step": 482320
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 1.9689293640468353e-05,
+      "loss": 0.7389,
+      "step": 482330
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 1.9688145710568606e-05,
+      "loss": 0.744,
+      "step": 482340
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 1.968699778066886e-05,
+      "loss": 0.7287,
+      "step": 482350
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 1.9685849850769113e-05,
+      "loss": 0.6672,
+      "step": 482360
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 1.9684701920869362e-05,
+      "loss": 0.774,
+      "step": 482370
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 1.968355399096962e-05,
+      "loss": 0.7566,
+      "step": 482380
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 1.968240606106987e-05,
+      "loss": 0.7656,
+      "step": 482390
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 1.968125813117012e-05,
+      "loss": 0.752,
+      "step": 482400
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 1.9680110201270375e-05,
+      "loss": 0.8146,
+      "step": 482410
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 1.9678962271370625e-05,
+      "loss": 0.7694,
+      "step": 482420
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 1.967781434147088e-05,
+      "loss": 0.7907,
+      "step": 482430
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 1.9676666411571135e-05,
+      "loss": 0.6605,
+      "step": 482440
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 1.9675518481671385e-05,
+      "loss": 0.7652,
+      "step": 482450
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 1.9674370551771635e-05,
+      "loss": 0.7745,
+      "step": 482460
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 1.967322262187189e-05,
+      "loss": 0.7149,
+      "step": 482470
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 1.967207469197214e-05,
+      "loss": 0.7403,
+      "step": 482480
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 1.9670926762072395e-05,
+      "loss": 0.7818,
+      "step": 482490
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 1.9669778832172648e-05,
+      "loss": 0.7274,
+      "step": 482500
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 1.96686309022729e-05,
+      "loss": 0.8066,
+      "step": 482510
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 1.966748297237315e-05,
+      "loss": 0.7529,
+      "step": 482520
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 1.9666335042473407e-05,
+      "loss": 0.7246,
+      "step": 482530
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 1.9665187112573657e-05,
+      "loss": 0.7802,
+      "step": 482540
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 1.9664039182673907e-05,
+      "loss": 0.7987,
+      "step": 482550
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 1.9662891252774164e-05,
+      "loss": 0.6929,
+      "step": 482560
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 1.9661743322874414e-05,
+      "loss": 0.7231,
+      "step": 482570
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 1.9660595392974667e-05,
+      "loss": 0.8394,
+      "step": 482580
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 1.9659447463074924e-05,
+      "loss": 0.6856,
+      "step": 482590
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 1.9658299533175173e-05,
+      "loss": 0.7604,
+      "step": 482600
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 1.9657151603275423e-05,
+      "loss": 0.6892,
+      "step": 482610
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 1.965600367337568e-05,
+      "loss": 0.7848,
+      "step": 482620
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 1.965485574347593e-05,
+      "loss": 0.7355,
+      "step": 482630
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 1.9653707813576183e-05,
+      "loss": 0.678,
+      "step": 482640
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 1.9652559883676433e-05,
+      "loss": 0.7808,
+      "step": 482650
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 1.965141195377669e-05,
+      "loss": 0.6975,
+      "step": 482660
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 1.965026402387694e-05,
+      "loss": 0.6653,
+      "step": 482670
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 1.964911609397719e-05,
+      "loss": 0.6749,
+      "step": 482680
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 1.9647968164077446e-05,
+      "loss": 0.7463,
+      "step": 482690
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 1.9646820234177696e-05,
+      "loss": 0.7633,
+      "step": 482700
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 1.964567230427795e-05,
+      "loss": 0.8089,
+      "step": 482710
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 1.9644524374378202e-05,
+      "loss": 0.7771,
+      "step": 482720
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 1.9643376444478455e-05,
+      "loss": 0.8251,
+      "step": 482730
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 1.9642228514578705e-05,
+      "loss": 0.7881,
+      "step": 482740
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 1.9641080584678962e-05,
+      "loss": 0.7986,
+      "step": 482750
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 1.9639932654779212e-05,
+      "loss": 0.8149,
+      "step": 482760
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 1.9638784724879465e-05,
+      "loss": 0.7378,
+      "step": 482770
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 1.9637636794979718e-05,
+      "loss": 0.773,
+      "step": 482780
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 1.963648886507997e-05,
+      "loss": 0.7762,
+      "step": 482790
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 1.963534093518022e-05,
+      "loss": 0.7421,
+      "step": 482800
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 1.9634193005280478e-05,
+      "loss": 0.7898,
+      "step": 482810
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 1.9633045075380728e-05,
+      "loss": 0.743,
+      "step": 482820
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 1.9631897145480978e-05,
+      "loss": 0.794,
+      "step": 482830
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 1.9630749215581234e-05,
+      "loss": 0.8102,
+      "step": 482840
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 1.9629601285681484e-05,
+      "loss": 0.8391,
+      "step": 482850
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 1.9628453355781737e-05,
+      "loss": 0.8122,
+      "step": 482860
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 1.962730542588199e-05,
+      "loss": 0.7464,
+      "step": 482870
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 1.9626157495982244e-05,
+      "loss": 0.8285,
+      "step": 482880
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 1.9625009566082494e-05,
+      "loss": 0.7258,
+      "step": 482890
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 1.962386163618275e-05,
+      "loss": 0.7529,
+      "step": 482900
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 1.9622713706283e-05,
+      "loss": 0.7718,
+      "step": 482910
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 1.9621565776383253e-05,
+      "loss": 0.7923,
+      "step": 482920
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 1.9620417846483507e-05,
+      "loss": 0.7426,
+      "step": 482930
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 1.961926991658376e-05,
+      "loss": 0.8393,
+      "step": 482940
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 1.961812198668401e-05,
+      "loss": 0.7808,
+      "step": 482950
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 1.9616974056784266e-05,
+      "loss": 0.8196,
+      "step": 482960
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 1.9615826126884516e-05,
+      "loss": 0.7274,
+      "step": 482970
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 1.9614678196984766e-05,
+      "loss": 0.8091,
+      "step": 482980
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 1.9613530267085023e-05,
+      "loss": 0.7899,
+      "step": 482990
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 1.9612382337185273e-05,
+      "loss": 0.7806,
+      "step": 483000
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 1.9611234407285526e-05,
+      "loss": 0.7591,
+      "step": 483010
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 1.961008647738578e-05,
+      "loss": 0.79,
+      "step": 483020
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 1.9608938547486032e-05,
+      "loss": 0.7568,
+      "step": 483030
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 1.9607790617586282e-05,
+      "loss": 0.8102,
+      "step": 483040
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 1.960664268768654e-05,
+      "loss": 0.7107,
+      "step": 483050
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 1.960549475778679e-05,
+      "loss": 0.6749,
+      "step": 483060
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 1.9604346827887042e-05,
+      "loss": 0.7693,
+      "step": 483070
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 1.9603198897987295e-05,
+      "loss": 0.7851,
+      "step": 483080
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 1.960205096808755e-05,
+      "loss": 0.7104,
+      "step": 483090
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 1.9600903038187798e-05,
+      "loss": 0.7604,
+      "step": 483100
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 1.9599755108288055e-05,
+      "loss": 0.8499,
+      "step": 483110
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 1.9598607178388305e-05,
+      "loss": 0.7875,
+      "step": 483120
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 1.9597459248488555e-05,
+      "loss": 0.7714,
+      "step": 483130
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 1.959631131858881e-05,
+      "loss": 0.7421,
+      "step": 483140
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 1.959516338868906e-05,
+      "loss": 0.7873,
+      "step": 483150
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 1.9594015458789314e-05,
+      "loss": 0.7468,
+      "step": 483160
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 1.9592867528889568e-05,
+      "loss": 0.773,
+      "step": 483170
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 1.959171959898982e-05,
+      "loss": 0.808,
+      "step": 483180
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 1.959057166909007e-05,
+      "loss": 0.8124,
+      "step": 483190
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 1.9589423739190327e-05,
+      "loss": 0.7587,
+      "step": 483200
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 1.9588275809290577e-05,
+      "loss": 0.7382,
+      "step": 483210
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 1.958712787939083e-05,
+      "loss": 0.8043,
+      "step": 483220
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 1.9585979949491084e-05,
+      "loss": 0.7742,
+      "step": 483230
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 1.9584832019591337e-05,
+      "loss": 0.7446,
+      "step": 483240
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 1.9583684089691587e-05,
+      "loss": 0.8025,
+      "step": 483250
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 1.9582536159791843e-05,
+      "loss": 0.7096,
+      "step": 483260
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 1.9581388229892093e-05,
+      "loss": 0.7603,
+      "step": 483270
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 1.9580240299992343e-05,
+      "loss": 0.7466,
+      "step": 483280
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 1.95790923700926e-05,
+      "loss": 0.8089,
+      "step": 483290
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 1.957794444019285e-05,
+      "loss": 0.699,
+      "step": 483300
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 1.9576796510293103e-05,
+      "loss": 0.6889,
+      "step": 483310
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 1.9575648580393353e-05,
+      "loss": 0.7635,
+      "step": 483320
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 1.957450065049361e-05,
+      "loss": 0.7811,
+      "step": 483330
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 1.957335272059386e-05,
+      "loss": 0.7474,
+      "step": 483340
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 1.9572204790694112e-05,
+      "loss": 0.7773,
+      "step": 483350
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 1.9571056860794366e-05,
+      "loss": 0.7529,
+      "step": 483360
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 1.956990893089462e-05,
+      "loss": 0.794,
+      "step": 483370
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 1.956876100099487e-05,
+      "loss": 0.7996,
+      "step": 483380
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 1.9567613071095125e-05,
+      "loss": 0.7828,
+      "step": 483390
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 1.9566465141195375e-05,
+      "loss": 0.8395,
+      "step": 483400
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 1.9565317211295625e-05,
+      "loss": 0.7816,
+      "step": 483410
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 1.956416928139588e-05,
+      "loss": 0.7874,
+      "step": 483420
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 1.956302135149613e-05,
+      "loss": 0.7452,
+      "step": 483430
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 1.9561873421596385e-05,
+      "loss": 0.7119,
+      "step": 483440
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 1.9560725491696638e-05,
+      "loss": 0.7244,
+      "step": 483450
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 1.955957756179689e-05,
+      "loss": 0.8359,
+      "step": 483460
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 1.955842963189714e-05,
+      "loss": 0.7691,
+      "step": 483470
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 1.9557281701997398e-05,
+      "loss": 0.7673,
+      "step": 483480
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 1.9556133772097648e-05,
+      "loss": 0.7535,
+      "step": 483490
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 1.95549858421979e-05,
+      "loss": 0.734,
+      "step": 483500
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 1.9553837912298154e-05,
+      "loss": 0.8173,
+      "step": 483510
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 1.9552689982398407e-05,
+      "loss": 0.7152,
+      "step": 483520
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 1.9551542052498657e-05,
+      "loss": 0.7663,
+      "step": 483530
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 1.9550394122598914e-05,
+      "loss": 0.7908,
+      "step": 483540
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 1.9549246192699164e-05,
+      "loss": 0.6695,
+      "step": 483550
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 1.9548098262799414e-05,
+      "loss": 0.6979,
+      "step": 483560
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 1.954695033289967e-05,
+      "loss": 0.7493,
+      "step": 483570
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 1.954580240299992e-05,
+      "loss": 0.7932,
+      "step": 483580
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 1.9544654473100173e-05,
+      "loss": 0.7382,
+      "step": 483590
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 1.9543506543200426e-05,
+      "loss": 0.746,
+      "step": 483600
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 1.954235861330068e-05,
+      "loss": 0.7295,
+      "step": 483610
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 1.954121068340093e-05,
+      "loss": 0.7287,
+      "step": 483620
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 1.9540062753501186e-05,
+      "loss": 0.7202,
+      "step": 483630
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 1.9538914823601436e-05,
+      "loss": 0.7123,
+      "step": 483640
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 1.953776689370169e-05,
+      "loss": 0.7521,
+      "step": 483650
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 1.9536618963801943e-05,
+      "loss": 0.7526,
+      "step": 483660
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 1.9535471033902196e-05,
+      "loss": 0.8013,
+      "step": 483670
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 1.9534323104002446e-05,
+      "loss": 0.8201,
+      "step": 483680
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 1.9533175174102702e-05,
+      "loss": 0.7701,
+      "step": 483690
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 1.9532027244202952e-05,
+      "loss": 0.7019,
+      "step": 483700
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 1.9530879314303202e-05,
+      "loss": 0.8675,
+      "step": 483710
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 1.952973138440346e-05,
+      "loss": 0.8569,
+      "step": 483720
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 1.952858345450371e-05,
+      "loss": 0.7561,
+      "step": 483730
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 1.9527435524603962e-05,
+      "loss": 0.7281,
+      "step": 483740
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 1.9526287594704215e-05,
+      "loss": 0.7632,
+      "step": 483750
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 1.9525139664804468e-05,
+      "loss": 0.7535,
+      "step": 483760
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 1.9523991734904718e-05,
+      "loss": 0.695,
+      "step": 483770
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 1.9522843805004975e-05,
+      "loss": 0.7143,
+      "step": 483780
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 1.9521695875105225e-05,
+      "loss": 0.6845,
+      "step": 483790
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 1.9520547945205478e-05,
+      "loss": 0.757,
+      "step": 483800
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 1.951940001530573e-05,
+      "loss": 0.7851,
+      "step": 483810
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 1.9518252085405984e-05,
+      "loss": 0.806,
+      "step": 483820
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 1.9517104155506234e-05,
+      "loss": 0.842,
+      "step": 483830
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 1.951595622560649e-05,
+      "loss": 0.7939,
+      "step": 483840
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 1.951480829570674e-05,
+      "loss": 0.6773,
+      "step": 483850
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 1.951366036580699e-05,
+      "loss": 0.7761,
+      "step": 483860
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 1.9512512435907247e-05,
+      "loss": 0.7978,
+      "step": 483870
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 1.9511364506007497e-05,
+      "loss": 0.8053,
+      "step": 483880
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 1.951021657610775e-05,
+      "loss": 0.8424,
+      "step": 483890
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 1.9509068646208003e-05,
+      "loss": 0.7338,
+      "step": 483900
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 1.9507920716308257e-05,
+      "loss": 0.7704,
+      "step": 483910
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 1.9506772786408507e-05,
+      "loss": 0.8106,
+      "step": 483920
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 1.9505624856508763e-05,
+      "loss": 0.8294,
+      "step": 483930
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 1.9504476926609013e-05,
+      "loss": 0.7697,
+      "step": 483940
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 1.9503328996709266e-05,
+      "loss": 0.7317,
+      "step": 483950
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 1.950218106680952e-05,
+      "loss": 0.7696,
+      "step": 483960
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 1.9501033136909773e-05,
+      "loss": 0.7592,
+      "step": 483970
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 1.9499885207010023e-05,
+      "loss": 0.7133,
+      "step": 483980
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 1.949873727711028e-05,
+      "loss": 0.7102,
+      "step": 483990
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 1.949758934721053e-05,
+      "loss": 0.8051,
+      "step": 484000
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 1.949644141731078e-05,
+      "loss": 0.6505,
+      "step": 484010
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 1.9495293487411032e-05,
+      "loss": 0.7256,
+      "step": 484020
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 1.9494145557511285e-05,
+      "loss": 0.7395,
+      "step": 484030
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 1.949299762761154e-05,
+      "loss": 0.816,
+      "step": 484040
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 1.949184969771179e-05,
+      "loss": 0.7514,
+      "step": 484050
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 1.9490701767812045e-05,
+      "loss": 0.8161,
+      "step": 484060
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 1.9489553837912295e-05,
+      "loss": 0.7925,
+      "step": 484070
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 1.9488405908012548e-05,
+      "loss": 0.8692,
+      "step": 484080
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 1.94872579781128e-05,
+      "loss": 0.7882,
+      "step": 484090
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 1.9486110048213055e-05,
+      "loss": 0.7802,
+      "step": 484100
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 1.9484962118313305e-05,
+      "loss": 0.7645,
+      "step": 484110
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 1.948381418841356e-05,
+      "loss": 0.7313,
+      "step": 484120
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 1.948266625851381e-05,
+      "loss": 0.7581,
+      "step": 484130
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 1.948151832861406e-05,
+      "loss": 0.7514,
+      "step": 484140
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 1.9480370398714318e-05,
+      "loss": 0.7092,
+      "step": 484150
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 1.9479222468814567e-05,
+      "loss": 0.7194,
+      "step": 484160
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 1.947807453891482e-05,
+      "loss": 0.7452,
+      "step": 484170
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 1.9476926609015074e-05,
+      "loss": 0.7446,
+      "step": 484180
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 1.9475778679115327e-05,
+      "loss": 0.7752,
+      "step": 484190
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 1.9474630749215577e-05,
+      "loss": 0.7285,
+      "step": 484200
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 1.9473482819315834e-05,
+      "loss": 0.7996,
+      "step": 484210
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 1.9472334889416083e-05,
+      "loss": 0.7689,
+      "step": 484220
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 1.9471186959516337e-05,
+      "loss": 0.7491,
+      "step": 484230
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 1.947003902961659e-05,
+      "loss": 0.7554,
+      "step": 484240
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 1.9468891099716843e-05,
+      "loss": 0.8048,
+      "step": 484250
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 1.9467743169817093e-05,
+      "loss": 0.7445,
+      "step": 484260
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 1.946659523991735e-05,
+      "loss": 0.8295,
+      "step": 484270
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 1.94654473100176e-05,
+      "loss": 0.796,
+      "step": 484280
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 1.946429938011785e-05,
+      "loss": 0.7806,
+      "step": 484290
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 1.9463151450218106e-05,
+      "loss": 0.7312,
+      "step": 484300
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 1.9462003520318356e-05,
+      "loss": 0.7984,
+      "step": 484310
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 1.946085559041861e-05,
+      "loss": 0.7547,
+      "step": 484320
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 1.9459707660518862e-05,
+      "loss": 0.7878,
+      "step": 484330
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 1.9458559730619116e-05,
+      "loss": 0.7443,
+      "step": 484340
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 1.9457411800719365e-05,
+      "loss": 0.7358,
+      "step": 484350
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 1.9456263870819622e-05,
+      "loss": 0.7737,
+      "step": 484360
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 1.9455115940919872e-05,
+      "loss": 0.7075,
+      "step": 484370
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 1.9453968011020125e-05,
+      "loss": 0.6801,
+      "step": 484380
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 1.945282008112038e-05,
+      "loss": 0.7371,
+      "step": 484390
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 1.945167215122063e-05,
+      "loss": 0.7861,
+      "step": 484400
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 1.945052422132088e-05,
+      "loss": 0.8404,
+      "step": 484410
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 1.9449376291421138e-05,
+      "loss": 0.7482,
+      "step": 484420
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 1.9448228361521388e-05,
+      "loss": 0.8581,
+      "step": 484430
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 1.9447080431621638e-05,
+      "loss": 0.7329,
+      "step": 484440
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 1.9445932501721895e-05,
+      "loss": 0.7769,
+      "step": 484450
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 1.9444784571822144e-05,
+      "loss": 0.7826,
+      "step": 484460
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 1.9443636641922398e-05,
+      "loss": 0.8027,
+      "step": 484470
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 1.944248871202265e-05,
+      "loss": 0.7964,
+      "step": 484480
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 1.9441340782122904e-05,
+      "loss": 0.773,
+      "step": 484490
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 1.9440192852223154e-05,
+      "loss": 0.7466,
+      "step": 484500
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 1.943904492232341e-05,
+      "loss": 0.8319,
+      "step": 484510
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 1.943789699242366e-05,
+      "loss": 0.7772,
+      "step": 484520
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 1.9436749062523914e-05,
+      "loss": 0.8508,
+      "step": 484530
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 1.9435601132624167e-05,
+      "loss": 0.805,
+      "step": 484540
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 1.943445320272442e-05,
+      "loss": 0.7304,
+      "step": 484550
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 1.943330527282467e-05,
+      "loss": 0.7339,
+      "step": 484560
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 1.9432157342924927e-05,
+      "loss": 0.6941,
+      "step": 484570
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 1.9431009413025176e-05,
+      "loss": 0.7297,
+      "step": 484580
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 1.9429861483125426e-05,
+      "loss": 0.696,
+      "step": 484590
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 1.9428713553225683e-05,
+      "loss": 0.7237,
+      "step": 484600
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 1.9427565623325933e-05,
+      "loss": 0.8259,
+      "step": 484610
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 1.9426417693426186e-05,
+      "loss": 0.813,
+      "step": 484620
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 1.942526976352644e-05,
+      "loss": 0.7382,
+      "step": 484630
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 1.9424121833626693e-05,
+      "loss": 0.7418,
+      "step": 484640
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 1.9422973903726942e-05,
+      "loss": 0.7564,
+      "step": 484650
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 1.94218259738272e-05,
+      "loss": 0.6779,
+      "step": 484660
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 1.942067804392745e-05,
+      "loss": 0.7572,
+      "step": 484670
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 1.9419530114027702e-05,
+      "loss": 0.7113,
+      "step": 484680
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 1.9418382184127952e-05,
+      "loss": 0.7888,
+      "step": 484690
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 1.941723425422821e-05,
+      "loss": 0.7845,
+      "step": 484700
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 1.941608632432846e-05,
+      "loss": 0.6761,
+      "step": 484710
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 1.941493839442871e-05,
+      "loss": 0.751,
+      "step": 484720
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 1.9413790464528965e-05,
+      "loss": 0.6391,
+      "step": 484730
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 1.9412642534629215e-05,
+      "loss": 0.811,
+      "step": 484740
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 1.9411494604729468e-05,
+      "loss": 0.7441,
+      "step": 484750
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 1.941034667482972e-05,
+      "loss": 0.7381,
+      "step": 484760
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 1.9409198744929975e-05,
+      "loss": 0.8048,
+      "step": 484770
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 1.9408050815030224e-05,
+      "loss": 0.7788,
+      "step": 484780
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 1.940690288513048e-05,
+      "loss": 0.793,
+      "step": 484790
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 1.940575495523073e-05,
+      "loss": 0.732,
+      "step": 484800
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 1.9404607025330984e-05,
+      "loss": 0.7129,
+      "step": 484810
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 1.9403459095431237e-05,
+      "loss": 0.7462,
+      "step": 484820
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 1.940231116553149e-05,
+      "loss": 0.7994,
+      "step": 484830
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 1.940116323563174e-05,
+      "loss": 0.7418,
+      "step": 484840
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 1.9400015305731997e-05,
+      "loss": 0.7934,
+      "step": 484850
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 1.9398867375832247e-05,
+      "loss": 0.8243,
+      "step": 484860
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 1.9397719445932497e-05,
+      "loss": 0.7772,
+      "step": 484870
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 1.9396571516032753e-05,
+      "loss": 0.7612,
+      "step": 484880
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 1.9395423586133003e-05,
+      "loss": 0.7015,
+      "step": 484890
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 1.9394275656233257e-05,
+      "loss": 0.7688,
+      "step": 484900
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 1.939312772633351e-05,
+      "loss": 0.7583,
+      "step": 484910
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 1.9391979796433763e-05,
+      "loss": 0.7147,
+      "step": 484920
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 1.9390831866534013e-05,
+      "loss": 0.6353,
+      "step": 484930
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 1.938968393663427e-05,
+      "loss": 0.8915,
+      "step": 484940
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 1.938853600673452e-05,
+      "loss": 0.8194,
+      "step": 484950
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 1.9387388076834773e-05,
+      "loss": 0.7965,
+      "step": 484960
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 1.9386240146935026e-05,
+      "loss": 0.7346,
+      "step": 484970
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 1.938509221703528e-05,
+      "loss": 0.7877,
+      "step": 484980
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 1.938394428713553e-05,
+      "loss": 0.7551,
+      "step": 484990
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 1.9382796357235786e-05,
+      "loss": 0.8085,
+      "step": 485000
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 1.9381648427336035e-05,
+      "loss": 0.7347,
+      "step": 485010
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 1.9380500497436285e-05,
+      "loss": 0.6992,
+      "step": 485020
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 1.9379352567536542e-05,
+      "loss": 0.7481,
+      "step": 485030
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 1.9378204637636792e-05,
+      "loss": 0.814,
+      "step": 485040
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 1.9377056707737045e-05,
+      "loss": 0.6967,
+      "step": 485050
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 1.9375908777837298e-05,
+      "loss": 0.7977,
+      "step": 485060
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 1.937476084793755e-05,
+      "loss": 0.7549,
+      "step": 485070
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 1.93736129180378e-05,
+      "loss": 0.7899,
+      "step": 485080
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 1.9372464988138058e-05,
+      "loss": 0.7777,
+      "step": 485090
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 1.9371317058238308e-05,
+      "loss": 0.6749,
+      "step": 485100
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 1.937016912833856e-05,
+      "loss": 0.7457,
+      "step": 485110
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 1.9369021198438814e-05,
+      "loss": 0.7286,
+      "step": 485120
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 1.9367873268539068e-05,
+      "loss": 0.7806,
+      "step": 485130
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 1.9366725338639317e-05,
+      "loss": 0.7852,
+      "step": 485140
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 1.9365577408739574e-05,
+      "loss": 0.7559,
+      "step": 485150
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 1.9364429478839824e-05,
+      "loss": 0.7347,
+      "step": 485160
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 1.9363281548940074e-05,
+      "loss": 0.75,
+      "step": 485170
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 1.936213361904033e-05,
+      "loss": 0.7817,
+      "step": 485180
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 1.936098568914058e-05,
+      "loss": 0.7034,
+      "step": 485190
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 1.9359837759240833e-05,
+      "loss": 0.7902,
+      "step": 485200
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 1.9358689829341087e-05,
+      "loss": 0.7793,
+      "step": 485210
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 1.935754189944134e-05,
+      "loss": 0.8282,
+      "step": 485220
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 1.935639396954159e-05,
+      "loss": 0.7768,
+      "step": 485230
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 1.9355246039641846e-05,
+      "loss": 0.7891,
+      "step": 485240
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 1.9354098109742096e-05,
+      "loss": 0.7564,
+      "step": 485250
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 1.935295017984235e-05,
+      "loss": 0.7651,
+      "step": 485260
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 1.9351802249942603e-05,
+      "loss": 0.8131,
+      "step": 485270
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 1.9350654320042856e-05,
+      "loss": 0.7799,
+      "step": 485280
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 1.9349506390143106e-05,
+      "loss": 0.6887,
+      "step": 485290
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 1.9348358460243363e-05,
+      "loss": 0.8481,
+      "step": 485300
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 1.9347210530343612e-05,
+      "loss": 0.7279,
+      "step": 485310
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 1.9346062600443862e-05,
+      "loss": 0.7956,
+      "step": 485320
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 1.934491467054412e-05,
+      "loss": 0.8469,
+      "step": 485330
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 1.934376674064437e-05,
+      "loss": 0.7133,
+      "step": 485340
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 1.9342618810744622e-05,
+      "loss": 0.742,
+      "step": 485350
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 1.9341470880844875e-05,
+      "loss": 0.7348,
+      "step": 485360
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 1.934032295094513e-05,
+      "loss": 0.7401,
+      "step": 485370
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 1.9339175021045378e-05,
+      "loss": 0.7136,
+      "step": 485380
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 1.933802709114563e-05,
+      "loss": 0.765,
+      "step": 485390
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 1.9336879161245885e-05,
+      "loss": 0.8102,
+      "step": 485400
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 1.9335731231346138e-05,
+      "loss": 0.7848,
+      "step": 485410
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 1.9334583301446388e-05,
+      "loss": 0.7409,
+      "step": 485420
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 1.9333435371546644e-05,
+      "loss": 0.7697,
+      "step": 485430
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 1.9332287441646894e-05,
+      "loss": 0.7778,
+      "step": 485440
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 1.9331139511747144e-05,
+      "loss": 0.8287,
+      "step": 485450
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 1.93299915818474e-05,
+      "loss": 0.7741,
+      "step": 485460
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 1.932884365194765e-05,
+      "loss": 0.7469,
+      "step": 485470
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 1.9327695722047904e-05,
+      "loss": 0.7538,
+      "step": 485480
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 1.9326547792148157e-05,
+      "loss": 0.6876,
+      "step": 485490
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 1.932539986224841e-05,
+      "loss": 0.8158,
+      "step": 485500
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 1.932425193234866e-05,
+      "loss": 0.732,
+      "step": 485510
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 1.9323104002448917e-05,
+      "loss": 0.7451,
+      "step": 485520
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 1.9321956072549167e-05,
+      "loss": 0.7206,
+      "step": 485530
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 1.932080814264942e-05,
+      "loss": 0.822,
+      "step": 485540
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 1.9319660212749673e-05,
+      "loss": 0.7602,
+      "step": 485550
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 1.9318512282849926e-05,
+      "loss": 0.7773,
+      "step": 485560
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 1.9317364352950176e-05,
+      "loss": 0.7541,
+      "step": 485570
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 1.9316216423050433e-05,
+      "loss": 0.7566,
+      "step": 485580
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 1.9315068493150683e-05,
+      "loss": 0.8649,
+      "step": 485590
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 1.9313920563250933e-05,
+      "loss": 0.7602,
+      "step": 485600
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 1.931277263335119e-05,
+      "loss": 0.8299,
+      "step": 485610
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 1.931162470345144e-05,
+      "loss": 0.7908,
+      "step": 485620
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 1.9310476773551692e-05,
+      "loss": 0.7233,
+      "step": 485630
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 1.9309328843651946e-05,
+      "loss": 0.7523,
+      "step": 485640
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 1.93081809137522e-05,
+      "loss": 0.7782,
+      "step": 485650
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 1.930703298385245e-05,
+      "loss": 0.8141,
+      "step": 485660
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 1.9305885053952705e-05,
+      "loss": 0.8145,
+      "step": 485670
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 1.9304737124052955e-05,
+      "loss": 0.7583,
+      "step": 485680
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 1.930358919415321e-05,
+      "loss": 0.7591,
+      "step": 485690
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 1.9302441264253462e-05,
+      "loss": 0.7558,
+      "step": 485700
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 1.9301293334353715e-05,
+      "loss": 0.7872,
+      "step": 485710
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 1.9300145404453965e-05,
+      "loss": 0.8017,
+      "step": 485720
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 1.929899747455422e-05,
+      "loss": 0.6858,
+      "step": 485730
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 1.929784954465447e-05,
+      "loss": 0.7565,
+      "step": 485740
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 1.929670161475472e-05,
+      "loss": 0.8045,
+      "step": 485750
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 1.9295553684854978e-05,
+      "loss": 0.8141,
+      "step": 485760
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 1.9294405754955228e-05,
+      "loss": 0.7843,
+      "step": 485770
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 1.929325782505548e-05,
+      "loss": 0.8116,
+      "step": 485780
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 1.9292109895155734e-05,
+      "loss": 0.7146,
+      "step": 485790
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 1.9290961965255987e-05,
+      "loss": 0.7158,
+      "step": 485800
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 1.9289814035356237e-05,
+      "loss": 0.7194,
+      "step": 485810
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 1.9288666105456494e-05,
+      "loss": 0.8094,
+      "step": 485820
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 1.9287518175556744e-05,
+      "loss": 0.7519,
+      "step": 485830
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 1.9286370245656997e-05,
+      "loss": 0.76,
+      "step": 485840
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 1.928522231575725e-05,
+      "loss": 0.6739,
+      "step": 485850
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 1.9284074385857503e-05,
+      "loss": 0.8053,
+      "step": 485860
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 1.9282926455957753e-05,
+      "loss": 0.7465,
+      "step": 485870
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 1.928177852605801e-05,
+      "loss": 0.6976,
+      "step": 485880
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 1.928063059615826e-05,
+      "loss": 0.7137,
+      "step": 485890
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 1.927948266625851e-05,
+      "loss": 0.6955,
+      "step": 485900
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 1.9278334736358766e-05,
+      "loss": 0.7046,
+      "step": 485910
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 1.9277186806459016e-05,
+      "loss": 0.7368,
+      "step": 485920
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 1.927603887655927e-05,
+      "loss": 0.8304,
+      "step": 485930
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 1.9274890946659523e-05,
+      "loss": 0.7509,
+      "step": 485940
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 1.9273743016759776e-05,
+      "loss": 0.7379,
+      "step": 485950
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 1.9272595086860026e-05,
+      "loss": 0.7441,
+      "step": 485960
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 1.9271447156960282e-05,
+      "loss": 0.8132,
+      "step": 485970
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 1.9270299227060532e-05,
+      "loss": 0.7723,
+      "step": 485980
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 1.9269151297160785e-05,
+      "loss": 0.7827,
+      "step": 485990
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 1.926800336726104e-05,
+      "loss": 0.7681,
+      "step": 486000
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 1.9266855437361292e-05,
+      "loss": 0.7971,
+      "step": 486010
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 1.9265707507461542e-05,
+      "loss": 0.7479,
+      "step": 486020
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 1.92645595775618e-05,
+      "loss": 0.7834,
+      "step": 486030
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 1.9263411647662048e-05,
+      "loss": 0.7599,
+      "step": 486040
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 1.9262263717762298e-05,
+      "loss": 0.696,
+      "step": 486050
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 1.926111578786255e-05,
+      "loss": 0.7527,
+      "step": 486060
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 1.9259967857962805e-05,
+      "loss": 0.8359,
+      "step": 486070
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 1.9258819928063058e-05,
+      "loss": 0.8156,
+      "step": 486080
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 1.9257671998163308e-05,
+      "loss": 0.7637,
+      "step": 486090
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 1.9256524068263564e-05,
+      "loss": 0.7633,
+      "step": 486100
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 1.9255376138363814e-05,
+      "loss": 0.7729,
+      "step": 486110
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 1.9254228208464067e-05,
+      "loss": 0.8231,
+      "step": 486120
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 1.925308027856432e-05,
+      "loss": 0.715,
+      "step": 486130
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 1.9251932348664574e-05,
+      "loss": 0.7542,
+      "step": 486140
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 1.9250784418764824e-05,
+      "loss": 0.7829,
+      "step": 486150
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 1.924963648886508e-05,
+      "loss": 0.851,
+      "step": 486160
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 1.924848855896533e-05,
+      "loss": 0.8008,
+      "step": 486170
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 1.924734062906558e-05,
+      "loss": 0.7299,
+      "step": 486180
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 1.9246192699165837e-05,
+      "loss": 0.8531,
+      "step": 486190
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 1.9245044769266087e-05,
+      "loss": 0.7443,
+      "step": 486200
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 1.924389683936634e-05,
+      "loss": 0.7892,
+      "step": 486210
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 1.9242748909466593e-05,
+      "loss": 0.7143,
+      "step": 486220
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 1.9241600979566846e-05,
+      "loss": 0.717,
+      "step": 486230
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 1.9240453049667096e-05,
+      "loss": 0.8016,
+      "step": 486240
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 1.9239305119767353e-05,
+      "loss": 0.7036,
+      "step": 486250
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 1.9238157189867603e-05,
+      "loss": 0.6789,
+      "step": 486260
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 1.9237009259967856e-05,
+      "loss": 0.72,
+      "step": 486270
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 1.923586133006811e-05,
+      "loss": 0.7361,
+      "step": 486280
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 1.9234713400168362e-05,
+      "loss": 0.7851,
+      "step": 486290
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 1.9233565470268612e-05,
+      "loss": 0.7286,
+      "step": 486300
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 1.923241754036887e-05,
+      "loss": 0.8183,
+      "step": 486310
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 1.923126961046912e-05,
+      "loss": 0.7773,
+      "step": 486320
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 1.923012168056937e-05,
+      "loss": 0.7854,
+      "step": 486330
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 1.9228973750669625e-05,
+      "loss": 0.7963,
+      "step": 486340
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 1.9227825820769875e-05,
+      "loss": 0.6722,
+      "step": 486350
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 1.9226677890870128e-05,
+      "loss": 0.6367,
+      "step": 486360
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 1.922552996097038e-05,
+      "loss": 0.702,
+      "step": 486370
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 1.9224382031070635e-05,
+      "loss": 0.7592,
+      "step": 486380
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 1.9223234101170885e-05,
+      "loss": 0.6889,
+      "step": 486390
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 1.922208617127114e-05,
+      "loss": 0.7458,
+      "step": 486400
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 1.922093824137139e-05,
+      "loss": 0.7914,
+      "step": 486410
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 1.9219790311471644e-05,
+      "loss": 0.8244,
+      "step": 486420
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 1.9218642381571898e-05,
+      "loss": 0.7495,
+      "step": 486430
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 1.921749445167215e-05,
+      "loss": 0.79,
+      "step": 486440
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 1.92163465217724e-05,
+      "loss": 0.7691,
+      "step": 486450
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 1.9215198591872657e-05,
+      "loss": 0.653,
+      "step": 486460
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 1.9214050661972907e-05,
+      "loss": 0.8007,
+      "step": 486470
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 1.9212902732073157e-05,
+      "loss": 0.6882,
+      "step": 486480
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 1.9211754802173414e-05,
+      "loss": 0.7937,
+      "step": 486490
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 1.9210606872273664e-05,
+      "loss": 0.7575,
+      "step": 486500
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 1.9209458942373917e-05,
+      "loss": 0.7203,
+      "step": 486510
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 1.920831101247417e-05,
+      "loss": 0.7721,
+      "step": 486520
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 1.9207163082574423e-05,
+      "loss": 0.6812,
+      "step": 486530
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 1.9206015152674673e-05,
+      "loss": 0.7703,
+      "step": 486540
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 1.920486722277493e-05,
+      "loss": 0.7137,
+      "step": 486550
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 1.920371929287518e-05,
+      "loss": 0.776,
+      "step": 486560
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 1.9202571362975433e-05,
+      "loss": 0.7158,
+      "step": 486570
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 1.9201423433075686e-05,
+      "loss": 0.7046,
+      "step": 486580
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 1.920027550317594e-05,
+      "loss": 0.7552,
+      "step": 486590
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 1.919912757327619e-05,
+      "loss": 0.8182,
+      "step": 486600
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 1.9197979643376446e-05,
+      "loss": 0.8032,
+      "step": 486610
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 1.9196831713476696e-05,
+      "loss": 0.7618,
+      "step": 486620
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 1.9195683783576945e-05,
+      "loss": 0.7219,
+      "step": 486630
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 1.9194535853677202e-05,
+      "loss": 0.7851,
+      "step": 486640
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 1.9193387923777452e-05,
+      "loss": 0.7593,
+      "step": 486650
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 1.9192239993877705e-05,
+      "loss": 0.7215,
+      "step": 486660
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 1.919109206397796e-05,
+      "loss": 0.6966,
+      "step": 486670
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 1.9189944134078212e-05,
+      "loss": 0.8032,
+      "step": 486680
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 1.918879620417846e-05,
+      "loss": 0.7681,
+      "step": 486690
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 1.9187648274278718e-05,
+      "loss": 0.7784,
+      "step": 486700
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 1.9186500344378968e-05,
+      "loss": 0.7587,
+      "step": 486710
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 1.918535241447922e-05,
+      "loss": 0.7421,
+      "step": 486720
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 1.9184204484579475e-05,
+      "loss": 0.7446,
+      "step": 486730
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 1.9183056554679728e-05,
+      "loss": 0.7934,
+      "step": 486740
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 1.9181908624779978e-05,
+      "loss": 0.7839,
+      "step": 486750
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 1.9180760694880227e-05,
+      "loss": 0.7732,
+      "step": 486760
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 1.9179612764980484e-05,
+      "loss": 0.784,
+      "step": 486770
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 1.9178464835080734e-05,
+      "loss": 0.7713,
+      "step": 486780
+    },
+    {
+      "epoch": 22.33,
+      "learning_rate": 1.9177316905180987e-05,
+      "loss": 0.7787,
+      "step": 486790
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 1.917616897528124e-05,
+      "loss": 0.7837,
+      "step": 486800
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 1.9175021045381494e-05,
+      "loss": 0.711,
+      "step": 486810
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 1.9173873115481744e-05,
+      "loss": 0.8061,
+      "step": 486820
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 1.9172725185582e-05,
+      "loss": 0.7555,
+      "step": 486830
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 1.917157725568225e-05,
+      "loss": 0.7817,
+      "step": 486840
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 1.9170429325782503e-05,
+      "loss": 0.7995,
+      "step": 486850
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 1.9169281395882757e-05,
+      "loss": 0.8116,
+      "step": 486860
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 1.916813346598301e-05,
+      "loss": 0.7498,
+      "step": 486870
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 1.916698553608326e-05,
+      "loss": 0.8255,
+      "step": 486880
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 1.9165837606183516e-05,
+      "loss": 0.7472,
+      "step": 486890
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 1.9164689676283766e-05,
+      "loss": 0.8678,
+      "step": 486900
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 1.9163541746384016e-05,
+      "loss": 0.7473,
+      "step": 486910
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 1.9162393816484273e-05,
+      "loss": 0.7576,
+      "step": 486920
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 1.9161245886584522e-05,
+      "loss": 0.7297,
+      "step": 486930
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 1.9160097956684776e-05,
+      "loss": 0.7405,
+      "step": 486940
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 1.915895002678503e-05,
+      "loss": 0.6847,
+      "step": 486950
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 1.9157802096885282e-05,
+      "loss": 0.8031,
+      "step": 486960
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 1.9156654166985532e-05,
+      "loss": 0.8289,
+      "step": 486970
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 1.915550623708579e-05,
+      "loss": 0.7157,
+      "step": 486980
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 1.915435830718604e-05,
+      "loss": 0.8345,
+      "step": 486990
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 1.9153210377286292e-05,
+      "loss": 0.7564,
+      "step": 487000
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 1.9152062447386545e-05,
+      "loss": 0.7831,
+      "step": 487010
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 1.9150914517486798e-05,
+      "loss": 0.7473,
+      "step": 487020
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 1.9149766587587048e-05,
+      "loss": 0.7815,
+      "step": 487030
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 1.9148618657687305e-05,
+      "loss": 0.7348,
+      "step": 487040
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 1.9147470727787555e-05,
+      "loss": 0.7827,
+      "step": 487050
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 1.9146322797887804e-05,
+      "loss": 0.7984,
+      "step": 487060
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 1.914517486798806e-05,
+      "loss": 0.7878,
+      "step": 487070
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 1.914402693808831e-05,
+      "loss": 0.718,
+      "step": 487080
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 1.9142879008188564e-05,
+      "loss": 0.7867,
+      "step": 487090
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 1.9141731078288817e-05,
+      "loss": 0.689,
+      "step": 487100
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 1.914058314838907e-05,
+      "loss": 0.7573,
+      "step": 487110
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 1.913943521848932e-05,
+      "loss": 0.7583,
+      "step": 487120
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 1.9138287288589577e-05,
+      "loss": 0.772,
+      "step": 487130
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 1.9137139358689827e-05,
+      "loss": 0.7323,
+      "step": 487140
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 1.913599142879008e-05,
+      "loss": 0.7907,
+      "step": 487150
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 1.9134843498890333e-05,
+      "loss": 0.804,
+      "step": 487160
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 1.9133695568990587e-05,
+      "loss": 0.7338,
+      "step": 487170
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 1.9132547639090837e-05,
+      "loss": 0.7511,
+      "step": 487180
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 1.9131399709191093e-05,
+      "loss": 0.8588,
+      "step": 487190
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 1.9130251779291343e-05,
+      "loss": 0.7414,
+      "step": 487200
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 1.9129103849391593e-05,
+      "loss": 0.7779,
+      "step": 487210
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 1.912795591949185e-05,
+      "loss": 0.8379,
+      "step": 487220
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 1.91268079895921e-05,
+      "loss": 0.8664,
+      "step": 487230
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 1.9125660059692353e-05,
+      "loss": 0.6882,
+      "step": 487240
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 1.9124512129792606e-05,
+      "loss": 0.7372,
+      "step": 487250
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 1.912336419989286e-05,
+      "loss": 0.7104,
+      "step": 487260
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 1.912221626999311e-05,
+      "loss": 0.725,
+      "step": 487270
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 1.9121068340093366e-05,
+      "loss": 0.7376,
+      "step": 487280
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 1.9119920410193615e-05,
+      "loss": 0.7609,
+      "step": 487290
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 1.911877248029387e-05,
+      "loss": 0.7496,
+      "step": 487300
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 1.9117624550394122e-05,
+      "loss": 0.7709,
+      "step": 487310
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 1.9116476620494375e-05,
+      "loss": 0.7543,
+      "step": 487320
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 1.9115328690594625e-05,
+      "loss": 0.819,
+      "step": 487330
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 1.911418076069488e-05,
+      "loss": 0.7756,
+      "step": 487340
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 1.911303283079513e-05,
+      "loss": 0.7001,
+      "step": 487350
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 1.911188490089538e-05,
+      "loss": 0.6877,
+      "step": 487360
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 1.9110736970995638e-05,
+      "loss": 0.8241,
+      "step": 487370
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 1.9109589041095888e-05,
+      "loss": 0.7032,
+      "step": 487380
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 1.910844111119614e-05,
+      "loss": 0.7901,
+      "step": 487390
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 1.9107293181296394e-05,
+      "loss": 0.7584,
+      "step": 487400
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 1.9106145251396648e-05,
+      "loss": 0.8049,
+      "step": 487410
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 1.9104997321496897e-05,
+      "loss": 0.6447,
+      "step": 487420
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 1.910384939159715e-05,
+      "loss": 0.8068,
+      "step": 487430
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 1.9102701461697404e-05,
+      "loss": 0.8514,
+      "step": 487440
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 1.9101553531797657e-05,
+      "loss": 0.75,
+      "step": 487450
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 1.9100405601897907e-05,
+      "loss": 0.8317,
+      "step": 487460
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 1.9099257671998164e-05,
+      "loss": 0.8074,
+      "step": 487470
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 1.9098109742098414e-05,
+      "loss": 0.8438,
+      "step": 487480
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 1.9096961812198663e-05,
+      "loss": 0.7982,
+      "step": 487490
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 1.909581388229892e-05,
+      "loss": 0.6794,
+      "step": 487500
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 1.909466595239917e-05,
+      "loss": 0.7472,
+      "step": 487510
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 1.9093518022499423e-05,
+      "loss": 0.758,
+      "step": 487520
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 1.9092370092599676e-05,
+      "loss": 0.8182,
+      "step": 487530
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 1.909122216269993e-05,
+      "loss": 0.7519,
+      "step": 487540
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 1.909007423280018e-05,
+      "loss": 0.7441,
+      "step": 487550
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 1.9088926302900436e-05,
+      "loss": 0.7633,
+      "step": 487560
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 1.9087778373000686e-05,
+      "loss": 0.7919,
+      "step": 487570
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 1.908663044310094e-05,
+      "loss": 0.7452,
+      "step": 487580
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 1.9085482513201192e-05,
+      "loss": 0.7193,
+      "step": 487590
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 1.9084334583301446e-05,
+      "loss": 0.7372,
+      "step": 487600
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 1.9083186653401695e-05,
+      "loss": 0.7445,
+      "step": 487610
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 1.9082038723501952e-05,
+      "loss": 0.8321,
+      "step": 487620
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 1.9080890793602202e-05,
+      "loss": 0.7074,
+      "step": 487630
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 1.9079742863702452e-05,
+      "loss": 0.819,
+      "step": 487640
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 1.907859493380271e-05,
+      "loss": 0.7468,
+      "step": 487650
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 1.907744700390296e-05,
+      "loss": 0.7183,
+      "step": 487660
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 1.907629907400321e-05,
+      "loss": 0.7887,
+      "step": 487670
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 1.9075151144103465e-05,
+      "loss": 0.8273,
+      "step": 487680
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 1.9074003214203718e-05,
+      "loss": 0.7347,
+      "step": 487690
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 1.9072855284303968e-05,
+      "loss": 0.7522,
+      "step": 487700
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 1.9071707354404225e-05,
+      "loss": 0.8272,
+      "step": 487710
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 1.9070559424504474e-05,
+      "loss": 0.7372,
+      "step": 487720
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 1.9069411494604728e-05,
+      "loss": 0.7729,
+      "step": 487730
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 1.906826356470498e-05,
+      "loss": 0.789,
+      "step": 487740
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 1.9067115634805234e-05,
+      "loss": 0.814,
+      "step": 487750
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 1.9065967704905484e-05,
+      "loss": 0.7875,
+      "step": 487760
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 1.906481977500574e-05,
+      "loss": 0.7382,
+      "step": 487770
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 1.906367184510599e-05,
+      "loss": 0.7524,
+      "step": 487780
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 1.906252391520624e-05,
+      "loss": 0.8218,
+      "step": 487790
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 1.9061375985306497e-05,
+      "loss": 0.7155,
+      "step": 487800
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 1.9060228055406747e-05,
+      "loss": 0.8007,
+      "step": 487810
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 1.9059080125507e-05,
+      "loss": 0.7657,
+      "step": 487820
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 1.9057932195607253e-05,
+      "loss": 0.6875,
+      "step": 487830
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 1.9056784265707507e-05,
+      "loss": 0.7733,
+      "step": 487840
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 1.9055636335807756e-05,
+      "loss": 0.7493,
+      "step": 487850
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 1.9054488405908013e-05,
+      "loss": 0.7923,
+      "step": 487860
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 1.9053340476008263e-05,
+      "loss": 0.7388,
+      "step": 487870
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 1.9052192546108516e-05,
+      "loss": 0.7186,
+      "step": 487880
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 1.905104461620877e-05,
+      "loss": 0.7944,
+      "step": 487890
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 1.9049896686309023e-05,
+      "loss": 0.7453,
+      "step": 487900
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 1.9048748756409272e-05,
+      "loss": 0.732,
+      "step": 487910
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 1.904760082650953e-05,
+      "loss": 0.7203,
+      "step": 487920
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 1.904645289660978e-05,
+      "loss": 0.774,
+      "step": 487930
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 1.904530496671003e-05,
+      "loss": 0.7524,
+      "step": 487940
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 1.9044157036810285e-05,
+      "loss": 0.8635,
+      "step": 487950
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 1.9043009106910535e-05,
+      "loss": 0.6831,
+      "step": 487960
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 1.904186117701079e-05,
+      "loss": 0.7113,
+      "step": 487970
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 1.9040713247111042e-05,
+      "loss": 0.8033,
+      "step": 487980
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 1.9039565317211295e-05,
+      "loss": 0.797,
+      "step": 487990
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 1.9038417387311545e-05,
+      "loss": 0.7424,
+      "step": 488000
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 1.90372694574118e-05,
+      "loss": 0.7878,
+      "step": 488010
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 1.903612152751205e-05,
+      "loss": 0.8056,
+      "step": 488020
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 1.9034973597612305e-05,
+      "loss": 0.794,
+      "step": 488030
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 1.9033825667712558e-05,
+      "loss": 0.8087,
+      "step": 488040
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 1.903267773781281e-05,
+      "loss": 0.8105,
+      "step": 488050
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 1.903152980791306e-05,
+      "loss": 0.7687,
+      "step": 488060
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 1.9030381878013318e-05,
+      "loss": 0.7539,
+      "step": 488070
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 1.9029233948113567e-05,
+      "loss": 0.7185,
+      "step": 488080
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 1.9028086018213817e-05,
+      "loss": 0.7187,
+      "step": 488090
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 1.9026938088314074e-05,
+      "loss": 0.7628,
+      "step": 488100
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 1.9025790158414324e-05,
+      "loss": 0.7604,
+      "step": 488110
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 1.9024642228514577e-05,
+      "loss": 0.7761,
+      "step": 488120
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 1.9023494298614827e-05,
+      "loss": 0.7051,
+      "step": 488130
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 1.9022346368715083e-05,
+      "loss": 0.7621,
+      "step": 488140
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 1.9021198438815333e-05,
+      "loss": 0.7967,
+      "step": 488150
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 1.9020050508915587e-05,
+      "loss": 0.7271,
+      "step": 488160
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 1.901890257901584e-05,
+      "loss": 0.6971,
+      "step": 488170
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 1.9017754649116093e-05,
+      "loss": 0.8048,
+      "step": 488180
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 1.9016606719216343e-05,
+      "loss": 0.7864,
+      "step": 488190
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 1.90154587893166e-05,
+      "loss": 0.7362,
+      "step": 488200
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 1.901431085941685e-05,
+      "loss": 0.6839,
+      "step": 488210
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 1.90131629295171e-05,
+      "loss": 0.7162,
+      "step": 488220
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 1.9012014999617356e-05,
+      "loss": 0.7815,
+      "step": 488230
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 1.9010867069717606e-05,
+      "loss": 0.7739,
+      "step": 488240
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 1.900971913981786e-05,
+      "loss": 0.7879,
+      "step": 488250
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 1.9008571209918112e-05,
+      "loss": 0.7342,
+      "step": 488260
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 1.9007423280018365e-05,
+      "loss": 0.8094,
+      "step": 488270
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 1.9006275350118615e-05,
+      "loss": 0.7142,
+      "step": 488280
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 1.9005127420218872e-05,
+      "loss": 0.6705,
+      "step": 488290
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 1.9003979490319122e-05,
+      "loss": 0.722,
+      "step": 488300
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 1.9002831560419375e-05,
+      "loss": 0.7627,
+      "step": 488310
+    },
+    {
+      "epoch": 22.4,
+      "learning_rate": 1.9001683630519628e-05,
+      "loss": 0.7687,
+      "step": 488320
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 1.900053570061988e-05,
+      "loss": 0.732,
+      "step": 488330
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 1.899938777072013e-05,
+      "loss": 0.7484,
+      "step": 488340
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 1.8998239840820388e-05,
+      "loss": 0.7987,
+      "step": 488350
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 1.8997091910920638e-05,
+      "loss": 0.7019,
+      "step": 488360
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 1.8995943981020888e-05,
+      "loss": 0.6685,
+      "step": 488370
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 1.8994796051121144e-05,
+      "loss": 0.7643,
+      "step": 488380
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 1.8993648121221394e-05,
+      "loss": 0.72,
+      "step": 488390
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 1.8992500191321647e-05,
+      "loss": 0.8462,
+      "step": 488400
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 1.89913522614219e-05,
+      "loss": 0.6659,
+      "step": 488410
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 1.8990204331522154e-05,
+      "loss": 0.855,
+      "step": 488420
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 1.8989056401622404e-05,
+      "loss": 0.7496,
+      "step": 488430
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 1.898790847172266e-05,
+      "loss": 0.7637,
+      "step": 488440
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 1.898676054182291e-05,
+      "loss": 0.8065,
+      "step": 488450
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 1.8985612611923163e-05,
+      "loss": 0.7129,
+      "step": 488460
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 1.8984464682023417e-05,
+      "loss": 0.7353,
+      "step": 488470
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 1.898331675212367e-05,
+      "loss": 0.8209,
+      "step": 488480
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 1.898216882222392e-05,
+      "loss": 0.7631,
+      "step": 488490
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 1.8981020892324176e-05,
+      "loss": 0.8301,
+      "step": 488500
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 1.8979872962424426e-05,
+      "loss": 0.7551,
+      "step": 488510
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 1.8978725032524676e-05,
+      "loss": 0.7364,
+      "step": 488520
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 1.8977577102624933e-05,
+      "loss": 0.6929,
+      "step": 488530
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 1.8976429172725183e-05,
+      "loss": 0.7518,
+      "step": 488540
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 1.8975281242825436e-05,
+      "loss": 0.8488,
+      "step": 488550
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 1.897413331292569e-05,
+      "loss": 0.7408,
+      "step": 488560
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 1.8972985383025942e-05,
+      "loss": 0.8315,
+      "step": 488570
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 1.8971837453126192e-05,
+      "loss": 0.7735,
+      "step": 488580
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 1.897068952322645e-05,
+      "loss": 0.8273,
+      "step": 488590
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 1.89695415933267e-05,
+      "loss": 0.8292,
+      "step": 488600
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 1.8968393663426952e-05,
+      "loss": 0.8212,
+      "step": 488610
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 1.8967245733527205e-05,
+      "loss": 0.8223,
+      "step": 488620
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 1.896609780362746e-05,
+      "loss": 0.8143,
+      "step": 488630
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 1.896494987372771e-05,
+      "loss": 0.7989,
+      "step": 488640
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 1.8963801943827965e-05,
+      "loss": 0.7372,
+      "step": 488650
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 1.8962654013928215e-05,
+      "loss": 0.816,
+      "step": 488660
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 1.8961506084028465e-05,
+      "loss": 0.7703,
+      "step": 488670
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 1.896035815412872e-05,
+      "loss": 0.7555,
+      "step": 488680
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 1.895921022422897e-05,
+      "loss": 0.7531,
+      "step": 488690
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 1.8958062294329224e-05,
+      "loss": 0.7233,
+      "step": 488700
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 1.8956914364429478e-05,
+      "loss": 0.7612,
+      "step": 488710
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 1.895576643452973e-05,
+      "loss": 0.7287,
+      "step": 488720
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 1.895461850462998e-05,
+      "loss": 0.7395,
+      "step": 488730
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 1.8953470574730237e-05,
+      "loss": 0.7785,
+      "step": 488740
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 1.8952322644830487e-05,
+      "loss": 0.7763,
+      "step": 488750
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 1.895117471493074e-05,
+      "loss": 0.7542,
+      "step": 488760
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 1.8950026785030994e-05,
+      "loss": 0.724,
+      "step": 488770
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 1.8948878855131247e-05,
+      "loss": 0.7832,
+      "step": 488780
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 1.8947730925231497e-05,
+      "loss": 0.7937,
+      "step": 488790
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 1.8946582995331747e-05,
+      "loss": 0.7428,
+      "step": 488800
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 1.8945435065432003e-05,
+      "loss": 0.786,
+      "step": 488810
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 1.8944287135532253e-05,
+      "loss": 0.7157,
+      "step": 488820
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 1.8943139205632506e-05,
+      "loss": 0.7091,
+      "step": 488830
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 1.894199127573276e-05,
+      "loss": 0.8611,
+      "step": 488840
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 1.8940843345833013e-05,
+      "loss": 0.6648,
+      "step": 488850
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 1.8939695415933263e-05,
+      "loss": 0.7751,
+      "step": 488860
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 1.893854748603352e-05,
+      "loss": 0.7288,
+      "step": 488870
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 1.893739955613377e-05,
+      "loss": 0.8371,
+      "step": 488880
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 1.8936251626234022e-05,
+      "loss": 0.8185,
+      "step": 488890
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 1.8935103696334276e-05,
+      "loss": 0.7877,
+      "step": 488900
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 1.893395576643453e-05,
+      "loss": 0.7862,
+      "step": 488910
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 1.893280783653478e-05,
+      "loss": 0.8231,
+      "step": 488920
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 1.8931659906635035e-05,
+      "loss": 0.8111,
+      "step": 488930
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 1.8930511976735285e-05,
+      "loss": 0.7656,
+      "step": 488940
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 1.8929364046835535e-05,
+      "loss": 0.8206,
+      "step": 488950
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 1.8928216116935792e-05,
+      "loss": 0.6705,
+      "step": 488960
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 1.892706818703604e-05,
+      "loss": 0.7888,
+      "step": 488970
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 1.8925920257136295e-05,
+      "loss": 0.6475,
+      "step": 488980
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 1.8924772327236548e-05,
+      "loss": 0.7885,
+      "step": 488990
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 1.89236243973368e-05,
+      "loss": 0.7724,
+      "step": 489000
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 1.892247646743705e-05,
+      "loss": 0.7971,
+      "step": 489010
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 1.8921328537537308e-05,
+      "loss": 0.7647,
+      "step": 489020
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 1.8920180607637558e-05,
+      "loss": 0.8304,
+      "step": 489030
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 1.891903267773781e-05,
+      "loss": 0.8207,
+      "step": 489040
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 1.8917884747838064e-05,
+      "loss": 0.6888,
+      "step": 489050
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 1.8916736817938317e-05,
+      "loss": 0.7958,
+      "step": 489060
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 1.8915588888038567e-05,
+      "loss": 0.7705,
+      "step": 489070
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 1.8914440958138824e-05,
+      "loss": 0.7954,
+      "step": 489080
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 1.8913293028239074e-05,
+      "loss": 0.8721,
+      "step": 489090
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 1.8912145098339324e-05,
+      "loss": 0.7533,
+      "step": 489100
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 1.891099716843958e-05,
+      "loss": 0.797,
+      "step": 489110
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 1.890984923853983e-05,
+      "loss": 0.7211,
+      "step": 489120
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 1.8908701308640083e-05,
+      "loss": 0.6773,
+      "step": 489130
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 1.8907553378740337e-05,
+      "loss": 0.809,
+      "step": 489140
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 1.890640544884059e-05,
+      "loss": 0.7573,
+      "step": 489150
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 1.890525751894084e-05,
+      "loss": 0.7304,
+      "step": 489160
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 1.8904109589041096e-05,
+      "loss": 0.7826,
+      "step": 489170
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 1.8902961659141346e-05,
+      "loss": 0.8305,
+      "step": 489180
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 1.89018137292416e-05,
+      "loss": 0.8316,
+      "step": 489190
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 1.8900665799341853e-05,
+      "loss": 0.809,
+      "step": 489200
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 1.8899517869442106e-05,
+      "loss": 0.7928,
+      "step": 489210
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 1.8898369939542356e-05,
+      "loss": 0.7778,
+      "step": 489220
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 1.8897222009642612e-05,
+      "loss": 0.7617,
+      "step": 489230
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 1.8896074079742862e-05,
+      "loss": 0.7383,
+      "step": 489240
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 1.8894926149843112e-05,
+      "loss": 0.7139,
+      "step": 489250
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 1.889377821994337e-05,
+      "loss": 0.7502,
+      "step": 489260
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 1.889263029004362e-05,
+      "loss": 0.7441,
+      "step": 489270
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 1.8891482360143872e-05,
+      "loss": 0.7304,
+      "step": 489280
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 1.8890334430244125e-05,
+      "loss": 0.7275,
+      "step": 489290
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 1.8889186500344378e-05,
+      "loss": 0.8002,
+      "step": 489300
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 1.8888038570444628e-05,
+      "loss": 0.7445,
+      "step": 489310
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 1.8886890640544885e-05,
+      "loss": 0.761,
+      "step": 489320
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 1.8885742710645135e-05,
+      "loss": 0.7576,
+      "step": 489330
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 1.8884594780745388e-05,
+      "loss": 0.6935,
+      "step": 489340
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 1.888344685084564e-05,
+      "loss": 0.7656,
+      "step": 489350
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 1.8882298920945894e-05,
+      "loss": 0.8002,
+      "step": 489360
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 1.8881150991046144e-05,
+      "loss": 0.8111,
+      "step": 489370
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 1.88800030611464e-05,
+      "loss": 0.8108,
+      "step": 489380
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 1.887885513124665e-05,
+      "loss": 0.7549,
+      "step": 489390
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 1.88777072013469e-05,
+      "loss": 0.788,
+      "step": 489400
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 1.8876559271447157e-05,
+      "loss": 0.7691,
+      "step": 489410
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 1.8875411341547407e-05,
+      "loss": 0.7356,
+      "step": 489420
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 1.887426341164766e-05,
+      "loss": 0.7232,
+      "step": 489430
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 1.8873115481747913e-05,
+      "loss": 0.8182,
+      "step": 489440
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 1.8871967551848167e-05,
+      "loss": 0.7721,
+      "step": 489450
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 1.8870819621948417e-05,
+      "loss": 0.694,
+      "step": 489460
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 1.8869671692048673e-05,
+      "loss": 0.6911,
+      "step": 489470
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 1.8868523762148923e-05,
+      "loss": 0.7909,
+      "step": 489480
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 1.8867375832249176e-05,
+      "loss": 0.8092,
+      "step": 489490
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 1.8866227902349426e-05,
+      "loss": 0.7395,
+      "step": 489500
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 1.8865079972449683e-05,
+      "loss": 0.7216,
+      "step": 489510
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 1.8863932042549933e-05,
+      "loss": 0.8274,
+      "step": 489520
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 1.8862784112650183e-05,
+      "loss": 0.8042,
+      "step": 489530
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 1.886163618275044e-05,
+      "loss": 0.6738,
+      "step": 489540
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 1.886048825285069e-05,
+      "loss": 0.7404,
+      "step": 489550
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 1.8859340322950942e-05,
+      "loss": 0.7311,
+      "step": 489560
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 1.8858192393051195e-05,
+      "loss": 0.7408,
+      "step": 489570
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 1.885704446315145e-05,
+      "loss": 0.8777,
+      "step": 489580
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 1.88558965332517e-05,
+      "loss": 0.701,
+      "step": 489590
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 1.8854748603351955e-05,
+      "loss": 0.7432,
+      "step": 489600
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 1.8853600673452205e-05,
+      "loss": 0.8155,
+      "step": 489610
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 1.8852452743552458e-05,
+      "loss": 0.8179,
+      "step": 489620
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 1.885130481365271e-05,
+      "loss": 0.7503,
+      "step": 489630
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 1.8850156883752965e-05,
+      "loss": 0.7791,
+      "step": 489640
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 1.8849008953853215e-05,
+      "loss": 0.78,
+      "step": 489650
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 1.884786102395347e-05,
+      "loss": 0.7992,
+      "step": 489660
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 1.884671309405372e-05,
+      "loss": 0.7902,
+      "step": 489670
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 1.884556516415397e-05,
+      "loss": 0.7401,
+      "step": 489680
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 1.8844417234254228e-05,
+      "loss": 0.7441,
+      "step": 489690
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 1.8843269304354477e-05,
+      "loss": 0.7817,
+      "step": 489700
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 1.884212137445473e-05,
+      "loss": 0.8462,
+      "step": 489710
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 1.8840973444554984e-05,
+      "loss": 0.7893,
+      "step": 489720
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 1.8839825514655237e-05,
+      "loss": 0.787,
+      "step": 489730
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 1.8838677584755487e-05,
+      "loss": 0.7845,
+      "step": 489740
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 1.8837529654855744e-05,
+      "loss": 0.7133,
+      "step": 489750
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 1.8836381724955994e-05,
+      "loss": 0.7658,
+      "step": 489760
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 1.8835233795056247e-05,
+      "loss": 0.8095,
+      "step": 489770
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 1.88340858651565e-05,
+      "loss": 0.761,
+      "step": 489780
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 1.8832937935256753e-05,
+      "loss": 0.7318,
+      "step": 489790
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 1.8831790005357003e-05,
+      "loss": 0.8003,
+      "step": 489800
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 1.883064207545726e-05,
+      "loss": 0.7115,
+      "step": 489810
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 1.882949414555751e-05,
+      "loss": 0.7481,
+      "step": 489820
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 1.882834621565776e-05,
+      "loss": 0.7631,
+      "step": 489830
+    },
+    {
+      "epoch": 22.47,
+      "learning_rate": 1.8827198285758016e-05,
+      "loss": 0.7694,
+      "step": 489840
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 1.8826050355858266e-05,
+      "loss": 0.738,
+      "step": 489850
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 1.882490242595852e-05,
+      "loss": 0.8205,
+      "step": 489860
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 1.8823754496058772e-05,
+      "loss": 0.7877,
+      "step": 489870
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 1.8822606566159026e-05,
+      "loss": 0.8088,
+      "step": 489880
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 1.8821458636259276e-05,
+      "loss": 0.7154,
+      "step": 489890
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 1.8820310706359532e-05,
+      "loss": 0.7881,
+      "step": 489900
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 1.8819162776459782e-05,
+      "loss": 0.781,
+      "step": 489910
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 1.8818014846560035e-05,
+      "loss": 0.7304,
+      "step": 489920
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 1.881686691666029e-05,
+      "loss": 0.76,
+      "step": 489930
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 1.8815718986760542e-05,
+      "loss": 0.8606,
+      "step": 489940
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 1.881457105686079e-05,
+      "loss": 0.8049,
+      "step": 489950
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 1.8813423126961048e-05,
+      "loss": 0.751,
+      "step": 489960
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 1.8812275197061298e-05,
+      "loss": 0.737,
+      "step": 489970
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 1.8811127267161548e-05,
+      "loss": 0.7519,
+      "step": 489980
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 1.8809979337261805e-05,
+      "loss": 0.7986,
+      "step": 489990
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 1.8808831407362054e-05,
+      "loss": 0.7426,
+      "step": 490000
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 1.8807683477462308e-05,
+      "loss": 0.7509,
+      "step": 490010
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 1.880653554756256e-05,
+      "loss": 0.7933,
+      "step": 490020
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 1.8805387617662814e-05,
+      "loss": 0.6885,
+      "step": 490030
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 1.8804239687763064e-05,
+      "loss": 0.7302,
+      "step": 490040
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 1.880309175786332e-05,
+      "loss": 0.7208,
+      "step": 490050
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 1.880194382796357e-05,
+      "loss": 0.7193,
+      "step": 490060
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 1.8800795898063824e-05,
+      "loss": 0.6104,
+      "step": 490070
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 1.8799647968164077e-05,
+      "loss": 0.699,
+      "step": 490080
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 1.879850003826433e-05,
+      "loss": 0.8112,
+      "step": 490090
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 1.879735210836458e-05,
+      "loss": 0.8141,
+      "step": 490100
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 1.8796204178464837e-05,
+      "loss": 0.7707,
+      "step": 490110
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 1.8795056248565087e-05,
+      "loss": 0.8086,
+      "step": 490120
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 1.8793908318665336e-05,
+      "loss": 0.6846,
+      "step": 490130
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 1.8792760388765593e-05,
+      "loss": 0.8005,
+      "step": 490140
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 1.8791612458865843e-05,
+      "loss": 0.7991,
+      "step": 490150
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 1.8790464528966096e-05,
+      "loss": 0.7635,
+      "step": 490160
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 1.8789316599066346e-05,
+      "loss": 0.8721,
+      "step": 490170
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 1.8788168669166603e-05,
+      "loss": 0.9139,
+      "step": 490180
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 1.8787020739266852e-05,
+      "loss": 0.8145,
+      "step": 490190
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 1.8785872809367106e-05,
+      "loss": 0.744,
+      "step": 490200
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 1.878472487946736e-05,
+      "loss": 0.7596,
+      "step": 490210
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 1.8783576949567612e-05,
+      "loss": 0.7377,
+      "step": 490220
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 1.8782429019667862e-05,
+      "loss": 0.8395,
+      "step": 490230
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 1.878128108976812e-05,
+      "loss": 0.8291,
+      "step": 490240
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 1.878013315986837e-05,
+      "loss": 0.8631,
+      "step": 490250
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 1.877898522996862e-05,
+      "loss": 0.7993,
+      "step": 490260
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 1.8777837300068875e-05,
+      "loss": 0.7054,
+      "step": 490270
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 1.8776689370169125e-05,
+      "loss": 0.7839,
+      "step": 490280
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 1.8775541440269378e-05,
+      "loss": 0.7192,
+      "step": 490290
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 1.877439351036963e-05,
+      "loss": 0.7979,
+      "step": 490300
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 1.8773245580469885e-05,
+      "loss": 0.6448,
+      "step": 490310
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 1.8772097650570134e-05,
+      "loss": 0.7978,
+      "step": 490320
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 1.877094972067039e-05,
+      "loss": 0.8202,
+      "step": 490330
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 1.876980179077064e-05,
+      "loss": 0.7351,
+      "step": 490340
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 1.8768653860870894e-05,
+      "loss": 0.7375,
+      "step": 490350
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 1.8767505930971147e-05,
+      "loss": 0.6957,
+      "step": 490360
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 1.87663580010714e-05,
+      "loss": 0.7238,
+      "step": 490370
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 1.876521007117165e-05,
+      "loss": 0.7692,
+      "step": 490380
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 1.8764062141271907e-05,
+      "loss": 0.7482,
+      "step": 490390
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 1.8762914211372157e-05,
+      "loss": 0.7814,
+      "step": 490400
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 1.8761766281472407e-05,
+      "loss": 0.7826,
+      "step": 490410
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 1.8760618351572663e-05,
+      "loss": 0.7881,
+      "step": 490420
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 1.8759470421672913e-05,
+      "loss": 0.8147,
+      "step": 490430
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 1.8758322491773167e-05,
+      "loss": 0.7103,
+      "step": 490440
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 1.875717456187342e-05,
+      "loss": 0.7409,
+      "step": 490450
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 1.8756026631973673e-05,
+      "loss": 0.7551,
+      "step": 490460
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 1.8754878702073923e-05,
+      "loss": 0.7852,
+      "step": 490470
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 1.875373077217418e-05,
+      "loss": 0.7969,
+      "step": 490480
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 1.875258284227443e-05,
+      "loss": 0.8059,
+      "step": 490490
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 1.8751434912374683e-05,
+      "loss": 0.791,
+      "step": 490500
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 1.8750286982474936e-05,
+      "loss": 0.7118,
+      "step": 490510
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 1.874913905257519e-05,
+      "loss": 0.8009,
+      "step": 490520
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 1.8747991122675442e-05,
+      "loss": 0.8217,
+      "step": 490530
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 1.8746843192775692e-05,
+      "loss": 0.703,
+      "step": 490540
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 1.8745695262875945e-05,
+      "loss": 0.7068,
+      "step": 490550
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 1.87445473329762e-05,
+      "loss": 0.7639,
+      "step": 490560
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 1.874339940307645e-05,
+      "loss": 0.715,
+      "step": 490570
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 1.8742251473176702e-05,
+      "loss": 0.7375,
+      "step": 490580
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 1.8741103543276955e-05,
+      "loss": 0.813,
+      "step": 490590
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 1.8739955613377208e-05,
+      "loss": 0.7958,
+      "step": 490600
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 1.873880768347746e-05,
+      "loss": 0.7946,
+      "step": 490610
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 1.8737659753577715e-05,
+      "loss": 0.7463,
+      "step": 490620
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 1.8736511823677965e-05,
+      "loss": 0.6859,
+      "step": 490630
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 1.8735363893778218e-05,
+      "loss": 0.7675,
+      "step": 490640
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 1.873421596387847e-05,
+      "loss": 0.7842,
+      "step": 490650
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 1.8733068033978724e-05,
+      "loss": 0.7001,
+      "step": 490660
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 1.8731920104078978e-05,
+      "loss": 0.7421,
+      "step": 490670
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 1.873077217417923e-05,
+      "loss": 0.7334,
+      "step": 490680
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 1.872962424427948e-05,
+      "loss": 0.7988,
+      "step": 490690
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 1.8728476314379734e-05,
+      "loss": 0.6939,
+      "step": 490700
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 1.8727328384479984e-05,
+      "loss": 0.7913,
+      "step": 490710
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 1.8726180454580237e-05,
+      "loss": 0.8154,
+      "step": 490720
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 1.872503252468049e-05,
+      "loss": 0.8013,
+      "step": 490730
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 1.8723884594780744e-05,
+      "loss": 0.7725,
+      "step": 490740
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 1.8722736664880997e-05,
+      "loss": 0.7873,
+      "step": 490750
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 1.872158873498125e-05,
+      "loss": 0.8068,
+      "step": 490760
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 1.87204408050815e-05,
+      "loss": 0.8054,
+      "step": 490770
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 1.8719292875181753e-05,
+      "loss": 0.8246,
+      "step": 490780
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 1.8718144945282006e-05,
+      "loss": 0.8163,
+      "step": 490790
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 1.871699701538226e-05,
+      "loss": 0.6543,
+      "step": 490800
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 1.8715849085482513e-05,
+      "loss": 0.7405,
+      "step": 490810
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 1.8714701155582766e-05,
+      "loss": 0.8745,
+      "step": 490820
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 1.8713553225683016e-05,
+      "loss": 0.8356,
+      "step": 490830
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 1.871240529578327e-05,
+      "loss": 0.6992,
+      "step": 490840
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 1.8711257365883522e-05,
+      "loss": 0.7488,
+      "step": 490850
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 1.8710109435983772e-05,
+      "loss": 0.7467,
+      "step": 490860
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 1.8708961506084026e-05,
+      "loss": 0.7514,
+      "step": 490870
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 1.870781357618428e-05,
+      "loss": 0.7105,
+      "step": 490880
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 1.8706665646284532e-05,
+      "loss": 0.825,
+      "step": 490890
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 1.8705517716384785e-05,
+      "loss": 0.8085,
+      "step": 490900
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 1.870436978648504e-05,
+      "loss": 0.7942,
+      "step": 490910
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 1.870322185658529e-05,
+      "loss": 0.7615,
+      "step": 490920
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 1.870207392668554e-05,
+      "loss": 0.7965,
+      "step": 490930
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 1.8700925996785795e-05,
+      "loss": 0.7608,
+      "step": 490940
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 1.8699778066886048e-05,
+      "loss": 0.8136,
+      "step": 490950
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 1.86986301369863e-05,
+      "loss": 0.6811,
+      "step": 490960
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 1.8697482207086555e-05,
+      "loss": 0.7307,
+      "step": 490970
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 1.8696334277186804e-05,
+      "loss": 0.7816,
+      "step": 490980
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 1.8695186347287058e-05,
+      "loss": 0.7064,
+      "step": 490990
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 1.869403841738731e-05,
+      "loss": 0.7665,
+      "step": 491000
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 1.869289048748756e-05,
+      "loss": 0.79,
+      "step": 491010
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 1.8691742557587814e-05,
+      "loss": 0.7238,
+      "step": 491020
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 1.8690594627688067e-05,
+      "loss": 0.713,
+      "step": 491030
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 1.868944669778832e-05,
+      "loss": 0.6814,
+      "step": 491040
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 1.8688298767888574e-05,
+      "loss": 0.7819,
+      "step": 491050
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 1.8687150837988824e-05,
+      "loss": 0.7906,
+      "step": 491060
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 1.8686002908089077e-05,
+      "loss": 0.7653,
+      "step": 491070
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 1.868485497818933e-05,
+      "loss": 0.7615,
+      "step": 491080
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 1.8683707048289583e-05,
+      "loss": 0.7275,
+      "step": 491090
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 1.8682559118389837e-05,
+      "loss": 0.7499,
+      "step": 491100
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 1.868141118849009e-05,
+      "loss": 0.9072,
+      "step": 491110
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 1.868026325859034e-05,
+      "loss": 0.7848,
+      "step": 491120
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 1.8679115328690593e-05,
+      "loss": 0.7904,
+      "step": 491130
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 1.8677967398790846e-05,
+      "loss": 0.7382,
+      "step": 491140
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 1.8676819468891096e-05,
+      "loss": 0.7594,
+      "step": 491150
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 1.867567153899135e-05,
+      "loss": 0.8096,
+      "step": 491160
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 1.8674523609091602e-05,
+      "loss": 0.7815,
+      "step": 491170
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 1.8673375679191856e-05,
+      "loss": 0.8452,
+      "step": 491180
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 1.867222774929211e-05,
+      "loss": 0.799,
+      "step": 491190
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 1.8671079819392362e-05,
+      "loss": 0.7199,
+      "step": 491200
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 1.8669931889492612e-05,
+      "loss": 0.755,
+      "step": 491210
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 1.8668783959592865e-05,
+      "loss": 0.7546,
+      "step": 491220
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 1.866763602969312e-05,
+      "loss": 0.7296,
+      "step": 491230
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 1.8666488099793372e-05,
+      "loss": 0.8052,
+      "step": 491240
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 1.8665340169893625e-05,
+      "loss": 0.763,
+      "step": 491250
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 1.8664192239993878e-05,
+      "loss": 0.7477,
+      "step": 491260
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 1.8663044310094128e-05,
+      "loss": 0.8518,
+      "step": 491270
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 1.866189638019438e-05,
+      "loss": 0.7729,
+      "step": 491280
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 1.8660748450294635e-05,
+      "loss": 0.7744,
+      "step": 491290
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 1.8659600520394884e-05,
+      "loss": 0.7674,
+      "step": 491300
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 1.8658452590495138e-05,
+      "loss": 0.754,
+      "step": 491310
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 1.865730466059539e-05,
+      "loss": 0.8093,
+      "step": 491320
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 1.8656156730695644e-05,
+      "loss": 0.7795,
+      "step": 491330
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 1.8655008800795897e-05,
+      "loss": 0.7648,
+      "step": 491340
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 1.865386087089615e-05,
+      "loss": 0.7688,
+      "step": 491350
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 1.86527129409964e-05,
+      "loss": 0.757,
+      "step": 491360
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 1.8651565011096654e-05,
+      "loss": 0.7369,
+      "step": 491370
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 1.8650417081196907e-05,
+      "loss": 0.8016,
+      "step": 491380
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 1.864926915129716e-05,
+      "loss": 0.6848,
+      "step": 491390
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 1.8648121221397413e-05,
+      "loss": 0.7973,
+      "step": 491400
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 1.8646973291497663e-05,
+      "loss": 0.7958,
+      "step": 491410
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 1.8645825361597917e-05,
+      "loss": 0.735,
+      "step": 491420
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 1.864467743169817e-05,
+      "loss": 0.824,
+      "step": 491430
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 1.864352950179842e-05,
+      "loss": 0.8535,
+      "step": 491440
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 1.8642381571898673e-05,
+      "loss": 0.7313,
+      "step": 491450
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 1.8641233641998926e-05,
+      "loss": 0.7475,
+      "step": 491460
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 1.864008571209918e-05,
+      "loss": 0.7476,
+      "step": 491470
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 1.8638937782199433e-05,
+      "loss": 0.776,
+      "step": 491480
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 1.8637789852299686e-05,
+      "loss": 0.8202,
+      "step": 491490
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 1.8636641922399936e-05,
+      "loss": 0.8174,
+      "step": 491500
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 1.863549399250019e-05,
+      "loss": 0.7412,
+      "step": 491510
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 1.8634346062600442e-05,
+      "loss": 0.7618,
+      "step": 491520
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 1.8633198132700695e-05,
+      "loss": 0.7915,
+      "step": 491530
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 1.863205020280095e-05,
+      "loss": 0.7875,
+      "step": 491540
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 1.8630902272901202e-05,
+      "loss": 0.7792,
+      "step": 491550
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 1.8629754343001452e-05,
+      "loss": 0.6966,
+      "step": 491560
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 1.8628606413101705e-05,
+      "loss": 0.765,
+      "step": 491570
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 1.8627458483201958e-05,
+      "loss": 0.8226,
+      "step": 491580
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 1.8626310553302208e-05,
+      "loss": 0.8033,
+      "step": 491590
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 1.862516262340246e-05,
+      "loss": 0.7396,
+      "step": 491600
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 1.8624014693502715e-05,
+      "loss": 0.6733,
+      "step": 491610
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 1.8622866763602968e-05,
+      "loss": 0.7457,
+      "step": 491620
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 1.862171883370322e-05,
+      "loss": 0.7716,
+      "step": 491630
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 1.8620570903803474e-05,
+      "loss": 0.8299,
+      "step": 491640
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 1.8619422973903724e-05,
+      "loss": 0.6863,
+      "step": 491650
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 1.8618275044003977e-05,
+      "loss": 0.8614,
+      "step": 491660
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 1.861712711410423e-05,
+      "loss": 0.7818,
+      "step": 491670
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 1.8615979184204484e-05,
+      "loss": 0.8404,
+      "step": 491680
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 1.8614831254304737e-05,
+      "loss": 0.7892,
+      "step": 491690
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 1.861368332440499e-05,
+      "loss": 0.7708,
+      "step": 491700
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 1.861253539450524e-05,
+      "loss": 0.7107,
+      "step": 491710
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 1.8611387464605494e-05,
+      "loss": 0.8049,
+      "step": 491720
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 1.8610239534705743e-05,
+      "loss": 0.7931,
+      "step": 491730
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 1.8609091604805997e-05,
+      "loss": 0.7486,
+      "step": 491740
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 1.860794367490625e-05,
+      "loss": 0.6929,
+      "step": 491750
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 1.8606795745006503e-05,
+      "loss": 0.7949,
+      "step": 491760
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 1.8605647815106756e-05,
+      "loss": 0.7616,
+      "step": 491770
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 1.860449988520701e-05,
+      "loss": 0.7964,
+      "step": 491780
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 1.860335195530726e-05,
+      "loss": 0.7505,
+      "step": 491790
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 1.8602204025407513e-05,
+      "loss": 0.7259,
+      "step": 491800
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 1.8601056095507766e-05,
+      "loss": 0.7691,
+      "step": 491810
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 1.859990816560802e-05,
+      "loss": 0.7491,
+      "step": 491820
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 1.8598760235708272e-05,
+      "loss": 0.7435,
+      "step": 491830
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 1.8597612305808526e-05,
+      "loss": 0.7635,
+      "step": 491840
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 1.8596464375908776e-05,
+      "loss": 0.8292,
+      "step": 491850
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 1.859531644600903e-05,
+      "loss": 0.7149,
+      "step": 491860
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 1.8594168516109282e-05,
+      "loss": 0.702,
+      "step": 491870
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 1.8593020586209532e-05,
+      "loss": 0.7888,
+      "step": 491880
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 1.8591872656309785e-05,
+      "loss": 0.7995,
+      "step": 491890
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 1.859072472641004e-05,
+      "loss": 0.7161,
+      "step": 491900
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 1.858957679651029e-05,
+      "loss": 0.7544,
+      "step": 491910
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 1.8588428866610545e-05,
+      "loss": 0.7343,
+      "step": 491920
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 1.8587280936710798e-05,
+      "loss": 0.6518,
+      "step": 491930
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 1.8586133006811048e-05,
+      "loss": 0.7889,
+      "step": 491940
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 1.85849850769113e-05,
+      "loss": 0.8516,
+      "step": 491950
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 1.8583837147011554e-05,
+      "loss": 0.7774,
+      "step": 491960
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 1.8582689217111808e-05,
+      "loss": 0.733,
+      "step": 491970
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 1.858154128721206e-05,
+      "loss": 0.7896,
+      "step": 491980
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 1.8580393357312314e-05,
+      "loss": 0.7484,
+      "step": 491990
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 1.8579245427412564e-05,
+      "loss": 0.7478,
+      "step": 492000
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 1.8578097497512817e-05,
+      "loss": 0.7873,
+      "step": 492010
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 1.857694956761307e-05,
+      "loss": 0.7551,
+      "step": 492020
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 1.857580163771332e-05,
+      "loss": 0.7945,
+      "step": 492030
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 1.8574653707813574e-05,
+      "loss": 0.8141,
+      "step": 492040
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 1.8573505777913827e-05,
+      "loss": 0.7932,
+      "step": 492050
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 1.857235784801408e-05,
+      "loss": 0.7483,
+      "step": 492060
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 1.8571209918114333e-05,
+      "loss": 0.7682,
+      "step": 492070
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 1.8570061988214583e-05,
+      "loss": 0.7798,
+      "step": 492080
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 1.8568914058314836e-05,
+      "loss": 0.76,
+      "step": 492090
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 1.856776612841509e-05,
+      "loss": 0.7946,
+      "step": 492100
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 1.8566618198515343e-05,
+      "loss": 0.8267,
+      "step": 492110
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 1.8565470268615596e-05,
+      "loss": 0.7781,
+      "step": 492120
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 1.856432233871585e-05,
+      "loss": 0.8237,
+      "step": 492130
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 1.85631744088161e-05,
+      "loss": 0.7945,
+      "step": 492140
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 1.8562026478916352e-05,
+      "loss": 0.7713,
+      "step": 492150
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 1.8560878549016606e-05,
+      "loss": 0.8465,
+      "step": 492160
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 1.8559730619116856e-05,
+      "loss": 0.6732,
+      "step": 492170
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 1.855858268921711e-05,
+      "loss": 0.8085,
+      "step": 492180
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 1.8557434759317362e-05,
+      "loss": 0.7128,
+      "step": 492190
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 1.8556286829417615e-05,
+      "loss": 0.8284,
+      "step": 492200
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 1.855513889951787e-05,
+      "loss": 0.7922,
+      "step": 492210
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 1.8553990969618122e-05,
+      "loss": 0.7034,
+      "step": 492220
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 1.855284303971837e-05,
+      "loss": 0.7492,
+      "step": 492230
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 1.8551695109818625e-05,
+      "loss": 0.8749,
+      "step": 492240
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 1.8550547179918878e-05,
+      "loss": 0.8011,
+      "step": 492250
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 1.854939925001913e-05,
+      "loss": 0.7862,
+      "step": 492260
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 1.8548251320119385e-05,
+      "loss": 0.8761,
+      "step": 492270
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 1.8547103390219638e-05,
+      "loss": 0.7647,
+      "step": 492280
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 1.8545955460319888e-05,
+      "loss": 0.8219,
+      "step": 492290
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 1.854480753042014e-05,
+      "loss": 0.7784,
+      "step": 492300
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 1.8543659600520394e-05,
+      "loss": 0.7151,
+      "step": 492310
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 1.8542511670620644e-05,
+      "loss": 0.6984,
+      "step": 492320
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 1.8541363740720897e-05,
+      "loss": 0.7932,
+      "step": 492330
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 1.854021581082115e-05,
+      "loss": 0.756,
+      "step": 492340
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 1.8539067880921404e-05,
+      "loss": 0.736,
+      "step": 492350
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 1.8537919951021657e-05,
+      "loss": 0.8197,
+      "step": 492360
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 1.853677202112191e-05,
+      "loss": 0.6907,
+      "step": 492370
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 1.853562409122216e-05,
+      "loss": 0.723,
+      "step": 492380
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 1.8534476161322413e-05,
+      "loss": 0.7491,
+      "step": 492390
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 1.8533328231422667e-05,
+      "loss": 0.8367,
+      "step": 492400
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 1.853218030152292e-05,
+      "loss": 0.7456,
+      "step": 492410
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 1.8531032371623173e-05,
+      "loss": 0.7286,
+      "step": 492420
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 1.8529884441723423e-05,
+      "loss": 0.8215,
+      "step": 492430
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 1.8528736511823676e-05,
+      "loss": 0.7266,
+      "step": 492440
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 1.852758858192393e-05,
+      "loss": 0.7916,
+      "step": 492450
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 1.852644065202418e-05,
+      "loss": 0.8554,
+      "step": 492460
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 1.8525292722124432e-05,
+      "loss": 0.7107,
+      "step": 492470
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 1.8524144792224686e-05,
+      "loss": 0.8339,
+      "step": 492480
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 1.852299686232494e-05,
+      "loss": 0.7544,
+      "step": 492490
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 1.8521848932425192e-05,
+      "loss": 0.8489,
+      "step": 492500
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 1.8520701002525445e-05,
+      "loss": 0.6864,
+      "step": 492510
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 1.8519553072625695e-05,
+      "loss": 0.7463,
+      "step": 492520
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 1.851840514272595e-05,
+      "loss": 0.8084,
+      "step": 492530
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 1.8517257212826202e-05,
+      "loss": 0.7522,
+      "step": 492540
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 1.8516109282926455e-05,
+      "loss": 0.765,
+      "step": 492550
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 1.8514961353026708e-05,
+      "loss": 0.8341,
+      "step": 492560
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 1.851381342312696e-05,
+      "loss": 0.7622,
+      "step": 492570
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 1.851266549322721e-05,
+      "loss": 0.8002,
+      "step": 492580
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 1.8511517563327465e-05,
+      "loss": 0.7402,
+      "step": 492590
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 1.8510369633427718e-05,
+      "loss": 0.8015,
+      "step": 492600
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 1.8509221703527968e-05,
+      "loss": 0.7803,
+      "step": 492610
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 1.850807377362822e-05,
+      "loss": 0.7173,
+      "step": 492620
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 1.8506925843728474e-05,
+      "loss": 0.7925,
+      "step": 492630
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 1.8505777913828727e-05,
+      "loss": 0.6962,
+      "step": 492640
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 1.850462998392898e-05,
+      "loss": 0.7159,
+      "step": 492650
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 1.8503482054029234e-05,
+      "loss": 0.7774,
+      "step": 492660
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 1.8502334124129484e-05,
+      "loss": 0.8149,
+      "step": 492670
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 1.8501186194229737e-05,
+      "loss": 0.817,
+      "step": 492680
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 1.850003826432999e-05,
+      "loss": 0.7688,
+      "step": 492690
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 1.8498890334430244e-05,
+      "loss": 0.7912,
+      "step": 492700
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 1.8497742404530497e-05,
+      "loss": 0.7424,
+      "step": 492710
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 1.849659447463075e-05,
+      "loss": 0.811,
+      "step": 492720
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 1.8495446544731e-05,
+      "loss": 0.8331,
+      "step": 492730
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 1.8494298614831253e-05,
+      "loss": 0.8132,
+      "step": 492740
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 1.8493150684931503e-05,
+      "loss": 0.7913,
+      "step": 492750
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 1.8492002755031756e-05,
+      "loss": 0.7849,
+      "step": 492760
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 1.849085482513201e-05,
+      "loss": 0.7885,
+      "step": 492770
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 1.8489706895232263e-05,
+      "loss": 0.7918,
+      "step": 492780
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 1.8488558965332516e-05,
+      "loss": 0.8359,
+      "step": 492790
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 1.848741103543277e-05,
+      "loss": 0.7531,
+      "step": 492800
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 1.848626310553302e-05,
+      "loss": 0.7209,
+      "step": 492810
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 1.8485115175633272e-05,
+      "loss": 0.7556,
+      "step": 492820
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 1.8483967245733526e-05,
+      "loss": 0.7509,
+      "step": 492830
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 1.848281931583378e-05,
+      "loss": 0.79,
+      "step": 492840
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 1.8481671385934032e-05,
+      "loss": 0.7753,
+      "step": 492850
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 1.8480523456034285e-05,
+      "loss": 0.7376,
+      "step": 492860
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 1.8479375526134535e-05,
+      "loss": 0.7481,
+      "step": 492870
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 1.847822759623479e-05,
+      "loss": 0.7281,
+      "step": 492880
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 1.847707966633504e-05,
+      "loss": 0.724,
+      "step": 492890
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 1.847593173643529e-05,
+      "loss": 0.7019,
+      "step": 492900
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 1.8474783806535545e-05,
+      "loss": 0.8024,
+      "step": 492910
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 1.8473635876635798e-05,
+      "loss": 0.8212,
+      "step": 492920
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 1.847248794673605e-05,
+      "loss": 0.7827,
+      "step": 492930
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 1.8471340016836304e-05,
+      "loss": 0.7019,
+      "step": 492940
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 1.8470192086936558e-05,
+      "loss": 0.844,
+      "step": 492950
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 1.8469044157036807e-05,
+      "loss": 0.7245,
+      "step": 492960
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 1.846789622713706e-05,
+      "loss": 0.7968,
+      "step": 492970
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 1.8466748297237314e-05,
+      "loss": 0.7871,
+      "step": 492980
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 1.8465600367337567e-05,
+      "loss": 0.8302,
+      "step": 492990
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 1.846445243743782e-05,
+      "loss": 0.7728,
+      "step": 493000
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 1.8463304507538074e-05,
+      "loss": 0.776,
+      "step": 493010
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 1.8462156577638324e-05,
+      "loss": 0.7168,
+      "step": 493020
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 1.8461008647738577e-05,
+      "loss": 0.7346,
+      "step": 493030
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 1.845986071783883e-05,
+      "loss": 0.7191,
+      "step": 493040
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 1.845871278793908e-05,
+      "loss": 0.7993,
+      "step": 493050
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 1.8457564858039333e-05,
+      "loss": 0.7504,
+      "step": 493060
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 1.8456416928139586e-05,
+      "loss": 0.799,
+      "step": 493070
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 1.845526899823984e-05,
+      "loss": 0.822,
+      "step": 493080
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 1.8454121068340093e-05,
+      "loss": 0.7366,
+      "step": 493090
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 1.8452973138440343e-05,
+      "loss": 0.8286,
+      "step": 493100
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 1.8451825208540596e-05,
+      "loss": 0.6628,
+      "step": 493110
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 1.845067727864085e-05,
+      "loss": 0.7182,
+      "step": 493120
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 1.8449529348741102e-05,
+      "loss": 0.7888,
+      "step": 493130
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 1.8448381418841356e-05,
+      "loss": 0.7781,
+      "step": 493140
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 1.844723348894161e-05,
+      "loss": 0.732,
+      "step": 493150
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 1.844608555904186e-05,
+      "loss": 0.7998,
+      "step": 493160
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 1.8444937629142112e-05,
+      "loss": 0.8276,
+      "step": 493170
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 1.8443789699242365e-05,
+      "loss": 0.7247,
+      "step": 493180
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 1.8442641769342615e-05,
+      "loss": 0.7686,
+      "step": 493190
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 1.844149383944287e-05,
+      "loss": 0.7947,
+      "step": 493200
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 1.844034590954312e-05,
+      "loss": 0.813,
+      "step": 493210
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 1.8439197979643375e-05,
+      "loss": 0.7868,
+      "step": 493220
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 1.8438050049743628e-05,
+      "loss": 0.7771,
+      "step": 493230
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 1.843690211984388e-05,
+      "loss": 0.6982,
+      "step": 493240
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 1.843575418994413e-05,
+      "loss": 0.861,
+      "step": 493250
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 1.8434606260044384e-05,
+      "loss": 0.78,
+      "step": 493260
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 1.8433458330144638e-05,
+      "loss": 0.7932,
+      "step": 493270
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 1.843231040024489e-05,
+      "loss": 0.8075,
+      "step": 493280
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 1.8431162470345144e-05,
+      "loss": 0.8094,
+      "step": 493290
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 1.8430014540445397e-05,
+      "loss": 0.786,
+      "step": 493300
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 1.8428866610545647e-05,
+      "loss": 0.7551,
+      "step": 493310
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 1.84277186806459e-05,
+      "loss": 0.7599,
+      "step": 493320
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 1.8426570750746154e-05,
+      "loss": 0.8039,
+      "step": 493330
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 1.8425422820846404e-05,
+      "loss": 0.8206,
+      "step": 493340
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 1.8424274890946657e-05,
+      "loss": 0.818,
+      "step": 493350
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 1.842312696104691e-05,
+      "loss": 0.7659,
+      "step": 493360
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 1.8421979031147163e-05,
+      "loss": 0.7572,
+      "step": 493370
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 1.8420831101247417e-05,
+      "loss": 0.7948,
+      "step": 493380
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 1.841968317134767e-05,
+      "loss": 0.801,
+      "step": 493390
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 1.841853524144792e-05,
+      "loss": 0.7561,
+      "step": 493400
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 1.8417387311548173e-05,
+      "loss": 0.813,
+      "step": 493410
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 1.8416239381648426e-05,
+      "loss": 0.7544,
+      "step": 493420
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 1.841509145174868e-05,
+      "loss": 0.7285,
+      "step": 493430
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 1.8413943521848933e-05,
+      "loss": 0.7937,
+      "step": 493440
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 1.8412795591949182e-05,
+      "loss": 0.6597,
+      "step": 493450
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 1.8411647662049436e-05,
+      "loss": 0.7285,
+      "step": 493460
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 1.841049973214969e-05,
+      "loss": 0.783,
+      "step": 493470
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 1.840935180224994e-05,
+      "loss": 0.8506,
+      "step": 493480
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 1.8408203872350192e-05,
+      "loss": 0.8249,
+      "step": 493490
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 1.8407055942450445e-05,
+      "loss": 0.8014,
+      "step": 493500
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 1.84059080125507e-05,
+      "loss": 0.7159,
+      "step": 493510
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 1.8404760082650952e-05,
+      "loss": 0.8569,
+      "step": 493520
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 1.8403612152751205e-05,
+      "loss": 0.8529,
+      "step": 493530
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 1.8402464222851455e-05,
+      "loss": 0.8018,
+      "step": 493540
+    },
+    {
+      "epoch": 22.64,
+      "learning_rate": 1.8401316292951708e-05,
+      "loss": 0.7368,
+      "step": 493550
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 1.840016836305196e-05,
+      "loss": 0.6573,
+      "step": 493560
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 1.8399020433152215e-05,
+      "loss": 0.7726,
+      "step": 493570
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 1.8397872503252468e-05,
+      "loss": 0.7327,
+      "step": 493580
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 1.839672457335272e-05,
+      "loss": 0.7722,
+      "step": 493590
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 1.839557664345297e-05,
+      "loss": 0.7262,
+      "step": 493600
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 1.8394428713553224e-05,
+      "loss": 0.8256,
+      "step": 493610
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 1.8393280783653477e-05,
+      "loss": 0.6584,
+      "step": 493620
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 1.8392132853753727e-05,
+      "loss": 0.8811,
+      "step": 493630
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 1.839098492385398e-05,
+      "loss": 0.8051,
+      "step": 493640
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 1.8389836993954234e-05,
+      "loss": 0.8238,
+      "step": 493650
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 1.8388689064054487e-05,
+      "loss": 0.7636,
+      "step": 493660
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 1.838754113415474e-05,
+      "loss": 0.8566,
+      "step": 493670
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 1.8386393204254994e-05,
+      "loss": 0.6924,
+      "step": 493680
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 1.8385245274355243e-05,
+      "loss": 0.776,
+      "step": 493690
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 1.8384097344455497e-05,
+      "loss": 0.811,
+      "step": 493700
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 1.838294941455575e-05,
+      "loss": 0.8,
+      "step": 493710
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 1.8381801484656003e-05,
+      "loss": 0.768,
+      "step": 493720
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 1.8380653554756256e-05,
+      "loss": 0.7427,
+      "step": 493730
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 1.837950562485651e-05,
+      "loss": 0.7766,
+      "step": 493740
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 1.837835769495676e-05,
+      "loss": 0.7852,
+      "step": 493750
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 1.8377209765057013e-05,
+      "loss": 0.7367,
+      "step": 493760
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 1.8376061835157266e-05,
+      "loss": 0.747,
+      "step": 493770
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 1.8374913905257516e-05,
+      "loss": 0.755,
+      "step": 493780
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 1.837376597535777e-05,
+      "loss": 0.8161,
+      "step": 493790
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 1.8372618045458022e-05,
+      "loss": 0.7396,
+      "step": 493800
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 1.8371470115558276e-05,
+      "loss": 0.7581,
+      "step": 493810
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 1.837032218565853e-05,
+      "loss": 0.7428,
+      "step": 493820
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 1.836917425575878e-05,
+      "loss": 0.7965,
+      "step": 493830
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 1.8368026325859032e-05,
+      "loss": 0.7974,
+      "step": 493840
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 1.8366878395959285e-05,
+      "loss": 0.7588,
+      "step": 493850
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 1.836573046605954e-05,
+      "loss": 0.7143,
+      "step": 493860
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 1.836458253615979e-05,
+      "loss": 0.7649,
+      "step": 493870
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 1.8363434606260045e-05,
+      "loss": 0.8097,
+      "step": 493880
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 1.8362286676360295e-05,
+      "loss": 0.803,
+      "step": 493890
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 1.8361138746460548e-05,
+      "loss": 0.8628,
+      "step": 493900
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 1.83599908165608e-05,
+      "loss": 0.8112,
+      "step": 493910
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 1.835884288666105e-05,
+      "loss": 0.7521,
+      "step": 493920
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 1.8357694956761304e-05,
+      "loss": 0.7853,
+      "step": 493930
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 1.8356547026861557e-05,
+      "loss": 0.7945,
+      "step": 493940
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 1.835539909696181e-05,
+      "loss": 0.8204,
+      "step": 493950
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 1.8354251167062064e-05,
+      "loss": 0.7926,
+      "step": 493960
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 1.8353103237162317e-05,
+      "loss": 0.8067,
+      "step": 493970
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 1.8351955307262567e-05,
+      "loss": 0.7285,
+      "step": 493980
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 1.835080737736282e-05,
+      "loss": 0.7941,
+      "step": 493990
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 1.8349659447463074e-05,
+      "loss": 0.8633,
+      "step": 494000
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 1.8348511517563327e-05,
+      "loss": 0.7013,
+      "step": 494010
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 1.834736358766358e-05,
+      "loss": 0.7404,
+      "step": 494020
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 1.8346215657763833e-05,
+      "loss": 0.7675,
+      "step": 494030
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 1.8345067727864083e-05,
+      "loss": 0.741,
+      "step": 494040
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 1.8343919797964336e-05,
+      "loss": 0.7784,
+      "step": 494050
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 1.834277186806459e-05,
+      "loss": 0.8587,
+      "step": 494060
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 1.834162393816484e-05,
+      "loss": 0.7978,
+      "step": 494070
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 1.8340476008265093e-05,
+      "loss": 0.6892,
+      "step": 494080
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 1.8339328078365346e-05,
+      "loss": 0.7473,
+      "step": 494090
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 1.83381801484656e-05,
+      "loss": 0.6993,
+      "step": 494100
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 1.8337032218565852e-05,
+      "loss": 0.8564,
+      "step": 494110
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 1.8335884288666102e-05,
+      "loss": 0.7138,
+      "step": 494120
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 1.8334736358766356e-05,
+      "loss": 0.7212,
+      "step": 494130
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 1.833358842886661e-05,
+      "loss": 0.6811,
+      "step": 494140
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 1.8332440498966862e-05,
+      "loss": 0.8662,
+      "step": 494150
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 1.8331292569067115e-05,
+      "loss": 0.76,
+      "step": 494160
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 1.833014463916737e-05,
+      "loss": 0.7803,
+      "step": 494170
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 1.832899670926762e-05,
+      "loss": 0.7039,
+      "step": 494180
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 1.832784877936787e-05,
+      "loss": 0.7419,
+      "step": 494190
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 1.8326700849468125e-05,
+      "loss": 0.8215,
+      "step": 494200
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 1.8325552919568375e-05,
+      "loss": 0.7967,
+      "step": 494210
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 1.8324404989668628e-05,
+      "loss": 0.8208,
+      "step": 494220
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 1.832325705976888e-05,
+      "loss": 0.7342,
+      "step": 494230
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 1.8322109129869134e-05,
+      "loss": 0.8697,
+      "step": 494240
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 1.8320961199969388e-05,
+      "loss": 0.7632,
+      "step": 494250
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 1.831981327006964e-05,
+      "loss": 0.7408,
+      "step": 494260
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 1.831866534016989e-05,
+      "loss": 0.7765,
+      "step": 494270
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 1.8317517410270144e-05,
+      "loss": 0.6976,
+      "step": 494280
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 1.8316369480370397e-05,
+      "loss": 0.6818,
+      "step": 494290
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 1.831522155047065e-05,
+      "loss": 0.8043,
+      "step": 494300
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 1.8314073620570904e-05,
+      "loss": 0.7541,
+      "step": 494310
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 1.8312925690671157e-05,
+      "loss": 0.738,
+      "step": 494320
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 1.8311777760771407e-05,
+      "loss": 0.8441,
+      "step": 494330
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 1.831062983087166e-05,
+      "loss": 0.7613,
+      "step": 494340
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 1.8309481900971913e-05,
+      "loss": 0.7692,
+      "step": 494350
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 1.8308333971072163e-05,
+      "loss": 0.7483,
+      "step": 494360
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 1.8307186041172416e-05,
+      "loss": 0.7885,
+      "step": 494370
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 1.830603811127267e-05,
+      "loss": 0.7631,
+      "step": 494380
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 1.8304890181372923e-05,
+      "loss": 0.8128,
+      "step": 494390
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 1.8303742251473176e-05,
+      "loss": 0.7581,
+      "step": 494400
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 1.830259432157343e-05,
+      "loss": 0.7751,
+      "step": 494410
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 1.830144639167368e-05,
+      "loss": 0.7284,
+      "step": 494420
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 1.8300298461773932e-05,
+      "loss": 0.7927,
+      "step": 494430
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 1.8299150531874186e-05,
+      "loss": 0.7878,
+      "step": 494440
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 1.829800260197444e-05,
+      "loss": 0.7159,
+      "step": 494450
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 1.8296854672074692e-05,
+      "loss": 0.7768,
+      "step": 494460
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 1.8295706742174942e-05,
+      "loss": 0.7764,
+      "step": 494470
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 1.8294558812275195e-05,
+      "loss": 0.669,
+      "step": 494480
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 1.829341088237545e-05,
+      "loss": 0.7099,
+      "step": 494490
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 1.82922629524757e-05,
+      "loss": 0.712,
+      "step": 494500
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 1.829111502257595e-05,
+      "loss": 0.7913,
+      "step": 494510
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 1.8289967092676205e-05,
+      "loss": 0.7533,
+      "step": 494520
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 1.8288819162776458e-05,
+      "loss": 0.72,
+      "step": 494530
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 1.828767123287671e-05,
+      "loss": 0.6981,
+      "step": 494540
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 1.8286523302976965e-05,
+      "loss": 0.8073,
+      "step": 494550
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 1.8285375373077214e-05,
+      "loss": 0.789,
+      "step": 494560
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 1.8284227443177468e-05,
+      "loss": 0.764,
+      "step": 494570
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 1.828307951327772e-05,
+      "loss": 0.8261,
+      "step": 494580
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 1.8281931583377974e-05,
+      "loss": 0.7897,
+      "step": 494590
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 1.8280783653478227e-05,
+      "loss": 0.875,
+      "step": 494600
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 1.827963572357848e-05,
+      "loss": 0.743,
+      "step": 494610
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 1.827848779367873e-05,
+      "loss": 0.8009,
+      "step": 494620
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 1.8277339863778984e-05,
+      "loss": 0.7439,
+      "step": 494630
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 1.8276191933879237e-05,
+      "loss": 0.8045,
+      "step": 494640
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 1.8275044003979487e-05,
+      "loss": 0.7901,
+      "step": 494650
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 1.827389607407974e-05,
+      "loss": 0.8215,
+      "step": 494660
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 1.8272748144179993e-05,
+      "loss": 0.8433,
+      "step": 494670
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 1.8271600214280247e-05,
+      "loss": 0.7663,
+      "step": 494680
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 1.82704522843805e-05,
+      "loss": 0.7923,
+      "step": 494690
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 1.8269304354480753e-05,
+      "loss": 0.7453,
+      "step": 494700
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 1.8268156424581003e-05,
+      "loss": 0.7461,
+      "step": 494710
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 1.8267008494681256e-05,
+      "loss": 0.7763,
+      "step": 494720
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 1.826586056478151e-05,
+      "loss": 0.8075,
+      "step": 494730
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 1.8264712634881763e-05,
+      "loss": 0.6875,
+      "step": 494740
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 1.8263564704982016e-05,
+      "loss": 0.8148,
+      "step": 494750
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 1.826241677508227e-05,
+      "loss": 0.7749,
+      "step": 494760
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 1.826126884518252e-05,
+      "loss": 0.6498,
+      "step": 494770
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 1.8260120915282772e-05,
+      "loss": 0.7399,
+      "step": 494780
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 1.8258972985383025e-05,
+      "loss": 0.7318,
+      "step": 494790
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 1.8257825055483275e-05,
+      "loss": 0.8133,
+      "step": 494800
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 1.825667712558353e-05,
+      "loss": 0.7823,
+      "step": 494810
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 1.8255529195683782e-05,
+      "loss": 0.8258,
+      "step": 494820
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 1.8254381265784035e-05,
+      "loss": 0.7652,
+      "step": 494830
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 1.825323333588429e-05,
+      "loss": 0.7846,
+      "step": 494840
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 1.8252085405984538e-05,
+      "loss": 0.778,
+      "step": 494850
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 1.825093747608479e-05,
+      "loss": 0.7676,
+      "step": 494860
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 1.8249789546185045e-05,
+      "loss": 0.8343,
+      "step": 494870
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 1.8248641616285298e-05,
+      "loss": 0.7542,
+      "step": 494880
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 1.824749368638555e-05,
+      "loss": 0.8131,
+      "step": 494890
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 1.8246345756485804e-05,
+      "loss": 0.7016,
+      "step": 494900
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 1.8245197826586054e-05,
+      "loss": 0.6572,
+      "step": 494910
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 1.8244049896686307e-05,
+      "loss": 0.7777,
+      "step": 494920
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 1.824290196678656e-05,
+      "loss": 0.6979,
+      "step": 494930
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 1.824175403688681e-05,
+      "loss": 0.8551,
+      "step": 494940
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 1.8240606106987064e-05,
+      "loss": 0.75,
+      "step": 494950
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 1.8239458177087317e-05,
+      "loss": 0.7603,
+      "step": 494960
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 1.823831024718757e-05,
+      "loss": 0.7714,
+      "step": 494970
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 1.8237162317287824e-05,
+      "loss": 0.8203,
+      "step": 494980
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 1.8236014387388077e-05,
+      "loss": 0.7605,
+      "step": 494990
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 1.8234866457488327e-05,
+      "loss": 0.7419,
+      "step": 495000
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 1.823371852758858e-05,
+      "loss": 0.7895,
+      "step": 495010
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 1.8232570597688833e-05,
+      "loss": 0.788,
+      "step": 495020
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 1.8231422667789086e-05,
+      "loss": 0.7859,
+      "step": 495030
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 1.823027473788934e-05,
+      "loss": 0.7572,
+      "step": 495040
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 1.8229126807989593e-05,
+      "loss": 0.8263,
+      "step": 495050
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 1.8227978878089843e-05,
+      "loss": 0.6857,
+      "step": 495060
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 1.8226830948190096e-05,
+      "loss": 0.824,
+      "step": 495070
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 1.822568301829035e-05,
+      "loss": 0.6901,
+      "step": 495080
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 1.82245350883906e-05,
+      "loss": 0.8233,
+      "step": 495090
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 1.8223387158490852e-05,
+      "loss": 0.7538,
+      "step": 495100
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 1.8222239228591106e-05,
+      "loss": 0.7995,
+      "step": 495110
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 1.822109129869136e-05,
+      "loss": 0.8419,
+      "step": 495120
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 1.8219943368791612e-05,
+      "loss": 0.8455,
+      "step": 495130
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 1.8218795438891865e-05,
+      "loss": 0.8423,
+      "step": 495140
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 1.8217647508992115e-05,
+      "loss": 0.7829,
+      "step": 495150
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 1.821649957909237e-05,
+      "loss": 0.8245,
+      "step": 495160
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 1.821535164919262e-05,
+      "loss": 0.753,
+      "step": 495170
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 1.8214203719292875e-05,
+      "loss": 0.6887,
+      "step": 495180
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 1.8213055789393128e-05,
+      "loss": 0.7288,
+      "step": 495190
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 1.8211907859493378e-05,
+      "loss": 0.7464,
+      "step": 495200
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 1.821075992959363e-05,
+      "loss": 0.7485,
+      "step": 495210
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 1.8209611999693884e-05,
+      "loss": 0.7768,
+      "step": 495220
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 1.8208464069794134e-05,
+      "loss": 0.7865,
+      "step": 495230
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 1.8207316139894388e-05,
+      "loss": 0.8503,
+      "step": 495240
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 1.820616820999464e-05,
+      "loss": 0.7864,
+      "step": 495250
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 1.8205020280094894e-05,
+      "loss": 0.8085,
+      "step": 495260
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 1.8203872350195147e-05,
+      "loss": 0.7318,
+      "step": 495270
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 1.82027244202954e-05,
+      "loss": 0.6802,
+      "step": 495280
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 1.820157649039565e-05,
+      "loss": 0.8119,
+      "step": 495290
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 1.8200428560495904e-05,
+      "loss": 0.8233,
+      "step": 495300
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 1.8199280630596157e-05,
+      "loss": 0.7563,
+      "step": 495310
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 1.819813270069641e-05,
+      "loss": 0.83,
+      "step": 495320
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 1.8196984770796663e-05,
+      "loss": 0.8066,
+      "step": 495330
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 1.8195836840896917e-05,
+      "loss": 0.7928,
+      "step": 495340
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 1.8194688910997166e-05,
+      "loss": 0.7431,
+      "step": 495350
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 1.819354098109742e-05,
+      "loss": 0.691,
+      "step": 495360
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 1.8192393051197673e-05,
+      "loss": 0.8838,
+      "step": 495370
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 1.8191245121297923e-05,
+      "loss": 0.6722,
+      "step": 495380
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 1.8190097191398176e-05,
+      "loss": 0.7686,
+      "step": 495390
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 1.818894926149843e-05,
+      "loss": 0.6289,
+      "step": 495400
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 1.8187801331598682e-05,
+      "loss": 0.7755,
+      "step": 495410
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 1.8186653401698936e-05,
+      "loss": 0.8484,
+      "step": 495420
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 1.818550547179919e-05,
+      "loss": 0.7991,
+      "step": 495430
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 1.818435754189944e-05,
+      "loss": 0.7881,
+      "step": 495440
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 1.8183209611999692e-05,
+      "loss": 0.8035,
+      "step": 495450
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 1.8182061682099945e-05,
+      "loss": 0.7255,
+      "step": 495460
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 1.81809137522002e-05,
+      "loss": 0.7898,
+      "step": 495470
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 1.8179765822300452e-05,
+      "loss": 0.7241,
+      "step": 495480
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 1.81786178924007e-05,
+      "loss": 0.8099,
+      "step": 495490
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 1.8177469962500955e-05,
+      "loss": 0.7996,
+      "step": 495500
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 1.8176322032601208e-05,
+      "loss": 0.8055,
+      "step": 495510
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 1.8175174102701458e-05,
+      "loss": 0.7362,
+      "step": 495520
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 1.817402617280171e-05,
+      "loss": 0.7323,
+      "step": 495530
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 1.8172878242901964e-05,
+      "loss": 0.857,
+      "step": 495540
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 1.8171730313002218e-05,
+      "loss": 0.693,
+      "step": 495550
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 1.817058238310247e-05,
+      "loss": 0.7994,
+      "step": 495560
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 1.8169434453202724e-05,
+      "loss": 0.8139,
+      "step": 495570
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 1.8168286523302974e-05,
+      "loss": 0.7787,
+      "step": 495580
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 1.8167138593403227e-05,
+      "loss": 0.7704,
+      "step": 495590
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 1.816599066350348e-05,
+      "loss": 0.7675,
+      "step": 495600
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 1.8164842733603734e-05,
+      "loss": 0.6655,
+      "step": 495610
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 1.8163694803703987e-05,
+      "loss": 0.7877,
+      "step": 495620
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 1.816254687380424e-05,
+      "loss": 0.6673,
+      "step": 495630
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 1.816139894390449e-05,
+      "loss": 0.7871,
+      "step": 495640
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 1.8160251014004743e-05,
+      "loss": 0.904,
+      "step": 495650
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 1.8159103084104997e-05,
+      "loss": 0.7897,
+      "step": 495660
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 1.8157955154205246e-05,
+      "loss": 0.7149,
+      "step": 495670
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 1.81568072243055e-05,
+      "loss": 0.7471,
+      "step": 495680
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 1.8155659294405753e-05,
+      "loss": 0.785,
+      "step": 495690
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 1.8154511364506006e-05,
+      "loss": 0.755,
+      "step": 495700
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 1.815336343460626e-05,
+      "loss": 0.8048,
+      "step": 495710
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 1.8152215504706513e-05,
+      "loss": 0.7511,
+      "step": 495720
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 1.8151067574806763e-05,
+      "loss": 0.8161,
+      "step": 495730
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 1.8149919644907016e-05,
+      "loss": 0.8055,
+      "step": 495740
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 1.814877171500727e-05,
+      "loss": 0.8558,
+      "step": 495750
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 1.8147623785107522e-05,
+      "loss": 0.7022,
+      "step": 495760
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 1.8146475855207775e-05,
+      "loss": 0.7573,
+      "step": 495770
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 1.814532792530803e-05,
+      "loss": 0.9117,
+      "step": 495780
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 1.814417999540828e-05,
+      "loss": 0.7927,
+      "step": 495790
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 1.8143032065508532e-05,
+      "loss": 0.7941,
+      "step": 495800
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 1.8141884135608785e-05,
+      "loss": 0.7924,
+      "step": 495810
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 1.8140736205709035e-05,
+      "loss": 0.7618,
+      "step": 495820
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 1.8139588275809288e-05,
+      "loss": 0.7448,
+      "step": 495830
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 1.813844034590954e-05,
+      "loss": 0.7763,
+      "step": 495840
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 1.8137292416009795e-05,
+      "loss": 0.7705,
+      "step": 495850
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 1.8136144486110048e-05,
+      "loss": 0.8189,
+      "step": 495860
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 1.8134996556210298e-05,
+      "loss": 0.7565,
+      "step": 495870
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 1.813384862631055e-05,
+      "loss": 0.7741,
+      "step": 495880
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 1.8132700696410804e-05,
+      "loss": 0.7332,
+      "step": 495890
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 1.8131552766511057e-05,
+      "loss": 0.7957,
+      "step": 495900
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 1.813040483661131e-05,
+      "loss": 0.7867,
+      "step": 495910
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 1.8129256906711564e-05,
+      "loss": 0.7991,
+      "step": 495920
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 1.8128108976811814e-05,
+      "loss": 0.7925,
+      "step": 495930
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 1.8126961046912067e-05,
+      "loss": 0.7844,
+      "step": 495940
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 1.812581311701232e-05,
+      "loss": 0.8052,
+      "step": 495950
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 1.812466518711257e-05,
+      "loss": 0.8691,
+      "step": 495960
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 1.8123517257212823e-05,
+      "loss": 0.7713,
+      "step": 495970
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 1.8122369327313077e-05,
+      "loss": 0.7774,
+      "step": 495980
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 1.812122139741333e-05,
+      "loss": 0.7401,
+      "step": 495990
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 1.8120073467513583e-05,
+      "loss": 0.7532,
+      "step": 496000
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 1.8118925537613836e-05,
+      "loss": 0.8099,
+      "step": 496010
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 1.8117777607714086e-05,
+      "loss": 0.7866,
+      "step": 496020
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 1.811662967781434e-05,
+      "loss": 0.8482,
+      "step": 496030
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 1.8115481747914593e-05,
+      "loss": 0.7498,
+      "step": 496040
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 1.8114333818014846e-05,
+      "loss": 0.7306,
+      "step": 496050
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 1.81131858881151e-05,
+      "loss": 0.7283,
+      "step": 496060
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 1.8112037958215352e-05,
+      "loss": 0.7069,
+      "step": 496070
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 1.8110890028315602e-05,
+      "loss": 0.7981,
+      "step": 496080
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 1.8109742098415856e-05,
+      "loss": 0.7408,
+      "step": 496090
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 1.810859416851611e-05,
+      "loss": 0.7831,
+      "step": 496100
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 1.810744623861636e-05,
+      "loss": 0.7908,
+      "step": 496110
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 1.8106298308716612e-05,
+      "loss": 0.7749,
+      "step": 496120
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 1.8105150378816865e-05,
+      "loss": 0.7576,
+      "step": 496130
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 1.810400244891712e-05,
+      "loss": 0.7356,
+      "step": 496140
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 1.810285451901737e-05,
+      "loss": 0.721,
+      "step": 496150
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 1.8101706589117625e-05,
+      "loss": 0.698,
+      "step": 496160
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 1.8100558659217875e-05,
+      "loss": 0.7692,
+      "step": 496170
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 1.8099410729318128e-05,
+      "loss": 0.7018,
+      "step": 496180
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 1.809826279941838e-05,
+      "loss": 0.6619,
+      "step": 496190
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 1.8097114869518634e-05,
+      "loss": 0.7312,
+      "step": 496200
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 1.8095966939618888e-05,
+      "loss": 0.8316,
+      "step": 496210
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 1.8094819009719138e-05,
+      "loss": 0.7565,
+      "step": 496220
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 1.809367107981939e-05,
+      "loss": 0.7411,
+      "step": 496230
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 1.8092523149919644e-05,
+      "loss": 0.7689,
+      "step": 496240
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 1.8091375220019894e-05,
+      "loss": 0.7697,
+      "step": 496250
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 1.8090227290120147e-05,
+      "loss": 0.7884,
+      "step": 496260
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 1.80890793602204e-05,
+      "loss": 0.7423,
+      "step": 496270
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 1.8087931430320654e-05,
+      "loss": 0.7248,
+      "step": 496280
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 1.8086783500420907e-05,
+      "loss": 0.8529,
+      "step": 496290
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 1.808563557052116e-05,
+      "loss": 0.7799,
+      "step": 496300
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 1.808448764062141e-05,
+      "loss": 0.8243,
+      "step": 496310
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 1.8083339710721663e-05,
+      "loss": 0.8096,
+      "step": 496320
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 1.8082191780821916e-05,
+      "loss": 0.8779,
+      "step": 496330
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 1.808104385092217e-05,
+      "loss": 0.7965,
+      "step": 496340
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 1.8079895921022423e-05,
+      "loss": 0.7701,
+      "step": 496350
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 1.8078747991122676e-05,
+      "loss": 0.767,
+      "step": 496360
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 1.8077600061222926e-05,
+      "loss": 0.8187,
+      "step": 496370
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 1.807645213132318e-05,
+      "loss": 0.7113,
+      "step": 496380
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 1.8075304201423432e-05,
+      "loss": 0.7493,
+      "step": 496390
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 1.8074156271523682e-05,
+      "loss": 0.8217,
+      "step": 496400
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 1.8073008341623936e-05,
+      "loss": 0.81,
+      "step": 496410
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 1.807186041172419e-05,
+      "loss": 0.7758,
+      "step": 496420
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 1.8070712481824442e-05,
+      "loss": 0.7173,
+      "step": 496430
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 1.8069564551924695e-05,
+      "loss": 0.7922,
+      "step": 496440
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 1.806841662202495e-05,
+      "loss": 0.7149,
+      "step": 496450
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 1.80672686921252e-05,
+      "loss": 0.8114,
+      "step": 496460
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 1.806612076222545e-05,
+      "loss": 0.7907,
+      "step": 496470
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 1.8064972832325705e-05,
+      "loss": 0.7673,
+      "step": 496480
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 1.8063824902425958e-05,
+      "loss": 0.7797,
+      "step": 496490
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 1.806267697252621e-05,
+      "loss": 0.7281,
+      "step": 496500
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 1.8061529042626465e-05,
+      "loss": 0.8014,
+      "step": 496510
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 1.8060381112726714e-05,
+      "loss": 0.7112,
+      "step": 496520
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 1.8059233182826968e-05,
+      "loss": 0.8144,
+      "step": 496530
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 1.8058085252927218e-05,
+      "loss": 0.7594,
+      "step": 496540
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 1.805693732302747e-05,
+      "loss": 0.8513,
+      "step": 496550
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 1.8055789393127724e-05,
+      "loss": 0.7035,
+      "step": 496560
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 1.8054641463227977e-05,
+      "loss": 0.7586,
+      "step": 496570
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 1.805349353332823e-05,
+      "loss": 0.7486,
+      "step": 496580
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 1.8052345603428484e-05,
+      "loss": 0.7842,
+      "step": 496590
+    },
+    {
+      "epoch": 22.78,
+      "learning_rate": 1.8051197673528734e-05,
+      "loss": 0.7181,
+      "step": 496600
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 1.8050049743628987e-05,
+      "loss": 0.8273,
+      "step": 496610
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 1.804890181372924e-05,
+      "loss": 0.7975,
+      "step": 496620
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 1.8047753883829493e-05,
+      "loss": 0.7325,
+      "step": 496630
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 1.8046605953929747e-05,
+      "loss": 0.7156,
+      "step": 496640
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 1.804545802403e-05,
+      "loss": 0.7791,
+      "step": 496650
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 1.804431009413025e-05,
+      "loss": 0.7955,
+      "step": 496660
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 1.8043162164230503e-05,
+      "loss": 0.7379,
+      "step": 496670
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 1.8042014234330756e-05,
+      "loss": 0.7593,
+      "step": 496680
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 1.8040866304431006e-05,
+      "loss": 0.7505,
+      "step": 496690
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 1.803971837453126e-05,
+      "loss": 0.7734,
+      "step": 496700
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 1.8038570444631513e-05,
+      "loss": 0.7938,
+      "step": 496710
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 1.8037422514731766e-05,
+      "loss": 0.7831,
+      "step": 496720
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 1.803627458483202e-05,
+      "loss": 0.8107,
+      "step": 496730
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 1.8035126654932272e-05,
+      "loss": 0.7678,
+      "step": 496740
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 1.8033978725032522e-05,
+      "loss": 0.8521,
+      "step": 496750
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 1.8032830795132775e-05,
+      "loss": 0.7384,
+      "step": 496760
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 1.803168286523303e-05,
+      "loss": 0.6779,
+      "step": 496770
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 1.8030534935333282e-05,
+      "loss": 0.7707,
+      "step": 496780
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 1.8029387005433535e-05,
+      "loss": 0.7792,
+      "step": 496790
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 1.802823907553379e-05,
+      "loss": 0.7956,
+      "step": 496800
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 1.8027091145634038e-05,
+      "loss": 0.7882,
+      "step": 496810
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 1.802594321573429e-05,
+      "loss": 0.8169,
+      "step": 496820
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 1.8024795285834545e-05,
+      "loss": 0.7478,
+      "step": 496830
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 1.8023647355934795e-05,
+      "loss": 0.7642,
+      "step": 496840
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 1.8022499426035048e-05,
+      "loss": 0.72,
+      "step": 496850
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 1.80213514961353e-05,
+      "loss": 0.7498,
+      "step": 496860
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 1.8020203566235554e-05,
+      "loss": 0.744,
+      "step": 496870
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 1.8019055636335807e-05,
+      "loss": 0.7641,
+      "step": 496880
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 1.8017907706436057e-05,
+      "loss": 0.7884,
+      "step": 496890
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 1.801675977653631e-05,
+      "loss": 0.7725,
+      "step": 496900
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 1.8015611846636564e-05,
+      "loss": 0.8437,
+      "step": 496910
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 1.8014463916736817e-05,
+      "loss": 0.8139,
+      "step": 496920
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 1.801331598683707e-05,
+      "loss": 0.7801,
+      "step": 496930
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 1.8012168056937324e-05,
+      "loss": 0.7472,
+      "step": 496940
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 1.8011020127037573e-05,
+      "loss": 0.8351,
+      "step": 496950
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 1.8009872197137827e-05,
+      "loss": 0.7258,
+      "step": 496960
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 1.800872426723808e-05,
+      "loss": 0.7908,
+      "step": 496970
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 1.800757633733833e-05,
+      "loss": 0.8001,
+      "step": 496980
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 1.8006428407438583e-05,
+      "loss": 0.7601,
+      "step": 496990
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 1.8005280477538836e-05,
+      "loss": 0.7396,
+      "step": 497000
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 1.800413254763909e-05,
+      "loss": 0.7024,
+      "step": 497010
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 1.8002984617739343e-05,
+      "loss": 0.7609,
+      "step": 497020
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 1.8001836687839596e-05,
+      "loss": 0.7327,
+      "step": 497030
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 1.8000688757939846e-05,
+      "loss": 0.6965,
+      "step": 497040
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 1.79995408280401e-05,
+      "loss": 0.8522,
+      "step": 497050
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 1.7998392898140352e-05,
+      "loss": 0.7924,
+      "step": 497060
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 1.7997244968240606e-05,
+      "loss": 0.7523,
+      "step": 497070
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 1.799609703834086e-05,
+      "loss": 0.7485,
+      "step": 497080
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 1.7994949108441112e-05,
+      "loss": 0.8166,
+      "step": 497090
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 1.7993801178541362e-05,
+      "loss": 0.7626,
+      "step": 497100
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 1.7992653248641615e-05,
+      "loss": 0.8206,
+      "step": 497110
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 1.799150531874187e-05,
+      "loss": 0.8138,
+      "step": 497120
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 1.7990357388842118e-05,
+      "loss": 0.7585,
+      "step": 497130
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 1.798920945894237e-05,
+      "loss": 0.7719,
+      "step": 497140
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 1.7988061529042625e-05,
+      "loss": 0.8213,
+      "step": 497150
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 1.7986913599142878e-05,
+      "loss": 0.8101,
+      "step": 497160
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 1.798576566924313e-05,
+      "loss": 0.8131,
+      "step": 497170
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 1.7984617739343384e-05,
+      "loss": 0.8246,
+      "step": 497180
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 1.7983469809443634e-05,
+      "loss": 0.8726,
+      "step": 497190
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 1.7982321879543888e-05,
+      "loss": 0.8529,
+      "step": 497200
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 1.798117394964414e-05,
+      "loss": 0.7374,
+      "step": 497210
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 1.7980026019744394e-05,
+      "loss": 0.8569,
+      "step": 497220
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 1.7978878089844647e-05,
+      "loss": 0.7904,
+      "step": 497230
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 1.7977730159944897e-05,
+      "loss": 0.8533,
+      "step": 497240
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 1.797658223004515e-05,
+      "loss": 0.724,
+      "step": 497250
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 1.7975434300145404e-05,
+      "loss": 0.7945,
+      "step": 497260
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 1.7974286370245653e-05,
+      "loss": 0.6518,
+      "step": 497270
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 1.7973138440345907e-05,
+      "loss": 0.7859,
+      "step": 497280
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 1.797199051044616e-05,
+      "loss": 0.7061,
+      "step": 497290
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 1.7970842580546413e-05,
+      "loss": 0.8025,
+      "step": 497300
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 1.7969694650646666e-05,
+      "loss": 0.8025,
+      "step": 497310
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 1.796854672074692e-05,
+      "loss": 0.6946,
+      "step": 497320
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 1.796739879084717e-05,
+      "loss": 0.7933,
+      "step": 497330
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 1.7966250860947423e-05,
+      "loss": 0.7247,
+      "step": 497340
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 1.7965102931047676e-05,
+      "loss": 0.7541,
+      "step": 497350
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 1.796395500114793e-05,
+      "loss": 0.7462,
+      "step": 497360
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 1.7962807071248182e-05,
+      "loss": 0.8355,
+      "step": 497370
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 1.7961659141348436e-05,
+      "loss": 0.7347,
+      "step": 497380
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 1.7960511211448686e-05,
+      "loss": 0.7738,
+      "step": 497390
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 1.795936328154894e-05,
+      "loss": 0.8192,
+      "step": 497400
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 1.7958215351649192e-05,
+      "loss": 0.7703,
+      "step": 497410
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 1.7957067421749442e-05,
+      "loss": 0.748,
+      "step": 497420
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 1.7955919491849695e-05,
+      "loss": 0.7476,
+      "step": 497430
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 1.795477156194995e-05,
+      "loss": 0.7625,
+      "step": 497440
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 1.79536236320502e-05,
+      "loss": 0.7963,
+      "step": 497450
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 1.7952475702150455e-05,
+      "loss": 0.7461,
+      "step": 497460
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 1.7951327772250708e-05,
+      "loss": 0.745,
+      "step": 497470
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 1.7950179842350958e-05,
+      "loss": 0.7683,
+      "step": 497480
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 1.794903191245121e-05,
+      "loss": 0.8344,
+      "step": 497490
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 1.7947883982551464e-05,
+      "loss": 0.7513,
+      "step": 497500
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 1.7946736052651718e-05,
+      "loss": 0.8249,
+      "step": 497510
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 1.794558812275197e-05,
+      "loss": 0.7496,
+      "step": 497520
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 1.7944440192852224e-05,
+      "loss": 0.7715,
+      "step": 497530
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 1.7943292262952474e-05,
+      "loss": 0.7673,
+      "step": 497540
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 1.7942144333052727e-05,
+      "loss": 0.7645,
+      "step": 497550
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 1.7940996403152977e-05,
+      "loss": 0.8049,
+      "step": 497560
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 1.793984847325323e-05,
+      "loss": 0.8182,
+      "step": 497570
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 1.7938700543353484e-05,
+      "loss": 0.8629,
+      "step": 497580
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 1.7937552613453737e-05,
+      "loss": 0.7713,
+      "step": 497590
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 1.793640468355399e-05,
+      "loss": 0.7344,
+      "step": 497600
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 1.7935256753654243e-05,
+      "loss": 0.8026,
+      "step": 497610
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 1.7934108823754493e-05,
+      "loss": 0.7044,
+      "step": 497620
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 1.7932960893854746e-05,
+      "loss": 0.794,
+      "step": 497630
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 1.7931812963955e-05,
+      "loss": 0.8536,
+      "step": 497640
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 1.7930665034055253e-05,
+      "loss": 0.7729,
+      "step": 497650
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 1.7929517104155506e-05,
+      "loss": 0.6899,
+      "step": 497660
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 1.792836917425576e-05,
+      "loss": 0.7968,
+      "step": 497670
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 1.792722124435601e-05,
+      "loss": 0.7942,
+      "step": 497680
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 1.7926073314456263e-05,
+      "loss": 0.811,
+      "step": 497690
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 1.7924925384556516e-05,
+      "loss": 0.7225,
+      "step": 497700
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 1.7923777454656766e-05,
+      "loss": 0.7766,
+      "step": 497710
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 1.792262952475702e-05,
+      "loss": 0.7335,
+      "step": 497720
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 1.7921481594857272e-05,
+      "loss": 0.8362,
+      "step": 497730
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 1.7920333664957525e-05,
+      "loss": 0.779,
+      "step": 497740
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 1.791918573505778e-05,
+      "loss": 0.7395,
+      "step": 497750
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 1.7918037805158032e-05,
+      "loss": 0.6928,
+      "step": 497760
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 1.791688987525828e-05,
+      "loss": 0.6701,
+      "step": 497770
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 1.7915741945358535e-05,
+      "loss": 0.7722,
+      "step": 497780
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 1.7914594015458788e-05,
+      "loss": 0.8191,
+      "step": 497790
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 1.791344608555904e-05,
+      "loss": 0.8823,
+      "step": 497800
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 1.7912298155659295e-05,
+      "loss": 0.7022,
+      "step": 497810
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 1.7911150225759548e-05,
+      "loss": 0.7985,
+      "step": 497820
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 1.7910002295859798e-05,
+      "loss": 0.7704,
+      "step": 497830
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 1.790885436596005e-05,
+      "loss": 0.8115,
+      "step": 497840
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 1.7907706436060304e-05,
+      "loss": 0.8056,
+      "step": 497850
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 1.7906558506160554e-05,
+      "loss": 0.6829,
+      "step": 497860
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 1.7905410576260807e-05,
+      "loss": 0.8027,
+      "step": 497870
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 1.790426264636106e-05,
+      "loss": 0.7125,
+      "step": 497880
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 1.7903114716461314e-05,
+      "loss": 0.8896,
+      "step": 497890
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 1.7901966786561567e-05,
+      "loss": 0.7841,
+      "step": 497900
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 1.7900818856661817e-05,
+      "loss": 0.7616,
+      "step": 497910
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 1.789967092676207e-05,
+      "loss": 0.72,
+      "step": 497920
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 1.7898522996862323e-05,
+      "loss": 0.7559,
+      "step": 497930
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 1.7897375066962577e-05,
+      "loss": 0.783,
+      "step": 497940
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 1.789622713706283e-05,
+      "loss": 0.8405,
+      "step": 497950
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 1.7895079207163083e-05,
+      "loss": 0.7386,
+      "step": 497960
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 1.7893931277263333e-05,
+      "loss": 0.7504,
+      "step": 497970
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 1.7892783347363586e-05,
+      "loss": 0.7793,
+      "step": 497980
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 1.789163541746384e-05,
+      "loss": 0.7817,
+      "step": 497990
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 1.789048748756409e-05,
+      "loss": 0.8444,
+      "step": 498000
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 1.7889339557664343e-05,
+      "loss": 0.7945,
+      "step": 498010
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 1.7888191627764596e-05,
+      "loss": 0.6991,
+      "step": 498020
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 1.788704369786485e-05,
+      "loss": 0.8332,
+      "step": 498030
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 1.7885895767965102e-05,
+      "loss": 0.8611,
+      "step": 498040
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 1.7884747838065356e-05,
+      "loss": 0.7423,
+      "step": 498050
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 1.7883599908165605e-05,
+      "loss": 0.7411,
+      "step": 498060
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 1.788245197826586e-05,
+      "loss": 0.7373,
+      "step": 498070
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 1.7881304048366112e-05,
+      "loss": 0.7304,
+      "step": 498080
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 1.7880156118466365e-05,
+      "loss": 0.8056,
+      "step": 498090
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 1.787900818856662e-05,
+      "loss": 0.791,
+      "step": 498100
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 1.787786025866687e-05,
+      "loss": 0.7053,
+      "step": 498110
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 1.787671232876712e-05,
+      "loss": 0.7932,
+      "step": 498120
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 1.7875564398867375e-05,
+      "loss": 0.8,
+      "step": 498130
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 1.7874416468967628e-05,
+      "loss": 0.7058,
+      "step": 498140
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 1.7873268539067878e-05,
+      "loss": 0.713,
+      "step": 498150
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 1.787212060916813e-05,
+      "loss": 0.7754,
+      "step": 498160
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 1.7870972679268384e-05,
+      "loss": 0.7527,
+      "step": 498170
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 1.7869824749368638e-05,
+      "loss": 0.7281,
+      "step": 498180
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 1.786867681946889e-05,
+      "loss": 0.7943,
+      "step": 498190
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 1.7867528889569144e-05,
+      "loss": 0.8578,
+      "step": 498200
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 1.7866380959669394e-05,
+      "loss": 0.7438,
+      "step": 498210
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 1.7865233029769647e-05,
+      "loss": 0.8193,
+      "step": 498220
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 1.78640850998699e-05,
+      "loss": 0.8206,
+      "step": 498230
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 1.7862937169970154e-05,
+      "loss": 0.8097,
+      "step": 498240
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 1.7861789240070407e-05,
+      "loss": 0.797,
+      "step": 498250
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 1.7860641310170657e-05,
+      "loss": 0.7366,
+      "step": 498260
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 1.785949338027091e-05,
+      "loss": 0.8688,
+      "step": 498270
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 1.7858345450371163e-05,
+      "loss": 0.7861,
+      "step": 498280
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 1.7857197520471413e-05,
+      "loss": 0.7613,
+      "step": 498290
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 1.7856049590571666e-05,
+      "loss": 0.7509,
+      "step": 498300
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 1.785490166067192e-05,
+      "loss": 0.7659,
+      "step": 498310
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 1.7853753730772173e-05,
+      "loss": 0.6915,
+      "step": 498320
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 1.7852605800872426e-05,
+      "loss": 0.7186,
+      "step": 498330
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 1.785145787097268e-05,
+      "loss": 0.7452,
+      "step": 498340
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 1.785030994107293e-05,
+      "loss": 0.7338,
+      "step": 498350
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 1.7849162011173182e-05,
+      "loss": 0.7702,
+      "step": 498360
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 1.7848014081273436e-05,
+      "loss": 0.7539,
+      "step": 498370
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 1.784686615137369e-05,
+      "loss": 0.7779,
+      "step": 498380
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 1.7845718221473942e-05,
+      "loss": 0.778,
+      "step": 498390
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 1.7844570291574195e-05,
+      "loss": 0.7853,
+      "step": 498400
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 1.7843422361674445e-05,
+      "loss": 0.7856,
+      "step": 498410
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 1.78422744317747e-05,
+      "loss": 0.67,
+      "step": 498420
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 1.784112650187495e-05,
+      "loss": 0.813,
+      "step": 498430
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 1.78399785719752e-05,
+      "loss": 0.799,
+      "step": 498440
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 1.7838830642075455e-05,
+      "loss": 0.7568,
+      "step": 498450
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 1.7837682712175708e-05,
+      "loss": 0.7682,
+      "step": 498460
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 1.783653478227596e-05,
+      "loss": 0.7032,
+      "step": 498470
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 1.7835386852376214e-05,
+      "loss": 0.754,
+      "step": 498480
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 1.7834238922476468e-05,
+      "loss": 0.8268,
+      "step": 498490
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 1.7833090992576718e-05,
+      "loss": 0.8557,
+      "step": 498500
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 1.783194306267697e-05,
+      "loss": 0.7248,
+      "step": 498510
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 1.7830795132777224e-05,
+      "loss": 0.8394,
+      "step": 498520
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 1.7829647202877477e-05,
+      "loss": 0.8142,
+      "step": 498530
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 1.782849927297773e-05,
+      "loss": 0.7392,
+      "step": 498540
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 1.7827351343077984e-05,
+      "loss": 0.7836,
+      "step": 498550
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 1.7826203413178234e-05,
+      "loss": 0.7675,
+      "step": 498560
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 1.7825055483278487e-05,
+      "loss": 0.7419,
+      "step": 498570
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 1.7823907553378737e-05,
+      "loss": 0.7567,
+      "step": 498580
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 1.782275962347899e-05,
+      "loss": 0.7277,
+      "step": 498590
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 1.7821611693579243e-05,
+      "loss": 0.7881,
+      "step": 498600
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 1.7820463763679496e-05,
+      "loss": 0.7516,
+      "step": 498610
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 1.781931583377975e-05,
+      "loss": 0.7854,
+      "step": 498620
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 1.7818167903880003e-05,
+      "loss": 0.9041,
+      "step": 498630
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 1.7817019973980253e-05,
+      "loss": 0.7313,
+      "step": 498640
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 1.7815872044080506e-05,
+      "loss": 0.887,
+      "step": 498650
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 1.781472411418076e-05,
+      "loss": 0.7881,
+      "step": 498660
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 1.7813576184281013e-05,
+      "loss": 0.7178,
+      "step": 498670
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 1.7812428254381266e-05,
+      "loss": 0.7288,
+      "step": 498680
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 1.781128032448152e-05,
+      "loss": 0.7695,
+      "step": 498690
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 1.781013239458177e-05,
+      "loss": 0.7734,
+      "step": 498700
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 1.7808984464682022e-05,
+      "loss": 0.7765,
+      "step": 498710
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 1.7807836534782275e-05,
+      "loss": 0.7001,
+      "step": 498720
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 1.7806688604882525e-05,
+      "loss": 0.8599,
+      "step": 498730
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 1.780554067498278e-05,
+      "loss": 0.7891,
+      "step": 498740
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 1.780439274508303e-05,
+      "loss": 0.7323,
+      "step": 498750
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 1.7803244815183285e-05,
+      "loss": 0.7827,
+      "step": 498760
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 1.7802096885283538e-05,
+      "loss": 0.7239,
+      "step": 498770
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 1.780094895538379e-05,
+      "loss": 0.7928,
+      "step": 498780
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 1.779980102548404e-05,
+      "loss": 0.8271,
+      "step": 498790
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 1.7798653095584294e-05,
+      "loss": 0.8315,
+      "step": 498800
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 1.7797505165684548e-05,
+      "loss": 0.8259,
+      "step": 498810
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 1.77963572357848e-05,
+      "loss": 0.7404,
+      "step": 498820
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 1.7795209305885054e-05,
+      "loss": 0.6885,
+      "step": 498830
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 1.7794061375985307e-05,
+      "loss": 0.7333,
+      "step": 498840
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 1.7792913446085557e-05,
+      "loss": 0.7637,
+      "step": 498850
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 1.779176551618581e-05,
+      "loss": 0.7623,
+      "step": 498860
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 1.7790617586286064e-05,
+      "loss": 0.7589,
+      "step": 498870
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 1.7789469656386314e-05,
+      "loss": 0.7007,
+      "step": 498880
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 1.7788321726486567e-05,
+      "loss": 0.6869,
+      "step": 498890
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 1.778717379658682e-05,
+      "loss": 0.7567,
+      "step": 498900
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 1.7786025866687073e-05,
+      "loss": 0.7608,
+      "step": 498910
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 1.7784877936787327e-05,
+      "loss": 0.7675,
+      "step": 498920
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 1.7783730006887576e-05,
+      "loss": 0.7695,
+      "step": 498930
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 1.778258207698783e-05,
+      "loss": 0.8138,
+      "step": 498940
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 1.7781434147088083e-05,
+      "loss": 0.7856,
+      "step": 498950
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 1.7780286217188336e-05,
+      "loss": 0.7689,
+      "step": 498960
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 1.777913828728859e-05,
+      "loss": 0.7305,
+      "step": 498970
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 1.7777990357388843e-05,
+      "loss": 0.7789,
+      "step": 498980
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 1.7776842427489093e-05,
+      "loss": 0.7255,
+      "step": 498990
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 1.7775694497589346e-05,
+      "loss": 0.7259,
+      "step": 499000
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 1.77745465676896e-05,
+      "loss": 0.7966,
+      "step": 499010
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 1.777339863778985e-05,
+      "loss": 0.7299,
+      "step": 499020
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 1.7772250707890102e-05,
+      "loss": 0.6999,
+      "step": 499030
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 1.7771102777990355e-05,
+      "loss": 0.8092,
+      "step": 499040
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 1.776995484809061e-05,
+      "loss": 0.8568,
+      "step": 499050
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 1.7768806918190862e-05,
+      "loss": 0.793,
+      "step": 499060
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 1.7767658988291115e-05,
+      "loss": 0.735,
+      "step": 499070
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 1.7766511058391365e-05,
+      "loss": 0.7561,
+      "step": 499080
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 1.7765363128491618e-05,
+      "loss": 0.7015,
+      "step": 499090
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 1.776421519859187e-05,
+      "loss": 0.8107,
+      "step": 499100
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 1.7763067268692125e-05,
+      "loss": 0.7916,
+      "step": 499110
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 1.7761919338792378e-05,
+      "loss": 0.7497,
+      "step": 499120
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 1.776077140889263e-05,
+      "loss": 0.8029,
+      "step": 499130
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 1.775962347899288e-05,
+      "loss": 0.7629,
+      "step": 499140
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 1.7758475549093134e-05,
+      "loss": 0.7256,
+      "step": 499150
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 1.7757327619193388e-05,
+      "loss": 0.6896,
+      "step": 499160
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 1.7756179689293637e-05,
+      "loss": 0.818,
+      "step": 499170
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 1.775503175939389e-05,
+      "loss": 0.8046,
+      "step": 499180
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 1.7753883829494144e-05,
+      "loss": 0.7357,
+      "step": 499190
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 1.7752735899594397e-05,
+      "loss": 0.8134,
+      "step": 499200
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 1.775158796969465e-05,
+      "loss": 0.7125,
+      "step": 499210
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 1.7750440039794904e-05,
+      "loss": 0.6901,
+      "step": 499220
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 1.7749292109895153e-05,
+      "loss": 0.7726,
+      "step": 499230
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 1.7748144179995407e-05,
+      "loss": 0.7486,
+      "step": 499240
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 1.774699625009566e-05,
+      "loss": 0.7602,
+      "step": 499250
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 1.7745848320195913e-05,
+      "loss": 0.7865,
+      "step": 499260
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 1.7744700390296166e-05,
+      "loss": 0.8049,
+      "step": 499270
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 1.7743552460396416e-05,
+      "loss": 0.6569,
+      "step": 499280
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 1.774240453049667e-05,
+      "loss": 0.752,
+      "step": 499290
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 1.7741256600596923e-05,
+      "loss": 0.7684,
+      "step": 499300
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 1.7740108670697173e-05,
+      "loss": 0.846,
+      "step": 499310
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 1.7738960740797426e-05,
+      "loss": 0.7615,
+      "step": 499320
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 1.773781281089768e-05,
+      "loss": 0.8166,
+      "step": 499330
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 1.7736664880997932e-05,
+      "loss": 0.7238,
+      "step": 499340
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 1.7735516951098186e-05,
+      "loss": 0.7723,
+      "step": 499350
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 1.773436902119844e-05,
+      "loss": 0.7147,
+      "step": 499360
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 1.773322109129869e-05,
+      "loss": 0.7646,
+      "step": 499370
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 1.7732073161398942e-05,
+      "loss": 0.7878,
+      "step": 499380
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 1.7730925231499195e-05,
+      "loss": 0.8488,
+      "step": 499390
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 1.7729777301599445e-05,
+      "loss": 0.7477,
+      "step": 499400
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 1.77286293716997e-05,
+      "loss": 0.7685,
+      "step": 499410
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 1.7727481441799955e-05,
+      "loss": 0.8191,
+      "step": 499420
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 1.7726333511900205e-05,
+      "loss": 0.7538,
+      "step": 499430
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 1.7725185582000458e-05,
+      "loss": 0.7453,
+      "step": 499440
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 1.772403765210071e-05,
+      "loss": 0.7735,
+      "step": 499450
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 1.772288972220096e-05,
+      "loss": 0.8026,
+      "step": 499460
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 1.7721741792301214e-05,
+      "loss": 0.7676,
+      "step": 499470
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 1.7720593862401468e-05,
+      "loss": 0.7198,
+      "step": 499480
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 1.771944593250172e-05,
+      "loss": 0.7583,
+      "step": 499490
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 1.7718298002601974e-05,
+      "loss": 0.8123,
+      "step": 499500
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 1.7717150072702227e-05,
+      "loss": 0.7988,
+      "step": 499510
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 1.7716002142802477e-05,
+      "loss": 0.6963,
+      "step": 499520
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 1.771485421290273e-05,
+      "loss": 0.7997,
+      "step": 499530
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 1.7713706283002984e-05,
+      "loss": 0.8417,
+      "step": 499540
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 1.7712558353103233e-05,
+      "loss": 0.8638,
+      "step": 499550
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 1.771141042320349e-05,
+      "loss": 0.7929,
+      "step": 499560
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 1.7710262493303743e-05,
+      "loss": 0.7345,
+      "step": 499570
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 1.7709114563403993e-05,
+      "loss": 0.843,
+      "step": 499580
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 1.7707966633504246e-05,
+      "loss": 0.7279,
+      "step": 499590
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 1.77068187036045e-05,
+      "loss": 0.7254,
+      "step": 499600
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 1.770567077370475e-05,
+      "loss": 0.7753,
+      "step": 499610
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 1.7704522843805003e-05,
+      "loss": 0.7243,
+      "step": 499620
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 1.7703374913905256e-05,
+      "loss": 0.7857,
+      "step": 499630
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 1.770222698400551e-05,
+      "loss": 0.7594,
+      "step": 499640
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 1.7701079054105763e-05,
+      "loss": 0.7571,
+      "step": 499650
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 1.7699931124206012e-05,
+      "loss": 0.7963,
+      "step": 499660
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 1.7698783194306266e-05,
+      "loss": 0.8116,
+      "step": 499670
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 1.769763526440652e-05,
+      "loss": 0.7716,
+      "step": 499680
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 1.769648733450677e-05,
+      "loss": 0.8397,
+      "step": 499690
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 1.7695339404607022e-05,
+      "loss": 0.786,
+      "step": 499700
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 1.769419147470728e-05,
+      "loss": 0.7509,
+      "step": 499710
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 1.769304354480753e-05,
+      "loss": 0.7969,
+      "step": 499720
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 1.769189561490778e-05,
+      "loss": 0.7924,
+      "step": 499730
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 1.7690747685008035e-05,
+      "loss": 0.6905,
+      "step": 499740
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 1.7689599755108285e-05,
+      "loss": 0.6767,
+      "step": 499750
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 1.7688451825208538e-05,
+      "loss": 0.7618,
+      "step": 499760
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 1.768730389530879e-05,
+      "loss": 0.7908,
+      "step": 499770
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 1.7686155965409044e-05,
+      "loss": 0.6845,
+      "step": 499780
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 1.7685008035509298e-05,
+      "loss": 0.7201,
+      "step": 499790
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 1.768386010560955e-05,
+      "loss": 0.7269,
+      "step": 499800
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 1.76827121757098e-05,
+      "loss": 0.7316,
+      "step": 499810
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 1.7681564245810054e-05,
+      "loss": 0.7878,
+      "step": 499820
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 1.7680416315910307e-05,
+      "loss": 0.8109,
+      "step": 499830
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 1.7679268386010557e-05,
+      "loss": 0.7099,
+      "step": 499840
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 1.767812045611081e-05,
+      "loss": 0.7077,
+      "step": 499850
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 1.7676972526211067e-05,
+      "loss": 0.8343,
+      "step": 499860
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 1.7675824596311317e-05,
+      "loss": 0.7506,
+      "step": 499870
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 1.767467666641157e-05,
+      "loss": 0.713,
+      "step": 499880
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 1.7673528736511823e-05,
+      "loss": 0.8139,
+      "step": 499890
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 1.7672380806612073e-05,
+      "loss": 0.7179,
+      "step": 499900
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 1.7671232876712326e-05,
+      "loss": 0.7761,
+      "step": 499910
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 1.767008494681258e-05,
+      "loss": 0.8239,
+      "step": 499920
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 1.7668937016912833e-05,
+      "loss": 0.7387,
+      "step": 499930
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 1.7667789087013086e-05,
+      "loss": 0.7916,
+      "step": 499940
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 1.7666641157113336e-05,
+      "loss": 0.8238,
+      "step": 499950
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 1.766549322721359e-05,
+      "loss": 0.8733,
+      "step": 499960
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 1.7664345297313843e-05,
+      "loss": 0.752,
+      "step": 499970
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 1.7663197367414092e-05,
+      "loss": 0.7332,
+      "step": 499980
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 1.7662049437514346e-05,
+      "loss": 0.7525,
+      "step": 499990
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 1.76609015076146e-05,
+      "loss": 0.8066,
+      "step": 500000
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 1.7659753577714852e-05,
+      "loss": 0.719,
+      "step": 500010
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 1.7658605647815105e-05,
+      "loss": 0.7649,
+      "step": 500020
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 1.765745771791536e-05,
+      "loss": 0.8069,
+      "step": 500030
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 1.765630978801561e-05,
+      "loss": 0.7554,
+      "step": 500040
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 1.7655161858115862e-05,
+      "loss": 0.7848,
+      "step": 500050
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 1.7654013928216115e-05,
+      "loss": 0.7559,
+      "step": 500060
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 1.7652865998316368e-05,
+      "loss": 0.8112,
+      "step": 500070
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 1.765171806841662e-05,
+      "loss": 0.7598,
+      "step": 500080
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 1.7650570138516875e-05,
+      "loss": 0.7763,
+      "step": 500090
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 1.7649422208617125e-05,
+      "loss": 0.8475,
+      "step": 500100
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 1.7648274278717378e-05,
+      "loss": 0.7571,
+      "step": 500110
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 1.764712634881763e-05,
+      "loss": 0.7925,
+      "step": 500120
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 1.764597841891788e-05,
+      "loss": 0.8742,
+      "step": 500130
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 1.7644830489018134e-05,
+      "loss": 0.8106,
+      "step": 500140
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 1.7643682559118387e-05,
+      "loss": 0.7833,
+      "step": 500150
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 1.764253462921864e-05,
+      "loss": 0.76,
+      "step": 500160
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 1.7641386699318894e-05,
+      "loss": 0.7874,
+      "step": 500170
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 1.7640238769419147e-05,
+      "loss": 0.8166,
+      "step": 500180
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 1.7639090839519397e-05,
+      "loss": 0.7381,
+      "step": 500190
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 1.763794290961965e-05,
+      "loss": 0.7798,
+      "step": 500200
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 1.7636794979719903e-05,
+      "loss": 0.7205,
+      "step": 500210
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 1.7635647049820157e-05,
+      "loss": 0.8231,
+      "step": 500220
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 1.763449911992041e-05,
+      "loss": 0.7637,
+      "step": 500230
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 1.7633351190020663e-05,
+      "loss": 0.7564,
+      "step": 500240
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 1.7632203260120913e-05,
+      "loss": 0.7477,
+      "step": 500250
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 1.7631055330221166e-05,
+      "loss": 0.8055,
+      "step": 500260
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 1.762990740032142e-05,
+      "loss": 0.729,
+      "step": 500270
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 1.762875947042167e-05,
+      "loss": 0.8368,
+      "step": 500280
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 1.7627611540521923e-05,
+      "loss": 0.7966,
+      "step": 500290
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 1.7626463610622176e-05,
+      "loss": 0.7479,
+      "step": 500300
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 1.762531568072243e-05,
+      "loss": 0.8458,
+      "step": 500310
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 1.7624167750822682e-05,
+      "loss": 0.7667,
+      "step": 500320
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 1.7623019820922932e-05,
+      "loss": 0.7909,
+      "step": 500330
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 1.7621871891023185e-05,
+      "loss": 0.7647,
+      "step": 500340
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 1.762072396112344e-05,
+      "loss": 0.7018,
+      "step": 500350
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 1.7619576031223692e-05,
+      "loss": 0.7696,
+      "step": 500360
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 1.7618428101323945e-05,
+      "loss": 0.7147,
+      "step": 500370
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 1.76172801714242e-05,
+      "loss": 0.7787,
+      "step": 500380
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 1.7616132241524448e-05,
+      "loss": 0.8097,
+      "step": 500390
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 1.76149843116247e-05,
+      "loss": 0.822,
+      "step": 500400
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 1.7613836381724955e-05,
+      "loss": 0.8656,
+      "step": 500410
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 1.7612688451825205e-05,
+      "loss": 0.783,
+      "step": 500420
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 1.7611540521925458e-05,
+      "loss": 0.7889,
+      "step": 500430
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 1.761039259202571e-05,
+      "loss": 0.7435,
+      "step": 500440
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 1.7609244662125964e-05,
+      "loss": 0.7633,
+      "step": 500450
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 1.7608096732226218e-05,
+      "loss": 0.7254,
+      "step": 500460
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 1.760694880232647e-05,
+      "loss": 0.8206,
+      "step": 500470
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 1.760580087242672e-05,
+      "loss": 0.8227,
+      "step": 500480
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 1.7604652942526974e-05,
+      "loss": 0.6907,
+      "step": 500490
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 1.7603505012627227e-05,
+      "loss": 0.7317,
+      "step": 500500
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 1.760235708272748e-05,
+      "loss": 0.7316,
+      "step": 500510
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 1.7601209152827734e-05,
+      "loss": 0.7957,
+      "step": 500520
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 1.7600061222927987e-05,
+      "loss": 0.7594,
+      "step": 500530
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 1.7598913293028237e-05,
+      "loss": 0.7404,
+      "step": 500540
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 1.759776536312849e-05,
+      "loss": 0.7116,
+      "step": 500550
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 1.7596617433228743e-05,
+      "loss": 0.8241,
+      "step": 500560
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 1.7595469503328993e-05,
+      "loss": 0.7771,
+      "step": 500570
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 1.7594321573429246e-05,
+      "loss": 0.8115,
+      "step": 500580
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 1.75931736435295e-05,
+      "loss": 0.6895,
+      "step": 500590
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 1.7592025713629753e-05,
+      "loss": 0.8089,
+      "step": 500600
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 1.7590877783730006e-05,
+      "loss": 0.6879,
+      "step": 500610
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 1.758972985383026e-05,
+      "loss": 0.7608,
+      "step": 500620
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 1.758858192393051e-05,
+      "loss": 0.8067,
+      "step": 500630
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 1.7587433994030762e-05,
+      "loss": 0.687,
+      "step": 500640
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 1.7586286064131016e-05,
+      "loss": 0.7331,
+      "step": 500650
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 1.758513813423127e-05,
+      "loss": 0.7829,
+      "step": 500660
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 1.7583990204331522e-05,
+      "loss": 0.7437,
+      "step": 500670
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 1.7582842274431772e-05,
+      "loss": 0.7813,
+      "step": 500680
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 1.7581694344532025e-05,
+      "loss": 0.689,
+      "step": 500690
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 1.758054641463228e-05,
+      "loss": 0.7757,
+      "step": 500700
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 1.7579398484732528e-05,
+      "loss": 0.7866,
+      "step": 500710
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 1.757825055483278e-05,
+      "loss": 0.7589,
+      "step": 500720
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 1.7577102624933035e-05,
+      "loss": 0.7752,
+      "step": 500730
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 1.7575954695033288e-05,
+      "loss": 0.7726,
+      "step": 500740
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 1.757480676513354e-05,
+      "loss": 0.8006,
+      "step": 500750
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 1.7573658835233794e-05,
+      "loss": 0.7247,
+      "step": 500760
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 1.7572510905334044e-05,
+      "loss": 0.774,
+      "step": 500770
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 1.7571362975434298e-05,
+      "loss": 0.7271,
+      "step": 500780
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 1.757021504553455e-05,
+      "loss": 0.8421,
+      "step": 500790
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 1.7569067115634804e-05,
+      "loss": 0.7084,
+      "step": 500800
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 1.7567919185735057e-05,
+      "loss": 0.7877,
+      "step": 500810
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 1.756677125583531e-05,
+      "loss": 0.7987,
+      "step": 500820
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 1.756562332593556e-05,
+      "loss": 0.7457,
+      "step": 500830
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 1.7564475396035814e-05,
+      "loss": 0.7894,
+      "step": 500840
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 1.7563327466136067e-05,
+      "loss": 0.8216,
+      "step": 500850
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 1.7562179536236317e-05,
+      "loss": 0.7769,
+      "step": 500860
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 1.756103160633657e-05,
+      "loss": 0.7399,
+      "step": 500870
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 1.7559883676436823e-05,
+      "loss": 0.7718,
+      "step": 500880
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 1.7558735746537076e-05,
+      "loss": 0.7953,
+      "step": 500890
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 1.755758781663733e-05,
+      "loss": 0.7125,
+      "step": 500900
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 1.7556439886737583e-05,
+      "loss": 0.7864,
+      "step": 500910
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 1.7555291956837833e-05,
+      "loss": 0.8245,
+      "step": 500920
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 1.7554144026938086e-05,
+      "loss": 0.8048,
+      "step": 500930
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 1.755299609703834e-05,
+      "loss": 0.7654,
+      "step": 500940
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 1.7551848167138593e-05,
+      "loss": 0.6827,
+      "step": 500950
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 1.7550700237238846e-05,
+      "loss": 0.8409,
+      "step": 500960
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 1.75495523073391e-05,
+      "loss": 0.7794,
+      "step": 500970
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 1.754840437743935e-05,
+      "loss": 0.7057,
+      "step": 500980
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 1.7547256447539602e-05,
+      "loss": 0.7551,
+      "step": 500990
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 1.7546108517639852e-05,
+      "loss": 0.7941,
+      "step": 501000
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 1.7544960587740105e-05,
+      "loss": 0.7473,
+      "step": 501010
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 1.754381265784036e-05,
+      "loss": 0.8084,
+      "step": 501020
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 1.7542664727940612e-05,
+      "loss": 0.7266,
+      "step": 501030
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 1.7541516798040865e-05,
+      "loss": 0.8383,
+      "step": 501040
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 1.7540368868141118e-05,
+      "loss": 0.73,
+      "step": 501050
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 1.7539220938241368e-05,
+      "loss": 0.757,
+      "step": 501060
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 1.753807300834162e-05,
+      "loss": 0.739,
+      "step": 501070
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 1.7536925078441875e-05,
+      "loss": 0.7507,
+      "step": 501080
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 1.7535777148542128e-05,
+      "loss": 0.794,
+      "step": 501090
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 1.753462921864238e-05,
+      "loss": 0.7569,
+      "step": 501100
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 1.7533481288742634e-05,
+      "loss": 0.7991,
+      "step": 501110
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 1.7532333358842884e-05,
+      "loss": 0.7456,
+      "step": 501120
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 1.7531185428943137e-05,
+      "loss": 0.7076,
+      "step": 501130
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 1.753003749904339e-05,
+      "loss": 0.8442,
+      "step": 501140
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 1.752888956914364e-05,
+      "loss": 0.727,
+      "step": 501150
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 1.7527741639243894e-05,
+      "loss": 0.7778,
+      "step": 501160
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 1.7526593709344147e-05,
+      "loss": 0.738,
+      "step": 501170
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 1.75254457794444e-05,
+      "loss": 0.7191,
+      "step": 501180
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 1.7524297849544653e-05,
+      "loss": 0.7556,
+      "step": 501190
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 1.7523149919644907e-05,
+      "loss": 0.7999,
+      "step": 501200
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 1.7522001989745157e-05,
+      "loss": 0.7629,
+      "step": 501210
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 1.752085405984541e-05,
+      "loss": 0.7817,
+      "step": 501220
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 1.7519706129945663e-05,
+      "loss": 0.7423,
+      "step": 501230
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 1.7518558200045916e-05,
+      "loss": 0.7693,
+      "step": 501240
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 1.751741027014617e-05,
+      "loss": 0.7481,
+      "step": 501250
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 1.7516262340246423e-05,
+      "loss": 0.7868,
+      "step": 501260
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 1.7515114410346673e-05,
+      "loss": 0.7492,
+      "step": 501270
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 1.7513966480446926e-05,
+      "loss": 0.7713,
+      "step": 501280
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 1.751281855054718e-05,
+      "loss": 0.6925,
+      "step": 501290
+    },
+    {
+      "epoch": 23.0,
+      "eval_cer": 0.668951120628692,
+      "eval_em": 0.007454017424975799,
+      "eval_f1": 0.007454017424975799,
+      "eval_loss": 1.328505516052246,
+      "eval_runtime": 1602.949,
+      "eval_samples_per_second": 6.444,
+      "eval_steps_per_second": 0.806,
+      "eval_wer": 0.9925459825750242,
+      "step": 501291
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 1.751167062064743e-05,
+      "loss": 0.7279,
+      "step": 501300
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 1.7510522690747682e-05,
+      "loss": 0.777,
+      "step": 501310
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 1.7509374760847935e-05,
+      "loss": 0.7765,
+      "step": 501320
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 1.750822683094819e-05,
+      "loss": 0.7117,
+      "step": 501330
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 1.7507078901048442e-05,
+      "loss": 0.6997,
+      "step": 501340
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 1.7505930971148692e-05,
+      "loss": 0.7278,
+      "step": 501350
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 1.7504783041248945e-05,
+      "loss": 0.7128,
+      "step": 501360
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 1.7503635111349198e-05,
+      "loss": 0.7256,
+      "step": 501370
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 1.750248718144945e-05,
+      "loss": 0.6483,
+      "step": 501380
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 1.7501339251549705e-05,
+      "loss": 0.8106,
+      "step": 501390
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.7500191321649958e-05,
+      "loss": 0.6985,
+      "step": 501400
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.7499043391750208e-05,
+      "loss": 0.7804,
+      "step": 501410
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.749789546185046e-05,
+      "loss": 0.6902,
+      "step": 501420
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.7496747531950714e-05,
+      "loss": 0.7353,
+      "step": 501430
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.7495599602050964e-05,
+      "loss": 0.698,
+      "step": 501440
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.7494451672151217e-05,
+      "loss": 0.686,
+      "step": 501450
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.749330374225147e-05,
+      "loss": 0.8109,
+      "step": 501460
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.7492155812351724e-05,
+      "loss": 0.7467,
+      "step": 501470
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.7491007882451977e-05,
+      "loss": 0.7866,
+      "step": 501480
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.748985995255223e-05,
+      "loss": 0.736,
+      "step": 501490
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.748871202265248e-05,
+      "loss": 0.6963,
+      "step": 501500
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.7487564092752733e-05,
+      "loss": 0.6626,
+      "step": 501510
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.7486416162852987e-05,
+      "loss": 0.6441,
+      "step": 501520
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.748526823295324e-05,
+      "loss": 0.7753,
+      "step": 501530
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.7484120303053493e-05,
+      "loss": 0.7019,
+      "step": 501540
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.7482972373153746e-05,
+      "loss": 0.7758,
+      "step": 501550
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.7481824443253996e-05,
+      "loss": 0.784,
+      "step": 501560
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.748067651335425e-05,
+      "loss": 0.7186,
+      "step": 501570
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.7479528583454503e-05,
+      "loss": 0.6993,
+      "step": 501580
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.7478380653554753e-05,
+      "loss": 0.7392,
+      "step": 501590
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.7477232723655006e-05,
+      "loss": 0.7152,
+      "step": 501600
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 1.747608479375526e-05,
+      "loss": 0.7754,
+      "step": 501610
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.7474936863855512e-05,
+      "loss": 0.7031,
+      "step": 501620
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.7473788933955766e-05,
+      "loss": 0.7387,
+      "step": 501630
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.747264100405602e-05,
+      "loss": 0.8116,
+      "step": 501640
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.747149307415627e-05,
+      "loss": 0.767,
+      "step": 501650
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.7470345144256522e-05,
+      "loss": 0.6916,
+      "step": 501660
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.7469197214356775e-05,
+      "loss": 0.669,
+      "step": 501670
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.746804928445703e-05,
+      "loss": 0.7332,
+      "step": 501680
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.746690135455728e-05,
+      "loss": 0.7346,
+      "step": 501690
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.746575342465753e-05,
+      "loss": 0.7032,
+      "step": 501700
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.7464605494757785e-05,
+      "loss": 0.7824,
+      "step": 501710
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.7463457564858038e-05,
+      "loss": 0.7192,
+      "step": 501720
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.7462309634958288e-05,
+      "loss": 0.7128,
+      "step": 501730
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.746116170505854e-05,
+      "loss": 0.7537,
+      "step": 501740
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.7460013775158794e-05,
+      "loss": 0.7471,
+      "step": 501750
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.7458865845259048e-05,
+      "loss": 0.7881,
+      "step": 501760
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.74577179153593e-05,
+      "loss": 0.6475,
+      "step": 501770
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.7456569985459554e-05,
+      "loss": 0.792,
+      "step": 501780
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.7455422055559804e-05,
+      "loss": 0.8076,
+      "step": 501790
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.7454274125660057e-05,
+      "loss": 0.7766,
+      "step": 501800
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.745312619576031e-05,
+      "loss": 0.6846,
+      "step": 501810
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.7451978265860564e-05,
+      "loss": 0.7143,
+      "step": 501820
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 1.7450830335960817e-05,
+      "loss": 0.7363,
+      "step": 501830
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.744968240606107e-05,
+      "loss": 0.7975,
+      "step": 501840
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.744853447616132e-05,
+      "loss": 0.7286,
+      "step": 501850
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.7447386546261573e-05,
+      "loss": 0.794,
+      "step": 501860
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.7446238616361826e-05,
+      "loss": 0.7284,
+      "step": 501870
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.7445090686462076e-05,
+      "loss": 0.7314,
+      "step": 501880
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.744394275656233e-05,
+      "loss": 0.7027,
+      "step": 501890
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.7442794826662583e-05,
+      "loss": 0.7222,
+      "step": 501900
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.7441646896762836e-05,
+      "loss": 0.7754,
+      "step": 501910
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.744049896686309e-05,
+      "loss": 0.6749,
+      "step": 501920
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.7439351036963343e-05,
+      "loss": 0.7644,
+      "step": 501930
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.7438203107063592e-05,
+      "loss": 0.7521,
+      "step": 501940
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.7437055177163846e-05,
+      "loss": 0.8401,
+      "step": 501950
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.74359072472641e-05,
+      "loss": 0.6418,
+      "step": 501960
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.7434759317364352e-05,
+      "loss": 0.8095,
+      "step": 501970
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.7433611387464605e-05,
+      "loss": 0.7452,
+      "step": 501980
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.743246345756486e-05,
+      "loss": 0.687,
+      "step": 501990
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.743131552766511e-05,
+      "loss": 0.7401,
+      "step": 502000
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.7430167597765362e-05,
+      "loss": 0.7811,
+      "step": 502010
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.742901966786561e-05,
+      "loss": 0.7886,
+      "step": 502020
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.7427871737965865e-05,
+      "loss": 0.7498,
+      "step": 502030
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.7426723808066118e-05,
+      "loss": 0.6856,
+      "step": 502040
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 1.742557587816637e-05,
+      "loss": 0.7119,
+      "step": 502050
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 1.7424427948266625e-05,
+      "loss": 0.6948,
+      "step": 502060
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 1.7423280018366878e-05,
+      "loss": 0.7752,
+      "step": 502070
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 1.7422132088467128e-05,
+      "loss": 0.8014,
+      "step": 502080
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 1.742098415856738e-05,
+      "loss": 0.7343,
+      "step": 502090
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 1.7419836228667634e-05,
+      "loss": 0.7679,
+      "step": 502100
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 1.7418688298767887e-05,
+      "loss": 0.8173,
+      "step": 502110
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 1.741754036886814e-05,
+      "loss": 0.8007,
+      "step": 502120
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 1.7416392438968394e-05,
+      "loss": 0.7897,
+      "step": 502130
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 1.7415244509068644e-05,
+      "loss": 0.7922,
+      "step": 502140
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 1.7414096579168897e-05,
+      "loss": 0.7972,
+      "step": 502150
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 1.741294864926915e-05,
+      "loss": 0.738,
+      "step": 502160
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 1.74118007193694e-05,
+      "loss": 0.8159,
+      "step": 502170
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 1.7410652789469653e-05,
+      "loss": 0.7076,
+      "step": 502180
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 1.7409504859569907e-05,
+      "loss": 0.7,
+      "step": 502190
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 1.740835692967016e-05,
+      "loss": 0.6855,
+      "step": 502200
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 1.7407208999770413e-05,
+      "loss": 0.7548,
+      "step": 502210
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 1.7406061069870666e-05,
+      "loss": 0.7695,
+      "step": 502220
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 1.7404913139970916e-05,
+      "loss": 0.8206,
+      "step": 502230
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 1.740376521007117e-05,
+      "loss": 0.7323,
+      "step": 502240
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 1.7402617280171423e-05,
+      "loss": 0.7529,
+      "step": 502250
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 1.7401469350271676e-05,
+      "loss": 0.7418,
+      "step": 502260
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 1.740032142037193e-05,
+      "loss": 0.7428,
+      "step": 502270
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 1.7399173490472182e-05,
+      "loss": 0.8127,
+      "step": 502280
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 1.7398025560572432e-05,
+      "loss": 0.7039,
+      "step": 502290
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 1.7396877630672685e-05,
+      "loss": 0.7173,
+      "step": 502300
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 1.739572970077294e-05,
+      "loss": 0.7606,
+      "step": 502310
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 1.739458177087319e-05,
+      "loss": 0.6923,
+      "step": 502320
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 1.7393433840973442e-05,
+      "loss": 0.7703,
+      "step": 502330
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 1.7392285911073695e-05,
+      "loss": 0.6922,
+      "step": 502340
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 1.7391137981173948e-05,
+      "loss": 0.7358,
+      "step": 502350
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 1.73899900512742e-05,
+      "loss": 0.666,
+      "step": 502360
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 1.738884212137445e-05,
+      "loss": 0.6871,
+      "step": 502370
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 1.7387694191474705e-05,
+      "loss": 0.7292,
+      "step": 502380
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 1.7386546261574958e-05,
+      "loss": 0.7239,
+      "step": 502390
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 1.738539833167521e-05,
+      "loss": 0.7366,
+      "step": 502400
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 1.7384250401775464e-05,
+      "loss": 0.7164,
+      "step": 502410
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 1.7383102471875718e-05,
+      "loss": 0.7491,
+      "step": 502420
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 1.7381954541975967e-05,
+      "loss": 0.7739,
+      "step": 502430
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 1.738080661207622e-05,
+      "loss": 0.7192,
+      "step": 502440
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 1.7379658682176474e-05,
+      "loss": 0.6753,
+      "step": 502450
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 1.7378510752276724e-05,
+      "loss": 0.7928,
+      "step": 502460
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 1.7377362822376977e-05,
+      "loss": 0.7827,
+      "step": 502470
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 1.737621489247723e-05,
+      "loss": 0.8243,
+      "step": 502480
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 1.7375066962577483e-05,
+      "loss": 0.7645,
+      "step": 502490
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 1.7373919032677737e-05,
+      "loss": 0.7063,
+      "step": 502500
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 1.737277110277799e-05,
+      "loss": 0.7306,
+      "step": 502510
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 1.737162317287824e-05,
+      "loss": 0.7873,
+      "step": 502520
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 1.7370475242978493e-05,
+      "loss": 0.7423,
+      "step": 502530
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 1.7369327313078746e-05,
+      "loss": 0.6482,
+      "step": 502540
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 1.7368179383179e-05,
+      "loss": 0.8011,
+      "step": 502550
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 1.7367031453279253e-05,
+      "loss": 0.7781,
+      "step": 502560
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 1.7365883523379506e-05,
+      "loss": 0.7317,
+      "step": 502570
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 1.7364735593479756e-05,
+      "loss": 0.7456,
+      "step": 502580
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 1.736358766358001e-05,
+      "loss": 0.7808,
+      "step": 502590
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 1.7362439733680262e-05,
+      "loss": 0.7927,
+      "step": 502600
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 1.7361291803780512e-05,
+      "loss": 0.8093,
+      "step": 502610
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 1.7360143873880765e-05,
+      "loss": 0.7427,
+      "step": 502620
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 1.735899594398102e-05,
+      "loss": 0.7256,
+      "step": 502630
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 1.7357848014081272e-05,
+      "loss": 0.7699,
+      "step": 502640
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 1.7356700084181525e-05,
+      "loss": 0.6916,
+      "step": 502650
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 1.735555215428178e-05,
+      "loss": 0.7797,
+      "step": 502660
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 1.7354404224382028e-05,
+      "loss": 0.7549,
+      "step": 502670
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 1.735325629448228e-05,
+      "loss": 0.7363,
+      "step": 502680
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 1.7352108364582535e-05,
+      "loss": 0.8135,
+      "step": 502690
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 1.7350960434682788e-05,
+      "loss": 0.7232,
+      "step": 502700
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 1.734981250478304e-05,
+      "loss": 0.8056,
+      "step": 502710
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 1.734866457488329e-05,
+      "loss": 0.7414,
+      "step": 502720
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 1.7347516644983544e-05,
+      "loss": 0.7451,
+      "step": 502730
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 1.7346368715083798e-05,
+      "loss": 0.7608,
+      "step": 502740
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 1.7345220785184047e-05,
+      "loss": 0.6782,
+      "step": 502750
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 1.73440728552843e-05,
+      "loss": 0.7527,
+      "step": 502760
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 1.7342924925384554e-05,
+      "loss": 0.7399,
+      "step": 502770
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 1.7341776995484807e-05,
+      "loss": 0.7371,
+      "step": 502780
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 1.734062906558506e-05,
+      "loss": 0.6769,
+      "step": 502790
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 1.7339481135685314e-05,
+      "loss": 0.7564,
+      "step": 502800
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 1.7338333205785563e-05,
+      "loss": 0.7307,
+      "step": 502810
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 1.7337185275885817e-05,
+      "loss": 0.6671,
+      "step": 502820
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 1.733603734598607e-05,
+      "loss": 0.726,
+      "step": 502830
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 1.7334889416086323e-05,
+      "loss": 0.6834,
+      "step": 502840
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 1.7333741486186576e-05,
+      "loss": 0.7288,
+      "step": 502850
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 1.733259355628683e-05,
+      "loss": 0.7255,
+      "step": 502860
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 1.733144562638708e-05,
+      "loss": 0.782,
+      "step": 502870
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 1.7330297696487333e-05,
+      "loss": 0.7377,
+      "step": 502880
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 1.7329149766587586e-05,
+      "loss": 0.7749,
+      "step": 502890
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 1.7328001836687836e-05,
+      "loss": 0.7643,
+      "step": 502900
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 1.732685390678809e-05,
+      "loss": 0.7803,
+      "step": 502910
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 1.7325705976888342e-05,
+      "loss": 0.6945,
+      "step": 502920
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 1.7324558046988596e-05,
+      "loss": 0.7366,
+      "step": 502930
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 1.732341011708885e-05,
+      "loss": 0.6859,
+      "step": 502940
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 1.7322262187189102e-05,
+      "loss": 0.7606,
+      "step": 502950
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 1.7321114257289352e-05,
+      "loss": 0.7638,
+      "step": 502960
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 1.7319966327389605e-05,
+      "loss": 0.7722,
+      "step": 502970
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 1.731881839748986e-05,
+      "loss": 0.7154,
+      "step": 502980
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 1.731767046759011e-05,
+      "loss": 0.7381,
+      "step": 502990
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 1.7316522537690365e-05,
+      "loss": 0.7574,
+      "step": 503000
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 1.7315374607790618e-05,
+      "loss": 0.7654,
+      "step": 503010
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 1.7314226677890868e-05,
+      "loss": 0.7651,
+      "step": 503020
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 1.731307874799112e-05,
+      "loss": 0.7464,
+      "step": 503030
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 1.731193081809137e-05,
+      "loss": 0.7687,
+      "step": 503040
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 1.7310782888191624e-05,
+      "loss": 0.7855,
+      "step": 503050
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 1.7309634958291878e-05,
+      "loss": 0.7759,
+      "step": 503060
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 1.730848702839213e-05,
+      "loss": 0.738,
+      "step": 503070
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 1.7307339098492384e-05,
+      "loss": 0.7274,
+      "step": 503080
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 1.7306191168592637e-05,
+      "loss": 0.7671,
+      "step": 503090
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 1.7305043238692887e-05,
+      "loss": 0.7652,
+      "step": 503100
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 1.730389530879314e-05,
+      "loss": 0.6729,
+      "step": 503110
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 1.7302747378893394e-05,
+      "loss": 0.7813,
+      "step": 503120
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 1.7301599448993647e-05,
+      "loss": 0.7081,
+      "step": 503130
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 1.73004515190939e-05,
+      "loss": 0.7077,
+      "step": 503140
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 1.7299303589194153e-05,
+      "loss": 0.7896,
+      "step": 503150
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 1.7298155659294403e-05,
+      "loss": 0.7858,
+      "step": 503160
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 1.7297007729394657e-05,
+      "loss": 0.787,
+      "step": 503170
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 1.729585979949491e-05,
+      "loss": 0.8145,
+      "step": 503180
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 1.729471186959516e-05,
+      "loss": 0.7317,
+      "step": 503190
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 1.7293563939695413e-05,
+      "loss": 0.7479,
+      "step": 503200
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 1.7292416009795666e-05,
+      "loss": 0.7628,
+      "step": 503210
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 1.729126807989592e-05,
+      "loss": 0.7903,
+      "step": 503220
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 1.7290120149996173e-05,
+      "loss": 0.7956,
+      "step": 503230
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 1.7288972220096426e-05,
+      "loss": 0.7737,
+      "step": 503240
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 1.7287824290196676e-05,
+      "loss": 0.7214,
+      "step": 503250
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 1.728667636029693e-05,
+      "loss": 0.7937,
+      "step": 503260
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 1.7285528430397182e-05,
+      "loss": 0.7839,
+      "step": 503270
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 1.7284380500497435e-05,
+      "loss": 0.8625,
+      "step": 503280
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 1.728323257059769e-05,
+      "loss": 0.7631,
+      "step": 503290
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 1.7282084640697942e-05,
+      "loss": 0.7534,
+      "step": 503300
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 1.7280936710798192e-05,
+      "loss": 0.6981,
+      "step": 503310
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 1.7279788780898445e-05,
+      "loss": 0.8136,
+      "step": 503320
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 1.7278640850998698e-05,
+      "loss": 0.7358,
+      "step": 503330
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 1.7277492921098948e-05,
+      "loss": 0.7778,
+      "step": 503340
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 1.72763449911992e-05,
+      "loss": 0.7696,
+      "step": 503350
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 1.7275197061299455e-05,
+      "loss": 0.7573,
+      "step": 503360
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 1.7274049131399708e-05,
+      "loss": 0.6526,
+      "step": 503370
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 1.727290120149996e-05,
+      "loss": 0.7839,
+      "step": 503380
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 1.727175327160021e-05,
+      "loss": 0.7646,
+      "step": 503390
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 1.7270605341700464e-05,
+      "loss": 0.6783,
+      "step": 503400
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 1.7269457411800717e-05,
+      "loss": 0.7608,
+      "step": 503410
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 1.726830948190097e-05,
+      "loss": 0.8081,
+      "step": 503420
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 1.7267161552001224e-05,
+      "loss": 0.7316,
+      "step": 503430
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 1.7266013622101477e-05,
+      "loss": 0.7484,
+      "step": 503440
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 1.7264865692201727e-05,
+      "loss": 0.6834,
+      "step": 503450
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 1.726371776230198e-05,
+      "loss": 0.7373,
+      "step": 503460
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 1.7262569832402233e-05,
+      "loss": 0.7784,
+      "step": 503470
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 1.7261421902502483e-05,
+      "loss": 0.756,
+      "step": 503480
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 1.7260273972602737e-05,
+      "loss": 0.7813,
+      "step": 503490
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 1.725912604270299e-05,
+      "loss": 0.7835,
+      "step": 503500
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 1.7257978112803243e-05,
+      "loss": 0.6962,
+      "step": 503510
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 1.7256830182903496e-05,
+      "loss": 0.735,
+      "step": 503520
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 1.725568225300375e-05,
+      "loss": 0.8027,
+      "step": 503530
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 1.7254534323104e-05,
+      "loss": 0.7161,
+      "step": 503540
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 1.7253386393204253e-05,
+      "loss": 0.7664,
+      "step": 503550
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 1.7252238463304506e-05,
+      "loss": 0.7864,
+      "step": 503560
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 1.725109053340476e-05,
+      "loss": 0.7615,
+      "step": 503570
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 1.7249942603505012e-05,
+      "loss": 0.7862,
+      "step": 503580
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 1.7248794673605266e-05,
+      "loss": 0.7688,
+      "step": 503590
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 1.7247646743705515e-05,
+      "loss": 0.8002,
+      "step": 503600
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 1.724649881380577e-05,
+      "loss": 0.6888,
+      "step": 503610
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 1.7245350883906022e-05,
+      "loss": 0.7104,
+      "step": 503620
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 1.7244202954006272e-05,
+      "loss": 0.754,
+      "step": 503630
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 1.7243055024106525e-05,
+      "loss": 0.7686,
+      "step": 503640
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 1.7241907094206778e-05,
+      "loss": 0.7665,
+      "step": 503650
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 1.724075916430703e-05,
+      "loss": 0.7624,
+      "step": 503660
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 1.7239611234407285e-05,
+      "loss": 0.7696,
+      "step": 503670
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 1.7238463304507538e-05,
+      "loss": 0.7482,
+      "step": 503680
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 1.7237315374607788e-05,
+      "loss": 0.7437,
+      "step": 503690
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 1.723616744470804e-05,
+      "loss": 0.748,
+      "step": 503700
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 1.7235019514808294e-05,
+      "loss": 0.7299,
+      "step": 503710
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 1.7233871584908548e-05,
+      "loss": 0.7121,
+      "step": 503720
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 1.72327236550088e-05,
+      "loss": 0.7745,
+      "step": 503730
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 1.723157572510905e-05,
+      "loss": 0.7838,
+      "step": 503740
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 1.7230427795209304e-05,
+      "loss": 0.7235,
+      "step": 503750
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 1.7229279865309557e-05,
+      "loss": 0.7457,
+      "step": 503760
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 1.7228131935409807e-05,
+      "loss": 0.7747,
+      "step": 503770
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 1.722698400551006e-05,
+      "loss": 0.7239,
+      "step": 503780
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 1.7225836075610313e-05,
+      "loss": 0.8072,
+      "step": 503790
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 1.7224688145710567e-05,
+      "loss": 0.7684,
+      "step": 503800
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 1.722354021581082e-05,
+      "loss": 0.6366,
+      "step": 503810
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 1.7222392285911073e-05,
+      "loss": 0.7364,
+      "step": 503820
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 1.7221244356011323e-05,
+      "loss": 0.7843,
+      "step": 503830
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 1.7220096426111576e-05,
+      "loss": 0.744,
+      "step": 503840
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 1.721894849621183e-05,
+      "loss": 0.7569,
+      "step": 503850
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 1.7217800566312083e-05,
+      "loss": 0.7755,
+      "step": 503860
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 1.7216652636412336e-05,
+      "loss": 0.7417,
+      "step": 503870
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 1.721550470651259e-05,
+      "loss": 0.7394,
+      "step": 503880
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 1.721435677661284e-05,
+      "loss": 0.711,
+      "step": 503890
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 1.7213208846713092e-05,
+      "loss": 0.7054,
+      "step": 503900
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 1.7212060916813346e-05,
+      "loss": 0.7462,
+      "step": 503910
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 1.7210912986913595e-05,
+      "loss": 0.757,
+      "step": 503920
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 1.720976505701385e-05,
+      "loss": 0.7696,
+      "step": 503930
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 1.7208617127114102e-05,
+      "loss": 0.7566,
+      "step": 503940
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 1.7207469197214355e-05,
+      "loss": 0.8093,
+      "step": 503950
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 1.720632126731461e-05,
+      "loss": 0.7014,
+      "step": 503960
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 1.720517333741486e-05,
+      "loss": 0.7382,
+      "step": 503970
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 1.720402540751511e-05,
+      "loss": 0.7461,
+      "step": 503980
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 1.7202877477615365e-05,
+      "loss": 0.7465,
+      "step": 503990
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 1.7201729547715618e-05,
+      "loss": 0.7244,
+      "step": 504000
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 1.720058161781587e-05,
+      "loss": 0.7946,
+      "step": 504010
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 1.7199433687916125e-05,
+      "loss": 0.6556,
+      "step": 504020
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 1.7198285758016378e-05,
+      "loss": 0.771,
+      "step": 504030
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 1.7197137828116628e-05,
+      "loss": 0.7726,
+      "step": 504040
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 1.719598989821688e-05,
+      "loss": 0.7588,
+      "step": 504050
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 1.7194841968317134e-05,
+      "loss": 0.771,
+      "step": 504060
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 1.7193694038417384e-05,
+      "loss": 0.7517,
+      "step": 504070
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 1.7192546108517637e-05,
+      "loss": 0.7457,
+      "step": 504080
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 1.719139817861789e-05,
+      "loss": 0.7711,
+      "step": 504090
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 1.7190250248718144e-05,
+      "loss": 0.7656,
+      "step": 504100
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 1.7189102318818397e-05,
+      "loss": 0.6546,
+      "step": 504110
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 1.7187954388918647e-05,
+      "loss": 0.7798,
+      "step": 504120
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 1.71868064590189e-05,
+      "loss": 0.7499,
+      "step": 504130
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 1.7185658529119153e-05,
+      "loss": 0.7267,
+      "step": 504140
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 1.7184510599219406e-05,
+      "loss": 0.7385,
+      "step": 504150
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 1.718336266931966e-05,
+      "loss": 0.8423,
+      "step": 504160
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 1.7182214739419913e-05,
+      "loss": 0.6854,
+      "step": 504170
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 1.7181066809520163e-05,
+      "loss": 0.746,
+      "step": 504180
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 1.7179918879620416e-05,
+      "loss": 0.7398,
+      "step": 504190
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 1.717877094972067e-05,
+      "loss": 0.7548,
+      "step": 504200
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 1.717762301982092e-05,
+      "loss": 0.7097,
+      "step": 504210
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 1.7176475089921172e-05,
+      "loss": 0.7131,
+      "step": 504220
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 1.7175327160021426e-05,
+      "loss": 0.7369,
+      "step": 504230
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 1.717417923012168e-05,
+      "loss": 0.8648,
+      "step": 504240
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 1.7173031300221932e-05,
+      "loss": 0.8076,
+      "step": 504250
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 1.7171883370322185e-05,
+      "loss": 0.7736,
+      "step": 504260
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 1.7170735440422435e-05,
+      "loss": 0.803,
+      "step": 504270
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 1.716958751052269e-05,
+      "loss": 0.7887,
+      "step": 504280
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 1.7168439580622942e-05,
+      "loss": 0.6837,
+      "step": 504290
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 1.7167291650723195e-05,
+      "loss": 0.7482,
+      "step": 504300
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 1.7166143720823448e-05,
+      "loss": 0.7122,
+      "step": 504310
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 1.71649957909237e-05,
+      "loss": 0.7012,
+      "step": 504320
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 1.716384786102395e-05,
+      "loss": 0.8386,
+      "step": 504330
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 1.7162699931124205e-05,
+      "loss": 0.7805,
+      "step": 504340
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 1.7161552001224458e-05,
+      "loss": 0.7011,
+      "step": 504350
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 1.7160404071324708e-05,
+      "loss": 0.7112,
+      "step": 504360
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 1.715925614142496e-05,
+      "loss": 0.7419,
+      "step": 504370
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 1.7158108211525214e-05,
+      "loss": 0.8085,
+      "step": 504380
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 1.7156960281625467e-05,
+      "loss": 0.6524,
+      "step": 504390
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 1.715581235172572e-05,
+      "loss": 0.7834,
+      "step": 504400
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 1.715466442182597e-05,
+      "loss": 0.7656,
+      "step": 504410
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 1.7153516491926224e-05,
+      "loss": 0.7662,
+      "step": 504420
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 1.7152368562026477e-05,
+      "loss": 0.8012,
+      "step": 504430
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 1.715122063212673e-05,
+      "loss": 0.7372,
+      "step": 504440
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 1.7150072702226983e-05,
+      "loss": 0.795,
+      "step": 504450
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 1.7148924772327237e-05,
+      "loss": 0.6877,
+      "step": 504460
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 1.7147776842427487e-05,
+      "loss": 0.7178,
+      "step": 504470
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 1.714662891252774e-05,
+      "loss": 0.7577,
+      "step": 504480
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 1.7145480982627993e-05,
+      "loss": 0.774,
+      "step": 504490
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 1.7144333052728243e-05,
+      "loss": 0.7448,
+      "step": 504500
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 1.7143185122828496e-05,
+      "loss": 0.7849,
+      "step": 504510
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 1.714203719292875e-05,
+      "loss": 0.8331,
+      "step": 504520
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 1.7140889263029003e-05,
+      "loss": 0.6964,
+      "step": 504530
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 1.7139741333129256e-05,
+      "loss": 0.7676,
+      "step": 504540
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 1.713859340322951e-05,
+      "loss": 0.8575,
+      "step": 504550
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 1.713744547332976e-05,
+      "loss": 0.653,
+      "step": 504560
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 1.7136297543430012e-05,
+      "loss": 0.7795,
+      "step": 504570
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 1.7135149613530265e-05,
+      "loss": 0.7522,
+      "step": 504580
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 1.713400168363052e-05,
+      "loss": 0.7929,
+      "step": 504590
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 1.7132853753730772e-05,
+      "loss": 0.7685,
+      "step": 504600
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 1.7131705823831025e-05,
+      "loss": 0.743,
+      "step": 504610
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 1.7130557893931275e-05,
+      "loss": 0.7442,
+      "step": 504620
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 1.7129409964031528e-05,
+      "loss": 0.6787,
+      "step": 504630
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 1.712826203413178e-05,
+      "loss": 0.7416,
+      "step": 504640
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 1.712711410423203e-05,
+      "loss": 0.7223,
+      "step": 504650
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 1.7125966174332285e-05,
+      "loss": 0.7713,
+      "step": 504660
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 1.7124818244432538e-05,
+      "loss": 0.7285,
+      "step": 504670
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 1.712367031453279e-05,
+      "loss": 0.712,
+      "step": 504680
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 1.7122522384633044e-05,
+      "loss": 0.7163,
+      "step": 504690
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 1.7121374454733298e-05,
+      "loss": 0.7619,
+      "step": 504700
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 1.7120226524833547e-05,
+      "loss": 0.6858,
+      "step": 504710
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 1.71190785949338e-05,
+      "loss": 0.7383,
+      "step": 504720
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 1.7117930665034054e-05,
+      "loss": 0.7741,
+      "step": 504730
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 1.7116782735134307e-05,
+      "loss": 0.7192,
+      "step": 504740
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 1.711563480523456e-05,
+      "loss": 0.7322,
+      "step": 504750
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 1.711448687533481e-05,
+      "loss": 0.7742,
+      "step": 504760
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 1.7113338945435063e-05,
+      "loss": 0.8051,
+      "step": 504770
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 1.7112191015535317e-05,
+      "loss": 0.6768,
+      "step": 504780
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 1.7111043085635567e-05,
+      "loss": 0.6881,
+      "step": 504790
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 1.710989515573582e-05,
+      "loss": 0.8192,
+      "step": 504800
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 1.7108747225836073e-05,
+      "loss": 0.8034,
+      "step": 504810
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 1.7107599295936326e-05,
+      "loss": 0.7048,
+      "step": 504820
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 1.710645136603658e-05,
+      "loss": 0.7104,
+      "step": 504830
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 1.7105303436136833e-05,
+      "loss": 0.7727,
+      "step": 504840
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 1.7104155506237083e-05,
+      "loss": 0.6661,
+      "step": 504850
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 1.7103007576337336e-05,
+      "loss": 0.8455,
+      "step": 504860
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 1.710185964643759e-05,
+      "loss": 0.7144,
+      "step": 504870
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 1.7100711716537842e-05,
+      "loss": 0.7885,
+      "step": 504880
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 1.7099563786638096e-05,
+      "loss": 0.7227,
+      "step": 504890
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 1.709841585673835e-05,
+      "loss": 0.644,
+      "step": 504900
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 1.70972679268386e-05,
+      "loss": 0.7889,
+      "step": 504910
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 1.7096119996938852e-05,
+      "loss": 0.7063,
+      "step": 504920
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 1.7094972067039105e-05,
+      "loss": 0.7607,
+      "step": 504930
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 1.7093824137139355e-05,
+      "loss": 0.7624,
+      "step": 504940
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 1.7092676207239608e-05,
+      "loss": 0.6444,
+      "step": 504950
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 1.709152827733986e-05,
+      "loss": 0.7107,
+      "step": 504960
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 1.7090380347440115e-05,
+      "loss": 0.8072,
+      "step": 504970
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 1.7089232417540368e-05,
+      "loss": 0.6926,
+      "step": 504980
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 1.708808448764062e-05,
+      "loss": 0.7428,
+      "step": 504990
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 1.708693655774087e-05,
+      "loss": 0.8146,
+      "step": 505000
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 1.7085788627841124e-05,
+      "loss": 0.7498,
+      "step": 505010
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 1.7084640697941378e-05,
+      "loss": 0.7158,
+      "step": 505020
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 1.708349276804163e-05,
+      "loss": 0.8278,
+      "step": 505030
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 1.7082344838141884e-05,
+      "loss": 0.7774,
+      "step": 505040
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 1.7081196908242137e-05,
+      "loss": 0.7983,
+      "step": 505050
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 1.7080048978342387e-05,
+      "loss": 0.7187,
+      "step": 505060
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 1.707890104844264e-05,
+      "loss": 0.7591,
+      "step": 505070
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 1.7077753118542894e-05,
+      "loss": 0.7215,
+      "step": 505080
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 1.7076605188643144e-05,
+      "loss": 0.8297,
+      "step": 505090
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 1.7075457258743397e-05,
+      "loss": 0.6591,
+      "step": 505100
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 1.707430932884365e-05,
+      "loss": 0.6509,
+      "step": 505110
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 1.7073161398943903e-05,
+      "loss": 0.6663,
+      "step": 505120
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 1.7072013469044156e-05,
+      "loss": 0.8114,
+      "step": 505130
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 1.7070865539144406e-05,
+      "loss": 0.8475,
+      "step": 505140
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 1.706971760924466e-05,
+      "loss": 0.7174,
+      "step": 505150
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 1.7068569679344913e-05,
+      "loss": 0.7517,
+      "step": 505160
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 1.7067421749445166e-05,
+      "loss": 0.7668,
+      "step": 505170
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 1.706627381954542e-05,
+      "loss": 0.7492,
+      "step": 505180
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 1.7065125889645673e-05,
+      "loss": 0.7288,
+      "step": 505190
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 1.7063977959745922e-05,
+      "loss": 0.7202,
+      "step": 505200
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 1.7062830029846176e-05,
+      "loss": 0.7503,
+      "step": 505210
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 1.706168209994643e-05,
+      "loss": 0.7072,
+      "step": 505220
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 1.706053417004668e-05,
+      "loss": 0.8337,
+      "step": 505230
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 1.7059386240146932e-05,
+      "loss": 0.8052,
+      "step": 505240
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 1.7058238310247185e-05,
+      "loss": 0.7882,
+      "step": 505250
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 1.705709038034744e-05,
+      "loss": 0.7241,
+      "step": 505260
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 1.7055942450447692e-05,
+      "loss": 0.757,
+      "step": 505270
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 1.7054794520547945e-05,
+      "loss": 0.7145,
+      "step": 505280
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 1.7053646590648195e-05,
+      "loss": 0.7387,
+      "step": 505290
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 1.7052498660748448e-05,
+      "loss": 0.6803,
+      "step": 505300
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 1.70513507308487e-05,
+      "loss": 0.7966,
+      "step": 505310
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 1.7050202800948955e-05,
+      "loss": 0.8496,
+      "step": 505320
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 1.7049054871049208e-05,
+      "loss": 0.6602,
+      "step": 505330
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 1.704790694114946e-05,
+      "loss": 0.7256,
+      "step": 505340
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 1.704675901124971e-05,
+      "loss": 0.7909,
+      "step": 505350
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 1.7045611081349964e-05,
+      "loss": 0.8066,
+      "step": 505360
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 1.7044463151450217e-05,
+      "loss": 0.7576,
+      "step": 505370
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 1.7043315221550467e-05,
+      "loss": 0.7747,
+      "step": 505380
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 1.704216729165072e-05,
+      "loss": 0.6946,
+      "step": 505390
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 1.7041019361750974e-05,
+      "loss": 0.7544,
+      "step": 505400
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 1.7039871431851227e-05,
+      "loss": 0.8498,
+      "step": 505410
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 1.703872350195148e-05,
+      "loss": 0.8376,
+      "step": 505420
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 1.7037575572051733e-05,
+      "loss": 0.7769,
+      "step": 505430
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 1.7036427642151983e-05,
+      "loss": 0.8231,
+      "step": 505440
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 1.7035279712252237e-05,
+      "loss": 0.7398,
+      "step": 505450
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 1.703413178235249e-05,
+      "loss": 0.7991,
+      "step": 505460
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 1.7032983852452743e-05,
+      "loss": 0.7223,
+      "step": 505470
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 1.7031835922552996e-05,
+      "loss": 0.7609,
+      "step": 505480
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 1.7030687992653246e-05,
+      "loss": 0.7904,
+      "step": 505490
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 1.70295400627535e-05,
+      "loss": 0.8106,
+      "step": 505500
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 1.7028392132853753e-05,
+      "loss": 0.8023,
+      "step": 505510
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 1.7027244202954002e-05,
+      "loss": 0.7572,
+      "step": 505520
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 1.7026096273054256e-05,
+      "loss": 0.7525,
+      "step": 505530
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 1.702494834315451e-05,
+      "loss": 0.7103,
+      "step": 505540
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 1.7023800413254762e-05,
+      "loss": 0.7541,
+      "step": 505550
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 1.7022652483355015e-05,
+      "loss": 0.7895,
+      "step": 505560
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 1.702150455345527e-05,
+      "loss": 0.7328,
+      "step": 505570
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 1.702035662355552e-05,
+      "loss": 0.7564,
+      "step": 505580
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 1.7019208693655772e-05,
+      "loss": 0.7096,
+      "step": 505590
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 1.7018060763756025e-05,
+      "loss": 0.825,
+      "step": 505600
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 1.7016912833856278e-05,
+      "loss": 0.8017,
+      "step": 505610
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 1.701576490395653e-05,
+      "loss": 0.7112,
+      "step": 505620
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 1.7014616974056785e-05,
+      "loss": 0.7858,
+      "step": 505630
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 1.7013469044157035e-05,
+      "loss": 0.6796,
+      "step": 505640
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 1.7012321114257288e-05,
+      "loss": 0.7815,
+      "step": 505650
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 1.701117318435754e-05,
+      "loss": 0.7642,
+      "step": 505660
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 1.701002525445779e-05,
+      "loss": 0.7166,
+      "step": 505670
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 1.7008877324558044e-05,
+      "loss": 0.87,
+      "step": 505680
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 1.7007729394658297e-05,
+      "loss": 0.7708,
+      "step": 505690
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 1.700658146475855e-05,
+      "loss": 0.7573,
+      "step": 505700
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 1.7005433534858804e-05,
+      "loss": 0.8092,
+      "step": 505710
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 1.7004285604959057e-05,
+      "loss": 0.8029,
+      "step": 505720
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 1.7003137675059307e-05,
+      "loss": 0.8132,
+      "step": 505730
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 1.700198974515956e-05,
+      "loss": 0.7168,
+      "step": 505740
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 1.7000841815259813e-05,
+      "loss": 0.8351,
+      "step": 505750
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 1.6999693885360067e-05,
+      "loss": 0.7646,
+      "step": 505760
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 1.699854595546032e-05,
+      "loss": 0.807,
+      "step": 505770
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 1.6997398025560573e-05,
+      "loss": 0.7299,
+      "step": 505780
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 1.6996250095660823e-05,
+      "loss": 0.8102,
+      "step": 505790
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 1.6995102165761076e-05,
+      "loss": 0.7664,
+      "step": 505800
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 1.6993954235861326e-05,
+      "loss": 0.7601,
+      "step": 505810
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 1.699280630596158e-05,
+      "loss": 0.7393,
+      "step": 505820
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 1.6991658376061833e-05,
+      "loss": 0.7933,
+      "step": 505830
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 1.6990510446162086e-05,
+      "loss": 0.7245,
+      "step": 505840
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 1.698936251626234e-05,
+      "loss": 0.7338,
+      "step": 505850
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 1.6988214586362592e-05,
+      "loss": 0.7852,
+      "step": 505860
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 1.6987066656462842e-05,
+      "loss": 0.7361,
+      "step": 505870
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 1.6985918726563095e-05,
+      "loss": 0.6831,
+      "step": 505880
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 1.698477079666335e-05,
+      "loss": 0.7242,
+      "step": 505890
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 1.6983622866763602e-05,
+      "loss": 0.734,
+      "step": 505900
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 1.6982474936863855e-05,
+      "loss": 0.7639,
+      "step": 505910
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 1.698132700696411e-05,
+      "loss": 0.7772,
+      "step": 505920
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 1.6980179077064358e-05,
+      "loss": 0.7735,
+      "step": 505930
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 1.697903114716461e-05,
+      "loss": 0.7437,
+      "step": 505940
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 1.6977883217264865e-05,
+      "loss": 0.748,
+      "step": 505950
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 1.6976735287365115e-05,
+      "loss": 0.7428,
+      "step": 505960
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 1.6975587357465368e-05,
+      "loss": 0.7489,
+      "step": 505970
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 1.697443942756562e-05,
+      "loss": 0.7165,
+      "step": 505980
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 1.6973291497665874e-05,
+      "loss": 0.877,
+      "step": 505990
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 1.6972143567766128e-05,
+      "loss": 0.6974,
+      "step": 506000
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 1.697099563786638e-05,
+      "loss": 0.763,
+      "step": 506010
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 1.696984770796663e-05,
+      "loss": 0.7072,
+      "step": 506020
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 1.6968699778066884e-05,
+      "loss": 0.7694,
+      "step": 506030
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 1.6967551848167137e-05,
+      "loss": 0.7798,
+      "step": 506040
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 1.696640391826739e-05,
+      "loss": 0.7313,
+      "step": 506050
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 1.6965255988367644e-05,
+      "loss": 0.7078,
+      "step": 506060
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 1.6964108058467897e-05,
+      "loss": 0.7969,
+      "step": 506070
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 1.6962960128568147e-05,
+      "loss": 0.7538,
+      "step": 506080
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 1.69618121986684e-05,
+      "loss": 0.7513,
+      "step": 506090
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 1.6960664268768653e-05,
+      "loss": 0.7054,
+      "step": 506100
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 1.6959516338868903e-05,
+      "loss": 0.7277,
+      "step": 506110
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 1.6958368408969156e-05,
+      "loss": 0.8349,
+      "step": 506120
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 1.695722047906941e-05,
+      "loss": 0.7955,
+      "step": 506130
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 1.6956072549169663e-05,
+      "loss": 0.7211,
+      "step": 506140
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 1.6954924619269916e-05,
+      "loss": 0.7628,
+      "step": 506150
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 1.6953776689370166e-05,
+      "loss": 0.7504,
+      "step": 506160
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 1.695262875947042e-05,
+      "loss": 0.7396,
+      "step": 506170
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 1.6951480829570672e-05,
+      "loss": 0.741,
+      "step": 506180
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 1.6950332899670926e-05,
+      "loss": 0.7386,
+      "step": 506190
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 1.694918496977118e-05,
+      "loss": 0.6791,
+      "step": 506200
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 1.6948037039871432e-05,
+      "loss": 0.8528,
+      "step": 506210
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 1.6946889109971682e-05,
+      "loss": 0.7234,
+      "step": 506220
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 1.6945741180071935e-05,
+      "loss": 0.7252,
+      "step": 506230
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 1.694459325017219e-05,
+      "loss": 0.756,
+      "step": 506240
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 1.694344532027244e-05,
+      "loss": 0.709,
+      "step": 506250
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 1.694229739037269e-05,
+      "loss": 0.8262,
+      "step": 506260
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 1.6941149460472945e-05,
+      "loss": 0.7248,
+      "step": 506270
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 1.6940001530573198e-05,
+      "loss": 0.6893,
+      "step": 506280
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 1.693885360067345e-05,
+      "loss": 0.6734,
+      "step": 506290
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 1.6937705670773705e-05,
+      "loss": 0.7267,
+      "step": 506300
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 1.6936557740873954e-05,
+      "loss": 0.7826,
+      "step": 506310
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 1.6935409810974208e-05,
+      "loss": 0.6562,
+      "step": 506320
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 1.693426188107446e-05,
+      "loss": 0.7119,
+      "step": 506330
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 1.6933113951174714e-05,
+      "loss": 0.773,
+      "step": 506340
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 1.6931966021274967e-05,
+      "loss": 0.7531,
+      "step": 506350
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 1.693081809137522e-05,
+      "loss": 0.7723,
+      "step": 506360
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 1.692967016147547e-05,
+      "loss": 0.7964,
+      "step": 506370
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 1.6928522231575724e-05,
+      "loss": 0.7022,
+      "step": 506380
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 1.6927374301675977e-05,
+      "loss": 0.73,
+      "step": 506390
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 1.6926226371776227e-05,
+      "loss": 0.8395,
+      "step": 506400
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 1.692507844187648e-05,
+      "loss": 0.7501,
+      "step": 506410
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 1.6923930511976733e-05,
+      "loss": 0.8376,
+      "step": 506420
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 1.6922782582076987e-05,
+      "loss": 0.7557,
+      "step": 506430
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 1.692163465217724e-05,
+      "loss": 0.6898,
+      "step": 506440
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 1.6920486722277493e-05,
+      "loss": 0.7904,
+      "step": 506450
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 1.6919338792377743e-05,
+      "loss": 0.7577,
+      "step": 506460
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 1.6918190862477996e-05,
+      "loss": 0.7104,
+      "step": 506470
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 1.691704293257825e-05,
+      "loss": 0.7612,
+      "step": 506480
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 1.6915895002678503e-05,
+      "loss": 0.7399,
+      "step": 506490
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 1.6914747072778756e-05,
+      "loss": 0.783,
+      "step": 506500
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 1.6913599142879006e-05,
+      "loss": 0.7138,
+      "step": 506510
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 1.691245121297926e-05,
+      "loss": 0.7275,
+      "step": 506520
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 1.6911303283079512e-05,
+      "loss": 0.6836,
+      "step": 506530
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 1.6910155353179762e-05,
+      "loss": 0.7098,
+      "step": 506540
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 1.6909007423280015e-05,
+      "loss": 0.752,
+      "step": 506550
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 1.690785949338027e-05,
+      "loss": 0.7126,
+      "step": 506560
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 1.6906711563480522e-05,
+      "loss": 0.6588,
+      "step": 506570
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 1.6905563633580775e-05,
+      "loss": 0.6931,
+      "step": 506580
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 1.6904415703681028e-05,
+      "loss": 0.7346,
+      "step": 506590
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 1.6903267773781278e-05,
+      "loss": 0.7685,
+      "step": 506600
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 1.690211984388153e-05,
+      "loss": 0.7051,
+      "step": 506610
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 1.6900971913981785e-05,
+      "loss": 0.749,
+      "step": 506620
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 1.6899823984082038e-05,
+      "loss": 0.7931,
+      "step": 506630
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 1.689867605418229e-05,
+      "loss": 0.7418,
+      "step": 506640
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 1.6897528124282544e-05,
+      "loss": 0.7759,
+      "step": 506650
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 1.6896380194382794e-05,
+      "loss": 0.713,
+      "step": 506660
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 1.6895232264483047e-05,
+      "loss": 0.6959,
+      "step": 506670
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 1.68940843345833e-05,
+      "loss": 0.6871,
+      "step": 506680
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 1.689293640468355e-05,
+      "loss": 0.7271,
+      "step": 506690
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 1.6891788474783804e-05,
+      "loss": 0.7838,
+      "step": 506700
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 1.6890640544884057e-05,
+      "loss": 0.8067,
+      "step": 506710
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 1.688949261498431e-05,
+      "loss": 0.6986,
+      "step": 506720
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 1.6888344685084563e-05,
+      "loss": 0.7765,
+      "step": 506730
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 1.6887196755184817e-05,
+      "loss": 0.7654,
+      "step": 506740
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 1.6886048825285067e-05,
+      "loss": 0.8327,
+      "step": 506750
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 1.688490089538532e-05,
+      "loss": 0.7196,
+      "step": 506760
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 1.6883752965485573e-05,
+      "loss": 0.7378,
+      "step": 506770
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 1.6882605035585826e-05,
+      "loss": 0.7325,
+      "step": 506780
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 1.688145710568608e-05,
+      "loss": 0.7451,
+      "step": 506790
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 1.6880309175786333e-05,
+      "loss": 0.7989,
+      "step": 506800
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 1.6879161245886583e-05,
+      "loss": 0.7209,
+      "step": 506810
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 1.6878013315986836e-05,
+      "loss": 0.7475,
+      "step": 506820
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 1.6876865386087086e-05,
+      "loss": 0.7938,
+      "step": 506830
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 1.687571745618734e-05,
+      "loss": 0.7409,
+      "step": 506840
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 1.6874569526287592e-05,
+      "loss": 0.6926,
+      "step": 506850
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 1.6873421596387845e-05,
+      "loss": 0.7209,
+      "step": 506860
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 1.68722736664881e-05,
+      "loss": 0.7358,
+      "step": 506870
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 1.6871125736588352e-05,
+      "loss": 0.799,
+      "step": 506880
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 1.6869977806688602e-05,
+      "loss": 0.6594,
+      "step": 506890
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 1.6868829876788855e-05,
+      "loss": 0.7943,
+      "step": 506900
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 1.6867681946889108e-05,
+      "loss": 0.7078,
+      "step": 506910
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 1.686653401698936e-05,
+      "loss": 0.7822,
+      "step": 506920
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 1.6865386087089615e-05,
+      "loss": 0.7901,
+      "step": 506930
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 1.6864238157189868e-05,
+      "loss": 0.732,
+      "step": 506940
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 1.6863090227290118e-05,
+      "loss": 0.705,
+      "step": 506950
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 1.686194229739037e-05,
+      "loss": 0.8206,
+      "step": 506960
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 1.6860794367490624e-05,
+      "loss": 0.7947,
+      "step": 506970
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 1.6859646437590874e-05,
+      "loss": 0.7218,
+      "step": 506980
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 1.6858498507691127e-05,
+      "loss": 0.729,
+      "step": 506990
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 1.685735057779138e-05,
+      "loss": 0.7559,
+      "step": 507000
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 1.6856202647891634e-05,
+      "loss": 0.8163,
+      "step": 507010
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 1.6855054717991887e-05,
+      "loss": 0.7676,
+      "step": 507020
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 1.685390678809214e-05,
+      "loss": 0.7547,
+      "step": 507030
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 1.685275885819239e-05,
+      "loss": 0.7271,
+      "step": 507040
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 1.6851610928292644e-05,
+      "loss": 0.7167,
+      "step": 507050
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 1.6850462998392897e-05,
+      "loss": 0.7407,
+      "step": 507060
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 1.684931506849315e-05,
+      "loss": 0.7977,
+      "step": 507070
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 1.6848167138593403e-05,
+      "loss": 0.6297,
+      "step": 507080
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 1.6847019208693656e-05,
+      "loss": 0.7875,
+      "step": 507090
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 1.6845871278793906e-05,
+      "loss": 0.8599,
+      "step": 507100
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 1.684472334889416e-05,
+      "loss": 0.7326,
+      "step": 507110
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 1.6843575418994413e-05,
+      "loss": 0.7962,
+      "step": 507120
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 1.6842427489094663e-05,
+      "loss": 0.6746,
+      "step": 507130
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 1.6841279559194916e-05,
+      "loss": 0.7883,
+      "step": 507140
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 1.684013162929517e-05,
+      "loss": 0.7418,
+      "step": 507150
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 1.6838983699395422e-05,
+      "loss": 0.7083,
+      "step": 507160
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 1.6837835769495676e-05,
+      "loss": 0.8494,
+      "step": 507170
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 1.6836687839595926e-05,
+      "loss": 0.7404,
+      "step": 507180
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 1.683553990969618e-05,
+      "loss": 0.7762,
+      "step": 507190
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 1.6834391979796432e-05,
+      "loss": 0.773,
+      "step": 507200
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 1.6833244049896685e-05,
+      "loss": 0.7425,
+      "step": 507210
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 1.683209611999694e-05,
+      "loss": 0.804,
+      "step": 507220
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 1.6830948190097192e-05,
+      "loss": 0.7759,
+      "step": 507230
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 1.682980026019744e-05,
+      "loss": 0.6573,
+      "step": 507240
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 1.6828652330297695e-05,
+      "loss": 0.7504,
+      "step": 507250
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 1.6827504400397948e-05,
+      "loss": 0.7399,
+      "step": 507260
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 1.6826356470498198e-05,
+      "loss": 0.7424,
+      "step": 507270
+    },
+    {
+      "epoch": 23.27,
+      "learning_rate": 1.682520854059845e-05,
+      "loss": 0.7807,
+      "step": 507280
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 1.6824060610698704e-05,
+      "loss": 0.7843,
+      "step": 507290
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 1.6822912680798958e-05,
+      "loss": 0.8028,
+      "step": 507300
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 1.682176475089921e-05,
+      "loss": 0.7498,
+      "step": 507310
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 1.6820616820999464e-05,
+      "loss": 0.81,
+      "step": 507320
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 1.6819468891099714e-05,
+      "loss": 0.6987,
+      "step": 507330
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 1.6818320961199967e-05,
+      "loss": 0.7257,
+      "step": 507340
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 1.681717303130022e-05,
+      "loss": 0.719,
+      "step": 507350
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 1.6816025101400474e-05,
+      "loss": 0.6944,
+      "step": 507360
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 1.6814877171500727e-05,
+      "loss": 0.7953,
+      "step": 507370
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 1.681372924160098e-05,
+      "loss": 0.7867,
+      "step": 507380
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 1.681258131170123e-05,
+      "loss": 0.7588,
+      "step": 507390
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 1.6811433381801483e-05,
+      "loss": 0.7192,
+      "step": 507400
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 1.6810285451901737e-05,
+      "loss": 0.6346,
+      "step": 507410
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 1.6809137522001986e-05,
+      "loss": 0.7571,
+      "step": 507420
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 1.680798959210224e-05,
+      "loss": 0.7733,
+      "step": 507430
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 1.6806841662202493e-05,
+      "loss": 0.7503,
+      "step": 507440
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 1.6805693732302746e-05,
+      "loss": 0.7224,
+      "step": 507450
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 1.6804545802403e-05,
+      "loss": 0.7477,
+      "step": 507460
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 1.6803397872503253e-05,
+      "loss": 0.7823,
+      "step": 507470
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 1.6802249942603502e-05,
+      "loss": 0.7356,
+      "step": 507480
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 1.6801102012703756e-05,
+      "loss": 0.8118,
+      "step": 507490
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 1.679995408280401e-05,
+      "loss": 0.7634,
+      "step": 507500
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 1.6798806152904262e-05,
+      "loss": 0.6671,
+      "step": 507510
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 1.6797658223004515e-05,
+      "loss": 0.7225,
+      "step": 507520
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 1.6796510293104765e-05,
+      "loss": 0.772,
+      "step": 507530
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 1.679536236320502e-05,
+      "loss": 0.7641,
+      "step": 507540
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 1.6794214433305272e-05,
+      "loss": 0.7052,
+      "step": 507550
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 1.679306650340552e-05,
+      "loss": 0.7385,
+      "step": 507560
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 1.6791918573505775e-05,
+      "loss": 0.7286,
+      "step": 507570
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 1.6790770643606028e-05,
+      "loss": 0.8479,
+      "step": 507580
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 1.678962271370628e-05,
+      "loss": 0.7515,
+      "step": 507590
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 1.6788474783806535e-05,
+      "loss": 0.7987,
+      "step": 507600
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 1.6787326853906788e-05,
+      "loss": 0.7584,
+      "step": 507610
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 1.6786178924007038e-05,
+      "loss": 0.6306,
+      "step": 507620
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 1.678503099410729e-05,
+      "loss": 0.7429,
+      "step": 507630
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 1.6783883064207544e-05,
+      "loss": 0.6969,
+      "step": 507640
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 1.6782735134307797e-05,
+      "loss": 0.7839,
+      "step": 507650
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 1.678158720440805e-05,
+      "loss": 0.7617,
+      "step": 507660
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 1.6780439274508304e-05,
+      "loss": 0.7587,
+      "step": 507670
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 1.6779291344608554e-05,
+      "loss": 0.6908,
+      "step": 507680
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 1.6778143414708807e-05,
+      "loss": 0.8186,
+      "step": 507690
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 1.677699548480906e-05,
+      "loss": 0.7237,
+      "step": 507700
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 1.677584755490931e-05,
+      "loss": 0.8451,
+      "step": 507710
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 1.6774699625009563e-05,
+      "loss": 0.7695,
+      "step": 507720
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 1.6773551695109817e-05,
+      "loss": 0.793,
+      "step": 507730
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 1.677240376521007e-05,
+      "loss": 0.7207,
+      "step": 507740
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 1.6771255835310323e-05,
+      "loss": 0.7936,
+      "step": 507750
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 1.6770107905410576e-05,
+      "loss": 0.6991,
+      "step": 507760
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 1.6768959975510826e-05,
+      "loss": 0.8073,
+      "step": 507770
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 1.676781204561108e-05,
+      "loss": 0.7594,
+      "step": 507780
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 1.6766664115711333e-05,
+      "loss": 0.7894,
+      "step": 507790
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 1.6765516185811586e-05,
+      "loss": 0.7669,
+      "step": 507800
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 1.676436825591184e-05,
+      "loss": 0.7748,
+      "step": 507810
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 1.6763220326012092e-05,
+      "loss": 0.6927,
+      "step": 507820
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 1.6762072396112342e-05,
+      "loss": 0.7211,
+      "step": 507830
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 1.6760924466212595e-05,
+      "loss": 0.6841,
+      "step": 507840
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 1.6759776536312845e-05,
+      "loss": 0.7011,
+      "step": 507850
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 1.67586286064131e-05,
+      "loss": 0.7236,
+      "step": 507860
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 1.6757480676513352e-05,
+      "loss": 0.8194,
+      "step": 507870
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 1.6756332746613605e-05,
+      "loss": 0.6654,
+      "step": 507880
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 1.6755184816713858e-05,
+      "loss": 0.7943,
+      "step": 507890
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 1.675403688681411e-05,
+      "loss": 0.6673,
+      "step": 507900
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 1.675288895691436e-05,
+      "loss": 0.7185,
+      "step": 507910
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 1.6751741027014615e-05,
+      "loss": 0.7807,
+      "step": 507920
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 1.6750593097114868e-05,
+      "loss": 0.7594,
+      "step": 507930
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 1.674944516721512e-05,
+      "loss": 0.658,
+      "step": 507940
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 1.6748297237315374e-05,
+      "loss": 0.7897,
+      "step": 507950
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 1.6747149307415628e-05,
+      "loss": 0.6892,
+      "step": 507960
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 1.6746001377515877e-05,
+      "loss": 0.6936,
+      "step": 507970
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 1.674485344761613e-05,
+      "loss": 0.7061,
+      "step": 507980
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 1.6743705517716384e-05,
+      "loss": 0.7374,
+      "step": 507990
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 1.6742557587816634e-05,
+      "loss": 0.7279,
+      "step": 508000
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 1.6741409657916887e-05,
+      "loss": 0.7073,
+      "step": 508010
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 1.674026172801714e-05,
+      "loss": 0.7532,
+      "step": 508020
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 1.6739113798117394e-05,
+      "loss": 0.766,
+      "step": 508030
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 1.6737965868217647e-05,
+      "loss": 0.7342,
+      "step": 508040
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 1.67368179383179e-05,
+      "loss": 0.7155,
+      "step": 508050
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 1.673567000841815e-05,
+      "loss": 0.7718,
+      "step": 508060
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 1.6734522078518403e-05,
+      "loss": 0.8408,
+      "step": 508070
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 1.6733374148618656e-05,
+      "loss": 0.776,
+      "step": 508080
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 1.673222621871891e-05,
+      "loss": 0.7757,
+      "step": 508090
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 1.6731078288819163e-05,
+      "loss": 0.727,
+      "step": 508100
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 1.6729930358919416e-05,
+      "loss": 0.7703,
+      "step": 508110
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 1.6728782429019666e-05,
+      "loss": 0.7298,
+      "step": 508120
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 1.672763449911992e-05,
+      "loss": 0.7441,
+      "step": 508130
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 1.6726486569220172e-05,
+      "loss": 0.7267,
+      "step": 508140
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 1.6725338639320422e-05,
+      "loss": 0.7687,
+      "step": 508150
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 1.6724190709420675e-05,
+      "loss": 0.7144,
+      "step": 508160
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 1.672304277952093e-05,
+      "loss": 0.7347,
+      "step": 508170
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 1.6721894849621182e-05,
+      "loss": 0.7061,
+      "step": 508180
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 1.6720746919721435e-05,
+      "loss": 0.8497,
+      "step": 508190
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 1.6719598989821685e-05,
+      "loss": 0.7382,
+      "step": 508200
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 1.671845105992194e-05,
+      "loss": 0.7131,
+      "step": 508210
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 1.671730313002219e-05,
+      "loss": 0.7476,
+      "step": 508220
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 1.6716155200122445e-05,
+      "loss": 0.7503,
+      "step": 508230
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 1.6715007270222698e-05,
+      "loss": 0.6863,
+      "step": 508240
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 1.671385934032295e-05,
+      "loss": 0.6824,
+      "step": 508250
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 1.67127114104232e-05,
+      "loss": 0.6795,
+      "step": 508260
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 1.6711563480523454e-05,
+      "loss": 0.6856,
+      "step": 508270
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 1.6710415550623708e-05,
+      "loss": 0.7516,
+      "step": 508280
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 1.6709267620723957e-05,
+      "loss": 0.7871,
+      "step": 508290
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 1.670811969082421e-05,
+      "loss": 0.7755,
+      "step": 508300
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 1.6706971760924464e-05,
+      "loss": 0.7445,
+      "step": 508310
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 1.6705823831024717e-05,
+      "loss": 0.7204,
+      "step": 508320
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 1.670467590112497e-05,
+      "loss": 0.7093,
+      "step": 508330
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 1.6703527971225224e-05,
+      "loss": 0.6983,
+      "step": 508340
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 1.6702380041325474e-05,
+      "loss": 0.8626,
+      "step": 508350
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 1.6701232111425727e-05,
+      "loss": 0.7481,
+      "step": 508360
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 1.670008418152598e-05,
+      "loss": 0.6873,
+      "step": 508370
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 1.6698936251626233e-05,
+      "loss": 0.6433,
+      "step": 508380
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 1.6697788321726487e-05,
+      "loss": 0.7523,
+      "step": 508390
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 1.669664039182674e-05,
+      "loss": 0.7357,
+      "step": 508400
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 1.669549246192699e-05,
+      "loss": 0.7714,
+      "step": 508410
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 1.6694344532027243e-05,
+      "loss": 0.6761,
+      "step": 508420
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 1.6693196602127496e-05,
+      "loss": 0.7114,
+      "step": 508430
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 1.6692048672227746e-05,
+      "loss": 0.7651,
+      "step": 508440
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 1.6690900742328e-05,
+      "loss": 0.7186,
+      "step": 508450
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 1.6689752812428252e-05,
+      "loss": 0.7107,
+      "step": 508460
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 1.6688604882528506e-05,
+      "loss": 0.7873,
+      "step": 508470
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 1.668745695262876e-05,
+      "loss": 0.8056,
+      "step": 508480
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 1.6686309022729012e-05,
+      "loss": 0.708,
+      "step": 508490
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 1.6685161092829262e-05,
+      "loss": 0.7179,
+      "step": 508500
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 1.6684013162929515e-05,
+      "loss": 0.7415,
+      "step": 508510
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 1.668286523302977e-05,
+      "loss": 0.7256,
+      "step": 508520
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 1.6681717303130022e-05,
+      "loss": 0.7674,
+      "step": 508530
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 1.6680569373230275e-05,
+      "loss": 0.7499,
+      "step": 508540
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 1.6679421443330525e-05,
+      "loss": 0.7847,
+      "step": 508550
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 1.6678273513430778e-05,
+      "loss": 0.8124,
+      "step": 508560
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 1.667712558353103e-05,
+      "loss": 0.7122,
+      "step": 508570
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 1.667597765363128e-05,
+      "loss": 0.6995,
+      "step": 508580
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 1.6674829723731534e-05,
+      "loss": 0.7514,
+      "step": 508590
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 1.6673681793831788e-05,
+      "loss": 0.705,
+      "step": 508600
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 1.667253386393204e-05,
+      "loss": 0.7905,
+      "step": 508610
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 1.6671385934032294e-05,
+      "loss": 0.7551,
+      "step": 508620
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 1.6670238004132547e-05,
+      "loss": 0.6668,
+      "step": 508630
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 1.6669090074232797e-05,
+      "loss": 0.7695,
+      "step": 508640
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 1.666794214433305e-05,
+      "loss": 0.7697,
+      "step": 508650
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 1.6666794214433304e-05,
+      "loss": 0.8118,
+      "step": 508660
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 1.6665646284533557e-05,
+      "loss": 0.6969,
+      "step": 508670
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 1.666449835463381e-05,
+      "loss": 0.8298,
+      "step": 508680
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 1.6663350424734063e-05,
+      "loss": 0.7359,
+      "step": 508690
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 1.6662202494834313e-05,
+      "loss": 0.7663,
+      "step": 508700
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 1.6661054564934567e-05,
+      "loss": 0.7707,
+      "step": 508710
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 1.665990663503482e-05,
+      "loss": 0.7531,
+      "step": 508720
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 1.665875870513507e-05,
+      "loss": 0.7078,
+      "step": 508730
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 1.6657610775235323e-05,
+      "loss": 0.7181,
+      "step": 508740
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 1.6656462845335576e-05,
+      "loss": 0.7845,
+      "step": 508750
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 1.665531491543583e-05,
+      "loss": 0.7303,
+      "step": 508760
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 1.6654166985536083e-05,
+      "loss": 0.771,
+      "step": 508770
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 1.6653019055636336e-05,
+      "loss": 0.7659,
+      "step": 508780
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 1.6651871125736586e-05,
+      "loss": 0.751,
+      "step": 508790
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 1.665072319583684e-05,
+      "loss": 0.6909,
+      "step": 508800
+    },
+    {
+      "epoch": 23.34,
+      "learning_rate": 1.6649575265937092e-05,
+      "loss": 0.7174,
+      "step": 508810
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 1.6648427336037345e-05,
+      "loss": 0.7147,
+      "step": 508820
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 1.66472794061376e-05,
+      "loss": 0.7419,
+      "step": 508830
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 1.6646131476237852e-05,
+      "loss": 0.6946,
+      "step": 508840
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 1.6644983546338102e-05,
+      "loss": 0.6858,
+      "step": 508850
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 1.6643835616438355e-05,
+      "loss": 0.6652,
+      "step": 508860
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 1.6642687686538608e-05,
+      "loss": 0.7522,
+      "step": 508870
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 1.6641539756638858e-05,
+      "loss": 0.6846,
+      "step": 508880
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 1.664039182673911e-05,
+      "loss": 0.7748,
+      "step": 508890
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 1.6639243896839365e-05,
+      "loss": 0.7924,
+      "step": 508900
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 1.6638095966939618e-05,
+      "loss": 0.716,
+      "step": 508910
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 1.663694803703987e-05,
+      "loss": 0.7636,
+      "step": 508920
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 1.663580010714012e-05,
+      "loss": 0.7162,
+      "step": 508930
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 1.6634652177240374e-05,
+      "loss": 0.7384,
+      "step": 508940
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 1.6633504247340627e-05,
+      "loss": 0.7108,
+      "step": 508950
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 1.663235631744088e-05,
+      "loss": 0.7078,
+      "step": 508960
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 1.6631208387541134e-05,
+      "loss": 0.6947,
+      "step": 508970
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 1.6630060457641387e-05,
+      "loss": 0.7784,
+      "step": 508980
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 1.6628912527741637e-05,
+      "loss": 0.7013,
+      "step": 508990
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 1.662776459784189e-05,
+      "loss": 0.7491,
+      "step": 509000
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 1.6626616667942144e-05,
+      "loss": 0.7498,
+      "step": 509010
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 1.6625468738042393e-05,
+      "loss": 0.9053,
+      "step": 509020
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 1.6624320808142647e-05,
+      "loss": 0.7324,
+      "step": 509030
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 1.66231728782429e-05,
+      "loss": 0.8102,
+      "step": 509040
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 1.6622024948343153e-05,
+      "loss": 0.6888,
+      "step": 509050
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 1.6620877018443406e-05,
+      "loss": 0.7817,
+      "step": 509060
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 1.661972908854366e-05,
+      "loss": 0.7311,
+      "step": 509070
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 1.661858115864391e-05,
+      "loss": 0.8406,
+      "step": 509080
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 1.6617433228744163e-05,
+      "loss": 0.7298,
+      "step": 509090
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 1.6616285298844416e-05,
+      "loss": 0.769,
+      "step": 509100
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 1.661513736894467e-05,
+      "loss": 0.7547,
+      "step": 509110
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 1.6613989439044922e-05,
+      "loss": 0.7573,
+      "step": 509120
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 1.6612841509145176e-05,
+      "loss": 0.8283,
+      "step": 509130
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 1.6611693579245425e-05,
+      "loss": 0.7849,
+      "step": 509140
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 1.661054564934568e-05,
+      "loss": 0.7619,
+      "step": 509150
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 1.6609397719445932e-05,
+      "loss": 0.6965,
+      "step": 509160
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 1.6608249789546182e-05,
+      "loss": 0.7662,
+      "step": 509170
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 1.6607101859646435e-05,
+      "loss": 0.7243,
+      "step": 509180
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 1.660595392974669e-05,
+      "loss": 0.6676,
+      "step": 509190
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 1.660480599984694e-05,
+      "loss": 0.7604,
+      "step": 509200
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 1.6603658069947195e-05,
+      "loss": 0.7065,
+      "step": 509210
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 1.6602510140047445e-05,
+      "loss": 0.7928,
+      "step": 509220
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 1.6601362210147698e-05,
+      "loss": 0.8082,
+      "step": 509230
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 1.660021428024795e-05,
+      "loss": 0.7871,
+      "step": 509240
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 1.6599066350348204e-05,
+      "loss": 0.8141,
+      "step": 509250
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 1.6597918420448458e-05,
+      "loss": 0.7047,
+      "step": 509260
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 1.659677049054871e-05,
+      "loss": 0.7626,
+      "step": 509270
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 1.659562256064896e-05,
+      "loss": 0.6895,
+      "step": 509280
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 1.6594474630749214e-05,
+      "loss": 0.7443,
+      "step": 509290
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 1.6593326700849467e-05,
+      "loss": 0.7276,
+      "step": 509300
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 1.6592178770949717e-05,
+      "loss": 0.8007,
+      "step": 509310
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 1.659103084104997e-05,
+      "loss": 0.7234,
+      "step": 509320
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 1.6589882911150224e-05,
+      "loss": 0.6753,
+      "step": 509330
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 1.6588734981250477e-05,
+      "loss": 0.7808,
+      "step": 509340
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 1.658758705135073e-05,
+      "loss": 0.7184,
+      "step": 509350
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 1.6586439121450983e-05,
+      "loss": 0.7995,
+      "step": 509360
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 1.6585291191551233e-05,
+      "loss": 0.7648,
+      "step": 509370
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 1.6584143261651486e-05,
+      "loss": 0.6894,
+      "step": 509380
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 1.658299533175174e-05,
+      "loss": 0.7329,
+      "step": 509390
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 1.6581847401851993e-05,
+      "loss": 0.7342,
+      "step": 509400
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 1.6580699471952246e-05,
+      "loss": 0.7843,
+      "step": 509410
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 1.65795515420525e-05,
+      "loss": 0.7219,
+      "step": 509420
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 1.657840361215275e-05,
+      "loss": 0.8042,
+      "step": 509430
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 1.6577255682253002e-05,
+      "loss": 0.7048,
+      "step": 509440
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 1.6576107752353256e-05,
+      "loss": 0.7639,
+      "step": 509450
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 1.6574959822453506e-05,
+      "loss": 0.7527,
+      "step": 509460
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 1.657381189255376e-05,
+      "loss": 0.7442,
+      "step": 509470
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 1.6572663962654012e-05,
+      "loss": 0.7627,
+      "step": 509480
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 1.6571516032754265e-05,
+      "loss": 0.7509,
+      "step": 509490
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 1.657036810285452e-05,
+      "loss": 0.7413,
+      "step": 509500
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 1.6569220172954772e-05,
+      "loss": 0.7991,
+      "step": 509510
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 1.656807224305502e-05,
+      "loss": 0.7444,
+      "step": 509520
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 1.6566924313155275e-05,
+      "loss": 0.7608,
+      "step": 509530
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 1.6565776383255528e-05,
+      "loss": 0.7178,
+      "step": 509540
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 1.656462845335578e-05,
+      "loss": 0.7295,
+      "step": 509550
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 1.6563480523456035e-05,
+      "loss": 0.7616,
+      "step": 509560
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 1.6562332593556284e-05,
+      "loss": 0.7931,
+      "step": 509570
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 1.6561184663656538e-05,
+      "loss": 0.7208,
+      "step": 509580
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 1.656003673375679e-05,
+      "loss": 0.7155,
+      "step": 509590
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 1.655888880385704e-05,
+      "loss": 0.6968,
+      "step": 509600
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 1.6557740873957294e-05,
+      "loss": 0.7173,
+      "step": 509610
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 1.6556592944057547e-05,
+      "loss": 0.7137,
+      "step": 509620
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 1.65554450141578e-05,
+      "loss": 0.7509,
+      "step": 509630
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 1.6554297084258054e-05,
+      "loss": 0.7489,
+      "step": 509640
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 1.6553149154358307e-05,
+      "loss": 0.7447,
+      "step": 509650
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 1.6552001224458557e-05,
+      "loss": 0.7309,
+      "step": 509660
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 1.655085329455881e-05,
+      "loss": 0.763,
+      "step": 509670
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 1.6549705364659063e-05,
+      "loss": 0.7116,
+      "step": 509680
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 1.6548557434759317e-05,
+      "loss": 0.7424,
+      "step": 509690
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 1.654740950485957e-05,
+      "loss": 0.7323,
+      "step": 509700
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 1.6546261574959823e-05,
+      "loss": 0.7307,
+      "step": 509710
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 1.6545113645060073e-05,
+      "loss": 0.7939,
+      "step": 509720
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 1.6543965715160326e-05,
+      "loss": 0.7502,
+      "step": 509730
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 1.654281778526058e-05,
+      "loss": 0.7297,
+      "step": 509740
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 1.654166985536083e-05,
+      "loss": 0.734,
+      "step": 509750
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 1.6540521925461082e-05,
+      "loss": 0.7234,
+      "step": 509760
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 1.6539373995561336e-05,
+      "loss": 0.6934,
+      "step": 509770
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 1.653822606566159e-05,
+      "loss": 0.6914,
+      "step": 509780
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 1.6537078135761842e-05,
+      "loss": 0.724,
+      "step": 509790
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 1.6535930205862095e-05,
+      "loss": 0.6605,
+      "step": 509800
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 1.6534782275962345e-05,
+      "loss": 0.7176,
+      "step": 509810
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 1.65336343460626e-05,
+      "loss": 0.7122,
+      "step": 509820
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 1.6532486416162852e-05,
+      "loss": 0.7758,
+      "step": 509830
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 1.6531338486263105e-05,
+      "loss": 0.751,
+      "step": 509840
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 1.6530190556363358e-05,
+      "loss": 0.7764,
+      "step": 509850
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 1.652904262646361e-05,
+      "loss": 0.8482,
+      "step": 509860
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 1.652789469656386e-05,
+      "loss": 0.7844,
+      "step": 509870
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 1.6526746766664115e-05,
+      "loss": 0.8051,
+      "step": 509880
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 1.6525598836764368e-05,
+      "loss": 0.7422,
+      "step": 509890
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 1.6524450906864618e-05,
+      "loss": 0.7776,
+      "step": 509900
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 1.652330297696487e-05,
+      "loss": 0.7497,
+      "step": 509910
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 1.6522155047065124e-05,
+      "loss": 0.7382,
+      "step": 509920
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 1.6521007117165377e-05,
+      "loss": 0.7401,
+      "step": 509930
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 1.651985918726563e-05,
+      "loss": 0.7807,
+      "step": 509940
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 1.651871125736588e-05,
+      "loss": 0.7591,
+      "step": 509950
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 1.6517563327466134e-05,
+      "loss": 0.7633,
+      "step": 509960
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 1.6516415397566387e-05,
+      "loss": 0.7811,
+      "step": 509970
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 1.651526746766664e-05,
+      "loss": 0.6739,
+      "step": 509980
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 1.6514119537766893e-05,
+      "loss": 0.7495,
+      "step": 509990
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 1.6512971607867147e-05,
+      "loss": 0.7485,
+      "step": 510000
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 1.6511823677967397e-05,
+      "loss": 0.7702,
+      "step": 510010
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 1.651067574806765e-05,
+      "loss": 0.7651,
+      "step": 510020
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 1.6509527818167903e-05,
+      "loss": 0.7527,
+      "step": 510030
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 1.6508379888268153e-05,
+      "loss": 0.7536,
+      "step": 510040
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 1.6507231958368406e-05,
+      "loss": 0.6798,
+      "step": 510050
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 1.650608402846866e-05,
+      "loss": 0.7503,
+      "step": 510060
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 1.6504936098568913e-05,
+      "loss": 0.8475,
+      "step": 510070
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 1.6503788168669166e-05,
+      "loss": 0.7613,
+      "step": 510080
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 1.650264023876942e-05,
+      "loss": 0.7696,
+      "step": 510090
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 1.650149230886967e-05,
+      "loss": 0.7372,
+      "step": 510100
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 1.6500344378969922e-05,
+      "loss": 0.7623,
+      "step": 510110
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 1.6499196449070175e-05,
+      "loss": 0.7321,
+      "step": 510120
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 1.649804851917043e-05,
+      "loss": 0.6861,
+      "step": 510130
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 1.6496900589270682e-05,
+      "loss": 0.7781,
+      "step": 510140
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 1.6495752659370935e-05,
+      "loss": 0.8238,
+      "step": 510150
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 1.6494604729471185e-05,
+      "loss": 0.7643,
+      "step": 510160
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 1.649345679957144e-05,
+      "loss": 0.7064,
+      "step": 510170
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 1.649230886967169e-05,
+      "loss": 0.7445,
+      "step": 510180
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 1.649116093977194e-05,
+      "loss": 0.78,
+      "step": 510190
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 1.6490013009872195e-05,
+      "loss": 0.7163,
+      "step": 510200
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 1.6488865079972448e-05,
+      "loss": 0.7564,
+      "step": 510210
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 1.64877171500727e-05,
+      "loss": 0.7387,
+      "step": 510220
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 1.6486569220172954e-05,
+      "loss": 0.79,
+      "step": 510230
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 1.6485421290273208e-05,
+      "loss": 0.7623,
+      "step": 510240
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 1.6484273360373457e-05,
+      "loss": 0.8149,
+      "step": 510250
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 1.648312543047371e-05,
+      "loss": 0.7623,
+      "step": 510260
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 1.6481977500573964e-05,
+      "loss": 0.7975,
+      "step": 510270
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 1.6480829570674217e-05,
+      "loss": 0.6946,
+      "step": 510280
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 1.647968164077447e-05,
+      "loss": 0.7283,
+      "step": 510290
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 1.647853371087472e-05,
+      "loss": 0.7316,
+      "step": 510300
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 1.6477385780974974e-05,
+      "loss": 0.7187,
+      "step": 510310
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 1.6476237851075227e-05,
+      "loss": 0.7518,
+      "step": 510320
+    },
+    {
+      "epoch": 23.41,
+      "learning_rate": 1.6475089921175477e-05,
+      "loss": 0.799,
+      "step": 510330
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 1.647394199127573e-05,
+      "loss": 0.7654,
+      "step": 510340
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 1.6472794061375983e-05,
+      "loss": 0.7475,
+      "step": 510350
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 1.6471646131476236e-05,
+      "loss": 0.8031,
+      "step": 510360
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 1.647049820157649e-05,
+      "loss": 0.7674,
+      "step": 510370
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 1.6469350271676743e-05,
+      "loss": 0.7271,
+      "step": 510380
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 1.6468202341776993e-05,
+      "loss": 0.8123,
+      "step": 510390
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 1.6467054411877246e-05,
+      "loss": 0.7911,
+      "step": 510400
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 1.64659064819775e-05,
+      "loss": 0.6935,
+      "step": 510410
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 1.6464758552077752e-05,
+      "loss": 0.6964,
+      "step": 510420
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 1.6463610622178006e-05,
+      "loss": 0.768,
+      "step": 510430
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 1.646246269227826e-05,
+      "loss": 0.7073,
+      "step": 510440
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 1.646131476237851e-05,
+      "loss": 0.7892,
+      "step": 510450
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 1.6460166832478762e-05,
+      "loss": 0.6929,
+      "step": 510460
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 1.6459018902579015e-05,
+      "loss": 0.7462,
+      "step": 510470
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 1.6457870972679265e-05,
+      "loss": 0.7017,
+      "step": 510480
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 1.645672304277952e-05,
+      "loss": 0.7254,
+      "step": 510490
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 1.645557511287977e-05,
+      "loss": 0.8213,
+      "step": 510500
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 1.6454427182980025e-05,
+      "loss": 0.7189,
+      "step": 510510
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 1.6453279253080278e-05,
+      "loss": 0.8826,
+      "step": 510520
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 1.645213132318053e-05,
+      "loss": 0.7609,
+      "step": 510530
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 1.645098339328078e-05,
+      "loss": 0.7316,
+      "step": 510540
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 1.6449835463381034e-05,
+      "loss": 0.7203,
+      "step": 510550
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 1.6448687533481288e-05,
+      "loss": 0.6952,
+      "step": 510560
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 1.644753960358154e-05,
+      "loss": 0.7568,
+      "step": 510570
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 1.6446391673681794e-05,
+      "loss": 0.785,
+      "step": 510580
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 1.6445243743782044e-05,
+      "loss": 0.7499,
+      "step": 510590
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 1.6444095813882297e-05,
+      "loss": 0.7335,
+      "step": 510600
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 1.644294788398255e-05,
+      "loss": 0.7936,
+      "step": 510610
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 1.64417999540828e-05,
+      "loss": 0.7192,
+      "step": 510620
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 1.6440652024183054e-05,
+      "loss": 0.6833,
+      "step": 510630
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 1.6439504094283307e-05,
+      "loss": 0.8103,
+      "step": 510640
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 1.643835616438356e-05,
+      "loss": 0.7434,
+      "step": 510650
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 1.6437208234483813e-05,
+      "loss": 0.752,
+      "step": 510660
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 1.6436060304584067e-05,
+      "loss": 0.6402,
+      "step": 510670
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 1.6434912374684316e-05,
+      "loss": 0.7962,
+      "step": 510680
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 1.643376444478457e-05,
+      "loss": 0.7943,
+      "step": 510690
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 1.6432616514884823e-05,
+      "loss": 0.7305,
+      "step": 510700
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 1.6431468584985076e-05,
+      "loss": 0.8433,
+      "step": 510710
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 1.643032065508533e-05,
+      "loss": 0.7991,
+      "step": 510720
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 1.6429172725185583e-05,
+      "loss": 0.6799,
+      "step": 510730
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 1.6428024795285832e-05,
+      "loss": 0.7511,
+      "step": 510740
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 1.6426876865386086e-05,
+      "loss": 0.7845,
+      "step": 510750
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 1.642572893548634e-05,
+      "loss": 0.7545,
+      "step": 510760
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 1.642458100558659e-05,
+      "loss": 0.729,
+      "step": 510770
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 1.6423433075686842e-05,
+      "loss": 0.7118,
+      "step": 510780
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 1.6422285145787095e-05,
+      "loss": 0.7515,
+      "step": 510790
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 1.642113721588735e-05,
+      "loss": 0.7603,
+      "step": 510800
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 1.6419989285987602e-05,
+      "loss": 0.6856,
+      "step": 510810
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 1.6418841356087855e-05,
+      "loss": 0.8034,
+      "step": 510820
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 1.6417693426188105e-05,
+      "loss": 0.7703,
+      "step": 510830
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 1.6416545496288358e-05,
+      "loss": 0.7431,
+      "step": 510840
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 1.641539756638861e-05,
+      "loss": 0.7076,
+      "step": 510850
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 1.6414249636488865e-05,
+      "loss": 0.7779,
+      "step": 510860
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 1.6413101706589118e-05,
+      "loss": 0.7485,
+      "step": 510870
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 1.641195377668937e-05,
+      "loss": 0.7344,
+      "step": 510880
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 1.641080584678962e-05,
+      "loss": 0.7771,
+      "step": 510890
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 1.6409657916889874e-05,
+      "loss": 0.788,
+      "step": 510900
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 1.6408509986990127e-05,
+      "loss": 0.7192,
+      "step": 510910
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 1.6407362057090377e-05,
+      "loss": 0.6927,
+      "step": 510920
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 1.640621412719063e-05,
+      "loss": 0.738,
+      "step": 510930
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 1.6405066197290884e-05,
+      "loss": 0.7537,
+      "step": 510940
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 1.6403918267391137e-05,
+      "loss": 0.6783,
+      "step": 510950
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 1.640277033749139e-05,
+      "loss": 0.8161,
+      "step": 510960
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 1.640162240759164e-05,
+      "loss": 0.8272,
+      "step": 510970
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 1.6400474477691893e-05,
+      "loss": 0.8129,
+      "step": 510980
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 1.6399326547792147e-05,
+      "loss": 0.7333,
+      "step": 510990
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 1.63981786178924e-05,
+      "loss": 0.7604,
+      "step": 511000
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 1.6397030687992653e-05,
+      "loss": 0.7725,
+      "step": 511010
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 1.6395882758092906e-05,
+      "loss": 0.7522,
+      "step": 511020
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 1.6394734828193156e-05,
+      "loss": 0.766,
+      "step": 511030
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 1.639358689829341e-05,
+      "loss": 0.7682,
+      "step": 511040
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 1.6392438968393663e-05,
+      "loss": 0.7828,
+      "step": 511050
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 1.6391291038493913e-05,
+      "loss": 0.809,
+      "step": 511060
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 1.6390143108594166e-05,
+      "loss": 0.7401,
+      "step": 511070
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 1.638899517869442e-05,
+      "loss": 0.7709,
+      "step": 511080
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 1.6387847248794672e-05,
+      "loss": 0.6878,
+      "step": 511090
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 1.6386699318894925e-05,
+      "loss": 0.7571,
+      "step": 511100
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 1.638555138899518e-05,
+      "loss": 0.7401,
+      "step": 511110
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 1.638440345909543e-05,
+      "loss": 0.7144,
+      "step": 511120
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 1.6383255529195682e-05,
+      "loss": 0.7248,
+      "step": 511130
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 1.6382107599295935e-05,
+      "loss": 0.7797,
+      "step": 511140
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 1.6380959669396188e-05,
+      "loss": 0.7719,
+      "step": 511150
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 1.637981173949644e-05,
+      "loss": 0.7657,
+      "step": 511160
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 1.6378663809596695e-05,
+      "loss": 0.6836,
+      "step": 511170
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 1.6377515879696945e-05,
+      "loss": 0.7008,
+      "step": 511180
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 1.6376367949797198e-05,
+      "loss": 0.7521,
+      "step": 511190
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 1.637522001989745e-05,
+      "loss": 0.8472,
+      "step": 511200
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 1.63740720899977e-05,
+      "loss": 0.72,
+      "step": 511210
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 1.6372924160097954e-05,
+      "loss": 0.7571,
+      "step": 511220
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 1.6371776230198207e-05,
+      "loss": 0.7987,
+      "step": 511230
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 1.637062830029846e-05,
+      "loss": 0.7338,
+      "step": 511240
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 1.6369480370398714e-05,
+      "loss": 0.7163,
+      "step": 511250
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 1.6368332440498967e-05,
+      "loss": 0.7383,
+      "step": 511260
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 1.6367184510599217e-05,
+      "loss": 0.7407,
+      "step": 511270
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 1.636603658069947e-05,
+      "loss": 0.7254,
+      "step": 511280
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 1.6364888650799724e-05,
+      "loss": 0.7502,
+      "step": 511290
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 1.6363740720899977e-05,
+      "loss": 0.7163,
+      "step": 511300
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 1.636259279100023e-05,
+      "loss": 0.7926,
+      "step": 511310
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 1.636144486110048e-05,
+      "loss": 0.665,
+      "step": 511320
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 1.6360296931200733e-05,
+      "loss": 0.6613,
+      "step": 511330
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 1.6359149001300986e-05,
+      "loss": 0.7688,
+      "step": 511340
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 1.6358001071401236e-05,
+      "loss": 0.788,
+      "step": 511350
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 1.635685314150149e-05,
+      "loss": 0.6797,
+      "step": 511360
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 1.6355705211601743e-05,
+      "loss": 0.7586,
+      "step": 511370
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 1.6354557281701996e-05,
+      "loss": 0.7738,
+      "step": 511380
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 1.635340935180225e-05,
+      "loss": 0.5928,
+      "step": 511390
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 1.6352261421902502e-05,
+      "loss": 0.8564,
+      "step": 511400
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 1.6351113492002752e-05,
+      "loss": 0.7637,
+      "step": 511410
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 1.6349965562103006e-05,
+      "loss": 0.8103,
+      "step": 511420
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 1.634881763220326e-05,
+      "loss": 0.7388,
+      "step": 511430
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 1.6347669702303512e-05,
+      "loss": 0.7666,
+      "step": 511440
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 1.6346521772403765e-05,
+      "loss": 0.8392,
+      "step": 511450
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 1.634537384250402e-05,
+      "loss": 0.7652,
+      "step": 511460
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 1.634422591260427e-05,
+      "loss": 0.7035,
+      "step": 511470
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 1.634307798270452e-05,
+      "loss": 0.6642,
+      "step": 511480
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 1.6341930052804775e-05,
+      "loss": 0.7908,
+      "step": 511490
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 1.6340782122905025e-05,
+      "loss": 0.7776,
+      "step": 511500
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 1.6339634193005278e-05,
+      "loss": 0.7101,
+      "step": 511510
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 1.633848626310553e-05,
+      "loss": 0.8327,
+      "step": 511520
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 1.6337338333205784e-05,
+      "loss": 0.6862,
+      "step": 511530
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 1.6336190403306038e-05,
+      "loss": 0.7262,
+      "step": 511540
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 1.633504247340629e-05,
+      "loss": 0.7374,
+      "step": 511550
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 1.633389454350654e-05,
+      "loss": 0.7717,
+      "step": 511560
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 1.6332746613606794e-05,
+      "loss": 0.7886,
+      "step": 511570
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 1.6331598683707047e-05,
+      "loss": 0.7225,
+      "step": 511580
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 1.63304507538073e-05,
+      "loss": 0.7208,
+      "step": 511590
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 1.6329302823907554e-05,
+      "loss": 0.7358,
+      "step": 511600
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 1.6328154894007807e-05,
+      "loss": 0.6943,
+      "step": 511610
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 1.6327006964108057e-05,
+      "loss": 0.729,
+      "step": 511620
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 1.632585903420831e-05,
+      "loss": 0.7618,
+      "step": 511630
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 1.632471110430856e-05,
+      "loss": 0.7369,
+      "step": 511640
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 1.6323563174408813e-05,
+      "loss": 0.7633,
+      "step": 511650
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 1.6322415244509066e-05,
+      "loss": 0.7502,
+      "step": 511660
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 1.632126731460932e-05,
+      "loss": 0.7418,
+      "step": 511670
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 1.6320119384709573e-05,
+      "loss": 0.7176,
+      "step": 511680
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 1.6318971454809826e-05,
+      "loss": 0.8413,
+      "step": 511690
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 1.6317823524910076e-05,
+      "loss": 0.7598,
+      "step": 511700
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 1.631667559501033e-05,
+      "loss": 0.7155,
+      "step": 511710
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 1.6315527665110582e-05,
+      "loss": 0.7588,
+      "step": 511720
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 1.6314379735210836e-05,
+      "loss": 0.7377,
+      "step": 511730
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 1.631323180531109e-05,
+      "loss": 0.7287,
+      "step": 511740
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 1.6312083875411342e-05,
+      "loss": 0.6486,
+      "step": 511750
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 1.6310935945511592e-05,
+      "loss": 0.8574,
+      "step": 511760
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 1.6309788015611845e-05,
+      "loss": 0.7343,
+      "step": 511770
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 1.63086400857121e-05,
+      "loss": 0.7609,
+      "step": 511780
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 1.630749215581235e-05,
+      "loss": 0.7241,
+      "step": 511790
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 1.63063442259126e-05,
+      "loss": 0.7737,
+      "step": 511800
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 1.6305196296012855e-05,
+      "loss": 0.7299,
+      "step": 511810
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 1.6304048366113108e-05,
+      "loss": 0.7496,
+      "step": 511820
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 1.630290043621336e-05,
+      "loss": 0.7484,
+      "step": 511830
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 1.6301752506313615e-05,
+      "loss": 0.6814,
+      "step": 511840
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 1.6300604576413864e-05,
+      "loss": 0.7499,
+      "step": 511850
+    },
+    {
+      "epoch": 23.48,
+      "learning_rate": 1.6299456646514118e-05,
+      "loss": 0.7371,
+      "step": 511860
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 1.629830871661437e-05,
+      "loss": 0.7784,
+      "step": 511870
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 1.6297160786714624e-05,
+      "loss": 0.7921,
+      "step": 511880
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 1.6296012856814877e-05,
+      "loss": 0.7831,
+      "step": 511890
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 1.629486492691513e-05,
+      "loss": 0.7882,
+      "step": 511900
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 1.629371699701538e-05,
+      "loss": 0.7665,
+      "step": 511910
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 1.6292569067115634e-05,
+      "loss": 0.7063,
+      "step": 511920
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 1.6291421137215887e-05,
+      "loss": 0.7179,
+      "step": 511930
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 1.6290273207316137e-05,
+      "loss": 0.7109,
+      "step": 511940
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 1.628912527741639e-05,
+      "loss": 0.815,
+      "step": 511950
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 1.6287977347516643e-05,
+      "loss": 0.8558,
+      "step": 511960
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 1.6286829417616897e-05,
+      "loss": 0.764,
+      "step": 511970
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 1.628568148771715e-05,
+      "loss": 0.7616,
+      "step": 511980
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 1.62845335578174e-05,
+      "loss": 0.8139,
+      "step": 511990
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 1.6283385627917653e-05,
+      "loss": 0.6526,
+      "step": 512000
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 1.6282237698017906e-05,
+      "loss": 0.7265,
+      "step": 512010
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 1.628108976811816e-05,
+      "loss": 0.7223,
+      "step": 512020
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 1.6279941838218413e-05,
+      "loss": 0.7162,
+      "step": 512030
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 1.6278793908318666e-05,
+      "loss": 0.7723,
+      "step": 512040
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 1.6277645978418916e-05,
+      "loss": 0.7832,
+      "step": 512050
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 1.627649804851917e-05,
+      "loss": 0.8101,
+      "step": 512060
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 1.6275350118619422e-05,
+      "loss": 0.7416,
+      "step": 512070
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 1.6274202188719672e-05,
+      "loss": 0.7811,
+      "step": 512080
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 1.6273054258819925e-05,
+      "loss": 0.66,
+      "step": 512090
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 1.627190632892018e-05,
+      "loss": 0.6988,
+      "step": 512100
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 1.6270758399020432e-05,
+      "loss": 0.7296,
+      "step": 512110
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 1.6269610469120685e-05,
+      "loss": 0.7887,
+      "step": 512120
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 1.6268462539220938e-05,
+      "loss": 0.747,
+      "step": 512130
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 1.6267314609321188e-05,
+      "loss": 0.6448,
+      "step": 512140
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 1.626616667942144e-05,
+      "loss": 0.7263,
+      "step": 512150
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 1.6265018749521695e-05,
+      "loss": 0.7186,
+      "step": 512160
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 1.6263870819621948e-05,
+      "loss": 0.7144,
+      "step": 512170
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 1.62627228897222e-05,
+      "loss": 0.7604,
+      "step": 512180
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 1.6261574959822454e-05,
+      "loss": 0.7677,
+      "step": 512190
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 1.6260427029922704e-05,
+      "loss": 0.6654,
+      "step": 512200
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 1.6259279100022957e-05,
+      "loss": 0.7668,
+      "step": 512210
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 1.625813117012321e-05,
+      "loss": 0.7774,
+      "step": 512220
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 1.625698324022346e-05,
+      "loss": 0.7937,
+      "step": 512230
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 1.6255835310323714e-05,
+      "loss": 0.6913,
+      "step": 512240
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 1.6254687380423967e-05,
+      "loss": 0.7504,
+      "step": 512250
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 1.625353945052422e-05,
+      "loss": 0.803,
+      "step": 512260
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 1.6252391520624474e-05,
+      "loss": 0.7215,
+      "step": 512270
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 1.6251243590724727e-05,
+      "loss": 0.7099,
+      "step": 512280
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 1.6250095660824977e-05,
+      "loss": 0.7825,
+      "step": 512290
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 1.624894773092523e-05,
+      "loss": 0.7026,
+      "step": 512300
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 1.6247799801025483e-05,
+      "loss": 0.7012,
+      "step": 512310
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 1.6246651871125736e-05,
+      "loss": 0.7256,
+      "step": 512320
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 1.624550394122599e-05,
+      "loss": 0.707,
+      "step": 512330
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 1.624435601132624e-05,
+      "loss": 0.6911,
+      "step": 512340
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 1.6243208081426493e-05,
+      "loss": 0.743,
+      "step": 512350
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 1.6242060151526746e-05,
+      "loss": 0.77,
+      "step": 512360
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 1.6240912221626996e-05,
+      "loss": 0.7203,
+      "step": 512370
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 1.623976429172725e-05,
+      "loss": 0.7133,
+      "step": 512380
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 1.6238616361827502e-05,
+      "loss": 0.7663,
+      "step": 512390
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 1.6237468431927756e-05,
+      "loss": 0.7461,
+      "step": 512400
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 1.623632050202801e-05,
+      "loss": 0.7798,
+      "step": 512410
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 1.6235172572128262e-05,
+      "loss": 0.8921,
+      "step": 512420
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 1.6234024642228512e-05,
+      "loss": 0.7338,
+      "step": 512430
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 1.6232876712328765e-05,
+      "loss": 0.7013,
+      "step": 512440
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 1.623172878242902e-05,
+      "loss": 0.7489,
+      "step": 512450
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 1.623058085252927e-05,
+      "loss": 0.7086,
+      "step": 512460
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 1.6229432922629525e-05,
+      "loss": 0.7996,
+      "step": 512470
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 1.6228284992729778e-05,
+      "loss": 0.6892,
+      "step": 512480
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 1.6227137062830028e-05,
+      "loss": 0.7991,
+      "step": 512490
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 1.622598913293028e-05,
+      "loss": 0.7388,
+      "step": 512500
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 1.6224841203030534e-05,
+      "loss": 0.7478,
+      "step": 512510
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 1.6223693273130784e-05,
+      "loss": 0.7722,
+      "step": 512520
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 1.6222545343231038e-05,
+      "loss": 0.7308,
+      "step": 512530
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 1.622139741333129e-05,
+      "loss": 0.7963,
+      "step": 512540
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 1.6220249483431544e-05,
+      "loss": 0.6797,
+      "step": 512550
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 1.6219101553531797e-05,
+      "loss": 0.8958,
+      "step": 512560
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 1.621795362363205e-05,
+      "loss": 0.7506,
+      "step": 512570
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 1.62168056937323e-05,
+      "loss": 0.6487,
+      "step": 512580
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 1.6215657763832554e-05,
+      "loss": 0.8237,
+      "step": 512590
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 1.6214509833932807e-05,
+      "loss": 0.7627,
+      "step": 512600
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 1.621336190403306e-05,
+      "loss": 0.8356,
+      "step": 512610
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 1.6212213974133313e-05,
+      "loss": 0.767,
+      "step": 512620
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 1.6211066044233567e-05,
+      "loss": 0.6746,
+      "step": 512630
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 1.6209918114333816e-05,
+      "loss": 0.7654,
+      "step": 512640
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 1.620877018443407e-05,
+      "loss": 0.8063,
+      "step": 512650
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 1.620762225453432e-05,
+      "loss": 0.6913,
+      "step": 512660
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 1.6206474324634573e-05,
+      "loss": 0.7265,
+      "step": 512670
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 1.6205326394734826e-05,
+      "loss": 0.8081,
+      "step": 512680
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 1.620417846483508e-05,
+      "loss": 0.7944,
+      "step": 512690
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 1.6203030534935332e-05,
+      "loss": 0.7974,
+      "step": 512700
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 1.6201882605035586e-05,
+      "loss": 0.7104,
+      "step": 512710
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 1.6200734675135836e-05,
+      "loss": 0.7843,
+      "step": 512720
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 1.619958674523609e-05,
+      "loss": 0.8184,
+      "step": 512730
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 1.6198438815336342e-05,
+      "loss": 0.7706,
+      "step": 512740
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 1.6197290885436595e-05,
+      "loss": 0.7787,
+      "step": 512750
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 1.619614295553685e-05,
+      "loss": 0.7384,
+      "step": 512760
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 1.6194995025637102e-05,
+      "loss": 0.7184,
+      "step": 512770
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 1.619384709573735e-05,
+      "loss": 0.6768,
+      "step": 512780
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 1.6192699165837605e-05,
+      "loss": 0.7109,
+      "step": 512790
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 1.6191551235937858e-05,
+      "loss": 0.7937,
+      "step": 512800
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 1.6190403306038108e-05,
+      "loss": 0.7201,
+      "step": 512810
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 1.618925537613836e-05,
+      "loss": 0.7956,
+      "step": 512820
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 1.6188107446238614e-05,
+      "loss": 0.7859,
+      "step": 512830
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 1.6186959516338868e-05,
+      "loss": 0.7432,
+      "step": 512840
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 1.618581158643912e-05,
+      "loss": 0.7798,
+      "step": 512850
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 1.6184663656539374e-05,
+      "loss": 0.6786,
+      "step": 512860
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 1.6183515726639624e-05,
+      "loss": 0.7799,
+      "step": 512870
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 1.6182367796739877e-05,
+      "loss": 0.7484,
+      "step": 512880
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 1.618121986684013e-05,
+      "loss": 0.7535,
+      "step": 512890
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 1.6180071936940384e-05,
+      "loss": 0.7306,
+      "step": 512900
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 1.6178924007040637e-05,
+      "loss": 0.6288,
+      "step": 512910
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 1.617777607714089e-05,
+      "loss": 0.7004,
+      "step": 512920
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 1.617662814724114e-05,
+      "loss": 0.7178,
+      "step": 512930
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 1.6175480217341393e-05,
+      "loss": 0.7901,
+      "step": 512940
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 1.6174332287441647e-05,
+      "loss": 0.7971,
+      "step": 512950
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 1.6173184357541896e-05,
+      "loss": 0.7332,
+      "step": 512960
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 1.617203642764215e-05,
+      "loss": 0.6934,
+      "step": 512970
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 1.6170888497742403e-05,
+      "loss": 0.727,
+      "step": 512980
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 1.6169740567842656e-05,
+      "loss": 0.791,
+      "step": 512990
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 1.616859263794291e-05,
+      "loss": 0.8137,
+      "step": 513000
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 1.616744470804316e-05,
+      "loss": 0.7428,
+      "step": 513010
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 1.6166296778143413e-05,
+      "loss": 0.6979,
+      "step": 513020
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 1.6165148848243666e-05,
+      "loss": 0.7857,
+      "step": 513030
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 1.616400091834392e-05,
+      "loss": 0.735,
+      "step": 513040
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 1.6162852988444172e-05,
+      "loss": 0.7684,
+      "step": 513050
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 1.6161705058544425e-05,
+      "loss": 0.7677,
+      "step": 513060
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 1.6160557128644675e-05,
+      "loss": 0.715,
+      "step": 513070
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 1.615940919874493e-05,
+      "loss": 0.7535,
+      "step": 513080
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 1.6158261268845182e-05,
+      "loss": 0.7089,
+      "step": 513090
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 1.615711333894543e-05,
+      "loss": 0.8163,
+      "step": 513100
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 1.6155965409045685e-05,
+      "loss": 0.7708,
+      "step": 513110
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 1.6154817479145938e-05,
+      "loss": 0.7934,
+      "step": 513120
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 1.615366954924619e-05,
+      "loss": 0.7456,
+      "step": 513130
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 1.6152521619346445e-05,
+      "loss": 0.8533,
+      "step": 513140
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 1.6151373689446698e-05,
+      "loss": 0.8124,
+      "step": 513150
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 1.6150225759546948e-05,
+      "loss": 0.826,
+      "step": 513160
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 1.61490778296472e-05,
+      "loss": 0.7328,
+      "step": 513170
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 1.6147929899747454e-05,
+      "loss": 0.7122,
+      "step": 513180
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 1.6146781969847707e-05,
+      "loss": 0.8152,
+      "step": 513190
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 1.614563403994796e-05,
+      "loss": 0.767,
+      "step": 513200
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 1.6144486110048214e-05,
+      "loss": 0.754,
+      "step": 513210
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 1.6143338180148464e-05,
+      "loss": 0.7583,
+      "step": 513220
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 1.6142190250248717e-05,
+      "loss": 0.7336,
+      "step": 513230
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 1.614104232034897e-05,
+      "loss": 0.668,
+      "step": 513240
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 1.613989439044922e-05,
+      "loss": 0.7123,
+      "step": 513250
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 1.6138746460549473e-05,
+      "loss": 0.7785,
+      "step": 513260
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 1.6137598530649727e-05,
+      "loss": 0.7496,
+      "step": 513270
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 1.613645060074998e-05,
+      "loss": 0.7239,
+      "step": 513280
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 1.6135302670850233e-05,
+      "loss": 0.7549,
+      "step": 513290
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 1.6134154740950486e-05,
+      "loss": 0.7406,
+      "step": 513300
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 1.6133006811050736e-05,
+      "loss": 0.737,
+      "step": 513310
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 1.613185888115099e-05,
+      "loss": 0.7566,
+      "step": 513320
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 1.6130710951251243e-05,
+      "loss": 0.7439,
+      "step": 513330
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 1.6129563021351496e-05,
+      "loss": 0.7438,
+      "step": 513340
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 1.612841509145175e-05,
+      "loss": 0.6804,
+      "step": 513350
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 1.6127267161552e-05,
+      "loss": 0.8207,
+      "step": 513360
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 1.6126119231652252e-05,
+      "loss": 0.7872,
+      "step": 513370
+    },
+    {
+      "epoch": 23.55,
+      "learning_rate": 1.6124971301752506e-05,
+      "loss": 0.7223,
+      "step": 513380
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 1.6123823371852755e-05,
+      "loss": 0.7832,
+      "step": 513390
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 1.612267544195301e-05,
+      "loss": 0.8242,
+      "step": 513400
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 1.6121527512053262e-05,
+      "loss": 0.7092,
+      "step": 513410
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 1.6120379582153515e-05,
+      "loss": 0.7141,
+      "step": 513420
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 1.611923165225377e-05,
+      "loss": 0.8006,
+      "step": 513430
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 1.611808372235402e-05,
+      "loss": 0.7403,
+      "step": 513440
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 1.611693579245427e-05,
+      "loss": 0.7667,
+      "step": 513450
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 1.6115787862554525e-05,
+      "loss": 0.7331,
+      "step": 513460
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 1.6114639932654778e-05,
+      "loss": 0.6988,
+      "step": 513470
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 1.611349200275503e-05,
+      "loss": 0.6829,
+      "step": 513480
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 1.6112344072855284e-05,
+      "loss": 0.7906,
+      "step": 513490
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 1.6111196142955538e-05,
+      "loss": 0.8003,
+      "step": 513500
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 1.6110048213055787e-05,
+      "loss": 0.7779,
+      "step": 513510
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 1.610890028315604e-05,
+      "loss": 0.7518,
+      "step": 513520
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 1.6107752353256294e-05,
+      "loss": 0.7635,
+      "step": 513530
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 1.6106604423356544e-05,
+      "loss": 0.7488,
+      "step": 513540
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 1.6105456493456797e-05,
+      "loss": 0.7458,
+      "step": 513550
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 1.610430856355705e-05,
+      "loss": 0.6781,
+      "step": 513560
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 1.6103160633657304e-05,
+      "loss": 0.8682,
+      "step": 513570
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 1.6102012703757557e-05,
+      "loss": 0.7229,
+      "step": 513580
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 1.610086477385781e-05,
+      "loss": 0.6597,
+      "step": 513590
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 1.609971684395806e-05,
+      "loss": 0.6943,
+      "step": 513600
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 1.6098568914058313e-05,
+      "loss": 0.8251,
+      "step": 513610
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 1.6097420984158566e-05,
+      "loss": 0.7341,
+      "step": 513620
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 1.609627305425882e-05,
+      "loss": 0.7313,
+      "step": 513630
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 1.6095125124359073e-05,
+      "loss": 0.722,
+      "step": 513640
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 1.6093977194459326e-05,
+      "loss": 0.8287,
+      "step": 513650
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 1.6092829264559576e-05,
+      "loss": 0.7921,
+      "step": 513660
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 1.609168133465983e-05,
+      "loss": 0.7222,
+      "step": 513670
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 1.609053340476008e-05,
+      "loss": 0.7943,
+      "step": 513680
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 1.6089385474860332e-05,
+      "loss": 0.6839,
+      "step": 513690
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 1.6088237544960586e-05,
+      "loss": 0.6763,
+      "step": 513700
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 1.608708961506084e-05,
+      "loss": 0.7222,
+      "step": 513710
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 1.6085941685161092e-05,
+      "loss": 0.7196,
+      "step": 513720
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 1.6084793755261345e-05,
+      "loss": 0.7087,
+      "step": 513730
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 1.6083645825361595e-05,
+      "loss": 0.7181,
+      "step": 513740
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 1.608249789546185e-05,
+      "loss": 0.6748,
+      "step": 513750
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 1.60813499655621e-05,
+      "loss": 0.7659,
+      "step": 513760
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 1.6080202035662355e-05,
+      "loss": 0.6674,
+      "step": 513770
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 1.6079054105762608e-05,
+      "loss": 0.7463,
+      "step": 513780
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 1.607790617586286e-05,
+      "loss": 0.6871,
+      "step": 513790
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 1.607675824596311e-05,
+      "loss": 0.7101,
+      "step": 513800
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 1.6075610316063364e-05,
+      "loss": 0.7711,
+      "step": 513810
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 1.6074462386163618e-05,
+      "loss": 0.7103,
+      "step": 513820
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 1.6073314456263868e-05,
+      "loss": 0.6644,
+      "step": 513830
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 1.607216652636412e-05,
+      "loss": 0.6706,
+      "step": 513840
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 1.6071018596464374e-05,
+      "loss": 0.7809,
+      "step": 513850
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 1.6069870666564627e-05,
+      "loss": 0.7885,
+      "step": 513860
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 1.606872273666488e-05,
+      "loss": 0.7982,
+      "step": 513870
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 1.6067574806765134e-05,
+      "loss": 0.7394,
+      "step": 513880
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 1.6066426876865384e-05,
+      "loss": 0.6813,
+      "step": 513890
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 1.6065278946965637e-05,
+      "loss": 0.6769,
+      "step": 513900
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 1.606413101706589e-05,
+      "loss": 0.798,
+      "step": 513910
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 1.6062983087166143e-05,
+      "loss": 0.6733,
+      "step": 513920
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 1.6061835157266397e-05,
+      "loss": 0.699,
+      "step": 513930
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 1.606068722736665e-05,
+      "loss": 0.715,
+      "step": 513940
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 1.60595392974669e-05,
+      "loss": 0.7606,
+      "step": 513950
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 1.6058391367567153e-05,
+      "loss": 0.7728,
+      "step": 513960
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 1.6057243437667406e-05,
+      "loss": 0.7678,
+      "step": 513970
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 1.6056095507767656e-05,
+      "loss": 0.7752,
+      "step": 513980
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 1.605494757786791e-05,
+      "loss": 0.7779,
+      "step": 513990
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 1.6053799647968162e-05,
+      "loss": 0.7812,
+      "step": 514000
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 1.6052651718068416e-05,
+      "loss": 0.7153,
+      "step": 514010
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 1.605150378816867e-05,
+      "loss": 0.6388,
+      "step": 514020
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 1.605035585826892e-05,
+      "loss": 0.7098,
+      "step": 514030
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 1.6049207928369172e-05,
+      "loss": 0.8065,
+      "step": 514040
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 1.6048059998469425e-05,
+      "loss": 0.801,
+      "step": 514050
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 1.604691206856968e-05,
+      "loss": 0.7626,
+      "step": 514060
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 1.6045764138669932e-05,
+      "loss": 0.8227,
+      "step": 514070
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 1.6044616208770185e-05,
+      "loss": 0.7125,
+      "step": 514080
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 1.6043468278870435e-05,
+      "loss": 0.7192,
+      "step": 514090
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 1.6042320348970688e-05,
+      "loss": 0.7201,
+      "step": 514100
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 1.604117241907094e-05,
+      "loss": 0.7438,
+      "step": 514110
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 1.604002448917119e-05,
+      "loss": 0.7308,
+      "step": 514120
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 1.6038876559271444e-05,
+      "loss": 0.7823,
+      "step": 514130
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 1.6037728629371698e-05,
+      "loss": 0.7267,
+      "step": 514140
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 1.603658069947195e-05,
+      "loss": 0.7998,
+      "step": 514150
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 1.6035432769572204e-05,
+      "loss": 0.7626,
+      "step": 514160
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 1.6034284839672457e-05,
+      "loss": 0.7717,
+      "step": 514170
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 1.6033136909772707e-05,
+      "loss": 0.7464,
+      "step": 514180
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 1.603198897987296e-05,
+      "loss": 0.7011,
+      "step": 514190
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 1.6030841049973214e-05,
+      "loss": 0.7479,
+      "step": 514200
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 1.6029693120073467e-05,
+      "loss": 0.7429,
+      "step": 514210
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 1.602854519017372e-05,
+      "loss": 0.7861,
+      "step": 514220
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 1.6027397260273974e-05,
+      "loss": 0.7815,
+      "step": 514230
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 1.6026249330374223e-05,
+      "loss": 0.7687,
+      "step": 514240
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 1.6025101400474477e-05,
+      "loss": 0.7812,
+      "step": 514250
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 1.602395347057473e-05,
+      "loss": 0.6734,
+      "step": 514260
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 1.602280554067498e-05,
+      "loss": 0.6952,
+      "step": 514270
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 1.6021657610775233e-05,
+      "loss": 0.7493,
+      "step": 514280
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 1.6020509680875486e-05,
+      "loss": 0.6983,
+      "step": 514290
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 1.601936175097574e-05,
+      "loss": 0.7271,
+      "step": 514300
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 1.6018213821075993e-05,
+      "loss": 0.8395,
+      "step": 514310
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 1.6017065891176246e-05,
+      "loss": 0.6992,
+      "step": 514320
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 1.6015917961276496e-05,
+      "loss": 0.6888,
+      "step": 514330
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 1.601477003137675e-05,
+      "loss": 0.8231,
+      "step": 514340
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 1.6013622101477002e-05,
+      "loss": 0.6996,
+      "step": 514350
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 1.6012474171577256e-05,
+      "loss": 0.7337,
+      "step": 514360
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 1.601132624167751e-05,
+      "loss": 0.7287,
+      "step": 514370
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 1.601017831177776e-05,
+      "loss": 0.8239,
+      "step": 514380
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 1.6009030381878012e-05,
+      "loss": 0.7823,
+      "step": 514390
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 1.6007882451978265e-05,
+      "loss": 0.8207,
+      "step": 514400
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 1.6006734522078515e-05,
+      "loss": 0.7513,
+      "step": 514410
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 1.6005586592178768e-05,
+      "loss": 0.7312,
+      "step": 514420
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 1.600443866227902e-05,
+      "loss": 0.7468,
+      "step": 514430
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 1.6003290732379275e-05,
+      "loss": 0.8073,
+      "step": 514440
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 1.6002142802479528e-05,
+      "loss": 0.766,
+      "step": 514450
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 1.600099487257978e-05,
+      "loss": 0.7389,
+      "step": 514460
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 1.599984694268003e-05,
+      "loss": 0.7381,
+      "step": 514470
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 1.5998699012780284e-05,
+      "loss": 0.7808,
+      "step": 514480
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 1.5997551082880537e-05,
+      "loss": 0.8271,
+      "step": 514490
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 1.599640315298079e-05,
+      "loss": 0.7521,
+      "step": 514500
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 1.5995255223081044e-05,
+      "loss": 0.678,
+      "step": 514510
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 1.5994107293181297e-05,
+      "loss": 0.8101,
+      "step": 514520
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 1.5992959363281547e-05,
+      "loss": 0.787,
+      "step": 514530
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 1.59918114333818e-05,
+      "loss": 0.7808,
+      "step": 514540
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 1.5990663503482054e-05,
+      "loss": 0.7636,
+      "step": 514550
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 1.5989515573582303e-05,
+      "loss": 0.7595,
+      "step": 514560
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 1.5988367643682557e-05,
+      "loss": 0.756,
+      "step": 514570
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 1.598721971378281e-05,
+      "loss": 0.7555,
+      "step": 514580
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 1.5986071783883063e-05,
+      "loss": 0.7755,
+      "step": 514590
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 1.5984923853983316e-05,
+      "loss": 0.7128,
+      "step": 514600
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 1.598377592408357e-05,
+      "loss": 0.798,
+      "step": 514610
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 1.598262799418382e-05,
+      "loss": 0.7863,
+      "step": 514620
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 1.5981480064284073e-05,
+      "loss": 0.7412,
+      "step": 514630
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 1.5980332134384326e-05,
+      "loss": 0.7632,
+      "step": 514640
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 1.597918420448458e-05,
+      "loss": 0.7458,
+      "step": 514650
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 1.5978036274584832e-05,
+      "loss": 0.691,
+      "step": 514660
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 1.5976888344685086e-05,
+      "loss": 0.7765,
+      "step": 514670
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 1.5975740414785336e-05,
+      "loss": 0.7167,
+      "step": 514680
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 1.597459248488559e-05,
+      "loss": 0.7407,
+      "step": 514690
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 1.5973444554985842e-05,
+      "loss": 0.7552,
+      "step": 514700
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 1.5972296625086092e-05,
+      "loss": 0.755,
+      "step": 514710
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 1.5971148695186345e-05,
+      "loss": 0.7149,
+      "step": 514720
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 1.59700007652866e-05,
+      "loss": 0.7478,
+      "step": 514730
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 1.596885283538685e-05,
+      "loss": 0.6744,
+      "step": 514740
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 1.5967704905487105e-05,
+      "loss": 0.7658,
+      "step": 514750
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 1.5966556975587355e-05,
+      "loss": 0.8364,
+      "step": 514760
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 1.5965409045687608e-05,
+      "loss": 0.7605,
+      "step": 514770
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 1.596426111578786e-05,
+      "loss": 0.7425,
+      "step": 514780
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 1.5963113185888114e-05,
+      "loss": 0.7508,
+      "step": 514790
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 1.5961965255988368e-05,
+      "loss": 0.7999,
+      "step": 514800
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 1.596081732608862e-05,
+      "loss": 0.6982,
+      "step": 514810
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 1.595966939618887e-05,
+      "loss": 0.8012,
+      "step": 514820
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 1.5958521466289124e-05,
+      "loss": 0.708,
+      "step": 514830
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 1.5957373536389377e-05,
+      "loss": 0.6997,
+      "step": 514840
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 1.5956225606489627e-05,
+      "loss": 0.7343,
+      "step": 514850
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 1.595507767658988e-05,
+      "loss": 0.8173,
+      "step": 514860
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 1.5953929746690134e-05,
+      "loss": 0.7297,
+      "step": 514870
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 1.5952781816790387e-05,
+      "loss": 0.7651,
+      "step": 514880
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 1.595163388689064e-05,
+      "loss": 0.7299,
+      "step": 514890
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 1.5950485956990893e-05,
+      "loss": 0.7769,
+      "step": 514900
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 1.5949338027091143e-05,
+      "loss": 0.7216,
+      "step": 514910
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 1.5948190097191396e-05,
+      "loss": 0.7886,
+      "step": 514920
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 1.594704216729165e-05,
+      "loss": 0.7395,
+      "step": 514930
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 1.5945894237391903e-05,
+      "loss": 0.8191,
+      "step": 514940
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 1.5944746307492156e-05,
+      "loss": 0.7541,
+      "step": 514950
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 1.594359837759241e-05,
+      "loss": 0.7498,
+      "step": 514960
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 1.594245044769266e-05,
+      "loss": 0.7503,
+      "step": 514970
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 1.5941302517792912e-05,
+      "loss": 0.7421,
+      "step": 514980
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 1.5940154587893166e-05,
+      "loss": 0.6877,
+      "step": 514990
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 1.5939006657993416e-05,
+      "loss": 0.72,
+      "step": 515000
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 1.593785872809367e-05,
+      "loss": 0.7318,
+      "step": 515010
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 1.5936710798193922e-05,
+      "loss": 0.7596,
+      "step": 515020
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 1.5935562868294175e-05,
+      "loss": 0.7662,
+      "step": 515030
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 1.593441493839443e-05,
+      "loss": 0.7672,
+      "step": 515040
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 1.593326700849468e-05,
+      "loss": 0.7849,
+      "step": 515050
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 1.593211907859493e-05,
+      "loss": 0.6946,
+      "step": 515060
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 1.5930971148695185e-05,
+      "loss": 0.6837,
+      "step": 515070
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 1.5929823218795438e-05,
+      "loss": 0.7548,
+      "step": 515080
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 1.592867528889569e-05,
+      "loss": 0.759,
+      "step": 515090
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 1.5927527358995945e-05,
+      "loss": 0.6969,
+      "step": 515100
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 1.5926379429096194e-05,
+      "loss": 0.715,
+      "step": 515110
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 1.5925231499196448e-05,
+      "loss": 0.7095,
+      "step": 515120
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 1.59240835692967e-05,
+      "loss": 0.7531,
+      "step": 515130
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 1.592293563939695e-05,
+      "loss": 0.6997,
+      "step": 515140
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 1.5921787709497204e-05,
+      "loss": 0.7203,
+      "step": 515150
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 1.5920639779597457e-05,
+      "loss": 0.7484,
+      "step": 515160
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 1.591949184969771e-05,
+      "loss": 0.7501,
+      "step": 515170
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 1.5918343919797964e-05,
+      "loss": 0.7928,
+      "step": 515180
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 1.5917195989898217e-05,
+      "loss": 0.7316,
+      "step": 515190
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 1.5916048059998467e-05,
+      "loss": 0.7443,
+      "step": 515200
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 1.591490013009872e-05,
+      "loss": 0.8028,
+      "step": 515210
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 1.5913752200198973e-05,
+      "loss": 0.7014,
+      "step": 515220
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 1.5912604270299227e-05,
+      "loss": 0.7569,
+      "step": 515230
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 1.591145634039948e-05,
+      "loss": 0.6703,
+      "step": 515240
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 1.5910308410499733e-05,
+      "loss": 0.7899,
+      "step": 515250
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 1.5909160480599983e-05,
+      "loss": 0.7715,
+      "step": 515260
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 1.5908012550700236e-05,
+      "loss": 0.7322,
+      "step": 515270
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 1.590686462080049e-05,
+      "loss": 0.7856,
+      "step": 515280
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 1.590571669090074e-05,
+      "loss": 0.6444,
+      "step": 515290
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 1.5904568761000993e-05,
+      "loss": 0.7638,
+      "step": 515300
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 1.5903420831101246e-05,
+      "loss": 0.7876,
+      "step": 515310
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 1.59022729012015e-05,
+      "loss": 0.7691,
+      "step": 515320
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 1.5901124971301752e-05,
+      "loss": 0.7245,
+      "step": 515330
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 1.5899977041402006e-05,
+      "loss": 0.6933,
+      "step": 515340
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 1.5898829111502255e-05,
+      "loss": 0.7569,
+      "step": 515350
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 1.589768118160251e-05,
+      "loss": 0.7477,
+      "step": 515360
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 1.5896533251702762e-05,
+      "loss": 0.7737,
+      "step": 515370
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 1.5895385321803015e-05,
+      "loss": 0.7066,
+      "step": 515380
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 1.589423739190327e-05,
+      "loss": 0.7839,
+      "step": 515390
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 1.5893089462003518e-05,
+      "loss": 0.794,
+      "step": 515400
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 1.589194153210377e-05,
+      "loss": 0.7719,
+      "step": 515410
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 1.5890793602204025e-05,
+      "loss": 0.7259,
+      "step": 515420
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 1.5889645672304275e-05,
+      "loss": 0.7195,
+      "step": 515430
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 1.5888497742404528e-05,
+      "loss": 0.8063,
+      "step": 515440
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 1.588734981250478e-05,
+      "loss": 0.7421,
+      "step": 515450
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 1.5886201882605034e-05,
+      "loss": 0.7766,
+      "step": 515460
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 1.5885053952705287e-05,
+      "loss": 0.7023,
+      "step": 515470
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 1.588390602280554e-05,
+      "loss": 0.7327,
+      "step": 515480
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 1.588275809290579e-05,
+      "loss": 0.7375,
+      "step": 515490
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 1.5881610163006044e-05,
+      "loss": 0.708,
+      "step": 515500
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 1.5880462233106297e-05,
+      "loss": 0.75,
+      "step": 515510
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 1.587931430320655e-05,
+      "loss": 0.6871,
+      "step": 515520
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 1.5878166373306804e-05,
+      "loss": 0.7748,
+      "step": 515530
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 1.5877018443407057e-05,
+      "loss": 0.7749,
+      "step": 515540
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 1.5875870513507307e-05,
+      "loss": 0.7011,
+      "step": 515550
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 1.587472258360756e-05,
+      "loss": 0.6739,
+      "step": 515560
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 1.5873574653707813e-05,
+      "loss": 0.6894,
+      "step": 515570
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 1.5872426723808063e-05,
+      "loss": 0.7067,
+      "step": 515580
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 1.5871278793908316e-05,
+      "loss": 0.7199,
+      "step": 515590
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 1.587013086400857e-05,
+      "loss": 0.7927,
+      "step": 515600
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 1.5868982934108823e-05,
+      "loss": 0.7381,
+      "step": 515610
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 1.5867835004209076e-05,
+      "loss": 0.7348,
+      "step": 515620
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 1.586668707430933e-05,
+      "loss": 0.7933,
+      "step": 515630
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 1.586553914440958e-05,
+      "loss": 0.6931,
+      "step": 515640
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 1.5864391214509832e-05,
+      "loss": 0.7521,
+      "step": 515650
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 1.5863243284610086e-05,
+      "loss": 0.7756,
+      "step": 515660
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 1.586209535471034e-05,
+      "loss": 0.7471,
+      "step": 515670
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 1.5860947424810592e-05,
+      "loss": 0.7111,
+      "step": 515680
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 1.5859799494910845e-05,
+      "loss": 0.7701,
+      "step": 515690
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 1.5858651565011095e-05,
+      "loss": 0.857,
+      "step": 515700
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 1.585750363511135e-05,
+      "loss": 0.7993,
+      "step": 515710
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 1.58563557052116e-05,
+      "loss": 0.6615,
+      "step": 515720
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 1.585520777531185e-05,
+      "loss": 0.6792,
+      "step": 515730
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 1.5854059845412105e-05,
+      "loss": 0.7625,
+      "step": 515740
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 1.5852911915512358e-05,
+      "loss": 0.7361,
+      "step": 515750
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 1.585176398561261e-05,
+      "loss": 0.7296,
+      "step": 515760
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 1.5850616055712864e-05,
+      "loss": 0.7822,
+      "step": 515770
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 1.5849468125813114e-05,
+      "loss": 0.7234,
+      "step": 515780
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 1.5848320195913368e-05,
+      "loss": 0.857,
+      "step": 515790
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 1.584717226601362e-05,
+      "loss": 0.6956,
+      "step": 515800
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 1.5846024336113874e-05,
+      "loss": 0.737,
+      "step": 515810
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 1.5844876406214127e-05,
+      "loss": 0.8742,
+      "step": 515820
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 1.584372847631438e-05,
+      "loss": 0.7164,
+      "step": 515830
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 1.584258054641463e-05,
+      "loss": 0.7414,
+      "step": 515840
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 1.5841432616514884e-05,
+      "loss": 0.7145,
+      "step": 515850
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 1.5840284686615137e-05,
+      "loss": 0.7729,
+      "step": 515860
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 1.5839136756715387e-05,
+      "loss": 0.7075,
+      "step": 515870
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 1.583798882681564e-05,
+      "loss": 0.8486,
+      "step": 515880
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 1.5836840896915893e-05,
+      "loss": 0.8455,
+      "step": 515890
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 1.5835692967016146e-05,
+      "loss": 0.7983,
+      "step": 515900
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 1.58345450371164e-05,
+      "loss": 0.7296,
+      "step": 515910
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 1.5833397107216653e-05,
+      "loss": 0.8071,
+      "step": 515920
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 1.5832249177316903e-05,
+      "loss": 0.717,
+      "step": 515930
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 1.5831101247417156e-05,
+      "loss": 0.8022,
+      "step": 515940
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 1.582995331751741e-05,
+      "loss": 0.7158,
+      "step": 515950
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 1.5828805387617662e-05,
+      "loss": 0.681,
+      "step": 515960
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 1.5827657457717916e-05,
+      "loss": 0.7878,
+      "step": 515970
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 1.582650952781817e-05,
+      "loss": 0.7548,
+      "step": 515980
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 1.582536159791842e-05,
+      "loss": 0.7949,
+      "step": 515990
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 1.5824213668018672e-05,
+      "loss": 0.7894,
+      "step": 516000
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 1.5823065738118925e-05,
+      "loss": 0.7793,
+      "step": 516010
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 1.5821917808219175e-05,
+      "loss": 0.6429,
+      "step": 516020
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 1.582076987831943e-05,
+      "loss": 0.7206,
+      "step": 516030
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 1.581962194841968e-05,
+      "loss": 0.7269,
+      "step": 516040
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 1.5818474018519935e-05,
+      "loss": 0.7092,
+      "step": 516050
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 1.5817326088620188e-05,
+      "loss": 0.6662,
+      "step": 516060
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 1.581617815872044e-05,
+      "loss": 0.7737,
+      "step": 516070
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 1.581503022882069e-05,
+      "loss": 0.7842,
+      "step": 516080
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 1.5813882298920944e-05,
+      "loss": 0.7561,
+      "step": 516090
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 1.5812734369021198e-05,
+      "loss": 0.7554,
+      "step": 516100
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 1.581158643912145e-05,
+      "loss": 0.6949,
+      "step": 516110
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 1.5810438509221704e-05,
+      "loss": 0.7477,
+      "step": 516120
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 1.5809290579321954e-05,
+      "loss": 0.7311,
+      "step": 516130
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 1.5808142649422207e-05,
+      "loss": 0.7593,
+      "step": 516140
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 1.580699471952246e-05,
+      "loss": 0.7373,
+      "step": 516150
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 1.580584678962271e-05,
+      "loss": 0.7316,
+      "step": 516160
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 1.5804698859722964e-05,
+      "loss": 0.7308,
+      "step": 516170
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 1.5803550929823217e-05,
+      "loss": 0.8182,
+      "step": 516180
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 1.580240299992347e-05,
+      "loss": 0.7668,
+      "step": 516190
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 1.5801255070023723e-05,
+      "loss": 0.8038,
+      "step": 516200
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 1.5800107140123977e-05,
+      "loss": 0.7291,
+      "step": 516210
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 1.5798959210224226e-05,
+      "loss": 0.7019,
+      "step": 516220
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 1.579781128032448e-05,
+      "loss": 0.7879,
+      "step": 516230
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 1.5796663350424733e-05,
+      "loss": 0.6684,
+      "step": 516240
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 1.5795515420524986e-05,
+      "loss": 0.732,
+      "step": 516250
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 1.579436749062524e-05,
+      "loss": 0.8072,
+      "step": 516260
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 1.5793219560725493e-05,
+      "loss": 0.601,
+      "step": 516270
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 1.5792071630825743e-05,
+      "loss": 0.7524,
+      "step": 516280
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 1.5790923700925996e-05,
+      "loss": 0.7339,
+      "step": 516290
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 1.578977577102625e-05,
+      "loss": 0.7502,
+      "step": 516300
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 1.57886278411265e-05,
+      "loss": 0.8494,
+      "step": 516310
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 1.5787479911226752e-05,
+      "loss": 0.7423,
+      "step": 516320
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 1.5786331981327005e-05,
+      "loss": 0.7385,
+      "step": 516330
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 1.578518405142726e-05,
+      "loss": 0.7522,
+      "step": 516340
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 1.5784036121527512e-05,
+      "loss": 0.7071,
+      "step": 516350
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 1.5782888191627765e-05,
+      "loss": 0.7529,
+      "step": 516360
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 1.5781740261728015e-05,
+      "loss": 0.7151,
+      "step": 516370
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 1.5780592331828268e-05,
+      "loss": 0.716,
+      "step": 516380
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 1.577944440192852e-05,
+      "loss": 0.6427,
+      "step": 516390
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 1.5778296472028775e-05,
+      "loss": 0.7409,
+      "step": 516400
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 1.5777148542129028e-05,
+      "loss": 0.826,
+      "step": 516410
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 1.577600061222928e-05,
+      "loss": 0.7505,
+      "step": 516420
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 1.577485268232953e-05,
+      "loss": 0.7751,
+      "step": 516430
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 1.5773704752429784e-05,
+      "loss": 0.7617,
+      "step": 516440
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 1.5772556822530034e-05,
+      "loss": 0.7965,
+      "step": 516450
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 1.5771408892630287e-05,
+      "loss": 0.7399,
+      "step": 516460
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 1.577026096273054e-05,
+      "loss": 0.8125,
+      "step": 516470
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 1.5769113032830794e-05,
+      "loss": 0.7628,
+      "step": 516480
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 1.5767965102931047e-05,
+      "loss": 0.7136,
+      "step": 516490
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 1.57668171730313e-05,
+      "loss": 0.8129,
+      "step": 516500
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 1.576566924313155e-05,
+      "loss": 0.7512,
+      "step": 516510
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 1.5764521313231803e-05,
+      "loss": 0.7571,
+      "step": 516520
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 1.5763373383332057e-05,
+      "loss": 0.7327,
+      "step": 516530
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 1.576222545343231e-05,
+      "loss": 0.7642,
+      "step": 516540
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 1.5761077523532563e-05,
+      "loss": 0.6944,
+      "step": 516550
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 1.5759929593632816e-05,
+      "loss": 0.7069,
+      "step": 516560
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 1.5758781663733066e-05,
+      "loss": 0.7263,
+      "step": 516570
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 1.575763373383332e-05,
+      "loss": 0.7966,
+      "step": 516580
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 1.5756485803933573e-05,
+      "loss": 0.7512,
+      "step": 516590
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 1.5755337874033823e-05,
+      "loss": 0.7137,
+      "step": 516600
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 1.5754189944134076e-05,
+      "loss": 0.7583,
+      "step": 516610
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 1.575304201423433e-05,
+      "loss": 0.7824,
+      "step": 516620
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 1.5751894084334582e-05,
+      "loss": 0.7173,
+      "step": 516630
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 1.5750746154434836e-05,
+      "loss": 0.8199,
+      "step": 516640
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 1.574959822453509e-05,
+      "loss": 0.7929,
+      "step": 516650
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 1.574845029463534e-05,
+      "loss": 0.7494,
+      "step": 516660
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 1.5747302364735592e-05,
+      "loss": 0.765,
+      "step": 516670
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 1.5746154434835845e-05,
+      "loss": 0.7482,
+      "step": 516680
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 1.57450065049361e-05,
+      "loss": 0.768,
+      "step": 516690
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 1.574385857503635e-05,
+      "loss": 0.7513,
+      "step": 516700
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 1.5742710645136605e-05,
+      "loss": 0.8058,
+      "step": 516710
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 1.5741562715236855e-05,
+      "loss": 0.6728,
+      "step": 516720
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 1.5740414785337108e-05,
+      "loss": 0.8192,
+      "step": 516730
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 1.573926685543736e-05,
+      "loss": 0.7439,
+      "step": 516740
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 1.573811892553761e-05,
+      "loss": 0.7861,
+      "step": 516750
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 1.5736970995637864e-05,
+      "loss": 0.7079,
+      "step": 516760
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 1.5735823065738118e-05,
+      "loss": 0.7344,
+      "step": 516770
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 1.573467513583837e-05,
+      "loss": 0.7749,
+      "step": 516780
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 1.5733527205938624e-05,
+      "loss": 0.8186,
+      "step": 516790
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 1.5732379276038874e-05,
+      "loss": 0.7549,
+      "step": 516800
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 1.5731231346139127e-05,
+      "loss": 0.7344,
+      "step": 516810
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 1.573008341623938e-05,
+      "loss": 0.7461,
+      "step": 516820
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 1.5728935486339634e-05,
+      "loss": 0.7122,
+      "step": 516830
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 1.5727787556439887e-05,
+      "loss": 0.6991,
+      "step": 516840
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 1.572663962654014e-05,
+      "loss": 0.7487,
+      "step": 516850
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 1.572549169664039e-05,
+      "loss": 0.8124,
+      "step": 516860
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 1.5724343766740643e-05,
+      "loss": 0.6681,
+      "step": 516870
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 1.5723195836840896e-05,
+      "loss": 0.7132,
+      "step": 516880
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 1.5722047906941146e-05,
+      "loss": 0.7596,
+      "step": 516890
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 1.57208999770414e-05,
+      "loss": 0.8224,
+      "step": 516900
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 1.5719752047141653e-05,
+      "loss": 0.7637,
+      "step": 516910
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 1.5718604117241906e-05,
+      "loss": 0.7211,
+      "step": 516920
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 1.571745618734216e-05,
+      "loss": 0.8144,
+      "step": 516930
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 1.5716308257442412e-05,
+      "loss": 0.7919,
+      "step": 516940
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 1.5715160327542662e-05,
+      "loss": 0.7593,
+      "step": 516950
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 1.5714012397642916e-05,
+      "loss": 0.7413,
+      "step": 516960
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 1.571286446774317e-05,
+      "loss": 0.6543,
+      "step": 516970
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 1.5711716537843422e-05,
+      "loss": 0.7594,
+      "step": 516980
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 1.5710568607943675e-05,
+      "loss": 0.6839,
+      "step": 516990
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 1.570942067804393e-05,
+      "loss": 0.7005,
+      "step": 517000
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 1.570827274814418e-05,
+      "loss": 0.77,
+      "step": 517010
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 1.570712481824443e-05,
+      "loss": 0.7348,
+      "step": 517020
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 1.5705976888344685e-05,
+      "loss": 0.7858,
+      "step": 517030
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 1.5704828958444935e-05,
+      "loss": 0.6854,
+      "step": 517040
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 1.5703681028545188e-05,
+      "loss": 0.8024,
+      "step": 517050
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 1.570253309864544e-05,
+      "loss": 0.783,
+      "step": 517060
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 1.5701385168745694e-05,
+      "loss": 0.8101,
+      "step": 517070
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 1.5700237238845948e-05,
+      "loss": 0.7631,
+      "step": 517080
+    },
+    {
+      "epoch": 23.72,
+      "learning_rate": 1.56990893089462e-05,
+      "loss": 0.6965,
+      "step": 517090
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 1.569794137904645e-05,
+      "loss": 0.709,
+      "step": 517100
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 1.5696793449146704e-05,
+      "loss": 0.7591,
+      "step": 517110
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 1.5695645519246954e-05,
+      "loss": 0.7973,
+      "step": 517120
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 1.569449758934721e-05,
+      "loss": 0.7516,
+      "step": 517130
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 1.5693349659447464e-05,
+      "loss": 0.66,
+      "step": 517140
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 1.5692201729547714e-05,
+      "loss": 0.7402,
+      "step": 517150
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 1.5691053799647967e-05,
+      "loss": 0.7731,
+      "step": 517160
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 1.568990586974822e-05,
+      "loss": 0.7634,
+      "step": 517170
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 1.568875793984847e-05,
+      "loss": 0.7791,
+      "step": 517180
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 1.5687610009948723e-05,
+      "loss": 0.7936,
+      "step": 517190
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 1.5686462080048976e-05,
+      "loss": 0.8404,
+      "step": 517200
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 1.568531415014923e-05,
+      "loss": 0.737,
+      "step": 517210
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 1.5684166220249483e-05,
+      "loss": 0.8135,
+      "step": 517220
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 1.5683018290349736e-05,
+      "loss": 0.7166,
+      "step": 517230
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 1.5681870360449986e-05,
+      "loss": 0.7155,
+      "step": 517240
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 1.568072243055024e-05,
+      "loss": 0.7653,
+      "step": 517250
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 1.5679574500650493e-05,
+      "loss": 0.8041,
+      "step": 517260
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 1.5678426570750746e-05,
+      "loss": 0.7033,
+      "step": 517270
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 1.5677278640851e-05,
+      "loss": 0.7572,
+      "step": 517280
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 1.5676130710951252e-05,
+      "loss": 0.7593,
+      "step": 517290
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 1.5674982781051502e-05,
+      "loss": 0.7249,
+      "step": 517300
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 1.5673834851151755e-05,
+      "loss": 0.7972,
+      "step": 517310
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 1.567268692125201e-05,
+      "loss": 0.7776,
+      "step": 517320
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 1.567153899135226e-05,
+      "loss": 0.6806,
+      "step": 517330
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 1.567039106145251e-05,
+      "loss": 0.7496,
+      "step": 517340
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 1.5669243131552765e-05,
+      "loss": 0.7433,
+      "step": 517350
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 1.5668095201653018e-05,
+      "loss": 0.7165,
+      "step": 517360
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 1.566694727175327e-05,
+      "loss": 0.7502,
+      "step": 517370
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 1.5665799341853525e-05,
+      "loss": 0.7553,
+      "step": 517380
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 1.5664651411953775e-05,
+      "loss": 0.7203,
+      "step": 517390
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 1.5663503482054028e-05,
+      "loss": 0.7152,
+      "step": 517400
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 1.566235555215428e-05,
+      "loss": 0.7338,
+      "step": 517410
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 1.5661207622254534e-05,
+      "loss": 0.7481,
+      "step": 517420
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 1.5660059692354787e-05,
+      "loss": 0.8124,
+      "step": 517430
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 1.565891176245504e-05,
+      "loss": 0.7095,
+      "step": 517440
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 1.565776383255529e-05,
+      "loss": 0.7493,
+      "step": 517450
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 1.5656615902655544e-05,
+      "loss": 0.711,
+      "step": 517460
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 1.5655467972755794e-05,
+      "loss": 0.7779,
+      "step": 517470
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 1.5654320042856047e-05,
+      "loss": 0.7216,
+      "step": 517480
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 1.56531721129563e-05,
+      "loss": 0.7729,
+      "step": 517490
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 1.5652024183056553e-05,
+      "loss": 0.7378,
+      "step": 517500
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 1.5650876253156807e-05,
+      "loss": 0.7571,
+      "step": 517510
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 1.564972832325706e-05,
+      "loss": 0.7149,
+      "step": 517520
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 1.564858039335731e-05,
+      "loss": 0.7353,
+      "step": 517530
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 1.5647432463457563e-05,
+      "loss": 0.7554,
+      "step": 517540
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 1.5646284533557816e-05,
+      "loss": 0.7645,
+      "step": 517550
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 1.5645136603658066e-05,
+      "loss": 0.7601,
+      "step": 517560
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 1.5643988673758323e-05,
+      "loss": 0.8153,
+      "step": 517570
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 1.5642840743858576e-05,
+      "loss": 0.7093,
+      "step": 517580
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 1.5641692813958826e-05,
+      "loss": 0.7586,
+      "step": 517590
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 1.564054488405908e-05,
+      "loss": 0.6948,
+      "step": 517600
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 1.5639396954159332e-05,
+      "loss": 0.7058,
+      "step": 517610
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 1.5638249024259582e-05,
+      "loss": 0.7314,
+      "step": 517620
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 1.5637101094359835e-05,
+      "loss": 0.7179,
+      "step": 517630
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 1.563595316446009e-05,
+      "loss": 0.7694,
+      "step": 517640
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 1.5634805234560342e-05,
+      "loss": 0.7659,
+      "step": 517650
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 1.5633657304660595e-05,
+      "loss": 0.7338,
+      "step": 517660
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 1.563250937476085e-05,
+      "loss": 0.7469,
+      "step": 517670
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 1.5631361444861098e-05,
+      "loss": 0.8084,
+      "step": 517680
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 1.563021351496135e-05,
+      "loss": 0.7792,
+      "step": 517690
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 1.5629065585061605e-05,
+      "loss": 0.7206,
+      "step": 517700
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 1.5627917655161855e-05,
+      "loss": 0.788,
+      "step": 517710
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 1.562676972526211e-05,
+      "loss": 0.7487,
+      "step": 517720
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 1.5625621795362364e-05,
+      "loss": 0.733,
+      "step": 517730
+    },
+    {
+      "epoch": 23.75,
+      "learning_rate": 1.5624473865462614e-05,
+      "loss": 0.7192,
+      "step": 517740
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 1.5623325935562868e-05,
+      "loss": 0.771,
+      "step": 517750
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 1.562217800566312e-05,
+      "loss": 0.7347,
+      "step": 517760
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 1.562103007576337e-05,
+      "loss": 0.7641,
+      "step": 517770
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 1.5619882145863624e-05,
+      "loss": 0.7396,
+      "step": 517780
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 1.5618734215963877e-05,
+      "loss": 0.7739,
+      "step": 517790
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 1.561758628606413e-05,
+      "loss": 0.7297,
+      "step": 517800
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 1.5616438356164384e-05,
+      "loss": 0.8145,
+      "step": 517810
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 1.5615290426264633e-05,
+      "loss": 0.7318,
+      "step": 517820
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 1.5614142496364887e-05,
+      "loss": 0.7569,
+      "step": 517830
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 1.561299456646514e-05,
+      "loss": 0.8283,
+      "step": 517840
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 1.561184663656539e-05,
+      "loss": 0.7586,
+      "step": 517850
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 1.5610698706665643e-05,
+      "loss": 0.7144,
+      "step": 517860
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 1.56095507767659e-05,
+      "loss": 0.7433,
+      "step": 517870
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 1.560840284686615e-05,
+      "loss": 0.7741,
+      "step": 517880
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 1.5607254916966403e-05,
+      "loss": 0.7131,
+      "step": 517890
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 1.5606106987066656e-05,
+      "loss": 0.7215,
+      "step": 517900
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 1.5604959057166906e-05,
+      "loss": 0.7151,
+      "step": 517910
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 1.560381112726716e-05,
+      "loss": 0.7854,
+      "step": 517920
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 1.5602663197367412e-05,
+      "loss": 0.7575,
+      "step": 517930
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 1.5601515267467666e-05,
+      "loss": 0.7728,
+      "step": 517940
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 1.560036733756792e-05,
+      "loss": 0.6422,
+      "step": 517950
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 1.5599219407668172e-05,
+      "loss": 0.6917,
+      "step": 517960
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 1.5598071477768422e-05,
+      "loss": 0.8589,
+      "step": 517970
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 1.5596923547868675e-05,
+      "loss": 0.7847,
+      "step": 517980
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 1.559577561796893e-05,
+      "loss": 0.735,
+      "step": 517990
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 1.5594627688069178e-05,
+      "loss": 0.724,
+      "step": 518000
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 1.559347975816943e-05,
+      "loss": 0.8399,
+      "step": 518010
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 1.5592331828269688e-05,
+      "loss": 0.7704,
+      "step": 518020
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 1.5591183898369938e-05,
+      "loss": 0.7655,
+      "step": 518030
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 1.559003596847019e-05,
+      "loss": 0.7528,
+      "step": 518040
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 1.5588888038570444e-05,
+      "loss": 0.7519,
+      "step": 518050
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 1.5587740108670694e-05,
+      "loss": 0.7272,
+      "step": 518060
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 1.5586592178770948e-05,
+      "loss": 0.7214,
+      "step": 518070
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 1.55854442488712e-05,
+      "loss": 0.7601,
+      "step": 518080
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 1.5584296318971454e-05,
+      "loss": 0.6701,
+      "step": 518090
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 1.5583148389071707e-05,
+      "loss": 0.7319,
+      "step": 518100
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 1.558200045917196e-05,
+      "loss": 0.7772,
+      "step": 518110
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 1.558085252927221e-05,
+      "loss": 0.8122,
+      "step": 518120
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 1.5579704599372464e-05,
+      "loss": 0.6736,
+      "step": 518130
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 1.5578556669472713e-05,
+      "loss": 0.8172,
+      "step": 518140
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 1.5577408739572967e-05,
+      "loss": 0.7582,
+      "step": 518150
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 1.557626080967322e-05,
+      "loss": 0.6774,
+      "step": 518160
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 1.5575112879773473e-05,
+      "loss": 0.7683,
+      "step": 518170
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 1.5573964949873726e-05,
+      "loss": 0.7971,
+      "step": 518180
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 1.557281701997398e-05,
+      "loss": 0.7387,
+      "step": 518190
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 1.557166909007423e-05,
+      "loss": 0.7208,
+      "step": 518200
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 1.5570521160174483e-05,
+      "loss": 0.8533,
+      "step": 518210
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 1.5569373230274736e-05,
+      "loss": 0.7424,
+      "step": 518220
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 1.556822530037499e-05,
+      "loss": 0.7064,
+      "step": 518230
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 1.5567077370475243e-05,
+      "loss": 0.7274,
+      "step": 518240
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 1.5565929440575496e-05,
+      "loss": 0.7903,
+      "step": 518250
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 1.5564781510675746e-05,
+      "loss": 0.7617,
+      "step": 518260
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 1.5563633580776e-05,
+      "loss": 0.7559,
+      "step": 518270
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 1.5562485650876252e-05,
+      "loss": 0.7628,
+      "step": 518280
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 1.5561337720976502e-05,
+      "loss": 0.6689,
+      "step": 518290
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 1.5560189791076755e-05,
+      "loss": 0.7434,
+      "step": 518300
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 1.555904186117701e-05,
+      "loss": 0.748,
+      "step": 518310
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 1.555789393127726e-05,
+      "loss": 0.699,
+      "step": 518320
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 1.5556746001377515e-05,
+      "loss": 0.7675,
+      "step": 518330
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 1.5555598071477768e-05,
+      "loss": 0.6979,
+      "step": 518340
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 1.5554450141578018e-05,
+      "loss": 0.7367,
+      "step": 518350
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 1.555330221167827e-05,
+      "loss": 0.7687,
+      "step": 518360
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 1.5552154281778525e-05,
+      "loss": 0.8686,
+      "step": 518370
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 1.5551006351878778e-05,
+      "loss": 0.7678,
+      "step": 518380
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 1.554985842197903e-05,
+      "loss": 0.7669,
+      "step": 518390
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 1.5548710492079284e-05,
+      "loss": 0.6526,
+      "step": 518400
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 1.5547562562179534e-05,
+      "loss": 0.7775,
+      "step": 518410
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 1.5546414632279787e-05,
+      "loss": 0.7363,
+      "step": 518420
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 1.554526670238004e-05,
+      "loss": 0.7866,
+      "step": 518430
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 1.554411877248029e-05,
+      "loss": 0.702,
+      "step": 518440
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 1.5542970842580544e-05,
+      "loss": 0.7641,
+      "step": 518450
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 1.5541822912680797e-05,
+      "loss": 0.7437,
+      "step": 518460
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 1.554067498278105e-05,
+      "loss": 0.6973,
+      "step": 518470
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 1.5539527052881303e-05,
+      "loss": 0.7486,
+      "step": 518480
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 1.5538379122981553e-05,
+      "loss": 0.7943,
+      "step": 518490
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 1.5537231193081806e-05,
+      "loss": 0.7431,
+      "step": 518500
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 1.553608326318206e-05,
+      "loss": 0.7931,
+      "step": 518510
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 1.5534935333282313e-05,
+      "loss": 0.7912,
+      "step": 518520
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 1.5533787403382566e-05,
+      "loss": 0.7159,
+      "step": 518530
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 1.553263947348282e-05,
+      "loss": 0.7052,
+      "step": 518540
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 1.553149154358307e-05,
+      "loss": 0.8001,
+      "step": 518550
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 1.5530343613683323e-05,
+      "loss": 0.7314,
+      "step": 518560
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 1.5529195683783576e-05,
+      "loss": 0.7981,
+      "step": 518570
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 1.5528047753883826e-05,
+      "loss": 0.7359,
+      "step": 518580
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 1.552689982398408e-05,
+      "loss": 0.7121,
+      "step": 518590
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 1.5525751894084332e-05,
+      "loss": 0.7854,
+      "step": 518600
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 1.5524603964184585e-05,
+      "loss": 0.7121,
+      "step": 518610
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 1.552345603428484e-05,
+      "loss": 0.6924,
+      "step": 518620
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 1.5522308104385092e-05,
+      "loss": 0.7738,
+      "step": 518630
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 1.5521160174485342e-05,
+      "loss": 0.7995,
+      "step": 518640
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 1.5520012244585595e-05,
+      "loss": 0.6581,
+      "step": 518650
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 1.5518864314685848e-05,
+      "loss": 0.7943,
+      "step": 518660
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 1.55177163847861e-05,
+      "loss": 0.8153,
+      "step": 518670
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 1.5516568454886355e-05,
+      "loss": 0.7791,
+      "step": 518680
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 1.5515420524986608e-05,
+      "loss": 0.7561,
+      "step": 518690
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 1.5514272595086858e-05,
+      "loss": 0.7577,
+      "step": 518700
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 1.551312466518711e-05,
+      "loss": 0.8,
+      "step": 518710
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 1.5511976735287364e-05,
+      "loss": 0.7974,
+      "step": 518720
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 1.5510828805387614e-05,
+      "loss": 0.7534,
+      "step": 518730
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 1.5509680875487867e-05,
+      "loss": 0.7648,
+      "step": 518740
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 1.550853294558812e-05,
+      "loss": 0.7952,
+      "step": 518750
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 1.5507385015688374e-05,
+      "loss": 0.7757,
+      "step": 518760
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 1.5506237085788627e-05,
+      "loss": 0.7601,
+      "step": 518770
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 1.550508915588888e-05,
+      "loss": 0.8036,
+      "step": 518780
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 1.550394122598913e-05,
+      "loss": 0.6948,
+      "step": 518790
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 1.5502793296089383e-05,
+      "loss": 0.7557,
+      "step": 518800
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 1.5501645366189637e-05,
+      "loss": 0.7738,
+      "step": 518810
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 1.550049743628989e-05,
+      "loss": 0.75,
+      "step": 518820
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 1.5499349506390143e-05,
+      "loss": 0.7883,
+      "step": 518830
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 1.5498201576490393e-05,
+      "loss": 0.7992,
+      "step": 518840
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 1.5497053646590646e-05,
+      "loss": 0.7679,
+      "step": 518850
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 1.54959057166909e-05,
+      "loss": 0.8113,
+      "step": 518860
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 1.549475778679115e-05,
+      "loss": 0.7207,
+      "step": 518870
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 1.5493609856891403e-05,
+      "loss": 0.7274,
+      "step": 518880
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 1.5492461926991656e-05,
+      "loss": 0.7059,
+      "step": 518890
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 1.549131399709191e-05,
+      "loss": 0.7292,
+      "step": 518900
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 1.5490166067192162e-05,
+      "loss": 0.7199,
+      "step": 518910
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 1.5489018137292416e-05,
+      "loss": 0.7188,
+      "step": 518920
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 1.5487870207392665e-05,
+      "loss": 0.8602,
+      "step": 518930
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 1.548672227749292e-05,
+      "loss": 0.7398,
+      "step": 518940
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 1.5485574347593172e-05,
+      "loss": 0.7869,
+      "step": 518950
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 1.5484426417693425e-05,
+      "loss": 0.7144,
+      "step": 518960
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 1.548327848779368e-05,
+      "loss": 0.7366,
+      "step": 518970
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 1.548213055789393e-05,
+      "loss": 0.7165,
+      "step": 518980
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 1.548098262799418e-05,
+      "loss": 0.8116,
+      "step": 518990
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 1.5479834698094435e-05,
+      "loss": 0.8516,
+      "step": 519000
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 1.5478686768194688e-05,
+      "loss": 0.7435,
+      "step": 519010
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 1.5477538838294938e-05,
+      "loss": 0.7478,
+      "step": 519020
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 1.547639090839519e-05,
+      "loss": 0.8059,
+      "step": 519030
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 1.5475242978495444e-05,
+      "loss": 0.7356,
+      "step": 519040
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 1.5474095048595698e-05,
+      "loss": 0.7502,
+      "step": 519050
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 1.547294711869595e-05,
+      "loss": 0.7754,
+      "step": 519060
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 1.5471799188796204e-05,
+      "loss": 0.6599,
+      "step": 519070
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 1.5470651258896454e-05,
+      "loss": 0.7405,
+      "step": 519080
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 1.5469503328996707e-05,
+      "loss": 0.7578,
+      "step": 519090
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 1.546835539909696e-05,
+      "loss": 0.8084,
+      "step": 519100
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 1.5467207469197214e-05,
+      "loss": 0.7668,
+      "step": 519110
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 1.5466059539297467e-05,
+      "loss": 0.7569,
+      "step": 519120
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 1.546491160939772e-05,
+      "loss": 0.7369,
+      "step": 519130
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 1.546376367949797e-05,
+      "loss": 0.6483,
+      "step": 519140
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 1.5462615749598223e-05,
+      "loss": 0.7714,
+      "step": 519150
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 1.5461467819698476e-05,
+      "loss": 0.7498,
+      "step": 519160
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 1.5460319889798726e-05,
+      "loss": 0.8295,
+      "step": 519170
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 1.545917195989898e-05,
+      "loss": 0.7798,
+      "step": 519180
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 1.5458024029999233e-05,
+      "loss": 0.7246,
+      "step": 519190
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 1.5456876100099486e-05,
+      "loss": 0.757,
+      "step": 519200
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 1.545572817019974e-05,
+      "loss": 0.6828,
+      "step": 519210
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 1.545458024029999e-05,
+      "loss": 0.6487,
+      "step": 519220
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 1.5453432310400242e-05,
+      "loss": 0.7413,
+      "step": 519230
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 1.5452284380500496e-05,
+      "loss": 0.8003,
+      "step": 519240
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 1.545113645060075e-05,
+      "loss": 0.7564,
+      "step": 519250
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 1.5449988520701002e-05,
+      "loss": 0.7935,
+      "step": 519260
+    },
+    {
+      "epoch": 23.82,
+      "learning_rate": 1.5448840590801255e-05,
+      "loss": 0.7412,
+      "step": 519270
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 1.5447692660901505e-05,
+      "loss": 0.6538,
+      "step": 519280
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 1.544654473100176e-05,
+      "loss": 0.7608,
+      "step": 519290
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 1.544539680110201e-05,
+      "loss": 0.7307,
+      "step": 519300
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 1.544424887120226e-05,
+      "loss": 0.7969,
+      "step": 519310
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 1.5443100941302515e-05,
+      "loss": 0.7363,
+      "step": 519320
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 1.5441953011402768e-05,
+      "loss": 0.7126,
+      "step": 519330
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 1.544080508150302e-05,
+      "loss": 0.7516,
+      "step": 519340
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 1.5439657151603274e-05,
+      "loss": 0.7629,
+      "step": 519350
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 1.5438509221703528e-05,
+      "loss": 0.7679,
+      "step": 519360
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 1.5437361291803778e-05,
+      "loss": 0.8077,
+      "step": 519370
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 1.543621336190403e-05,
+      "loss": 0.7538,
+      "step": 519380
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 1.5435065432004284e-05,
+      "loss": 0.8146,
+      "step": 519390
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 1.5433917502104537e-05,
+      "loss": 0.7733,
+      "step": 519400
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 1.543276957220479e-05,
+      "loss": 0.8448,
+      "step": 519410
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 1.5431621642305044e-05,
+      "loss": 0.8323,
+      "step": 519420
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 1.5430473712405294e-05,
+      "loss": 0.728,
+      "step": 519430
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 1.5429325782505547e-05,
+      "loss": 0.7832,
+      "step": 519440
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 1.54281778526058e-05,
+      "loss": 0.7663,
+      "step": 519450
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 1.542702992270605e-05,
+      "loss": 0.7401,
+      "step": 519460
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 1.5425881992806303e-05,
+      "loss": 0.7541,
+      "step": 519470
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 1.5424734062906556e-05,
+      "loss": 0.6689,
+      "step": 519480
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 1.542358613300681e-05,
+      "loss": 0.8482,
+      "step": 519490
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 1.5422438203107063e-05,
+      "loss": 0.7038,
+      "step": 519500
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 1.5421290273207313e-05,
+      "loss": 0.7197,
+      "step": 519510
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 1.5420142343307566e-05,
+      "loss": 0.7516,
+      "step": 519520
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 1.541899441340782e-05,
+      "loss": 0.7421,
+      "step": 519530
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 1.5417846483508073e-05,
+      "loss": 0.7377,
+      "step": 519540
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 1.5416698553608326e-05,
+      "loss": 0.8058,
+      "step": 519550
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 1.541555062370858e-05,
+      "loss": 0.7582,
+      "step": 519560
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 1.541440269380883e-05,
+      "loss": 0.7532,
+      "step": 519570
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 1.5413254763909082e-05,
+      "loss": 0.8243,
+      "step": 519580
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 1.5412106834009335e-05,
+      "loss": 0.7483,
+      "step": 519590
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 1.5410958904109585e-05,
+      "loss": 0.7709,
+      "step": 519600
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 1.540981097420984e-05,
+      "loss": 0.8305,
+      "step": 519610
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 1.5408663044310092e-05,
+      "loss": 0.7307,
+      "step": 519620
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 1.5407515114410345e-05,
+      "loss": 0.7859,
+      "step": 519630
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 1.5406367184510598e-05,
+      "loss": 0.703,
+      "step": 519640
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 1.540521925461085e-05,
+      "loss": 0.7256,
+      "step": 519650
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 1.54040713247111e-05,
+      "loss": 0.8449,
+      "step": 519660
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 1.5402923394811355e-05,
+      "loss": 0.764,
+      "step": 519670
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 1.5401775464911608e-05,
+      "loss": 0.8208,
+      "step": 519680
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 1.540062753501186e-05,
+      "loss": 0.8118,
+      "step": 519690
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 1.5399479605112114e-05,
+      "loss": 0.6599,
+      "step": 519700
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 1.5398331675212368e-05,
+      "loss": 0.7418,
+      "step": 519710
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 1.5397183745312617e-05,
+      "loss": 0.8177,
+      "step": 519720
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 1.539603581541287e-05,
+      "loss": 0.8122,
+      "step": 519730
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 1.5394887885513124e-05,
+      "loss": 0.7109,
+      "step": 519740
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 1.5393739955613374e-05,
+      "loss": 0.8054,
+      "step": 519750
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 1.5392592025713627e-05,
+      "loss": 0.769,
+      "step": 519760
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 1.539144409581388e-05,
+      "loss": 0.8101,
+      "step": 519770
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 1.5390296165914133e-05,
+      "loss": 0.7648,
+      "step": 519780
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 1.5389148236014387e-05,
+      "loss": 0.7646,
+      "step": 519790
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 1.538800030611464e-05,
+      "loss": 0.7813,
+      "step": 519800
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 1.538685237621489e-05,
+      "loss": 0.6835,
+      "step": 519810
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 1.5385704446315143e-05,
+      "loss": 0.7945,
+      "step": 519820
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 1.5384556516415396e-05,
+      "loss": 0.7747,
+      "step": 519830
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 1.538340858651565e-05,
+      "loss": 0.7838,
+      "step": 519840
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 1.5382260656615903e-05,
+      "loss": 0.7746,
+      "step": 519850
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 1.5381112726716153e-05,
+      "loss": 0.7429,
+      "step": 519860
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 1.5379964796816406e-05,
+      "loss": 0.8003,
+      "step": 519870
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 1.537881686691666e-05,
+      "loss": 0.7827,
+      "step": 519880
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 1.537766893701691e-05,
+      "loss": 0.8082,
+      "step": 519890
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 1.5376521007117162e-05,
+      "loss": 0.7249,
+      "step": 519900
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 1.5375373077217415e-05,
+      "loss": 0.7434,
+      "step": 519910
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 1.537422514731767e-05,
+      "loss": 0.7247,
+      "step": 519920
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 1.5373077217417922e-05,
+      "loss": 0.8378,
+      "step": 519930
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 1.5371929287518175e-05,
+      "loss": 0.7327,
+      "step": 519940
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 1.5370781357618425e-05,
+      "loss": 0.732,
+      "step": 519950
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 1.5369633427718678e-05,
+      "loss": 0.7808,
+      "step": 519960
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 1.536848549781893e-05,
+      "loss": 0.7686,
+      "step": 519970
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 1.5367337567919185e-05,
+      "loss": 0.7399,
+      "step": 519980
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 1.5366189638019438e-05,
+      "loss": 0.6908,
+      "step": 519990
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 1.536504170811969e-05,
+      "loss": 0.7747,
+      "step": 520000
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 1.536389377821994e-05,
+      "loss": 0.7844,
+      "step": 520010
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 1.5362745848320194e-05,
+      "loss": 0.7582,
+      "step": 520020
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 1.5361597918420448e-05,
+      "loss": 0.7336,
+      "step": 520030
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 1.5360449988520697e-05,
+      "loss": 0.7448,
+      "step": 520040
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 1.535930205862095e-05,
+      "loss": 0.7879,
+      "step": 520050
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 1.5358154128721204e-05,
+      "loss": 0.6754,
+      "step": 520060
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 1.5357006198821457e-05,
+      "loss": 0.7684,
+      "step": 520070
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 1.535585826892171e-05,
+      "loss": 0.8569,
+      "step": 520080
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 1.5354710339021964e-05,
+      "loss": 0.8204,
+      "step": 520090
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 1.5353562409122213e-05,
+      "loss": 0.807,
+      "step": 520100
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 1.5352414479222467e-05,
+      "loss": 0.7,
+      "step": 520110
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 1.535126654932272e-05,
+      "loss": 0.7646,
+      "step": 520120
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 1.5350118619422973e-05,
+      "loss": 0.663,
+      "step": 520130
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 1.5348970689523226e-05,
+      "loss": 0.6954,
+      "step": 520140
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 1.534782275962348e-05,
+      "loss": 0.7054,
+      "step": 520150
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 1.534667482972373e-05,
+      "loss": 0.793,
+      "step": 520160
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 1.5345526899823983e-05,
+      "loss": 0.6602,
+      "step": 520170
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 1.5344378969924236e-05,
+      "loss": 0.7471,
+      "step": 520180
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 1.5343231040024486e-05,
+      "loss": 0.7511,
+      "step": 520190
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 1.534208311012474e-05,
+      "loss": 0.6731,
+      "step": 520200
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 1.5340935180224992e-05,
+      "loss": 0.6804,
+      "step": 520210
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 1.5339787250325246e-05,
+      "loss": 0.8101,
+      "step": 520220
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 1.53386393204255e-05,
+      "loss": 0.7684,
+      "step": 520230
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 1.533749139052575e-05,
+      "loss": 0.8679,
+      "step": 520240
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 1.5336343460626002e-05,
+      "loss": 0.7892,
+      "step": 520250
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 1.5335195530726255e-05,
+      "loss": 0.7236,
+      "step": 520260
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 1.533404760082651e-05,
+      "loss": 0.727,
+      "step": 520270
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 1.533289967092676e-05,
+      "loss": 0.7309,
+      "step": 520280
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 1.5331751741027015e-05,
+      "loss": 0.8019,
+      "step": 520290
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 1.5330603811127265e-05,
+      "loss": 0.7434,
+      "step": 520300
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 1.5329455881227518e-05,
+      "loss": 0.7053,
+      "step": 520310
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 1.532830795132777e-05,
+      "loss": 0.8089,
+      "step": 520320
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 1.532716002142802e-05,
+      "loss": 0.8199,
+      "step": 520330
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 1.5326012091528274e-05,
+      "loss": 0.768,
+      "step": 520340
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 1.5324864161628528e-05,
+      "loss": 0.7208,
+      "step": 520350
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 1.532371623172878e-05,
+      "loss": 0.7574,
+      "step": 520360
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 1.5322568301829034e-05,
+      "loss": 0.8246,
+      "step": 520370
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 1.5321420371929287e-05,
+      "loss": 0.7832,
+      "step": 520380
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 1.5320272442029537e-05,
+      "loss": 0.7925,
+      "step": 520390
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 1.531912451212979e-05,
+      "loss": 0.7835,
+      "step": 520400
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 1.5317976582230044e-05,
+      "loss": 0.7594,
+      "step": 520410
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 1.5316828652330297e-05,
+      "loss": 0.8031,
+      "step": 520420
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 1.531568072243055e-05,
+      "loss": 0.6949,
+      "step": 520430
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 1.5314532792530803e-05,
+      "loss": 0.7566,
+      "step": 520440
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 1.5313384862631053e-05,
+      "loss": 0.6904,
+      "step": 520450
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 1.5312236932731306e-05,
+      "loss": 0.7709,
+      "step": 520460
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 1.531108900283156e-05,
+      "loss": 0.7412,
+      "step": 520470
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 1.530994107293181e-05,
+      "loss": 0.8006,
+      "step": 520480
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 1.5308793143032063e-05,
+      "loss": 0.7082,
+      "step": 520490
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 1.5307645213132316e-05,
+      "loss": 0.8479,
+      "step": 520500
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 1.530649728323257e-05,
+      "loss": 0.7019,
+      "step": 520510
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 1.5305349353332823e-05,
+      "loss": 0.7857,
+      "step": 520520
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 1.5304201423433076e-05,
+      "loss": 0.805,
+      "step": 520530
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 1.5303053493533326e-05,
+      "loss": 0.681,
+      "step": 520540
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 1.530190556363358e-05,
+      "loss": 0.705,
+      "step": 520550
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 1.5300757633733832e-05,
+      "loss": 0.8557,
+      "step": 520560
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 1.5299609703834085e-05,
+      "loss": 0.7453,
+      "step": 520570
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 1.529846177393434e-05,
+      "loss": 0.8689,
+      "step": 520580
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 1.529731384403459e-05,
+      "loss": 0.6981,
+      "step": 520590
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 1.5296165914134842e-05,
+      "loss": 0.6721,
+      "step": 520600
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 1.5295017984235095e-05,
+      "loss": 0.6941,
+      "step": 520610
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 1.5293870054335345e-05,
+      "loss": 0.6789,
+      "step": 520620
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 1.5292722124435598e-05,
+      "loss": 0.75,
+      "step": 520630
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 1.529157419453585e-05,
+      "loss": 0.7196,
+      "step": 520640
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 1.5290426264636105e-05,
+      "loss": 0.7297,
+      "step": 520650
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 1.5289278334736358e-05,
+      "loss": 0.7689,
+      "step": 520660
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 1.528813040483661e-05,
+      "loss": 0.7836,
+      "step": 520670
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 1.528698247493686e-05,
+      "loss": 0.707,
+      "step": 520680
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 1.5285834545037114e-05,
+      "loss": 0.7,
+      "step": 520690
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 1.5284686615137367e-05,
+      "loss": 0.7918,
+      "step": 520700
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 1.528353868523762e-05,
+      "loss": 0.7197,
+      "step": 520710
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 1.5282390755337874e-05,
+      "loss": 0.7713,
+      "step": 520720
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 1.5281242825438127e-05,
+      "loss": 0.7463,
+      "step": 520730
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 1.5280094895538377e-05,
+      "loss": 0.7686,
+      "step": 520740
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 1.527894696563863e-05,
+      "loss": 0.6512,
+      "step": 520750
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 1.5277799035738883e-05,
+      "loss": 0.8331,
+      "step": 520760
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 1.5276651105839133e-05,
+      "loss": 0.7184,
+      "step": 520770
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 1.5275503175939387e-05,
+      "loss": 0.8164,
+      "step": 520780
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 1.527435524603964e-05,
+      "loss": 0.7689,
+      "step": 520790
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 1.5273207316139893e-05,
+      "loss": 0.7846,
+      "step": 520800
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 1.5272059386240146e-05,
+      "loss": 0.7457,
+      "step": 520810
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 1.52709114563404e-05,
+      "loss": 0.7935,
+      "step": 520820
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 1.526976352644065e-05,
+      "loss": 0.7942,
+      "step": 520830
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 1.5268615596540903e-05,
+      "loss": 0.7972,
+      "step": 520840
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 1.5267467666641156e-05,
+      "loss": 0.7593,
+      "step": 520850
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 1.526631973674141e-05,
+      "loss": 0.8342,
+      "step": 520860
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 1.5265171806841662e-05,
+      "loss": 0.7104,
+      "step": 520870
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 1.5264023876941916e-05,
+      "loss": 0.7648,
+      "step": 520880
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 1.5262875947042165e-05,
+      "loss": 0.7615,
+      "step": 520890
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 1.526172801714242e-05,
+      "loss": 0.8068,
+      "step": 520900
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 1.526058008724267e-05,
+      "loss": 0.6959,
+      "step": 520910
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 1.5259432157342922e-05,
+      "loss": 0.7482,
+      "step": 520920
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 1.5258284227443177e-05,
+      "loss": 0.745,
+      "step": 520930
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 1.5257136297543428e-05,
+      "loss": 0.7414,
+      "step": 520940
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 1.5255988367643681e-05,
+      "loss": 0.6734,
+      "step": 520950
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 1.5254840437743935e-05,
+      "loss": 0.8099,
+      "step": 520960
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 1.5253692507844186e-05,
+      "loss": 0.7668,
+      "step": 520970
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 1.525254457794444e-05,
+      "loss": 0.7593,
+      "step": 520980
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 1.5251396648044693e-05,
+      "loss": 0.7173,
+      "step": 520990
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 1.5250248718144943e-05,
+      "loss": 0.7567,
+      "step": 521000
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 1.5249100788245196e-05,
+      "loss": 0.6783,
+      "step": 521010
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 1.5247952858345449e-05,
+      "loss": 0.83,
+      "step": 521020
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 1.52468049284457e-05,
+      "loss": 0.8391,
+      "step": 521030
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 1.5245656998545954e-05,
+      "loss": 0.7465,
+      "step": 521040
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 1.5244509068646207e-05,
+      "loss": 0.7859,
+      "step": 521050
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 1.5243361138746459e-05,
+      "loss": 0.743,
+      "step": 521060
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 1.5242213208846712e-05,
+      "loss": 0.7071,
+      "step": 521070
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 1.5241065278946965e-05,
+      "loss": 0.7038,
+      "step": 521080
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 1.5239917349047217e-05,
+      "loss": 0.7458,
+      "step": 521090
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 1.523876941914747e-05,
+      "loss": 0.7146,
+      "step": 521100
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 1.5237621489247723e-05,
+      "loss": 0.713,
+      "step": 521110
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 1.5236473559347975e-05,
+      "loss": 0.7315,
+      "step": 521120
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 1.5235325629448228e-05,
+      "loss": 0.7207,
+      "step": 521130
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 1.5234177699548481e-05,
+      "loss": 0.7134,
+      "step": 521140
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 1.5233029769648731e-05,
+      "loss": 0.7749,
+      "step": 521150
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 1.5231881839748984e-05,
+      "loss": 0.77,
+      "step": 521160
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 1.5230733909849238e-05,
+      "loss": 0.7404,
+      "step": 521170
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 1.5229585979949489e-05,
+      "loss": 0.7806,
+      "step": 521180
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 1.5228438050049742e-05,
+      "loss": 0.8197,
+      "step": 521190
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 1.5227290120149996e-05,
+      "loss": 0.7903,
+      "step": 521200
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 1.5226142190250247e-05,
+      "loss": 0.7906,
+      "step": 521210
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 1.52249942603505e-05,
+      "loss": 0.8044,
+      "step": 521220
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 1.5223846330450752e-05,
+      "loss": 0.7601,
+      "step": 521230
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 1.5222698400551005e-05,
+      "loss": 0.7896,
+      "step": 521240
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 1.5221550470651258e-05,
+      "loss": 0.7667,
+      "step": 521250
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 1.522040254075151e-05,
+      "loss": 0.8247,
+      "step": 521260
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 1.5219254610851763e-05,
+      "loss": 0.6469,
+      "step": 521270
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 1.5218106680952016e-05,
+      "loss": 0.7453,
+      "step": 521280
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 1.5216958751052266e-05,
+      "loss": 0.8019,
+      "step": 521290
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 1.521581082115252e-05,
+      "loss": 0.7889,
+      "step": 521300
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 1.5214662891252773e-05,
+      "loss": 0.7458,
+      "step": 521310
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 1.5213514961353024e-05,
+      "loss": 0.859,
+      "step": 521320
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 1.5212367031453278e-05,
+      "loss": 0.8395,
+      "step": 521330
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 1.521121910155353e-05,
+      "loss": 0.7227,
+      "step": 521340
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 1.5210071171653782e-05,
+      "loss": 0.6941,
+      "step": 521350
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 1.5208923241754036e-05,
+      "loss": 0.763,
+      "step": 521360
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 1.5207775311854289e-05,
+      "loss": 0.8071,
+      "step": 521370
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 1.520662738195454e-05,
+      "loss": 0.7891,
+      "step": 521380
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 1.5205479452054794e-05,
+      "loss": 0.7196,
+      "step": 521390
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 1.5204331522155047e-05,
+      "loss": 0.7242,
+      "step": 521400
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 1.5203183592255298e-05,
+      "loss": 0.7392,
+      "step": 521410
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 1.5202035662355552e-05,
+      "loss": 0.7283,
+      "step": 521420
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 1.5200887732455805e-05,
+      "loss": 0.7807,
+      "step": 521430
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 1.5199739802556055e-05,
+      "loss": 0.7884,
+      "step": 521440
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 1.5198591872656308e-05,
+      "loss": 0.7877,
+      "step": 521450
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 1.5197443942756561e-05,
+      "loss": 0.6829,
+      "step": 521460
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 1.5196296012856813e-05,
+      "loss": 0.7489,
+      "step": 521470
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 1.5195148082957066e-05,
+      "loss": 0.8275,
+      "step": 521480
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 1.519400015305732e-05,
+      "loss": 0.7971,
+      "step": 521490
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 1.5192852223157571e-05,
+      "loss": 0.7673,
+      "step": 521500
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 1.5191704293257824e-05,
+      "loss": 0.7528,
+      "step": 521510
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 1.5190556363358077e-05,
+      "loss": 0.7373,
+      "step": 521520
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 1.5189408433458329e-05,
+      "loss": 0.7668,
+      "step": 521530
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 1.5188260503558582e-05,
+      "loss": 0.7334,
+      "step": 521540
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 1.5187112573658835e-05,
+      "loss": 0.7507,
+      "step": 521550
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 1.5185964643759087e-05,
+      "loss": 0.8793,
+      "step": 521560
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 1.518481671385934e-05,
+      "loss": 0.7154,
+      "step": 521570
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 1.518366878395959e-05,
+      "loss": 0.7427,
+      "step": 521580
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 1.5182520854059843e-05,
+      "loss": 0.7332,
+      "step": 521590
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 1.5181372924160096e-05,
+      "loss": 0.7683,
+      "step": 521600
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 1.5180224994260348e-05,
+      "loss": 0.7269,
+      "step": 521610
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 1.5179077064360601e-05,
+      "loss": 0.7941,
+      "step": 521620
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 1.5177929134460855e-05,
+      "loss": 0.7624,
+      "step": 521630
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 1.5176781204561106e-05,
+      "loss": 0.8151,
+      "step": 521640
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 1.517563327466136e-05,
+      "loss": 0.7851,
+      "step": 521650
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 1.5174485344761613e-05,
+      "loss": 0.6692,
+      "step": 521660
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 1.5173337414861864e-05,
+      "loss": 0.7248,
+      "step": 521670
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 1.5172189484962117e-05,
+      "loss": 0.6467,
+      "step": 521680
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 1.517104155506237e-05,
+      "loss": 0.7318,
+      "step": 521690
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 1.5169893625162622e-05,
+      "loss": 0.7429,
+      "step": 521700
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 1.5168745695262875e-05,
+      "loss": 0.7888,
+      "step": 521710
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 1.5167597765363129e-05,
+      "loss": 0.6963,
+      "step": 521720
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 1.5166449835463378e-05,
+      "loss": 0.736,
+      "step": 521730
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 1.5165301905563632e-05,
+      "loss": 0.6942,
+      "step": 521740
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 1.5164153975663885e-05,
+      "loss": 0.7209,
+      "step": 521750
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 1.5163006045764137e-05,
+      "loss": 0.8114,
+      "step": 521760
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 1.516185811586439e-05,
+      "loss": 0.7614,
+      "step": 521770
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 1.5160710185964643e-05,
+      "loss": 0.7633,
+      "step": 521780
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 1.5159562256064895e-05,
+      "loss": 0.7752,
+      "step": 521790
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 1.5158414326165148e-05,
+      "loss": 0.7593,
+      "step": 521800
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 1.5157266396265401e-05,
+      "loss": 0.7707,
+      "step": 521810
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 1.5156118466365653e-05,
+      "loss": 0.8399,
+      "step": 521820
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 1.5154970536465906e-05,
+      "loss": 0.7534,
+      "step": 521830
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 1.5153822606566159e-05,
+      "loss": 0.7864,
+      "step": 521840
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 1.515267467666641e-05,
+      "loss": 0.7344,
+      "step": 521850
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 1.5151526746766664e-05,
+      "loss": 0.812,
+      "step": 521860
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 1.5150378816866917e-05,
+      "loss": 0.7704,
+      "step": 521870
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 1.5149230886967167e-05,
+      "loss": 0.7556,
+      "step": 521880
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 1.514808295706742e-05,
+      "loss": 0.7899,
+      "step": 521890
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 1.5146935027167673e-05,
+      "loss": 0.7938,
+      "step": 521900
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 1.5145787097267925e-05,
+      "loss": 0.8464,
+      "step": 521910
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 1.5144639167368178e-05,
+      "loss": 0.8097,
+      "step": 521920
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 1.514349123746843e-05,
+      "loss": 0.7508,
+      "step": 521930
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 1.5142343307568683e-05,
+      "loss": 0.6962,
+      "step": 521940
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 1.5141195377668936e-05,
+      "loss": 0.7899,
+      "step": 521950
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 1.5140047447769188e-05,
+      "loss": 0.7503,
+      "step": 521960
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 1.5138899517869441e-05,
+      "loss": 0.7639,
+      "step": 521970
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 1.5137751587969694e-05,
+      "loss": 0.8072,
+      "step": 521980
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 1.5136603658069944e-05,
+      "loss": 0.6964,
+      "step": 521990
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 1.5135455728170199e-05,
+      "loss": 0.8024,
+      "step": 522000
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 1.5134307798270452e-05,
+      "loss": 0.7881,
+      "step": 522010
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 1.5133159868370702e-05,
+      "loss": 0.742,
+      "step": 522020
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 1.5132011938470955e-05,
+      "loss": 0.7986,
+      "step": 522030
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 1.5130864008571209e-05,
+      "loss": 0.7898,
+      "step": 522040
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 1.512971607867146e-05,
+      "loss": 0.8339,
+      "step": 522050
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 1.5128568148771713e-05,
+      "loss": 0.7537,
+      "step": 522060
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 1.5127420218871967e-05,
+      "loss": 0.7678,
+      "step": 522070
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 1.5126272288972218e-05,
+      "loss": 0.8285,
+      "step": 522080
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 1.5125124359072471e-05,
+      "loss": 0.7723,
+      "step": 522090
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 1.5123976429172725e-05,
+      "loss": 0.745,
+      "step": 522100
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 1.5122828499272976e-05,
+      "loss": 0.7152,
+      "step": 522110
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 1.512168056937323e-05,
+      "loss": 0.8537,
+      "step": 522120
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 1.5120532639473483e-05,
+      "loss": 0.8116,
+      "step": 522130
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 1.5119384709573733e-05,
+      "loss": 0.6835,
+      "step": 522140
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 1.5118236779673988e-05,
+      "loss": 0.7212,
+      "step": 522150
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 1.511708884977424e-05,
+      "loss": 0.7112,
+      "step": 522160
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 1.511594091987449e-05,
+      "loss": 0.7512,
+      "step": 522170
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 1.5114792989974744e-05,
+      "loss": 0.801,
+      "step": 522180
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 1.5113645060074997e-05,
+      "loss": 0.7483,
+      "step": 522190
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 1.5112497130175249e-05,
+      "loss": 0.7495,
+      "step": 522200
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 1.5111349200275502e-05,
+      "loss": 0.7496,
+      "step": 522210
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 1.5110201270375755e-05,
+      "loss": 0.803,
+      "step": 522220
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 1.5109053340476007e-05,
+      "loss": 0.7401,
+      "step": 522230
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 1.510790541057626e-05,
+      "loss": 0.7697,
+      "step": 522240
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 1.5106757480676513e-05,
+      "loss": 0.7707,
+      "step": 522250
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 1.5105609550776765e-05,
+      "loss": 0.7944,
+      "step": 522260
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 1.5104461620877018e-05,
+      "loss": 0.7536,
+      "step": 522270
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 1.5103313690977268e-05,
+      "loss": 0.7374,
+      "step": 522280
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 1.5102165761077521e-05,
+      "loss": 0.7853,
+      "step": 522290
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 1.5101017831177776e-05,
+      "loss": 0.8518,
+      "step": 522300
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 1.5099869901278026e-05,
+      "loss": 0.7418,
+      "step": 522310
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 1.5098721971378279e-05,
+      "loss": 0.7085,
+      "step": 522320
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 1.5097574041478532e-05,
+      "loss": 0.787,
+      "step": 522330
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 1.5096426111578784e-05,
+      "loss": 0.6886,
+      "step": 522340
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 1.5095278181679037e-05,
+      "loss": 0.7602,
+      "step": 522350
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 1.509413025177929e-05,
+      "loss": 0.7312,
+      "step": 522360
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 1.5092982321879542e-05,
+      "loss": 0.7465,
+      "step": 522370
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 1.5091834391979795e-05,
+      "loss": 0.804,
+      "step": 522380
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 1.5090686462080048e-05,
+      "loss": 0.6926,
+      "step": 522390
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 1.50895385321803e-05,
+      "loss": 0.7817,
+      "step": 522400
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 1.5088390602280553e-05,
+      "loss": 0.8049,
+      "step": 522410
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 1.5087242672380806e-05,
+      "loss": 0.7388,
+      "step": 522420
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 1.5086094742481056e-05,
+      "loss": 0.7349,
+      "step": 522430
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 1.508494681258131e-05,
+      "loss": 0.7676,
+      "step": 522440
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 1.5083798882681565e-05,
+      "loss": 0.7909,
+      "step": 522450
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 1.5082650952781814e-05,
+      "loss": 0.7725,
+      "step": 522460
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 1.5081503022882068e-05,
+      "loss": 0.7711,
+      "step": 522470
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 1.508035509298232e-05,
+      "loss": 0.772,
+      "step": 522480
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 1.5079207163082572e-05,
+      "loss": 0.7624,
+      "step": 522490
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 1.5078059233182826e-05,
+      "loss": 0.7342,
+      "step": 522500
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 1.5076911303283079e-05,
+      "loss": 0.8088,
+      "step": 522510
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 1.507576337338333e-05,
+      "loss": 0.8084,
+      "step": 522520
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 1.5074615443483584e-05,
+      "loss": 0.7899,
+      "step": 522530
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 1.5073467513583837e-05,
+      "loss": 0.6904,
+      "step": 522540
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 1.5072319583684088e-05,
+      "loss": 0.6725,
+      "step": 522550
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 1.5071171653784342e-05,
+      "loss": 0.7689,
+      "step": 522560
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 1.5070023723884595e-05,
+      "loss": 0.7608,
+      "step": 522570
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 1.5068875793984845e-05,
+      "loss": 0.7458,
+      "step": 522580
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 1.5067727864085098e-05,
+      "loss": 0.786,
+      "step": 522590
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 1.506657993418535e-05,
+      "loss": 0.7333,
+      "step": 522600
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 1.5065432004285603e-05,
+      "loss": 0.7271,
+      "step": 522610
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 1.5064284074385856e-05,
+      "loss": 0.7143,
+      "step": 522620
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 1.5063136144486108e-05,
+      "loss": 0.724,
+      "step": 522630
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 1.5061988214586361e-05,
+      "loss": 0.7333,
+      "step": 522640
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 1.5060840284686614e-05,
+      "loss": 0.7426,
+      "step": 522650
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 1.5059692354786866e-05,
+      "loss": 0.7656,
+      "step": 522660
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 1.5058544424887119e-05,
+      "loss": 0.7338,
+      "step": 522670
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 1.5057396494987372e-05,
+      "loss": 0.7755,
+      "step": 522680
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 1.5056248565087624e-05,
+      "loss": 0.6352,
+      "step": 522690
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 1.5055100635187877e-05,
+      "loss": 0.7621,
+      "step": 522700
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 1.505395270528813e-05,
+      "loss": 0.7609,
+      "step": 522710
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 1.505280477538838e-05,
+      "loss": 0.7589,
+      "step": 522720
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 1.5051656845488633e-05,
+      "loss": 0.768,
+      "step": 522730
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 1.5050508915588887e-05,
+      "loss": 0.7503,
+      "step": 522740
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 1.5049360985689138e-05,
+      "loss": 0.7843,
+      "step": 522750
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 1.5048213055789391e-05,
+      "loss": 0.7667,
+      "step": 522760
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 1.5047065125889645e-05,
+      "loss": 0.797,
+      "step": 522770
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 1.5045917195989896e-05,
+      "loss": 0.6791,
+      "step": 522780
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 1.504476926609015e-05,
+      "loss": 0.7652,
+      "step": 522790
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 1.5043621336190403e-05,
+      "loss": 0.7656,
+      "step": 522800
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 1.5042473406290654e-05,
+      "loss": 0.7679,
+      "step": 522810
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 1.5041325476390907e-05,
+      "loss": 0.7635,
+      "step": 522820
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 1.504017754649116e-05,
+      "loss": 0.8032,
+      "step": 522830
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 1.5039029616591412e-05,
+      "loss": 0.7972,
+      "step": 522840
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 1.5037881686691665e-05,
+      "loss": 0.7649,
+      "step": 522850
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 1.5036733756791919e-05,
+      "loss": 0.717,
+      "step": 522860
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 1.5035585826892168e-05,
+      "loss": 0.7322,
+      "step": 522870
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 1.5034437896992422e-05,
+      "loss": 0.7762,
+      "step": 522880
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 1.5033289967092675e-05,
+      "loss": 0.7514,
+      "step": 522890
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 1.5032142037192927e-05,
+      "loss": 0.7216,
+      "step": 522900
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 1.503099410729318e-05,
+      "loss": 0.7087,
+      "step": 522910
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 1.5029846177393433e-05,
+      "loss": 0.748,
+      "step": 522920
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 1.5028698247493685e-05,
+      "loss": 0.7385,
+      "step": 522930
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 1.5027550317593938e-05,
+      "loss": 0.6853,
+      "step": 522940
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 1.502640238769419e-05,
+      "loss": 0.8203,
+      "step": 522950
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 1.5025254457794443e-05,
+      "loss": 0.7389,
+      "step": 522960
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 1.5024106527894696e-05,
+      "loss": 0.7796,
+      "step": 522970
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 1.5022958597994947e-05,
+      "loss": 0.7632,
+      "step": 522980
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 1.50218106680952e-05,
+      "loss": 0.8172,
+      "step": 522990
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 1.5020662738195454e-05,
+      "loss": 0.6624,
+      "step": 523000
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 1.5019514808295704e-05,
+      "loss": 0.7873,
+      "step": 523010
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 1.5018366878395957e-05,
+      "loss": 0.7267,
+      "step": 523020
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 1.501721894849621e-05,
+      "loss": 0.746,
+      "step": 523030
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 1.5016071018596462e-05,
+      "loss": 0.6847,
+      "step": 523040
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 1.5014923088696715e-05,
+      "loss": 0.6305,
+      "step": 523050
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 1.5013775158796968e-05,
+      "loss": 0.7732,
+      "step": 523060
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 1.501262722889722e-05,
+      "loss": 0.6603,
+      "step": 523070
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 1.5011479298997473e-05,
+      "loss": 0.7978,
+      "step": 523080
+    },
+    {
+      "epoch": 24.0,
+      "eval_cer": 0.6688580036613497,
+      "eval_em": 0.007357212003872217,
+      "eval_f1": 0.007357212003872217,
+      "eval_loss": 1.3419605493545532,
+      "eval_runtime": 1595.2109,
+      "eval_samples_per_second": 6.476,
+      "eval_steps_per_second": 0.81,
+      "eval_wer": 0.9926427879961278,
+      "step": 523086
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 1.5010331369097726e-05,
+      "loss": 0.6753,
+      "step": 523090
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 1.5009183439197978e-05,
+      "loss": 0.7075,
+      "step": 523100
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 1.5008035509298231e-05,
+      "loss": 0.7645,
+      "step": 523110
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 1.5006887579398484e-05,
+      "loss": 0.7844,
+      "step": 523120
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 1.5005739649498736e-05,
+      "loss": 0.6993,
+      "step": 523130
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 1.5004591719598989e-05,
+      "loss": 0.6798,
+      "step": 523140
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 1.5003443789699242e-05,
+      "loss": 0.7117,
+      "step": 523150
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 1.5002295859799492e-05,
+      "loss": 0.7221,
+      "step": 523160
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 1.5001147929899745e-05,
+      "loss": 0.6729,
+      "step": 523170
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 1.4999999999999999e-05,
+      "loss": 0.7567,
+      "step": 523180
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 1.499885207010025e-05,
+      "loss": 0.7173,
+      "step": 523190
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 1.4997704140200503e-05,
+      "loss": 0.5935,
+      "step": 523200
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 1.4996556210300757e-05,
+      "loss": 0.7185,
+      "step": 523210
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 1.4995408280401008e-05,
+      "loss": 0.7267,
+      "step": 523220
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 1.4994260350501262e-05,
+      "loss": 0.6324,
+      "step": 523230
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 1.4993112420601515e-05,
+      "loss": 0.7334,
+      "step": 523240
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 1.4991964490701766e-05,
+      "loss": 0.6597,
+      "step": 523250
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 1.499081656080202e-05,
+      "loss": 0.741,
+      "step": 523260
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 1.4989668630902273e-05,
+      "loss": 0.662,
+      "step": 523270
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 1.4988520701002524e-05,
+      "loss": 0.7096,
+      "step": 523280
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 1.4987372771102778e-05,
+      "loss": 0.6494,
+      "step": 523290
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 1.4986224841203027e-05,
+      "loss": 0.6993,
+      "step": 523300
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 1.498507691130328e-05,
+      "loss": 0.6724,
+      "step": 523310
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 1.4983928981403534e-05,
+      "loss": 0.7962,
+      "step": 523320
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 1.4982781051503785e-05,
+      "loss": 0.7173,
+      "step": 523330
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 1.4981633121604039e-05,
+      "loss": 0.6999,
+      "step": 523340
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 1.4980485191704292e-05,
+      "loss": 0.7147,
+      "step": 523350
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 1.4979337261804543e-05,
+      "loss": 0.7156,
+      "step": 523360
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 1.4978189331904797e-05,
+      "loss": 0.7008,
+      "step": 523370
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 1.497704140200505e-05,
+      "loss": 0.7201,
+      "step": 523380
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 1.4975893472105302e-05,
+      "loss": 0.7309,
+      "step": 523390
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 1.4974745542205555e-05,
+      "loss": 0.745,
+      "step": 523400
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 1.4973597612305808e-05,
+      "loss": 0.7405,
+      "step": 523410
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.497244968240606e-05,
+      "loss": 0.746,
+      "step": 523420
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.4971301752506313e-05,
+      "loss": 0.7383,
+      "step": 523430
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.4970153822606566e-05,
+      "loss": 0.6917,
+      "step": 523440
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.4969005892706816e-05,
+      "loss": 0.6293,
+      "step": 523450
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.4967857962807069e-05,
+      "loss": 0.6647,
+      "step": 523460
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.4966710032907322e-05,
+      "loss": 0.7117,
+      "step": 523470
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.4965562103007574e-05,
+      "loss": 0.7529,
+      "step": 523480
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.4964414173107827e-05,
+      "loss": 0.7073,
+      "step": 523490
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.496326624320808e-05,
+      "loss": 0.8113,
+      "step": 523500
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.4962118313308332e-05,
+      "loss": 0.7413,
+      "step": 523510
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.4960970383408585e-05,
+      "loss": 0.7908,
+      "step": 523520
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.4959822453508838e-05,
+      "loss": 0.7495,
+      "step": 523530
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.495867452360909e-05,
+      "loss": 0.664,
+      "step": 523540
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.4957526593709343e-05,
+      "loss": 0.6832,
+      "step": 523550
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.4956378663809596e-05,
+      "loss": 0.7845,
+      "step": 523560
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.4955230733909848e-05,
+      "loss": 0.6852,
+      "step": 523570
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.4954082804010101e-05,
+      "loss": 0.7295,
+      "step": 523580
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.4952934874110355e-05,
+      "loss": 0.7065,
+      "step": 523590
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.4951786944210604e-05,
+      "loss": 0.7728,
+      "step": 523600
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.4950639014310858e-05,
+      "loss": 0.6992,
+      "step": 523610
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.4949491084411111e-05,
+      "loss": 0.7039,
+      "step": 523620
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 1.4948343154511362e-05,
+      "loss": 0.7828,
+      "step": 523630
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.4947195224611616e-05,
+      "loss": 0.6922,
+      "step": 523640
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.4946047294711867e-05,
+      "loss": 0.762,
+      "step": 523650
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.494489936481212e-05,
+      "loss": 0.7271,
+      "step": 523660
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.4943751434912374e-05,
+      "loss": 0.7386,
+      "step": 523670
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.4942603505012625e-05,
+      "loss": 0.7474,
+      "step": 523680
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.4941455575112878e-05,
+      "loss": 0.7534,
+      "step": 523690
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.4940307645213132e-05,
+      "loss": 0.7766,
+      "step": 523700
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.4939159715313383e-05,
+      "loss": 0.8563,
+      "step": 523710
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.4938011785413637e-05,
+      "loss": 0.7138,
+      "step": 523720
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.493686385551389e-05,
+      "loss": 0.7629,
+      "step": 523730
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.493571592561414e-05,
+      "loss": 0.7452,
+      "step": 523740
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.4934567995714393e-05,
+      "loss": 0.6775,
+      "step": 523750
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.4933420065814646e-05,
+      "loss": 0.7576,
+      "step": 523760
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.4932272135914898e-05,
+      "loss": 0.7957,
+      "step": 523770
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.4931124206015151e-05,
+      "loss": 0.7364,
+      "step": 523780
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.4929976276115404e-05,
+      "loss": 0.756,
+      "step": 523790
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.4928828346215656e-05,
+      "loss": 0.7437,
+      "step": 523800
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.4927680416315909e-05,
+      "loss": 0.7143,
+      "step": 523810
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.4926532486416162e-05,
+      "loss": 0.6528,
+      "step": 523820
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.4925384556516414e-05,
+      "loss": 0.737,
+      "step": 523830
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 1.4924236626616667e-05,
+      "loss": 0.7955,
+      "step": 523840
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.492308869671692e-05,
+      "loss": 0.6388,
+      "step": 523850
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.4921940766817172e-05,
+      "loss": 0.7626,
+      "step": 523860
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.4920792836917425e-05,
+      "loss": 0.7965,
+      "step": 523870
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.4919644907017678e-05,
+      "loss": 0.7487,
+      "step": 523880
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.4918496977117928e-05,
+      "loss": 0.7012,
+      "step": 523890
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.4917349047218181e-05,
+      "loss": 0.6421,
+      "step": 523900
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.4916201117318435e-05,
+      "loss": 0.6911,
+      "step": 523910
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.4915053187418686e-05,
+      "loss": 0.7259,
+      "step": 523920
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.491390525751894e-05,
+      "loss": 0.7741,
+      "step": 523930
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.4912757327619193e-05,
+      "loss": 0.7569,
+      "step": 523940
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.4911609397719444e-05,
+      "loss": 0.7616,
+      "step": 523950
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.4910461467819697e-05,
+      "loss": 0.7021,
+      "step": 523960
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.4909313537919949e-05,
+      "loss": 0.7384,
+      "step": 523970
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.4908165608020202e-05,
+      "loss": 0.7283,
+      "step": 523980
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.4907017678120455e-05,
+      "loss": 0.6794,
+      "step": 523990
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.4905869748220707e-05,
+      "loss": 0.7571,
+      "step": 524000
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.490472181832096e-05,
+      "loss": 0.7122,
+      "step": 524010
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.4903573888421213e-05,
+      "loss": 0.8144,
+      "step": 524020
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.4902425958521463e-05,
+      "loss": 0.6306,
+      "step": 524030
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.4901278028621717e-05,
+      "loss": 0.7515,
+      "step": 524040
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.490013009872197e-05,
+      "loss": 0.6952,
+      "step": 524050
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 1.4898982168822221e-05,
+      "loss": 0.7155,
+      "step": 524060
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.4897834238922475e-05,
+      "loss": 0.7307,
+      "step": 524070
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.4896686309022728e-05,
+      "loss": 0.7213,
+      "step": 524080
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.489553837912298e-05,
+      "loss": 0.7584,
+      "step": 524090
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.4894390449223233e-05,
+      "loss": 0.7052,
+      "step": 524100
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.4893242519323486e-05,
+      "loss": 0.7258,
+      "step": 524110
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.4892094589423737e-05,
+      "loss": 0.7448,
+      "step": 524120
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.489094665952399e-05,
+      "loss": 0.6987,
+      "step": 524130
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.4889798729624244e-05,
+      "loss": 0.7182,
+      "step": 524140
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.4888650799724495e-05,
+      "loss": 0.7385,
+      "step": 524150
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.4887502869824749e-05,
+      "loss": 0.6777,
+      "step": 524160
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.4886354939925002e-05,
+      "loss": 0.8013,
+      "step": 524170
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.4885207010025252e-05,
+      "loss": 0.7316,
+      "step": 524180
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.4884059080125505e-05,
+      "loss": 0.6604,
+      "step": 524190
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.4882911150225758e-05,
+      "loss": 0.6823,
+      "step": 524200
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.488176322032601e-05,
+      "loss": 0.7977,
+      "step": 524210
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.4880615290426263e-05,
+      "loss": 0.7884,
+      "step": 524220
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.4879467360526516e-05,
+      "loss": 0.6798,
+      "step": 524230
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.4878319430626768e-05,
+      "loss": 0.7549,
+      "step": 524240
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.4877171500727021e-05,
+      "loss": 0.7337,
+      "step": 524250
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.4876023570827274e-05,
+      "loss": 0.8013,
+      "step": 524260
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.4874875640927526e-05,
+      "loss": 0.6543,
+      "step": 524270
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 1.4873727711027779e-05,
+      "loss": 0.7154,
+      "step": 524280
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.4872579781128032e-05,
+      "loss": 0.7214,
+      "step": 524290
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.4871431851228284e-05,
+      "loss": 0.7331,
+      "step": 524300
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.4870283921328537e-05,
+      "loss": 0.6844,
+      "step": 524310
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.4869135991428787e-05,
+      "loss": 0.6651,
+      "step": 524320
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.486798806152904e-05,
+      "loss": 0.6812,
+      "step": 524330
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.4866840131629293e-05,
+      "loss": 0.753,
+      "step": 524340
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.4865692201729545e-05,
+      "loss": 0.7191,
+      "step": 524350
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.4864544271829798e-05,
+      "loss": 0.6711,
+      "step": 524360
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.4863396341930052e-05,
+      "loss": 0.7156,
+      "step": 524370
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.4862248412030303e-05,
+      "loss": 0.7535,
+      "step": 524380
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.4861100482130556e-05,
+      "loss": 0.7892,
+      "step": 524390
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.485995255223081e-05,
+      "loss": 0.6859,
+      "step": 524400
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.4858804622331061e-05,
+      "loss": 0.7226,
+      "step": 524410
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.4857656692431314e-05,
+      "loss": 0.7106,
+      "step": 524420
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.4856508762531568e-05,
+      "loss": 0.665,
+      "step": 524430
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.4855360832631819e-05,
+      "loss": 0.69,
+      "step": 524440
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.4854212902732072e-05,
+      "loss": 0.7022,
+      "step": 524450
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.4853064972832326e-05,
+      "loss": 0.7312,
+      "step": 524460
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.4851917042932575e-05,
+      "loss": 0.6679,
+      "step": 524470
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.4850769113032829e-05,
+      "loss": 0.7437,
+      "step": 524480
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.4849621183133082e-05,
+      "loss": 0.7592,
+      "step": 524490
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 1.4848473253233334e-05,
+      "loss": 0.7472,
+      "step": 524500
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 1.4847325323333587e-05,
+      "loss": 0.7745,
+      "step": 524510
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 1.484617739343384e-05,
+      "loss": 0.7039,
+      "step": 524520
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 1.4845029463534092e-05,
+      "loss": 0.6583,
+      "step": 524530
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 1.4843881533634345e-05,
+      "loss": 0.744,
+      "step": 524540
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 1.4842733603734598e-05,
+      "loss": 0.7298,
+      "step": 524550
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 1.484158567383485e-05,
+      "loss": 0.7603,
+      "step": 524560
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 1.4840437743935103e-05,
+      "loss": 0.6933,
+      "step": 524570
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 1.4839289814035356e-05,
+      "loss": 0.682,
+      "step": 524580
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 1.4838141884135608e-05,
+      "loss": 0.6208,
+      "step": 524590
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 1.4836993954235861e-05,
+      "loss": 0.6934,
+      "step": 524600
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 1.4835846024336114e-05,
+      "loss": 0.781,
+      "step": 524610
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 1.4834698094436364e-05,
+      "loss": 0.8117,
+      "step": 524620
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 1.4833550164536617e-05,
+      "loss": 0.675,
+      "step": 524630
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 1.483240223463687e-05,
+      "loss": 0.729,
+      "step": 524640
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 1.4831254304737122e-05,
+      "loss": 0.764,
+      "step": 524650
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 1.4830106374837375e-05,
+      "loss": 0.6945,
+      "step": 524660
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 1.4828958444937627e-05,
+      "loss": 0.8512,
+      "step": 524670
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 1.482781051503788e-05,
+      "loss": 0.7012,
+      "step": 524680
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 1.4826662585138133e-05,
+      "loss": 0.7255,
+      "step": 524690
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 1.4825514655238385e-05,
+      "loss": 0.6822,
+      "step": 524700
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 1.4824366725338638e-05,
+      "loss": 0.6936,
+      "step": 524710
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 1.4823218795438891e-05,
+      "loss": 0.6938,
+      "step": 524720
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 1.4822070865539143e-05,
+      "loss": 0.6993,
+      "step": 524730
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 1.4820922935639396e-05,
+      "loss": 0.7686,
+      "step": 524740
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 1.481977500573965e-05,
+      "loss": 0.738,
+      "step": 524750
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 1.48186270758399e-05,
+      "loss": 0.7545,
+      "step": 524760
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 1.4817479145940152e-05,
+      "loss": 0.7149,
+      "step": 524770
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 1.4816331216040406e-05,
+      "loss": 0.7707,
+      "step": 524780
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 1.4815183286140657e-05,
+      "loss": 0.7525,
+      "step": 524790
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 1.481403535624091e-05,
+      "loss": 0.7007,
+      "step": 524800
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 1.4812887426341164e-05,
+      "loss": 0.7204,
+      "step": 524810
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 1.4811739496441415e-05,
+      "loss": 0.7133,
+      "step": 524820
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 1.4810591566541668e-05,
+      "loss": 0.7246,
+      "step": 524830
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 1.4809443636641922e-05,
+      "loss": 0.6975,
+      "step": 524840
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 1.4808295706742173e-05,
+      "loss": 0.6551,
+      "step": 524850
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 1.4807147776842427e-05,
+      "loss": 0.6943,
+      "step": 524860
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 1.480599984694268e-05,
+      "loss": 0.7306,
+      "step": 524870
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 1.4804851917042931e-05,
+      "loss": 0.7113,
+      "step": 524880
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 1.4803703987143185e-05,
+      "loss": 0.7461,
+      "step": 524890
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 1.4802556057243438e-05,
+      "loss": 0.7144,
+      "step": 524900
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 1.4801408127343688e-05,
+      "loss": 0.7172,
+      "step": 524910
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 1.4800260197443941e-05,
+      "loss": 0.6651,
+      "step": 524920
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 1.4799112267544194e-05,
+      "loss": 0.6743,
+      "step": 524930
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.4797964337644446e-05,
+      "loss": 0.7558,
+      "step": 524940
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.4796816407744699e-05,
+      "loss": 0.7181,
+      "step": 524950
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.4795668477844952e-05,
+      "loss": 0.7273,
+      "step": 524960
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.4794520547945204e-05,
+      "loss": 0.698,
+      "step": 524970
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.4793372618045457e-05,
+      "loss": 0.7449,
+      "step": 524980
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.479222468814571e-05,
+      "loss": 0.7222,
+      "step": 524990
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.4791076758245962e-05,
+      "loss": 0.737,
+      "step": 525000
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.4789928828346215e-05,
+      "loss": 0.7273,
+      "step": 525010
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.4788780898446467e-05,
+      "loss": 0.6824,
+      "step": 525020
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.478763296854672e-05,
+      "loss": 0.7427,
+      "step": 525030
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.4786485038646973e-05,
+      "loss": 0.7279,
+      "step": 525040
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.4785337108747223e-05,
+      "loss": 0.7039,
+      "step": 525050
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.4784189178847476e-05,
+      "loss": 0.7133,
+      "step": 525060
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.478304124894773e-05,
+      "loss": 0.7045,
+      "step": 525070
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.4781893319047981e-05,
+      "loss": 0.684,
+      "step": 525080
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.4780745389148234e-05,
+      "loss": 0.7597,
+      "step": 525090
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.4779597459248487e-05,
+      "loss": 0.7323,
+      "step": 525100
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.4778449529348739e-05,
+      "loss": 0.705,
+      "step": 525110
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.4777301599448992e-05,
+      "loss": 0.716,
+      "step": 525120
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.4776153669549245e-05,
+      "loss": 0.7675,
+      "step": 525130
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.4775005739649497e-05,
+      "loss": 0.7411,
+      "step": 525140
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 1.477385780974975e-05,
+      "loss": 0.8469,
+      "step": 525150
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.4772709879850003e-05,
+      "loss": 0.7102,
+      "step": 525160
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.4771561949950255e-05,
+      "loss": 0.7107,
+      "step": 525170
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.4770414020050508e-05,
+      "loss": 0.6727,
+      "step": 525180
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.4769266090150762e-05,
+      "loss": 0.721,
+      "step": 525190
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.4768118160251011e-05,
+      "loss": 0.6753,
+      "step": 525200
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.4766970230351265e-05,
+      "loss": 0.6789,
+      "step": 525210
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.4765822300451518e-05,
+      "loss": 0.6603,
+      "step": 525220
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.476467437055177e-05,
+      "loss": 0.7656,
+      "step": 525230
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.4763526440652023e-05,
+      "loss": 0.7569,
+      "step": 525240
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.4762378510752276e-05,
+      "loss": 0.8059,
+      "step": 525250
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.4761230580852527e-05,
+      "loss": 0.7792,
+      "step": 525260
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.476008265095278e-05,
+      "loss": 0.7204,
+      "step": 525270
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.4758934721053034e-05,
+      "loss": 0.6802,
+      "step": 525280
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.4757786791153285e-05,
+      "loss": 0.7157,
+      "step": 525290
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.4756638861253539e-05,
+      "loss": 0.7554,
+      "step": 525300
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.4755490931353792e-05,
+      "loss": 0.7689,
+      "step": 525310
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.4754343001454043e-05,
+      "loss": 0.7355,
+      "step": 525320
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.4753195071554297e-05,
+      "loss": 0.6986,
+      "step": 525330
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.475204714165455e-05,
+      "loss": 0.765,
+      "step": 525340
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.47508992117548e-05,
+      "loss": 0.7123,
+      "step": 525350
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.4749751281855053e-05,
+      "loss": 0.7628,
+      "step": 525360
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 1.4748603351955305e-05,
+      "loss": 0.7839,
+      "step": 525370
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.4747455422055558e-05,
+      "loss": 0.708,
+      "step": 525380
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.4746307492155811e-05,
+      "loss": 0.7182,
+      "step": 525390
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.4745159562256063e-05,
+      "loss": 0.824,
+      "step": 525400
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.4744011632356316e-05,
+      "loss": 0.7051,
+      "step": 525410
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.4742863702456569e-05,
+      "loss": 0.7358,
+      "step": 525420
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.474171577255682e-05,
+      "loss": 0.7891,
+      "step": 525430
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.4740567842657074e-05,
+      "loss": 0.6766,
+      "step": 525440
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.4739419912757327e-05,
+      "loss": 0.7954,
+      "step": 525450
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.4738271982857579e-05,
+      "loss": 0.6975,
+      "step": 525460
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.4737124052957832e-05,
+      "loss": 0.7399,
+      "step": 525470
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.4735976123058085e-05,
+      "loss": 0.7762,
+      "step": 525480
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.4734828193158335e-05,
+      "loss": 0.7538,
+      "step": 525490
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.4733680263258588e-05,
+      "loss": 0.6806,
+      "step": 525500
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.4732532333358842e-05,
+      "loss": 0.7371,
+      "step": 525510
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.4731384403459093e-05,
+      "loss": 0.69,
+      "step": 525520
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.4730236473559346e-05,
+      "loss": 0.787,
+      "step": 525530
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.47290885436596e-05,
+      "loss": 0.7453,
+      "step": 525540
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.4727940613759851e-05,
+      "loss": 0.6566,
+      "step": 525550
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.4726792683860104e-05,
+      "loss": 0.7419,
+      "step": 525560
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.4725644753960358e-05,
+      "loss": 0.7059,
+      "step": 525570
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.472449682406061e-05,
+      "loss": 0.729,
+      "step": 525580
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 1.4723348894160862e-05,
+      "loss": 0.7348,
+      "step": 525590
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.4722200964261116e-05,
+      "loss": 0.6518,
+      "step": 525600
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.4721053034361367e-05,
+      "loss": 0.7214,
+      "step": 525610
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.471990510446162e-05,
+      "loss": 0.6903,
+      "step": 525620
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.4718757174561874e-05,
+      "loss": 0.7468,
+      "step": 525630
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.4717609244662124e-05,
+      "loss": 0.8078,
+      "step": 525640
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.4716461314762377e-05,
+      "loss": 0.69,
+      "step": 525650
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.471531338486263e-05,
+      "loss": 0.842,
+      "step": 525660
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.4714165454962882e-05,
+      "loss": 0.7349,
+      "step": 525670
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.4713017525063135e-05,
+      "loss": 0.7369,
+      "step": 525680
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.4711869595163386e-05,
+      "loss": 0.6733,
+      "step": 525690
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.471072166526364e-05,
+      "loss": 0.7094,
+      "step": 525700
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.4709573735363893e-05,
+      "loss": 0.7217,
+      "step": 525710
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.4708425805464144e-05,
+      "loss": 0.6289,
+      "step": 525720
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.4707277875564398e-05,
+      "loss": 0.7943,
+      "step": 525730
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.4706129945664651e-05,
+      "loss": 0.7239,
+      "step": 525740
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.4704982015764902e-05,
+      "loss": 0.7792,
+      "step": 525750
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.4703834085865156e-05,
+      "loss": 0.7164,
+      "step": 525760
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.4702686155965409e-05,
+      "loss": 0.7115,
+      "step": 525770
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.4701538226065659e-05,
+      "loss": 0.7621,
+      "step": 525780
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.4700390296165912e-05,
+      "loss": 0.7334,
+      "step": 525790
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.4699242366266165e-05,
+      "loss": 0.7085,
+      "step": 525800
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 1.4698094436366417e-05,
+      "loss": 0.7472,
+      "step": 525810
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 1.469694650646667e-05,
+      "loss": 0.6684,
+      "step": 525820
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 1.4695798576566923e-05,
+      "loss": 0.782,
+      "step": 525830
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 1.4694650646667175e-05,
+      "loss": 0.6931,
+      "step": 525840
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 1.4693502716767428e-05,
+      "loss": 0.6852,
+      "step": 525850
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 1.4692354786867681e-05,
+      "loss": 0.7893,
+      "step": 525860
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 1.4691206856967933e-05,
+      "loss": 0.6343,
+      "step": 525870
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 1.4690058927068186e-05,
+      "loss": 0.7432,
+      "step": 525880
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 1.468891099716844e-05,
+      "loss": 0.7342,
+      "step": 525890
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 1.4687763067268691e-05,
+      "loss": 0.7306,
+      "step": 525900
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 1.4686615137368944e-05,
+      "loss": 0.6833,
+      "step": 525910
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 1.4685467207469197e-05,
+      "loss": 0.7043,
+      "step": 525920
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 1.4684319277569447e-05,
+      "loss": 0.6837,
+      "step": 525930
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 1.46831713476697e-05,
+      "loss": 0.7775,
+      "step": 525940
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 1.4682023417769954e-05,
+      "loss": 0.7061,
+      "step": 525950
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 1.4680875487870205e-05,
+      "loss": 0.7526,
+      "step": 525960
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 1.4679727557970459e-05,
+      "loss": 0.6906,
+      "step": 525970
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 1.4678579628070712e-05,
+      "loss": 0.7302,
+      "step": 525980
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 1.4677431698170963e-05,
+      "loss": 0.5917,
+      "step": 525990
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 1.4676283768271217e-05,
+      "loss": 0.7225,
+      "step": 526000
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 1.467513583837147e-05,
+      "loss": 0.7088,
+      "step": 526010
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 1.4673987908471721e-05,
+      "loss": 0.7688,
+      "step": 526020
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.4672839978571975e-05,
+      "loss": 0.7423,
+      "step": 526030
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.4671692048672226e-05,
+      "loss": 0.7865,
+      "step": 526040
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.467054411877248e-05,
+      "loss": 0.7278,
+      "step": 526050
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.4669396188872733e-05,
+      "loss": 0.7856,
+      "step": 526060
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.4668248258972982e-05,
+      "loss": 0.7641,
+      "step": 526070
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.4667100329073236e-05,
+      "loss": 0.7763,
+      "step": 526080
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.4665952399173489e-05,
+      "loss": 0.6984,
+      "step": 526090
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.466480446927374e-05,
+      "loss": 0.7892,
+      "step": 526100
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.4663656539373994e-05,
+      "loss": 0.7273,
+      "step": 526110
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.4662508609474247e-05,
+      "loss": 0.6885,
+      "step": 526120
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.4661360679574499e-05,
+      "loss": 0.7589,
+      "step": 526130
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.4660212749674752e-05,
+      "loss": 0.7714,
+      "step": 526140
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.4659064819775005e-05,
+      "loss": 0.777,
+      "step": 526150
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.4657916889875257e-05,
+      "loss": 0.7601,
+      "step": 526160
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.465676895997551e-05,
+      "loss": 0.7595,
+      "step": 526170
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.4655621030075763e-05,
+      "loss": 0.7487,
+      "step": 526180
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.4654473100176015e-05,
+      "loss": 0.78,
+      "step": 526190
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.4653325170276268e-05,
+      "loss": 0.7616,
+      "step": 526200
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.4652177240376521e-05,
+      "loss": 0.7812,
+      "step": 526210
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.4651029310476771e-05,
+      "loss": 0.7245,
+      "step": 526220
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.4649881380577024e-05,
+      "loss": 0.692,
+      "step": 526230
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 1.4648733450677277e-05,
+      "loss": 0.7907,
+      "step": 526240
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 1.4647585520777529e-05,
+      "loss": 0.6054,
+      "step": 526250
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 1.4646437590877782e-05,
+      "loss": 0.7428,
+      "step": 526260
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 1.4645289660978035e-05,
+      "loss": 0.7262,
+      "step": 526270
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 1.4644141731078287e-05,
+      "loss": 0.7549,
+      "step": 526280
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 1.464299380117854e-05,
+      "loss": 0.775,
+      "step": 526290
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 1.4641845871278793e-05,
+      "loss": 0.7452,
+      "step": 526300
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 1.4640697941379045e-05,
+      "loss": 0.7319,
+      "step": 526310
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 1.4639550011479298e-05,
+      "loss": 0.7132,
+      "step": 526320
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 1.4638402081579552e-05,
+      "loss": 0.7411,
+      "step": 526330
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 1.4637254151679803e-05,
+      "loss": 0.7376,
+      "step": 526340
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 1.4636106221780056e-05,
+      "loss": 0.7168,
+      "step": 526350
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 1.463495829188031e-05,
+      "loss": 0.7473,
+      "step": 526360
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 1.463381036198056e-05,
+      "loss": 0.7703,
+      "step": 526370
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 1.4632662432080813e-05,
+      "loss": 0.6977,
+      "step": 526380
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 1.4631514502181064e-05,
+      "loss": 0.7048,
+      "step": 526390
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 1.4630366572281317e-05,
+      "loss": 0.731,
+      "step": 526400
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 1.462921864238157e-05,
+      "loss": 0.6928,
+      "step": 526410
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 1.4628070712481822e-05,
+      "loss": 0.7107,
+      "step": 526420
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 1.4626922782582075e-05,
+      "loss": 0.7601,
+      "step": 526430
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 1.4625774852682329e-05,
+      "loss": 0.7924,
+      "step": 526440
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 1.462462692278258e-05,
+      "loss": 0.6744,
+      "step": 526450
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 1.4623478992882834e-05,
+      "loss": 0.6933,
+      "step": 526460
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.4622331062983087e-05,
+      "loss": 0.6689,
+      "step": 526470
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.4621183133083338e-05,
+      "loss": 0.7719,
+      "step": 526480
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.4620035203183592e-05,
+      "loss": 0.7858,
+      "step": 526490
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.4618887273283845e-05,
+      "loss": 0.773,
+      "step": 526500
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.4617739343384095e-05,
+      "loss": 0.7316,
+      "step": 526510
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.4616591413484348e-05,
+      "loss": 0.7901,
+      "step": 526520
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.4615443483584601e-05,
+      "loss": 0.6852,
+      "step": 526530
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.4614295553684853e-05,
+      "loss": 0.6869,
+      "step": 526540
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.4613147623785106e-05,
+      "loss": 0.7911,
+      "step": 526550
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.4611999693885359e-05,
+      "loss": 0.6497,
+      "step": 526560
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.461085176398561e-05,
+      "loss": 0.6778,
+      "step": 526570
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.4609703834085864e-05,
+      "loss": 0.7876,
+      "step": 526580
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.4608555904186117e-05,
+      "loss": 0.6472,
+      "step": 526590
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.4607407974286369e-05,
+      "loss": 0.7,
+      "step": 526600
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.4606260044386622e-05,
+      "loss": 0.7162,
+      "step": 526610
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.4605112114486875e-05,
+      "loss": 0.7121,
+      "step": 526620
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.4603964184587127e-05,
+      "loss": 0.7643,
+      "step": 526630
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.460281625468738e-05,
+      "loss": 0.7635,
+      "step": 526640
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.4601668324787633e-05,
+      "loss": 0.7207,
+      "step": 526650
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.4600520394887883e-05,
+      "loss": 0.7093,
+      "step": 526660
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.4599372464988136e-05,
+      "loss": 0.7595,
+      "step": 526670
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 1.459822453508839e-05,
+      "loss": 0.751,
+      "step": 526680
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.4597076605188641e-05,
+      "loss": 0.7717,
+      "step": 526690
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.4595928675288894e-05,
+      "loss": 0.6612,
+      "step": 526700
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.4594780745389148e-05,
+      "loss": 0.6198,
+      "step": 526710
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.45936328154894e-05,
+      "loss": 0.7287,
+      "step": 526720
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.4592484885589652e-05,
+      "loss": 0.6838,
+      "step": 526730
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.4591336955689904e-05,
+      "loss": 0.8094,
+      "step": 526740
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.4590189025790157e-05,
+      "loss": 0.7586,
+      "step": 526750
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.458904109589041e-05,
+      "loss": 0.7768,
+      "step": 526760
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.4587893165990662e-05,
+      "loss": 0.7265,
+      "step": 526770
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.4586745236090915e-05,
+      "loss": 0.7217,
+      "step": 526780
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.4585597306191168e-05,
+      "loss": 0.7266,
+      "step": 526790
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.4584449376291418e-05,
+      "loss": 0.7738,
+      "step": 526800
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.4583301446391672e-05,
+      "loss": 0.7106,
+      "step": 526810
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.4582153516491925e-05,
+      "loss": 0.7514,
+      "step": 526820
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.4581005586592176e-05,
+      "loss": 0.7684,
+      "step": 526830
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.457985765669243e-05,
+      "loss": 0.8114,
+      "step": 526840
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.4578709726792683e-05,
+      "loss": 0.7006,
+      "step": 526850
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.4577561796892934e-05,
+      "loss": 0.6361,
+      "step": 526860
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.4576413866993188e-05,
+      "loss": 0.6454,
+      "step": 526870
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.4575265937093441e-05,
+      "loss": 0.6393,
+      "step": 526880
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.4574118007193692e-05,
+      "loss": 0.7412,
+      "step": 526890
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 1.4572970077293946e-05,
+      "loss": 0.6616,
+      "step": 526900
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 1.4571822147394199e-05,
+      "loss": 0.7015,
+      "step": 526910
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 1.457067421749445e-05,
+      "loss": 0.6779,
+      "step": 526920
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 1.4569526287594704e-05,
+      "loss": 0.7363,
+      "step": 526930
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 1.4568378357694957e-05,
+      "loss": 0.7379,
+      "step": 526940
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 1.4567230427795207e-05,
+      "loss": 0.71,
+      "step": 526950
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 1.456608249789546e-05,
+      "loss": 0.7818,
+      "step": 526960
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 1.4564934567995713e-05,
+      "loss": 0.82,
+      "step": 526970
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 1.4563786638095965e-05,
+      "loss": 0.663,
+      "step": 526980
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 1.4562638708196218e-05,
+      "loss": 0.7153,
+      "step": 526990
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 1.4561490778296471e-05,
+      "loss": 0.6359,
+      "step": 527000
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 1.4560342848396723e-05,
+      "loss": 0.6786,
+      "step": 527010
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 1.4559194918496976e-05,
+      "loss": 0.6981,
+      "step": 527020
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 1.455804698859723e-05,
+      "loss": 0.6838,
+      "step": 527030
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 1.4556899058697481e-05,
+      "loss": 0.7079,
+      "step": 527040
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 1.4555751128797734e-05,
+      "loss": 0.7208,
+      "step": 527050
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 1.4554603198897986e-05,
+      "loss": 0.6907,
+      "step": 527060
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 1.4553455268998239e-05,
+      "loss": 0.7359,
+      "step": 527070
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 1.4552307339098492e-05,
+      "loss": 0.6669,
+      "step": 527080
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 1.4551159409198742e-05,
+      "loss": 0.7164,
+      "step": 527090
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 1.4550011479298995e-05,
+      "loss": 0.7706,
+      "step": 527100
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 1.4548863549399249e-05,
+      "loss": 0.7741,
+      "step": 527110
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.45477156194995e-05,
+      "loss": 0.7386,
+      "step": 527120
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.4546567689599753e-05,
+      "loss": 0.7227,
+      "step": 527130
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.4545419759700007e-05,
+      "loss": 0.6967,
+      "step": 527140
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.4544271829800258e-05,
+      "loss": 0.6832,
+      "step": 527150
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.4543123899900511e-05,
+      "loss": 0.7087,
+      "step": 527160
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.4541975970000765e-05,
+      "loss": 0.6762,
+      "step": 527170
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.4540828040101016e-05,
+      "loss": 0.7796,
+      "step": 527180
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.453968011020127e-05,
+      "loss": 0.688,
+      "step": 527190
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.4538532180301523e-05,
+      "loss": 0.7284,
+      "step": 527200
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.4537384250401774e-05,
+      "loss": 0.6318,
+      "step": 527210
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.4536236320502027e-05,
+      "loss": 0.7859,
+      "step": 527220
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.453508839060228e-05,
+      "loss": 0.7489,
+      "step": 527230
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.453394046070253e-05,
+      "loss": 0.7369,
+      "step": 527240
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.4532792530802784e-05,
+      "loss": 0.7355,
+      "step": 527250
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.4531644600903037e-05,
+      "loss": 0.6946,
+      "step": 527260
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.4530496671003289e-05,
+      "loss": 0.7474,
+      "step": 527270
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.4529348741103542e-05,
+      "loss": 0.6709,
+      "step": 527280
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.4528200811203795e-05,
+      "loss": 0.7817,
+      "step": 527290
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.4527052881304047e-05,
+      "loss": 0.7124,
+      "step": 527300
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.45259049514043e-05,
+      "loss": 0.7117,
+      "step": 527310
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.4524757021504553e-05,
+      "loss": 0.713,
+      "step": 527320
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 1.4523609091604805e-05,
+      "loss": 0.7457,
+      "step": 527330
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 1.4522461161705058e-05,
+      "loss": 0.7316,
+      "step": 527340
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 1.4521313231805311e-05,
+      "loss": 0.7986,
+      "step": 527350
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 1.4520165301905563e-05,
+      "loss": 0.7432,
+      "step": 527360
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 1.4519017372005816e-05,
+      "loss": 0.7733,
+      "step": 527370
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 1.4517869442106069e-05,
+      "loss": 0.7926,
+      "step": 527380
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 1.4516721512206319e-05,
+      "loss": 0.7136,
+      "step": 527390
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 1.4515573582306572e-05,
+      "loss": 0.7689,
+      "step": 527400
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 1.4514425652406824e-05,
+      "loss": 0.7603,
+      "step": 527410
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 1.4513277722507077e-05,
+      "loss": 0.7706,
+      "step": 527420
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 1.451212979260733e-05,
+      "loss": 0.646,
+      "step": 527430
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 1.4510981862707582e-05,
+      "loss": 0.7596,
+      "step": 527440
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 1.4509833932807835e-05,
+      "loss": 0.8307,
+      "step": 527450
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 1.4508686002908088e-05,
+      "loss": 0.7163,
+      "step": 527460
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 1.450753807300834e-05,
+      "loss": 0.669,
+      "step": 527470
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 1.4506390143108593e-05,
+      "loss": 0.7292,
+      "step": 527480
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 1.4505242213208846e-05,
+      "loss": 0.6788,
+      "step": 527490
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 1.4504094283309098e-05,
+      "loss": 0.7702,
+      "step": 527500
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 1.4502946353409351e-05,
+      "loss": 0.6993,
+      "step": 527510
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 1.4501798423509604e-05,
+      "loss": 0.7995,
+      "step": 527520
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 1.4500650493609854e-05,
+      "loss": 0.8549,
+      "step": 527530
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 1.4499502563710107e-05,
+      "loss": 0.8014,
+      "step": 527540
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 1.449835463381036e-05,
+      "loss": 0.782,
+      "step": 527550
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.4497206703910612e-05,
+      "loss": 0.7505,
+      "step": 527560
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.4496058774010865e-05,
+      "loss": 0.8614,
+      "step": 527570
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.4494910844111119e-05,
+      "loss": 0.6999,
+      "step": 527580
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.449376291421137e-05,
+      "loss": 0.7431,
+      "step": 527590
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.4492614984311624e-05,
+      "loss": 0.8171,
+      "step": 527600
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.4491467054411877e-05,
+      "loss": 0.7456,
+      "step": 527610
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.4490319124512128e-05,
+      "loss": 0.6962,
+      "step": 527620
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.4489171194612382e-05,
+      "loss": 0.7478,
+      "step": 527630
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.4488023264712635e-05,
+      "loss": 0.7419,
+      "step": 527640
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.4486875334812886e-05,
+      "loss": 0.6563,
+      "step": 527650
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.448572740491314e-05,
+      "loss": 0.6824,
+      "step": 527660
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.4484579475013393e-05,
+      "loss": 0.848,
+      "step": 527670
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.4483431545113643e-05,
+      "loss": 0.6861,
+      "step": 527680
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.4482283615213896e-05,
+      "loss": 0.7413,
+      "step": 527690
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.448113568531415e-05,
+      "loss": 0.6646,
+      "step": 527700
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.44799877554144e-05,
+      "loss": 0.6916,
+      "step": 527710
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.4478839825514654e-05,
+      "loss": 0.8239,
+      "step": 527720
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.4477691895614907e-05,
+      "loss": 0.7071,
+      "step": 527730
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.4476543965715159e-05,
+      "loss": 0.6796,
+      "step": 527740
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.4475396035815412e-05,
+      "loss": 0.7683,
+      "step": 527750
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.4474248105915664e-05,
+      "loss": 0.793,
+      "step": 527760
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 1.4473100176015917e-05,
+      "loss": 0.7143,
+      "step": 527770
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.447195224611617e-05,
+      "loss": 0.7515,
+      "step": 527780
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.4470804316216422e-05,
+      "loss": 0.8325,
+      "step": 527790
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.4469656386316675e-05,
+      "loss": 0.7469,
+      "step": 527800
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.4468508456416928e-05,
+      "loss": 0.7378,
+      "step": 527810
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.4467360526517178e-05,
+      "loss": 0.774,
+      "step": 527820
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.4466212596617431e-05,
+      "loss": 0.7779,
+      "step": 527830
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.4465064666717684e-05,
+      "loss": 0.8067,
+      "step": 527840
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.4463916736817936e-05,
+      "loss": 0.6912,
+      "step": 527850
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.446276880691819e-05,
+      "loss": 0.7865,
+      "step": 527860
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.4461620877018442e-05,
+      "loss": 0.7501,
+      "step": 527870
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.4460472947118694e-05,
+      "loss": 0.7286,
+      "step": 527880
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.4459325017218947e-05,
+      "loss": 0.7693,
+      "step": 527890
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.44581770873192e-05,
+      "loss": 0.7797,
+      "step": 527900
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.4457029157419452e-05,
+      "loss": 0.6909,
+      "step": 527910
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.4455881227519705e-05,
+      "loss": 0.7433,
+      "step": 527920
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.4454733297619958e-05,
+      "loss": 0.6686,
+      "step": 527930
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.445358536772021e-05,
+      "loss": 0.7677,
+      "step": 527940
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.4452437437820463e-05,
+      "loss": 0.694,
+      "step": 527950
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.4451289507920717e-05,
+      "loss": 0.7698,
+      "step": 527960
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.4450141578020966e-05,
+      "loss": 0.7243,
+      "step": 527970
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.444899364812122e-05,
+      "loss": 0.6763,
+      "step": 527980
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 1.4447845718221473e-05,
+      "loss": 0.7631,
+      "step": 527990
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 1.4446697788321724e-05,
+      "loss": 0.7433,
+      "step": 528000
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 1.4445549858421978e-05,
+      "loss": 0.8074,
+      "step": 528010
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 1.4444401928522231e-05,
+      "loss": 0.7294,
+      "step": 528020
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 1.4443253998622482e-05,
+      "loss": 0.6869,
+      "step": 528030
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 1.4442106068722736e-05,
+      "loss": 0.6642,
+      "step": 528040
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 1.4440958138822989e-05,
+      "loss": 0.7571,
+      "step": 528050
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 1.443981020892324e-05,
+      "loss": 0.7576,
+      "step": 528060
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 1.4438662279023494e-05,
+      "loss": 0.7739,
+      "step": 528070
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 1.4437514349123747e-05,
+      "loss": 0.7146,
+      "step": 528080
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 1.4436366419223999e-05,
+      "loss": 0.7529,
+      "step": 528090
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 1.4435218489324252e-05,
+      "loss": 0.6684,
+      "step": 528100
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 1.4434070559424502e-05,
+      "loss": 0.6988,
+      "step": 528110
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 1.4432922629524755e-05,
+      "loss": 0.6683,
+      "step": 528120
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 1.4431774699625008e-05,
+      "loss": 0.7369,
+      "step": 528130
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 1.443062676972526e-05,
+      "loss": 0.7248,
+      "step": 528140
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 1.4429478839825513e-05,
+      "loss": 0.7772,
+      "step": 528150
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 1.4428330909925766e-05,
+      "loss": 0.7667,
+      "step": 528160
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 1.4427182980026018e-05,
+      "loss": 0.79,
+      "step": 528170
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 1.4426035050126271e-05,
+      "loss": 0.7538,
+      "step": 528180
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 1.4424887120226524e-05,
+      "loss": 0.6668,
+      "step": 528190
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 1.4423739190326776e-05,
+      "loss": 0.7246,
+      "step": 528200
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.4422591260427029e-05,
+      "loss": 0.6731,
+      "step": 528210
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.4421443330527282e-05,
+      "loss": 0.7959,
+      "step": 528220
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.4420295400627534e-05,
+      "loss": 0.7054,
+      "step": 528230
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.4419147470727787e-05,
+      "loss": 0.7757,
+      "step": 528240
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.441799954082804e-05,
+      "loss": 0.773,
+      "step": 528250
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.441685161092829e-05,
+      "loss": 0.73,
+      "step": 528260
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.4415703681028543e-05,
+      "loss": 0.7156,
+      "step": 528270
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.4414555751128797e-05,
+      "loss": 0.7024,
+      "step": 528280
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.4413407821229048e-05,
+      "loss": 0.7191,
+      "step": 528290
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.4412259891329301e-05,
+      "loss": 0.7028,
+      "step": 528300
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.4411111961429555e-05,
+      "loss": 0.6926,
+      "step": 528310
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.4409964031529806e-05,
+      "loss": 0.7638,
+      "step": 528320
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.440881610163006e-05,
+      "loss": 0.6585,
+      "step": 528330
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.4407668171730313e-05,
+      "loss": 0.8046,
+      "step": 528340
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.4406520241830564e-05,
+      "loss": 0.6991,
+      "step": 528350
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.4405372311930817e-05,
+      "loss": 0.7554,
+      "step": 528360
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.440422438203107e-05,
+      "loss": 0.6671,
+      "step": 528370
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.4403076452131322e-05,
+      "loss": 0.7359,
+      "step": 528380
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.4401928522231575e-05,
+      "loss": 0.6658,
+      "step": 528390
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.4400780592331829e-05,
+      "loss": 0.7659,
+      "step": 528400
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.4399632662432079e-05,
+      "loss": 0.739,
+      "step": 528410
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.4398484732532332e-05,
+      "loss": 0.7495,
+      "step": 528420
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.4397336802632585e-05,
+      "loss": 0.7257,
+      "step": 528430
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.4396188872732837e-05,
+      "loss": 0.7323,
+      "step": 528440
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.439504094283309e-05,
+      "loss": 0.8046,
+      "step": 528450
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.4393893012933341e-05,
+      "loss": 0.8172,
+      "step": 528460
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.4392745083033595e-05,
+      "loss": 0.7618,
+      "step": 528470
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.4391597153133848e-05,
+      "loss": 0.6688,
+      "step": 528480
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.43904492232341e-05,
+      "loss": 0.7332,
+      "step": 528490
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.4389301293334353e-05,
+      "loss": 0.6802,
+      "step": 528500
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.4388153363434606e-05,
+      "loss": 0.716,
+      "step": 528510
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.4387005433534857e-05,
+      "loss": 0.7412,
+      "step": 528520
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.438585750363511e-05,
+      "loss": 0.7697,
+      "step": 528530
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.4384709573735364e-05,
+      "loss": 0.7373,
+      "step": 528540
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.4383561643835614e-05,
+      "loss": 0.765,
+      "step": 528550
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.4382413713935867e-05,
+      "loss": 0.8024,
+      "step": 528560
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.438126578403612e-05,
+      "loss": 0.8256,
+      "step": 528570
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.4380117854136372e-05,
+      "loss": 0.6968,
+      "step": 528580
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.4378969924236625e-05,
+      "loss": 0.7968,
+      "step": 528590
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.4377821994336878e-05,
+      "loss": 0.7341,
+      "step": 528600
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.437667406443713e-05,
+      "loss": 0.7721,
+      "step": 528610
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.4375526134537383e-05,
+      "loss": 0.7749,
+      "step": 528620
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.4374378204637636e-05,
+      "loss": 0.7941,
+      "step": 528630
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 1.4373230274737888e-05,
+      "loss": 0.7464,
+      "step": 528640
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.4372082344838141e-05,
+      "loss": 0.7362,
+      "step": 528650
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.4370934414938394e-05,
+      "loss": 0.743,
+      "step": 528660
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.4369786485038646e-05,
+      "loss": 0.721,
+      "step": 528670
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.43686385551389e-05,
+      "loss": 0.7026,
+      "step": 528680
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.4367490625239152e-05,
+      "loss": 0.7285,
+      "step": 528690
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.4366342695339402e-05,
+      "loss": 0.7468,
+      "step": 528700
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.4365194765439656e-05,
+      "loss": 0.7416,
+      "step": 528710
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.4364046835539909e-05,
+      "loss": 0.7291,
+      "step": 528720
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.436289890564016e-05,
+      "loss": 0.7121,
+      "step": 528730
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.4361750975740414e-05,
+      "loss": 0.8079,
+      "step": 528740
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.4360603045840667e-05,
+      "loss": 0.7242,
+      "step": 528750
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.4359455115940918e-05,
+      "loss": 0.7415,
+      "step": 528760
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.4358307186041172e-05,
+      "loss": 0.677,
+      "step": 528770
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.4357159256141423e-05,
+      "loss": 0.77,
+      "step": 528780
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.4356011326241676e-05,
+      "loss": 0.7582,
+      "step": 528790
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.435486339634193e-05,
+      "loss": 0.7617,
+      "step": 528800
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.4353715466442181e-05,
+      "loss": 0.7325,
+      "step": 528810
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.4352567536542434e-05,
+      "loss": 0.7263,
+      "step": 528820
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.4351419606642688e-05,
+      "loss": 0.7437,
+      "step": 528830
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.4350271676742937e-05,
+      "loss": 0.8122,
+      "step": 528840
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.434912374684319e-05,
+      "loss": 0.6877,
+      "step": 528850
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 1.4347975816943444e-05,
+      "loss": 0.7953,
+      "step": 528860
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.4346827887043696e-05,
+      "loss": 0.7395,
+      "step": 528870
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.4345679957143949e-05,
+      "loss": 0.7357,
+      "step": 528880
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.4344532027244202e-05,
+      "loss": 0.7808,
+      "step": 528890
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.4343384097344454e-05,
+      "loss": 0.7283,
+      "step": 528900
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.4342236167444707e-05,
+      "loss": 0.7127,
+      "step": 528910
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.434108823754496e-05,
+      "loss": 0.7017,
+      "step": 528920
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.4339940307645212e-05,
+      "loss": 0.7639,
+      "step": 528930
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.4338792377745465e-05,
+      "loss": 0.6851,
+      "step": 528940
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.4337644447845718e-05,
+      "loss": 0.8356,
+      "step": 528950
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.433649651794597e-05,
+      "loss": 0.7042,
+      "step": 528960
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.4335348588046223e-05,
+      "loss": 0.7153,
+      "step": 528970
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.4334200658146476e-05,
+      "loss": 0.6983,
+      "step": 528980
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.4333052728246726e-05,
+      "loss": 0.6629,
+      "step": 528990
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.433190479834698e-05,
+      "loss": 0.7213,
+      "step": 529000
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.4330756868447232e-05,
+      "loss": 0.7104,
+      "step": 529010
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.4329608938547484e-05,
+      "loss": 0.6978,
+      "step": 529020
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.4328461008647737e-05,
+      "loss": 0.685,
+      "step": 529030
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.432731307874799e-05,
+      "loss": 0.737,
+      "step": 529040
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.4326165148848242e-05,
+      "loss": 0.7712,
+      "step": 529050
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.4325017218948495e-05,
+      "loss": 0.7199,
+      "step": 529060
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.4323869289048749e-05,
+      "loss": 0.7605,
+      "step": 529070
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 1.4322721359149e-05,
+      "loss": 0.7262,
+      "step": 529080
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 1.4321573429249253e-05,
+      "loss": 0.7458,
+      "step": 529090
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 1.4320425499349507e-05,
+      "loss": 0.8173,
+      "step": 529100
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 1.4319277569449758e-05,
+      "loss": 0.7732,
+      "step": 529110
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 1.4318129639550011e-05,
+      "loss": 0.7557,
+      "step": 529120
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 1.4316981709650261e-05,
+      "loss": 0.7315,
+      "step": 529130
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 1.4315833779750514e-05,
+      "loss": 0.7971,
+      "step": 529140
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 1.4314685849850768e-05,
+      "loss": 0.7008,
+      "step": 529150
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 1.431353791995102e-05,
+      "loss": 0.7537,
+      "step": 529160
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 1.4312389990051272e-05,
+      "loss": 0.7244,
+      "step": 529170
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 1.4311242060151526e-05,
+      "loss": 0.7592,
+      "step": 529180
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 1.4310094130251777e-05,
+      "loss": 0.7301,
+      "step": 529190
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 1.430894620035203e-05,
+      "loss": 0.7684,
+      "step": 529200
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 1.4307798270452284e-05,
+      "loss": 0.7728,
+      "step": 529210
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 1.4306650340552535e-05,
+      "loss": 0.7485,
+      "step": 529220
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 1.4305502410652789e-05,
+      "loss": 0.6907,
+      "step": 529230
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 1.4304354480753042e-05,
+      "loss": 0.7967,
+      "step": 529240
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 1.4303206550853293e-05,
+      "loss": 0.6835,
+      "step": 529250
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 1.4302058620953547e-05,
+      "loss": 0.7754,
+      "step": 529260
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 1.43009106910538e-05,
+      "loss": 0.7545,
+      "step": 529270
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 1.429976276115405e-05,
+      "loss": 0.7127,
+      "step": 529280
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 1.4298614831254303e-05,
+      "loss": 0.7459,
+      "step": 529290
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.4297466901354556e-05,
+      "loss": 0.7056,
+      "step": 529300
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.4296318971454808e-05,
+      "loss": 0.7621,
+      "step": 529310
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.4295171041555061e-05,
+      "loss": 0.7443,
+      "step": 529320
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.4294023111655314e-05,
+      "loss": 0.6617,
+      "step": 529330
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.4292875181755566e-05,
+      "loss": 0.7268,
+      "step": 529340
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.4291727251855819e-05,
+      "loss": 0.7181,
+      "step": 529350
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.4290579321956072e-05,
+      "loss": 0.7859,
+      "step": 529360
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.4289431392056324e-05,
+      "loss": 0.7275,
+      "step": 529370
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.4288283462156577e-05,
+      "loss": 0.7608,
+      "step": 529380
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.428713553225683e-05,
+      "loss": 0.7927,
+      "step": 529390
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.4285987602357082e-05,
+      "loss": 0.661,
+      "step": 529400
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.4284839672457335e-05,
+      "loss": 0.763,
+      "step": 529410
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.4283691742557588e-05,
+      "loss": 0.7312,
+      "step": 529420
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.4282543812657838e-05,
+      "loss": 0.7313,
+      "step": 529430
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.4281395882758091e-05,
+      "loss": 0.8107,
+      "step": 529440
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.4280247952858345e-05,
+      "loss": 0.7961,
+      "step": 529450
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.4279100022958596e-05,
+      "loss": 0.7267,
+      "step": 529460
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.427795209305885e-05,
+      "loss": 0.6624,
+      "step": 529470
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.4276804163159101e-05,
+      "loss": 0.6893,
+      "step": 529480
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.4275656233259354e-05,
+      "loss": 0.72,
+      "step": 529490
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.4274508303359607e-05,
+      "loss": 0.7817,
+      "step": 529500
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 1.4273360373459859e-05,
+      "loss": 0.7118,
+      "step": 529510
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.4272212443560112e-05,
+      "loss": 0.7427,
+      "step": 529520
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.4271064513660365e-05,
+      "loss": 0.7418,
+      "step": 529530
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.4269916583760617e-05,
+      "loss": 0.7361,
+      "step": 529540
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.426876865386087e-05,
+      "loss": 0.8217,
+      "step": 529550
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.4267620723961124e-05,
+      "loss": 0.748,
+      "step": 529560
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.4266472794061373e-05,
+      "loss": 0.6797,
+      "step": 529570
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.4265324864161627e-05,
+      "loss": 0.7179,
+      "step": 529580
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.426417693426188e-05,
+      "loss": 0.7017,
+      "step": 529590
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.4263029004362131e-05,
+      "loss": 0.6958,
+      "step": 529600
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.4261881074462385e-05,
+      "loss": 0.7584,
+      "step": 529610
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.4260733144562638e-05,
+      "loss": 0.7125,
+      "step": 529620
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.425958521466289e-05,
+      "loss": 0.6624,
+      "step": 529630
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.4258437284763143e-05,
+      "loss": 0.7291,
+      "step": 529640
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.4257289354863396e-05,
+      "loss": 0.74,
+      "step": 529650
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.4256141424963647e-05,
+      "loss": 0.7716,
+      "step": 529660
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.42549934950639e-05,
+      "loss": 0.6906,
+      "step": 529670
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.4253845565164154e-05,
+      "loss": 0.6774,
+      "step": 529680
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.4252697635264405e-05,
+      "loss": 0.7668,
+      "step": 529690
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.4251549705364659e-05,
+      "loss": 0.6559,
+      "step": 529700
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.4250401775464912e-05,
+      "loss": 0.6987,
+      "step": 529710
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.4249253845565162e-05,
+      "loss": 0.7754,
+      "step": 529720
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 1.4248105915665415e-05,
+      "loss": 0.7581,
+      "step": 529730
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.4246957985765668e-05,
+      "loss": 0.7488,
+      "step": 529740
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.424581005586592e-05,
+      "loss": 0.7661,
+      "step": 529750
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.4244662125966173e-05,
+      "loss": 0.7277,
+      "step": 529760
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.4243514196066426e-05,
+      "loss": 0.6427,
+      "step": 529770
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.4242366266166678e-05,
+      "loss": 0.6878,
+      "step": 529780
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.4241218336266931e-05,
+      "loss": 0.7204,
+      "step": 529790
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.4240070406367184e-05,
+      "loss": 0.7028,
+      "step": 529800
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.4238922476467436e-05,
+      "loss": 0.6762,
+      "step": 529810
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.423777454656769e-05,
+      "loss": 0.7665,
+      "step": 529820
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.423662661666794e-05,
+      "loss": 0.6828,
+      "step": 529830
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.4235478686768194e-05,
+      "loss": 0.7198,
+      "step": 529840
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.4234330756868447e-05,
+      "loss": 0.7066,
+      "step": 529850
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.4233182826968697e-05,
+      "loss": 0.7626,
+      "step": 529860
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.423203489706895e-05,
+      "loss": 0.7595,
+      "step": 529870
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.4230886967169204e-05,
+      "loss": 0.7096,
+      "step": 529880
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.4229739037269455e-05,
+      "loss": 0.6721,
+      "step": 529890
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.4228591107369708e-05,
+      "loss": 0.7281,
+      "step": 529900
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.4227443177469962e-05,
+      "loss": 0.6923,
+      "step": 529910
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.4226295247570213e-05,
+      "loss": 0.7653,
+      "step": 529920
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.4225147317670466e-05,
+      "loss": 0.8394,
+      "step": 529930
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.422399938777072e-05,
+      "loss": 0.7326,
+      "step": 529940
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 1.4222851457870971e-05,
+      "loss": 0.7331,
+      "step": 529950
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.4221703527971224e-05,
+      "loss": 0.7749,
+      "step": 529960
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.4220555598071478e-05,
+      "loss": 0.7166,
+      "step": 529970
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.421940766817173e-05,
+      "loss": 0.8529,
+      "step": 529980
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.4218259738271982e-05,
+      "loss": 0.7024,
+      "step": 529990
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.4217111808372236e-05,
+      "loss": 0.7054,
+      "step": 530000
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.4215963878472486e-05,
+      "loss": 0.7005,
+      "step": 530010
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.4214815948572739e-05,
+      "loss": 0.7785,
+      "step": 530020
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.4213668018672992e-05,
+      "loss": 0.6634,
+      "step": 530030
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.4212520088773244e-05,
+      "loss": 0.751,
+      "step": 530040
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.4211372158873497e-05,
+      "loss": 0.6827,
+      "step": 530050
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.421022422897375e-05,
+      "loss": 0.8095,
+      "step": 530060
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.4209076299074002e-05,
+      "loss": 0.8556,
+      "step": 530070
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.4207928369174255e-05,
+      "loss": 0.7256,
+      "step": 530080
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.4206780439274508e-05,
+      "loss": 0.7429,
+      "step": 530090
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.420563250937476e-05,
+      "loss": 0.6931,
+      "step": 530100
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.4204484579475013e-05,
+      "loss": 0.7795,
+      "step": 530110
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.4203336649575266e-05,
+      "loss": 0.7569,
+      "step": 530120
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.4202188719675518e-05,
+      "loss": 0.7492,
+      "step": 530130
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.4201040789775771e-05,
+      "loss": 0.706,
+      "step": 530140
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.419989285987602e-05,
+      "loss": 0.7592,
+      "step": 530150
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.4198744929976274e-05,
+      "loss": 0.6895,
+      "step": 530160
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 1.4197597000076527e-05,
+      "loss": 0.8023,
+      "step": 530170
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 1.4196449070176779e-05,
+      "loss": 0.775,
+      "step": 530180
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 1.4195301140277032e-05,
+      "loss": 0.6145,
+      "step": 530190
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 1.4194153210377285e-05,
+      "loss": 0.6692,
+      "step": 530200
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 1.4193005280477537e-05,
+      "loss": 0.7546,
+      "step": 530210
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 1.419185735057779e-05,
+      "loss": 0.681,
+      "step": 530220
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 1.4190709420678043e-05,
+      "loss": 0.7722,
+      "step": 530230
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 1.4189561490778295e-05,
+      "loss": 0.7507,
+      "step": 530240
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 1.4188413560878548e-05,
+      "loss": 0.673,
+      "step": 530250
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 1.4187265630978801e-05,
+      "loss": 0.6093,
+      "step": 530260
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 1.4186117701079053e-05,
+      "loss": 0.7689,
+      "step": 530270
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 1.4184969771179306e-05,
+      "loss": 0.7352,
+      "step": 530280
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 1.418382184127956e-05,
+      "loss": 0.7665,
+      "step": 530290
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 1.418267391137981e-05,
+      "loss": 0.7216,
+      "step": 530300
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 1.4181525981480062e-05,
+      "loss": 0.7693,
+      "step": 530310
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 1.4180378051580316e-05,
+      "loss": 0.6809,
+      "step": 530320
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 1.4179230121680567e-05,
+      "loss": 0.7371,
+      "step": 530330
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 1.417808219178082e-05,
+      "loss": 0.7133,
+      "step": 530340
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 1.4176934261881074e-05,
+      "loss": 0.7134,
+      "step": 530350
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 1.4175786331981325e-05,
+      "loss": 0.7099,
+      "step": 530360
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 1.4174638402081579e-05,
+      "loss": 0.7583,
+      "step": 530370
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 1.4173490472181832e-05,
+      "loss": 0.8314,
+      "step": 530380
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.4172342542282083e-05,
+      "loss": 0.7205,
+      "step": 530390
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.4171194612382337e-05,
+      "loss": 0.6573,
+      "step": 530400
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.417004668248259e-05,
+      "loss": 0.7278,
+      "step": 530410
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.4168898752582841e-05,
+      "loss": 0.7328,
+      "step": 530420
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.4167750822683095e-05,
+      "loss": 0.736,
+      "step": 530430
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.4166602892783348e-05,
+      "loss": 0.7142,
+      "step": 530440
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.4165454962883598e-05,
+      "loss": 0.723,
+      "step": 530450
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.4164307032983851e-05,
+      "loss": 0.7716,
+      "step": 530460
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.4163159103084104e-05,
+      "loss": 0.727,
+      "step": 530470
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.4162011173184356e-05,
+      "loss": 0.742,
+      "step": 530480
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.4160863243284609e-05,
+      "loss": 0.7243,
+      "step": 530490
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.415971531338486e-05,
+      "loss": 0.7556,
+      "step": 530500
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.4158567383485114e-05,
+      "loss": 0.7278,
+      "step": 530510
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.4157419453585367e-05,
+      "loss": 0.7842,
+      "step": 530520
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.4156271523685619e-05,
+      "loss": 0.7326,
+      "step": 530530
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.4155123593785872e-05,
+      "loss": 0.7,
+      "step": 530540
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.4153975663886125e-05,
+      "loss": 0.7477,
+      "step": 530550
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.4152827733986377e-05,
+      "loss": 0.729,
+      "step": 530560
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.415167980408663e-05,
+      "loss": 0.7706,
+      "step": 530570
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.4150531874186883e-05,
+      "loss": 0.6808,
+      "step": 530580
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.4149383944287133e-05,
+      "loss": 0.7959,
+      "step": 530590
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 1.4148236014387386e-05,
+      "loss": 0.7668,
+      "step": 530600
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.414708808448764e-05,
+      "loss": 0.7828,
+      "step": 530610
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.4145940154587891e-05,
+      "loss": 0.7204,
+      "step": 530620
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.4144792224688144e-05,
+      "loss": 0.7386,
+      "step": 530630
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.4143644294788397e-05,
+      "loss": 0.7151,
+      "step": 530640
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.4142496364888649e-05,
+      "loss": 0.7472,
+      "step": 530650
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.4141348434988902e-05,
+      "loss": 0.775,
+      "step": 530660
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.4140200505089155e-05,
+      "loss": 0.8496,
+      "step": 530670
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.4139052575189407e-05,
+      "loss": 0.6582,
+      "step": 530680
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.413790464528966e-05,
+      "loss": 0.7234,
+      "step": 530690
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.4136756715389914e-05,
+      "loss": 0.7708,
+      "step": 530700
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.4135608785490165e-05,
+      "loss": 0.744,
+      "step": 530710
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.4134460855590418e-05,
+      "loss": 0.7201,
+      "step": 530720
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.4133312925690672e-05,
+      "loss": 0.6986,
+      "step": 530730
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.4132164995790921e-05,
+      "loss": 0.785,
+      "step": 530740
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.4131017065891175e-05,
+      "loss": 0.7292,
+      "step": 530750
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.4129869135991428e-05,
+      "loss": 0.6825,
+      "step": 530760
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.412872120609168e-05,
+      "loss": 0.821,
+      "step": 530770
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.4127573276191933e-05,
+      "loss": 0.7317,
+      "step": 530780
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.4126425346292186e-05,
+      "loss": 0.745,
+      "step": 530790
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.4125277416392437e-05,
+      "loss": 0.7816,
+      "step": 530800
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.412412948649269e-05,
+      "loss": 0.7825,
+      "step": 530810
+    },
+    {
+      "epoch": 24.35,
+      "learning_rate": 1.4122981556592944e-05,
+      "loss": 0.7352,
+      "step": 530820
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.4121833626693196e-05,
+      "loss": 0.7526,
+      "step": 530830
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.4120685696793449e-05,
+      "loss": 0.6604,
+      "step": 530840
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.4119537766893699e-05,
+      "loss": 0.7549,
+      "step": 530850
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.4118389836993954e-05,
+      "loss": 0.6829,
+      "step": 530860
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.4117241907094207e-05,
+      "loss": 0.7394,
+      "step": 530870
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.4116093977194457e-05,
+      "loss": 0.7629,
+      "step": 530880
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.411494604729471e-05,
+      "loss": 0.7223,
+      "step": 530890
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.4113798117394963e-05,
+      "loss": 0.6933,
+      "step": 530900
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.4112650187495215e-05,
+      "loss": 0.7254,
+      "step": 530910
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.4111502257595468e-05,
+      "loss": 0.7497,
+      "step": 530920
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.4110354327695721e-05,
+      "loss": 0.7347,
+      "step": 530930
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.4109206397795973e-05,
+      "loss": 0.7289,
+      "step": 530940
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.4108058467896226e-05,
+      "loss": 0.7661,
+      "step": 530950
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.410691053799648e-05,
+      "loss": 0.7355,
+      "step": 530960
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.410576260809673e-05,
+      "loss": 0.7632,
+      "step": 530970
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.4104614678196984e-05,
+      "loss": 0.7437,
+      "step": 530980
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.4103466748297237e-05,
+      "loss": 0.6491,
+      "step": 530990
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.4102318818397487e-05,
+      "loss": 0.6485,
+      "step": 531000
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.4101170888497742e-05,
+      "loss": 0.7496,
+      "step": 531010
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.4100022958597995e-05,
+      "loss": 0.7153,
+      "step": 531020
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.4098875028698245e-05,
+      "loss": 0.7064,
+      "step": 531030
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 1.4097727098798498e-05,
+      "loss": 0.8155,
+      "step": 531040
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 1.4096579168898752e-05,
+      "loss": 0.7087,
+      "step": 531050
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 1.4095431238999003e-05,
+      "loss": 0.7053,
+      "step": 531060
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 1.4094283309099256e-05,
+      "loss": 0.689,
+      "step": 531070
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 1.409313537919951e-05,
+      "loss": 0.7063,
+      "step": 531080
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 1.4091987449299761e-05,
+      "loss": 0.7428,
+      "step": 531090
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 1.4090839519400014e-05,
+      "loss": 0.7634,
+      "step": 531100
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 1.4089691589500268e-05,
+      "loss": 0.7322,
+      "step": 531110
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 1.408854365960052e-05,
+      "loss": 0.809,
+      "step": 531120
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 1.4087395729700772e-05,
+      "loss": 0.6907,
+      "step": 531130
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 1.4086247799801026e-05,
+      "loss": 0.7177,
+      "step": 531140
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 1.4085099869901276e-05,
+      "loss": 0.7424,
+      "step": 531150
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 1.408395194000153e-05,
+      "loss": 0.765,
+      "step": 531160
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 1.4082804010101784e-05,
+      "loss": 0.7047,
+      "step": 531170
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 1.4081656080202034e-05,
+      "loss": 0.7462,
+      "step": 531180
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 1.4080508150302287e-05,
+      "loss": 0.7297,
+      "step": 531190
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 1.4079360220402538e-05,
+      "loss": 0.7506,
+      "step": 531200
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 1.4078212290502792e-05,
+      "loss": 0.7667,
+      "step": 531210
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 1.4077064360603045e-05,
+      "loss": 0.7715,
+      "step": 531220
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 1.4075916430703296e-05,
+      "loss": 0.7014,
+      "step": 531230
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 1.407476850080355e-05,
+      "loss": 0.7287,
+      "step": 531240
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 1.4073620570903803e-05,
+      "loss": 0.7598,
+      "step": 531250
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.4072472641004054e-05,
+      "loss": 0.6628,
+      "step": 531260
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.4071324711104308e-05,
+      "loss": 0.6662,
+      "step": 531270
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.4070176781204561e-05,
+      "loss": 0.7183,
+      "step": 531280
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.406902885130481e-05,
+      "loss": 0.6994,
+      "step": 531290
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.4067880921405064e-05,
+      "loss": 0.7463,
+      "step": 531300
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.4066732991505319e-05,
+      "loss": 0.7256,
+      "step": 531310
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.4065585061605569e-05,
+      "loss": 0.6996,
+      "step": 531320
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.4064437131705822e-05,
+      "loss": 0.7251,
+      "step": 531330
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.4063289201806075e-05,
+      "loss": 0.7949,
+      "step": 531340
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.4062141271906327e-05,
+      "loss": 0.7298,
+      "step": 531350
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.406099334200658e-05,
+      "loss": 0.7218,
+      "step": 531360
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.4059845412106833e-05,
+      "loss": 0.7808,
+      "step": 531370
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.4058697482207085e-05,
+      "loss": 0.7122,
+      "step": 531380
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.4057549552307338e-05,
+      "loss": 0.7035,
+      "step": 531390
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.4056401622407591e-05,
+      "loss": 0.8107,
+      "step": 531400
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.4055253692507843e-05,
+      "loss": 0.6166,
+      "step": 531410
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.4054105762608096e-05,
+      "loss": 0.7405,
+      "step": 531420
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.405295783270835e-05,
+      "loss": 0.6898,
+      "step": 531430
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.40518099028086e-05,
+      "loss": 0.7028,
+      "step": 531440
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.4050661972908854e-05,
+      "loss": 0.7577,
+      "step": 531450
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.4049514043009107e-05,
+      "loss": 0.6994,
+      "step": 531460
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 1.4048366113109357e-05,
+      "loss": 0.7935,
+      "step": 531470
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.404721818320961e-05,
+      "loss": 0.7124,
+      "step": 531480
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.4046070253309864e-05,
+      "loss": 0.7037,
+      "step": 531490
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.4044922323410115e-05,
+      "loss": 0.7518,
+      "step": 531500
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.4043774393510369e-05,
+      "loss": 0.7538,
+      "step": 531510
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.404262646361062e-05,
+      "loss": 0.6826,
+      "step": 531520
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.4041478533710873e-05,
+      "loss": 0.7137,
+      "step": 531530
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.4040330603811127e-05,
+      "loss": 0.7742,
+      "step": 531540
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.4039182673911378e-05,
+      "loss": 0.6914,
+      "step": 531550
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.4038034744011631e-05,
+      "loss": 0.7559,
+      "step": 531560
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.4036886814111885e-05,
+      "loss": 0.8645,
+      "step": 531570
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.4035738884212134e-05,
+      "loss": 0.7889,
+      "step": 531580
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.4034590954312388e-05,
+      "loss": 0.7348,
+      "step": 531590
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.4033443024412643e-05,
+      "loss": 0.7849,
+      "step": 531600
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.4032295094512893e-05,
+      "loss": 0.7954,
+      "step": 531610
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.4031147164613146e-05,
+      "loss": 0.7311,
+      "step": 531620
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.4029999234713399e-05,
+      "loss": 0.7783,
+      "step": 531630
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.402885130481365e-05,
+      "loss": 0.7312,
+      "step": 531640
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.4027703374913904e-05,
+      "loss": 0.7732,
+      "step": 531650
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.4026555445014157e-05,
+      "loss": 0.7741,
+      "step": 531660
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.4025407515114409e-05,
+      "loss": 0.7414,
+      "step": 531670
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.4024259585214662e-05,
+      "loss": 0.7252,
+      "step": 531680
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 1.4023111655314915e-05,
+      "loss": 0.7457,
+      "step": 531690
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.4021963725415167e-05,
+      "loss": 0.7161,
+      "step": 531700
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.402081579551542e-05,
+      "loss": 0.7684,
+      "step": 531710
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.4019667865615673e-05,
+      "loss": 0.7378,
+      "step": 531720
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.4018519935715923e-05,
+      "loss": 0.6902,
+      "step": 531730
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.4017372005816176e-05,
+      "loss": 0.74,
+      "step": 531740
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.4016224075916431e-05,
+      "loss": 0.6409,
+      "step": 531750
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.4015076146016681e-05,
+      "loss": 0.6802,
+      "step": 531760
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.4013928216116934e-05,
+      "loss": 0.7911,
+      "step": 531770
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.4012780286217187e-05,
+      "loss": 0.7622,
+      "step": 531780
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.4011632356317439e-05,
+      "loss": 0.7762,
+      "step": 531790
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.4010484426417692e-05,
+      "loss": 0.7313,
+      "step": 531800
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.4009336496517946e-05,
+      "loss": 0.7834,
+      "step": 531810
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.4008188566618197e-05,
+      "loss": 0.7141,
+      "step": 531820
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.400704063671845e-05,
+      "loss": 0.6646,
+      "step": 531830
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.4005892706818704e-05,
+      "loss": 0.6668,
+      "step": 531840
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.4004744776918955e-05,
+      "loss": 0.7244,
+      "step": 531850
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.4003596847019208e-05,
+      "loss": 0.7276,
+      "step": 531860
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.4002448917119458e-05,
+      "loss": 0.723,
+      "step": 531870
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.4001300987219711e-05,
+      "loss": 0.6807,
+      "step": 531880
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.4000153057319965e-05,
+      "loss": 0.7948,
+      "step": 531890
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.3999005127420216e-05,
+      "loss": 0.7085,
+      "step": 531900
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 1.399785719752047e-05,
+      "loss": 0.7672,
+      "step": 531910
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 1.3996709267620723e-05,
+      "loss": 0.8184,
+      "step": 531920
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 1.3995561337720974e-05,
+      "loss": 0.7726,
+      "step": 531930
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 1.3994413407821227e-05,
+      "loss": 0.678,
+      "step": 531940
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 1.399326547792148e-05,
+      "loss": 0.6992,
+      "step": 531950
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 1.3992117548021732e-05,
+      "loss": 0.719,
+      "step": 531960
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 1.3990969618121986e-05,
+      "loss": 0.7849,
+      "step": 531970
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 1.3989821688222239e-05,
+      "loss": 0.7047,
+      "step": 531980
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 1.398867375832249e-05,
+      "loss": 0.6733,
+      "step": 531990
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 1.3987525828422744e-05,
+      "loss": 0.7024,
+      "step": 532000
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 1.3986377898522997e-05,
+      "loss": 0.8123,
+      "step": 532010
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 1.3985229968623247e-05,
+      "loss": 0.7387,
+      "step": 532020
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 1.39840820387235e-05,
+      "loss": 0.7368,
+      "step": 532030
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 1.3982934108823753e-05,
+      "loss": 0.7046,
+      "step": 532040
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 1.3981786178924005e-05,
+      "loss": 0.6888,
+      "step": 532050
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 1.3980638249024258e-05,
+      "loss": 0.7405,
+      "step": 532060
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 1.3979490319124511e-05,
+      "loss": 0.7829,
+      "step": 532070
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 1.3978342389224763e-05,
+      "loss": 0.7215,
+      "step": 532080
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 1.3977194459325016e-05,
+      "loss": 0.7673,
+      "step": 532090
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 1.397604652942527e-05,
+      "loss": 0.6673,
+      "step": 532100
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 1.397489859952552e-05,
+      "loss": 0.7234,
+      "step": 532110
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 1.3973750669625774e-05,
+      "loss": 0.7766,
+      "step": 532120
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 1.3972602739726027e-05,
+      "loss": 0.7371,
+      "step": 532130
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 1.3971454809826279e-05,
+      "loss": 0.7272,
+      "step": 532140
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 1.3970306879926532e-05,
+      "loss": 0.6631,
+      "step": 532150
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 1.3969158950026785e-05,
+      "loss": 0.7594,
+      "step": 532160
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 1.3968011020127035e-05,
+      "loss": 0.7301,
+      "step": 532170
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 1.3966863090227288e-05,
+      "loss": 0.7675,
+      "step": 532180
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 1.3965715160327542e-05,
+      "loss": 0.7283,
+      "step": 532190
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 1.3964567230427793e-05,
+      "loss": 0.7544,
+      "step": 532200
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 1.3963419300528046e-05,
+      "loss": 0.7936,
+      "step": 532210
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 1.3962271370628298e-05,
+      "loss": 0.7854,
+      "step": 532220
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 1.3961123440728551e-05,
+      "loss": 0.6952,
+      "step": 532230
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 1.3959975510828804e-05,
+      "loss": 0.7318,
+      "step": 532240
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 1.3958827580929056e-05,
+      "loss": 0.7531,
+      "step": 532250
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 1.395767965102931e-05,
+      "loss": 0.7499,
+      "step": 532260
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 1.3956531721129562e-05,
+      "loss": 0.6893,
+      "step": 532270
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 1.3955383791229814e-05,
+      "loss": 0.7216,
+      "step": 532280
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 1.3954235861330067e-05,
+      "loss": 0.7451,
+      "step": 532290
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 1.395308793143032e-05,
+      "loss": 0.7947,
+      "step": 532300
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 1.395194000153057e-05,
+      "loss": 0.7843,
+      "step": 532310
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 1.3950792071630824e-05,
+      "loss": 0.714,
+      "step": 532320
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 1.3949644141731077e-05,
+      "loss": 0.7453,
+      "step": 532330
+    },
+    {
+      "epoch": 24.42,
+      "learning_rate": 1.3948496211831328e-05,
+      "loss": 0.7808,
+      "step": 532340
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.3947348281931582e-05,
+      "loss": 0.7507,
+      "step": 532350
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.3946200352031835e-05,
+      "loss": 0.8235,
+      "step": 532360
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.3945052422132086e-05,
+      "loss": 0.7187,
+      "step": 532370
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.394390449223234e-05,
+      "loss": 0.8698,
+      "step": 532380
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.3942756562332593e-05,
+      "loss": 0.6935,
+      "step": 532390
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.3941608632432844e-05,
+      "loss": 0.7426,
+      "step": 532400
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.3940460702533098e-05,
+      "loss": 0.7442,
+      "step": 532410
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.3939312772633351e-05,
+      "loss": 0.6912,
+      "step": 532420
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.3938164842733602e-05,
+      "loss": 0.7551,
+      "step": 532430
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.3937016912833856e-05,
+      "loss": 0.7102,
+      "step": 532440
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.3935868982934109e-05,
+      "loss": 0.7607,
+      "step": 532450
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.3934721053034359e-05,
+      "loss": 0.6921,
+      "step": 532460
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.3933573123134612e-05,
+      "loss": 0.7726,
+      "step": 532470
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.3932425193234865e-05,
+      "loss": 0.8144,
+      "step": 532480
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.3931277263335117e-05,
+      "loss": 0.7824,
+      "step": 532490
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.393012933343537e-05,
+      "loss": 0.6938,
+      "step": 532500
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.3928981403535623e-05,
+      "loss": 0.808,
+      "step": 532510
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.3927833473635875e-05,
+      "loss": 0.7425,
+      "step": 532520
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.3926685543736128e-05,
+      "loss": 0.712,
+      "step": 532530
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.3925537613836381e-05,
+      "loss": 0.7733,
+      "step": 532540
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.3924389683936633e-05,
+      "loss": 0.7547,
+      "step": 532550
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 1.3923241754036886e-05,
+      "loss": 0.7227,
+      "step": 532560
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.3922093824137138e-05,
+      "loss": 0.7215,
+      "step": 532570
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.3920945894237391e-05,
+      "loss": 0.6925,
+      "step": 532580
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.3919797964337644e-05,
+      "loss": 0.7487,
+      "step": 532590
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.3918650034437894e-05,
+      "loss": 0.776,
+      "step": 532600
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.3917502104538147e-05,
+      "loss": 0.7722,
+      "step": 532610
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.39163541746384e-05,
+      "loss": 0.6473,
+      "step": 532620
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.3915206244738652e-05,
+      "loss": 0.7844,
+      "step": 532630
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.3914058314838905e-05,
+      "loss": 0.7162,
+      "step": 532640
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.3912910384939159e-05,
+      "loss": 0.6989,
+      "step": 532650
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.391176245503941e-05,
+      "loss": 0.7263,
+      "step": 532660
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.3910614525139663e-05,
+      "loss": 0.7294,
+      "step": 532670
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.3909466595239917e-05,
+      "loss": 0.7646,
+      "step": 532680
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.3908318665340168e-05,
+      "loss": 0.7536,
+      "step": 532690
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.3907170735440421e-05,
+      "loss": 0.7324,
+      "step": 532700
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.3906022805540675e-05,
+      "loss": 0.7044,
+      "step": 532710
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.3904874875640926e-05,
+      "loss": 0.6781,
+      "step": 532720
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.390372694574118e-05,
+      "loss": 0.6814,
+      "step": 532730
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.3902579015841433e-05,
+      "loss": 0.7286,
+      "step": 532740
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.3901431085941683e-05,
+      "loss": 0.7028,
+      "step": 532750
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.3900283156041936e-05,
+      "loss": 0.7816,
+      "step": 532760
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.3899135226142189e-05,
+      "loss": 0.7592,
+      "step": 532770
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 1.389798729624244e-05,
+      "loss": 0.666,
+      "step": 532780
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.3896839366342694e-05,
+      "loss": 0.7679,
+      "step": 532790
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.3895691436442947e-05,
+      "loss": 0.7398,
+      "step": 532800
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.3894543506543199e-05,
+      "loss": 0.7352,
+      "step": 532810
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.3893395576643452e-05,
+      "loss": 0.6861,
+      "step": 532820
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.3892247646743705e-05,
+      "loss": 0.6829,
+      "step": 532830
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.3891099716843957e-05,
+      "loss": 0.6707,
+      "step": 532840
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.388995178694421e-05,
+      "loss": 0.7171,
+      "step": 532850
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.3888803857044463e-05,
+      "loss": 0.782,
+      "step": 532860
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.3887655927144715e-05,
+      "loss": 0.6931,
+      "step": 532870
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.3886507997244968e-05,
+      "loss": 0.7489,
+      "step": 532880
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.3885360067345221e-05,
+      "loss": 0.8362,
+      "step": 532890
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.3884212137445471e-05,
+      "loss": 0.6819,
+      "step": 532900
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.3883064207545724e-05,
+      "loss": 0.763,
+      "step": 532910
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.3881916277645976e-05,
+      "loss": 0.7445,
+      "step": 532920
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.3880768347746229e-05,
+      "loss": 0.7504,
+      "step": 532930
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.3879620417846482e-05,
+      "loss": 0.6424,
+      "step": 532940
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.3878472487946734e-05,
+      "loss": 0.6909,
+      "step": 532950
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.3877324558046987e-05,
+      "loss": 0.7721,
+      "step": 532960
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.387617662814724e-05,
+      "loss": 0.7203,
+      "step": 532970
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.3875028698247492e-05,
+      "loss": 0.7445,
+      "step": 532980
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.3873880768347745e-05,
+      "loss": 0.7612,
+      "step": 532990
+    },
+    {
+      "epoch": 24.45,
+      "learning_rate": 1.3872732838447998e-05,
+      "loss": 0.7637,
+      "step": 533000
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 1.387158490854825e-05,
+      "loss": 0.7177,
+      "step": 533010
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 1.3870436978648503e-05,
+      "loss": 0.7927,
+      "step": 533020
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 1.3869289048748756e-05,
+      "loss": 0.782,
+      "step": 533030
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 1.3868141118849006e-05,
+      "loss": 0.7868,
+      "step": 533040
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 1.386699318894926e-05,
+      "loss": 0.7321,
+      "step": 533050
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 1.3865845259049513e-05,
+      "loss": 0.6944,
+      "step": 533060
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 1.3864697329149764e-05,
+      "loss": 0.8365,
+      "step": 533070
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 1.3863549399250018e-05,
+      "loss": 0.6778,
+      "step": 533080
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 1.386240146935027e-05,
+      "loss": 0.6922,
+      "step": 533090
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 1.3861253539450522e-05,
+      "loss": 0.7002,
+      "step": 533100
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 1.3860105609550776e-05,
+      "loss": 0.716,
+      "step": 533110
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 1.3858957679651029e-05,
+      "loss": 0.731,
+      "step": 533120
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 1.385780974975128e-05,
+      "loss": 0.7147,
+      "step": 533130
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 1.3856661819851534e-05,
+      "loss": 0.7497,
+      "step": 533140
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 1.3855513889951787e-05,
+      "loss": 0.7018,
+      "step": 533150
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 1.3854365960052038e-05,
+      "loss": 0.8403,
+      "step": 533160
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 1.3853218030152292e-05,
+      "loss": 0.7976,
+      "step": 533170
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 1.3852070100252545e-05,
+      "loss": 0.695,
+      "step": 533180
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 1.3850922170352795e-05,
+      "loss": 0.7212,
+      "step": 533190
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 1.3849774240453048e-05,
+      "loss": 0.7021,
+      "step": 533200
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 1.3848626310553301e-05,
+      "loss": 0.7294,
+      "step": 533210
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 1.3847478380653553e-05,
+      "loss": 0.6795,
+      "step": 533220
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 1.3846330450753806e-05,
+      "loss": 0.6701,
+      "step": 533230
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 1.3845182520854058e-05,
+      "loss": 0.7775,
+      "step": 533240
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 1.384403459095431e-05,
+      "loss": 0.6688,
+      "step": 533250
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 1.3842886661054564e-05,
+      "loss": 0.6948,
+      "step": 533260
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 1.3841738731154816e-05,
+      "loss": 0.728,
+      "step": 533270
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 1.3840590801255069e-05,
+      "loss": 0.7057,
+      "step": 533280
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 1.3839442871355322e-05,
+      "loss": 0.7122,
+      "step": 533290
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 1.3838294941455574e-05,
+      "loss": 0.7103,
+      "step": 533300
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 1.3837147011555827e-05,
+      "loss": 0.7017,
+      "step": 533310
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 1.383599908165608e-05,
+      "loss": 0.7396,
+      "step": 533320
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 1.383485115175633e-05,
+      "loss": 0.7711,
+      "step": 533330
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 1.3833703221856583e-05,
+      "loss": 0.7502,
+      "step": 533340
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 1.3832555291956836e-05,
+      "loss": 0.7803,
+      "step": 533350
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 1.3831407362057088e-05,
+      "loss": 0.7684,
+      "step": 533360
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 1.3830259432157341e-05,
+      "loss": 0.745,
+      "step": 533370
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 1.3829111502257594e-05,
+      "loss": 0.8002,
+      "step": 533380
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 1.3827963572357846e-05,
+      "loss": 0.6935,
+      "step": 533390
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 1.38268156424581e-05,
+      "loss": 0.6904,
+      "step": 533400
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 1.3825667712558352e-05,
+      "loss": 0.6804,
+      "step": 533410
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 1.3824519782658604e-05,
+      "loss": 0.6949,
+      "step": 533420
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 1.3823371852758857e-05,
+      "loss": 0.772,
+      "step": 533430
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.382222392285911e-05,
+      "loss": 0.6998,
+      "step": 533440
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.3821075992959362e-05,
+      "loss": 0.7736,
+      "step": 533450
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.3819928063059615e-05,
+      "loss": 0.7671,
+      "step": 533460
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.3818780133159869e-05,
+      "loss": 0.7429,
+      "step": 533470
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.3817632203260118e-05,
+      "loss": 0.7067,
+      "step": 533480
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.3816484273360372e-05,
+      "loss": 0.7191,
+      "step": 533490
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.3815336343460625e-05,
+      "loss": 0.7686,
+      "step": 533500
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.3814188413560876e-05,
+      "loss": 0.6904,
+      "step": 533510
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.381304048366113e-05,
+      "loss": 0.7328,
+      "step": 533520
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.3811892553761383e-05,
+      "loss": 0.741,
+      "step": 533530
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.3810744623861634e-05,
+      "loss": 0.7136,
+      "step": 533540
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.3809596693961888e-05,
+      "loss": 0.7357,
+      "step": 533550
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.3808448764062141e-05,
+      "loss": 0.6957,
+      "step": 533560
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.3807300834162393e-05,
+      "loss": 0.6568,
+      "step": 533570
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.3806152904262646e-05,
+      "loss": 0.7677,
+      "step": 533580
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.3805004974362897e-05,
+      "loss": 0.8036,
+      "step": 533590
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.380385704446315e-05,
+      "loss": 0.7186,
+      "step": 533600
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.3802709114563404e-05,
+      "loss": 0.6815,
+      "step": 533610
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.3801561184663654e-05,
+      "loss": 0.7593,
+      "step": 533620
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.3800413254763907e-05,
+      "loss": 0.7277,
+      "step": 533630
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.379926532486416e-05,
+      "loss": 0.7022,
+      "step": 533640
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 1.3798117394964412e-05,
+      "loss": 0.687,
+      "step": 533650
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.3796969465064665e-05,
+      "loss": 0.7029,
+      "step": 533660
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.3795821535164918e-05,
+      "loss": 0.6985,
+      "step": 533670
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.379467360526517e-05,
+      "loss": 0.6821,
+      "step": 533680
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.3793525675365423e-05,
+      "loss": 0.762,
+      "step": 533690
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.3792377745465676e-05,
+      "loss": 0.7344,
+      "step": 533700
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.3791229815565928e-05,
+      "loss": 0.7218,
+      "step": 533710
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.3790081885666181e-05,
+      "loss": 0.8804,
+      "step": 533720
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.3788933955766434e-05,
+      "loss": 0.749,
+      "step": 533730
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.3787786025866686e-05,
+      "loss": 0.7762,
+      "step": 533740
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.3786638095966939e-05,
+      "loss": 0.7537,
+      "step": 533750
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.3785490166067192e-05,
+      "loss": 0.7577,
+      "step": 533760
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.3784342236167442e-05,
+      "loss": 0.6863,
+      "step": 533770
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.3783194306267695e-05,
+      "loss": 0.8275,
+      "step": 533780
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.3782046376367949e-05,
+      "loss": 0.7531,
+      "step": 533790
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.37808984464682e-05,
+      "loss": 0.7656,
+      "step": 533800
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.3779750516568453e-05,
+      "loss": 0.7529,
+      "step": 533810
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.3778602586668707e-05,
+      "loss": 0.7856,
+      "step": 533820
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.3777454656768958e-05,
+      "loss": 0.7084,
+      "step": 533830
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.3776306726869211e-05,
+      "loss": 0.7918,
+      "step": 533840
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.3775158796969465e-05,
+      "loss": 0.7155,
+      "step": 533850
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.3774010867069716e-05,
+      "loss": 0.7539,
+      "step": 533860
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 1.377286293716997e-05,
+      "loss": 0.673,
+      "step": 533870
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 1.3771715007270223e-05,
+      "loss": 0.7272,
+      "step": 533880
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 1.3770567077370474e-05,
+      "loss": 0.7051,
+      "step": 533890
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 1.3769419147470727e-05,
+      "loss": 0.7664,
+      "step": 533900
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 1.376827121757098e-05,
+      "loss": 0.7705,
+      "step": 533910
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 1.376712328767123e-05,
+      "loss": 0.842,
+      "step": 533920
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 1.3765975357771484e-05,
+      "loss": 0.7224,
+      "step": 533930
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 1.3764827427871735e-05,
+      "loss": 0.774,
+      "step": 533940
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 1.3763679497971989e-05,
+      "loss": 0.7638,
+      "step": 533950
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 1.3762531568072242e-05,
+      "loss": 0.7165,
+      "step": 533960
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 1.3761383638172493e-05,
+      "loss": 0.7121,
+      "step": 533970
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 1.3760235708272747e-05,
+      "loss": 0.7727,
+      "step": 533980
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 1.3759087778373e-05,
+      "loss": 0.808,
+      "step": 533990
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 1.3757939848473251e-05,
+      "loss": 0.696,
+      "step": 534000
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 1.3756791918573505e-05,
+      "loss": 0.7687,
+      "step": 534010
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 1.3755643988673758e-05,
+      "loss": 0.6641,
+      "step": 534020
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 1.375449605877401e-05,
+      "loss": 0.7178,
+      "step": 534030
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 1.3753348128874263e-05,
+      "loss": 0.7945,
+      "step": 534040
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 1.3752200198974516e-05,
+      "loss": 0.7133,
+      "step": 534050
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 1.3751052269074766e-05,
+      "loss": 0.7767,
+      "step": 534060
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 1.3749904339175019e-05,
+      "loss": 0.7695,
+      "step": 534070
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 1.3748756409275272e-05,
+      "loss": 0.7708,
+      "step": 534080
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 1.3747608479375524e-05,
+      "loss": 0.7546,
+      "step": 534090
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 1.3746460549475777e-05,
+      "loss": 0.7626,
+      "step": 534100
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 1.374531261957603e-05,
+      "loss": 0.7094,
+      "step": 534110
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 1.3744164689676282e-05,
+      "loss": 0.7297,
+      "step": 534120
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 1.3743016759776535e-05,
+      "loss": 0.7578,
+      "step": 534130
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 1.3741868829876788e-05,
+      "loss": 0.7869,
+      "step": 534140
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 1.374072089997704e-05,
+      "loss": 0.6969,
+      "step": 534150
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 1.3739572970077293e-05,
+      "loss": 0.7907,
+      "step": 534160
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 1.3738425040177546e-05,
+      "loss": 0.7102,
+      "step": 534170
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 1.3737277110277798e-05,
+      "loss": 0.7152,
+      "step": 534180
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 1.3736129180378051e-05,
+      "loss": 0.6749,
+      "step": 534190
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 1.3734981250478304e-05,
+      "loss": 0.7404,
+      "step": 534200
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 1.3733833320578554e-05,
+      "loss": 0.7257,
+      "step": 534210
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 1.3732685390678808e-05,
+      "loss": 0.7487,
+      "step": 534220
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 1.373153746077906e-05,
+      "loss": 0.7028,
+      "step": 534230
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 1.3730389530879312e-05,
+      "loss": 0.6919,
+      "step": 534240
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 1.3729241600979566e-05,
+      "loss": 0.7354,
+      "step": 534250
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 1.3728093671079819e-05,
+      "loss": 0.7204,
+      "step": 534260
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 1.372694574118007e-05,
+      "loss": 0.7446,
+      "step": 534270
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 1.3725797811280324e-05,
+      "loss": 0.7104,
+      "step": 534280
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 1.3724649881380575e-05,
+      "loss": 0.7979,
+      "step": 534290
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 1.3723501951480828e-05,
+      "loss": 0.7193,
+      "step": 534300
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 1.3722354021581082e-05,
+      "loss": 0.6502,
+      "step": 534310
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 1.3721206091681333e-05,
+      "loss": 0.7356,
+      "step": 534320
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 1.3720058161781586e-05,
+      "loss": 0.7708,
+      "step": 534330
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 1.371891023188184e-05,
+      "loss": 0.704,
+      "step": 534340
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 1.371776230198209e-05,
+      "loss": 0.7578,
+      "step": 534350
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 1.3716614372082343e-05,
+      "loss": 0.736,
+      "step": 534360
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 1.3715466442182596e-05,
+      "loss": 0.6557,
+      "step": 534370
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 1.3714318512282848e-05,
+      "loss": 0.6754,
+      "step": 534380
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 1.37131705823831e-05,
+      "loss": 0.6617,
+      "step": 534390
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 1.3712022652483354e-05,
+      "loss": 0.7571,
+      "step": 534400
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 1.3710874722583606e-05,
+      "loss": 0.673,
+      "step": 534410
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 1.3709726792683859e-05,
+      "loss": 0.7587,
+      "step": 534420
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 1.3708578862784112e-05,
+      "loss": 0.8564,
+      "step": 534430
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 1.3707430932884364e-05,
+      "loss": 0.7173,
+      "step": 534440
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 1.3706283002984617e-05,
+      "loss": 0.7439,
+      "step": 534450
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 1.370513507308487e-05,
+      "loss": 0.7403,
+      "step": 534460
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 1.3703987143185122e-05,
+      "loss": 0.7012,
+      "step": 534470
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 1.3702839213285375e-05,
+      "loss": 0.7626,
+      "step": 534480
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 1.3701691283385628e-05,
+      "loss": 0.6507,
+      "step": 534490
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 1.3700543353485878e-05,
+      "loss": 0.778,
+      "step": 534500
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 1.3699395423586131e-05,
+      "loss": 0.6858,
+      "step": 534510
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 1.3698247493686384e-05,
+      "loss": 0.7261,
+      "step": 534520
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 1.3697099563786636e-05,
+      "loss": 0.7069,
+      "step": 534530
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 1.369595163388689e-05,
+      "loss": 0.7619,
+      "step": 534540
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 1.3694803703987143e-05,
+      "loss": 0.7185,
+      "step": 534550
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 1.3693655774087394e-05,
+      "loss": 0.6807,
+      "step": 534560
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 1.3692507844187647e-05,
+      "loss": 0.6722,
+      "step": 534570
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 1.36913599142879e-05,
+      "loss": 0.7261,
+      "step": 534580
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 1.3690211984388152e-05,
+      "loss": 0.7172,
+      "step": 534590
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 1.3689064054488405e-05,
+      "loss": 0.7871,
+      "step": 534600
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 1.3687916124588657e-05,
+      "loss": 0.7645,
+      "step": 534610
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 1.368676819468891e-05,
+      "loss": 0.7762,
+      "step": 534620
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 1.3685620264789163e-05,
+      "loss": 0.7436,
+      "step": 534630
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 1.3684472334889413e-05,
+      "loss": 0.8176,
+      "step": 534640
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 1.3683324404989666e-05,
+      "loss": 0.7361,
+      "step": 534650
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 1.368217647508992e-05,
+      "loss": 0.7364,
+      "step": 534660
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 1.3681028545190171e-05,
+      "loss": 0.7532,
+      "step": 534670
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 1.3679880615290424e-05,
+      "loss": 0.7164,
+      "step": 534680
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 1.3678732685390678e-05,
+      "loss": 0.7763,
+      "step": 534690
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 1.367758475549093e-05,
+      "loss": 0.7084,
+      "step": 534700
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 1.3676436825591183e-05,
+      "loss": 0.8028,
+      "step": 534710
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 1.3675288895691436e-05,
+      "loss": 0.7804,
+      "step": 534720
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 1.3674140965791687e-05,
+      "loss": 0.7815,
+      "step": 534730
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 1.367299303589194e-05,
+      "loss": 0.7406,
+      "step": 534740
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 1.3671845105992194e-05,
+      "loss": 0.7642,
+      "step": 534750
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 1.3670697176092445e-05,
+      "loss": 0.7678,
+      "step": 534760
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 1.3669549246192699e-05,
+      "loss": 0.8511,
+      "step": 534770
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 1.3668401316292952e-05,
+      "loss": 0.6422,
+      "step": 534780
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 1.3667253386393202e-05,
+      "loss": 0.7145,
+      "step": 534790
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 1.3666105456493455e-05,
+      "loss": 0.7577,
+      "step": 534800
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 1.3664957526593708e-05,
+      "loss": 0.7126,
+      "step": 534810
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 1.366380959669396e-05,
+      "loss": 0.7748,
+      "step": 534820
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 1.3662661666794213e-05,
+      "loss": 0.6789,
+      "step": 534830
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 1.3661513736894466e-05,
+      "loss": 0.6945,
+      "step": 534840
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 1.3660365806994718e-05,
+      "loss": 0.7514,
+      "step": 534850
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 1.3659217877094971e-05,
+      "loss": 0.7477,
+      "step": 534860
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 1.3658069947195224e-05,
+      "loss": 0.722,
+      "step": 534870
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 1.3656922017295476e-05,
+      "loss": 0.7934,
+      "step": 534880
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 1.3655774087395729e-05,
+      "loss": 0.6793,
+      "step": 534890
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 1.3654626157495982e-05,
+      "loss": 0.7784,
+      "step": 534900
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 1.3653478227596234e-05,
+      "loss": 0.7427,
+      "step": 534910
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 1.3652330297696487e-05,
+      "loss": 0.7722,
+      "step": 534920
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 1.365118236779674e-05,
+      "loss": 0.7214,
+      "step": 534930
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 1.365003443789699e-05,
+      "loss": 0.7455,
+      "step": 534940
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 1.3648886507997243e-05,
+      "loss": 0.7545,
+      "step": 534950
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 1.3647738578097495e-05,
+      "loss": 0.813,
+      "step": 534960
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 1.3646590648197748e-05,
+      "loss": 0.6844,
+      "step": 534970
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 1.3645442718298001e-05,
+      "loss": 0.7185,
+      "step": 534980
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 1.3644294788398253e-05,
+      "loss": 0.7512,
+      "step": 534990
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 1.3643146858498506e-05,
+      "loss": 0.7021,
+      "step": 535000
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 1.364199892859876e-05,
+      "loss": 0.6623,
+      "step": 535010
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 1.3640850998699011e-05,
+      "loss": 0.6899,
+      "step": 535020
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 1.3639703068799264e-05,
+      "loss": 0.7185,
+      "step": 535030
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 1.3638555138899517e-05,
+      "loss": 0.7383,
+      "step": 535040
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 1.3637407208999769e-05,
+      "loss": 0.7544,
+      "step": 535050
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 1.3636259279100022e-05,
+      "loss": 0.8085,
+      "step": 535060
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 1.3635111349200276e-05,
+      "loss": 0.6192,
+      "step": 535070
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 1.3633963419300525e-05,
+      "loss": 0.687,
+      "step": 535080
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 1.3632815489400779e-05,
+      "loss": 0.6876,
+      "step": 535090
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 1.3631667559501032e-05,
+      "loss": 0.725,
+      "step": 535100
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 1.3630519629601283e-05,
+      "loss": 0.7969,
+      "step": 535110
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 1.3629371699701537e-05,
+      "loss": 0.6716,
+      "step": 535120
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 1.362822376980179e-05,
+      "loss": 0.6986,
+      "step": 535130
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 1.3627075839902041e-05,
+      "loss": 0.7918,
+      "step": 535140
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 1.3625927910002295e-05,
+      "loss": 0.7714,
+      "step": 535150
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 1.3624779980102548e-05,
+      "loss": 0.7997,
+      "step": 535160
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 1.36236320502028e-05,
+      "loss": 0.7612,
+      "step": 535170
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 1.3622484120303053e-05,
+      "loss": 0.7819,
+      "step": 535180
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 1.3621336190403306e-05,
+      "loss": 0.7658,
+      "step": 535190
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 1.3620188260503558e-05,
+      "loss": 0.6981,
+      "step": 535200
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 1.361904033060381e-05,
+      "loss": 0.7247,
+      "step": 535210
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 1.3617892400704064e-05,
+      "loss": 0.7421,
+      "step": 535220
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 1.3616744470804314e-05,
+      "loss": 0.6805,
+      "step": 535230
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 1.3615596540904567e-05,
+      "loss": 0.8112,
+      "step": 535240
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 1.361444861100482e-05,
+      "loss": 0.7572,
+      "step": 535250
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 1.3613300681105072e-05,
+      "loss": 0.679,
+      "step": 535260
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 1.3612152751205325e-05,
+      "loss": 0.7281,
+      "step": 535270
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 1.3611004821305578e-05,
+      "loss": 0.7175,
+      "step": 535280
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 1.360985689140583e-05,
+      "loss": 0.6597,
+      "step": 535290
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 1.3608708961506083e-05,
+      "loss": 0.8213,
+      "step": 535300
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 1.3607561031606335e-05,
+      "loss": 0.6883,
+      "step": 535310
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 1.3606413101706588e-05,
+      "loss": 0.7843,
+      "step": 535320
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 1.3605265171806841e-05,
+      "loss": 0.6886,
+      "step": 535330
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 1.3604117241907093e-05,
+      "loss": 0.6991,
+      "step": 535340
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 1.3602969312007346e-05,
+      "loss": 0.6614,
+      "step": 535350
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 1.36018213821076e-05,
+      "loss": 0.6973,
+      "step": 535360
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 1.3600673452207849e-05,
+      "loss": 0.7338,
+      "step": 535370
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 1.3599525522308102e-05,
+      "loss": 0.7402,
+      "step": 535380
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 1.3598377592408356e-05,
+      "loss": 0.6728,
+      "step": 535390
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 1.3597229662508607e-05,
+      "loss": 0.7105,
+      "step": 535400
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 1.359608173260886e-05,
+      "loss": 0.7749,
+      "step": 535410
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 1.3594933802709114e-05,
+      "loss": 0.7418,
+      "step": 535420
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 1.3593785872809365e-05,
+      "loss": 0.7551,
+      "step": 535430
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 1.3592637942909618e-05,
+      "loss": 0.7184,
+      "step": 535440
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 1.3591490013009872e-05,
+      "loss": 0.7322,
+      "step": 535450
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 1.3590342083110123e-05,
+      "loss": 0.7903,
+      "step": 535460
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 1.3589194153210376e-05,
+      "loss": 0.7261,
+      "step": 535470
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 1.358804622331063e-05,
+      "loss": 0.7405,
+      "step": 535480
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 1.3586898293410881e-05,
+      "loss": 0.7185,
+      "step": 535490
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 1.3585750363511134e-05,
+      "loss": 0.8042,
+      "step": 535500
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 1.3584602433611388e-05,
+      "loss": 0.807,
+      "step": 535510
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 1.3583454503711638e-05,
+      "loss": 0.777,
+      "step": 535520
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 1.358230657381189e-05,
+      "loss": 0.7775,
+      "step": 535530
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 1.3581158643912144e-05,
+      "loss": 0.736,
+      "step": 535540
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 1.3580010714012396e-05,
+      "loss": 0.7716,
+      "step": 535550
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 1.3578862784112649e-05,
+      "loss": 0.7384,
+      "step": 535560
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 1.3577714854212902e-05,
+      "loss": 0.7094,
+      "step": 535570
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 1.3576566924313154e-05,
+      "loss": 0.6844,
+      "step": 535580
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 1.3575418994413407e-05,
+      "loss": 0.7379,
+      "step": 535590
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 1.357427106451366e-05,
+      "loss": 0.7339,
+      "step": 535600
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 1.3573123134613912e-05,
+      "loss": 0.6837,
+      "step": 535610
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 1.3571975204714165e-05,
+      "loss": 0.7008,
+      "step": 535620
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 1.3570827274814418e-05,
+      "loss": 0.6618,
+      "step": 535630
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 1.356967934491467e-05,
+      "loss": 0.7008,
+      "step": 535640
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 1.3568531415014923e-05,
+      "loss": 0.7032,
+      "step": 535650
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 1.3567383485115173e-05,
+      "loss": 0.6967,
+      "step": 535660
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 1.3566235555215426e-05,
+      "loss": 0.7541,
+      "step": 535670
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 1.356508762531568e-05,
+      "loss": 0.6967,
+      "step": 535680
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 1.356393969541593e-05,
+      "loss": 0.744,
+      "step": 535690
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 1.3562791765516184e-05,
+      "loss": 0.7016,
+      "step": 535700
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 1.3561643835616437e-05,
+      "loss": 0.7277,
+      "step": 535710
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 1.3560495905716689e-05,
+      "loss": 0.7607,
+      "step": 535720
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 1.3559347975816942e-05,
+      "loss": 0.7627,
+      "step": 535730
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 1.3558200045917195e-05,
+      "loss": 0.7589,
+      "step": 535740
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 1.3557052116017447e-05,
+      "loss": 0.6858,
+      "step": 535750
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 1.35559041861177e-05,
+      "loss": 0.7182,
+      "step": 535760
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 1.3554756256217953e-05,
+      "loss": 0.7715,
+      "step": 535770
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 1.3553608326318205e-05,
+      "loss": 0.7176,
+      "step": 535780
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 1.3552460396418458e-05,
+      "loss": 0.6868,
+      "step": 535790
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 1.3551312466518711e-05,
+      "loss": 0.8029,
+      "step": 535800
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 1.3550164536618961e-05,
+      "loss": 0.7196,
+      "step": 535810
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 1.3549016606719215e-05,
+      "loss": 0.7104,
+      "step": 535820
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 1.3547868676819468e-05,
+      "loss": 0.7298,
+      "step": 535830
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 1.354672074691972e-05,
+      "loss": 0.6913,
+      "step": 535840
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 1.3545572817019973e-05,
+      "loss": 0.7228,
+      "step": 535850
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 1.3544424887120226e-05,
+      "loss": 0.8227,
+      "step": 535860
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 1.3543276957220477e-05,
+      "loss": 0.7831,
+      "step": 535870
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 1.354212902732073e-05,
+      "loss": 0.6828,
+      "step": 535880
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 1.3540981097420984e-05,
+      "loss": 0.7084,
+      "step": 535890
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 1.3539833167521235e-05,
+      "loss": 0.721,
+      "step": 535900
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 1.3538685237621489e-05,
+      "loss": 0.7775,
+      "step": 535910
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 1.3537537307721742e-05,
+      "loss": 0.7574,
+      "step": 535920
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 1.3536389377821993e-05,
+      "loss": 0.7833,
+      "step": 535930
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 1.3535241447922247e-05,
+      "loss": 0.7682,
+      "step": 535940
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 1.35340935180225e-05,
+      "loss": 0.751,
+      "step": 535950
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 1.353294558812275e-05,
+      "loss": 0.7248,
+      "step": 535960
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 1.3531797658223003e-05,
+      "loss": 0.7059,
+      "step": 535970
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 1.3530649728323255e-05,
+      "loss": 0.758,
+      "step": 535980
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 1.3529501798423508e-05,
+      "loss": 0.8204,
+      "step": 535990
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 1.3528353868523761e-05,
+      "loss": 0.7192,
+      "step": 536000
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 1.3527205938624013e-05,
+      "loss": 0.6793,
+      "step": 536010
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 1.3526058008724266e-05,
+      "loss": 0.6275,
+      "step": 536020
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 1.3524910078824519e-05,
+      "loss": 0.7622,
+      "step": 536030
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 1.352376214892477e-05,
+      "loss": 0.7039,
+      "step": 536040
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 1.3522614219025024e-05,
+      "loss": 0.7214,
+      "step": 536050
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 1.3521466289125277e-05,
+      "loss": 0.7323,
+      "step": 536060
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 1.3520318359225529e-05,
+      "loss": 0.7305,
+      "step": 536070
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 1.3519170429325782e-05,
+      "loss": 0.6924,
+      "step": 536080
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 1.3518022499426035e-05,
+      "loss": 0.6605,
+      "step": 536090
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 1.3516874569526285e-05,
+      "loss": 0.7454,
+      "step": 536100
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 1.3515726639626538e-05,
+      "loss": 0.6716,
+      "step": 536110
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 1.3514578709726791e-05,
+      "loss": 0.7002,
+      "step": 536120
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 1.3513430779827043e-05,
+      "loss": 0.8081,
+      "step": 536130
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 1.3512282849927296e-05,
+      "loss": 0.7374,
+      "step": 536140
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 1.351113492002755e-05,
+      "loss": 0.7123,
+      "step": 536150
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 1.3509986990127801e-05,
+      "loss": 0.7675,
+      "step": 536160
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 1.3508839060228054e-05,
+      "loss": 0.7133,
+      "step": 536170
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 1.3507691130328308e-05,
+      "loss": 0.7272,
+      "step": 536180
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 1.3506543200428559e-05,
+      "loss": 0.7122,
+      "step": 536190
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 1.3505395270528812e-05,
+      "loss": 0.7698,
+      "step": 536200
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 1.3504247340629066e-05,
+      "loss": 0.6907,
+      "step": 536210
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 1.3503099410729317e-05,
+      "loss": 0.7857,
+      "step": 536220
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 1.350195148082957e-05,
+      "loss": 0.6641,
+      "step": 536230
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 1.3500803550929824e-05,
+      "loss": 0.6424,
+      "step": 536240
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 1.3499655621030073e-05,
+      "loss": 0.7417,
+      "step": 536250
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 1.3498507691130327e-05,
+      "loss": 0.7421,
+      "step": 536260
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 1.349735976123058e-05,
+      "loss": 0.775,
+      "step": 536270
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 1.3496211831330831e-05,
+      "loss": 0.6725,
+      "step": 536280
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 1.3495063901431085e-05,
+      "loss": 0.7431,
+      "step": 536290
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 1.3493915971531338e-05,
+      "loss": 0.768,
+      "step": 536300
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 1.349276804163159e-05,
+      "loss": 0.7383,
+      "step": 536310
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 1.3491620111731843e-05,
+      "loss": 0.7371,
+      "step": 536320
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 1.3490472181832094e-05,
+      "loss": 0.7603,
+      "step": 536330
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 1.3489324251932348e-05,
+      "loss": 0.7732,
+      "step": 536340
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 1.34881763220326e-05,
+      "loss": 0.7227,
+      "step": 536350
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 1.3487028392132852e-05,
+      "loss": 0.6602,
+      "step": 536360
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 1.3485880462233106e-05,
+      "loss": 0.8163,
+      "step": 536370
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 1.3484732532333359e-05,
+      "loss": 0.7235,
+      "step": 536380
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 1.3483584602433609e-05,
+      "loss": 0.7053,
+      "step": 536390
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 1.3482436672533862e-05,
+      "loss": 0.8057,
+      "step": 536400
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 1.3481288742634115e-05,
+      "loss": 0.712,
+      "step": 536410
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 1.3480140812734367e-05,
+      "loss": 0.8043,
+      "step": 536420
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 1.347899288283462e-05,
+      "loss": 0.8046,
+      "step": 536430
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 1.3477844952934873e-05,
+      "loss": 0.7508,
+      "step": 536440
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 1.3476697023035125e-05,
+      "loss": 0.7172,
+      "step": 536450
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 1.3475549093135378e-05,
+      "loss": 0.7123,
+      "step": 536460
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 1.3474401163235631e-05,
+      "loss": 0.7248,
+      "step": 536470
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 1.3473253233335883e-05,
+      "loss": 0.6976,
+      "step": 536480
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 1.3472105303436136e-05,
+      "loss": 0.7032,
+      "step": 536490
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 1.347095737353639e-05,
+      "loss": 0.7459,
+      "step": 536500
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 1.346980944363664e-05,
+      "loss": 0.7522,
+      "step": 536510
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 1.3468661513736894e-05,
+      "loss": 0.7387,
+      "step": 536520
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 1.3467513583837147e-05,
+      "loss": 0.7611,
+      "step": 536530
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 1.3466365653937397e-05,
+      "loss": 0.824,
+      "step": 536540
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 1.346521772403765e-05,
+      "loss": 0.8082,
+      "step": 536550
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 1.3464069794137904e-05,
+      "loss": 0.7594,
+      "step": 536560
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 1.3462921864238155e-05,
+      "loss": 0.7041,
+      "step": 536570
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 1.3461773934338408e-05,
+      "loss": 0.7442,
+      "step": 536580
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 1.3460626004438662e-05,
+      "loss": 0.766,
+      "step": 536590
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 1.3459478074538913e-05,
+      "loss": 0.7436,
+      "step": 536600
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 1.3458330144639166e-05,
+      "loss": 0.6418,
+      "step": 536610
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 1.345718221473942e-05,
+      "loss": 0.8018,
+      "step": 536620
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 1.3456034284839671e-05,
+      "loss": 0.6186,
+      "step": 536630
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 1.3454886354939924e-05,
+      "loss": 0.7181,
+      "step": 536640
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 1.3453738425040178e-05,
+      "loss": 0.7422,
+      "step": 536650
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 1.345259049514043e-05,
+      "loss": 0.6917,
+      "step": 536660
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 1.3451442565240683e-05,
+      "loss": 0.7155,
+      "step": 536670
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 1.3450294635340932e-05,
+      "loss": 0.717,
+      "step": 536680
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 1.3449146705441186e-05,
+      "loss": 0.7646,
+      "step": 536690
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 1.3447998775541439e-05,
+      "loss": 0.7916,
+      "step": 536700
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 1.344685084564169e-05,
+      "loss": 0.7097,
+      "step": 536710
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 1.3445702915741944e-05,
+      "loss": 0.6979,
+      "step": 536720
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 1.3444554985842197e-05,
+      "loss": 0.7799,
+      "step": 536730
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 1.3443407055942448e-05,
+      "loss": 0.6604,
+      "step": 536740
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 1.3442259126042702e-05,
+      "loss": 0.6408,
+      "step": 536750
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 1.3441111196142955e-05,
+      "loss": 0.7513,
+      "step": 536760
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 1.3439963266243206e-05,
+      "loss": 0.7405,
+      "step": 536770
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 1.343881533634346e-05,
+      "loss": 0.7708,
+      "step": 536780
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 1.3437667406443713e-05,
+      "loss": 0.6963,
+      "step": 536790
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 1.3436519476543964e-05,
+      "loss": 0.7607,
+      "step": 536800
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 1.3435371546644218e-05,
+      "loss": 0.7412,
+      "step": 536810
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 1.3434223616744471e-05,
+      "loss": 0.6871,
+      "step": 536820
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 1.343307568684472e-05,
+      "loss": 0.694,
+      "step": 536830
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 1.3431927756944974e-05,
+      "loss": 0.7185,
+      "step": 536840
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 1.3430779827045227e-05,
+      "loss": 0.7535,
+      "step": 536850
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 1.3429631897145479e-05,
+      "loss": 0.7177,
+      "step": 536860
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 1.3428483967245732e-05,
+      "loss": 0.7266,
+      "step": 536870
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 1.3427336037345985e-05,
+      "loss": 0.7712,
+      "step": 536880
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 1.3426188107446237e-05,
+      "loss": 0.6759,
+      "step": 536890
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 1.342504017754649e-05,
+      "loss": 0.7852,
+      "step": 536900
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 1.3423892247646743e-05,
+      "loss": 0.7884,
+      "step": 536910
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 1.3422744317746995e-05,
+      "loss": 0.7839,
+      "step": 536920
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 1.3421596387847248e-05,
+      "loss": 0.7962,
+      "step": 536930
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 1.3420448457947501e-05,
+      "loss": 0.7408,
+      "step": 536940
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 1.3419300528047753e-05,
+      "loss": 0.7232,
+      "step": 536950
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 1.3418152598148006e-05,
+      "loss": 0.7564,
+      "step": 536960
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 1.341700466824826e-05,
+      "loss": 0.8145,
+      "step": 536970
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 1.341585673834851e-05,
+      "loss": 0.7165,
+      "step": 536980
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 1.3414708808448763e-05,
+      "loss": 0.8333,
+      "step": 536990
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 1.3413560878549016e-05,
+      "loss": 0.7097,
+      "step": 537000
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 1.3412412948649267e-05,
+      "loss": 0.7871,
+      "step": 537010
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 1.341126501874952e-05,
+      "loss": 0.7388,
+      "step": 537020
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 1.3410117088849772e-05,
+      "loss": 0.752,
+      "step": 537030
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 1.3408969158950025e-05,
+      "loss": 0.6686,
+      "step": 537040
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 1.3407821229050279e-05,
+      "loss": 0.7276,
+      "step": 537050
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 1.340667329915053e-05,
+      "loss": 0.6726,
+      "step": 537060
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 1.3405525369250783e-05,
+      "loss": 0.7463,
+      "step": 537070
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 1.3404377439351037e-05,
+      "loss": 0.7221,
+      "step": 537080
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 1.3403229509451288e-05,
+      "loss": 0.7176,
+      "step": 537090
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 1.3402081579551541e-05,
+      "loss": 0.7353,
+      "step": 537100
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 1.3400933649651795e-05,
+      "loss": 0.7324,
+      "step": 537110
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 1.3399785719752045e-05,
+      "loss": 0.6677,
+      "step": 537120
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 1.3398637789852298e-05,
+      "loss": 0.6727,
+      "step": 537130
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 1.3397489859952551e-05,
+      "loss": 0.788,
+      "step": 537140
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 1.3396341930052803e-05,
+      "loss": 0.7645,
+      "step": 537150
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 1.3395194000153056e-05,
+      "loss": 0.7285,
+      "step": 537160
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 1.3394046070253309e-05,
+      "loss": 0.7088,
+      "step": 537170
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 1.339289814035356e-05,
+      "loss": 0.7256,
+      "step": 537180
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 1.3391750210453814e-05,
+      "loss": 0.8184,
+      "step": 537190
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 1.3390602280554067e-05,
+      "loss": 0.7597,
+      "step": 537200
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 1.3389454350654319e-05,
+      "loss": 0.685,
+      "step": 537210
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 1.3388306420754572e-05,
+      "loss": 0.7584,
+      "step": 537220
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 1.3387158490854825e-05,
+      "loss": 0.7662,
+      "step": 537230
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 1.3386010560955077e-05,
+      "loss": 0.7106,
+      "step": 537240
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 1.338486263105533e-05,
+      "loss": 0.7446,
+      "step": 537250
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 1.3383714701155583e-05,
+      "loss": 0.6908,
+      "step": 537260
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 1.3382566771255833e-05,
+      "loss": 0.7027,
+      "step": 537270
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 1.3381418841356086e-05,
+      "loss": 0.7505,
+      "step": 537280
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 1.338027091145634e-05,
+      "loss": 0.7855,
+      "step": 537290
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 1.3379122981556591e-05,
+      "loss": 0.7493,
+      "step": 537300
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 1.3377975051656844e-05,
+      "loss": 0.7182,
+      "step": 537310
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 1.3376827121757098e-05,
+      "loss": 0.7463,
+      "step": 537320
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 1.3375679191857349e-05,
+      "loss": 0.7894,
+      "step": 537330
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 1.3374531261957602e-05,
+      "loss": 0.7267,
+      "step": 537340
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 1.3373383332057856e-05,
+      "loss": 0.7683,
+      "step": 537350
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 1.3372235402158107e-05,
+      "loss": 0.6899,
+      "step": 537360
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 1.337108747225836e-05,
+      "loss": 0.7681,
+      "step": 537370
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 1.3369939542358612e-05,
+      "loss": 0.7562,
+      "step": 537380
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 1.3368791612458865e-05,
+      "loss": 0.687,
+      "step": 537390
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 1.3367643682559118e-05,
+      "loss": 0.7259,
+      "step": 537400
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 1.3366495752659368e-05,
+      "loss": 0.7523,
+      "step": 537410
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 1.3365347822759621e-05,
+      "loss": 0.7696,
+      "step": 537420
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 1.3364199892859875e-05,
+      "loss": 0.7937,
+      "step": 537430
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 1.3363051962960126e-05,
+      "loss": 0.7274,
+      "step": 537440
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 1.336190403306038e-05,
+      "loss": 0.8151,
+      "step": 537450
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 1.3360756103160633e-05,
+      "loss": 0.721,
+      "step": 537460
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 1.3359608173260884e-05,
+      "loss": 0.6799,
+      "step": 537470
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 1.3358460243361138e-05,
+      "loss": 0.7519,
+      "step": 537480
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 1.335731231346139e-05,
+      "loss": 0.7128,
+      "step": 537490
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 1.3356164383561642e-05,
+      "loss": 0.7252,
+      "step": 537500
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 1.3355016453661896e-05,
+      "loss": 0.7756,
+      "step": 537510
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 1.3353868523762149e-05,
+      "loss": 0.7684,
+      "step": 537520
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 1.33527205938624e-05,
+      "loss": 0.7994,
+      "step": 537530
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 1.3351572663962654e-05,
+      "loss": 0.7271,
+      "step": 537540
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 1.3350424734062907e-05,
+      "loss": 0.7397,
+      "step": 537550
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 1.3349276804163157e-05,
+      "loss": 0.7219,
+      "step": 537560
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 1.334812887426341e-05,
+      "loss": 0.7674,
+      "step": 537570
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 1.3346980944363663e-05,
+      "loss": 0.6941,
+      "step": 537580
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 1.3345833014463915e-05,
+      "loss": 0.6942,
+      "step": 537590
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 1.3344685084564168e-05,
+      "loss": 0.7589,
+      "step": 537600
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 1.3343537154664421e-05,
+      "loss": 0.72,
+      "step": 537610
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 1.3342389224764673e-05,
+      "loss": 0.7124,
+      "step": 537620
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 1.3341241294864926e-05,
+      "loss": 0.8116,
+      "step": 537630
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 1.334009336496518e-05,
+      "loss": 0.7809,
+      "step": 537640
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 1.333894543506543e-05,
+      "loss": 0.7272,
+      "step": 537650
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 1.3337797505165684e-05,
+      "loss": 0.8394,
+      "step": 537660
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 1.3336649575265937e-05,
+      "loss": 0.7537,
+      "step": 537670
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 1.3335501645366189e-05,
+      "loss": 0.7033,
+      "step": 537680
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 1.3334353715466442e-05,
+      "loss": 0.7579,
+      "step": 537690
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 1.3333205785566692e-05,
+      "loss": 0.7892,
+      "step": 537700
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 1.3332057855666945e-05,
+      "loss": 0.7164,
+      "step": 537710
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 1.3330909925767198e-05,
+      "loss": 0.7661,
+      "step": 537720
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 1.332976199586745e-05,
+      "loss": 0.6687,
+      "step": 537730
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 1.3328614065967703e-05,
+      "loss": 0.717,
+      "step": 537740
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 1.3327466136067956e-05,
+      "loss": 0.7657,
+      "step": 537750
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 1.3326318206168208e-05,
+      "loss": 0.7438,
+      "step": 537760
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 1.3325170276268461e-05,
+      "loss": 0.7032,
+      "step": 537770
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 1.3324022346368714e-05,
+      "loss": 0.7173,
+      "step": 537780
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 1.3322874416468966e-05,
+      "loss": 0.7588,
+      "step": 537790
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 1.332172648656922e-05,
+      "loss": 0.8063,
+      "step": 537800
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 1.3320578556669473e-05,
+      "loss": 0.6649,
+      "step": 537810
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 1.3319430626769724e-05,
+      "loss": 0.7679,
+      "step": 537820
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 1.3318282696869977e-05,
+      "loss": 0.727,
+      "step": 537830
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 1.331713476697023e-05,
+      "loss": 0.7374,
+      "step": 537840
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 1.331598683707048e-05,
+      "loss": 0.8162,
+      "step": 537850
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 1.3314838907170734e-05,
+      "loss": 0.7491,
+      "step": 537860
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 1.3313690977270987e-05,
+      "loss": 0.8148,
+      "step": 537870
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 1.3312543047371238e-05,
+      "loss": 0.7199,
+      "step": 537880
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 1.3311395117471492e-05,
+      "loss": 0.7822,
+      "step": 537890
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 1.3310247187571745e-05,
+      "loss": 0.6815,
+      "step": 537900
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 1.3309099257671996e-05,
+      "loss": 0.7158,
+      "step": 537910
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 1.330795132777225e-05,
+      "loss": 0.7266,
+      "step": 537920
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 1.3306803397872503e-05,
+      "loss": 0.7007,
+      "step": 537930
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 1.3305655467972755e-05,
+      "loss": 0.7353,
+      "step": 537940
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 1.3304507538073008e-05,
+      "loss": 0.7655,
+      "step": 537950
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 1.3303359608173261e-05,
+      "loss": 0.7903,
+      "step": 537960
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 1.3302211678273513e-05,
+      "loss": 0.6892,
+      "step": 537970
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 1.3301063748373766e-05,
+      "loss": 0.7314,
+      "step": 537980
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 1.3299915818474019e-05,
+      "loss": 0.6352,
+      "step": 537990
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 1.3298767888574269e-05,
+      "loss": 0.681,
+      "step": 538000
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 1.3297619958674522e-05,
+      "loss": 0.7142,
+      "step": 538010
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 1.3296472028774775e-05,
+      "loss": 0.6511,
+      "step": 538020
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 1.3295324098875027e-05,
+      "loss": 0.7873,
+      "step": 538030
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 1.329417616897528e-05,
+      "loss": 0.6768,
+      "step": 538040
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 1.3293028239075532e-05,
+      "loss": 0.7479,
+      "step": 538050
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 1.3291880309175785e-05,
+      "loss": 0.8301,
+      "step": 538060
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 1.3290732379276038e-05,
+      "loss": 0.667,
+      "step": 538070
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 1.328958444937629e-05,
+      "loss": 0.7773,
+      "step": 538080
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 1.3288436519476543e-05,
+      "loss": 0.7257,
+      "step": 538090
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 1.3287288589576796e-05,
+      "loss": 0.789,
+      "step": 538100
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 1.3286140659677048e-05,
+      "loss": 0.7297,
+      "step": 538110
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 1.3284992729777301e-05,
+      "loss": 0.831,
+      "step": 538120
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 1.3283844799877554e-05,
+      "loss": 0.7859,
+      "step": 538130
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 1.3282696869977804e-05,
+      "loss": 0.7965,
+      "step": 538140
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 1.3281548940078057e-05,
+      "loss": 0.7838,
+      "step": 538150
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 1.328040101017831e-05,
+      "loss": 0.6556,
+      "step": 538160
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 1.3279253080278562e-05,
+      "loss": 0.7079,
+      "step": 538170
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 1.3278105150378815e-05,
+      "loss": 0.7186,
+      "step": 538180
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 1.3276957220479069e-05,
+      "loss": 0.7106,
+      "step": 538190
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 1.327580929057932e-05,
+      "loss": 0.7417,
+      "step": 538200
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 1.3274661360679573e-05,
+      "loss": 0.7343,
+      "step": 538210
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 1.3273513430779827e-05,
+      "loss": 0.7453,
+      "step": 538220
+    },
+    {
+      "epoch": 24.69,
+      "learning_rate": 1.3272365500880078e-05,
+      "loss": 0.7631,
+      "step": 538230
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 1.3271217570980331e-05,
+      "loss": 0.7217,
+      "step": 538240
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 1.3270069641080585e-05,
+      "loss": 0.7748,
+      "step": 538250
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 1.3268921711180836e-05,
+      "loss": 0.7048,
+      "step": 538260
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 1.326777378128109e-05,
+      "loss": 0.6787,
+      "step": 538270
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 1.3266625851381343e-05,
+      "loss": 0.8727,
+      "step": 538280
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 1.3265477921481593e-05,
+      "loss": 0.7154,
+      "step": 538290
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 1.3264329991581846e-05,
+      "loss": 0.8504,
+      "step": 538300
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 1.3263182061682099e-05,
+      "loss": 0.7315,
+      "step": 538310
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 1.326203413178235e-05,
+      "loss": 0.7099,
+      "step": 538320
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 1.3260886201882604e-05,
+      "loss": 0.6333,
+      "step": 538330
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 1.3259738271982857e-05,
+      "loss": 0.7565,
+      "step": 538340
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 1.3258590342083109e-05,
+      "loss": 0.7554,
+      "step": 538350
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 1.3257442412183362e-05,
+      "loss": 0.7443,
+      "step": 538360
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 1.3256294482283615e-05,
+      "loss": 0.6663,
+      "step": 538370
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 1.3255146552383867e-05,
+      "loss": 0.7601,
+      "step": 538380
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 1.325399862248412e-05,
+      "loss": 0.8517,
+      "step": 538390
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 1.3252850692584371e-05,
+      "loss": 0.7855,
+      "step": 538400
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 1.3251702762684625e-05,
+      "loss": 0.72,
+      "step": 538410
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 1.3250554832784878e-05,
+      "loss": 0.7655,
+      "step": 538420
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 1.3249406902885128e-05,
+      "loss": 0.7139,
+      "step": 538430
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 1.3248258972985381e-05,
+      "loss": 0.6663,
+      "step": 538440
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 1.3247111043085634e-05,
+      "loss": 0.7356,
+      "step": 538450
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 1.3245963113185886e-05,
+      "loss": 0.7619,
+      "step": 538460
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 1.3244815183286139e-05,
+      "loss": 0.7594,
+      "step": 538470
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 1.3243667253386392e-05,
+      "loss": 0.7634,
+      "step": 538480
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 1.3242519323486644e-05,
+      "loss": 0.8032,
+      "step": 538490
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 1.3241371393586897e-05,
+      "loss": 0.7481,
+      "step": 538500
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 1.324022346368715e-05,
+      "loss": 0.7142,
+      "step": 538510
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 1.3239075533787402e-05,
+      "loss": 0.7355,
+      "step": 538520
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 1.3237927603887655e-05,
+      "loss": 0.7312,
+      "step": 538530
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 1.3236779673987908e-05,
+      "loss": 0.7137,
+      "step": 538540
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 1.323563174408816e-05,
+      "loss": 0.7565,
+      "step": 538550
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 1.3234483814188413e-05,
+      "loss": 0.7086,
+      "step": 538560
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 1.3233335884288666e-05,
+      "loss": 0.7257,
+      "step": 538570
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 1.3232187954388916e-05,
+      "loss": 0.7762,
+      "step": 538580
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 1.323104002448917e-05,
+      "loss": 0.686,
+      "step": 538590
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 1.3229892094589423e-05,
+      "loss": 0.7538,
+      "step": 538600
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 1.3228744164689674e-05,
+      "loss": 0.7376,
+      "step": 538610
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 1.3227596234789928e-05,
+      "loss": 0.7337,
+      "step": 538620
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 1.322644830489018e-05,
+      "loss": 0.7267,
+      "step": 538630
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 1.3225300374990432e-05,
+      "loss": 0.7412,
+      "step": 538640
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 1.3224152445090686e-05,
+      "loss": 0.7789,
+      "step": 538650
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 1.3223004515190939e-05,
+      "loss": 0.7253,
+      "step": 538660
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 1.322185658529119e-05,
+      "loss": 0.7185,
+      "step": 538670
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 1.3220708655391444e-05,
+      "loss": 0.7536,
+      "step": 538680
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 1.3219560725491697e-05,
+      "loss": 0.7357,
+      "step": 538690
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 1.3218412795591948e-05,
+      "loss": 0.6792,
+      "step": 538700
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 1.3217264865692202e-05,
+      "loss": 0.7278,
+      "step": 538710
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 1.3216116935792455e-05,
+      "loss": 0.7522,
+      "step": 538720
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 1.3214969005892705e-05,
+      "loss": 0.7367,
+      "step": 538730
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 1.3213821075992958e-05,
+      "loss": 0.7706,
+      "step": 538740
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 1.321267314609321e-05,
+      "loss": 0.7686,
+      "step": 538750
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 1.3211525216193463e-05,
+      "loss": 0.7067,
+      "step": 538760
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 1.3210377286293716e-05,
+      "loss": 0.7823,
+      "step": 538770
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 1.3209229356393968e-05,
+      "loss": 0.7437,
+      "step": 538780
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 1.320808142649422e-05,
+      "loss": 0.7046,
+      "step": 538790
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 1.3206933496594474e-05,
+      "loss": 0.7245,
+      "step": 538800
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 1.3205785566694726e-05,
+      "loss": 0.6835,
+      "step": 538810
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 1.3204637636794979e-05,
+      "loss": 0.7585,
+      "step": 538820
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 1.3203489706895232e-05,
+      "loss": 0.8197,
+      "step": 538830
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 1.3202341776995484e-05,
+      "loss": 0.7579,
+      "step": 538840
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 1.3201193847095737e-05,
+      "loss": 0.7972,
+      "step": 538850
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 1.320004591719599e-05,
+      "loss": 0.7855,
+      "step": 538860
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 1.319889798729624e-05,
+      "loss": 0.715,
+      "step": 538870
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 1.3197750057396493e-05,
+      "loss": 0.7702,
+      "step": 538880
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 1.3196602127496746e-05,
+      "loss": 0.701,
+      "step": 538890
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 1.3195454197596998e-05,
+      "loss": 0.7434,
+      "step": 538900
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 1.3194306267697251e-05,
+      "loss": 0.7362,
+      "step": 538910
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 1.3193158337797505e-05,
+      "loss": 0.6846,
+      "step": 538920
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 1.3192010407897756e-05,
+      "loss": 0.718,
+      "step": 538930
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 1.319086247799801e-05,
+      "loss": 0.7341,
+      "step": 538940
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 1.3189714548098263e-05,
+      "loss": 0.691,
+      "step": 538950
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 1.3188566618198514e-05,
+      "loss": 0.7263,
+      "step": 538960
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 1.3187418688298767e-05,
+      "loss": 0.7869,
+      "step": 538970
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 1.318627075839902e-05,
+      "loss": 0.7431,
+      "step": 538980
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 1.3185122828499272e-05,
+      "loss": 0.6949,
+      "step": 538990
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 1.3183974898599525e-05,
+      "loss": 0.8073,
+      "step": 539000
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 1.3182826968699779e-05,
+      "loss": 0.7341,
+      "step": 539010
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 1.3181679038800028e-05,
+      "loss": 0.7713,
+      "step": 539020
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 1.3180531108900282e-05,
+      "loss": 0.7234,
+      "step": 539030
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 1.3179383179000535e-05,
+      "loss": 0.7451,
+      "step": 539040
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 1.3178235249100786e-05,
+      "loss": 0.7526,
+      "step": 539050
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 1.317708731920104e-05,
+      "loss": 0.7821,
+      "step": 539060
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 1.3175939389301291e-05,
+      "loss": 0.7508,
+      "step": 539070
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 1.3174791459401545e-05,
+      "loss": 0.6786,
+      "step": 539080
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 1.3173643529501798e-05,
+      "loss": 0.6752,
+      "step": 539090
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 1.317249559960205e-05,
+      "loss": 0.6823,
+      "step": 539100
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 1.3171347669702303e-05,
+      "loss": 0.7322,
+      "step": 539110
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 1.3170199739802556e-05,
+      "loss": 0.7052,
+      "step": 539120
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 1.3169051809902807e-05,
+      "loss": 0.6807,
+      "step": 539130
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 1.316790388000306e-05,
+      "loss": 0.6598,
+      "step": 539140
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 1.3166755950103314e-05,
+      "loss": 0.7367,
+      "step": 539150
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 1.3165608020203564e-05,
+      "loss": 0.7752,
+      "step": 539160
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 1.3164460090303817e-05,
+      "loss": 0.7302,
+      "step": 539170
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 1.316331216040407e-05,
+      "loss": 0.7574,
+      "step": 539180
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 1.3162164230504322e-05,
+      "loss": 0.8031,
+      "step": 539190
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 1.3161016300604575e-05,
+      "loss": 0.7438,
+      "step": 539200
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 1.3159868370704828e-05,
+      "loss": 0.7183,
+      "step": 539210
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 1.315872044080508e-05,
+      "loss": 0.7995,
+      "step": 539220
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 1.3157572510905333e-05,
+      "loss": 0.8081,
+      "step": 539230
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 1.3156424581005586e-05,
+      "loss": 0.762,
+      "step": 539240
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 1.3155276651105838e-05,
+      "loss": 0.7923,
+      "step": 539250
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 1.3154128721206091e-05,
+      "loss": 0.7367,
+      "step": 539260
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 1.3152980791306344e-05,
+      "loss": 0.7754,
+      "step": 539270
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 1.3151832861406596e-05,
+      "loss": 0.7799,
+      "step": 539280
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 1.3150684931506849e-05,
+      "loss": 0.6523,
+      "step": 539290
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 1.3149537001607102e-05,
+      "loss": 0.6693,
+      "step": 539300
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 1.3148389071707352e-05,
+      "loss": 0.7584,
+      "step": 539310
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 1.3147241141807605e-05,
+      "loss": 0.62,
+      "step": 539320
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 1.3146093211907859e-05,
+      "loss": 0.6797,
+      "step": 539330
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 1.314494528200811e-05,
+      "loss": 0.7455,
+      "step": 539340
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 1.3143797352108363e-05,
+      "loss": 0.7402,
+      "step": 539350
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 1.3142649422208617e-05,
+      "loss": 0.7328,
+      "step": 539360
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 1.3141501492308868e-05,
+      "loss": 0.8007,
+      "step": 539370
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 1.3140353562409121e-05,
+      "loss": 0.7381,
+      "step": 539380
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 1.3139205632509375e-05,
+      "loss": 0.7284,
+      "step": 539390
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 1.3138057702609626e-05,
+      "loss": 0.7572,
+      "step": 539400
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 1.313690977270988e-05,
+      "loss": 0.7348,
+      "step": 539410
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 1.3135761842810131e-05,
+      "loss": 0.7252,
+      "step": 539420
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 1.3134613912910384e-05,
+      "loss": 0.7423,
+      "step": 539430
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 1.3133465983010638e-05,
+      "loss": 0.8147,
+      "step": 539440
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 1.3132318053110887e-05,
+      "loss": 0.7419,
+      "step": 539450
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 1.313117012321114e-05,
+      "loss": 0.7243,
+      "step": 539460
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 1.3130022193311394e-05,
+      "loss": 0.7546,
+      "step": 539470
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 1.3128874263411645e-05,
+      "loss": 0.7538,
+      "step": 539480
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 1.3127726333511899e-05,
+      "loss": 0.742,
+      "step": 539490
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 1.3126578403612152e-05,
+      "loss": 0.7269,
+      "step": 539500
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 1.3125430473712403e-05,
+      "loss": 0.8136,
+      "step": 539510
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 1.3124282543812657e-05,
+      "loss": 0.6754,
+      "step": 539520
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 1.312313461391291e-05,
+      "loss": 0.7975,
+      "step": 539530
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 1.3121986684013161e-05,
+      "loss": 0.814,
+      "step": 539540
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 1.3120838754113415e-05,
+      "loss": 0.7724,
+      "step": 539550
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 1.3119690824213668e-05,
+      "loss": 0.6956,
+      "step": 539560
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 1.311854289431392e-05,
+      "loss": 0.6952,
+      "step": 539570
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 1.3117394964414173e-05,
+      "loss": 0.7395,
+      "step": 539580
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 1.3116247034514426e-05,
+      "loss": 0.7134,
+      "step": 539590
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 1.3115099104614676e-05,
+      "loss": 0.6802,
+      "step": 539600
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 1.3113951174714929e-05,
+      "loss": 0.6901,
+      "step": 539610
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 1.3112803244815182e-05,
+      "loss": 0.6792,
+      "step": 539620
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 1.3111655314915434e-05,
+      "loss": 0.7538,
+      "step": 539630
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 1.3110507385015687e-05,
+      "loss": 0.7322,
+      "step": 539640
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 1.310935945511594e-05,
+      "loss": 0.6954,
+      "step": 539650
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 1.3108211525216192e-05,
+      "loss": 0.6938,
+      "step": 539660
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 1.3107063595316445e-05,
+      "loss": 0.7284,
+      "step": 539670
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 1.3105915665416698e-05,
+      "loss": 0.6897,
+      "step": 539680
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 1.310476773551695e-05,
+      "loss": 0.6807,
+      "step": 539690
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 1.3103619805617203e-05,
+      "loss": 0.642,
+      "step": 539700
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 1.3102471875717456e-05,
+      "loss": 0.7207,
+      "step": 539710
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 1.3101323945817708e-05,
+      "loss": 0.6919,
+      "step": 539720
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 1.3100176015917961e-05,
+      "loss": 0.7097,
+      "step": 539730
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 1.3099028086018214e-05,
+      "loss": 0.6885,
+      "step": 539740
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 1.3097880156118464e-05,
+      "loss": 0.7487,
+      "step": 539750
+    },
+    {
+      "epoch": 24.76,
+      "learning_rate": 1.3096732226218718e-05,
+      "loss": 0.7073,
+      "step": 539760
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 1.3095584296318969e-05,
+      "loss": 0.6861,
+      "step": 539770
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 1.3094436366419222e-05,
+      "loss": 0.7287,
+      "step": 539780
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 1.3093288436519476e-05,
+      "loss": 0.7567,
+      "step": 539790
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 1.3092140506619727e-05,
+      "loss": 0.7117,
+      "step": 539800
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 1.309099257671998e-05,
+      "loss": 0.7748,
+      "step": 539810
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 1.3089844646820234e-05,
+      "loss": 0.7592,
+      "step": 539820
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 1.3088696716920485e-05,
+      "loss": 0.7419,
+      "step": 539830
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 1.3087548787020738e-05,
+      "loss": 0.8129,
+      "step": 539840
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 1.3086400857120992e-05,
+      "loss": 0.801,
+      "step": 539850
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 1.3085252927221243e-05,
+      "loss": 0.7329,
+      "step": 539860
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 1.3084104997321496e-05,
+      "loss": 0.7708,
+      "step": 539870
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 1.308295706742175e-05,
+      "loss": 0.7134,
+      "step": 539880
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 1.3081809137522e-05,
+      "loss": 0.781,
+      "step": 539890
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 1.3080661207622253e-05,
+      "loss": 0.7433,
+      "step": 539900
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 1.3079513277722506e-05,
+      "loss": 0.7236,
+      "step": 539910
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 1.3078365347822758e-05,
+      "loss": 0.739,
+      "step": 539920
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 1.307721741792301e-05,
+      "loss": 0.6817,
+      "step": 539930
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 1.3076069488023264e-05,
+      "loss": 0.6953,
+      "step": 539940
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 1.3074921558123516e-05,
+      "loss": 0.7832,
+      "step": 539950
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 1.3073773628223769e-05,
+      "loss": 0.7448,
+      "step": 539960
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 1.3072625698324022e-05,
+      "loss": 0.716,
+      "step": 539970
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 1.3071477768424274e-05,
+      "loss": 0.7321,
+      "step": 539980
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 1.3070329838524527e-05,
+      "loss": 0.7685,
+      "step": 539990
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 1.306918190862478e-05,
+      "loss": 0.7488,
+      "step": 540000
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 1.3068033978725032e-05,
+      "loss": 0.7496,
+      "step": 540010
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 1.3066886048825285e-05,
+      "loss": 0.7219,
+      "step": 540020
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 1.3065738118925538e-05,
+      "loss": 0.7308,
+      "step": 540030
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 1.3064590189025788e-05,
+      "loss": 0.6739,
+      "step": 540040
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 1.3063442259126041e-05,
+      "loss": 0.745,
+      "step": 540050
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 1.3062294329226295e-05,
+      "loss": 0.7718,
+      "step": 540060
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 1.3061146399326546e-05,
+      "loss": 0.7389,
+      "step": 540070
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 1.30599984694268e-05,
+      "loss": 0.7837,
+      "step": 540080
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 1.3058850539527053e-05,
+      "loss": 0.7384,
+      "step": 540090
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 1.3057702609627304e-05,
+      "loss": 0.7773,
+      "step": 540100
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 1.3056554679727557e-05,
+      "loss": 0.6962,
+      "step": 540110
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 1.3055406749827809e-05,
+      "loss": 0.7621,
+      "step": 540120
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 1.3054258819928062e-05,
+      "loss": 0.7361,
+      "step": 540130
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 1.3053110890028315e-05,
+      "loss": 0.7735,
+      "step": 540140
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 1.3051962960128565e-05,
+      "loss": 0.6997,
+      "step": 540150
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 1.305081503022882e-05,
+      "loss": 0.7737,
+      "step": 540160
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 1.3049667100329073e-05,
+      "loss": 0.6987,
+      "step": 540170
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 1.3048519170429323e-05,
+      "loss": 0.7952,
+      "step": 540180
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 1.3047371240529577e-05,
+      "loss": 0.7604,
+      "step": 540190
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 1.304622331062983e-05,
+      "loss": 0.7603,
+      "step": 540200
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 1.3045075380730081e-05,
+      "loss": 0.7987,
+      "step": 540210
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 1.3043927450830335e-05,
+      "loss": 0.7574,
+      "step": 540220
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 1.3042779520930588e-05,
+      "loss": 0.7211,
+      "step": 540230
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 1.304163159103084e-05,
+      "loss": 0.6951,
+      "step": 540240
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 1.3040483661131093e-05,
+      "loss": 0.7758,
+      "step": 540250
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 1.3039335731231346e-05,
+      "loss": 0.784,
+      "step": 540260
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 1.3038187801331597e-05,
+      "loss": 0.7217,
+      "step": 540270
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 1.303703987143185e-05,
+      "loss": 0.7836,
+      "step": 540280
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 1.3035891941532104e-05,
+      "loss": 0.7431,
+      "step": 540290
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 1.3034744011632354e-05,
+      "loss": 0.8108,
+      "step": 540300
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 1.3033596081732609e-05,
+      "loss": 0.779,
+      "step": 540310
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 1.3032448151832862e-05,
+      "loss": 0.8308,
+      "step": 540320
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 1.3031300221933112e-05,
+      "loss": 0.7506,
+      "step": 540330
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 1.3030152292033365e-05,
+      "loss": 0.6637,
+      "step": 540340
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 1.3029004362133618e-05,
+      "loss": 0.7886,
+      "step": 540350
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 1.302785643223387e-05,
+      "loss": 0.8115,
+      "step": 540360
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 1.3026708502334123e-05,
+      "loss": 0.6985,
+      "step": 540370
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 1.3025560572434376e-05,
+      "loss": 0.6923,
+      "step": 540380
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 1.3024412642534628e-05,
+      "loss": 0.7008,
+      "step": 540390
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 1.3023264712634881e-05,
+      "loss": 0.7405,
+      "step": 540400
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 1.3022116782735134e-05,
+      "loss": 0.7974,
+      "step": 540410
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 1.3020968852835386e-05,
+      "loss": 0.6874,
+      "step": 540420
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 1.3019820922935639e-05,
+      "loss": 0.6987,
+      "step": 540430
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 1.3018672993035892e-05,
+      "loss": 0.6933,
+      "step": 540440
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 1.3017525063136142e-05,
+      "loss": 0.7242,
+      "step": 540450
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 1.3016377133236397e-05,
+      "loss": 0.7002,
+      "step": 540460
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 1.3015229203336647e-05,
+      "loss": 0.6368,
+      "step": 540470
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 1.30140812734369e-05,
+      "loss": 0.7453,
+      "step": 540480
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 1.3012933343537153e-05,
+      "loss": 0.7448,
+      "step": 540490
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 1.3011785413637405e-05,
+      "loss": 0.7154,
+      "step": 540500
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 1.3010637483737658e-05,
+      "loss": 0.7697,
+      "step": 540510
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 1.3009489553837911e-05,
+      "loss": 0.7235,
+      "step": 540520
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 1.3008341623938163e-05,
+      "loss": 0.7231,
+      "step": 540530
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 1.3007193694038416e-05,
+      "loss": 0.7176,
+      "step": 540540
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 1.300604576413867e-05,
+      "loss": 0.7641,
+      "step": 540550
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 1.3004897834238921e-05,
+      "loss": 0.6976,
+      "step": 540560
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 1.3003749904339174e-05,
+      "loss": 0.6945,
+      "step": 540570
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 1.3002601974439428e-05,
+      "loss": 0.7432,
+      "step": 540580
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 1.3001454044539677e-05,
+      "loss": 0.7232,
+      "step": 540590
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 1.300030611463993e-05,
+      "loss": 0.6682,
+      "step": 540600
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 1.2999158184740186e-05,
+      "loss": 0.7518,
+      "step": 540610
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 1.2998010254840435e-05,
+      "loss": 0.7118,
+      "step": 540620
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 1.2996862324940689e-05,
+      "loss": 0.7036,
+      "step": 540630
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 1.2995714395040942e-05,
+      "loss": 0.6999,
+      "step": 540640
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 1.2994566465141193e-05,
+      "loss": 0.7743,
+      "step": 540650
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 1.2993418535241447e-05,
+      "loss": 0.698,
+      "step": 540660
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 1.29922706053417e-05,
+      "loss": 0.7261,
+      "step": 540670
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 1.2991122675441952e-05,
+      "loss": 0.7585,
+      "step": 540680
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 1.2989974745542205e-05,
+      "loss": 0.648,
+      "step": 540690
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 1.2988826815642458e-05,
+      "loss": 0.7706,
+      "step": 540700
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 1.298767888574271e-05,
+      "loss": 0.7062,
+      "step": 540710
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 1.2986530955842963e-05,
+      "loss": 0.7498,
+      "step": 540720
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 1.2985383025943216e-05,
+      "loss": 0.6871,
+      "step": 540730
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 1.2984235096043466e-05,
+      "loss": 0.7227,
+      "step": 540740
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 1.2983087166143719e-05,
+      "loss": 0.6241,
+      "step": 540750
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 1.2981939236243974e-05,
+      "loss": 0.7271,
+      "step": 540760
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 1.2980791306344224e-05,
+      "loss": 0.7635,
+      "step": 540770
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 1.2979643376444477e-05,
+      "loss": 0.7012,
+      "step": 540780
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 1.2978495446544729e-05,
+      "loss": 0.749,
+      "step": 540790
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 1.2977347516644982e-05,
+      "loss": 0.8107,
+      "step": 540800
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 1.2976199586745235e-05,
+      "loss": 0.6446,
+      "step": 540810
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 1.2975051656845487e-05,
+      "loss": 0.6598,
+      "step": 540820
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 1.297390372694574e-05,
+      "loss": 0.7392,
+      "step": 540830
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 1.2972755797045993e-05,
+      "loss": 0.7878,
+      "step": 540840
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 1.2971607867146245e-05,
+      "loss": 0.7741,
+      "step": 540850
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 1.2970459937246498e-05,
+      "loss": 0.7157,
+      "step": 540860
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 1.2969312007346751e-05,
+      "loss": 0.73,
+      "step": 540870
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 1.2968164077447001e-05,
+      "loss": 0.7509,
+      "step": 540880
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 1.2967016147547254e-05,
+      "loss": 0.7699,
+      "step": 540890
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 1.2965868217647508e-05,
+      "loss": 0.7467,
+      "step": 540900
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 1.2964720287747759e-05,
+      "loss": 0.7996,
+      "step": 540910
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 1.2963572357848012e-05,
+      "loss": 0.7356,
+      "step": 540920
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 1.2962424427948266e-05,
+      "loss": 0.7881,
+      "step": 540930
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 1.2961276498048517e-05,
+      "loss": 0.7202,
+      "step": 540940
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 1.296012856814877e-05,
+      "loss": 0.7806,
+      "step": 540950
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 1.2958980638249024e-05,
+      "loss": 0.6873,
+      "step": 540960
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 1.2957832708349275e-05,
+      "loss": 0.6833,
+      "step": 540970
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 1.2956684778449528e-05,
+      "loss": 0.7749,
+      "step": 540980
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 1.2955536848549782e-05,
+      "loss": 0.8334,
+      "step": 540990
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 1.2954388918650033e-05,
+      "loss": 0.7045,
+      "step": 541000
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 1.2953240988750286e-05,
+      "loss": 0.6935,
+      "step": 541010
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 1.295209305885054e-05,
+      "loss": 0.6626,
+      "step": 541020
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 1.295094512895079e-05,
+      "loss": 0.7455,
+      "step": 541030
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 1.2949797199051043e-05,
+      "loss": 0.7677,
+      "step": 541040
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 1.2948649269151296e-05,
+      "loss": 0.7874,
+      "step": 541050
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 1.2947501339251548e-05,
+      "loss": 0.726,
+      "step": 541060
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 1.2946353409351801e-05,
+      "loss": 0.6886,
+      "step": 541070
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 1.2945205479452054e-05,
+      "loss": 0.7093,
+      "step": 541080
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 1.2944057549552306e-05,
+      "loss": 0.7235,
+      "step": 541090
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 1.2942909619652559e-05,
+      "loss": 0.7216,
+      "step": 541100
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 1.2941761689752812e-05,
+      "loss": 0.7822,
+      "step": 541110
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 1.2940613759853064e-05,
+      "loss": 0.6934,
+      "step": 541120
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 1.2939465829953317e-05,
+      "loss": 0.7839,
+      "step": 541130
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 1.2938317900053568e-05,
+      "loss": 0.7491,
+      "step": 541140
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 1.2937169970153822e-05,
+      "loss": 0.7139,
+      "step": 541150
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 1.2936022040254075e-05,
+      "loss": 0.7275,
+      "step": 541160
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 1.2934874110354325e-05,
+      "loss": 0.7469,
+      "step": 541170
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 1.2933726180454578e-05,
+      "loss": 0.7623,
+      "step": 541180
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 1.2932578250554831e-05,
+      "loss": 0.6997,
+      "step": 541190
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 1.2931430320655083e-05,
+      "loss": 0.6646,
+      "step": 541200
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 1.2930282390755336e-05,
+      "loss": 0.7415,
+      "step": 541210
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 1.292913446085559e-05,
+      "loss": 0.7191,
+      "step": 541220
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 1.2927986530955841e-05,
+      "loss": 0.7407,
+      "step": 541230
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 1.2926838601056094e-05,
+      "loss": 0.7985,
+      "step": 541240
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 1.2925690671156347e-05,
+      "loss": 0.7907,
+      "step": 541250
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 1.2924542741256599e-05,
+      "loss": 0.7444,
+      "step": 541260
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 1.2923394811356852e-05,
+      "loss": 0.7034,
+      "step": 541270
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 1.2922246881457105e-05,
+      "loss": 0.7129,
+      "step": 541280
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 1.2921098951557357e-05,
+      "loss": 0.7156,
+      "step": 541290
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 1.291995102165761e-05,
+      "loss": 0.6897,
+      "step": 541300
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 1.2918803091757863e-05,
+      "loss": 0.7391,
+      "step": 541310
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 1.2917655161858113e-05,
+      "loss": 0.7154,
+      "step": 541320
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 1.2916507231958367e-05,
+      "loss": 0.751,
+      "step": 541330
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 1.291535930205862e-05,
+      "loss": 0.7429,
+      "step": 541340
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 1.2914211372158871e-05,
+      "loss": 0.7398,
+      "step": 541350
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 1.2913063442259125e-05,
+      "loss": 0.6963,
+      "step": 541360
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 1.2911915512359378e-05,
+      "loss": 0.7641,
+      "step": 541370
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 1.291076758245963e-05,
+      "loss": 0.7352,
+      "step": 541380
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 1.2909619652559883e-05,
+      "loss": 0.7,
+      "step": 541390
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 1.2908471722660136e-05,
+      "loss": 0.7234,
+      "step": 541400
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 1.2907323792760387e-05,
+      "loss": 0.695,
+      "step": 541410
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 1.290617586286064e-05,
+      "loss": 0.7511,
+      "step": 541420
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 1.2905027932960894e-05,
+      "loss": 0.7251,
+      "step": 541430
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 1.2903880003061145e-05,
+      "loss": 0.7164,
+      "step": 541440
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 1.2902732073161399e-05,
+      "loss": 0.6749,
+      "step": 541450
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 1.2901584143261652e-05,
+      "loss": 0.8052,
+      "step": 541460
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 1.2900436213361902e-05,
+      "loss": 0.7249,
+      "step": 541470
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 1.2899288283462155e-05,
+      "loss": 0.7356,
+      "step": 541480
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 1.2898140353562407e-05,
+      "loss": 0.7602,
+      "step": 541490
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 1.289699242366266e-05,
+      "loss": 0.7144,
+      "step": 541500
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 1.2895844493762913e-05,
+      "loss": 0.725,
+      "step": 541510
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 1.2894696563863165e-05,
+      "loss": 0.6502,
+      "step": 541520
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 1.2893548633963418e-05,
+      "loss": 0.7115,
+      "step": 541530
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 1.2892400704063671e-05,
+      "loss": 0.7516,
+      "step": 541540
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 1.2891252774163923e-05,
+      "loss": 0.7624,
+      "step": 541550
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 1.2890104844264176e-05,
+      "loss": 0.7174,
+      "step": 541560
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 1.2888956914364429e-05,
+      "loss": 0.7696,
+      "step": 541570
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 1.288780898446468e-05,
+      "loss": 0.7618,
+      "step": 541580
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 1.2886661054564934e-05,
+      "loss": 0.7877,
+      "step": 541590
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 1.2885513124665187e-05,
+      "loss": 0.6961,
+      "step": 541600
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 1.2884365194765437e-05,
+      "loss": 0.7384,
+      "step": 541610
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 1.288321726486569e-05,
+      "loss": 0.6837,
+      "step": 541620
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 1.2882069334965943e-05,
+      "loss": 0.7104,
+      "step": 541630
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 1.2880921405066195e-05,
+      "loss": 0.754,
+      "step": 541640
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 1.2879773475166448e-05,
+      "loss": 0.8098,
+      "step": 541650
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 1.2878625545266702e-05,
+      "loss": 0.7771,
+      "step": 541660
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 1.2877477615366953e-05,
+      "loss": 0.7024,
+      "step": 541670
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 1.2876329685467206e-05,
+      "loss": 0.6928,
+      "step": 541680
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 1.287518175556746e-05,
+      "loss": 0.74,
+      "step": 541690
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 1.2874033825667711e-05,
+      "loss": 0.7792,
+      "step": 541700
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 1.2872885895767964e-05,
+      "loss": 0.6702,
+      "step": 541710
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 1.2871737965868218e-05,
+      "loss": 0.7136,
+      "step": 541720
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 1.2870590035968469e-05,
+      "loss": 0.7242,
+      "step": 541730
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 1.2869442106068722e-05,
+      "loss": 0.6741,
+      "step": 541740
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 1.2868294176168976e-05,
+      "loss": 0.733,
+      "step": 541750
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 1.2867146246269225e-05,
+      "loss": 0.7933,
+      "step": 541760
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 1.2865998316369479e-05,
+      "loss": 0.8094,
+      "step": 541770
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 1.2864850386469732e-05,
+      "loss": 0.7421,
+      "step": 541780
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 1.2863702456569983e-05,
+      "loss": 0.7701,
+      "step": 541790
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 1.2862554526670237e-05,
+      "loss": 0.7214,
+      "step": 541800
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 1.286140659677049e-05,
+      "loss": 0.7005,
+      "step": 541810
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 1.2860258666870742e-05,
+      "loss": 0.8064,
+      "step": 541820
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 1.2859110736970995e-05,
+      "loss": 0.8283,
+      "step": 541830
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 1.2857962807071246e-05,
+      "loss": 0.7211,
+      "step": 541840
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 1.28568148771715e-05,
+      "loss": 0.7512,
+      "step": 541850
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 1.2855666947271753e-05,
+      "loss": 0.6592,
+      "step": 541860
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 1.2854519017372004e-05,
+      "loss": 0.6836,
+      "step": 541870
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 1.2853371087472258e-05,
+      "loss": 0.7064,
+      "step": 541880
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 1.285222315757251e-05,
+      "loss": 0.7774,
+      "step": 541890
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 1.285107522767276e-05,
+      "loss": 0.7666,
+      "step": 541900
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 1.2849927297773014e-05,
+      "loss": 0.7272,
+      "step": 541910
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 1.2848779367873267e-05,
+      "loss": 0.6555,
+      "step": 541920
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 1.2847631437973519e-05,
+      "loss": 0.7524,
+      "step": 541930
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 1.2846483508073772e-05,
+      "loss": 0.742,
+      "step": 541940
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 1.2845335578174025e-05,
+      "loss": 0.7693,
+      "step": 541950
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 1.2844187648274277e-05,
+      "loss": 0.723,
+      "step": 541960
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 1.284303971837453e-05,
+      "loss": 0.7806,
+      "step": 541970
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 1.2841891788474783e-05,
+      "loss": 0.7115,
+      "step": 541980
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 1.2840743858575035e-05,
+      "loss": 0.7265,
+      "step": 541990
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 1.2839595928675288e-05,
+      "loss": 0.7244,
+      "step": 542000
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 1.2838447998775541e-05,
+      "loss": 0.7714,
+      "step": 542010
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 1.2837300068875793e-05,
+      "loss": 0.7241,
+      "step": 542020
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 1.2836152138976046e-05,
+      "loss": 0.8051,
+      "step": 542030
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 1.28350042090763e-05,
+      "loss": 0.6786,
+      "step": 542040
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 1.283385627917655e-05,
+      "loss": 0.7141,
+      "step": 542050
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 1.2832708349276802e-05,
+      "loss": 0.695,
+      "step": 542060
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 1.2831560419377056e-05,
+      "loss": 0.7657,
+      "step": 542070
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 1.2830412489477307e-05,
+      "loss": 0.7811,
+      "step": 542080
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 1.282926455957756e-05,
+      "loss": 0.7436,
+      "step": 542090
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 1.2828116629677814e-05,
+      "loss": 0.7315,
+      "step": 542100
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 1.2826968699778065e-05,
+      "loss": 0.7886,
+      "step": 542110
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 1.2825820769878318e-05,
+      "loss": 0.7475,
+      "step": 542120
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 1.2824672839978572e-05,
+      "loss": 0.7336,
+      "step": 542130
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 1.2823524910078823e-05,
+      "loss": 0.7877,
+      "step": 542140
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 1.2822376980179077e-05,
+      "loss": 0.6372,
+      "step": 542150
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 1.2821229050279328e-05,
+      "loss": 0.6621,
+      "step": 542160
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 1.2820081120379581e-05,
+      "loss": 0.8179,
+      "step": 542170
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 1.2818933190479835e-05,
+      "loss": 0.7889,
+      "step": 542180
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 1.2817785260580084e-05,
+      "loss": 0.7588,
+      "step": 542190
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 1.2816637330680338e-05,
+      "loss": 0.705,
+      "step": 542200
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 1.2815489400780591e-05,
+      "loss": 0.776,
+      "step": 542210
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 1.2814341470880842e-05,
+      "loss": 0.6977,
+      "step": 542220
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 1.2813193540981096e-05,
+      "loss": 0.7185,
+      "step": 542230
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 1.2812045611081349e-05,
+      "loss": 0.6596,
+      "step": 542240
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 1.28108976811816e-05,
+      "loss": 0.6894,
+      "step": 542250
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 1.2809749751281854e-05,
+      "loss": 0.7155,
+      "step": 542260
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 1.2808601821382107e-05,
+      "loss": 0.7547,
+      "step": 542270
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 1.2807453891482358e-05,
+      "loss": 0.7085,
+      "step": 542280
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 1.2806305961582612e-05,
+      "loss": 0.7126,
+      "step": 542290
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 1.2805158031682865e-05,
+      "loss": 0.7159,
+      "step": 542300
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 1.2804010101783117e-05,
+      "loss": 0.7022,
+      "step": 542310
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 1.280286217188337e-05,
+      "loss": 0.7476,
+      "step": 542320
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 1.2801714241983623e-05,
+      "loss": 0.8089,
+      "step": 542330
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 1.2800566312083873e-05,
+      "loss": 0.7814,
+      "step": 542340
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 1.2799418382184126e-05,
+      "loss": 0.8118,
+      "step": 542350
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 1.279827045228438e-05,
+      "loss": 0.8513,
+      "step": 542360
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 1.2797122522384631e-05,
+      "loss": 0.736,
+      "step": 542370
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 1.2795974592484884e-05,
+      "loss": 0.7402,
+      "step": 542380
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 1.2794826662585137e-05,
+      "loss": 0.7082,
+      "step": 542390
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 1.2793678732685389e-05,
+      "loss": 0.7357,
+      "step": 542400
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 1.2792530802785642e-05,
+      "loss": 0.8373,
+      "step": 542410
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 1.2791382872885895e-05,
+      "loss": 0.7012,
+      "step": 542420
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 1.2790234942986147e-05,
+      "loss": 0.7824,
+      "step": 542430
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 1.27890870130864e-05,
+      "loss": 0.7558,
+      "step": 542440
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 1.2787939083186653e-05,
+      "loss": 0.6653,
+      "step": 542450
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 1.2786791153286905e-05,
+      "loss": 0.7514,
+      "step": 542460
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 1.2785643223387158e-05,
+      "loss": 0.7166,
+      "step": 542470
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 1.2784495293487411e-05,
+      "loss": 0.7313,
+      "step": 542480
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 1.2783347363587661e-05,
+      "loss": 0.7359,
+      "step": 542490
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 1.2782199433687915e-05,
+      "loss": 0.7794,
+      "step": 542500
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 1.2781051503788166e-05,
+      "loss": 0.7794,
+      "step": 542510
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 1.277990357388842e-05,
+      "loss": 0.7753,
+      "step": 542520
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 1.2778755643988673e-05,
+      "loss": 0.7824,
+      "step": 542530
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 1.2777607714088924e-05,
+      "loss": 0.7888,
+      "step": 542540
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 1.2776459784189177e-05,
+      "loss": 0.7176,
+      "step": 542550
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 1.277531185428943e-05,
+      "loss": 0.8002,
+      "step": 542560
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 1.2774163924389682e-05,
+      "loss": 0.7356,
+      "step": 542570
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 1.2773015994489935e-05,
+      "loss": 0.7091,
+      "step": 542580
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 1.2771868064590189e-05,
+      "loss": 0.7718,
+      "step": 542590
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 1.277072013469044e-05,
+      "loss": 0.6901,
+      "step": 542600
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 1.2769572204790693e-05,
+      "loss": 0.7518,
+      "step": 542610
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 1.2768424274890947e-05,
+      "loss": 0.7055,
+      "step": 542620
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 1.2767276344991197e-05,
+      "loss": 0.6496,
+      "step": 542630
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 1.276612841509145e-05,
+      "loss": 0.7479,
+      "step": 542640
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 1.2764980485191703e-05,
+      "loss": 0.7264,
+      "step": 542650
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 1.2763832555291955e-05,
+      "loss": 0.753,
+      "step": 542660
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 1.2762684625392208e-05,
+      "loss": 0.6968,
+      "step": 542670
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 1.2761536695492461e-05,
+      "loss": 0.8069,
+      "step": 542680
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 1.2760388765592713e-05,
+      "loss": 0.8078,
+      "step": 542690
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 1.2759240835692966e-05,
+      "loss": 0.7591,
+      "step": 542700
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 1.2758092905793219e-05,
+      "loss": 0.7118,
+      "step": 542710
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 1.275694497589347e-05,
+      "loss": 0.7787,
+      "step": 542720
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 1.2755797045993724e-05,
+      "loss": 0.7674,
+      "step": 542730
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 1.2754649116093977e-05,
+      "loss": 0.7858,
+      "step": 542740
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 1.2753501186194229e-05,
+      "loss": 0.7033,
+      "step": 542750
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 1.2752353256294482e-05,
+      "loss": 0.6705,
+      "step": 542760
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 1.2751205326394735e-05,
+      "loss": 0.7773,
+      "step": 542770
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 1.2750057396494985e-05,
+      "loss": 0.7182,
+      "step": 542780
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 1.2748909466595238e-05,
+      "loss": 0.6924,
+      "step": 542790
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 1.2747761536695492e-05,
+      "loss": 0.7233,
+      "step": 542800
+    },
+    {
+      "epoch": 24.9,
+      "learning_rate": 1.2746613606795743e-05,
+      "loss": 0.7377,
+      "step": 542810
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 1.2745465676895996e-05,
+      "loss": 0.7465,
+      "step": 542820
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 1.274431774699625e-05,
+      "loss": 0.7501,
+      "step": 542830
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 1.2743169817096501e-05,
+      "loss": 0.7139,
+      "step": 542840
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 1.2742021887196754e-05,
+      "loss": 0.7828,
+      "step": 542850
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 1.2740873957297006e-05,
+      "loss": 0.8149,
+      "step": 542860
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 1.2739726027397259e-05,
+      "loss": 0.7017,
+      "step": 542870
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 1.2738578097497512e-05,
+      "loss": 0.8412,
+      "step": 542880
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 1.2737430167597764e-05,
+      "loss": 0.8581,
+      "step": 542890
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 1.2736282237698017e-05,
+      "loss": 0.7324,
+      "step": 542900
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 1.273513430779827e-05,
+      "loss": 0.7876,
+      "step": 542910
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 1.273398637789852e-05,
+      "loss": 0.763,
+      "step": 542920
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 1.2732838447998774e-05,
+      "loss": 0.7748,
+      "step": 542930
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 1.2731690518099027e-05,
+      "loss": 0.7185,
+      "step": 542940
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 1.2730542588199278e-05,
+      "loss": 0.6933,
+      "step": 542950
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 1.2729394658299532e-05,
+      "loss": 0.7933,
+      "step": 542960
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 1.2728246728399785e-05,
+      "loss": 0.7477,
+      "step": 542970
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 1.2727098798500036e-05,
+      "loss": 0.7888,
+      "step": 542980
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 1.272595086860029e-05,
+      "loss": 0.7725,
+      "step": 542990
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 1.2724802938700543e-05,
+      "loss": 0.7112,
+      "step": 543000
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 1.2723655008800794e-05,
+      "loss": 0.845,
+      "step": 543010
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 1.2722507078901048e-05,
+      "loss": 0.7625,
+      "step": 543020
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.2721359149001301e-05,
+      "loss": 0.7304,
+      "step": 543030
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.2720211219101552e-05,
+      "loss": 0.7852,
+      "step": 543040
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.2719063289201806e-05,
+      "loss": 0.6872,
+      "step": 543050
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.2717915359302059e-05,
+      "loss": 0.7242,
+      "step": 543060
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.2716767429402309e-05,
+      "loss": 0.7526,
+      "step": 543070
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.2715619499502562e-05,
+      "loss": 0.7287,
+      "step": 543080
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.2714471569602815e-05,
+      "loss": 0.7894,
+      "step": 543090
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.2713323639703067e-05,
+      "loss": 0.7393,
+      "step": 543100
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.271217570980332e-05,
+      "loss": 0.8125,
+      "step": 543110
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.2711027779903573e-05,
+      "loss": 0.7118,
+      "step": 543120
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.2709879850003825e-05,
+      "loss": 0.783,
+      "step": 543130
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.2708731920104078e-05,
+      "loss": 0.7746,
+      "step": 543140
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.2707583990204331e-05,
+      "loss": 0.6541,
+      "step": 543150
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.2706436060304583e-05,
+      "loss": 0.7158,
+      "step": 543160
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.2705288130404836e-05,
+      "loss": 0.6901,
+      "step": 543170
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.270414020050509e-05,
+      "loss": 0.6623,
+      "step": 543180
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.2702992270605341e-05,
+      "loss": 0.7467,
+      "step": 543190
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.2701844340705594e-05,
+      "loss": 0.7164,
+      "step": 543200
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.2700696410805844e-05,
+      "loss": 0.7333,
+      "step": 543210
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.2699548480906097e-05,
+      "loss": 0.7956,
+      "step": 543220
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.269840055100635e-05,
+      "loss": 0.8161,
+      "step": 543230
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 1.2697252621106602e-05,
+      "loss": 0.7667,
+      "step": 543240
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 1.2696104691206855e-05,
+      "loss": 0.7403,
+      "step": 543250
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 1.2694956761307108e-05,
+      "loss": 0.8105,
+      "step": 543260
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 1.269380883140736e-05,
+      "loss": 0.7353,
+      "step": 543270
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 1.2692660901507613e-05,
+      "loss": 0.8144,
+      "step": 543280
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 1.2691512971607867e-05,
+      "loss": 0.6282,
+      "step": 543290
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 1.2690365041708118e-05,
+      "loss": 0.8344,
+      "step": 543300
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 1.2689217111808371e-05,
+      "loss": 0.6845,
+      "step": 543310
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 1.2688069181908625e-05,
+      "loss": 0.7519,
+      "step": 543320
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 1.2686921252008876e-05,
+      "loss": 0.6957,
+      "step": 543330
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 1.268577332210913e-05,
+      "loss": 0.7663,
+      "step": 543340
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 1.2684625392209383e-05,
+      "loss": 0.7766,
+      "step": 543350
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 1.2683477462309632e-05,
+      "loss": 0.7888,
+      "step": 543360
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 1.2682329532409886e-05,
+      "loss": 0.7508,
+      "step": 543370
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 1.2681181602510139e-05,
+      "loss": 0.7575,
+      "step": 543380
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 1.268003367261039e-05,
+      "loss": 0.6858,
+      "step": 543390
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 1.2678885742710644e-05,
+      "loss": 0.7914,
+      "step": 543400
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 1.2677737812810897e-05,
+      "loss": 0.7635,
+      "step": 543410
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 1.2676589882911149e-05,
+      "loss": 0.7502,
+      "step": 543420
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 1.2675441953011402e-05,
+      "loss": 0.7567,
+      "step": 543430
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 1.2674294023111655e-05,
+      "loss": 0.7778,
+      "step": 543440
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 1.2673146093211907e-05,
+      "loss": 0.7713,
+      "step": 543450
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 1.267199816331216e-05,
+      "loss": 0.6603,
+      "step": 543460
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.2670850233412413e-05,
+      "loss": 0.7488,
+      "step": 543470
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.2669702303512665e-05,
+      "loss": 0.7648,
+      "step": 543480
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.2668554373612918e-05,
+      "loss": 0.8306,
+      "step": 543490
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.2667406443713171e-05,
+      "loss": 0.7295,
+      "step": 543500
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.2666258513813421e-05,
+      "loss": 0.7616,
+      "step": 543510
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.2665110583913674e-05,
+      "loss": 0.7924,
+      "step": 543520
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.2663962654013926e-05,
+      "loss": 0.8193,
+      "step": 543530
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.2662814724114179e-05,
+      "loss": 0.7618,
+      "step": 543540
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.2661666794214432e-05,
+      "loss": 0.7221,
+      "step": 543550
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.2660518864314684e-05,
+      "loss": 0.7475,
+      "step": 543560
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.2659370934414937e-05,
+      "loss": 0.735,
+      "step": 543570
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.265822300451519e-05,
+      "loss": 0.7507,
+      "step": 543580
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.2657075074615442e-05,
+      "loss": 0.8138,
+      "step": 543590
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.2655927144715695e-05,
+      "loss": 0.8381,
+      "step": 543600
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.2654779214815948e-05,
+      "loss": 0.6915,
+      "step": 543610
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.26536312849162e-05,
+      "loss": 0.6502,
+      "step": 543620
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.2652483355016453e-05,
+      "loss": 0.7776,
+      "step": 543630
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.2651335425116706e-05,
+      "loss": 0.7242,
+      "step": 543640
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.2650187495216956e-05,
+      "loss": 0.7218,
+      "step": 543650
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.264903956531721e-05,
+      "loss": 0.7073,
+      "step": 543660
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.2647891635417463e-05,
+      "loss": 0.744,
+      "step": 543670
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 1.2646743705517714e-05,
+      "loss": 0.7707,
+      "step": 543680
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.2645595775617967e-05,
+      "loss": 0.809,
+      "step": 543690
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.264444784571822e-05,
+      "loss": 0.8135,
+      "step": 543700
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.2643299915818472e-05,
+      "loss": 0.7374,
+      "step": 543710
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.2642151985918725e-05,
+      "loss": 0.7139,
+      "step": 543720
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.2641004056018979e-05,
+      "loss": 0.7076,
+      "step": 543730
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.263985612611923e-05,
+      "loss": 0.7773,
+      "step": 543740
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.2638708196219483e-05,
+      "loss": 0.7428,
+      "step": 543750
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.2637560266319737e-05,
+      "loss": 0.6947,
+      "step": 543760
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.2636412336419988e-05,
+      "loss": 0.8077,
+      "step": 543770
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.2635264406520242e-05,
+      "loss": 0.8124,
+      "step": 543780
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.2634116476620495e-05,
+      "loss": 0.6812,
+      "step": 543790
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.2632968546720745e-05,
+      "loss": 0.7718,
+      "step": 543800
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.2631820616820998e-05,
+      "loss": 0.7905,
+      "step": 543810
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.2630672686921251e-05,
+      "loss": 0.7994,
+      "step": 543820
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.2629524757021503e-05,
+      "loss": 0.7402,
+      "step": 543830
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.2628376827121756e-05,
+      "loss": 0.7852,
+      "step": 543840
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.2627228897222009e-05,
+      "loss": 0.7889,
+      "step": 543850
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.262608096732226e-05,
+      "loss": 0.7376,
+      "step": 543860
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.2624933037422514e-05,
+      "loss": 0.6951,
+      "step": 543870
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.2623785107522765e-05,
+      "loss": 0.7239,
+      "step": 543880
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.2622637177623019e-05,
+      "loss": 0.7668,
+      "step": 543890
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 1.2621489247723272e-05,
+      "loss": 0.8064,
+      "step": 543900
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 1.2620341317823524e-05,
+      "loss": 0.7717,
+      "step": 543910
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 1.2619193387923777e-05,
+      "loss": 0.8133,
+      "step": 543920
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 1.261804545802403e-05,
+      "loss": 0.6639,
+      "step": 543930
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 1.261689752812428e-05,
+      "loss": 0.8293,
+      "step": 543940
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 1.2615749598224533e-05,
+      "loss": 0.6711,
+      "step": 543950
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 1.2614601668324786e-05,
+      "loss": 0.7486,
+      "step": 543960
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 1.2613453738425038e-05,
+      "loss": 0.851,
+      "step": 543970
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 1.2612305808525291e-05,
+      "loss": 0.6962,
+      "step": 543980
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 1.2611157878625544e-05,
+      "loss": 0.7248,
+      "step": 543990
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 1.2610009948725796e-05,
+      "loss": 0.7508,
+      "step": 544000
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 1.2608862018826049e-05,
+      "loss": 0.8379,
+      "step": 544010
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 1.2607714088926302e-05,
+      "loss": 0.7855,
+      "step": 544020
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 1.2606566159026554e-05,
+      "loss": 0.6899,
+      "step": 544030
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 1.2605418229126807e-05,
+      "loss": 0.6935,
+      "step": 544040
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 1.260427029922706e-05,
+      "loss": 0.7818,
+      "step": 544050
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 1.2603122369327312e-05,
+      "loss": 0.6878,
+      "step": 544060
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 1.2601974439427565e-05,
+      "loss": 0.7456,
+      "step": 544070
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 1.2600826509527818e-05,
+      "loss": 0.7565,
+      "step": 544080
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 1.2599678579628068e-05,
+      "loss": 0.7711,
+      "step": 544090
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 1.2598530649728322e-05,
+      "loss": 0.7405,
+      "step": 544100
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 1.2597382719828575e-05,
+      "loss": 0.8398,
+      "step": 544110
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 1.2596234789928826e-05,
+      "loss": 0.7103,
+      "step": 544120
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 1.259508686002908e-05,
+      "loss": 0.676,
+      "step": 544130
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 1.2593938930129333e-05,
+      "loss": 0.7689,
+      "step": 544140
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 1.2592791000229584e-05,
+      "loss": 0.7917,
+      "step": 544150
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 1.2591643070329838e-05,
+      "loss": 0.718,
+      "step": 544160
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 1.2590495140430091e-05,
+      "loss": 0.7552,
+      "step": 544170
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 1.2589347210530342e-05,
+      "loss": 0.7206,
+      "step": 544180
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 1.2588199280630596e-05,
+      "loss": 0.6776,
+      "step": 544190
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 1.2587051350730849e-05,
+      "loss": 0.7249,
+      "step": 544200
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 1.25859034208311e-05,
+      "loss": 0.721,
+      "step": 544210
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 1.2584755490931354e-05,
+      "loss": 0.7583,
+      "step": 544220
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 1.2583607561031604e-05,
+      "loss": 0.7718,
+      "step": 544230
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 1.2582459631131857e-05,
+      "loss": 0.7437,
+      "step": 544240
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 1.258131170123211e-05,
+      "loss": 0.7444,
+      "step": 544250
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 1.2580163771332362e-05,
+      "loss": 0.6636,
+      "step": 544260
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 1.2579015841432615e-05,
+      "loss": 0.7207,
+      "step": 544270
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 1.2577867911532868e-05,
+      "loss": 0.7549,
+      "step": 544280
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 1.257671998163312e-05,
+      "loss": 0.7292,
+      "step": 544290
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 1.2575572051733373e-05,
+      "loss": 0.7321,
+      "step": 544300
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 1.2574424121833626e-05,
+      "loss": 0.6909,
+      "step": 544310
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 1.2573276191933878e-05,
+      "loss": 0.7052,
+      "step": 544320
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 1.2572128262034131e-05,
+      "loss": 0.7561,
+      "step": 544330
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 1.2570980332134384e-05,
+      "loss": 0.6426,
+      "step": 544340
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 1.2569832402234636e-05,
+      "loss": 0.7155,
+      "step": 544350
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 1.2568684472334889e-05,
+      "loss": 0.6471,
+      "step": 544360
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 1.2567536542435142e-05,
+      "loss": 0.795,
+      "step": 544370
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 1.2566388612535392e-05,
+      "loss": 0.7831,
+      "step": 544380
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 1.2565240682635645e-05,
+      "loss": 0.8091,
+      "step": 544390
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 1.2564092752735898e-05,
+      "loss": 0.7192,
+      "step": 544400
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 1.256294482283615e-05,
+      "loss": 0.6179,
+      "step": 544410
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 1.2561796892936403e-05,
+      "loss": 0.7658,
+      "step": 544420
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 1.2560648963036657e-05,
+      "loss": 0.7554,
+      "step": 544430
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 1.2559501033136908e-05,
+      "loss": 0.7829,
+      "step": 544440
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 1.2558353103237161e-05,
+      "loss": 0.7289,
+      "step": 544450
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 1.2557205173337415e-05,
+      "loss": 0.6781,
+      "step": 544460
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 1.2556057243437666e-05,
+      "loss": 0.7612,
+      "step": 544470
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 1.255490931353792e-05,
+      "loss": 0.7638,
+      "step": 544480
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 1.2553761383638173e-05,
+      "loss": 0.8055,
+      "step": 544490
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 1.2552613453738424e-05,
+      "loss": 0.8317,
+      "step": 544500
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 1.2551465523838677e-05,
+      "loss": 0.7551,
+      "step": 544510
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 1.255031759393893e-05,
+      "loss": 0.6648,
+      "step": 544520
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 1.254916966403918e-05,
+      "loss": 0.7573,
+      "step": 544530
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 1.2548021734139434e-05,
+      "loss": 0.7487,
+      "step": 544540
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 1.2546873804239687e-05,
+      "loss": 0.7219,
+      "step": 544550
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 1.2545725874339939e-05,
+      "loss": 0.7869,
+      "step": 544560
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 1.2544577944440192e-05,
+      "loss": 0.7613,
+      "step": 544570
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 1.2543430014540443e-05,
+      "loss": 0.7866,
+      "step": 544580
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 1.2542282084640697e-05,
+      "loss": 0.6841,
+      "step": 544590
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 1.254113415474095e-05,
+      "loss": 0.6871,
+      "step": 544600
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 1.2539986224841201e-05,
+      "loss": 0.7513,
+      "step": 544610
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 1.2538838294941455e-05,
+      "loss": 0.7261,
+      "step": 544620
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 1.2537690365041708e-05,
+      "loss": 0.759,
+      "step": 544630
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 1.253654243514196e-05,
+      "loss": 0.7158,
+      "step": 544640
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 1.2535394505242213e-05,
+      "loss": 0.6545,
+      "step": 544650
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 1.2534246575342466e-05,
+      "loss": 0.7871,
+      "step": 544660
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 1.2533098645442716e-05,
+      "loss": 0.7265,
+      "step": 544670
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 1.2531950715542969e-05,
+      "loss": 0.8085,
+      "step": 544680
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 1.2530802785643222e-05,
+      "loss": 0.7525,
+      "step": 544690
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 1.2529654855743474e-05,
+      "loss": 0.7644,
+      "step": 544700
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 1.2528506925843727e-05,
+      "loss": 0.7643,
+      "step": 544710
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 1.252735899594398e-05,
+      "loss": 0.7032,
+      "step": 544720
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 1.2526211066044232e-05,
+      "loss": 0.6948,
+      "step": 544730
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 1.2525063136144485e-05,
+      "loss": 0.6968,
+      "step": 544740
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 1.2523915206244738e-05,
+      "loss": 0.7236,
+      "step": 544750
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 1.252276727634499e-05,
+      "loss": 0.6955,
+      "step": 544760
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 1.2521619346445243e-05,
+      "loss": 0.8069,
+      "step": 544770
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 1.2520471416545496e-05,
+      "loss": 0.6862,
+      "step": 544780
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 1.2519323486645748e-05,
+      "loss": 0.6898,
+      "step": 544790
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 1.2518175556746001e-05,
+      "loss": 0.7203,
+      "step": 544800
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 1.2517027626846254e-05,
+      "loss": 0.741,
+      "step": 544810
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 1.2515879696946504e-05,
+      "loss": 0.8504,
+      "step": 544820
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 1.2514731767046757e-05,
+      "loss": 0.7667,
+      "step": 544830
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 1.251358383714701e-05,
+      "loss": 0.8155,
+      "step": 544840
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 1.2512435907247262e-05,
+      "loss": 0.6997,
+      "step": 544850
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 1.2511287977347515e-05,
+      "loss": 0.8112,
+      "step": 544860
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 1.2510140047447769e-05,
+      "loss": 0.8093,
+      "step": 544870
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 1.250899211754802e-05,
+      "loss": 0.7203,
+      "step": 544880
+    },
+    {
+      "epoch": 25.0,
+      "eval_cer": 0.6687594920082421,
+      "eval_em": 0.007357212003872217,
+      "eval_f1": 0.007357212003872217,
+      "eval_loss": 1.3492690324783325,
+      "eval_runtime": 1398.549,
+      "eval_samples_per_second": 7.386,
+      "eval_steps_per_second": 0.924,
+      "eval_wer": 0.9926427879961278,
+      "step": 544882
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 1.2507844187648273e-05,
+      "loss": 0.726,
+      "step": 544890
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 1.2506696257748527e-05,
+      "loss": 0.772,
+      "step": 544900
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 1.2505548327848778e-05,
+      "loss": 0.6536,
+      "step": 544910
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 1.2504400397949032e-05,
+      "loss": 0.6948,
+      "step": 544920
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 1.2503252468049283e-05,
+      "loss": 0.7236,
+      "step": 544930
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 1.2502104538149536e-05,
+      "loss": 0.7086,
+      "step": 544940
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 1.250095660824979e-05,
+      "loss": 0.7267,
+      "step": 544950
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 1.249980867835004e-05,
+      "loss": 0.7305,
+      "step": 544960
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 1.2498660748450293e-05,
+      "loss": 0.7089,
+      "step": 544970
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 1.2497512818550546e-05,
+      "loss": 0.6505,
+      "step": 544980
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 1.2496364888650797e-05,
+      "loss": 0.6555,
+      "step": 544990
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 1.249521695875105e-05,
+      "loss": 0.5903,
+      "step": 545000
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 1.2494069028851304e-05,
+      "loss": 0.6605,
+      "step": 545010
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 1.2492921098951555e-05,
+      "loss": 0.6285,
+      "step": 545020
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 1.2491773169051809e-05,
+      "loss": 0.7428,
+      "step": 545030
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 1.2490625239152062e-05,
+      "loss": 0.7114,
+      "step": 545040
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 1.2489477309252314e-05,
+      "loss": 0.7796,
+      "step": 545050
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 1.2488329379352567e-05,
+      "loss": 0.6511,
+      "step": 545060
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 1.248718144945282e-05,
+      "loss": 0.6792,
+      "step": 545070
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 1.2486033519553072e-05,
+      "loss": 0.6794,
+      "step": 545080
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 1.2484885589653325e-05,
+      "loss": 0.6964,
+      "step": 545090
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 1.2483737659753578e-05,
+      "loss": 0.6938,
+      "step": 545100
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 1.2482589729853828e-05,
+      "loss": 0.7842,
+      "step": 545110
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 1.2481441799954081e-05,
+      "loss": 0.7617,
+      "step": 545120
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 1.2480293870054334e-05,
+      "loss": 0.6796,
+      "step": 545130
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 1.2479145940154586e-05,
+      "loss": 0.733,
+      "step": 545140
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 1.247799801025484e-05,
+      "loss": 0.6943,
+      "step": 545150
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 1.2476850080355092e-05,
+      "loss": 0.6204,
+      "step": 545160
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 1.2475702150455344e-05,
+      "loss": 0.7598,
+      "step": 545170
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 1.2474554220555597e-05,
+      "loss": 0.6112,
+      "step": 545180
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 1.247340629065585e-05,
+      "loss": 0.7188,
+      "step": 545190
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 1.2472258360756102e-05,
+      "loss": 0.772,
+      "step": 545200
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 1.2471110430856355e-05,
+      "loss": 0.7937,
+      "step": 545210
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 1.2469962500956608e-05,
+      "loss": 0.6852,
+      "step": 545220
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 1.246881457105686e-05,
+      "loss": 0.7054,
+      "step": 545230
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 1.2467666641157113e-05,
+      "loss": 0.7408,
+      "step": 545240
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 1.2466518711257363e-05,
+      "loss": 0.7198,
+      "step": 545250
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 1.2465370781357616e-05,
+      "loss": 0.6663,
+      "step": 545260
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 1.246422285145787e-05,
+      "loss": 0.6676,
+      "step": 545270
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 1.2463074921558121e-05,
+      "loss": 0.6938,
+      "step": 545280
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 1.2461926991658374e-05,
+      "loss": 0.6892,
+      "step": 545290
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 1.2460779061758628e-05,
+      "loss": 0.7477,
+      "step": 545300
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 1.245963113185888e-05,
+      "loss": 0.7284,
+      "step": 545310
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 1.2458483201959132e-05,
+      "loss": 0.7199,
+      "step": 545320
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 1.2457335272059386e-05,
+      "loss": 0.7019,
+      "step": 545330
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 1.2456187342159637e-05,
+      "loss": 0.7301,
+      "step": 545340
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 1.245503941225989e-05,
+      "loss": 0.7216,
+      "step": 545350
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 1.2453891482360144e-05,
+      "loss": 0.7115,
+      "step": 545360
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 1.2452743552460395e-05,
+      "loss": 0.7716,
+      "step": 545370
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 1.2451595622560648e-05,
+      "loss": 0.7246,
+      "step": 545380
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 1.2450447692660902e-05,
+      "loss": 0.6799,
+      "step": 545390
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 1.2449299762761152e-05,
+      "loss": 0.6833,
+      "step": 545400
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 1.2448151832861405e-05,
+      "loss": 0.6595,
+      "step": 545410
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 1.2447003902961658e-05,
+      "loss": 0.6706,
+      "step": 545420
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 1.244585597306191e-05,
+      "loss": 0.6912,
+      "step": 545430
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 1.2444708043162163e-05,
+      "loss": 0.752,
+      "step": 545440
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 1.2443560113262416e-05,
+      "loss": 0.6981,
+      "step": 545450
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 1.2442412183362668e-05,
+      "loss": 0.706,
+      "step": 545460
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 1.2441264253462921e-05,
+      "loss": 0.7258,
+      "step": 545470
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 1.2440116323563174e-05,
+      "loss": 0.7144,
+      "step": 545480
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 1.2438968393663426e-05,
+      "loss": 0.6494,
+      "step": 545490
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 1.2437820463763679e-05,
+      "loss": 0.7273,
+      "step": 545500
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 1.2436672533863932e-05,
+      "loss": 0.7581,
+      "step": 545510
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 1.2435524603964184e-05,
+      "loss": 0.6304,
+      "step": 545520
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 1.2434376674064437e-05,
+      "loss": 0.6976,
+      "step": 545530
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 1.243322874416469e-05,
+      "loss": 0.7745,
+      "step": 545540
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 1.243208081426494e-05,
+      "loss": 0.7133,
+      "step": 545550
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 1.2430932884365193e-05,
+      "loss": 0.6371,
+      "step": 545560
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 1.2429784954465447e-05,
+      "loss": 0.7106,
+      "step": 545570
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 1.2428637024565698e-05,
+      "loss": 0.762,
+      "step": 545580
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 1.2427489094665951e-05,
+      "loss": 0.6997,
+      "step": 545590
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 1.2426341164766203e-05,
+      "loss": 0.7215,
+      "step": 545600
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 1.2425193234866456e-05,
+      "loss": 0.6595,
+      "step": 545610
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 1.242404530496671e-05,
+      "loss": 0.735,
+      "step": 545620
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 1.2422897375066961e-05,
+      "loss": 0.6798,
+      "step": 545630
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 1.2421749445167214e-05,
+      "loss": 0.8091,
+      "step": 545640
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 1.2420601515267467e-05,
+      "loss": 0.6635,
+      "step": 545650
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 1.2419453585367719e-05,
+      "loss": 0.7103,
+      "step": 545660
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 1.2418305655467972e-05,
+      "loss": 0.7492,
+      "step": 545670
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 1.2417157725568225e-05,
+      "loss": 0.7283,
+      "step": 545680
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 1.2416009795668475e-05,
+      "loss": 0.6983,
+      "step": 545690
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 1.2414861865768729e-05,
+      "loss": 0.733,
+      "step": 545700
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 1.2413713935868982e-05,
+      "loss": 0.7131,
+      "step": 545710
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 1.2412566005969233e-05,
+      "loss": 0.6926,
+      "step": 545720
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 1.2411418076069487e-05,
+      "loss": 0.6588,
+      "step": 545730
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 1.241027014616974e-05,
+      "loss": 0.6749,
+      "step": 545740
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 1.2409122216269991e-05,
+      "loss": 0.768,
+      "step": 545750
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 1.2407974286370245e-05,
+      "loss": 0.772,
+      "step": 545760
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 1.2406826356470498e-05,
+      "loss": 0.7599,
+      "step": 545770
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 1.240567842657075e-05,
+      "loss": 0.7307,
+      "step": 545780
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 1.2404530496671003e-05,
+      "loss": 0.7131,
+      "step": 545790
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 1.2403382566771256e-05,
+      "loss": 0.7439,
+      "step": 545800
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 1.2402234636871507e-05,
+      "loss": 0.7417,
+      "step": 545810
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 1.240108670697176e-05,
+      "loss": 0.7695,
+      "step": 545820
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 1.2399938777072014e-05,
+      "loss": 0.7995,
+      "step": 545830
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 1.2398790847172264e-05,
+      "loss": 0.7487,
+      "step": 545840
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 1.2397642917272517e-05,
+      "loss": 0.6414,
+      "step": 545850
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 1.239649498737277e-05,
+      "loss": 0.6761,
+      "step": 545860
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 1.2395347057473022e-05,
+      "loss": 0.7053,
+      "step": 545870
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 1.2394199127573275e-05,
+      "loss": 0.6951,
+      "step": 545880
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 1.2393051197673528e-05,
+      "loss": 0.7447,
+      "step": 545890
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 1.239190326777378e-05,
+      "loss": 0.7378,
+      "step": 545900
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 1.2390755337874033e-05,
+      "loss": 0.7214,
+      "step": 545910
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 1.2389607407974286e-05,
+      "loss": 0.7016,
+      "step": 545920
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 1.2388459478074538e-05,
+      "loss": 0.7461,
+      "step": 545930
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 1.2387311548174791e-05,
+      "loss": 0.728,
+      "step": 545940
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 1.2386163618275043e-05,
+      "loss": 0.7168,
+      "step": 545950
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 1.2385015688375296e-05,
+      "loss": 0.7654,
+      "step": 545960
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 1.2383867758475549e-05,
+      "loss": 0.7255,
+      "step": 545970
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 1.2382719828575799e-05,
+      "loss": 0.7124,
+      "step": 545980
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 1.2381571898676052e-05,
+      "loss": 0.7204,
+      "step": 545990
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 1.2380423968776305e-05,
+      "loss": 0.684,
+      "step": 546000
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 1.2379276038876557e-05,
+      "loss": 0.6651,
+      "step": 546010
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 1.237812810897681e-05,
+      "loss": 0.723,
+      "step": 546020
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 1.2376980179077064e-05,
+      "loss": 0.7875,
+      "step": 546030
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 1.2375832249177315e-05,
+      "loss": 0.7564,
+      "step": 546040
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 1.2374684319277568e-05,
+      "loss": 0.7294,
+      "step": 546050
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 1.2373536389377822e-05,
+      "loss": 0.7177,
+      "step": 546060
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 1.2372388459478073e-05,
+      "loss": 0.7283,
+      "step": 546070
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 1.2371240529578326e-05,
+      "loss": 0.7529,
+      "step": 546080
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 1.237009259967858e-05,
+      "loss": 0.7112,
+      "step": 546090
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 1.2368944669778831e-05,
+      "loss": 0.6505,
+      "step": 546100
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 1.2367796739879084e-05,
+      "loss": 0.6812,
+      "step": 546110
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 1.2366648809979338e-05,
+      "loss": 0.7375,
+      "step": 546120
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 1.2365500880079587e-05,
+      "loss": 0.7207,
+      "step": 546130
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 1.236435295017984e-05,
+      "loss": 0.6788,
+      "step": 546140
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 1.2363205020280094e-05,
+      "loss": 0.7021,
+      "step": 546150
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 1.2362057090380345e-05,
+      "loss": 0.7607,
+      "step": 546160
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 1.2360909160480599e-05,
+      "loss": 0.7566,
+      "step": 546170
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 1.2359761230580852e-05,
+      "loss": 0.8077,
+      "step": 546180
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 1.2358613300681104e-05,
+      "loss": 0.7221,
+      "step": 546190
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 1.2357465370781357e-05,
+      "loss": 0.755,
+      "step": 546200
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 1.235631744088161e-05,
+      "loss": 0.7538,
+      "step": 546210
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 1.2355169510981862e-05,
+      "loss": 0.6837,
+      "step": 546220
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 1.2354021581082115e-05,
+      "loss": 0.6485,
+      "step": 546230
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 1.2352873651182368e-05,
+      "loss": 0.695,
+      "step": 546240
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 1.235172572128262e-05,
+      "loss": 0.6324,
+      "step": 546250
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 1.2350577791382873e-05,
+      "loss": 0.771,
+      "step": 546260
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 1.2349429861483126e-05,
+      "loss": 0.6829,
+      "step": 546270
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 1.2348281931583376e-05,
+      "loss": 0.7549,
+      "step": 546280
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 1.234713400168363e-05,
+      "loss": 0.7182,
+      "step": 546290
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 1.234598607178388e-05,
+      "loss": 0.6725,
+      "step": 546300
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 1.2344838141884134e-05,
+      "loss": 0.7181,
+      "step": 546310
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 1.2343690211984387e-05,
+      "loss": 0.766,
+      "step": 546320
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 1.2342542282084639e-05,
+      "loss": 0.7367,
+      "step": 546330
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 1.2341394352184892e-05,
+      "loss": 0.7219,
+      "step": 546340
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 1.2340246422285145e-05,
+      "loss": 0.8204,
+      "step": 546350
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 1.2339098492385397e-05,
+      "loss": 0.6709,
+      "step": 546360
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 1.233795056248565e-05,
+      "loss": 0.7188,
+      "step": 546370
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 1.2336802632585903e-05,
+      "loss": 0.7033,
+      "step": 546380
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 1.2335654702686155e-05,
+      "loss": 0.7456,
+      "step": 546390
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 1.2334506772786408e-05,
+      "loss": 0.7186,
+      "step": 546400
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 1.2333358842886661e-05,
+      "loss": 0.7366,
+      "step": 546410
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 1.2332210912986911e-05,
+      "loss": 0.7342,
+      "step": 546420
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 1.2331062983087164e-05,
+      "loss": 0.6928,
+      "step": 546430
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 1.2329915053187418e-05,
+      "loss": 0.7119,
+      "step": 546440
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 1.232876712328767e-05,
+      "loss": 0.7047,
+      "step": 546450
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 1.2327619193387922e-05,
+      "loss": 0.7283,
+      "step": 546460
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 1.2326471263488176e-05,
+      "loss": 0.6635,
+      "step": 546470
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 1.2325323333588427e-05,
+      "loss": 0.6795,
+      "step": 546480
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 1.232417540368868e-05,
+      "loss": 0.7819,
+      "step": 546490
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 1.2323027473788934e-05,
+      "loss": 0.7486,
+      "step": 546500
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 1.2321879543889185e-05,
+      "loss": 0.7407,
+      "step": 546510
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 1.2320731613989439e-05,
+      "loss": 0.6912,
+      "step": 546520
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 1.2319583684089692e-05,
+      "loss": 0.6767,
+      "step": 546530
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 1.2318435754189943e-05,
+      "loss": 0.7004,
+      "step": 546540
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 1.2317287824290197e-05,
+      "loss": 0.677,
+      "step": 546550
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 1.231613989439045e-05,
+      "loss": 0.7231,
+      "step": 546560
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 1.23149919644907e-05,
+      "loss": 0.705,
+      "step": 546570
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 1.2313844034590953e-05,
+      "loss": 0.7171,
+      "step": 546580
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 1.2312696104691206e-05,
+      "loss": 0.729,
+      "step": 546590
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 1.2311548174791458e-05,
+      "loss": 0.8079,
+      "step": 546600
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 1.2310400244891711e-05,
+      "loss": 0.7424,
+      "step": 546610
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 1.2309252314991962e-05,
+      "loss": 0.7513,
+      "step": 546620
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 1.2308104385092216e-05,
+      "loss": 0.6687,
+      "step": 546630
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 1.2306956455192469e-05,
+      "loss": 0.749,
+      "step": 546640
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 1.230580852529272e-05,
+      "loss": 0.7009,
+      "step": 546650
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 1.2304660595392974e-05,
+      "loss": 0.6868,
+      "step": 546660
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 1.2303512665493227e-05,
+      "loss": 0.723,
+      "step": 546670
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 1.2302364735593479e-05,
+      "loss": 0.6878,
+      "step": 546680
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 1.2301216805693732e-05,
+      "loss": 0.7577,
+      "step": 546690
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 1.2300068875793985e-05,
+      "loss": 0.7426,
+      "step": 546700
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 1.2298920945894235e-05,
+      "loss": 0.7619,
+      "step": 546710
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 1.2297773015994488e-05,
+      "loss": 0.7455,
+      "step": 546720
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 1.2296625086094741e-05,
+      "loss": 0.6555,
+      "step": 546730
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 1.2295477156194993e-05,
+      "loss": 0.7174,
+      "step": 546740
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 1.2294329226295246e-05,
+      "loss": 0.7225,
+      "step": 546750
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 1.22931812963955e-05,
+      "loss": 0.6575,
+      "step": 546760
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 1.2292033366495751e-05,
+      "loss": 0.7421,
+      "step": 546770
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 1.2290885436596004e-05,
+      "loss": 0.8025,
+      "step": 546780
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 1.2289737506696257e-05,
+      "loss": 0.7655,
+      "step": 546790
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 1.2288589576796509e-05,
+      "loss": 0.825,
+      "step": 546800
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 1.2287441646896762e-05,
+      "loss": 0.6769,
+      "step": 546810
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 1.2286293716997015e-05,
+      "loss": 0.7443,
+      "step": 546820
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 1.2285145787097267e-05,
+      "loss": 0.7383,
+      "step": 546830
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 1.228399785719752e-05,
+      "loss": 0.7355,
+      "step": 546840
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 1.2282849927297773e-05,
+      "loss": 0.6568,
+      "step": 546850
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 1.2281701997398023e-05,
+      "loss": 0.7609,
+      "step": 546860
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 1.2280554067498277e-05,
+      "loss": 0.6192,
+      "step": 546870
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 1.227940613759853e-05,
+      "loss": 0.7291,
+      "step": 546880
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 1.2278258207698781e-05,
+      "loss": 0.6796,
+      "step": 546890
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 1.2277110277799035e-05,
+      "loss": 0.6961,
+      "step": 546900
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 1.2275962347899288e-05,
+      "loss": 0.75,
+      "step": 546910
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 1.227481441799954e-05,
+      "loss": 0.6728,
+      "step": 546920
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 1.2273666488099793e-05,
+      "loss": 0.7168,
+      "step": 546930
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 1.2272518558200046e-05,
+      "loss": 0.6451,
+      "step": 546940
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 1.2271370628300297e-05,
+      "loss": 0.7048,
+      "step": 546950
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 1.227022269840055e-05,
+      "loss": 0.7747,
+      "step": 546960
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 1.2269074768500802e-05,
+      "loss": 0.6844,
+      "step": 546970
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 1.2267926838601055e-05,
+      "loss": 0.7102,
+      "step": 546980
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 1.2266778908701309e-05,
+      "loss": 0.7422,
+      "step": 546990
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 1.2265630978801559e-05,
+      "loss": 0.6653,
+      "step": 547000
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 1.2264483048901812e-05,
+      "loss": 0.7807,
+      "step": 547010
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 1.2263335119002065e-05,
+      "loss": 0.747,
+      "step": 547020
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 1.2262187189102317e-05,
+      "loss": 0.7279,
+      "step": 547030
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 1.226103925920257e-05,
+      "loss": 0.6767,
+      "step": 547040
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 1.2259891329302823e-05,
+      "loss": 0.7857,
+      "step": 547050
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 1.2258743399403075e-05,
+      "loss": 0.7321,
+      "step": 547060
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 1.2257595469503328e-05,
+      "loss": 0.6357,
+      "step": 547070
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 1.2256447539603581e-05,
+      "loss": 0.6664,
+      "step": 547080
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 1.2255299609703833e-05,
+      "loss": 0.8388,
+      "step": 547090
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 1.2254151679804086e-05,
+      "loss": 0.6868,
+      "step": 547100
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 1.225300374990434e-05,
+      "loss": 0.7736,
+      "step": 547110
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 1.225185582000459e-05,
+      "loss": 0.7464,
+      "step": 547120
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 1.2250707890104844e-05,
+      "loss": 0.7209,
+      "step": 547130
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 1.2249559960205097e-05,
+      "loss": 0.7348,
+      "step": 547140
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 1.2248412030305347e-05,
+      "loss": 0.7499,
+      "step": 547150
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 1.22472641004056e-05,
+      "loss": 0.7791,
+      "step": 547160
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 1.2246116170505854e-05,
+      "loss": 0.685,
+      "step": 547170
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 1.2244968240606105e-05,
+      "loss": 0.74,
+      "step": 547180
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 1.2243820310706358e-05,
+      "loss": 0.7245,
+      "step": 547190
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 1.2242672380806612e-05,
+      "loss": 0.6835,
+      "step": 547200
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 1.2241524450906863e-05,
+      "loss": 0.7047,
+      "step": 547210
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 1.2240376521007116e-05,
+      "loss": 0.7213,
+      "step": 547220
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 1.223922859110737e-05,
+      "loss": 0.7165,
+      "step": 547230
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 1.2238080661207621e-05,
+      "loss": 0.8203,
+      "step": 547240
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 1.2236932731307874e-05,
+      "loss": 0.7269,
+      "step": 547250
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 1.2235784801408128e-05,
+      "loss": 0.7043,
+      "step": 547260
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 1.223463687150838e-05,
+      "loss": 0.6939,
+      "step": 547270
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 1.2233488941608632e-05,
+      "loss": 0.7227,
+      "step": 547280
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 1.2232341011708886e-05,
+      "loss": 0.7478,
+      "step": 547290
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 1.2231193081809136e-05,
+      "loss": 0.7107,
+      "step": 547300
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 1.2230045151909389e-05,
+      "loss": 0.7313,
+      "step": 547310
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 1.222889722200964e-05,
+      "loss": 0.6708,
+      "step": 547320
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 1.2227749292109894e-05,
+      "loss": 0.7138,
+      "step": 547330
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 1.2226601362210147e-05,
+      "loss": 0.7245,
+      "step": 547340
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 1.2225453432310398e-05,
+      "loss": 0.6684,
+      "step": 547350
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 1.2224305502410652e-05,
+      "loss": 0.7084,
+      "step": 547360
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 1.2223157572510905e-05,
+      "loss": 0.741,
+      "step": 547370
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 1.2222009642611156e-05,
+      "loss": 0.7265,
+      "step": 547380
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 1.222086171271141e-05,
+      "loss": 0.7717,
+      "step": 547390
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 1.2219713782811663e-05,
+      "loss": 0.7233,
+      "step": 547400
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 1.2218565852911914e-05,
+      "loss": 0.6442,
+      "step": 547410
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 1.2217417923012168e-05,
+      "loss": 0.695,
+      "step": 547420
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 1.2216269993112421e-05,
+      "loss": 0.7483,
+      "step": 547430
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 1.221512206321267e-05,
+      "loss": 0.7129,
+      "step": 547440
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 1.2213974133312924e-05,
+      "loss": 0.7159,
+      "step": 547450
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 1.2212826203413177e-05,
+      "loss": 0.7175,
+      "step": 547460
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 1.2211678273513429e-05,
+      "loss": 0.7616,
+      "step": 547470
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 1.2210530343613682e-05,
+      "loss": 0.7951,
+      "step": 547480
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 1.2209382413713935e-05,
+      "loss": 0.7484,
+      "step": 547490
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 1.2208234483814187e-05,
+      "loss": 0.6726,
+      "step": 547500
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 1.220708655391444e-05,
+      "loss": 0.707,
+      "step": 547510
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 1.2205938624014693e-05,
+      "loss": 0.7492,
+      "step": 547520
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 1.2204790694114945e-05,
+      "loss": 0.6829,
+      "step": 547530
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 1.2203642764215198e-05,
+      "loss": 0.6926,
+      "step": 547540
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 1.2202494834315451e-05,
+      "loss": 0.81,
+      "step": 547550
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 1.2201346904415703e-05,
+      "loss": 0.7476,
+      "step": 547560
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 1.2200198974515956e-05,
+      "loss": 0.7783,
+      "step": 547570
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 1.219905104461621e-05,
+      "loss": 0.779,
+      "step": 547580
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 1.219790311471646e-05,
+      "loss": 0.6807,
+      "step": 547590
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 1.2196755184816712e-05,
+      "loss": 0.7094,
+      "step": 547600
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 1.2195607254916966e-05,
+      "loss": 0.6712,
+      "step": 547610
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 1.2194459325017217e-05,
+      "loss": 0.7584,
+      "step": 547620
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 1.219331139511747e-05,
+      "loss": 0.7068,
+      "step": 547630
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 1.2192163465217724e-05,
+      "loss": 0.6851,
+      "step": 547640
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 1.2191015535317975e-05,
+      "loss": 0.7313,
+      "step": 547650
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 1.2189867605418229e-05,
+      "loss": 0.7113,
+      "step": 547660
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 1.218871967551848e-05,
+      "loss": 0.7711,
+      "step": 547670
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 1.2187571745618733e-05,
+      "loss": 0.7118,
+      "step": 547680
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 1.2186423815718987e-05,
+      "loss": 0.7658,
+      "step": 547690
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 1.2185275885819238e-05,
+      "loss": 0.7324,
+      "step": 547700
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 1.2184127955919491e-05,
+      "loss": 0.7498,
+      "step": 547710
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 1.2182980026019745e-05,
+      "loss": 0.7477,
+      "step": 547720
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 1.2181832096119994e-05,
+      "loss": 0.6679,
+      "step": 547730
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 1.2180684166220248e-05,
+      "loss": 0.693,
+      "step": 547740
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 1.2179536236320501e-05,
+      "loss": 0.7237,
+      "step": 547750
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 1.2178388306420752e-05,
+      "loss": 0.772,
+      "step": 547760
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 1.2177240376521006e-05,
+      "loss": 0.7519,
+      "step": 547770
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 1.2176092446621259e-05,
+      "loss": 0.7374,
+      "step": 547780
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 1.217494451672151e-05,
+      "loss": 0.7859,
+      "step": 547790
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 1.2173796586821764e-05,
+      "loss": 0.678,
+      "step": 547800
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 1.2172648656922017e-05,
+      "loss": 0.6879,
+      "step": 547810
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 1.2171500727022269e-05,
+      "loss": 0.7034,
+      "step": 547820
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 1.2170352797122522e-05,
+      "loss": 0.8143,
+      "step": 547830
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 1.2169204867222775e-05,
+      "loss": 0.7822,
+      "step": 547840
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 1.2168056937323027e-05,
+      "loss": 0.7735,
+      "step": 547850
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 1.216690900742328e-05,
+      "loss": 0.7103,
+      "step": 547860
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 1.2165761077523533e-05,
+      "loss": 0.7067,
+      "step": 547870
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 1.2164613147623783e-05,
+      "loss": 0.6176,
+      "step": 547880
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 1.2163465217724036e-05,
+      "loss": 0.724,
+      "step": 547890
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 1.216231728782429e-05,
+      "loss": 0.7243,
+      "step": 547900
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 1.2161169357924541e-05,
+      "loss": 0.7036,
+      "step": 547910
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 1.2160021428024794e-05,
+      "loss": 0.6829,
+      "step": 547920
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 1.2158873498125047e-05,
+      "loss": 0.7815,
+      "step": 547930
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 1.2157725568225299e-05,
+      "loss": 0.6573,
+      "step": 547940
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 1.2156577638325552e-05,
+      "loss": 0.7196,
+      "step": 547950
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 1.2155429708425805e-05,
+      "loss": 0.618,
+      "step": 547960
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 1.2154281778526057e-05,
+      "loss": 0.757,
+      "step": 547970
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 1.215313384862631e-05,
+      "loss": 0.692,
+      "step": 547980
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 1.2151985918726562e-05,
+      "loss": 0.7433,
+      "step": 547990
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 1.2150837988826815e-05,
+      "loss": 0.7384,
+      "step": 548000
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 1.2149690058927068e-05,
+      "loss": 0.6775,
+      "step": 548010
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 1.2148542129027318e-05,
+      "loss": 0.7467,
+      "step": 548020
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 1.2147394199127571e-05,
+      "loss": 0.6089,
+      "step": 548030
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 1.2146246269227825e-05,
+      "loss": 0.6891,
+      "step": 548040
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 1.2145098339328076e-05,
+      "loss": 0.6868,
+      "step": 548050
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 1.214395040942833e-05,
+      "loss": 0.7179,
+      "step": 548060
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 1.2142802479528583e-05,
+      "loss": 0.6836,
+      "step": 548070
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 1.2141654549628834e-05,
+      "loss": 0.6957,
+      "step": 548080
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 1.2140506619729087e-05,
+      "loss": 0.6449,
+      "step": 548090
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 1.213935868982934e-05,
+      "loss": 0.6747,
+      "step": 548100
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 1.2138210759929592e-05,
+      "loss": 0.7223,
+      "step": 548110
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 1.2137062830029845e-05,
+      "loss": 0.7107,
+      "step": 548120
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 1.2135914900130099e-05,
+      "loss": 0.6898,
+      "step": 548130
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 1.213476697023035e-05,
+      "loss": 0.6906,
+      "step": 548140
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 1.2133619040330604e-05,
+      "loss": 0.7511,
+      "step": 548150
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 1.2132471110430857e-05,
+      "loss": 0.6837,
+      "step": 548160
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 1.2131323180531107e-05,
+      "loss": 0.7681,
+      "step": 548170
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 1.213017525063136e-05,
+      "loss": 0.6703,
+      "step": 548180
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 1.2129027320731613e-05,
+      "loss": 0.7962,
+      "step": 548190
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 1.2127879390831865e-05,
+      "loss": 0.6793,
+      "step": 548200
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 1.2126731460932118e-05,
+      "loss": 0.7477,
+      "step": 548210
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 1.2125583531032371e-05,
+      "loss": 0.6355,
+      "step": 548220
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 1.2124435601132623e-05,
+      "loss": 0.725,
+      "step": 548230
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 1.2123287671232876e-05,
+      "loss": 0.7725,
+      "step": 548240
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 1.212213974133313e-05,
+      "loss": 0.7455,
+      "step": 548250
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 1.212099181143338e-05,
+      "loss": 0.7951,
+      "step": 548260
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 1.2119843881533634e-05,
+      "loss": 0.7696,
+      "step": 548270
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 1.2118695951633887e-05,
+      "loss": 0.6547,
+      "step": 548280
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 1.2117548021734139e-05,
+      "loss": 0.6822,
+      "step": 548290
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 1.2116400091834392e-05,
+      "loss": 0.7754,
+      "step": 548300
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 1.2115252161934645e-05,
+      "loss": 0.6816,
+      "step": 548310
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 1.2114104232034895e-05,
+      "loss": 0.6724,
+      "step": 548320
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 1.2112956302135148e-05,
+      "loss": 0.684,
+      "step": 548330
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 1.21118083722354e-05,
+      "loss": 0.7345,
+      "step": 548340
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 1.2110660442335653e-05,
+      "loss": 0.7462,
+      "step": 548350
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 1.2109512512435906e-05,
+      "loss": 0.7052,
+      "step": 548360
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 1.2108364582536158e-05,
+      "loss": 0.7015,
+      "step": 548370
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 1.2107216652636411e-05,
+      "loss": 0.7266,
+      "step": 548380
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 1.2106068722736664e-05,
+      "loss": 0.689,
+      "step": 548390
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 1.2104920792836916e-05,
+      "loss": 0.7237,
+      "step": 548400
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 1.210377286293717e-05,
+      "loss": 0.713,
+      "step": 548410
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 1.2102624933037422e-05,
+      "loss": 0.714,
+      "step": 548420
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 1.2101477003137674e-05,
+      "loss": 0.713,
+      "step": 548430
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 1.2100329073237927e-05,
+      "loss": 0.7651,
+      "step": 548440
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 1.209918114333818e-05,
+      "loss": 0.7276,
+      "step": 548450
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 1.209803321343843e-05,
+      "loss": 0.6836,
+      "step": 548460
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 1.2096885283538684e-05,
+      "loss": 0.7297,
+      "step": 548470
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 1.2095737353638937e-05,
+      "loss": 0.6965,
+      "step": 548480
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 1.2094589423739188e-05,
+      "loss": 0.7146,
+      "step": 548490
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 1.2093441493839442e-05,
+      "loss": 0.6856,
+      "step": 548500
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 1.2092293563939695e-05,
+      "loss": 0.6736,
+      "step": 548510
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 1.2091145634039946e-05,
+      "loss": 0.7155,
+      "step": 548520
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 1.20899977041402e-05,
+      "loss": 0.7348,
+      "step": 548530
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 1.2088849774240453e-05,
+      "loss": 0.6959,
+      "step": 548540
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 1.2087701844340704e-05,
+      "loss": 0.7322,
+      "step": 548550
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 1.2086553914440958e-05,
+      "loss": 0.6517,
+      "step": 548560
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 1.2085405984541211e-05,
+      "loss": 0.6682,
+      "step": 548570
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 1.2084258054641462e-05,
+      "loss": 0.6588,
+      "step": 548580
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 1.2083110124741716e-05,
+      "loss": 0.6811,
+      "step": 548590
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 1.2081962194841969e-05,
+      "loss": 0.6718,
+      "step": 548600
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 1.2080814264942219e-05,
+      "loss": 0.7534,
+      "step": 548610
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 1.2079666335042472e-05,
+      "loss": 0.7282,
+      "step": 548620
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 1.2078518405142725e-05,
+      "loss": 0.7119,
+      "step": 548630
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 1.2077370475242977e-05,
+      "loss": 0.7108,
+      "step": 548640
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 1.207622254534323e-05,
+      "loss": 0.7931,
+      "step": 548650
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 1.2075074615443483e-05,
+      "loss": 0.7249,
+      "step": 548660
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 1.2073926685543735e-05,
+      "loss": 0.7183,
+      "step": 548670
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 1.2072778755643988e-05,
+      "loss": 0.6993,
+      "step": 548680
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 1.207163082574424e-05,
+      "loss": 0.778,
+      "step": 548690
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 1.2070482895844493e-05,
+      "loss": 0.6732,
+      "step": 548700
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 1.2069334965944746e-05,
+      "loss": 0.7258,
+      "step": 548710
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 1.2068187036044998e-05,
+      "loss": 0.7881,
+      "step": 548720
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 1.2067039106145251e-05,
+      "loss": 0.7483,
+      "step": 548730
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 1.2065891176245504e-05,
+      "loss": 0.7063,
+      "step": 548740
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 1.2064743246345754e-05,
+      "loss": 0.7468,
+      "step": 548750
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 1.2063595316446007e-05,
+      "loss": 0.7607,
+      "step": 548760
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 1.206244738654626e-05,
+      "loss": 0.6524,
+      "step": 548770
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 1.2061299456646512e-05,
+      "loss": 0.7272,
+      "step": 548780
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 1.2060151526746765e-05,
+      "loss": 0.8053,
+      "step": 548790
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 1.2059003596847019e-05,
+      "loss": 0.6846,
+      "step": 548800
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 1.205785566694727e-05,
+      "loss": 0.6775,
+      "step": 548810
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 1.2056707737047523e-05,
+      "loss": 0.8287,
+      "step": 548820
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 1.2055559807147777e-05,
+      "loss": 0.6309,
+      "step": 548830
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 1.2054411877248028e-05,
+      "loss": 0.6416,
+      "step": 548840
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 1.2053263947348281e-05,
+      "loss": 0.701,
+      "step": 548850
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 1.2052116017448535e-05,
+      "loss": 0.646,
+      "step": 548860
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 1.2050968087548786e-05,
+      "loss": 0.7325,
+      "step": 548870
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 1.204982015764904e-05,
+      "loss": 0.6197,
+      "step": 548880
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 1.2048672227749293e-05,
+      "loss": 0.7533,
+      "step": 548890
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 1.2047524297849542e-05,
+      "loss": 0.6592,
+      "step": 548900
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 1.2046376367949796e-05,
+      "loss": 0.7547,
+      "step": 548910
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 1.2045228438050049e-05,
+      "loss": 0.665,
+      "step": 548920
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 1.20440805081503e-05,
+      "loss": 0.6438,
+      "step": 548930
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 1.2042932578250554e-05,
+      "loss": 0.6676,
+      "step": 548940
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 1.2041784648350807e-05,
+      "loss": 0.7915,
+      "step": 548950
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 1.2040636718451059e-05,
+      "loss": 0.7897,
+      "step": 548960
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 1.2039488788551312e-05,
+      "loss": 0.7199,
+      "step": 548970
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 1.2038340858651565e-05,
+      "loss": 0.6703,
+      "step": 548980
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 1.2037192928751817e-05,
+      "loss": 0.6854,
+      "step": 548990
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 1.203604499885207e-05,
+      "loss": 0.7431,
+      "step": 549000
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 1.2034897068952323e-05,
+      "loss": 0.7759,
+      "step": 549010
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 1.2033749139052575e-05,
+      "loss": 0.734,
+      "step": 549020
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 1.2032601209152828e-05,
+      "loss": 0.7595,
+      "step": 549030
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 1.2031453279253078e-05,
+      "loss": 0.7234,
+      "step": 549040
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 1.2030305349353331e-05,
+      "loss": 0.7635,
+      "step": 549050
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 1.2029157419453584e-05,
+      "loss": 0.7555,
+      "step": 549060
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 1.2028009489553836e-05,
+      "loss": 0.7889,
+      "step": 549070
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 1.2026861559654089e-05,
+      "loss": 0.7066,
+      "step": 549080
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 1.2025713629754342e-05,
+      "loss": 0.8399,
+      "step": 549090
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 1.2024565699854594e-05,
+      "loss": 0.7448,
+      "step": 549100
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 1.2023417769954847e-05,
+      "loss": 0.7405,
+      "step": 549110
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 1.20222698400551e-05,
+      "loss": 0.7691,
+      "step": 549120
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 1.2021121910155352e-05,
+      "loss": 0.7113,
+      "step": 549130
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 1.2019973980255605e-05,
+      "loss": 0.7647,
+      "step": 549140
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 1.2018826050355858e-05,
+      "loss": 0.7022,
+      "step": 549150
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 1.2017678120456108e-05,
+      "loss": 0.6877,
+      "step": 549160
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 1.2016530190556363e-05,
+      "loss": 0.7647,
+      "step": 549170
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 1.2015382260656616e-05,
+      "loss": 0.742,
+      "step": 549180
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 1.2014234330756866e-05,
+      "loss": 0.7803,
+      "step": 549190
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 1.201308640085712e-05,
+      "loss": 0.7005,
+      "step": 549200
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 1.2011938470957373e-05,
+      "loss": 0.7427,
+      "step": 549210
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 1.2010790541057624e-05,
+      "loss": 0.6276,
+      "step": 549220
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 1.2009642611157877e-05,
+      "loss": 0.6729,
+      "step": 549230
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 1.200849468125813e-05,
+      "loss": 0.6409,
+      "step": 549240
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 1.2007346751358382e-05,
+      "loss": 0.7843,
+      "step": 549250
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 1.2006198821458636e-05,
+      "loss": 0.7228,
+      "step": 549260
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 1.2005050891558889e-05,
+      "loss": 0.7434,
+      "step": 549270
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 1.200390296165914e-05,
+      "loss": 0.6823,
+      "step": 549280
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 1.2002755031759394e-05,
+      "loss": 0.7724,
+      "step": 549290
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 1.2001607101859647e-05,
+      "loss": 0.7508,
+      "step": 549300
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 1.2000459171959897e-05,
+      "loss": 0.7705,
+      "step": 549310
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 1.1999311242060152e-05,
+      "loss": 0.7545,
+      "step": 549320
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 1.1998163312160405e-05,
+      "loss": 0.8027,
+      "step": 549330
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 1.1997015382260655e-05,
+      "loss": 0.6946,
+      "step": 549340
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 1.1995867452360908e-05,
+      "loss": 0.6976,
+      "step": 549350
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 1.1994719522461161e-05,
+      "loss": 0.7688,
+      "step": 549360
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 1.1993571592561413e-05,
+      "loss": 0.7495,
+      "step": 549370
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 1.1992423662661666e-05,
+      "loss": 0.6975,
+      "step": 549380
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 1.1991275732761917e-05,
+      "loss": 0.6939,
+      "step": 549390
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 1.199012780286217e-05,
+      "loss": 0.6286,
+      "step": 549400
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 1.1988979872962424e-05,
+      "loss": 0.6538,
+      "step": 549410
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 1.1987831943062676e-05,
+      "loss": 0.7006,
+      "step": 549420
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 1.1986684013162929e-05,
+      "loss": 0.7173,
+      "step": 549430
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 1.1985536083263182e-05,
+      "loss": 0.7543,
+      "step": 549440
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 1.1984388153363432e-05,
+      "loss": 0.71,
+      "step": 549450
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 1.1983240223463685e-05,
+      "loss": 0.7102,
+      "step": 549460
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 1.198209229356394e-05,
+      "loss": 0.6955,
+      "step": 549470
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 1.198094436366419e-05,
+      "loss": 0.7196,
+      "step": 549480
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 1.1979796433764443e-05,
+      "loss": 0.7363,
+      "step": 549490
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 1.1978648503864696e-05,
+      "loss": 0.6766,
+      "step": 549500
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 1.1977500573964948e-05,
+      "loss": 0.765,
+      "step": 549510
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 1.1976352644065201e-05,
+      "loss": 0.7226,
+      "step": 549520
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 1.1975204714165454e-05,
+      "loss": 0.749,
+      "step": 549530
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 1.1974056784265706e-05,
+      "loss": 0.6807,
+      "step": 549540
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 1.197290885436596e-05,
+      "loss": 0.7674,
+      "step": 549550
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 1.1971760924466212e-05,
+      "loss": 0.653,
+      "step": 549560
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 1.1970612994566464e-05,
+      "loss": 0.6729,
+      "step": 549570
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 1.1969465064666717e-05,
+      "loss": 0.7239,
+      "step": 549580
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 1.196831713476697e-05,
+      "loss": 0.7176,
+      "step": 549590
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 1.196716920486722e-05,
+      "loss": 0.7556,
+      "step": 549600
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 1.1966021274967474e-05,
+      "loss": 0.7547,
+      "step": 549610
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 1.1964873345067729e-05,
+      "loss": 0.7434,
+      "step": 549620
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 1.1963725415167978e-05,
+      "loss": 0.7397,
+      "step": 549630
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 1.1962577485268232e-05,
+      "loss": 0.6863,
+      "step": 549640
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 1.1961429555368485e-05,
+      "loss": 0.69,
+      "step": 549650
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 1.1960281625468736e-05,
+      "loss": 0.7411,
+      "step": 549660
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 1.195913369556899e-05,
+      "loss": 0.639,
+      "step": 549670
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 1.1957985765669243e-05,
+      "loss": 0.6937,
+      "step": 549680
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 1.1956837835769494e-05,
+      "loss": 0.703,
+      "step": 549690
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 1.1955689905869748e-05,
+      "loss": 0.6913,
+      "step": 549700
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 1.195454197597e-05,
+      "loss": 0.7498,
+      "step": 549710
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 1.1953394046070252e-05,
+      "loss": 0.7459,
+      "step": 549720
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 1.1952246116170506e-05,
+      "loss": 0.7166,
+      "step": 549730
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 1.1951098186270756e-05,
+      "loss": 0.7011,
+      "step": 549740
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 1.1949950256371009e-05,
+      "loss": 0.7753,
+      "step": 549750
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 1.1948802326471264e-05,
+      "loss": 0.6998,
+      "step": 549760
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 1.1947654396571514e-05,
+      "loss": 0.7807,
+      "step": 549770
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 1.1946506466671767e-05,
+      "loss": 0.7233,
+      "step": 549780
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 1.194535853677202e-05,
+      "loss": 0.7136,
+      "step": 549790
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 1.1944210606872272e-05,
+      "loss": 0.7215,
+      "step": 549800
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 1.1943062676972525e-05,
+      "loss": 0.7737,
+      "step": 549810
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 1.1941914747072778e-05,
+      "loss": 0.6843,
+      "step": 549820
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 1.194076681717303e-05,
+      "loss": 0.7492,
+      "step": 549830
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 1.1939618887273283e-05,
+      "loss": 0.7867,
+      "step": 549840
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 1.1938470957373536e-05,
+      "loss": 0.742,
+      "step": 549850
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 1.1937323027473788e-05,
+      "loss": 0.6481,
+      "step": 549860
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 1.1936175097574041e-05,
+      "loss": 0.6915,
+      "step": 549870
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 1.1935027167674294e-05,
+      "loss": 0.74,
+      "step": 549880
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 1.1933879237774544e-05,
+      "loss": 0.7381,
+      "step": 549890
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 1.1932731307874797e-05,
+      "loss": 0.6943,
+      "step": 549900
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 1.1931583377975052e-05,
+      "loss": 0.7006,
+      "step": 549910
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 1.1930435448075302e-05,
+      "loss": 0.7464,
+      "step": 549920
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 1.1929287518175555e-05,
+      "loss": 0.6813,
+      "step": 549930
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 1.1928139588275809e-05,
+      "loss": 0.8141,
+      "step": 549940
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 1.192699165837606e-05,
+      "loss": 0.7482,
+      "step": 549950
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 1.1925843728476313e-05,
+      "loss": 0.6822,
+      "step": 549960
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 1.1924695798576567e-05,
+      "loss": 0.7694,
+      "step": 549970
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 1.1923547868676818e-05,
+      "loss": 0.7687,
+      "step": 549980
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 1.1922399938777071e-05,
+      "loss": 0.6634,
+      "step": 549990
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 1.1921252008877325e-05,
+      "loss": 0.6574,
+      "step": 550000
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 1.1920104078977576e-05,
+      "loss": 0.6904,
+      "step": 550010
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 1.191895614907783e-05,
+      "loss": 0.6046,
+      "step": 550020
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 1.1917808219178083e-05,
+      "loss": 0.7357,
+      "step": 550030
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 1.1916660289278333e-05,
+      "loss": 0.7631,
+      "step": 550040
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 1.1915512359378586e-05,
+      "loss": 0.7624,
+      "step": 550050
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 1.1914364429478837e-05,
+      "loss": 0.6956,
+      "step": 550060
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 1.191321649957909e-05,
+      "loss": 0.7308,
+      "step": 550070
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 1.1912068569679344e-05,
+      "loss": 0.683,
+      "step": 550080
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 1.1910920639779595e-05,
+      "loss": 0.692,
+      "step": 550090
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 1.1909772709879849e-05,
+      "loss": 0.7486,
+      "step": 550100
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 1.1908624779980102e-05,
+      "loss": 0.7408,
+      "step": 550110
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 1.1907476850080353e-05,
+      "loss": 0.7932,
+      "step": 550120
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 1.1906328920180607e-05,
+      "loss": 0.7059,
+      "step": 550130
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 1.190518099028086e-05,
+      "loss": 0.6884,
+      "step": 550140
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 1.1904033060381111e-05,
+      "loss": 0.6895,
+      "step": 550150
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 1.1902885130481365e-05,
+      "loss": 0.8071,
+      "step": 550160
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 1.1901737200581618e-05,
+      "loss": 0.7777,
+      "step": 550170
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 1.1900589270681868e-05,
+      "loss": 0.6856,
+      "step": 550180
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 1.1899441340782121e-05,
+      "loss": 0.7067,
+      "step": 550190
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 1.1898293410882374e-05,
+      "loss": 0.7278,
+      "step": 550200
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 1.1897145480982626e-05,
+      "loss": 0.7763,
+      "step": 550210
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 1.1895997551082879e-05,
+      "loss": 0.6778,
+      "step": 550220
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 1.1894849621183132e-05,
+      "loss": 0.6772,
+      "step": 550230
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 1.1893701691283384e-05,
+      "loss": 0.7279,
+      "step": 550240
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 1.1892553761383637e-05,
+      "loss": 0.7558,
+      "step": 550250
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 1.189140583148389e-05,
+      "loss": 0.7001,
+      "step": 550260
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 1.1890257901584142e-05,
+      "loss": 0.7385,
+      "step": 550270
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 1.1889109971684395e-05,
+      "loss": 0.6956,
+      "step": 550280
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 1.1887962041784648e-05,
+      "loss": 0.7052,
+      "step": 550290
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 1.18868141118849e-05,
+      "loss": 0.6492,
+      "step": 550300
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 1.1885666181985153e-05,
+      "loss": 0.6719,
+      "step": 550310
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 1.1884518252085406e-05,
+      "loss": 0.6801,
+      "step": 550320
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 1.1883370322185656e-05,
+      "loss": 0.7047,
+      "step": 550330
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 1.188222239228591e-05,
+      "loss": 0.8453,
+      "step": 550340
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 1.1881074462386163e-05,
+      "loss": 0.7521,
+      "step": 550350
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 1.1879926532486414e-05,
+      "loss": 0.7496,
+      "step": 550360
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 1.1878778602586667e-05,
+      "loss": 0.7122,
+      "step": 550370
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 1.187763067268692e-05,
+      "loss": 0.7372,
+      "step": 550380
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 1.1876482742787172e-05,
+      "loss": 0.7785,
+      "step": 550390
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 1.1875334812887426e-05,
+      "loss": 0.7549,
+      "step": 550400
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 1.1874186882987677e-05,
+      "loss": 0.6606,
+      "step": 550410
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 1.187303895308793e-05,
+      "loss": 0.7211,
+      "step": 550420
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 1.1871891023188184e-05,
+      "loss": 0.6763,
+      "step": 550430
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 1.1870743093288435e-05,
+      "loss": 0.7063,
+      "step": 550440
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 1.1869595163388688e-05,
+      "loss": 0.7097,
+      "step": 550450
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 1.1868447233488942e-05,
+      "loss": 0.7103,
+      "step": 550460
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 1.1867299303589191e-05,
+      "loss": 0.6641,
+      "step": 550470
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 1.1866151373689445e-05,
+      "loss": 0.6697,
+      "step": 550480
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 1.1865003443789698e-05,
+      "loss": 0.7369,
+      "step": 550490
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 1.186385551388995e-05,
+      "loss": 0.655,
+      "step": 550500
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 1.1862707583990203e-05,
+      "loss": 0.7648,
+      "step": 550510
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 1.1861559654090456e-05,
+      "loss": 0.7573,
+      "step": 550520
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 1.1860411724190708e-05,
+      "loss": 0.7794,
+      "step": 550530
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 1.185926379429096e-05,
+      "loss": 0.7217,
+      "step": 550540
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 1.1858115864391214e-05,
+      "loss": 0.7726,
+      "step": 550550
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 1.1856967934491466e-05,
+      "loss": 0.7556,
+      "step": 550560
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 1.1855820004591719e-05,
+      "loss": 0.6903,
+      "step": 550570
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 1.1854672074691972e-05,
+      "loss": 0.6901,
+      "step": 550580
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 1.1853524144792224e-05,
+      "loss": 0.737,
+      "step": 550590
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 1.1852376214892477e-05,
+      "loss": 0.7227,
+      "step": 550600
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 1.185122828499273e-05,
+      "loss": 0.6887,
+      "step": 550610
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 1.185008035509298e-05,
+      "loss": 0.7501,
+      "step": 550620
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 1.1848932425193233e-05,
+      "loss": 0.7269,
+      "step": 550630
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 1.1847784495293486e-05,
+      "loss": 0.6911,
+      "step": 550640
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 1.1846636565393738e-05,
+      "loss": 0.6203,
+      "step": 550650
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 1.1845488635493991e-05,
+      "loss": 0.678,
+      "step": 550660
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 1.1844340705594244e-05,
+      "loss": 0.7246,
+      "step": 550670
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 1.1843192775694496e-05,
+      "loss": 0.7274,
+      "step": 550680
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 1.184204484579475e-05,
+      "loss": 0.706,
+      "step": 550690
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 1.1840896915895002e-05,
+      "loss": 0.7022,
+      "step": 550700
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 1.1839748985995254e-05,
+      "loss": 0.7395,
+      "step": 550710
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 1.1838601056095507e-05,
+      "loss": 0.7107,
+      "step": 550720
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 1.183745312619576e-05,
+      "loss": 0.7594,
+      "step": 550730
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 1.1836305196296012e-05,
+      "loss": 0.7397,
+      "step": 550740
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 1.1835157266396265e-05,
+      "loss": 0.7563,
+      "step": 550750
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 1.1834009336496515e-05,
+      "loss": 0.7631,
+      "step": 550760
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 1.1832861406596768e-05,
+      "loss": 0.7431,
+      "step": 550770
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 1.1831713476697022e-05,
+      "loss": 0.7493,
+      "step": 550780
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 1.1830565546797273e-05,
+      "loss": 0.7272,
+      "step": 550790
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 1.1829417616897526e-05,
+      "loss": 0.7294,
+      "step": 550800
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 1.182826968699778e-05,
+      "loss": 0.7251,
+      "step": 550810
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 1.1827121757098031e-05,
+      "loss": 0.7566,
+      "step": 550820
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 1.1825973827198284e-05,
+      "loss": 0.7313,
+      "step": 550830
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 1.1824825897298538e-05,
+      "loss": 0.6557,
+      "step": 550840
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 1.182367796739879e-05,
+      "loss": 0.7822,
+      "step": 550850
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 1.1822530037499042e-05,
+      "loss": 0.6793,
+      "step": 550860
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 1.1821382107599296e-05,
+      "loss": 0.7323,
+      "step": 550870
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 1.1820234177699547e-05,
+      "loss": 0.7745,
+      "step": 550880
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 1.18190862477998e-05,
+      "loss": 0.6913,
+      "step": 550890
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 1.1817938317900054e-05,
+      "loss": 0.7718,
+      "step": 550900
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 1.1816790388000304e-05,
+      "loss": 0.7184,
+      "step": 550910
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 1.1815642458100557e-05,
+      "loss": 0.7014,
+      "step": 550920
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 1.181449452820081e-05,
+      "loss": 0.7356,
+      "step": 550930
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 1.1813346598301062e-05,
+      "loss": 0.7585,
+      "step": 550940
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 1.1812198668401315e-05,
+      "loss": 0.6602,
+      "step": 550950
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 1.1811050738501568e-05,
+      "loss": 0.7483,
+      "step": 550960
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 1.180990280860182e-05,
+      "loss": 0.6965,
+      "step": 550970
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 1.1808754878702073e-05,
+      "loss": 0.6937,
+      "step": 550980
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 1.1807606948802326e-05,
+      "loss": 0.7223,
+      "step": 550990
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 1.1806459018902578e-05,
+      "loss": 0.7581,
+      "step": 551000
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 1.1805311089002831e-05,
+      "loss": 0.7423,
+      "step": 551010
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 1.1804163159103084e-05,
+      "loss": 0.6727,
+      "step": 551020
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 1.1803015229203336e-05,
+      "loss": 0.8038,
+      "step": 551030
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 1.1801867299303589e-05,
+      "loss": 0.6215,
+      "step": 551040
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 1.1800719369403842e-05,
+      "loss": 0.7144,
+      "step": 551050
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 1.1799571439504092e-05,
+      "loss": 0.702,
+      "step": 551060
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 1.1798423509604345e-05,
+      "loss": 0.725,
+      "step": 551070
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 1.1797275579704597e-05,
+      "loss": 0.6974,
+      "step": 551080
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 1.179612764980485e-05,
+      "loss": 0.6684,
+      "step": 551090
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 1.1794979719905103e-05,
+      "loss": 0.7291,
+      "step": 551100
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 1.1793831790005355e-05,
+      "loss": 0.757,
+      "step": 551110
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 1.1792683860105608e-05,
+      "loss": 0.7671,
+      "step": 551120
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 1.1791535930205861e-05,
+      "loss": 0.6763,
+      "step": 551130
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 1.1790388000306113e-05,
+      "loss": 0.6741,
+      "step": 551140
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 1.1789240070406366e-05,
+      "loss": 0.6862,
+      "step": 551150
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 1.178809214050662e-05,
+      "loss": 0.7287,
+      "step": 551160
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 1.1786944210606871e-05,
+      "loss": 0.639,
+      "step": 551170
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 1.1785796280707124e-05,
+      "loss": 0.6569,
+      "step": 551180
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 1.1784648350807377e-05,
+      "loss": 0.8404,
+      "step": 551190
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 1.1783500420907627e-05,
+      "loss": 0.7219,
+      "step": 551200
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 1.178235249100788e-05,
+      "loss": 0.7483,
+      "step": 551210
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 1.1781204561108134e-05,
+      "loss": 0.7244,
+      "step": 551220
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 1.1780056631208385e-05,
+      "loss": 0.7868,
+      "step": 551230
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 1.1778908701308639e-05,
+      "loss": 0.7351,
+      "step": 551240
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 1.1777760771408892e-05,
+      "loss": 0.7114,
+      "step": 551250
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 1.1776612841509143e-05,
+      "loss": 0.6805,
+      "step": 551260
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 1.1775464911609397e-05,
+      "loss": 0.7778,
+      "step": 551270
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 1.177431698170965e-05,
+      "loss": 0.6184,
+      "step": 551280
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 1.1773169051809901e-05,
+      "loss": 0.6869,
+      "step": 551290
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 1.1772021121910155e-05,
+      "loss": 0.6152,
+      "step": 551300
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 1.1770873192010408e-05,
+      "loss": 0.6694,
+      "step": 551310
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 1.176972526211066e-05,
+      "loss": 0.688,
+      "step": 551320
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 1.1768577332210913e-05,
+      "loss": 0.7341,
+      "step": 551330
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 1.1767429402311166e-05,
+      "loss": 0.7413,
+      "step": 551340
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 1.1766281472411416e-05,
+      "loss": 0.7055,
+      "step": 551350
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 1.1765133542511669e-05,
+      "loss": 0.8402,
+      "step": 551360
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 1.1763985612611922e-05,
+      "loss": 0.8151,
+      "step": 551370
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 1.1762837682712174e-05,
+      "loss": 0.7048,
+      "step": 551380
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 1.1761689752812427e-05,
+      "loss": 0.7125,
+      "step": 551390
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 1.176054182291268e-05,
+      "loss": 0.7749,
+      "step": 551400
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 1.1759393893012932e-05,
+      "loss": 0.7456,
+      "step": 551410
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 1.1758245963113185e-05,
+      "loss": 0.7123,
+      "step": 551420
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 1.1757098033213437e-05,
+      "loss": 0.719,
+      "step": 551430
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 1.175595010331369e-05,
+      "loss": 0.7655,
+      "step": 551440
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 1.1754802173413943e-05,
+      "loss": 0.7837,
+      "step": 551450
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 1.1753654243514195e-05,
+      "loss": 0.6541,
+      "step": 551460
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 1.1752506313614448e-05,
+      "loss": 0.7279,
+      "step": 551470
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 1.1751358383714701e-05,
+      "loss": 0.7127,
+      "step": 551480
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 1.1750210453814951e-05,
+      "loss": 0.7869,
+      "step": 551490
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 1.1749062523915204e-05,
+      "loss": 0.6956,
+      "step": 551500
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 1.1747914594015458e-05,
+      "loss": 0.7006,
+      "step": 551510
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 1.1746766664115709e-05,
+      "loss": 0.7592,
+      "step": 551520
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 1.1745618734215962e-05,
+      "loss": 0.6924,
+      "step": 551530
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 1.1744470804316216e-05,
+      "loss": 0.6288,
+      "step": 551540
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 1.1743322874416467e-05,
+      "loss": 0.7701,
+      "step": 551550
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 1.174217494451672e-05,
+      "loss": 0.7039,
+      "step": 551560
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 1.1741027014616974e-05,
+      "loss": 0.8345,
+      "step": 551570
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 1.1739879084717225e-05,
+      "loss": 0.729,
+      "step": 551580
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 1.1738731154817478e-05,
+      "loss": 0.6745,
+      "step": 551590
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 1.1737583224917732e-05,
+      "loss": 0.6538,
+      "step": 551600
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 1.1736435295017983e-05,
+      "loss": 0.7705,
+      "step": 551610
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 1.1735287365118236e-05,
+      "loss": 0.7784,
+      "step": 551620
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 1.173413943521849e-05,
+      "loss": 0.7591,
+      "step": 551630
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 1.173299150531874e-05,
+      "loss": 0.6835,
+      "step": 551640
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 1.1731843575418993e-05,
+      "loss": 0.7823,
+      "step": 551650
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 1.1730695645519246e-05,
+      "loss": 0.713,
+      "step": 551660
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 1.1729547715619498e-05,
+      "loss": 0.6953,
+      "step": 551670
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 1.172839978571975e-05,
+      "loss": 0.762,
+      "step": 551680
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 1.1727251855820004e-05,
+      "loss": 0.6722,
+      "step": 551690
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 1.1726103925920256e-05,
+      "loss": 0.7438,
+      "step": 551700
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 1.1724955996020509e-05,
+      "loss": 0.6708,
+      "step": 551710
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 1.1723808066120762e-05,
+      "loss": 0.6721,
+      "step": 551720
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 1.1722660136221014e-05,
+      "loss": 0.6601,
+      "step": 551730
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 1.1721512206321267e-05,
+      "loss": 0.7191,
+      "step": 551740
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 1.172036427642152e-05,
+      "loss": 0.7376,
+      "step": 551750
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 1.1719216346521772e-05,
+      "loss": 0.7141,
+      "step": 551760
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 1.1718068416622025e-05,
+      "loss": 0.723,
+      "step": 551770
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 1.1716920486722275e-05,
+      "loss": 0.7333,
+      "step": 551780
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 1.1715772556822528e-05,
+      "loss": 0.6985,
+      "step": 551790
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 1.1714624626922781e-05,
+      "loss": 0.713,
+      "step": 551800
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 1.1713476697023033e-05,
+      "loss": 0.6616,
+      "step": 551810
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 1.1712328767123286e-05,
+      "loss": 0.6725,
+      "step": 551820
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 1.171118083722354e-05,
+      "loss": 0.6189,
+      "step": 551830
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 1.171003290732379e-05,
+      "loss": 0.7244,
+      "step": 551840
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 1.1708884977424044e-05,
+      "loss": 0.7329,
+      "step": 551850
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 1.1707737047524297e-05,
+      "loss": 0.76,
+      "step": 551860
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 1.1706589117624549e-05,
+      "loss": 0.8016,
+      "step": 551870
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 1.1705441187724802e-05,
+      "loss": 0.7155,
+      "step": 551880
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 1.1704293257825055e-05,
+      "loss": 0.7043,
+      "step": 551890
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 1.1703145327925307e-05,
+      "loss": 0.7277,
+      "step": 551900
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 1.170199739802556e-05,
+      "loss": 0.7831,
+      "step": 551910
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 1.1700849468125813e-05,
+      "loss": 0.6879,
+      "step": 551920
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 1.1699701538226063e-05,
+      "loss": 0.7626,
+      "step": 551930
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 1.1698553608326316e-05,
+      "loss": 0.6923,
+      "step": 551940
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 1.169740567842657e-05,
+      "loss": 0.703,
+      "step": 551950
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 1.1696257748526821e-05,
+      "loss": 0.7095,
+      "step": 551960
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 1.1695109818627074e-05,
+      "loss": 0.7037,
+      "step": 551970
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 1.1693961888727328e-05,
+      "loss": 0.6925,
+      "step": 551980
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 1.169281395882758e-05,
+      "loss": 0.7052,
+      "step": 551990
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 1.1691666028927833e-05,
+      "loss": 0.7469,
+      "step": 552000
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 1.1690518099028086e-05,
+      "loss": 0.7489,
+      "step": 552010
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 1.1689370169128337e-05,
+      "loss": 0.6828,
+      "step": 552020
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 1.168822223922859e-05,
+      "loss": 0.6849,
+      "step": 552030
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 1.1687074309328844e-05,
+      "loss": 0.6601,
+      "step": 552040
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 1.1685926379429095e-05,
+      "loss": 0.7372,
+      "step": 552050
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 1.1684778449529349e-05,
+      "loss": 0.7207,
+      "step": 552060
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 1.1683630519629602e-05,
+      "loss": 0.8506,
+      "step": 552070
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 1.1682482589729852e-05,
+      "loss": 0.725,
+      "step": 552080
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 1.1681334659830105e-05,
+      "loss": 0.736,
+      "step": 552090
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 1.1680186729930358e-05,
+      "loss": 0.665,
+      "step": 552100
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 1.167903880003061e-05,
+      "loss": 0.7015,
+      "step": 552110
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 1.1677890870130863e-05,
+      "loss": 0.7085,
+      "step": 552120
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 1.1676742940231114e-05,
+      "loss": 0.7287,
+      "step": 552130
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 1.1675595010331368e-05,
+      "loss": 0.7311,
+      "step": 552140
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 1.1674447080431621e-05,
+      "loss": 0.7347,
+      "step": 552150
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 1.1673299150531873e-05,
+      "loss": 0.7434,
+      "step": 552160
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 1.1672151220632126e-05,
+      "loss": 0.692,
+      "step": 552170
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 1.1671003290732379e-05,
+      "loss": 0.742,
+      "step": 552180
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 1.166985536083263e-05,
+      "loss": 0.682,
+      "step": 552190
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 1.1668707430932884e-05,
+      "loss": 0.7962,
+      "step": 552200
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 1.1667559501033137e-05,
+      "loss": 0.6754,
+      "step": 552210
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 1.1666411571133387e-05,
+      "loss": 0.6838,
+      "step": 552220
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 1.166526364123364e-05,
+      "loss": 0.8214,
+      "step": 552230
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 1.1664115711333893e-05,
+      "loss": 0.6757,
+      "step": 552240
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 1.1662967781434145e-05,
+      "loss": 0.7744,
+      "step": 552250
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 1.1661819851534398e-05,
+      "loss": 0.7131,
+      "step": 552260
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 1.1660671921634651e-05,
+      "loss": 0.7185,
+      "step": 552270
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 1.1659523991734903e-05,
+      "loss": 0.6861,
+      "step": 552280
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 1.1658376061835156e-05,
+      "loss": 0.7371,
+      "step": 552290
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 1.165722813193541e-05,
+      "loss": 0.8136,
+      "step": 552300
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 1.1656080202035661e-05,
+      "loss": 0.725,
+      "step": 552310
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 1.1654932272135914e-05,
+      "loss": 0.6594,
+      "step": 552320
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 1.1653784342236167e-05,
+      "loss": 0.6674,
+      "step": 552330
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 1.1652636412336419e-05,
+      "loss": 0.7473,
+      "step": 552340
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 1.1651488482436672e-05,
+      "loss": 0.655,
+      "step": 552350
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 1.1650340552536926e-05,
+      "loss": 0.6722,
+      "step": 552360
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 1.1649192622637175e-05,
+      "loss": 0.7056,
+      "step": 552370
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 1.1648044692737429e-05,
+      "loss": 0.7454,
+      "step": 552380
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 1.1646896762837682e-05,
+      "loss": 0.7075,
+      "step": 552390
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 1.1645748832937933e-05,
+      "loss": 0.7019,
+      "step": 552400
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 1.1644600903038187e-05,
+      "loss": 0.7872,
+      "step": 552410
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 1.164345297313844e-05,
+      "loss": 0.7314,
+      "step": 552420
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 1.1642305043238691e-05,
+      "loss": 0.6711,
+      "step": 552430
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 1.1641157113338945e-05,
+      "loss": 0.7427,
+      "step": 552440
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 1.1640009183439198e-05,
+      "loss": 0.7295,
+      "step": 552450
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 1.163886125353945e-05,
+      "loss": 0.6851,
+      "step": 552460
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 1.1637713323639703e-05,
+      "loss": 0.74,
+      "step": 552470
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 1.1636565393739954e-05,
+      "loss": 0.753,
+      "step": 552480
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 1.1635417463840207e-05,
+      "loss": 0.7074,
+      "step": 552490
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 1.163426953394046e-05,
+      "loss": 0.7148,
+      "step": 552500
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 1.163312160404071e-05,
+      "loss": 0.6746,
+      "step": 552510
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 1.1631973674140964e-05,
+      "loss": 0.6966,
+      "step": 552520
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 1.1630825744241217e-05,
+      "loss": 0.7187,
+      "step": 552530
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 1.1629677814341469e-05,
+      "loss": 0.7844,
+      "step": 552540
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 1.1628529884441722e-05,
+      "loss": 0.7796,
+      "step": 552550
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 1.1627381954541975e-05,
+      "loss": 0.6857,
+      "step": 552560
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 1.1626234024642227e-05,
+      "loss": 0.7379,
+      "step": 552570
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 1.162508609474248e-05,
+      "loss": 0.697,
+      "step": 552580
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 1.1623938164842733e-05,
+      "loss": 0.6928,
+      "step": 552590
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 1.1622790234942985e-05,
+      "loss": 0.7203,
+      "step": 552600
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 1.1621642305043238e-05,
+      "loss": 0.7363,
+      "step": 552610
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 1.1620494375143491e-05,
+      "loss": 0.7879,
+      "step": 552620
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 1.1619346445243743e-05,
+      "loss": 0.7275,
+      "step": 552630
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 1.1618198515343996e-05,
+      "loss": 0.6813,
+      "step": 552640
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 1.161705058544425e-05,
+      "loss": 0.7305,
+      "step": 552650
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 1.1615902655544499e-05,
+      "loss": 0.6885,
+      "step": 552660
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 1.1614754725644752e-05,
+      "loss": 0.6642,
+      "step": 552670
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 1.1613606795745006e-05,
+      "loss": 0.7594,
+      "step": 552680
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 1.1612458865845257e-05,
+      "loss": 0.7127,
+      "step": 552690
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 1.161131093594551e-05,
+      "loss": 0.6848,
+      "step": 552700
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 1.1610163006045764e-05,
+      "loss": 0.7554,
+      "step": 552710
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 1.1609015076146015e-05,
+      "loss": 0.745,
+      "step": 552720
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 1.1607867146246268e-05,
+      "loss": 0.6591,
+      "step": 552730
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 1.1606719216346522e-05,
+      "loss": 0.7048,
+      "step": 552740
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 1.1605571286446773e-05,
+      "loss": 0.6603,
+      "step": 552750
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 1.1604423356547026e-05,
+      "loss": 0.6792,
+      "step": 552760
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 1.160327542664728e-05,
+      "loss": 0.738,
+      "step": 552770
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 1.1602127496747531e-05,
+      "loss": 0.7376,
+      "step": 552780
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 1.1600979566847784e-05,
+      "loss": 0.7311,
+      "step": 552790
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 1.1599831636948034e-05,
+      "loss": 0.7651,
+      "step": 552800
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 1.1598683707048288e-05,
+      "loss": 0.7048,
+      "step": 552810
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 1.159753577714854e-05,
+      "loss": 0.7626,
+      "step": 552820
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 1.1596387847248792e-05,
+      "loss": 0.7414,
+      "step": 552830
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 1.1595239917349046e-05,
+      "loss": 0.7165,
+      "step": 552840
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 1.1594091987449299e-05,
+      "loss": 0.7362,
+      "step": 552850
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 1.159294405754955e-05,
+      "loss": 0.7296,
+      "step": 552860
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 1.1591796127649804e-05,
+      "loss": 0.6719,
+      "step": 552870
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 1.1590648197750057e-05,
+      "loss": 0.6392,
+      "step": 552880
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 1.1589500267850308e-05,
+      "loss": 0.7812,
+      "step": 552890
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 1.1588352337950562e-05,
+      "loss": 0.7637,
+      "step": 552900
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 1.1587204408050815e-05,
+      "loss": 0.7276,
+      "step": 552910
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 1.1586056478151066e-05,
+      "loss": 0.6627,
+      "step": 552920
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 1.158490854825132e-05,
+      "loss": 0.7626,
+      "step": 552930
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 1.1583760618351573e-05,
+      "loss": 0.705,
+      "step": 552940
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 1.1582612688451823e-05,
+      "loss": 0.6994,
+      "step": 552950
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 1.1581464758552076e-05,
+      "loss": 0.6847,
+      "step": 552960
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 1.158031682865233e-05,
+      "loss": 0.6707,
+      "step": 552970
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 1.157916889875258e-05,
+      "loss": 0.7221,
+      "step": 552980
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 1.1578020968852834e-05,
+      "loss": 0.7207,
+      "step": 552990
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 1.1576873038953087e-05,
+      "loss": 0.6282,
+      "step": 553000
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 1.1575725109053339e-05,
+      "loss": 0.711,
+      "step": 553010
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 1.1574577179153592e-05,
+      "loss": 0.7911,
+      "step": 553020
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 1.1573429249253845e-05,
+      "loss": 0.7808,
+      "step": 553030
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 1.1572281319354097e-05,
+      "loss": 0.8161,
+      "step": 553040
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 1.157113338945435e-05,
+      "loss": 0.7347,
+      "step": 553050
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 1.1569985459554603e-05,
+      "loss": 0.7006,
+      "step": 553060
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 1.1568837529654855e-05,
+      "loss": 0.7303,
+      "step": 553070
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 1.1567689599755108e-05,
+      "loss": 0.6978,
+      "step": 553080
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 1.1566541669855361e-05,
+      "loss": 0.6691,
+      "step": 553090
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 1.1565393739955611e-05,
+      "loss": 0.6745,
+      "step": 553100
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 1.1564245810055864e-05,
+      "loss": 0.7835,
+      "step": 553110
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 1.1563097880156118e-05,
+      "loss": 0.6822,
+      "step": 553120
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 1.156194995025637e-05,
+      "loss": 0.6634,
+      "step": 553130
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 1.1560802020356623e-05,
+      "loss": 0.7149,
+      "step": 553140
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 1.1559654090456874e-05,
+      "loss": 0.7079,
+      "step": 553150
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 1.1558506160557127e-05,
+      "loss": 0.7722,
+      "step": 553160
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 1.155735823065738e-05,
+      "loss": 0.7935,
+      "step": 553170
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 1.1556210300757632e-05,
+      "loss": 0.762,
+      "step": 553180
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 1.1555062370857885e-05,
+      "loss": 0.7163,
+      "step": 553190
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 1.1553914440958139e-05,
+      "loss": 0.7943,
+      "step": 553200
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 1.155276651105839e-05,
+      "loss": 0.7044,
+      "step": 553210
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 1.1551618581158643e-05,
+      "loss": 0.7183,
+      "step": 553220
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 1.1550470651258897e-05,
+      "loss": 0.7673,
+      "step": 553230
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 1.1549322721359146e-05,
+      "loss": 0.7391,
+      "step": 553240
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 1.15481747914594e-05,
+      "loss": 0.7605,
+      "step": 553250
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 1.1547026861559653e-05,
+      "loss": 0.6821,
+      "step": 553260
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 1.1545878931659905e-05,
+      "loss": 0.7823,
+      "step": 553270
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 1.1544731001760158e-05,
+      "loss": 0.7459,
+      "step": 553280
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 1.1543583071860411e-05,
+      "loss": 0.6996,
+      "step": 553290
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 1.1542435141960663e-05,
+      "loss": 0.6489,
+      "step": 553300
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 1.1541287212060916e-05,
+      "loss": 0.7249,
+      "step": 553310
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 1.1540139282161169e-05,
+      "loss": 0.5903,
+      "step": 553320
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 1.153899135226142e-05,
+      "loss": 0.688,
+      "step": 553330
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 1.1537843422361674e-05,
+      "loss": 0.7162,
+      "step": 553340
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 1.1536695492461927e-05,
+      "loss": 0.7584,
+      "step": 553350
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 1.1535547562562179e-05,
+      "loss": 0.7527,
+      "step": 553360
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 1.1534399632662432e-05,
+      "loss": 0.6618,
+      "step": 553370
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 1.1533251702762685e-05,
+      "loss": 0.719,
+      "step": 553380
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 1.1532103772862935e-05,
+      "loss": 0.7064,
+      "step": 553390
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 1.1530955842963188e-05,
+      "loss": 0.7701,
+      "step": 553400
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 1.1529807913063441e-05,
+      "loss": 0.7221,
+      "step": 553410
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 1.1528659983163693e-05,
+      "loss": 0.7529,
+      "step": 553420
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 1.1527512053263946e-05,
+      "loss": 0.7288,
+      "step": 553430
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 1.15263641233642e-05,
+      "loss": 0.7131,
+      "step": 553440
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 1.1525216193464451e-05,
+      "loss": 0.6818,
+      "step": 553450
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 1.1524068263564704e-05,
+      "loss": 0.7304,
+      "step": 553460
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 1.1522920333664957e-05,
+      "loss": 0.7104,
+      "step": 553470
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 1.1521772403765209e-05,
+      "loss": 0.6663,
+      "step": 553480
+    },
+    {
+      "epoch": 25.39,
+      "learning_rate": 1.1520624473865462e-05,
+      "loss": 0.7018,
+      "step": 553490
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 1.1519476543965714e-05,
+      "loss": 0.7161,
+      "step": 553500
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 1.1518328614065967e-05,
+      "loss": 0.726,
+      "step": 553510
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 1.151718068416622e-05,
+      "loss": 0.7833,
+      "step": 553520
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 1.151603275426647e-05,
+      "loss": 0.639,
+      "step": 553530
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 1.1514884824366723e-05,
+      "loss": 0.6473,
+      "step": 553540
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 1.1513736894466977e-05,
+      "loss": 0.7463,
+      "step": 553550
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 1.1512588964567228e-05,
+      "loss": 0.7035,
+      "step": 553560
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 1.1511441034667481e-05,
+      "loss": 0.7326,
+      "step": 553570
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 1.1510293104767735e-05,
+      "loss": 0.7949,
+      "step": 553580
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 1.1509145174867986e-05,
+      "loss": 0.7086,
+      "step": 553590
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 1.150799724496824e-05,
+      "loss": 0.6995,
+      "step": 553600
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 1.1506849315068493e-05,
+      "loss": 0.703,
+      "step": 553610
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 1.1505701385168744e-05,
+      "loss": 0.6958,
+      "step": 553620
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 1.1504553455268998e-05,
+      "loss": 0.7372,
+      "step": 553630
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 1.150340552536925e-05,
+      "loss": 0.761,
+      "step": 553640
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 1.1502257595469502e-05,
+      "loss": 0.765,
+      "step": 553650
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 1.1501109665569756e-05,
+      "loss": 0.6844,
+      "step": 553660
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 1.1499961735670009e-05,
+      "loss": 0.668,
+      "step": 553670
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 1.1498813805770259e-05,
+      "loss": 0.7094,
+      "step": 553680
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 1.1497665875870512e-05,
+      "loss": 0.7108,
+      "step": 553690
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 1.1496517945970765e-05,
+      "loss": 0.739,
+      "step": 553700
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 1.1495370016071017e-05,
+      "loss": 0.6787,
+      "step": 553710
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 1.149422208617127e-05,
+      "loss": 0.7101,
+      "step": 553720
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 1.1493074156271523e-05,
+      "loss": 0.6896,
+      "step": 553730
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 1.1491926226371775e-05,
+      "loss": 0.6804,
+      "step": 553740
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 1.1490778296472028e-05,
+      "loss": 0.7242,
+      "step": 553750
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 1.1489630366572281e-05,
+      "loss": 0.7072,
+      "step": 553760
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 1.1488482436672533e-05,
+      "loss": 0.6822,
+      "step": 553770
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 1.1487334506772786e-05,
+      "loss": 0.7443,
+      "step": 553780
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 1.148618657687304e-05,
+      "loss": 0.6987,
+      "step": 553790
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 1.148503864697329e-05,
+      "loss": 0.7209,
+      "step": 553800
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 1.1483890717073544e-05,
+      "loss": 0.7168,
+      "step": 553810
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 1.1482742787173797e-05,
+      "loss": 0.6888,
+      "step": 553820
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 1.1481594857274047e-05,
+      "loss": 0.7856,
+      "step": 553830
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 1.14804469273743e-05,
+      "loss": 0.7146,
+      "step": 553840
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 1.1479298997474552e-05,
+      "loss": 0.7958,
+      "step": 553850
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 1.1478151067574805e-05,
+      "loss": 0.7731,
+      "step": 553860
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 1.1477003137675058e-05,
+      "loss": 0.6415,
+      "step": 553870
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 1.147585520777531e-05,
+      "loss": 0.7402,
+      "step": 553880
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 1.1474707277875563e-05,
+      "loss": 0.711,
+      "step": 553890
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 1.1473559347975816e-05,
+      "loss": 0.7121,
+      "step": 553900
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 1.1472411418076068e-05,
+      "loss": 0.7371,
+      "step": 553910
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 1.1471263488176321e-05,
+      "loss": 0.8232,
+      "step": 553920
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 1.1470115558276574e-05,
+      "loss": 0.7572,
+      "step": 553930
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 1.1468967628376826e-05,
+      "loss": 0.7223,
+      "step": 553940
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 1.146781969847708e-05,
+      "loss": 0.6883,
+      "step": 553950
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 1.1466671768577332e-05,
+      "loss": 0.6984,
+      "step": 553960
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 1.1465523838677582e-05,
+      "loss": 0.7426,
+      "step": 553970
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 1.1464375908777836e-05,
+      "loss": 0.7429,
+      "step": 553980
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 1.1463227978878089e-05,
+      "loss": 0.6961,
+      "step": 553990
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 1.146208004897834e-05,
+      "loss": 0.7378,
+      "step": 554000
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 1.1460932119078594e-05,
+      "loss": 0.7534,
+      "step": 554010
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 1.1459784189178847e-05,
+      "loss": 0.7479,
+      "step": 554020
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 1.1458636259279098e-05,
+      "loss": 0.6917,
+      "step": 554030
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 1.1457488329379352e-05,
+      "loss": 0.6798,
+      "step": 554040
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 1.1456340399479605e-05,
+      "loss": 0.7411,
+      "step": 554050
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 1.1455192469579856e-05,
+      "loss": 0.7266,
+      "step": 554060
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 1.145404453968011e-05,
+      "loss": 0.6327,
+      "step": 554070
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 1.1452896609780363e-05,
+      "loss": 0.7376,
+      "step": 554080
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 1.1451748679880614e-05,
+      "loss": 0.7631,
+      "step": 554090
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 1.1450600749980868e-05,
+      "loss": 0.7066,
+      "step": 554100
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 1.1449452820081121e-05,
+      "loss": 0.7226,
+      "step": 554110
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 1.144830489018137e-05,
+      "loss": 0.7333,
+      "step": 554120
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 1.1447156960281624e-05,
+      "loss": 0.7254,
+      "step": 554130
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 1.1446009030381877e-05,
+      "loss": 0.7477,
+      "step": 554140
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 1.1444861100482129e-05,
+      "loss": 0.6889,
+      "step": 554150
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 1.1443713170582382e-05,
+      "loss": 0.7393,
+      "step": 554160
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 1.1442565240682634e-05,
+      "loss": 0.6784,
+      "step": 554170
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 1.1441417310782887e-05,
+      "loss": 0.7099,
+      "step": 554180
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 1.144026938088314e-05,
+      "loss": 0.8095,
+      "step": 554190
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 1.1439121450983392e-05,
+      "loss": 0.6744,
+      "step": 554200
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 1.1437973521083645e-05,
+      "loss": 0.7488,
+      "step": 554210
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 1.1436825591183898e-05,
+      "loss": 0.6299,
+      "step": 554220
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 1.143567766128415e-05,
+      "loss": 0.7126,
+      "step": 554230
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 1.1434529731384403e-05,
+      "loss": 0.706,
+      "step": 554240
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 1.1433381801484656e-05,
+      "loss": 0.6357,
+      "step": 554250
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 1.1432233871584906e-05,
+      "loss": 0.6921,
+      "step": 554260
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 1.143108594168516e-05,
+      "loss": 0.7443,
+      "step": 554270
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 1.1429938011785413e-05,
+      "loss": 0.7065,
+      "step": 554280
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 1.1428790081885664e-05,
+      "loss": 0.648,
+      "step": 554290
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 1.1427642151985917e-05,
+      "loss": 0.6927,
+      "step": 554300
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 1.142649422208617e-05,
+      "loss": 0.7237,
+      "step": 554310
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 1.1425346292186422e-05,
+      "loss": 0.6974,
+      "step": 554320
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 1.1424198362286675e-05,
+      "loss": 0.7061,
+      "step": 554330
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 1.1423050432386929e-05,
+      "loss": 0.7662,
+      "step": 554340
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 1.142190250248718e-05,
+      "loss": 0.7483,
+      "step": 554350
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 1.1420754572587433e-05,
+      "loss": 0.6848,
+      "step": 554360
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 1.1419606642687687e-05,
+      "loss": 0.707,
+      "step": 554370
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 1.1418458712787938e-05,
+      "loss": 0.7374,
+      "step": 554380
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 1.1417310782888191e-05,
+      "loss": 0.8356,
+      "step": 554390
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 1.1416162852988445e-05,
+      "loss": 0.7165,
+      "step": 554400
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 1.1415014923088695e-05,
+      "loss": 0.7337,
+      "step": 554410
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 1.1413866993188948e-05,
+      "loss": 0.7402,
+      "step": 554420
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 1.1412719063289201e-05,
+      "loss": 0.7253,
+      "step": 554430
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 1.1411571133389453e-05,
+      "loss": 0.8605,
+      "step": 554440
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 1.1410423203489706e-05,
+      "loss": 0.7461,
+      "step": 554450
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 1.1409275273589959e-05,
+      "loss": 0.7476,
+      "step": 554460
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 1.140812734369021e-05,
+      "loss": 0.793,
+      "step": 554470
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 1.1406979413790464e-05,
+      "loss": 0.7184,
+      "step": 554480
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 1.1405831483890717e-05,
+      "loss": 0.6942,
+      "step": 554490
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 1.1404683553990969e-05,
+      "loss": 0.6915,
+      "step": 554500
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 1.1403535624091222e-05,
+      "loss": 0.7578,
+      "step": 554510
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 1.1402387694191473e-05,
+      "loss": 0.7323,
+      "step": 554520
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 1.1401239764291727e-05,
+      "loss": 0.6884,
+      "step": 554530
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 1.140009183439198e-05,
+      "loss": 0.7368,
+      "step": 554540
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 1.139894390449223e-05,
+      "loss": 0.635,
+      "step": 554550
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 1.1397795974592483e-05,
+      "loss": 0.612,
+      "step": 554560
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 1.1396648044692736e-05,
+      "loss": 0.6644,
+      "step": 554570
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 1.1395500114792988e-05,
+      "loss": 0.7161,
+      "step": 554580
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 1.1394352184893241e-05,
+      "loss": 0.6927,
+      "step": 554590
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 1.1393204254993494e-05,
+      "loss": 0.6864,
+      "step": 554600
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 1.1392056325093746e-05,
+      "loss": 0.6817,
+      "step": 554610
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 1.1390908395193999e-05,
+      "loss": 0.7426,
+      "step": 554620
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 1.1389760465294252e-05,
+      "loss": 0.6859,
+      "step": 554630
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 1.1388612535394504e-05,
+      "loss": 0.754,
+      "step": 554640
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 1.1387464605494757e-05,
+      "loss": 0.6245,
+      "step": 554650
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 1.138631667559501e-05,
+      "loss": 0.7609,
+      "step": 554660
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 1.1385168745695262e-05,
+      "loss": 0.6663,
+      "step": 554670
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 1.1384020815795515e-05,
+      "loss": 0.6938,
+      "step": 554680
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 1.1382872885895768e-05,
+      "loss": 0.6714,
+      "step": 554690
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 1.1381724955996018e-05,
+      "loss": 0.6806,
+      "step": 554700
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 1.1380577026096271e-05,
+      "loss": 0.7037,
+      "step": 554710
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 1.1379429096196525e-05,
+      "loss": 0.7309,
+      "step": 554720
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 1.1378281166296776e-05,
+      "loss": 0.7258,
+      "step": 554730
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 1.137713323639703e-05,
+      "loss": 0.7551,
+      "step": 554740
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 1.1375985306497283e-05,
+      "loss": 0.6817,
+      "step": 554750
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 1.1374837376597534e-05,
+      "loss": 0.7586,
+      "step": 554760
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 1.1373689446697788e-05,
+      "loss": 0.6677,
+      "step": 554770
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 1.137254151679804e-05,
+      "loss": 0.7227,
+      "step": 554780
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 1.1371393586898292e-05,
+      "loss": 0.7827,
+      "step": 554790
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 1.1370245656998546e-05,
+      "loss": 0.7544,
+      "step": 554800
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 1.1369097727098799e-05,
+      "loss": 0.6814,
+      "step": 554810
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 1.136794979719905e-05,
+      "loss": 0.6937,
+      "step": 554820
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 1.1366801867299304e-05,
+      "loss": 0.7862,
+      "step": 554830
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 1.1365653937399557e-05,
+      "loss": 0.6861,
+      "step": 554840
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 1.1364506007499807e-05,
+      "loss": 0.7729,
+      "step": 554850
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 1.136335807760006e-05,
+      "loss": 0.7331,
+      "step": 554860
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 1.1362210147700311e-05,
+      "loss": 0.756,
+      "step": 554870
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 1.1361062217800565e-05,
+      "loss": 0.7201,
+      "step": 554880
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 1.1359914287900818e-05,
+      "loss": 0.6904,
+      "step": 554890
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 1.135876635800107e-05,
+      "loss": 0.7173,
+      "step": 554900
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 1.1357618428101323e-05,
+      "loss": 0.6965,
+      "step": 554910
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 1.1356470498201576e-05,
+      "loss": 0.7815,
+      "step": 554920
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 1.1355322568301828e-05,
+      "loss": 0.6871,
+      "step": 554930
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 1.135417463840208e-05,
+      "loss": 0.667,
+      "step": 554940
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 1.1353026708502334e-05,
+      "loss": 0.672,
+      "step": 554950
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 1.1351878778602586e-05,
+      "loss": 0.6817,
+      "step": 554960
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 1.1350730848702839e-05,
+      "loss": 0.7167,
+      "step": 554970
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 1.1349582918803092e-05,
+      "loss": 0.8217,
+      "step": 554980
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 1.1348434988903342e-05,
+      "loss": 0.7084,
+      "step": 554990
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 1.1347287059003595e-05,
+      "loss": 0.6911,
+      "step": 555000
+    },
+    {
+      "epoch": 25.46,
+      "learning_rate": 1.1346139129103848e-05,
+      "loss": 0.7494,
+      "step": 555010
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 1.13449911992041e-05,
+      "loss": 0.6756,
+      "step": 555020
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 1.1343843269304353e-05,
+      "loss": 0.7814,
+      "step": 555030
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 1.1342695339404606e-05,
+      "loss": 0.6801,
+      "step": 555040
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 1.1341547409504858e-05,
+      "loss": 0.7117,
+      "step": 555050
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 1.1340399479605111e-05,
+      "loss": 0.7584,
+      "step": 555060
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 1.1339251549705364e-05,
+      "loss": 0.6928,
+      "step": 555070
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 1.1338103619805616e-05,
+      "loss": 0.7928,
+      "step": 555080
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 1.133695568990587e-05,
+      "loss": 0.6796,
+      "step": 555090
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 1.1335807760006123e-05,
+      "loss": 0.8259,
+      "step": 555100
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 1.1334659830106374e-05,
+      "loss": 0.6809,
+      "step": 555110
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 1.1333511900206627e-05,
+      "loss": 0.6491,
+      "step": 555120
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 1.133236397030688e-05,
+      "loss": 0.731,
+      "step": 555130
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 1.133121604040713e-05,
+      "loss": 0.7167,
+      "step": 555140
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 1.1330068110507384e-05,
+      "loss": 0.7219,
+      "step": 555150
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 1.1328920180607637e-05,
+      "loss": 0.7775,
+      "step": 555160
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 1.1327772250707888e-05,
+      "loss": 0.6826,
+      "step": 555170
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 1.1326624320808142e-05,
+      "loss": 0.6807,
+      "step": 555180
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 1.1325476390908395e-05,
+      "loss": 0.7839,
+      "step": 555190
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 1.1324328461008646e-05,
+      "loss": 0.7086,
+      "step": 555200
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 1.13231805311089e-05,
+      "loss": 0.7113,
+      "step": 555210
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 1.1322032601209151e-05,
+      "loss": 0.6566,
+      "step": 555220
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 1.1320884671309404e-05,
+      "loss": 0.6777,
+      "step": 555230
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 1.1319736741409658e-05,
+      "loss": 0.7657,
+      "step": 555240
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 1.131858881150991e-05,
+      "loss": 0.7139,
+      "step": 555250
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 1.1317440881610163e-05,
+      "loss": 0.7769,
+      "step": 555260
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 1.1316292951710416e-05,
+      "loss": 0.7134,
+      "step": 555270
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 1.1315145021810666e-05,
+      "loss": 0.7672,
+      "step": 555280
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 1.1313997091910919e-05,
+      "loss": 0.7252,
+      "step": 555290
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 1.1312849162011172e-05,
+      "loss": 0.6928,
+      "step": 555300
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 1.1311701232111424e-05,
+      "loss": 0.7448,
+      "step": 555310
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 1.1310553302211677e-05,
+      "loss": 0.7557,
+      "step": 555320
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 1.130940537231193e-05,
+      "loss": 0.7761,
+      "step": 555330
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 1.1308257442412182e-05,
+      "loss": 0.6318,
+      "step": 555340
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 1.1307109512512435e-05,
+      "loss": 0.6684,
+      "step": 555350
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 1.1305961582612688e-05,
+      "loss": 0.7426,
+      "step": 555360
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 1.130481365271294e-05,
+      "loss": 0.7825,
+      "step": 555370
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 1.1303665722813193e-05,
+      "loss": 0.7641,
+      "step": 555380
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 1.1302517792913446e-05,
+      "loss": 0.701,
+      "step": 555390
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 1.1301369863013698e-05,
+      "loss": 0.7058,
+      "step": 555400
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 1.1300221933113951e-05,
+      "loss": 0.7421,
+      "step": 555410
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 1.1299074003214204e-05,
+      "loss": 0.7223,
+      "step": 555420
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 1.1297926073314454e-05,
+      "loss": 0.6824,
+      "step": 555430
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 1.1296778143414707e-05,
+      "loss": 0.6815,
+      "step": 555440
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 1.129563021351496e-05,
+      "loss": 0.7861,
+      "step": 555450
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 1.1294482283615212e-05,
+      "loss": 0.7181,
+      "step": 555460
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 1.1293334353715465e-05,
+      "loss": 0.6821,
+      "step": 555470
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 1.1292186423815719e-05,
+      "loss": 0.6998,
+      "step": 555480
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 1.129103849391597e-05,
+      "loss": 0.8303,
+      "step": 555490
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 1.1289890564016223e-05,
+      "loss": 0.6986,
+      "step": 555500
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 1.1288742634116477e-05,
+      "loss": 0.7431,
+      "step": 555510
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 1.1287594704216728e-05,
+      "loss": 0.671,
+      "step": 555520
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 1.1286446774316981e-05,
+      "loss": 0.7219,
+      "step": 555530
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 1.1285298844417233e-05,
+      "loss": 0.7047,
+      "step": 555540
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 1.1284150914517486e-05,
+      "loss": 0.7666,
+      "step": 555550
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 1.128300298461774e-05,
+      "loss": 0.6277,
+      "step": 555560
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 1.128185505471799e-05,
+      "loss": 0.7449,
+      "step": 555570
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 1.1280707124818243e-05,
+      "loss": 0.7234,
+      "step": 555580
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 1.1279559194918496e-05,
+      "loss": 0.7041,
+      "step": 555590
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 1.1278411265018747e-05,
+      "loss": 0.6979,
+      "step": 555600
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 1.1277263335119e-05,
+      "loss": 0.7139,
+      "step": 555610
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 1.1276115405219254e-05,
+      "loss": 0.7811,
+      "step": 555620
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 1.1274967475319505e-05,
+      "loss": 0.7376,
+      "step": 555630
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 1.1273819545419759e-05,
+      "loss": 0.7156,
+      "step": 555640
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 1.1272671615520012e-05,
+      "loss": 0.7914,
+      "step": 555650
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 1.1271523685620263e-05,
+      "loss": 0.6735,
+      "step": 555660
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 1.1270375755720517e-05,
+      "loss": 0.6573,
+      "step": 555670
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 1.126922782582077e-05,
+      "loss": 0.702,
+      "step": 555680
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 1.1268079895921021e-05,
+      "loss": 0.8242,
+      "step": 555690
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 1.1266931966021275e-05,
+      "loss": 0.7125,
+      "step": 555700
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 1.1265784036121528e-05,
+      "loss": 0.7575,
+      "step": 555710
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 1.1264636106221778e-05,
+      "loss": 0.7341,
+      "step": 555720
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 1.1263488176322031e-05,
+      "loss": 0.7048,
+      "step": 555730
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 1.1262340246422284e-05,
+      "loss": 0.7399,
+      "step": 555740
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 1.1261192316522536e-05,
+      "loss": 0.6786,
+      "step": 555750
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 1.1260044386622789e-05,
+      "loss": 0.7517,
+      "step": 555760
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 1.1258896456723042e-05,
+      "loss": 0.7586,
+      "step": 555770
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 1.1257748526823294e-05,
+      "loss": 0.671,
+      "step": 555780
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 1.1256600596923547e-05,
+      "loss": 0.7333,
+      "step": 555790
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 1.12554526670238e-05,
+      "loss": 0.6859,
+      "step": 555800
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 1.1254304737124052e-05,
+      "loss": 0.7174,
+      "step": 555810
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 1.1253156807224305e-05,
+      "loss": 0.6912,
+      "step": 555820
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 1.1252008877324558e-05,
+      "loss": 0.7205,
+      "step": 555830
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 1.125086094742481e-05,
+      "loss": 0.702,
+      "step": 555840
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 1.1249713017525063e-05,
+      "loss": 0.7072,
+      "step": 555850
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 1.1248565087625316e-05,
+      "loss": 0.7064,
+      "step": 555860
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 1.1247417157725566e-05,
+      "loss": 0.8655,
+      "step": 555870
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 1.124626922782582e-05,
+      "loss": 0.6949,
+      "step": 555880
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 1.1245121297926071e-05,
+      "loss": 0.7633,
+      "step": 555890
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 1.1243973368026324e-05,
+      "loss": 0.7576,
+      "step": 555900
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 1.1242825438126578e-05,
+      "loss": 0.7285,
+      "step": 555910
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 1.1241677508226829e-05,
+      "loss": 0.7033,
+      "step": 555920
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 1.1240529578327082e-05,
+      "loss": 0.7078,
+      "step": 555930
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 1.1239381648427336e-05,
+      "loss": 0.707,
+      "step": 555940
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 1.1238233718527587e-05,
+      "loss": 0.6819,
+      "step": 555950
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 1.123708578862784e-05,
+      "loss": 0.6916,
+      "step": 555960
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 1.1235937858728094e-05,
+      "loss": 0.6315,
+      "step": 555970
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 1.1234789928828345e-05,
+      "loss": 0.695,
+      "step": 555980
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 1.1233641998928598e-05,
+      "loss": 0.7848,
+      "step": 555990
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 1.1232494069028852e-05,
+      "loss": 0.7149,
+      "step": 556000
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 1.1231346139129101e-05,
+      "loss": 0.7422,
+      "step": 556010
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 1.1230198209229355e-05,
+      "loss": 0.7,
+      "step": 556020
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 1.1229050279329608e-05,
+      "loss": 0.735,
+      "step": 556030
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 1.122790234942986e-05,
+      "loss": 0.7572,
+      "step": 556040
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 1.1226754419530113e-05,
+      "loss": 0.7708,
+      "step": 556050
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 1.1225606489630366e-05,
+      "loss": 0.6918,
+      "step": 556060
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 1.1224458559730618e-05,
+      "loss": 0.6709,
+      "step": 556070
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 1.122331062983087e-05,
+      "loss": 0.7106,
+      "step": 556080
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 1.1222162699931124e-05,
+      "loss": 0.6583,
+      "step": 556090
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 1.1221014770031376e-05,
+      "loss": 0.6936,
+      "step": 556100
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 1.1219866840131629e-05,
+      "loss": 0.6667,
+      "step": 556110
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 1.1218718910231882e-05,
+      "loss": 0.7087,
+      "step": 556120
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 1.1217570980332134e-05,
+      "loss": 0.7615,
+      "step": 556130
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 1.1216423050432387e-05,
+      "loss": 0.701,
+      "step": 556140
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 1.121527512053264e-05,
+      "loss": 0.7819,
+      "step": 556150
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 1.121412719063289e-05,
+      "loss": 0.7299,
+      "step": 556160
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 1.1212979260733143e-05,
+      "loss": 0.7345,
+      "step": 556170
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 1.1211831330833396e-05,
+      "loss": 0.7165,
+      "step": 556180
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 1.1210683400933648e-05,
+      "loss": 0.7507,
+      "step": 556190
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 1.1209535471033901e-05,
+      "loss": 0.7365,
+      "step": 556200
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 1.1208387541134154e-05,
+      "loss": 0.7419,
+      "step": 556210
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 1.1207239611234406e-05,
+      "loss": 0.7204,
+      "step": 556220
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 1.120609168133466e-05,
+      "loss": 0.7137,
+      "step": 556230
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 1.120494375143491e-05,
+      "loss": 0.7249,
+      "step": 556240
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 1.1203795821535164e-05,
+      "loss": 0.7111,
+      "step": 556250
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 1.1202647891635417e-05,
+      "loss": 0.7006,
+      "step": 556260
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 1.1201499961735669e-05,
+      "loss": 0.7453,
+      "step": 556270
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 1.1200352031835922e-05,
+      "loss": 0.7404,
+      "step": 556280
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 1.1199204101936175e-05,
+      "loss": 0.7235,
+      "step": 556290
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 1.1198056172036425e-05,
+      "loss": 0.7209,
+      "step": 556300
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 1.1196908242136678e-05,
+      "loss": 0.69,
+      "step": 556310
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 1.1195760312236932e-05,
+      "loss": 0.7448,
+      "step": 556320
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 1.1194612382337183e-05,
+      "loss": 0.7885,
+      "step": 556330
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 1.1193464452437436e-05,
+      "loss": 0.7605,
+      "step": 556340
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 1.119231652253769e-05,
+      "loss": 0.8371,
+      "step": 556350
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 1.1191168592637941e-05,
+      "loss": 0.6983,
+      "step": 556360
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 1.1190020662738195e-05,
+      "loss": 0.678,
+      "step": 556370
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 1.1188872732838448e-05,
+      "loss": 0.6864,
+      "step": 556380
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 1.11877248029387e-05,
+      "loss": 0.7567,
+      "step": 556390
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 1.1186576873038953e-05,
+      "loss": 0.6825,
+      "step": 556400
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 1.1185428943139206e-05,
+      "loss": 0.6994,
+      "step": 556410
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 1.1184281013239457e-05,
+      "loss": 0.7728,
+      "step": 556420
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 1.118313308333971e-05,
+      "loss": 0.7603,
+      "step": 556430
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 1.1181985153439964e-05,
+      "loss": 0.7245,
+      "step": 556440
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 1.1180837223540214e-05,
+      "loss": 0.6371,
+      "step": 556450
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 1.1179689293640467e-05,
+      "loss": 0.7522,
+      "step": 556460
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 1.117854136374072e-05,
+      "loss": 0.7291,
+      "step": 556470
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 1.1177393433840972e-05,
+      "loss": 0.6507,
+      "step": 556480
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 1.1176245503941225e-05,
+      "loss": 0.6818,
+      "step": 556490
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 1.1175097574041478e-05,
+      "loss": 0.6553,
+      "step": 556500
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 1.117394964414173e-05,
+      "loss": 0.8297,
+      "step": 556510
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 1.1172801714241983e-05,
+      "loss": 0.6566,
+      "step": 556520
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 1.1171653784342236e-05,
+      "loss": 0.7012,
+      "step": 556530
+    },
+    {
+      "epoch": 25.53,
+      "learning_rate": 1.1170505854442488e-05,
+      "loss": 0.7325,
+      "step": 556540
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 1.1169357924542741e-05,
+      "loss": 0.6669,
+      "step": 556550
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 1.1168209994642994e-05,
+      "loss": 0.7167,
+      "step": 556560
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 1.1167062064743246e-05,
+      "loss": 0.7465,
+      "step": 556570
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 1.1165914134843499e-05,
+      "loss": 0.7884,
+      "step": 556580
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 1.1164766204943749e-05,
+      "loss": 0.7425,
+      "step": 556590
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 1.1163618275044002e-05,
+      "loss": 0.7257,
+      "step": 556600
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 1.1162470345144255e-05,
+      "loss": 0.7534,
+      "step": 556610
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 1.1161322415244507e-05,
+      "loss": 0.6461,
+      "step": 556620
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 1.116017448534476e-05,
+      "loss": 0.7558,
+      "step": 556630
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 1.1159026555445013e-05,
+      "loss": 0.7256,
+      "step": 556640
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 1.1157878625545265e-05,
+      "loss": 0.6655,
+      "step": 556650
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 1.1156730695645518e-05,
+      "loss": 0.7674,
+      "step": 556660
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 1.1155582765745771e-05,
+      "loss": 0.6576,
+      "step": 556670
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 1.1154434835846023e-05,
+      "loss": 0.797,
+      "step": 556680
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 1.1153286905946276e-05,
+      "loss": 0.7492,
+      "step": 556690
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 1.115213897604653e-05,
+      "loss": 0.6599,
+      "step": 556700
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 1.1150991046146781e-05,
+      "loss": 0.6585,
+      "step": 556710
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 1.1149843116247034e-05,
+      "loss": 0.69,
+      "step": 556720
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 1.1148695186347288e-05,
+      "loss": 0.7277,
+      "step": 556730
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 1.1147547256447537e-05,
+      "loss": 0.7631,
+      "step": 556740
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 1.114639932654779e-05,
+      "loss": 0.7261,
+      "step": 556750
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 1.1145251396648044e-05,
+      "loss": 0.718,
+      "step": 556760
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 1.1144103466748295e-05,
+      "loss": 0.6808,
+      "step": 556770
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 1.1142955536848549e-05,
+      "loss": 0.654,
+      "step": 556780
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 1.1141807606948802e-05,
+      "loss": 0.6844,
+      "step": 556790
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 1.1140659677049053e-05,
+      "loss": 0.7213,
+      "step": 556800
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 1.1139511747149307e-05,
+      "loss": 0.7956,
+      "step": 556810
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 1.113836381724956e-05,
+      "loss": 0.6675,
+      "step": 556820
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 1.1137215887349811e-05,
+      "loss": 0.7286,
+      "step": 556830
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 1.1136067957450065e-05,
+      "loss": 0.7134,
+      "step": 556840
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 1.1134920027550318e-05,
+      "loss": 0.6904,
+      "step": 556850
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 1.113377209765057e-05,
+      "loss": 0.7839,
+      "step": 556860
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 1.1132624167750823e-05,
+      "loss": 0.6618,
+      "step": 556870
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 1.1131476237851076e-05,
+      "loss": 0.7061,
+      "step": 556880
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 1.1130328307951326e-05,
+      "loss": 0.7876,
+      "step": 556890
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 1.1129180378051579e-05,
+      "loss": 0.708,
+      "step": 556900
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 1.1128032448151832e-05,
+      "loss": 0.7908,
+      "step": 556910
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 1.1126884518252084e-05,
+      "loss": 0.7174,
+      "step": 556920
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 1.1125736588352337e-05,
+      "loss": 0.7157,
+      "step": 556930
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 1.1124588658452589e-05,
+      "loss": 0.6875,
+      "step": 556940
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 1.1123440728552842e-05,
+      "loss": 0.731,
+      "step": 556950
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 1.1122292798653095e-05,
+      "loss": 0.6483,
+      "step": 556960
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 1.1121144868753347e-05,
+      "loss": 0.8177,
+      "step": 556970
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 1.11199969388536e-05,
+      "loss": 0.839,
+      "step": 556980
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 1.1118849008953853e-05,
+      "loss": 0.6911,
+      "step": 556990
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 1.1117701079054105e-05,
+      "loss": 0.6983,
+      "step": 557000
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 1.1116553149154358e-05,
+      "loss": 0.6502,
+      "step": 557010
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 1.1115405219254611e-05,
+      "loss": 0.694,
+      "step": 557020
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 1.1114257289354861e-05,
+      "loss": 0.7363,
+      "step": 557030
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 1.1113109359455114e-05,
+      "loss": 0.6897,
+      "step": 557040
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 1.1111961429555368e-05,
+      "loss": 0.678,
+      "step": 557050
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 1.1110813499655619e-05,
+      "loss": 0.7787,
+      "step": 557060
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 1.1109665569755872e-05,
+      "loss": 0.7094,
+      "step": 557070
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 1.1108517639856126e-05,
+      "loss": 0.7245,
+      "step": 557080
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 1.1107369709956377e-05,
+      "loss": 0.7194,
+      "step": 557090
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 1.110622178005663e-05,
+      "loss": 0.7271,
+      "step": 557100
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 1.1105073850156884e-05,
+      "loss": 0.6713,
+      "step": 557110
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 1.1103925920257135e-05,
+      "loss": 0.6252,
+      "step": 557120
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 1.1102777990357388e-05,
+      "loss": 0.8147,
+      "step": 557130
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 1.1101630060457642e-05,
+      "loss": 0.7186,
+      "step": 557140
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 1.1100482130557893e-05,
+      "loss": 0.6773,
+      "step": 557150
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 1.1099334200658146e-05,
+      "loss": 0.7869,
+      "step": 557160
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 1.10981862707584e-05,
+      "loss": 0.6902,
+      "step": 557170
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 1.109703834085865e-05,
+      "loss": 0.6912,
+      "step": 557180
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 1.1095890410958903e-05,
+      "loss": 0.7641,
+      "step": 557190
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 1.1094742481059156e-05,
+      "loss": 0.7522,
+      "step": 557200
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 1.1093594551159408e-05,
+      "loss": 0.8058,
+      "step": 557210
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 1.109244662125966e-05,
+      "loss": 0.7066,
+      "step": 557220
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 1.1091298691359914e-05,
+      "loss": 0.6708,
+      "step": 557230
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 1.1090150761460166e-05,
+      "loss": 0.6634,
+      "step": 557240
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 1.1089002831560419e-05,
+      "loss": 0.6666,
+      "step": 557250
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 1.108785490166067e-05,
+      "loss": 0.6584,
+      "step": 557260
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 1.1086706971760924e-05,
+      "loss": 0.7031,
+      "step": 557270
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 1.1085559041861177e-05,
+      "loss": 0.7679,
+      "step": 557280
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 1.1084411111961428e-05,
+      "loss": 0.716,
+      "step": 557290
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 1.1083263182061682e-05,
+      "loss": 0.7042,
+      "step": 557300
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 1.1082115252161935e-05,
+      "loss": 0.7644,
+      "step": 557310
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 1.1080967322262185e-05,
+      "loss": 0.6983,
+      "step": 557320
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 1.1079819392362438e-05,
+      "loss": 0.7433,
+      "step": 557330
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 1.1078671462462691e-05,
+      "loss": 0.6925,
+      "step": 557340
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 1.1077523532562943e-05,
+      "loss": 0.7085,
+      "step": 557350
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 1.1076375602663196e-05,
+      "loss": 0.7428,
+      "step": 557360
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 1.107522767276345e-05,
+      "loss": 0.7372,
+      "step": 557370
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 1.10740797428637e-05,
+      "loss": 0.7744,
+      "step": 557380
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 1.1072931812963954e-05,
+      "loss": 0.7425,
+      "step": 557390
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 1.1071783883064207e-05,
+      "loss": 0.7989,
+      "step": 557400
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 1.1070635953164459e-05,
+      "loss": 0.7098,
+      "step": 557410
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 1.1069488023264712e-05,
+      "loss": 0.7059,
+      "step": 557420
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 1.1068340093364965e-05,
+      "loss": 0.7669,
+      "step": 557430
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 1.1067192163465217e-05,
+      "loss": 0.7193,
+      "step": 557440
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 1.106604423356547e-05,
+      "loss": 0.7615,
+      "step": 557450
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 1.1064896303665723e-05,
+      "loss": 0.7163,
+      "step": 557460
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 1.1063748373765973e-05,
+      "loss": 0.7598,
+      "step": 557470
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 1.1062600443866226e-05,
+      "loss": 0.7034,
+      "step": 557480
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 1.106145251396648e-05,
+      "loss": 0.7249,
+      "step": 557490
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 1.1060304584066731e-05,
+      "loss": 0.7525,
+      "step": 557500
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 1.1059156654166985e-05,
+      "loss": 0.7069,
+      "step": 557510
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 1.1058008724267238e-05,
+      "loss": 0.802,
+      "step": 557520
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 1.105686079436749e-05,
+      "loss": 0.7162,
+      "step": 557530
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 1.1055712864467743e-05,
+      "loss": 0.7321,
+      "step": 557540
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 1.1054564934567996e-05,
+      "loss": 0.7277,
+      "step": 557550
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 1.1053417004668247e-05,
+      "loss": 0.7818,
+      "step": 557560
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 1.10522690747685e-05,
+      "loss": 0.7193,
+      "step": 557570
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 1.1051121144868754e-05,
+      "loss": 0.6873,
+      "step": 557580
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 1.1049973214969005e-05,
+      "loss": 0.7614,
+      "step": 557590
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 1.1048825285069259e-05,
+      "loss": 0.7809,
+      "step": 557600
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 1.1047677355169508e-05,
+      "loss": 0.7549,
+      "step": 557610
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 1.1046529425269762e-05,
+      "loss": 0.6418,
+      "step": 557620
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 1.1045381495370015e-05,
+      "loss": 0.7406,
+      "step": 557630
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 1.1044233565470267e-05,
+      "loss": 0.7293,
+      "step": 557640
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 1.104308563557052e-05,
+      "loss": 0.6599,
+      "step": 557650
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 1.1041937705670773e-05,
+      "loss": 0.6534,
+      "step": 557660
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 1.1040789775771025e-05,
+      "loss": 0.7054,
+      "step": 557670
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 1.1039641845871278e-05,
+      "loss": 0.7217,
+      "step": 557680
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 1.1038493915971531e-05,
+      "loss": 0.7715,
+      "step": 557690
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 1.1037345986071783e-05,
+      "loss": 0.7077,
+      "step": 557700
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 1.1036198056172036e-05,
+      "loss": 0.7843,
+      "step": 557710
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 1.1035050126272289e-05,
+      "loss": 0.6676,
+      "step": 557720
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 1.103390219637254e-05,
+      "loss": 0.6892,
+      "step": 557730
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 1.1032754266472794e-05,
+      "loss": 0.7214,
+      "step": 557740
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 1.1031606336573047e-05,
+      "loss": 0.686,
+      "step": 557750
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 1.1030458406673297e-05,
+      "loss": 0.6661,
+      "step": 557760
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 1.102931047677355e-05,
+      "loss": 0.6944,
+      "step": 557770
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 1.1028162546873803e-05,
+      "loss": 0.7713,
+      "step": 557780
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 1.1027014616974055e-05,
+      "loss": 0.6882,
+      "step": 557790
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 1.1025866687074308e-05,
+      "loss": 0.7536,
+      "step": 557800
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 1.1024718757174561e-05,
+      "loss": 0.6644,
+      "step": 557810
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 1.1023570827274813e-05,
+      "loss": 0.703,
+      "step": 557820
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 1.1022422897375066e-05,
+      "loss": 0.6746,
+      "step": 557830
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 1.102127496747532e-05,
+      "loss": 0.7589,
+      "step": 557840
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 1.1020127037575571e-05,
+      "loss": 0.7219,
+      "step": 557850
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 1.1018979107675824e-05,
+      "loss": 0.676,
+      "step": 557860
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 1.1017831177776078e-05,
+      "loss": 0.7533,
+      "step": 557870
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 1.1016683247876329e-05,
+      "loss": 0.7766,
+      "step": 557880
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 1.1015535317976582e-05,
+      "loss": 0.7282,
+      "step": 557890
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 1.1014387388076836e-05,
+      "loss": 0.7854,
+      "step": 557900
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 1.1013239458177085e-05,
+      "loss": 0.7418,
+      "step": 557910
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 1.1012091528277339e-05,
+      "loss": 0.7266,
+      "step": 557920
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 1.1010943598377592e-05,
+      "loss": 0.7933,
+      "step": 557930
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 1.1009795668477843e-05,
+      "loss": 0.8047,
+      "step": 557940
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 1.1008647738578097e-05,
+      "loss": 0.7613,
+      "step": 557950
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 1.1007499808678348e-05,
+      "loss": 0.744,
+      "step": 557960
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 1.1006351878778601e-05,
+      "loss": 0.7353,
+      "step": 557970
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 1.1005203948878855e-05,
+      "loss": 0.7271,
+      "step": 557980
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 1.1004056018979106e-05,
+      "loss": 0.6938,
+      "step": 557990
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 1.100290808907936e-05,
+      "loss": 0.688,
+      "step": 558000
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 1.1001760159179613e-05,
+      "loss": 0.7609,
+      "step": 558010
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 1.1000612229279863e-05,
+      "loss": 0.7514,
+      "step": 558020
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 1.0999464299380118e-05,
+      "loss": 0.6371,
+      "step": 558030
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 1.099831636948037e-05,
+      "loss": 0.6509,
+      "step": 558040
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 1.099716843958062e-05,
+      "loss": 0.6208,
+      "step": 558050
+    },
+    {
+      "epoch": 25.6,
+      "learning_rate": 1.0996020509680874e-05,
+      "loss": 0.7919,
+      "step": 558060
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 1.0994872579781127e-05,
+      "loss": 0.7732,
+      "step": 558070
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 1.0993724649881379e-05,
+      "loss": 0.7585,
+      "step": 558080
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 1.0992576719981632e-05,
+      "loss": 0.7676,
+      "step": 558090
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 1.0991428790081885e-05,
+      "loss": 0.6342,
+      "step": 558100
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 1.0990280860182137e-05,
+      "loss": 0.6955,
+      "step": 558110
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 1.098913293028239e-05,
+      "loss": 0.7354,
+      "step": 558120
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 1.0987985000382643e-05,
+      "loss": 0.6846,
+      "step": 558130
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 1.0986837070482895e-05,
+      "loss": 0.7389,
+      "step": 558140
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 1.0985689140583148e-05,
+      "loss": 0.7032,
+      "step": 558150
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 1.0984541210683401e-05,
+      "loss": 0.7738,
+      "step": 558160
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 1.0983393280783653e-05,
+      "loss": 0.7478,
+      "step": 558170
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 1.0982245350883906e-05,
+      "loss": 0.7431,
+      "step": 558180
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 1.098109742098416e-05,
+      "loss": 0.6513,
+      "step": 558190
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 1.0979949491084409e-05,
+      "loss": 0.7771,
+      "step": 558200
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 1.0978801561184662e-05,
+      "loss": 0.6587,
+      "step": 558210
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 1.0977653631284916e-05,
+      "loss": 0.7757,
+      "step": 558220
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 1.0976505701385167e-05,
+      "loss": 0.7716,
+      "step": 558230
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 1.097535777148542e-05,
+      "loss": 0.7354,
+      "step": 558240
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 1.0974209841585674e-05,
+      "loss": 0.7311,
+      "step": 558250
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 1.0973061911685925e-05,
+      "loss": 0.7439,
+      "step": 558260
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 1.0971913981786178e-05,
+      "loss": 0.6994,
+      "step": 558270
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 1.0970766051886432e-05,
+      "loss": 0.7827,
+      "step": 558280
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 1.0969618121986683e-05,
+      "loss": 0.7616,
+      "step": 558290
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 1.0968470192086936e-05,
+      "loss": 0.6961,
+      "step": 558300
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 1.0967322262187186e-05,
+      "loss": 0.7111,
+      "step": 558310
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 1.0966174332287441e-05,
+      "loss": 0.7517,
+      "step": 558320
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 1.0965026402387694e-05,
+      "loss": 0.7751,
+      "step": 558330
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 1.0963878472487944e-05,
+      "loss": 0.748,
+      "step": 558340
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 1.0962730542588198e-05,
+      "loss": 0.7678,
+      "step": 558350
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 1.096158261268845e-05,
+      "loss": 0.7613,
+      "step": 558360
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 1.0960434682788702e-05,
+      "loss": 0.7568,
+      "step": 558370
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 1.0959286752888956e-05,
+      "loss": 0.717,
+      "step": 558380
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 1.0958138822989209e-05,
+      "loss": 0.7041,
+      "step": 558390
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 1.095699089308946e-05,
+      "loss": 0.6396,
+      "step": 558400
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 1.0955842963189714e-05,
+      "loss": 0.6609,
+      "step": 558410
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 1.0954695033289967e-05,
+      "loss": 0.7256,
+      "step": 558420
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 1.0953547103390218e-05,
+      "loss": 0.6087,
+      "step": 558430
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 1.0952399173490472e-05,
+      "loss": 0.7056,
+      "step": 558440
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 1.0951251243590725e-05,
+      "loss": 0.7925,
+      "step": 558450
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 1.0950103313690975e-05,
+      "loss": 0.7018,
+      "step": 558460
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 1.094895538379123e-05,
+      "loss": 0.6618,
+      "step": 558470
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 1.0947807453891483e-05,
+      "loss": 0.7041,
+      "step": 558480
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 1.0946659523991733e-05,
+      "loss": 0.7485,
+      "step": 558490
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 1.0945511594091986e-05,
+      "loss": 0.7611,
+      "step": 558500
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 1.094436366419224e-05,
+      "loss": 0.7162,
+      "step": 558510
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 1.0943215734292491e-05,
+      "loss": 0.6512,
+      "step": 558520
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 1.0942067804392744e-05,
+      "loss": 0.7368,
+      "step": 558530
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 1.0940919874492997e-05,
+      "loss": 0.816,
+      "step": 558540
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 1.0939771944593249e-05,
+      "loss": 0.7174,
+      "step": 558550
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 1.0938624014693502e-05,
+      "loss": 0.6877,
+      "step": 558560
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 1.0937476084793755e-05,
+      "loss": 0.7282,
+      "step": 558570
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 1.0936328154894007e-05,
+      "loss": 0.7237,
+      "step": 558580
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 1.093518022499426e-05,
+      "loss": 0.7364,
+      "step": 558590
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 1.0934032295094513e-05,
+      "loss": 0.7421,
+      "step": 558600
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 1.0932884365194763e-05,
+      "loss": 0.7118,
+      "step": 558610
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 1.0931736435295018e-05,
+      "loss": 0.7682,
+      "step": 558620
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 1.0930588505395268e-05,
+      "loss": 0.7731,
+      "step": 558630
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 1.0929440575495521e-05,
+      "loss": 0.6731,
+      "step": 558640
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 1.0928292645595775e-05,
+      "loss": 0.7266,
+      "step": 558650
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 1.0927144715696026e-05,
+      "loss": 0.6954,
+      "step": 558660
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 1.092599678579628e-05,
+      "loss": 0.7607,
+      "step": 558670
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 1.0924848855896533e-05,
+      "loss": 0.7546,
+      "step": 558680
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 1.0923700925996784e-05,
+      "loss": 0.6112,
+      "step": 558690
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 1.0922552996097037e-05,
+      "loss": 0.6871,
+      "step": 558700
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 1.092140506619729e-05,
+      "loss": 0.7872,
+      "step": 558710
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 1.0920257136297542e-05,
+      "loss": 0.7251,
+      "step": 558720
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 1.0919109206397795e-05,
+      "loss": 0.6884,
+      "step": 558730
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 1.0917961276498049e-05,
+      "loss": 0.74,
+      "step": 558740
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 1.0916813346598298e-05,
+      "loss": 0.687,
+      "step": 558750
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 1.0915665416698552e-05,
+      "loss": 0.7612,
+      "step": 558760
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 1.0914517486798807e-05,
+      "loss": 0.6922,
+      "step": 558770
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 1.0913369556899057e-05,
+      "loss": 0.6579,
+      "step": 558780
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 1.091222162699931e-05,
+      "loss": 0.6602,
+      "step": 558790
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 1.0911073697099563e-05,
+      "loss": 0.73,
+      "step": 558800
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 1.0909925767199815e-05,
+      "loss": 0.7256,
+      "step": 558810
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 1.0908777837300068e-05,
+      "loss": 0.7235,
+      "step": 558820
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 1.0907629907400321e-05,
+      "loss": 0.7065,
+      "step": 558830
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 1.0906481977500573e-05,
+      "loss": 0.7041,
+      "step": 558840
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 1.0905334047600826e-05,
+      "loss": 0.7005,
+      "step": 558850
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 1.0904186117701079e-05,
+      "loss": 0.7727,
+      "step": 558860
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 1.090303818780133e-05,
+      "loss": 0.7778,
+      "step": 558870
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 1.0901890257901584e-05,
+      "loss": 0.6748,
+      "step": 558880
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 1.0900742328001837e-05,
+      "loss": 0.7455,
+      "step": 558890
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 1.0899594398102087e-05,
+      "loss": 0.669,
+      "step": 558900
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 1.089844646820234e-05,
+      "loss": 0.6774,
+      "step": 558910
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 1.0897298538302595e-05,
+      "loss": 0.6943,
+      "step": 558920
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 1.0896150608402845e-05,
+      "loss": 0.7437,
+      "step": 558930
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 1.0895002678503098e-05,
+      "loss": 0.7248,
+      "step": 558940
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 1.0893854748603351e-05,
+      "loss": 0.7277,
+      "step": 558950
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 1.0892706818703603e-05,
+      "loss": 0.7906,
+      "step": 558960
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 1.0891558888803856e-05,
+      "loss": 0.7845,
+      "step": 558970
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 1.0890410958904108e-05,
+      "loss": 0.6995,
+      "step": 558980
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 1.0889263029004361e-05,
+      "loss": 0.6893,
+      "step": 558990
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 1.0888115099104614e-05,
+      "loss": 0.7254,
+      "step": 559000
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 1.0886967169204866e-05,
+      "loss": 0.6904,
+      "step": 559010
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 1.0885819239305119e-05,
+      "loss": 0.6681,
+      "step": 559020
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 1.0884671309405372e-05,
+      "loss": 0.7427,
+      "step": 559030
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 1.0883523379505622e-05,
+      "loss": 0.816,
+      "step": 559040
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 1.0882375449605875e-05,
+      "loss": 0.7471,
+      "step": 559050
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 1.0881227519706129e-05,
+      "loss": 0.6907,
+      "step": 559060
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 1.088007958980638e-05,
+      "loss": 0.7611,
+      "step": 559070
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 1.0878931659906633e-05,
+      "loss": 0.736,
+      "step": 559080
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 1.0877783730006887e-05,
+      "loss": 0.6921,
+      "step": 559090
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 1.0876635800107138e-05,
+      "loss": 0.7496,
+      "step": 559100
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 1.0875487870207392e-05,
+      "loss": 0.7384,
+      "step": 559110
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 1.0874339940307645e-05,
+      "loss": 0.6715,
+      "step": 559120
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 1.0873192010407896e-05,
+      "loss": 0.7424,
+      "step": 559130
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 1.087204408050815e-05,
+      "loss": 0.7148,
+      "step": 559140
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 1.0870896150608403e-05,
+      "loss": 0.746,
+      "step": 559150
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 1.0869748220708654e-05,
+      "loss": 0.6925,
+      "step": 559160
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 1.0868600290808908e-05,
+      "loss": 0.7471,
+      "step": 559170
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 1.086745236090916e-05,
+      "loss": 0.758,
+      "step": 559180
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 1.086630443100941e-05,
+      "loss": 0.7547,
+      "step": 559190
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 1.0865156501109664e-05,
+      "loss": 0.6699,
+      "step": 559200
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 1.0864008571209917e-05,
+      "loss": 0.6934,
+      "step": 559210
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 1.0862860641310169e-05,
+      "loss": 0.7381,
+      "step": 559220
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 1.0861712711410422e-05,
+      "loss": 0.7539,
+      "step": 559230
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 1.0860564781510675e-05,
+      "loss": 0.6797,
+      "step": 559240
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 1.0859416851610927e-05,
+      "loss": 0.8174,
+      "step": 559250
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 1.085826892171118e-05,
+      "loss": 0.6602,
+      "step": 559260
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 1.0857120991811433e-05,
+      "loss": 0.7676,
+      "step": 559270
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 1.0855973061911685e-05,
+      "loss": 0.7019,
+      "step": 559280
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 1.0854825132011938e-05,
+      "loss": 0.7399,
+      "step": 559290
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 1.0853677202112191e-05,
+      "loss": 0.7511,
+      "step": 559300
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 1.0852529272212443e-05,
+      "loss": 0.7223,
+      "step": 559310
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 1.0851381342312696e-05,
+      "loss": 0.7573,
+      "step": 559320
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 1.0850233412412946e-05,
+      "loss": 0.8257,
+      "step": 559330
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 1.0849085482513199e-05,
+      "loss": 0.7735,
+      "step": 559340
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 1.0847937552613452e-05,
+      "loss": 0.6397,
+      "step": 559350
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 1.0846789622713704e-05,
+      "loss": 0.7533,
+      "step": 559360
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 1.0845641692813957e-05,
+      "loss": 0.7284,
+      "step": 559370
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 1.084449376291421e-05,
+      "loss": 0.6968,
+      "step": 559380
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 1.0843345833014462e-05,
+      "loss": 0.7188,
+      "step": 559390
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 1.0842197903114715e-05,
+      "loss": 0.695,
+      "step": 559400
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 1.0841049973214968e-05,
+      "loss": 0.7486,
+      "step": 559410
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 1.083990204331522e-05,
+      "loss": 0.7642,
+      "step": 559420
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 1.0838754113415473e-05,
+      "loss": 0.6654,
+      "step": 559430
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 1.0837606183515726e-05,
+      "loss": 0.7377,
+      "step": 559440
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 1.0836458253615978e-05,
+      "loss": 0.7028,
+      "step": 559450
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 1.0835310323716231e-05,
+      "loss": 0.7614,
+      "step": 559460
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 1.0834162393816485e-05,
+      "loss": 0.7659,
+      "step": 559470
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 1.0833014463916734e-05,
+      "loss": 0.7164,
+      "step": 559480
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 1.0831866534016988e-05,
+      "loss": 0.7008,
+      "step": 559490
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 1.0830718604117241e-05,
+      "loss": 0.7503,
+      "step": 559500
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 1.0829570674217492e-05,
+      "loss": 0.7447,
+      "step": 559510
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 1.0828422744317746e-05,
+      "loss": 0.6951,
+      "step": 559520
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 1.0827274814417999e-05,
+      "loss": 0.7152,
+      "step": 559530
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 1.082612688451825e-05,
+      "loss": 0.7358,
+      "step": 559540
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 1.0824978954618504e-05,
+      "loss": 0.7494,
+      "step": 559550
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 1.0823831024718757e-05,
+      "loss": 0.6958,
+      "step": 559560
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 1.0822683094819008e-05,
+      "loss": 0.678,
+      "step": 559570
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 1.0821535164919262e-05,
+      "loss": 0.7141,
+      "step": 559580
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 1.0820387235019515e-05,
+      "loss": 0.7444,
+      "step": 559590
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 1.0819239305119767e-05,
+      "loss": 0.753,
+      "step": 559600
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 1.081809137522002e-05,
+      "loss": 0.7268,
+      "step": 559610
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 1.0816943445320273e-05,
+      "loss": 0.6819,
+      "step": 559620
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 1.0815795515420523e-05,
+      "loss": 0.7709,
+      "step": 559630
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 1.0814647585520776e-05,
+      "loss": 0.7275,
+      "step": 559640
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 1.081349965562103e-05,
+      "loss": 0.7273,
+      "step": 559650
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 1.0812351725721281e-05,
+      "loss": 0.7761,
+      "step": 559660
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 1.0811203795821534e-05,
+      "loss": 0.694,
+      "step": 559670
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 1.0810055865921786e-05,
+      "loss": 0.7074,
+      "step": 559680
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 1.0808907936022039e-05,
+      "loss": 0.7164,
+      "step": 559690
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 1.0807760006122292e-05,
+      "loss": 0.748,
+      "step": 559700
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 1.0806612076222544e-05,
+      "loss": 0.6696,
+      "step": 559710
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 1.0805464146322797e-05,
+      "loss": 0.7577,
+      "step": 559720
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 1.080431621642305e-05,
+      "loss": 0.7358,
+      "step": 559730
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 1.0803168286523302e-05,
+      "loss": 0.7521,
+      "step": 559740
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 1.0802020356623555e-05,
+      "loss": 0.7437,
+      "step": 559750
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 1.0800872426723808e-05,
+      "loss": 0.7119,
+      "step": 559760
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 1.0799724496824058e-05,
+      "loss": 0.7465,
+      "step": 559770
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 1.0798576566924311e-05,
+      "loss": 0.7699,
+      "step": 559780
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 1.0797428637024565e-05,
+      "loss": 0.8309,
+      "step": 559790
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 1.0796280707124816e-05,
+      "loss": 0.696,
+      "step": 559800
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 1.079513277722507e-05,
+      "loss": 0.7486,
+      "step": 559810
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 1.0793984847325323e-05,
+      "loss": 0.7533,
+      "step": 559820
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 1.0792836917425574e-05,
+      "loss": 0.6792,
+      "step": 559830
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 1.0791688987525827e-05,
+      "loss": 0.6748,
+      "step": 559840
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 1.079054105762608e-05,
+      "loss": 0.7273,
+      "step": 559850
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 1.0789393127726332e-05,
+      "loss": 0.7138,
+      "step": 559860
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 1.0788245197826585e-05,
+      "loss": 0.7059,
+      "step": 559870
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 1.0787097267926839e-05,
+      "loss": 0.8239,
+      "step": 559880
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 1.078594933802709e-05,
+      "loss": 0.7245,
+      "step": 559890
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 1.0784801408127343e-05,
+      "loss": 0.6814,
+      "step": 559900
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 1.0783653478227597e-05,
+      "loss": 0.7616,
+      "step": 559910
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 1.0782505548327847e-05,
+      "loss": 0.7219,
+      "step": 559920
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 1.07813576184281e-05,
+      "loss": 0.7464,
+      "step": 559930
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 1.0780209688528353e-05,
+      "loss": 0.7536,
+      "step": 559940
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 1.0779061758628605e-05,
+      "loss": 0.7332,
+      "step": 559950
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 1.0777913828728858e-05,
+      "loss": 0.736,
+      "step": 559960
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 1.0776765898829111e-05,
+      "loss": 0.7213,
+      "step": 559970
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 1.0775617968929363e-05,
+      "loss": 0.6854,
+      "step": 559980
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 1.0774470039029616e-05,
+      "loss": 0.7629,
+      "step": 559990
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 1.0773322109129867e-05,
+      "loss": 0.7366,
+      "step": 560000
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 1.077217417923012e-05,
+      "loss": 0.7654,
+      "step": 560010
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 1.0771026249330374e-05,
+      "loss": 0.7118,
+      "step": 560020
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 1.0769878319430625e-05,
+      "loss": 0.7054,
+      "step": 560030
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 1.0768730389530879e-05,
+      "loss": 0.7133,
+      "step": 560040
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 1.0767582459631132e-05,
+      "loss": 0.663,
+      "step": 560050
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 1.0766434529731382e-05,
+      "loss": 0.7521,
+      "step": 560060
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 1.0765286599831635e-05,
+      "loss": 0.8179,
+      "step": 560070
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 1.0764138669931888e-05,
+      "loss": 0.7365,
+      "step": 560080
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 1.076299074003214e-05,
+      "loss": 0.6697,
+      "step": 560090
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 1.0761842810132393e-05,
+      "loss": 0.7802,
+      "step": 560100
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 1.0760694880232646e-05,
+      "loss": 0.7206,
+      "step": 560110
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 1.0759546950332898e-05,
+      "loss": 0.65,
+      "step": 560120
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 1.0758399020433151e-05,
+      "loss": 0.7541,
+      "step": 560130
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 1.0757251090533404e-05,
+      "loss": 0.7176,
+      "step": 560140
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 1.0756103160633656e-05,
+      "loss": 0.7207,
+      "step": 560150
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 1.0754955230733909e-05,
+      "loss": 0.6957,
+      "step": 560160
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 1.0753807300834162e-05,
+      "loss": 0.7747,
+      "step": 560170
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 1.0752659370934414e-05,
+      "loss": 0.7985,
+      "step": 560180
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 1.0751511441034667e-05,
+      "loss": 0.697,
+      "step": 560190
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 1.075036351113492e-05,
+      "loss": 0.7232,
+      "step": 560200
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 1.074921558123517e-05,
+      "loss": 0.7137,
+      "step": 560210
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 1.0748067651335423e-05,
+      "loss": 0.7517,
+      "step": 560220
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 1.0746919721435677e-05,
+      "loss": 0.749,
+      "step": 560230
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 1.0745771791535928e-05,
+      "loss": 0.757,
+      "step": 560240
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 1.0744623861636182e-05,
+      "loss": 0.7469,
+      "step": 560250
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 1.0743475931736435e-05,
+      "loss": 0.8307,
+      "step": 560260
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 1.0742328001836686e-05,
+      "loss": 0.6892,
+      "step": 560270
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 1.074118007193694e-05,
+      "loss": 0.7814,
+      "step": 560280
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 1.0740032142037193e-05,
+      "loss": 0.7067,
+      "step": 560290
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 1.0738884212137444e-05,
+      "loss": 0.6944,
+      "step": 560300
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 1.0737736282237698e-05,
+      "loss": 0.7428,
+      "step": 560310
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 1.073658835233795e-05,
+      "loss": 0.7117,
+      "step": 560320
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 1.0735440422438202e-05,
+      "loss": 0.6654,
+      "step": 560330
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 1.0734292492538456e-05,
+      "loss": 0.6842,
+      "step": 560340
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 1.0733144562638705e-05,
+      "loss": 0.7185,
+      "step": 560350
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 1.0731996632738959e-05,
+      "loss": 0.7442,
+      "step": 560360
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 1.0730848702839212e-05,
+      "loss": 0.7755,
+      "step": 560370
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 1.0729700772939464e-05,
+      "loss": 0.6538,
+      "step": 560380
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 1.0728552843039717e-05,
+      "loss": 0.7753,
+      "step": 560390
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 1.072740491313997e-05,
+      "loss": 0.7091,
+      "step": 560400
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 1.0726256983240222e-05,
+      "loss": 0.7286,
+      "step": 560410
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 1.0725109053340475e-05,
+      "loss": 0.7086,
+      "step": 560420
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 1.0723961123440728e-05,
+      "loss": 0.73,
+      "step": 560430
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 1.072281319354098e-05,
+      "loss": 0.6812,
+      "step": 560440
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 1.0721665263641233e-05,
+      "loss": 0.7125,
+      "step": 560450
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 1.0720517333741486e-05,
+      "loss": 0.7698,
+      "step": 560460
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 1.0719369403841738e-05,
+      "loss": 0.7211,
+      "step": 560470
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 1.0718221473941991e-05,
+      "loss": 0.7545,
+      "step": 560480
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 1.0717073544042244e-05,
+      "loss": 0.7394,
+      "step": 560490
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 1.0715925614142494e-05,
+      "loss": 0.7222,
+      "step": 560500
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 1.0714777684242747e-05,
+      "loss": 0.7521,
+      "step": 560510
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 1.0713629754343e-05,
+      "loss": 0.6885,
+      "step": 560520
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 1.0712481824443252e-05,
+      "loss": 0.672,
+      "step": 560530
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 1.0711333894543505e-05,
+      "loss": 0.8,
+      "step": 560540
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 1.0710185964643758e-05,
+      "loss": 0.7262,
+      "step": 560550
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 1.070903803474401e-05,
+      "loss": 0.6759,
+      "step": 560560
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 1.0707890104844263e-05,
+      "loss": 0.7076,
+      "step": 560570
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 1.0706742174944516e-05,
+      "loss": 0.708,
+      "step": 560580
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 1.0705594245044768e-05,
+      "loss": 0.7957,
+      "step": 560590
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 1.0704446315145021e-05,
+      "loss": 0.725,
+      "step": 560600
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 1.0703298385245275e-05,
+      "loss": 0.7323,
+      "step": 560610
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 1.0702150455345526e-05,
+      "loss": 0.7936,
+      "step": 560620
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 1.070100252544578e-05,
+      "loss": 0.7799,
+      "step": 560630
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 1.0699854595546033e-05,
+      "loss": 0.6684,
+      "step": 560640
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 1.0698706665646282e-05,
+      "loss": 0.7894,
+      "step": 560650
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 1.0697558735746536e-05,
+      "loss": 0.739,
+      "step": 560660
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 1.0696410805846789e-05,
+      "loss": 0.6899,
+      "step": 560670
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 1.069526287594704e-05,
+      "loss": 0.7307,
+      "step": 560680
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 1.0694114946047294e-05,
+      "loss": 0.7399,
+      "step": 560690
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 1.0692967016147545e-05,
+      "loss": 0.7303,
+      "step": 560700
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 1.0691819086247798e-05,
+      "loss": 0.7111,
+      "step": 560710
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 1.0690671156348052e-05,
+      "loss": 0.7367,
+      "step": 560720
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 1.0689523226448303e-05,
+      "loss": 0.7725,
+      "step": 560730
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 1.0688375296548557e-05,
+      "loss": 0.6856,
+      "step": 560740
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 1.068722736664881e-05,
+      "loss": 0.72,
+      "step": 560750
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 1.0686079436749061e-05,
+      "loss": 0.7191,
+      "step": 560760
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 1.0684931506849315e-05,
+      "loss": 0.7605,
+      "step": 560770
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 1.0683783576949568e-05,
+      "loss": 0.7373,
+      "step": 560780
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 1.0682635647049818e-05,
+      "loss": 0.7664,
+      "step": 560790
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 1.0681487717150071e-05,
+      "loss": 0.6482,
+      "step": 560800
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 1.0680339787250324e-05,
+      "loss": 0.7262,
+      "step": 560810
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 1.0679191857350576e-05,
+      "loss": 0.7528,
+      "step": 560820
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 1.0678043927450829e-05,
+      "loss": 0.7303,
+      "step": 560830
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 1.0676895997551082e-05,
+      "loss": 0.7722,
+      "step": 560840
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 1.0675748067651334e-05,
+      "loss": 0.856,
+      "step": 560850
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 1.0674600137751587e-05,
+      "loss": 0.8196,
+      "step": 560860
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 1.067345220785184e-05,
+      "loss": 0.7018,
+      "step": 560870
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 1.0672304277952092e-05,
+      "loss": 0.7466,
+      "step": 560880
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 1.0671156348052345e-05,
+      "loss": 0.6936,
+      "step": 560890
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 1.0670008418152598e-05,
+      "loss": 0.676,
+      "step": 560900
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 1.066886048825285e-05,
+      "loss": 0.7376,
+      "step": 560910
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 1.0667712558353103e-05,
+      "loss": 0.6936,
+      "step": 560920
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 1.0666564628453356e-05,
+      "loss": 0.7434,
+      "step": 560930
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 1.0665416698553606e-05,
+      "loss": 0.7396,
+      "step": 560940
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 1.066426876865386e-05,
+      "loss": 0.7439,
+      "step": 560950
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 1.0663120838754113e-05,
+      "loss": 0.7608,
+      "step": 560960
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 1.0661972908854364e-05,
+      "loss": 0.7417,
+      "step": 560970
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 1.0660824978954617e-05,
+      "loss": 0.7376,
+      "step": 560980
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 1.065967704905487e-05,
+      "loss": 0.685,
+      "step": 560990
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 1.0658529119155122e-05,
+      "loss": 0.6129,
+      "step": 561000
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 1.0657381189255375e-05,
+      "loss": 0.6968,
+      "step": 561010
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 1.0656233259355629e-05,
+      "loss": 0.6878,
+      "step": 561020
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 1.065508532945588e-05,
+      "loss": 0.7112,
+      "step": 561030
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 1.0653937399556133e-05,
+      "loss": 0.6767,
+      "step": 561040
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 1.0652789469656385e-05,
+      "loss": 0.6681,
+      "step": 561050
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 1.0651641539756638e-05,
+      "loss": 0.7773,
+      "step": 561060
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 1.0650493609856891e-05,
+      "loss": 0.7051,
+      "step": 561070
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 1.0649345679957141e-05,
+      "loss": 0.7843,
+      "step": 561080
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 1.0648197750057395e-05,
+      "loss": 0.7123,
+      "step": 561090
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 1.0647049820157648e-05,
+      "loss": 0.6691,
+      "step": 561100
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 1.06459018902579e-05,
+      "loss": 0.7784,
+      "step": 561110
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 1.0644753960358153e-05,
+      "loss": 0.7686,
+      "step": 561120
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 1.0643606030458406e-05,
+      "loss": 0.7222,
+      "step": 561130
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 1.0642458100558657e-05,
+      "loss": 0.7976,
+      "step": 561140
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 1.064131017065891e-05,
+      "loss": 0.665,
+      "step": 561150
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 1.0640162240759164e-05,
+      "loss": 0.7167,
+      "step": 561160
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 1.0639014310859415e-05,
+      "loss": 0.7198,
+      "step": 561170
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 1.0637866380959669e-05,
+      "loss": 0.6382,
+      "step": 561180
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 1.0636718451059922e-05,
+      "loss": 0.7288,
+      "step": 561190
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 1.0635570521160173e-05,
+      "loss": 0.8002,
+      "step": 561200
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 1.0634422591260427e-05,
+      "loss": 0.6897,
+      "step": 561210
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 1.063327466136068e-05,
+      "loss": 0.6804,
+      "step": 561220
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 1.063212673146093e-05,
+      "loss": 0.6219,
+      "step": 561230
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 1.0630978801561183e-05,
+      "loss": 0.6822,
+      "step": 561240
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 1.0629830871661436e-05,
+      "loss": 0.6916,
+      "step": 561250
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 1.0628682941761688e-05,
+      "loss": 0.6891,
+      "step": 561260
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 1.0627535011861941e-05,
+      "loss": 0.6891,
+      "step": 561270
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 1.0626387081962194e-05,
+      "loss": 0.7513,
+      "step": 561280
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 1.0625239152062446e-05,
+      "loss": 0.7084,
+      "step": 561290
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 1.0624091222162699e-05,
+      "loss": 0.6487,
+      "step": 561300
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 1.0622943292262952e-05,
+      "loss": 0.7477,
+      "step": 561310
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 1.0621795362363204e-05,
+      "loss": 0.6791,
+      "step": 561320
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 1.0620647432463457e-05,
+      "loss": 0.6781,
+      "step": 561330
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 1.061949950256371e-05,
+      "loss": 0.7539,
+      "step": 561340
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 1.0618351572663962e-05,
+      "loss": 0.668,
+      "step": 561350
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 1.0617203642764215e-05,
+      "loss": 0.7253,
+      "step": 561360
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 1.0616055712864468e-05,
+      "loss": 0.6421,
+      "step": 561370
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 1.0614907782964718e-05,
+      "loss": 0.8295,
+      "step": 561380
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 1.0613759853064972e-05,
+      "loss": 0.7322,
+      "step": 561390
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 1.0612611923165223e-05,
+      "loss": 0.7033,
+      "step": 561400
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 1.0611463993265476e-05,
+      "loss": 0.7102,
+      "step": 561410
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 1.061031606336573e-05,
+      "loss": 0.7375,
+      "step": 561420
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 1.0609168133465981e-05,
+      "loss": 0.7735,
+      "step": 561430
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 1.0608020203566234e-05,
+      "loss": 0.7449,
+      "step": 561440
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 1.0606872273666488e-05,
+      "loss": 0.7806,
+      "step": 561450
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 1.0605724343766739e-05,
+      "loss": 0.7581,
+      "step": 561460
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 1.0604576413866992e-05,
+      "loss": 0.7139,
+      "step": 561470
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 1.0603428483967246e-05,
+      "loss": 0.7681,
+      "step": 561480
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 1.0602280554067497e-05,
+      "loss": 0.6682,
+      "step": 561490
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 1.060113262416775e-05,
+      "loss": 0.7712,
+      "step": 561500
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 1.0599984694268004e-05,
+      "loss": 0.6982,
+      "step": 561510
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 1.0598836764368254e-05,
+      "loss": 0.689,
+      "step": 561520
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 1.0597688834468507e-05,
+      "loss": 0.7089,
+      "step": 561530
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 1.059654090456876e-05,
+      "loss": 0.7838,
+      "step": 561540
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 1.0595392974669012e-05,
+      "loss": 0.7533,
+      "step": 561550
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 1.0594245044769265e-05,
+      "loss": 0.7362,
+      "step": 561560
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 1.0593097114869518e-05,
+      "loss": 0.7571,
+      "step": 561570
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 1.059194918496977e-05,
+      "loss": 0.7306,
+      "step": 561580
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 1.0590801255070023e-05,
+      "loss": 0.7734,
+      "step": 561590
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 1.0589653325170276e-05,
+      "loss": 0.769,
+      "step": 561600
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 1.0588505395270528e-05,
+      "loss": 0.7185,
+      "step": 561610
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 1.0587357465370781e-05,
+      "loss": 0.7271,
+      "step": 561620
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 1.0586209535471034e-05,
+      "loss": 0.7408,
+      "step": 561630
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 1.0585061605571286e-05,
+      "loss": 0.8404,
+      "step": 561640
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 1.0583913675671539e-05,
+      "loss": 0.726,
+      "step": 561650
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 1.0582765745771792e-05,
+      "loss": 0.6778,
+      "step": 561660
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 1.0581617815872042e-05,
+      "loss": 0.7315,
+      "step": 561670
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 1.0580469885972295e-05,
+      "loss": 0.68,
+      "step": 561680
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 1.0579321956072548e-05,
+      "loss": 0.6924,
+      "step": 561690
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 1.05781740261728e-05,
+      "loss": 0.6934,
+      "step": 561700
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 1.0577026096273053e-05,
+      "loss": 0.735,
+      "step": 561710
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 1.0575878166373305e-05,
+      "loss": 0.7374,
+      "step": 561720
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 1.0574730236473558e-05,
+      "loss": 0.7359,
+      "step": 561730
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 1.0573582306573811e-05,
+      "loss": 0.7769,
+      "step": 561740
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 1.0572434376674063e-05,
+      "loss": 0.6896,
+      "step": 561750
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 1.0571286446774316e-05,
+      "loss": 0.7357,
+      "step": 561760
+    },
+    {
+      "epoch": 25.77,
+      "learning_rate": 1.057013851687457e-05,
+      "loss": 0.6522,
+      "step": 561770
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 1.0568990586974821e-05,
+      "loss": 0.7554,
+      "step": 561780
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 1.0567842657075074e-05,
+      "loss": 0.7213,
+      "step": 561790
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 1.0566694727175327e-05,
+      "loss": 0.6817,
+      "step": 561800
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 1.0565546797275577e-05,
+      "loss": 0.6878,
+      "step": 561810
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 1.056439886737583e-05,
+      "loss": 0.696,
+      "step": 561820
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 1.0563250937476084e-05,
+      "loss": 0.6082,
+      "step": 561830
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 1.0562103007576335e-05,
+      "loss": 0.692,
+      "step": 561840
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 1.0560955077676588e-05,
+      "loss": 0.6822,
+      "step": 561850
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 1.0559807147776842e-05,
+      "loss": 0.7869,
+      "step": 561860
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 1.0558659217877093e-05,
+      "loss": 0.6555,
+      "step": 561870
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 1.0557511287977347e-05,
+      "loss": 0.7827,
+      "step": 561880
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 1.05563633580776e-05,
+      "loss": 0.6898,
+      "step": 561890
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 1.0555215428177851e-05,
+      "loss": 0.657,
+      "step": 561900
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 1.0554067498278105e-05,
+      "loss": 0.652,
+      "step": 561910
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 1.0552919568378358e-05,
+      "loss": 0.767,
+      "step": 561920
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 1.055177163847861e-05,
+      "loss": 0.6717,
+      "step": 561930
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 1.0550623708578863e-05,
+      "loss": 0.6669,
+      "step": 561940
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 1.0549475778679116e-05,
+      "loss": 0.7229,
+      "step": 561950
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 1.0548327848779366e-05,
+      "loss": 0.6156,
+      "step": 561960
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 1.0547179918879619e-05,
+      "loss": 0.6635,
+      "step": 561970
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 1.0546031988979872e-05,
+      "loss": 0.7515,
+      "step": 561980
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 1.0544884059080124e-05,
+      "loss": 0.7977,
+      "step": 561990
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 1.0543736129180377e-05,
+      "loss": 0.7672,
+      "step": 562000
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 1.054258819928063e-05,
+      "loss": 0.7241,
+      "step": 562010
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 1.0541440269380882e-05,
+      "loss": 0.7652,
+      "step": 562020
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 1.0540292339481135e-05,
+      "loss": 0.7379,
+      "step": 562030
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 1.0539144409581388e-05,
+      "loss": 0.6949,
+      "step": 562040
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 1.053799647968164e-05,
+      "loss": 0.7682,
+      "step": 562050
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 1.0536848549781893e-05,
+      "loss": 0.7861,
+      "step": 562060
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 1.0535700619882145e-05,
+      "loss": 0.7361,
+      "step": 562070
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 1.0534552689982398e-05,
+      "loss": 0.7896,
+      "step": 562080
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 1.0533404760082651e-05,
+      "loss": 0.6882,
+      "step": 562090
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 1.0532256830182901e-05,
+      "loss": 0.7898,
+      "step": 562100
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 1.0531108900283154e-05,
+      "loss": 0.7147,
+      "step": 562110
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 1.0529960970383407e-05,
+      "loss": 0.7196,
+      "step": 562120
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 1.0528813040483659e-05,
+      "loss": 0.7012,
+      "step": 562130
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 1.0527665110583912e-05,
+      "loss": 0.7421,
+      "step": 562140
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 1.0526517180684165e-05,
+      "loss": 0.635,
+      "step": 562150
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 1.0525369250784417e-05,
+      "loss": 0.7219,
+      "step": 562160
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 1.052422132088467e-05,
+      "loss": 0.7404,
+      "step": 562170
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 1.0523073390984923e-05,
+      "loss": 0.754,
+      "step": 562180
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 1.0521925461085175e-05,
+      "loss": 0.7318,
+      "step": 562190
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 1.0520777531185428e-05,
+      "loss": 0.6621,
+      "step": 562200
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 1.0519629601285682e-05,
+      "loss": 0.7275,
+      "step": 562210
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 1.0518481671385933e-05,
+      "loss": 0.7137,
+      "step": 562220
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 1.0517333741486186e-05,
+      "loss": 0.6914,
+      "step": 562230
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 1.051618581158644e-05,
+      "loss": 0.722,
+      "step": 562240
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 1.051503788168669e-05,
+      "loss": 0.6066,
+      "step": 562250
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 1.0513889951786943e-05,
+      "loss": 0.8003,
+      "step": 562260
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 1.0512742021887196e-05,
+      "loss": 0.719,
+      "step": 562270
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 1.0511594091987447e-05,
+      "loss": 0.7804,
+      "step": 562280
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 1.05104461620877e-05,
+      "loss": 0.7134,
+      "step": 562290
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 1.0509298232187954e-05,
+      "loss": 0.681,
+      "step": 562300
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 1.0508150302288205e-05,
+      "loss": 0.735,
+      "step": 562310
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 1.0507002372388459e-05,
+      "loss": 0.7515,
+      "step": 562320
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 1.0505854442488712e-05,
+      "loss": 0.6865,
+      "step": 562330
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 1.0504706512588963e-05,
+      "loss": 0.7632,
+      "step": 562340
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 1.0503558582689217e-05,
+      "loss": 0.683,
+      "step": 562350
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 1.050241065278947e-05,
+      "loss": 0.6941,
+      "step": 562360
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 1.0501262722889722e-05,
+      "loss": 0.7209,
+      "step": 562370
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 1.0500114792989975e-05,
+      "loss": 0.7412,
+      "step": 562380
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 1.0498966863090228e-05,
+      "loss": 0.6965,
+      "step": 562390
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 1.0497818933190478e-05,
+      "loss": 0.73,
+      "step": 562400
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 1.0496671003290731e-05,
+      "loss": 0.7066,
+      "step": 562410
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 1.0495523073390983e-05,
+      "loss": 0.7089,
+      "step": 562420
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 1.0494375143491236e-05,
+      "loss": 0.7754,
+      "step": 562430
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 1.0493227213591489e-05,
+      "loss": 0.6734,
+      "step": 562440
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 1.049207928369174e-05,
+      "loss": 0.7521,
+      "step": 562450
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 1.0490931353791994e-05,
+      "loss": 0.7217,
+      "step": 562460
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 1.0489783423892247e-05,
+      "loss": 0.6875,
+      "step": 562470
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 1.0488635493992499e-05,
+      "loss": 0.7614,
+      "step": 562480
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 1.0487487564092752e-05,
+      "loss": 0.7182,
+      "step": 562490
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 1.0486339634193005e-05,
+      "loss": 0.6312,
+      "step": 562500
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 1.0485191704293257e-05,
+      "loss": 0.7186,
+      "step": 562510
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 1.048404377439351e-05,
+      "loss": 0.7109,
+      "step": 562520
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 1.0482895844493763e-05,
+      "loss": 0.6182,
+      "step": 562530
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 1.0481747914594013e-05,
+      "loss": 0.7033,
+      "step": 562540
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 1.0480599984694266e-05,
+      "loss": 0.714,
+      "step": 562550
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 1.047945205479452e-05,
+      "loss": 0.6374,
+      "step": 562560
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 1.0478304124894771e-05,
+      "loss": 0.696,
+      "step": 562570
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 1.0477156194995024e-05,
+      "loss": 0.7622,
+      "step": 562580
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 1.0476008265095278e-05,
+      "loss": 0.707,
+      "step": 562590
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 1.047486033519553e-05,
+      "loss": 0.8009,
+      "step": 562600
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 1.0473712405295782e-05,
+      "loss": 0.7169,
+      "step": 562610
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 1.0472564475396036e-05,
+      "loss": 0.6779,
+      "step": 562620
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 1.0471416545496287e-05,
+      "loss": 0.7702,
+      "step": 562630
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 1.047026861559654e-05,
+      "loss": 0.708,
+      "step": 562640
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 1.0469120685696794e-05,
+      "loss": 0.7063,
+      "step": 562650
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 1.0467972755797045e-05,
+      "loss": 0.6775,
+      "step": 562660
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 1.0466824825897298e-05,
+      "loss": 0.7664,
+      "step": 562670
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 1.0465676895997552e-05,
+      "loss": 0.7113,
+      "step": 562680
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 1.0464528966097802e-05,
+      "loss": 0.732,
+      "step": 562690
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 1.0463381036198055e-05,
+      "loss": 0.7413,
+      "step": 562700
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 1.0462233106298308e-05,
+      "loss": 0.7468,
+      "step": 562710
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 1.046108517639856e-05,
+      "loss": 0.7842,
+      "step": 562720
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 1.0459937246498813e-05,
+      "loss": 0.7448,
+      "step": 562730
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 1.0458789316599066e-05,
+      "loss": 0.7782,
+      "step": 562740
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 1.0457641386699318e-05,
+      "loss": 0.7368,
+      "step": 562750
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 1.0456493456799571e-05,
+      "loss": 0.6875,
+      "step": 562760
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 1.0455345526899822e-05,
+      "loss": 0.7672,
+      "step": 562770
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 1.0454197597000076e-05,
+      "loss": 0.7148,
+      "step": 562780
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 1.0453049667100329e-05,
+      "loss": 0.7334,
+      "step": 562790
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 1.045190173720058e-05,
+      "loss": 0.7126,
+      "step": 562800
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 1.0450753807300834e-05,
+      "loss": 0.693,
+      "step": 562810
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 1.0449605877401087e-05,
+      "loss": 0.6859,
+      "step": 562820
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 1.0448457947501337e-05,
+      "loss": 0.6997,
+      "step": 562830
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 1.044731001760159e-05,
+      "loss": 0.6784,
+      "step": 562840
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 1.0446162087701843e-05,
+      "loss": 0.7885,
+      "step": 562850
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 1.0445014157802095e-05,
+      "loss": 0.7024,
+      "step": 562860
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 1.0443866227902348e-05,
+      "loss": 0.7503,
+      "step": 562870
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 1.0442718298002601e-05,
+      "loss": 0.7113,
+      "step": 562880
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 1.0441570368102853e-05,
+      "loss": 0.6057,
+      "step": 562890
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 1.0440422438203106e-05,
+      "loss": 0.7259,
+      "step": 562900
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 1.043927450830336e-05,
+      "loss": 0.7855,
+      "step": 562910
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 1.0438126578403611e-05,
+      "loss": 0.6984,
+      "step": 562920
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 1.0436978648503864e-05,
+      "loss": 0.7123,
+      "step": 562930
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 1.0435830718604117e-05,
+      "loss": 0.7799,
+      "step": 562940
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 1.0434682788704369e-05,
+      "loss": 0.732,
+      "step": 562950
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 1.0433534858804622e-05,
+      "loss": 0.779,
+      "step": 562960
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 1.0432386928904875e-05,
+      "loss": 0.7626,
+      "step": 562970
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 1.0431238999005125e-05,
+      "loss": 0.7366,
+      "step": 562980
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 1.0430091069105379e-05,
+      "loss": 0.7295,
+      "step": 562990
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 1.0428943139205632e-05,
+      "loss": 0.8136,
+      "step": 563000
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 1.0427795209305883e-05,
+      "loss": 0.6763,
+      "step": 563010
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 1.0426647279406137e-05,
+      "loss": 0.7744,
+      "step": 563020
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 1.042549934950639e-05,
+      "loss": 0.7658,
+      "step": 563030
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 1.0424351419606641e-05,
+      "loss": 0.6588,
+      "step": 563040
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 1.0423203489706895e-05,
+      "loss": 0.7642,
+      "step": 563050
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 1.0422055559807148e-05,
+      "loss": 0.7567,
+      "step": 563060
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 1.04209076299074e-05,
+      "loss": 0.6942,
+      "step": 563070
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 1.0419759700007653e-05,
+      "loss": 0.724,
+      "step": 563080
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 1.0418611770107904e-05,
+      "loss": 0.6591,
+      "step": 563090
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 1.0417463840208157e-05,
+      "loss": 0.7372,
+      "step": 563100
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 1.041631591030841e-05,
+      "loss": 0.6896,
+      "step": 563110
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 1.041516798040866e-05,
+      "loss": 0.7614,
+      "step": 563120
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 1.0414020050508914e-05,
+      "loss": 0.765,
+      "step": 563130
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 1.0412872120609167e-05,
+      "loss": 0.7015,
+      "step": 563140
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 1.0411724190709419e-05,
+      "loss": 0.7139,
+      "step": 563150
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 1.0410576260809672e-05,
+      "loss": 0.741,
+      "step": 563160
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 1.0409428330909925e-05,
+      "loss": 0.7682,
+      "step": 563170
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 1.0408280401010177e-05,
+      "loss": 0.6753,
+      "step": 563180
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 1.040713247111043e-05,
+      "loss": 0.7219,
+      "step": 563190
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 1.0405984541210683e-05,
+      "loss": 0.7019,
+      "step": 563200
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 1.0404836611310935e-05,
+      "loss": 0.6476,
+      "step": 563210
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 1.0403688681411188e-05,
+      "loss": 0.7637,
+      "step": 563220
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 1.0402540751511441e-05,
+      "loss": 0.655,
+      "step": 563230
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 1.0401392821611693e-05,
+      "loss": 0.7395,
+      "step": 563240
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 1.0400244891711946e-05,
+      "loss": 0.6608,
+      "step": 563250
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 1.0399096961812199e-05,
+      "loss": 0.7187,
+      "step": 563260
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 1.0397949031912449e-05,
+      "loss": 0.7708,
+      "step": 563270
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 1.0396801102012702e-05,
+      "loss": 0.6684,
+      "step": 563280
+    },
+    {
+      "epoch": 25.84,
+      "learning_rate": 1.0395653172112955e-05,
+      "loss": 0.7482,
+      "step": 563290
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 1.0394505242213207e-05,
+      "loss": 0.7475,
+      "step": 563300
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 1.039335731231346e-05,
+      "loss": 0.6862,
+      "step": 563310
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 1.0392209382413713e-05,
+      "loss": 0.7774,
+      "step": 563320
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 1.0391061452513965e-05,
+      "loss": 0.7095,
+      "step": 563330
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 1.0389913522614218e-05,
+      "loss": 0.8224,
+      "step": 563340
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 1.0388765592714472e-05,
+      "loss": 0.7896,
+      "step": 563350
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 1.0387617662814723e-05,
+      "loss": 0.669,
+      "step": 563360
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 1.0386469732914976e-05,
+      "loss": 0.7392,
+      "step": 563370
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 1.038532180301523e-05,
+      "loss": 0.6935,
+      "step": 563380
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 1.0384173873115481e-05,
+      "loss": 0.7241,
+      "step": 563390
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 1.0383025943215734e-05,
+      "loss": 0.7339,
+      "step": 563400
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 1.0381878013315988e-05,
+      "loss": 0.712,
+      "step": 563410
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 1.0380730083416237e-05,
+      "loss": 0.7687,
+      "step": 563420
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 1.037958215351649e-05,
+      "loss": 0.7442,
+      "step": 563430
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 1.0378434223616742e-05,
+      "loss": 0.7167,
+      "step": 563440
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 1.0377286293716995e-05,
+      "loss": 0.7364,
+      "step": 563450
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 1.0376138363817249e-05,
+      "loss": 0.7776,
+      "step": 563460
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 1.03749904339175e-05,
+      "loss": 0.7996,
+      "step": 563470
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 1.0373842504017754e-05,
+      "loss": 0.7622,
+      "step": 563480
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 1.0372694574118007e-05,
+      "loss": 0.745,
+      "step": 563490
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 1.0371546644218258e-05,
+      "loss": 0.7134,
+      "step": 563500
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 1.0370398714318512e-05,
+      "loss": 0.7029,
+      "step": 563510
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 1.0369250784418765e-05,
+      "loss": 0.6396,
+      "step": 563520
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 1.0368102854519016e-05,
+      "loss": 0.7045,
+      "step": 563530
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 1.036695492461927e-05,
+      "loss": 0.6982,
+      "step": 563540
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 1.0365806994719523e-05,
+      "loss": 0.7798,
+      "step": 563550
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 1.0364659064819773e-05,
+      "loss": 0.7208,
+      "step": 563560
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 1.0363511134920026e-05,
+      "loss": 0.6639,
+      "step": 563570
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 1.036236320502028e-05,
+      "loss": 0.7208,
+      "step": 563580
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 1.036121527512053e-05,
+      "loss": 0.6824,
+      "step": 563590
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 1.0360067345220784e-05,
+      "loss": 0.6342,
+      "step": 563600
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 1.0358919415321037e-05,
+      "loss": 0.7775,
+      "step": 563610
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 1.0357771485421289e-05,
+      "loss": 0.7929,
+      "step": 563620
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 1.0356623555521542e-05,
+      "loss": 0.7723,
+      "step": 563630
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 1.0355475625621795e-05,
+      "loss": 0.6744,
+      "step": 563640
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 1.0354327695722047e-05,
+      "loss": 0.7401,
+      "step": 563650
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 1.03531797658223e-05,
+      "loss": 0.6969,
+      "step": 563660
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 1.0352031835922553e-05,
+      "loss": 0.7127,
+      "step": 563670
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 1.0350883906022805e-05,
+      "loss": 0.7347,
+      "step": 563680
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 1.0349735976123058e-05,
+      "loss": 0.7499,
+      "step": 563690
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 1.0348588046223311e-05,
+      "loss": 0.7463,
+      "step": 563700
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 1.0347440116323561e-05,
+      "loss": 0.7281,
+      "step": 563710
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 1.0346292186423814e-05,
+      "loss": 0.6704,
+      "step": 563720
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 1.0345144256524068e-05,
+      "loss": 0.7636,
+      "step": 563730
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 1.034399632662432e-05,
+      "loss": 0.7533,
+      "step": 563740
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 1.0342848396724572e-05,
+      "loss": 0.7223,
+      "step": 563750
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 1.0341700466824826e-05,
+      "loss": 0.805,
+      "step": 563760
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 1.0340552536925077e-05,
+      "loss": 0.732,
+      "step": 563770
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 1.033940460702533e-05,
+      "loss": 0.6695,
+      "step": 563780
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 1.0338256677125582e-05,
+      "loss": 0.7781,
+      "step": 563790
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 1.0337108747225835e-05,
+      "loss": 0.701,
+      "step": 563800
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 1.0335960817326088e-05,
+      "loss": 0.7485,
+      "step": 563810
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 1.033481288742634e-05,
+      "loss": 0.7471,
+      "step": 563820
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 1.0333664957526593e-05,
+      "loss": 0.7739,
+      "step": 563830
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 1.0332517027626847e-05,
+      "loss": 0.673,
+      "step": 563840
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 1.0331369097727096e-05,
+      "loss": 0.7059,
+      "step": 563850
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 1.033022116782735e-05,
+      "loss": 0.6899,
+      "step": 563860
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 1.0329073237927603e-05,
+      "loss": 0.6919,
+      "step": 563870
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 1.0327925308027854e-05,
+      "loss": 0.7213,
+      "step": 563880
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 1.0326777378128108e-05,
+      "loss": 0.6835,
+      "step": 563890
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 1.0325629448228361e-05,
+      "loss": 0.7809,
+      "step": 563900
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 1.0324481518328612e-05,
+      "loss": 0.7429,
+      "step": 563910
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 1.0323333588428866e-05,
+      "loss": 0.6,
+      "step": 563920
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 1.0322185658529119e-05,
+      "loss": 0.8018,
+      "step": 563930
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 1.032103772862937e-05,
+      "loss": 0.8257,
+      "step": 563940
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 1.0319889798729624e-05,
+      "loss": 0.6808,
+      "step": 563950
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 1.0318741868829877e-05,
+      "loss": 0.7339,
+      "step": 563960
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 1.0317593938930129e-05,
+      "loss": 0.7206,
+      "step": 563970
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 1.0316446009030382e-05,
+      "loss": 0.6391,
+      "step": 563980
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 1.0315298079130635e-05,
+      "loss": 0.7063,
+      "step": 563990
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 1.0314150149230885e-05,
+      "loss": 0.7528,
+      "step": 564000
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 1.0313002219331138e-05,
+      "loss": 0.6945,
+      "step": 564010
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 1.0311854289431391e-05,
+      "loss": 0.6667,
+      "step": 564020
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 1.0310706359531643e-05,
+      "loss": 0.6981,
+      "step": 564030
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 1.0309558429631896e-05,
+      "loss": 0.7572,
+      "step": 564040
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 1.030841049973215e-05,
+      "loss": 0.6763,
+      "step": 564050
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 1.0307262569832401e-05,
+      "loss": 0.712,
+      "step": 564060
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 1.0306114639932654e-05,
+      "loss": 0.6851,
+      "step": 564070
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 1.0304966710032907e-05,
+      "loss": 0.6582,
+      "step": 564080
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 1.0303818780133159e-05,
+      "loss": 0.7136,
+      "step": 564090
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 1.0302670850233412e-05,
+      "loss": 0.6851,
+      "step": 564100
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 1.0301522920333665e-05,
+      "loss": 0.7212,
+      "step": 564110
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 1.0300374990433917e-05,
+      "loss": 0.669,
+      "step": 564120
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 1.029922706053417e-05,
+      "loss": 0.7384,
+      "step": 564130
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 1.029807913063442e-05,
+      "loss": 0.7126,
+      "step": 564140
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 1.0296931200734673e-05,
+      "loss": 0.7732,
+      "step": 564150
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 1.0295783270834927e-05,
+      "loss": 0.7265,
+      "step": 564160
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 1.0294635340935178e-05,
+      "loss": 0.7467,
+      "step": 564170
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 1.0293487411035431e-05,
+      "loss": 0.6792,
+      "step": 564180
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 1.0292339481135685e-05,
+      "loss": 0.7397,
+      "step": 564190
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 1.0291191551235936e-05,
+      "loss": 0.684,
+      "step": 564200
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 1.029004362133619e-05,
+      "loss": 0.6674,
+      "step": 564210
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 1.0288895691436443e-05,
+      "loss": 0.7611,
+      "step": 564220
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 1.0287747761536694e-05,
+      "loss": 0.7265,
+      "step": 564230
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 1.0286599831636947e-05,
+      "loss": 0.6755,
+      "step": 564240
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 1.02854519017372e-05,
+      "loss": 0.7143,
+      "step": 564250
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 1.0284303971837452e-05,
+      "loss": 0.7283,
+      "step": 564260
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 1.0283156041937705e-05,
+      "loss": 0.7465,
+      "step": 564270
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 1.0282008112037959e-05,
+      "loss": 0.69,
+      "step": 564280
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 1.0280860182138209e-05,
+      "loss": 0.7279,
+      "step": 564290
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 1.0279712252238462e-05,
+      "loss": 0.8245,
+      "step": 564300
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 1.0278564322338715e-05,
+      "loss": 0.6854,
+      "step": 564310
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 1.0277416392438967e-05,
+      "loss": 0.6509,
+      "step": 564320
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 1.027626846253922e-05,
+      "loss": 0.7677,
+      "step": 564330
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 1.0275120532639473e-05,
+      "loss": 0.7397,
+      "step": 564340
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 1.0273972602739725e-05,
+      "loss": 0.7583,
+      "step": 564350
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 1.0272824672839978e-05,
+      "loss": 0.62,
+      "step": 564360
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 1.0271676742940231e-05,
+      "loss": 0.6918,
+      "step": 564370
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 1.0270528813040483e-05,
+      "loss": 0.7351,
+      "step": 564380
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 1.0269380883140736e-05,
+      "loss": 0.6759,
+      "step": 564390
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 1.0268232953240989e-05,
+      "loss": 0.8019,
+      "step": 564400
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 1.026708502334124e-05,
+      "loss": 0.6631,
+      "step": 564410
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 1.0265937093441494e-05,
+      "loss": 0.716,
+      "step": 564420
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 1.0264789163541747e-05,
+      "loss": 0.829,
+      "step": 564430
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 1.0263641233641997e-05,
+      "loss": 0.7598,
+      "step": 564440
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 1.026249330374225e-05,
+      "loss": 0.695,
+      "step": 564450
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 1.0261345373842504e-05,
+      "loss": 0.7293,
+      "step": 564460
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 1.0260197443942755e-05,
+      "loss": 0.6767,
+      "step": 564470
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 1.0259049514043008e-05,
+      "loss": 0.6741,
+      "step": 564480
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 1.025790158414326e-05,
+      "loss": 0.6662,
+      "step": 564490
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 1.0256753654243513e-05,
+      "loss": 0.7343,
+      "step": 564500
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 1.0255605724343766e-05,
+      "loss": 0.7695,
+      "step": 564510
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 1.0254457794444018e-05,
+      "loss": 0.7675,
+      "step": 564520
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 1.0253309864544271e-05,
+      "loss": 0.7132,
+      "step": 564530
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 1.0252161934644524e-05,
+      "loss": 0.697,
+      "step": 564540
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 1.0251014004744776e-05,
+      "loss": 0.7229,
+      "step": 564550
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 1.024986607484503e-05,
+      "loss": 0.7504,
+      "step": 564560
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 1.0248718144945282e-05,
+      "loss": 0.6862,
+      "step": 564570
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 1.0247570215045532e-05,
+      "loss": 0.6891,
+      "step": 564580
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 1.0246422285145785e-05,
+      "loss": 0.798,
+      "step": 564590
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 1.0245274355246039e-05,
+      "loss": 0.7237,
+      "step": 564600
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 1.024412642534629e-05,
+      "loss": 0.6827,
+      "step": 564610
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 1.0242978495446544e-05,
+      "loss": 0.6697,
+      "step": 564620
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 1.0241830565546797e-05,
+      "loss": 0.7232,
+      "step": 564630
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 1.0240682635647048e-05,
+      "loss": 0.7407,
+      "step": 564640
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 1.0239534705747302e-05,
+      "loss": 0.7259,
+      "step": 564650
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 1.0238386775847555e-05,
+      "loss": 0.6959,
+      "step": 564660
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 1.0237238845947806e-05,
+      "loss": 0.6634,
+      "step": 564670
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 1.023609091604806e-05,
+      "loss": 0.7243,
+      "step": 564680
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 1.0234942986148313e-05,
+      "loss": 0.6602,
+      "step": 564690
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 1.0233795056248564e-05,
+      "loss": 0.6538,
+      "step": 564700
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 1.0232647126348818e-05,
+      "loss": 0.8099,
+      "step": 564710
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 1.0231499196449071e-05,
+      "loss": 0.7391,
+      "step": 564720
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 1.023035126654932e-05,
+      "loss": 0.7435,
+      "step": 564730
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 1.0229203336649574e-05,
+      "loss": 0.6811,
+      "step": 564740
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 1.0228055406749827e-05,
+      "loss": 0.6971,
+      "step": 564750
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 1.0226907476850079e-05,
+      "loss": 0.7083,
+      "step": 564760
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 1.0225759546950332e-05,
+      "loss": 0.7228,
+      "step": 564770
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 1.0224611617050585e-05,
+      "loss": 0.7778,
+      "step": 564780
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 1.0223463687150837e-05,
+      "loss": 0.7343,
+      "step": 564790
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 1.022231575725109e-05,
+      "loss": 0.7483,
+      "step": 564800
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 1.0221167827351342e-05,
+      "loss": 0.8037,
+      "step": 564810
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 1.0220019897451595e-05,
+      "loss": 0.7136,
+      "step": 564820
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 1.0218871967551848e-05,
+      "loss": 0.7214,
+      "step": 564830
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 1.02177240376521e-05,
+      "loss": 0.7476,
+      "step": 564840
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 1.0216576107752353e-05,
+      "loss": 0.716,
+      "step": 564850
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 1.0215428177852606e-05,
+      "loss": 0.6809,
+      "step": 564860
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 1.0214280247952856e-05,
+      "loss": 0.7797,
+      "step": 564870
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 1.021313231805311e-05,
+      "loss": 0.6268,
+      "step": 564880
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 1.0211984388153362e-05,
+      "loss": 0.8454,
+      "step": 564890
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 1.0210836458253614e-05,
+      "loss": 0.804,
+      "step": 564900
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 1.0209688528353867e-05,
+      "loss": 0.6919,
+      "step": 564910
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 1.020854059845412e-05,
+      "loss": 0.6871,
+      "step": 564920
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 1.0207392668554372e-05,
+      "loss": 0.705,
+      "step": 564930
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 1.0206244738654625e-05,
+      "loss": 0.7362,
+      "step": 564940
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 1.0205096808754879e-05,
+      "loss": 0.6435,
+      "step": 564950
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 1.020394887885513e-05,
+      "loss": 0.6828,
+      "step": 564960
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 1.0202800948955383e-05,
+      "loss": 0.6905,
+      "step": 564970
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 1.0201653019055637e-05,
+      "loss": 0.7251,
+      "step": 564980
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 1.0200505089155888e-05,
+      "loss": 0.7346,
+      "step": 564990
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 1.0199357159256141e-05,
+      "loss": 0.6714,
+      "step": 565000
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 1.0198209229356395e-05,
+      "loss": 0.7577,
+      "step": 565010
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 1.0197061299456644e-05,
+      "loss": 0.6781,
+      "step": 565020
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 1.0195913369556898e-05,
+      "loss": 0.7221,
+      "step": 565030
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 1.0194765439657151e-05,
+      "loss": 0.8061,
+      "step": 565040
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 1.0193617509757402e-05,
+      "loss": 0.6742,
+      "step": 565050
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 1.0192469579857656e-05,
+      "loss": 0.761,
+      "step": 565060
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 1.0191321649957909e-05,
+      "loss": 0.7165,
+      "step": 565070
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 1.019017372005816e-05,
+      "loss": 0.7235,
+      "step": 565080
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 1.0189025790158414e-05,
+      "loss": 0.7912,
+      "step": 565090
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 1.0187877860258667e-05,
+      "loss": 0.7526,
+      "step": 565100
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 1.0186729930358919e-05,
+      "loss": 0.6758,
+      "step": 565110
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 1.0185582000459172e-05,
+      "loss": 0.7323,
+      "step": 565120
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 1.0184434070559425e-05,
+      "loss": 0.7196,
+      "step": 565130
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 1.0183286140659677e-05,
+      "loss": 0.6841,
+      "step": 565140
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 1.018213821075993e-05,
+      "loss": 0.7751,
+      "step": 565150
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 1.018099028086018e-05,
+      "loss": 0.7612,
+      "step": 565160
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 1.0179842350960433e-05,
+      "loss": 0.7731,
+      "step": 565170
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 1.0178694421060686e-05,
+      "loss": 0.8059,
+      "step": 565180
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 1.0177546491160938e-05,
+      "loss": 0.7148,
+      "step": 565190
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 1.0176398561261191e-05,
+      "loss": 0.7548,
+      "step": 565200
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 1.0175250631361444e-05,
+      "loss": 0.682,
+      "step": 565210
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 1.0174102701461696e-05,
+      "loss": 0.7025,
+      "step": 565220
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 1.0172954771561949e-05,
+      "loss": 0.6512,
+      "step": 565230
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 1.0171806841662202e-05,
+      "loss": 0.746,
+      "step": 565240
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 1.0170658911762454e-05,
+      "loss": 0.7611,
+      "step": 565250
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 1.0169510981862707e-05,
+      "loss": 0.7252,
+      "step": 565260
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 1.016836305196296e-05,
+      "loss": 0.8135,
+      "step": 565270
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 1.0167215122063212e-05,
+      "loss": 0.7669,
+      "step": 565280
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 1.0166067192163465e-05,
+      "loss": 0.6957,
+      "step": 565290
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 1.0164919262263718e-05,
+      "loss": 0.7831,
+      "step": 565300
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 1.0163771332363968e-05,
+      "loss": 0.6816,
+      "step": 565310
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 1.0162623402464221e-05,
+      "loss": 0.6407,
+      "step": 565320
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 1.0161475472564475e-05,
+      "loss": 0.6562,
+      "step": 565330
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 1.0160327542664726e-05,
+      "loss": 0.6962,
+      "step": 565340
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 1.015917961276498e-05,
+      "loss": 0.6291,
+      "step": 565350
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 1.0158031682865233e-05,
+      "loss": 0.7236,
+      "step": 565360
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 1.0156883752965484e-05,
+      "loss": 0.6499,
+      "step": 565370
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 1.0155735823065737e-05,
+      "loss": 0.6996,
+      "step": 565380
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 1.015458789316599e-05,
+      "loss": 0.7611,
+      "step": 565390
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 1.0153439963266242e-05,
+      "loss": 0.7957,
+      "step": 565400
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 1.0152292033366495e-05,
+      "loss": 0.7145,
+      "step": 565410
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 1.0151144103466749e-05,
+      "loss": 0.6923,
+      "step": 565420
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 1.0149996173567e-05,
+      "loss": 0.7497,
+      "step": 565430
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 1.0148848243667254e-05,
+      "loss": 0.8007,
+      "step": 565440
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 1.0147700313767507e-05,
+      "loss": 0.8121,
+      "step": 565450
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 1.0146552383867757e-05,
+      "loss": 0.6874,
+      "step": 565460
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 1.014540445396801e-05,
+      "loss": 0.786,
+      "step": 565470
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 1.0144256524068263e-05,
+      "loss": 0.712,
+      "step": 565480
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 1.0143108594168515e-05,
+      "loss": 0.7462,
+      "step": 565490
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 1.0141960664268768e-05,
+      "loss": 0.7585,
+      "step": 565500
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 1.014081273436902e-05,
+      "loss": 0.7171,
+      "step": 565510
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 1.0139664804469273e-05,
+      "loss": 0.7057,
+      "step": 565520
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 1.0138516874569526e-05,
+      "loss": 0.6754,
+      "step": 565530
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 1.0137368944669777e-05,
+      "loss": 0.6748,
+      "step": 565540
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 1.013622101477003e-05,
+      "loss": 0.8085,
+      "step": 565550
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 1.0135073084870284e-05,
+      "loss": 0.735,
+      "step": 565560
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 1.0133925154970535e-05,
+      "loss": 0.7453,
+      "step": 565570
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 1.0132777225070789e-05,
+      "loss": 0.7289,
+      "step": 565580
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 1.0131629295171042e-05,
+      "loss": 0.764,
+      "step": 565590
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 1.0130481365271292e-05,
+      "loss": 0.7444,
+      "step": 565600
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 1.0129333435371545e-05,
+      "loss": 0.7115,
+      "step": 565610
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 1.0128185505471798e-05,
+      "loss": 0.6347,
+      "step": 565620
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 1.012703757557205e-05,
+      "loss": 0.744,
+      "step": 565630
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 1.0125889645672303e-05,
+      "loss": 0.6891,
+      "step": 565640
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 1.0124741715772556e-05,
+      "loss": 0.7483,
+      "step": 565650
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 1.0123593785872808e-05,
+      "loss": 0.7218,
+      "step": 565660
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 1.0122445855973061e-05,
+      "loss": 0.7208,
+      "step": 565670
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 1.0121297926073314e-05,
+      "loss": 0.8636,
+      "step": 565680
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 1.0120149996173566e-05,
+      "loss": 0.7455,
+      "step": 565690
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 1.011900206627382e-05,
+      "loss": 0.736,
+      "step": 565700
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 1.0117854136374072e-05,
+      "loss": 0.7442,
+      "step": 565710
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 1.0116706206474324e-05,
+      "loss": 0.7391,
+      "step": 565720
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 1.0115558276574577e-05,
+      "loss": 0.71,
+      "step": 565730
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 1.011441034667483e-05,
+      "loss": 0.7898,
+      "step": 565740
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 1.011326241677508e-05,
+      "loss": 0.6397,
+      "step": 565750
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 1.0112114486875334e-05,
+      "loss": 0.7893,
+      "step": 565760
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 1.0110966556975587e-05,
+      "loss": 0.7743,
+      "step": 565770
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 1.0109818627075838e-05,
+      "loss": 0.7054,
+      "step": 565780
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 1.0108670697176092e-05,
+      "loss": 0.717,
+      "step": 565790
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 1.0107522767276345e-05,
+      "loss": 0.7898,
+      "step": 565800
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 1.0106374837376596e-05,
+      "loss": 0.6572,
+      "step": 565810
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 1.010522690747685e-05,
+      "loss": 0.6766,
+      "step": 565820
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 1.0104078977577103e-05,
+      "loss": 0.7716,
+      "step": 565830
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 1.0102931047677354e-05,
+      "loss": 0.801,
+      "step": 565840
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 1.0101783117777608e-05,
+      "loss": 0.8261,
+      "step": 565850
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 1.010063518787786e-05,
+      "loss": 0.7561,
+      "step": 565860
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 1.0099487257978112e-05,
+      "loss": 0.7023,
+      "step": 565870
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 1.0098339328078366e-05,
+      "loss": 0.7243,
+      "step": 565880
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 1.0097191398178616e-05,
+      "loss": 0.7138,
+      "step": 565890
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 1.0096043468278869e-05,
+      "loss": 0.6628,
+      "step": 565900
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 1.0094895538379122e-05,
+      "loss": 0.6501,
+      "step": 565910
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 1.0093747608479374e-05,
+      "loss": 0.7135,
+      "step": 565920
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 1.0092599678579627e-05,
+      "loss": 0.6868,
+      "step": 565930
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 1.009145174867988e-05,
+      "loss": 0.6601,
+      "step": 565940
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 1.0090303818780132e-05,
+      "loss": 0.7872,
+      "step": 565950
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 1.0089155888880385e-05,
+      "loss": 0.7711,
+      "step": 565960
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 1.0088007958980638e-05,
+      "loss": 0.7175,
+      "step": 565970
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 1.008686002908089e-05,
+      "loss": 0.7299,
+      "step": 565980
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 1.0085712099181143e-05,
+      "loss": 0.6875,
+      "step": 565990
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 1.0084564169281396e-05,
+      "loss": 0.5972,
+      "step": 566000
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 1.0083416239381648e-05,
+      "loss": 0.6974,
+      "step": 566010
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 1.0082268309481901e-05,
+      "loss": 0.6616,
+      "step": 566020
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 1.0081120379582154e-05,
+      "loss": 0.6798,
+      "step": 566030
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 1.0079972449682404e-05,
+      "loss": 0.7676,
+      "step": 566040
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 1.0078824519782657e-05,
+      "loss": 0.8459,
+      "step": 566050
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 1.007767658988291e-05,
+      "loss": 0.7165,
+      "step": 566060
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 1.0076528659983162e-05,
+      "loss": 0.7617,
+      "step": 566070
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 1.0075380730083415e-05,
+      "loss": 0.7255,
+      "step": 566080
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 1.0074232800183669e-05,
+      "loss": 0.7299,
+      "step": 566090
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 1.007308487028392e-05,
+      "loss": 0.706,
+      "step": 566100
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 1.0071936940384173e-05,
+      "loss": 0.7251,
+      "step": 566110
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 1.0070789010484427e-05,
+      "loss": 0.7614,
+      "step": 566120
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 1.0069641080584678e-05,
+      "loss": 0.7363,
+      "step": 566130
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 1.0068493150684931e-05,
+      "loss": 0.6973,
+      "step": 566140
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 1.0067345220785185e-05,
+      "loss": 0.671,
+      "step": 566150
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 1.0066197290885436e-05,
+      "loss": 0.7564,
+      "step": 566160
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 1.006504936098569e-05,
+      "loss": 0.7579,
+      "step": 566170
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 1.006390143108594e-05,
+      "loss": 0.7193,
+      "step": 566180
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 1.0062753501186192e-05,
+      "loss": 0.7197,
+      "step": 566190
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 1.0061605571286446e-05,
+      "loss": 0.6793,
+      "step": 566200
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 1.0060457641386697e-05,
+      "loss": 0.7253,
+      "step": 566210
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 1.005930971148695e-05,
+      "loss": 0.7244,
+      "step": 566220
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 1.0058161781587204e-05,
+      "loss": 0.8085,
+      "step": 566230
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 1.0057013851687455e-05,
+      "loss": 0.7851,
+      "step": 566240
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 1.0055865921787709e-05,
+      "loss": 0.6837,
+      "step": 566250
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 1.0054717991887962e-05,
+      "loss": 0.7994,
+      "step": 566260
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 1.0053570061988213e-05,
+      "loss": 0.6936,
+      "step": 566270
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 1.0052422132088467e-05,
+      "loss": 0.6457,
+      "step": 566280
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 1.005127420218872e-05,
+      "loss": 0.6718,
+      "step": 566290
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 1.0050126272288971e-05,
+      "loss": 0.701,
+      "step": 566300
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 1.0048978342389225e-05,
+      "loss": 0.7279,
+      "step": 566310
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 1.0047830412489478e-05,
+      "loss": 0.6767,
+      "step": 566320
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 1.0046682482589728e-05,
+      "loss": 0.674,
+      "step": 566330
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 1.0045534552689981e-05,
+      "loss": 0.6864,
+      "step": 566340
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 1.0044386622790234e-05,
+      "loss": 0.6844,
+      "step": 566350
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 1.0043238692890486e-05,
+      "loss": 0.6941,
+      "step": 566360
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 1.0042090762990739e-05,
+      "loss": 0.7867,
+      "step": 566370
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 1.0040942833090992e-05,
+      "loss": 0.6712,
+      "step": 566380
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 1.0039794903191244e-05,
+      "loss": 0.7101,
+      "step": 566390
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 1.0038646973291497e-05,
+      "loss": 0.6848,
+      "step": 566400
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 1.003749904339175e-05,
+      "loss": 0.8105,
+      "step": 566410
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 1.0036351113492002e-05,
+      "loss": 0.7919,
+      "step": 566420
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 1.0035203183592255e-05,
+      "loss": 0.7126,
+      "step": 566430
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 1.0034055253692508e-05,
+      "loss": 0.7409,
+      "step": 566440
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 1.003290732379276e-05,
+      "loss": 0.7556,
+      "step": 566450
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 1.0031759393893013e-05,
+      "loss": 0.6969,
+      "step": 566460
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 1.0030611463993266e-05,
+      "loss": 0.7103,
+      "step": 566470
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 1.0029463534093516e-05,
+      "loss": 0.6574,
+      "step": 566480
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 1.002831560419377e-05,
+      "loss": 0.7297,
+      "step": 566490
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 1.0027167674294023e-05,
+      "loss": 0.6642,
+      "step": 566500
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 1.0026019744394274e-05,
+      "loss": 0.6209,
+      "step": 566510
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 1.0024871814494527e-05,
+      "loss": 0.6596,
+      "step": 566520
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 1.0023723884594779e-05,
+      "loss": 0.7921,
+      "step": 566530
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 1.0022575954695032e-05,
+      "loss": 0.7241,
+      "step": 566540
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 1.0021428024795285e-05,
+      "loss": 0.6808,
+      "step": 566550
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 1.0020280094895537e-05,
+      "loss": 0.7684,
+      "step": 566560
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 1.001913216499579e-05,
+      "loss": 0.6698,
+      "step": 566570
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 1.0017984235096044e-05,
+      "loss": 0.6996,
+      "step": 566580
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 1.0016836305196295e-05,
+      "loss": 0.6868,
+      "step": 566590
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 1.0015688375296548e-05,
+      "loss": 0.758,
+      "step": 566600
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 1.0014540445396802e-05,
+      "loss": 0.7224,
+      "step": 566610
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 1.0013392515497051e-05,
+      "loss": 0.7241,
+      "step": 566620
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 1.0012244585597305e-05,
+      "loss": 0.6747,
+      "step": 566630
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 1.0011096655697558e-05,
+      "loss": 0.7593,
+      "step": 566640
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 1.000994872579781e-05,
+      "loss": 0.6613,
+      "step": 566650
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 1.0008800795898063e-05,
+      "loss": 0.8364,
+      "step": 566660
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 1.0007652865998316e-05,
+      "loss": 0.6809,
+      "step": 566670
+    },
+    {
+      "epoch": 26.0,
+      "eval_cer": 0.6688305611294126,
+      "eval_em": 0.007357212003872217,
+      "eval_f1": 0.007357212003872217,
+      "eval_loss": 1.3603357076644897,
+      "eval_runtime": 1595.3047,
+      "eval_samples_per_second": 6.475,
+      "eval_steps_per_second": 0.81,
+      "eval_wer": 0.9926427879961278,
+      "step": 566677
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 1.0006504936098567e-05,
+      "loss": 0.6377,
+      "step": 566680
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 1.000535700619882e-05,
+      "loss": 0.7207,
+      "step": 566690
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 1.0004209076299074e-05,
+      "loss": 0.6376,
+      "step": 566700
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 1.0003061146399326e-05,
+      "loss": 0.7155,
+      "step": 566710
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 1.0001913216499579e-05,
+      "loss": 0.6772,
+      "step": 566720
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 1.0000765286599832e-05,
+      "loss": 0.699,
+      "step": 566730
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 9.999617356700084e-06,
+      "loss": 0.8079,
+      "step": 566740
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 9.998469426800337e-06,
+      "loss": 0.705,
+      "step": 566750
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 9.99732149690059e-06,
+      "loss": 0.6679,
+      "step": 566760
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 9.99617356700084e-06,
+      "loss": 0.6775,
+      "step": 566770
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 9.995025637101093e-06,
+      "loss": 0.7288,
+      "step": 566780
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 9.993877707201346e-06,
+      "loss": 0.6945,
+      "step": 566790
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 9.992729777301598e-06,
+      "loss": 0.6927,
+      "step": 566800
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 9.991581847401851e-06,
+      "loss": 0.6396,
+      "step": 566810
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 9.990433917502104e-06,
+      "loss": 0.671,
+      "step": 566820
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 9.989285987602356e-06,
+      "loss": 0.6786,
+      "step": 566830
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 9.98813805770261e-06,
+      "loss": 0.7492,
+      "step": 566840
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 9.986990127802862e-06,
+      "loss": 0.655,
+      "step": 566850
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 9.985842197903114e-06,
+      "loss": 0.6854,
+      "step": 566860
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 9.984694268003367e-06,
+      "loss": 0.7888,
+      "step": 566870
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 9.983546338103619e-06,
+      "loss": 0.8303,
+      "step": 566880
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 9.982398408203872e-06,
+      "loss": 0.7269,
+      "step": 566890
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 9.981250478304125e-06,
+      "loss": 0.7261,
+      "step": 566900
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 9.980102548404375e-06,
+      "loss": 0.6813,
+      "step": 566910
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 9.978954618504628e-06,
+      "loss": 0.7133,
+      "step": 566920
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 9.977806688604882e-06,
+      "loss": 0.68,
+      "step": 566930
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 9.976658758705133e-06,
+      "loss": 0.6877,
+      "step": 566940
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 9.975510828805386e-06,
+      "loss": 0.7103,
+      "step": 566950
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 9.97436289890564e-06,
+      "loss": 0.7447,
+      "step": 566960
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 9.973214969005891e-06,
+      "loss": 0.7274,
+      "step": 566970
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 9.972067039106144e-06,
+      "loss": 0.7126,
+      "step": 566980
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 9.970919109206398e-06,
+      "loss": 0.6956,
+      "step": 566990
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 9.96977117930665e-06,
+      "loss": 0.6605,
+      "step": 567000
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 9.968623249406902e-06,
+      "loss": 0.694,
+      "step": 567010
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 9.967475319507156e-06,
+      "loss": 0.7105,
+      "step": 567020
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 9.966327389607407e-06,
+      "loss": 0.7342,
+      "step": 567030
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 9.96517945970766e-06,
+      "loss": 0.7421,
+      "step": 567040
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 9.964031529807914e-06,
+      "loss": 0.7021,
+      "step": 567050
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 9.962883599908164e-06,
+      "loss": 0.7606,
+      "step": 567060
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 9.961735670008417e-06,
+      "loss": 0.689,
+      "step": 567070
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 9.96058774010867e-06,
+      "loss": 0.6799,
+      "step": 567080
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 9.959439810208922e-06,
+      "loss": 0.6868,
+      "step": 567090
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 9.958291880309175e-06,
+      "loss": 0.6673,
+      "step": 567100
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 9.957143950409428e-06,
+      "loss": 0.6236,
+      "step": 567110
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 9.95599602050968e-06,
+      "loss": 0.6638,
+      "step": 567120
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 9.954848090609933e-06,
+      "loss": 0.679,
+      "step": 567130
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 9.953700160710186e-06,
+      "loss": 0.627,
+      "step": 567140
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 9.952552230810438e-06,
+      "loss": 0.7345,
+      "step": 567150
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 9.951404300910691e-06,
+      "loss": 0.7557,
+      "step": 567160
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 9.950256371010944e-06,
+      "loss": 0.7271,
+      "step": 567170
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 9.949108441111196e-06,
+      "loss": 0.7055,
+      "step": 567180
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 9.947960511211449e-06,
+      "loss": 0.6141,
+      "step": 567190
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 9.946812581311702e-06,
+      "loss": 0.636,
+      "step": 567200
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 9.945664651411952e-06,
+      "loss": 0.7236,
+      "step": 567210
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 9.944516721512205e-06,
+      "loss": 0.727,
+      "step": 567220
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 9.943368791612457e-06,
+      "loss": 0.6295,
+      "step": 567230
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 9.94222086171271e-06,
+      "loss": 0.7306,
+      "step": 567240
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 9.941072931812963e-06,
+      "loss": 0.6931,
+      "step": 567250
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 9.939925001913215e-06,
+      "loss": 0.8092,
+      "step": 567260
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 9.938777072013468e-06,
+      "loss": 0.7058,
+      "step": 567270
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 9.937629142113721e-06,
+      "loss": 0.7317,
+      "step": 567280
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 9.936481212213973e-06,
+      "loss": 0.7732,
+      "step": 567290
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 9.935333282314226e-06,
+      "loss": 0.6302,
+      "step": 567300
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 9.93418535241448e-06,
+      "loss": 0.7267,
+      "step": 567310
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 9.93303742251473e-06,
+      "loss": 0.6948,
+      "step": 567320
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 9.931889492614984e-06,
+      "loss": 0.6825,
+      "step": 567330
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 9.930741562715237e-06,
+      "loss": 0.8175,
+      "step": 567340
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 9.929593632815487e-06,
+      "loss": 0.6987,
+      "step": 567350
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 9.92844570291574e-06,
+      "loss": 0.6923,
+      "step": 567360
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 9.927297773015994e-06,
+      "loss": 0.6788,
+      "step": 567370
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 9.926149843116245e-06,
+      "loss": 0.751,
+      "step": 567380
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 9.925001913216499e-06,
+      "loss": 0.667,
+      "step": 567390
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 9.923853983316752e-06,
+      "loss": 0.7406,
+      "step": 567400
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 9.922706053417003e-06,
+      "loss": 0.7296,
+      "step": 567410
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 9.921558123517257e-06,
+      "loss": 0.6397,
+      "step": 567420
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 9.92041019361751e-06,
+      "loss": 0.7004,
+      "step": 567430
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 9.919262263717761e-06,
+      "loss": 0.7399,
+      "step": 567440
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 9.918114333818015e-06,
+      "loss": 0.6901,
+      "step": 567450
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 9.916966403918268e-06,
+      "loss": 0.7535,
+      "step": 567460
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 9.915818474018518e-06,
+      "loss": 0.6216,
+      "step": 567470
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 9.914670544118773e-06,
+      "loss": 0.7152,
+      "step": 567480
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 9.913522614219026e-06,
+      "loss": 0.7135,
+      "step": 567490
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 9.912374684319276e-06,
+      "loss": 0.688,
+      "step": 567500
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 9.911226754419529e-06,
+      "loss": 0.7524,
+      "step": 567510
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 9.910078824519782e-06,
+      "loss": 0.7567,
+      "step": 567520
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 9.908930894620034e-06,
+      "loss": 0.7098,
+      "step": 567530
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 9.907782964720287e-06,
+      "loss": 0.7448,
+      "step": 567540
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 9.906635034820539e-06,
+      "loss": 0.7806,
+      "step": 567550
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 9.905487104920792e-06,
+      "loss": 0.6753,
+      "step": 567560
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 9.904339175021045e-06,
+      "loss": 0.6653,
+      "step": 567570
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 9.903191245121297e-06,
+      "loss": 0.7994,
+      "step": 567580
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 9.90204331522155e-06,
+      "loss": 0.638,
+      "step": 567590
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 9.900895385321803e-06,
+      "loss": 0.6981,
+      "step": 567600
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 9.899747455422053e-06,
+      "loss": 0.723,
+      "step": 567610
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 9.898599525522306e-06,
+      "loss": 0.6914,
+      "step": 567620
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 9.897451595622561e-06,
+      "loss": 0.6717,
+      "step": 567630
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 9.896303665722811e-06,
+      "loss": 0.7301,
+      "step": 567640
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 9.895155735823064e-06,
+      "loss": 0.7313,
+      "step": 567650
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 9.894007805923317e-06,
+      "loss": 0.7469,
+      "step": 567660
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 9.892859876023569e-06,
+      "loss": 0.7472,
+      "step": 567670
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 9.891711946123822e-06,
+      "loss": 0.694,
+      "step": 567680
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 9.890564016224075e-06,
+      "loss": 0.6991,
+      "step": 567690
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 9.889416086324327e-06,
+      "loss": 0.7147,
+      "step": 567700
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 9.88826815642458e-06,
+      "loss": 0.7313,
+      "step": 567710
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 9.887120226524834e-06,
+      "loss": 0.708,
+      "step": 567720
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 9.885972296625085e-06,
+      "loss": 0.7229,
+      "step": 567730
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 9.884824366725338e-06,
+      "loss": 0.7073,
+      "step": 567740
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 9.883676436825592e-06,
+      "loss": 0.702,
+      "step": 567750
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 9.882528506925841e-06,
+      "loss": 0.7115,
+      "step": 567760
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 9.881380577026095e-06,
+      "loss": 0.7854,
+      "step": 567770
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 9.88023264712635e-06,
+      "loss": 0.7388,
+      "step": 567780
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 9.8790847172266e-06,
+      "loss": 0.6755,
+      "step": 567790
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 9.877936787326853e-06,
+      "loss": 0.6836,
+      "step": 567800
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 9.876788857427106e-06,
+      "loss": 0.7943,
+      "step": 567810
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 9.875640927527357e-06,
+      "loss": 0.7631,
+      "step": 567820
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 9.87449299762761e-06,
+      "loss": 0.743,
+      "step": 567830
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 9.873345067727864e-06,
+      "loss": 0.7312,
+      "step": 567840
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 9.872197137828116e-06,
+      "loss": 0.6991,
+      "step": 567850
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 9.871049207928369e-06,
+      "loss": 0.6736,
+      "step": 567860
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 9.869901278028622e-06,
+      "loss": 0.7297,
+      "step": 567870
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 9.868753348128874e-06,
+      "loss": 0.694,
+      "step": 567880
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 9.867605418229127e-06,
+      "loss": 0.6575,
+      "step": 567890
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 9.866457488329377e-06,
+      "loss": 0.6644,
+      "step": 567900
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 9.86530955842963e-06,
+      "loss": 0.735,
+      "step": 567910
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 9.864161628529883e-06,
+      "loss": 0.7315,
+      "step": 567920
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 9.863013698630135e-06,
+      "loss": 0.7043,
+      "step": 567930
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 9.861865768730388e-06,
+      "loss": 0.7096,
+      "step": 567940
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 9.860717838830641e-06,
+      "loss": 0.6215,
+      "step": 567950
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 9.859569908930893e-06,
+      "loss": 0.6956,
+      "step": 567960
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 9.858421979031146e-06,
+      "loss": 0.6916,
+      "step": 567970
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 9.8572740491314e-06,
+      "loss": 0.7622,
+      "step": 567980
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 9.85612611923165e-06,
+      "loss": 0.722,
+      "step": 567990
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 9.854978189331904e-06,
+      "loss": 0.7333,
+      "step": 568000
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 9.853830259432157e-06,
+      "loss": 0.7243,
+      "step": 568010
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 9.852682329532409e-06,
+      "loss": 0.7568,
+      "step": 568020
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 9.851534399632662e-06,
+      "loss": 0.725,
+      "step": 568030
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 9.850386469732915e-06,
+      "loss": 0.7155,
+      "step": 568040
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 9.849238539833165e-06,
+      "loss": 0.6886,
+      "step": 568050
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 9.848090609933418e-06,
+      "loss": 0.6627,
+      "step": 568060
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 9.846942680033673e-06,
+      "loss": 0.7197,
+      "step": 568070
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 9.845794750133923e-06,
+      "loss": 0.6952,
+      "step": 568080
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 9.844646820234176e-06,
+      "loss": 0.6903,
+      "step": 568090
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 9.84349889033443e-06,
+      "loss": 0.7146,
+      "step": 568100
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 9.842350960434681e-06,
+      "loss": 0.7515,
+      "step": 568110
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 9.841203030534934e-06,
+      "loss": 0.7232,
+      "step": 568120
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 9.840055100635188e-06,
+      "loss": 0.7052,
+      "step": 568130
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 9.83890717073544e-06,
+      "loss": 0.6908,
+      "step": 568140
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 9.837759240835692e-06,
+      "loss": 0.6708,
+      "step": 568150
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 9.836611310935946e-06,
+      "loss": 0.74,
+      "step": 568160
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 9.835463381036197e-06,
+      "loss": 0.7511,
+      "step": 568170
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 9.83431545113645e-06,
+      "loss": 0.7052,
+      "step": 568180
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 9.833167521236704e-06,
+      "loss": 0.7515,
+      "step": 568190
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 9.832019591336954e-06,
+      "loss": 0.659,
+      "step": 568200
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 9.830871661437207e-06,
+      "loss": 0.6914,
+      "step": 568210
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 9.829723731537462e-06,
+      "loss": 0.7268,
+      "step": 568220
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 9.828575801637712e-06,
+      "loss": 0.6856,
+      "step": 568230
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 9.827427871737965e-06,
+      "loss": 0.724,
+      "step": 568240
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 9.826279941838216e-06,
+      "loss": 0.7885,
+      "step": 568250
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 9.82513201193847e-06,
+      "loss": 0.6758,
+      "step": 568260
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 9.823984082038723e-06,
+      "loss": 0.7079,
+      "step": 568270
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 9.822836152138974e-06,
+      "loss": 0.7106,
+      "step": 568280
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 9.821688222239228e-06,
+      "loss": 0.7427,
+      "step": 568290
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 9.820540292339481e-06,
+      "loss": 0.7138,
+      "step": 568300
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 9.819392362439732e-06,
+      "loss": 0.6982,
+      "step": 568310
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 9.818244432539986e-06,
+      "loss": 0.7828,
+      "step": 568320
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 9.817096502640239e-06,
+      "loss": 0.7483,
+      "step": 568330
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 9.815948572740489e-06,
+      "loss": 0.7273,
+      "step": 568340
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 9.814800642840742e-06,
+      "loss": 0.7783,
+      "step": 568350
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 9.813652712940995e-06,
+      "loss": 0.6917,
+      "step": 568360
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 9.812504783041247e-06,
+      "loss": 0.7161,
+      "step": 568370
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 9.8113568531415e-06,
+      "loss": 0.7386,
+      "step": 568380
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 9.810208923241753e-06,
+      "loss": 0.7493,
+      "step": 568390
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 9.809060993342005e-06,
+      "loss": 0.708,
+      "step": 568400
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 9.807913063442258e-06,
+      "loss": 0.7283,
+      "step": 568410
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 9.806765133542511e-06,
+      "loss": 0.7157,
+      "step": 568420
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 9.805617203642763e-06,
+      "loss": 0.6676,
+      "step": 568430
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 9.804469273743016e-06,
+      "loss": 0.6869,
+      "step": 568440
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 9.80332134384327e-06,
+      "loss": 0.7201,
+      "step": 568450
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 9.802173413943521e-06,
+      "loss": 0.7568,
+      "step": 568460
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 9.801025484043774e-06,
+      "loss": 0.6478,
+      "step": 568470
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 9.799877554144027e-06,
+      "loss": 0.7665,
+      "step": 568480
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 9.798729624244277e-06,
+      "loss": 0.6716,
+      "step": 568490
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 9.79758169434453e-06,
+      "loss": 0.6458,
+      "step": 568500
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 9.796433764444784e-06,
+      "loss": 0.7099,
+      "step": 568510
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 9.795285834545035e-06,
+      "loss": 0.5807,
+      "step": 568520
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 9.794137904645289e-06,
+      "loss": 0.6451,
+      "step": 568530
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 9.792989974745542e-06,
+      "loss": 0.6236,
+      "step": 568540
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 9.791842044845793e-06,
+      "loss": 0.7944,
+      "step": 568550
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 9.790694114946047e-06,
+      "loss": 0.7147,
+      "step": 568560
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 9.7895461850463e-06,
+      "loss": 0.7116,
+      "step": 568570
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 9.788398255146551e-06,
+      "loss": 0.6558,
+      "step": 568580
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 9.787250325246805e-06,
+      "loss": 0.6458,
+      "step": 568590
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 9.786102395347056e-06,
+      "loss": 0.7587,
+      "step": 568600
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 9.78495446544731e-06,
+      "loss": 0.7627,
+      "step": 568610
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 9.783806535547563e-06,
+      "loss": 0.7114,
+      "step": 568620
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 9.782658605647813e-06,
+      "loss": 0.7491,
+      "step": 568630
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 9.781510675748066e-06,
+      "loss": 0.7224,
+      "step": 568640
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 9.780362745848319e-06,
+      "loss": 0.7014,
+      "step": 568650
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 9.77921481594857e-06,
+      "loss": 0.6777,
+      "step": 568660
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 9.778066886048824e-06,
+      "loss": 0.717,
+      "step": 568670
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 9.776918956149077e-06,
+      "loss": 0.7078,
+      "step": 568680
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 9.775771026249329e-06,
+      "loss": 0.6707,
+      "step": 568690
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 9.774623096349582e-06,
+      "loss": 0.7416,
+      "step": 568700
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 9.773475166449835e-06,
+      "loss": 0.6887,
+      "step": 568710
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 9.772327236550087e-06,
+      "loss": 0.6201,
+      "step": 568720
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 9.77117930665034e-06,
+      "loss": 0.671,
+      "step": 568730
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 9.770031376750593e-06,
+      "loss": 0.6741,
+      "step": 568740
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 9.768883446850845e-06,
+      "loss": 0.6543,
+      "step": 568750
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 9.767735516951098e-06,
+      "loss": 0.7306,
+      "step": 568760
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 9.766587587051351e-06,
+      "loss": 0.6702,
+      "step": 568770
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 9.765439657151601e-06,
+      "loss": 0.7403,
+      "step": 568780
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 9.764291727251854e-06,
+      "loss": 0.6793,
+      "step": 568790
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 9.763143797352107e-06,
+      "loss": 0.6508,
+      "step": 568800
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 9.761995867452359e-06,
+      "loss": 0.7133,
+      "step": 568810
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 9.760847937552612e-06,
+      "loss": 0.7169,
+      "step": 568820
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 9.759700007652866e-06,
+      "loss": 0.7232,
+      "step": 568830
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 9.758552077753117e-06,
+      "loss": 0.7151,
+      "step": 568840
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 9.75740414785337e-06,
+      "loss": 0.6537,
+      "step": 568850
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 9.756256217953624e-06,
+      "loss": 0.6654,
+      "step": 568860
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 9.755108288053875e-06,
+      "loss": 0.5749,
+      "step": 568870
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 9.753960358154128e-06,
+      "loss": 0.6996,
+      "step": 568880
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 9.752812428254382e-06,
+      "loss": 0.6416,
+      "step": 568890
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 9.751664498354633e-06,
+      "loss": 0.6829,
+      "step": 568900
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 9.750516568454886e-06,
+      "loss": 0.7141,
+      "step": 568910
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 9.74936863855514e-06,
+      "loss": 0.7262,
+      "step": 568920
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 9.74822070865539e-06,
+      "loss": 0.7668,
+      "step": 568930
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 9.747072778755643e-06,
+      "loss": 0.6967,
+      "step": 568940
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 9.745924848855894e-06,
+      "loss": 0.7272,
+      "step": 568950
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 9.744776918956148e-06,
+      "loss": 0.7453,
+      "step": 568960
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 9.7436289890564e-06,
+      "loss": 0.7545,
+      "step": 568970
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 9.742481059156652e-06,
+      "loss": 0.6989,
+      "step": 568980
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 9.741333129256906e-06,
+      "loss": 0.7083,
+      "step": 568990
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 9.740185199357159e-06,
+      "loss": 0.7127,
+      "step": 569000
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 9.73903726945741e-06,
+      "loss": 0.6466,
+      "step": 569010
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 9.737889339557664e-06,
+      "loss": 0.737,
+      "step": 569020
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 9.736741409657917e-06,
+      "loss": 0.6552,
+      "step": 569030
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 9.735593479758168e-06,
+      "loss": 0.652,
+      "step": 569040
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 9.734445549858422e-06,
+      "loss": 0.7496,
+      "step": 569050
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 9.733297619958675e-06,
+      "loss": 0.6568,
+      "step": 569060
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 9.732149690058925e-06,
+      "loss": 0.6633,
+      "step": 569070
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 9.731001760159178e-06,
+      "loss": 0.7726,
+      "step": 569080
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 9.729853830259431e-06,
+      "loss": 0.7244,
+      "step": 569090
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 9.728705900359683e-06,
+      "loss": 0.7062,
+      "step": 569100
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 9.727557970459936e-06,
+      "loss": 0.6652,
+      "step": 569110
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 9.72641004056019e-06,
+      "loss": 0.6831,
+      "step": 569120
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 9.72526211066044e-06,
+      "loss": 0.6496,
+      "step": 569130
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 9.724114180760694e-06,
+      "loss": 0.7136,
+      "step": 569140
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 9.722966250860947e-06,
+      "loss": 0.662,
+      "step": 569150
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 9.721818320961199e-06,
+      "loss": 0.7313,
+      "step": 569160
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 9.720670391061452e-06,
+      "loss": 0.7305,
+      "step": 569170
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 9.719522461161705e-06,
+      "loss": 0.6533,
+      "step": 569180
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 9.718374531261957e-06,
+      "loss": 0.7411,
+      "step": 569190
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 9.71722660136221e-06,
+      "loss": 0.7534,
+      "step": 569200
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 9.716078671462463e-06,
+      "loss": 0.7139,
+      "step": 569210
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 9.714930741562713e-06,
+      "loss": 0.7098,
+      "step": 569220
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 9.713782811662966e-06,
+      "loss": 0.7917,
+      "step": 569230
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 9.71263488176322e-06,
+      "loss": 0.6979,
+      "step": 569240
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 9.711486951863471e-06,
+      "loss": 0.7734,
+      "step": 569250
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 9.710339021963724e-06,
+      "loss": 0.7253,
+      "step": 569260
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 9.709191092063976e-06,
+      "loss": 0.7136,
+      "step": 569270
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 9.70804316216423e-06,
+      "loss": 0.7439,
+      "step": 569280
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 9.706895232264482e-06,
+      "loss": 0.6977,
+      "step": 569290
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 9.705747302364734e-06,
+      "loss": 0.6625,
+      "step": 569300
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 9.704599372464987e-06,
+      "loss": 0.7063,
+      "step": 569310
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 9.70345144256524e-06,
+      "loss": 0.7652,
+      "step": 569320
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 9.702303512665492e-06,
+      "loss": 0.6711,
+      "step": 569330
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 9.701155582765745e-06,
+      "loss": 0.7581,
+      "step": 569340
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 9.700007652865999e-06,
+      "loss": 0.7506,
+      "step": 569350
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 9.698859722966248e-06,
+      "loss": 0.7078,
+      "step": 569360
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 9.697711793066502e-06,
+      "loss": 0.6709,
+      "step": 569370
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 9.696563863166755e-06,
+      "loss": 0.6416,
+      "step": 569380
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 9.695415933267006e-06,
+      "loss": 0.7171,
+      "step": 569390
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 9.69426800336726e-06,
+      "loss": 0.6806,
+      "step": 569400
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 9.693120073467513e-06,
+      "loss": 0.7069,
+      "step": 569410
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 9.691972143567764e-06,
+      "loss": 0.7232,
+      "step": 569420
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 9.690824213668018e-06,
+      "loss": 0.6956,
+      "step": 569430
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 9.689676283768271e-06,
+      "loss": 0.6807,
+      "step": 569440
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 9.688528353868523e-06,
+      "loss": 0.6254,
+      "step": 569450
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 9.687380423968776e-06,
+      "loss": 0.7478,
+      "step": 569460
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 9.686232494069029e-06,
+      "loss": 0.6735,
+      "step": 569470
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 9.68508456416928e-06,
+      "loss": 0.6992,
+      "step": 569480
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 9.683936634269534e-06,
+      "loss": 0.6196,
+      "step": 569490
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 9.682788704369787e-06,
+      "loss": 0.7146,
+      "step": 569500
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 9.681640774470037e-06,
+      "loss": 0.7205,
+      "step": 569510
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 9.68049284457029e-06,
+      "loss": 0.6832,
+      "step": 569520
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 9.679344914670543e-06,
+      "loss": 0.6571,
+      "step": 569530
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 9.678196984770795e-06,
+      "loss": 0.721,
+      "step": 569540
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 9.677049054871048e-06,
+      "loss": 0.6942,
+      "step": 569550
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 9.675901124971301e-06,
+      "loss": 0.7315,
+      "step": 569560
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 9.674753195071553e-06,
+      "loss": 0.6687,
+      "step": 569570
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 9.673605265171806e-06,
+      "loss": 0.7377,
+      "step": 569580
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 9.67245733527206e-06,
+      "loss": 0.6552,
+      "step": 569590
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 9.671309405372311e-06,
+      "loss": 0.7585,
+      "step": 569600
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 9.670161475472564e-06,
+      "loss": 0.7269,
+      "step": 569610
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 9.669013545572816e-06,
+      "loss": 0.7745,
+      "step": 569620
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 9.667865615673069e-06,
+      "loss": 0.6997,
+      "step": 569630
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 9.666717685773322e-06,
+      "loss": 0.7247,
+      "step": 569640
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 9.665569755873572e-06,
+      "loss": 0.6941,
+      "step": 569650
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 9.664421825973825e-06,
+      "loss": 0.6886,
+      "step": 569660
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 9.663273896074079e-06,
+      "loss": 0.7383,
+      "step": 569670
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 9.66212596617433e-06,
+      "loss": 0.6476,
+      "step": 569680
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 9.660978036274583e-06,
+      "loss": 0.7309,
+      "step": 569690
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 9.659830106374837e-06,
+      "loss": 0.6738,
+      "step": 569700
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 9.658682176475088e-06,
+      "loss": 0.6451,
+      "step": 569710
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 9.657534246575341e-06,
+      "loss": 0.6554,
+      "step": 569720
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 9.656386316675595e-06,
+      "loss": 0.6703,
+      "step": 569730
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 9.655238386775846e-06,
+      "loss": 0.707,
+      "step": 569740
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 9.6540904568761e-06,
+      "loss": 0.6239,
+      "step": 569750
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 9.652942526976353e-06,
+      "loss": 0.6949,
+      "step": 569760
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 9.651794597076604e-06,
+      "loss": 0.6765,
+      "step": 569770
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 9.650646667176857e-06,
+      "loss": 0.6752,
+      "step": 569780
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 9.64949873727711e-06,
+      "loss": 0.7226,
+      "step": 569790
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 9.64835080737736e-06,
+      "loss": 0.615,
+      "step": 569800
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 9.647202877477614e-06,
+      "loss": 0.7354,
+      "step": 569810
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 9.646054947577867e-06,
+      "loss": 0.6181,
+      "step": 569820
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 9.644907017678119e-06,
+      "loss": 0.723,
+      "step": 569830
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 9.643759087778372e-06,
+      "loss": 0.6954,
+      "step": 569840
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 9.642611157878625e-06,
+      "loss": 0.7508,
+      "step": 569850
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 9.641463227978877e-06,
+      "loss": 0.68,
+      "step": 569860
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 9.64031529807913e-06,
+      "loss": 0.7237,
+      "step": 569870
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 9.639167368179383e-06,
+      "loss": 0.6538,
+      "step": 569880
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 9.638019438279635e-06,
+      "loss": 0.7263,
+      "step": 569890
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 9.636871508379888e-06,
+      "loss": 0.6975,
+      "step": 569900
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 9.635723578480141e-06,
+      "loss": 0.6892,
+      "step": 569910
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 9.634575648580393e-06,
+      "loss": 0.6671,
+      "step": 569920
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 9.633427718680646e-06,
+      "loss": 0.7105,
+      "step": 569930
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 9.6322797887809e-06,
+      "loss": 0.7611,
+      "step": 569940
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 9.631131858881149e-06,
+      "loss": 0.7113,
+      "step": 569950
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 9.629983928981402e-06,
+      "loss": 0.7255,
+      "step": 569960
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 9.628835999081654e-06,
+      "loss": 0.7219,
+      "step": 569970
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 9.627688069181907e-06,
+      "loss": 0.6547,
+      "step": 569980
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 9.62654013928216e-06,
+      "loss": 0.6596,
+      "step": 569990
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 9.625392209382412e-06,
+      "loss": 0.7216,
+      "step": 570000
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 9.624244279482665e-06,
+      "loss": 0.7257,
+      "step": 570010
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 9.623096349582918e-06,
+      "loss": 0.6817,
+      "step": 570020
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 9.62194841968317e-06,
+      "loss": 0.6733,
+      "step": 570030
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 9.620800489783423e-06,
+      "loss": 0.8363,
+      "step": 570040
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 9.619652559883676e-06,
+      "loss": 0.6814,
+      "step": 570050
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 9.618504629983928e-06,
+      "loss": 0.6715,
+      "step": 570060
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 9.617356700084181e-06,
+      "loss": 0.6719,
+      "step": 570070
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 9.616208770184434e-06,
+      "loss": 0.6427,
+      "step": 570080
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 9.615060840284684e-06,
+      "loss": 0.7352,
+      "step": 570090
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 9.613912910384938e-06,
+      "loss": 0.65,
+      "step": 570100
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 9.61276498048519e-06,
+      "loss": 0.7456,
+      "step": 570110
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 9.611617050585442e-06,
+      "loss": 0.7256,
+      "step": 570120
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 9.610469120685696e-06,
+      "loss": 0.7067,
+      "step": 570130
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 9.609321190785949e-06,
+      "loss": 0.7222,
+      "step": 570140
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 9.6081732608862e-06,
+      "loss": 0.7195,
+      "step": 570150
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 9.607025330986454e-06,
+      "loss": 0.707,
+      "step": 570160
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 9.605877401086707e-06,
+      "loss": 0.7016,
+      "step": 570170
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 9.604729471186958e-06,
+      "loss": 0.7829,
+      "step": 570180
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 9.603581541287212e-06,
+      "loss": 0.7127,
+      "step": 570190
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 9.602433611387465e-06,
+      "loss": 0.8012,
+      "step": 570200
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 9.601285681487716e-06,
+      "loss": 0.7353,
+      "step": 570210
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 9.60013775158797e-06,
+      "loss": 0.6623,
+      "step": 570220
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 9.598989821688223e-06,
+      "loss": 0.7625,
+      "step": 570230
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 9.597841891788473e-06,
+      "loss": 0.7029,
+      "step": 570240
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 9.596693961888726e-06,
+      "loss": 0.7776,
+      "step": 570250
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 9.59554603198898e-06,
+      "loss": 0.7079,
+      "step": 570260
+    },
+    {
+      "epoch": 26.16,
+      "learning_rate": 9.59439810208923e-06,
+      "loss": 0.7292,
+      "step": 570270
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 9.593250172189484e-06,
+      "loss": 0.7134,
+      "step": 570280
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 9.592102242289737e-06,
+      "loss": 0.688,
+      "step": 570290
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 9.590954312389989e-06,
+      "loss": 0.6914,
+      "step": 570300
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 9.589806382490242e-06,
+      "loss": 0.6601,
+      "step": 570310
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 9.588658452590494e-06,
+      "loss": 0.6964,
+      "step": 570320
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 9.587510522690747e-06,
+      "loss": 0.6317,
+      "step": 570330
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 9.586362592791e-06,
+      "loss": 0.6846,
+      "step": 570340
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 9.585214662891252e-06,
+      "loss": 0.7497,
+      "step": 570350
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 9.584066732991505e-06,
+      "loss": 0.6701,
+      "step": 570360
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 9.582918803091758e-06,
+      "loss": 0.6577,
+      "step": 570370
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 9.581770873192008e-06,
+      "loss": 0.7701,
+      "step": 570380
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 9.580622943292261e-06,
+      "loss": 0.6677,
+      "step": 570390
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 9.579475013392514e-06,
+      "loss": 0.7422,
+      "step": 570400
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 9.578327083492766e-06,
+      "loss": 0.7143,
+      "step": 570410
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 9.57717915359302e-06,
+      "loss": 0.7438,
+      "step": 570420
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 9.576031223693272e-06,
+      "loss": 0.744,
+      "step": 570430
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 9.574883293793524e-06,
+      "loss": 0.7758,
+      "step": 570440
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 9.573735363893777e-06,
+      "loss": 0.6514,
+      "step": 570450
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 9.57258743399403e-06,
+      "loss": 0.7254,
+      "step": 570460
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 9.571439504094282e-06,
+      "loss": 0.7349,
+      "step": 570470
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 9.570291574194535e-06,
+      "loss": 0.7877,
+      "step": 570480
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 9.569143644294789e-06,
+      "loss": 0.6964,
+      "step": 570490
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 9.56799571439504e-06,
+      "loss": 0.7569,
+      "step": 570500
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 9.566847784495293e-06,
+      "loss": 0.7182,
+      "step": 570510
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 9.565699854595547e-06,
+      "loss": 0.7191,
+      "step": 570520
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 9.564551924695796e-06,
+      "loss": 0.7405,
+      "step": 570530
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 9.56340399479605e-06,
+      "loss": 0.7461,
+      "step": 570540
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 9.562256064896303e-06,
+      "loss": 0.6792,
+      "step": 570550
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 9.561108134996554e-06,
+      "loss": 0.7159,
+      "step": 570560
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 9.559960205096808e-06,
+      "loss": 0.6709,
+      "step": 570570
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 9.558812275197061e-06,
+      "loss": 0.7082,
+      "step": 570580
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 9.557664345297313e-06,
+      "loss": 0.7019,
+      "step": 570590
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 9.556516415397566e-06,
+      "loss": 0.8158,
+      "step": 570600
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 9.555368485497819e-06,
+      "loss": 0.7794,
+      "step": 570610
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 9.55422055559807e-06,
+      "loss": 0.7753,
+      "step": 570620
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 9.553072625698324e-06,
+      "loss": 0.7036,
+      "step": 570630
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 9.551924695798575e-06,
+      "loss": 0.6914,
+      "step": 570640
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 9.550776765898829e-06,
+      "loss": 0.6941,
+      "step": 570650
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 9.549628835999082e-06,
+      "loss": 0.7416,
+      "step": 570660
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 9.548480906099332e-06,
+      "loss": 0.6444,
+      "step": 570670
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 9.547332976199585e-06,
+      "loss": 0.6832,
+      "step": 570680
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 9.546185046299838e-06,
+      "loss": 0.7365,
+      "step": 570690
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 9.54503711640009e-06,
+      "loss": 0.6927,
+      "step": 570700
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 9.543889186500343e-06,
+      "loss": 0.7755,
+      "step": 570710
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 9.542741256600596e-06,
+      "loss": 0.6959,
+      "step": 570720
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 9.541593326700848e-06,
+      "loss": 0.7578,
+      "step": 570730
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 9.540445396801101e-06,
+      "loss": 0.6291,
+      "step": 570740
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 9.539297466901354e-06,
+      "loss": 0.6773,
+      "step": 570750
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 9.538149537001606e-06,
+      "loss": 0.7227,
+      "step": 570760
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 9.537001607101859e-06,
+      "loss": 0.7436,
+      "step": 570770
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 9.535853677202112e-06,
+      "loss": 0.6735,
+      "step": 570780
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 9.534705747302364e-06,
+      "loss": 0.6756,
+      "step": 570790
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 9.533557817402617e-06,
+      "loss": 0.7134,
+      "step": 570800
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 9.53240988750287e-06,
+      "loss": 0.6642,
+      "step": 570810
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 9.53126195760312e-06,
+      "loss": 0.7152,
+      "step": 570820
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 9.530114027703373e-06,
+      "loss": 0.7416,
+      "step": 570830
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 9.528966097803627e-06,
+      "loss": 0.7001,
+      "step": 570840
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 9.527818167903878e-06,
+      "loss": 0.6316,
+      "step": 570850
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 9.526670238004131e-06,
+      "loss": 0.6927,
+      "step": 570860
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 9.525522308104385e-06,
+      "loss": 0.6345,
+      "step": 570870
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 9.524374378204636e-06,
+      "loss": 0.7117,
+      "step": 570880
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 9.52322644830489e-06,
+      "loss": 0.7367,
+      "step": 570890
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 9.522078518405143e-06,
+      "loss": 0.7119,
+      "step": 570900
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 9.520930588505394e-06,
+      "loss": 0.7041,
+      "step": 570910
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 9.519782658605647e-06,
+      "loss": 0.686,
+      "step": 570920
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 9.5186347287059e-06,
+      "loss": 0.7384,
+      "step": 570930
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 9.517486798806152e-06,
+      "loss": 0.7059,
+      "step": 570940
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 9.516338868906406e-06,
+      "loss": 0.723,
+      "step": 570950
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 9.515190939006659e-06,
+      "loss": 0.6919,
+      "step": 570960
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 9.514043009106909e-06,
+      "loss": 0.7125,
+      "step": 570970
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 9.512895079207162e-06,
+      "loss": 0.6507,
+      "step": 570980
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 9.511747149307413e-06,
+      "loss": 0.7022,
+      "step": 570990
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 9.510599219407667e-06,
+      "loss": 0.661,
+      "step": 571000
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 9.50945128950792e-06,
+      "loss": 0.7685,
+      "step": 571010
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 9.508303359608171e-06,
+      "loss": 0.8008,
+      "step": 571020
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 9.507155429708425e-06,
+      "loss": 0.6977,
+      "step": 571030
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 9.506007499808678e-06,
+      "loss": 0.7349,
+      "step": 571040
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 9.50485956990893e-06,
+      "loss": 0.7871,
+      "step": 571050
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 9.503711640009183e-06,
+      "loss": 0.7384,
+      "step": 571060
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 9.502563710109436e-06,
+      "loss": 0.6528,
+      "step": 571070
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 9.501415780209688e-06,
+      "loss": 0.7196,
+      "step": 571080
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 9.50026785030994e-06,
+      "loss": 0.6336,
+      "step": 571090
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 9.499119920410194e-06,
+      "loss": 0.7224,
+      "step": 571100
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 9.497971990510444e-06,
+      "loss": 0.6538,
+      "step": 571110
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 9.496824060610697e-06,
+      "loss": 0.733,
+      "step": 571120
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 9.49567613071095e-06,
+      "loss": 0.7596,
+      "step": 571130
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 9.494528200811202e-06,
+      "loss": 0.7018,
+      "step": 571140
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 9.493380270911455e-06,
+      "loss": 0.7518,
+      "step": 571150
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 9.492232341011708e-06,
+      "loss": 0.746,
+      "step": 571160
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 9.49108441111196e-06,
+      "loss": 0.6935,
+      "step": 571170
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 9.489936481212213e-06,
+      "loss": 0.6961,
+      "step": 571180
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 9.488788551312466e-06,
+      "loss": 0.7198,
+      "step": 571190
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 9.487640621412718e-06,
+      "loss": 0.6552,
+      "step": 571200
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 9.486492691512971e-06,
+      "loss": 0.6904,
+      "step": 571210
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 9.485344761613224e-06,
+      "loss": 0.6879,
+      "step": 571220
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 9.484196831713476e-06,
+      "loss": 0.7104,
+      "step": 571230
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 9.48304890181373e-06,
+      "loss": 0.6771,
+      "step": 571240
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 9.481900971913982e-06,
+      "loss": 0.713,
+      "step": 571250
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 9.480753042014232e-06,
+      "loss": 0.7325,
+      "step": 571260
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 9.479605112114486e-06,
+      "loss": 0.6307,
+      "step": 571270
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 9.478457182214739e-06,
+      "loss": 0.7154,
+      "step": 571280
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 9.47730925231499e-06,
+      "loss": 0.682,
+      "step": 571290
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 9.476161322415244e-06,
+      "loss": 0.7166,
+      "step": 571300
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 9.475013392515497e-06,
+      "loss": 0.6762,
+      "step": 571310
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 9.473865462615748e-06,
+      "loss": 0.7073,
+      "step": 571320
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 9.472717532716002e-06,
+      "loss": 0.6092,
+      "step": 571330
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 9.471569602816253e-06,
+      "loss": 0.7229,
+      "step": 571340
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 9.470421672916506e-06,
+      "loss": 0.7113,
+      "step": 571350
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 9.46927374301676e-06,
+      "loss": 0.749,
+      "step": 571360
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 9.468125813117011e-06,
+      "loss": 0.7244,
+      "step": 571370
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 9.466977883217264e-06,
+      "loss": 0.7161,
+      "step": 571380
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 9.465829953317518e-06,
+      "loss": 0.714,
+      "step": 571390
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 9.464682023417768e-06,
+      "loss": 0.7201,
+      "step": 571400
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 9.46353409351802e-06,
+      "loss": 0.7603,
+      "step": 571410
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 9.462386163618274e-06,
+      "loss": 0.7303,
+      "step": 571420
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 9.461238233718526e-06,
+      "loss": 0.6922,
+      "step": 571430
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 9.460090303818779e-06,
+      "loss": 0.7556,
+      "step": 571440
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 9.458942373919032e-06,
+      "loss": 0.7315,
+      "step": 571450
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 9.457794444019284e-06,
+      "loss": 0.7126,
+      "step": 571460
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 9.456646514119537e-06,
+      "loss": 0.6934,
+      "step": 571470
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 9.45549858421979e-06,
+      "loss": 0.6982,
+      "step": 571480
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 9.454350654320042e-06,
+      "loss": 0.6934,
+      "step": 571490
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 9.453202724420295e-06,
+      "loss": 0.6938,
+      "step": 571500
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 9.452054794520548e-06,
+      "loss": 0.6892,
+      "step": 571510
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 9.4509068646208e-06,
+      "loss": 0.7368,
+      "step": 571520
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 9.449758934721053e-06,
+      "loss": 0.6822,
+      "step": 571530
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 9.448611004821306e-06,
+      "loss": 0.709,
+      "step": 571540
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 9.447463074921556e-06,
+      "loss": 0.7232,
+      "step": 571550
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 9.44631514502181e-06,
+      "loss": 0.664,
+      "step": 571560
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 9.445167215122063e-06,
+      "loss": 0.8045,
+      "step": 571570
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 9.444019285222314e-06,
+      "loss": 0.8092,
+      "step": 571580
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 9.442871355322567e-06,
+      "loss": 0.7232,
+      "step": 571590
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 9.44172342542282e-06,
+      "loss": 0.638,
+      "step": 571600
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 9.440575495523072e-06,
+      "loss": 0.6678,
+      "step": 571610
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 9.439427565623325e-06,
+      "loss": 0.7299,
+      "step": 571620
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 9.438279635723579e-06,
+      "loss": 0.6814,
+      "step": 571630
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 9.43713170582383e-06,
+      "loss": 0.6819,
+      "step": 571640
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 9.435983775924083e-06,
+      "loss": 0.7599,
+      "step": 571650
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 9.434835846024337e-06,
+      "loss": 0.7309,
+      "step": 571660
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 9.433687916124588e-06,
+      "loss": 0.7165,
+      "step": 571670
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 9.432539986224841e-06,
+      "loss": 0.6601,
+      "step": 571680
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 9.431392056325091e-06,
+      "loss": 0.6552,
+      "step": 571690
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 9.430244126425344e-06,
+      "loss": 0.7278,
+      "step": 571700
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 9.429096196525598e-06,
+      "loss": 0.6345,
+      "step": 571710
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 9.42794826662585e-06,
+      "loss": 0.6433,
+      "step": 571720
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 9.426800336726103e-06,
+      "loss": 0.6662,
+      "step": 571730
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 9.425652406826356e-06,
+      "loss": 0.6687,
+      "step": 571740
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 9.424504476926607e-06,
+      "loss": 0.6696,
+      "step": 571750
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 9.42335654702686e-06,
+      "loss": 0.6518,
+      "step": 571760
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 9.422208617127114e-06,
+      "loss": 0.6934,
+      "step": 571770
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 9.421060687227365e-06,
+      "loss": 0.6344,
+      "step": 571780
+    },
+    {
+      "epoch": 26.23,
+      "learning_rate": 9.419912757327619e-06,
+      "loss": 0.7325,
+      "step": 571790
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 9.418764827427872e-06,
+      "loss": 0.7364,
+      "step": 571800
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 9.417616897528123e-06,
+      "loss": 0.7033,
+      "step": 571810
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 9.416468967628377e-06,
+      "loss": 0.6792,
+      "step": 571820
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 9.41532103772863e-06,
+      "loss": 0.7582,
+      "step": 571830
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 9.41417310782888e-06,
+      "loss": 0.699,
+      "step": 571840
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 9.413025177929133e-06,
+      "loss": 0.7128,
+      "step": 571850
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 9.411877248029386e-06,
+      "loss": 0.7757,
+      "step": 571860
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 9.410729318129638e-06,
+      "loss": 0.735,
+      "step": 571870
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 9.409581388229891e-06,
+      "loss": 0.7372,
+      "step": 571880
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 9.408433458330144e-06,
+      "loss": 0.7515,
+      "step": 571890
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 9.407285528430396e-06,
+      "loss": 0.8412,
+      "step": 571900
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 9.406137598530649e-06,
+      "loss": 0.7795,
+      "step": 571910
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 9.404989668630902e-06,
+      "loss": 0.7192,
+      "step": 571920
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 9.403841738731154e-06,
+      "loss": 0.716,
+      "step": 571930
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 9.402693808831407e-06,
+      "loss": 0.732,
+      "step": 571940
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 9.40154587893166e-06,
+      "loss": 0.7393,
+      "step": 571950
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 9.400397949031912e-06,
+      "loss": 0.7236,
+      "step": 571960
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 9.399250019132165e-06,
+      "loss": 0.7136,
+      "step": 571970
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 9.398102089232418e-06,
+      "loss": 0.6404,
+      "step": 571980
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 9.396954159332668e-06,
+      "loss": 0.7094,
+      "step": 571990
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 9.395806229432921e-06,
+      "loss": 0.6599,
+      "step": 572000
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 9.394658299533173e-06,
+      "loss": 0.762,
+      "step": 572010
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 9.393510369633426e-06,
+      "loss": 0.7299,
+      "step": 572020
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 9.39236243973368e-06,
+      "loss": 0.7374,
+      "step": 572030
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 9.391214509833931e-06,
+      "loss": 0.6721,
+      "step": 572040
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 9.390066579934184e-06,
+      "loss": 0.6394,
+      "step": 572050
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 9.388918650034438e-06,
+      "loss": 0.6476,
+      "step": 572060
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 9.387770720134689e-06,
+      "loss": 0.7215,
+      "step": 572070
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 9.386622790234942e-06,
+      "loss": 0.6417,
+      "step": 572080
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 9.385474860335196e-06,
+      "loss": 0.6954,
+      "step": 572090
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 9.384326930435447e-06,
+      "loss": 0.6922,
+      "step": 572100
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 9.3831790005357e-06,
+      "loss": 0.7141,
+      "step": 572110
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 9.382031070635954e-06,
+      "loss": 0.7902,
+      "step": 572120
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 9.380883140736203e-06,
+      "loss": 0.793,
+      "step": 572130
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 9.379735210836457e-06,
+      "loss": 0.7394,
+      "step": 572140
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 9.37858728093671e-06,
+      "loss": 0.7001,
+      "step": 572150
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 9.377439351036961e-06,
+      "loss": 0.6186,
+      "step": 572160
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 9.376291421137215e-06,
+      "loss": 0.7979,
+      "step": 572170
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 9.375143491237468e-06,
+      "loss": 0.7379,
+      "step": 572180
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 9.373995561337721e-06,
+      "loss": 0.7197,
+      "step": 572190
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 9.372847631437973e-06,
+      "loss": 0.6594,
+      "step": 572200
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 9.371699701538224e-06,
+      "loss": 0.8005,
+      "step": 572210
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 9.370551771638478e-06,
+      "loss": 0.6545,
+      "step": 572220
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 9.36940384173873e-06,
+      "loss": 0.6517,
+      "step": 572230
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 9.368255911838982e-06,
+      "loss": 0.7265,
+      "step": 572240
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 9.367107981939236e-06,
+      "loss": 0.7719,
+      "step": 572250
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 9.365960052039489e-06,
+      "loss": 0.7531,
+      "step": 572260
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 9.36481212213974e-06,
+      "loss": 0.6981,
+      "step": 572270
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 9.363664192239992e-06,
+      "loss": 0.6838,
+      "step": 572280
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 9.362516262340245e-06,
+      "loss": 0.6684,
+      "step": 572290
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 9.361368332440498e-06,
+      "loss": 0.693,
+      "step": 572300
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 9.36022040254075e-06,
+      "loss": 0.7118,
+      "step": 572310
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 9.359072472641003e-06,
+      "loss": 0.6749,
+      "step": 572320
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 9.357924542741256e-06,
+      "loss": 0.7199,
+      "step": 572330
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 9.356776612841508e-06,
+      "loss": 0.7565,
+      "step": 572340
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 9.355628682941761e-06,
+      "loss": 0.7229,
+      "step": 572350
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 9.354480753042013e-06,
+      "loss": 0.7355,
+      "step": 572360
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 9.353332823142266e-06,
+      "loss": 0.7063,
+      "step": 572370
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 9.35218489324252e-06,
+      "loss": 0.6918,
+      "step": 572380
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 9.35103696334277e-06,
+      "loss": 0.7168,
+      "step": 572390
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 9.349889033443024e-06,
+      "loss": 0.659,
+      "step": 572400
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 9.348741103543277e-06,
+      "loss": 0.6565,
+      "step": 572410
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 9.347593173643529e-06,
+      "loss": 0.6935,
+      "step": 572420
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 9.34644524374378e-06,
+      "loss": 0.6471,
+      "step": 572430
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 9.345297313844034e-06,
+      "loss": 0.6615,
+      "step": 572440
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 9.344149383944287e-06,
+      "loss": 0.6746,
+      "step": 572450
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 9.343001454044538e-06,
+      "loss": 0.7203,
+      "step": 572460
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 9.341853524144792e-06,
+      "loss": 0.6938,
+      "step": 572470
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 9.340705594245045e-06,
+      "loss": 0.7341,
+      "step": 572480
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 9.339557664345296e-06,
+      "loss": 0.6781,
+      "step": 572490
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 9.338409734445548e-06,
+      "loss": 0.6881,
+      "step": 572500
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 9.337261804545801e-06,
+      "loss": 0.624,
+      "step": 572510
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 9.336113874646054e-06,
+      "loss": 0.7411,
+      "step": 572520
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 9.334965944746306e-06,
+      "loss": 0.7341,
+      "step": 572530
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 9.33381801484656e-06,
+      "loss": 0.7322,
+      "step": 572540
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 9.332670084946813e-06,
+      "loss": 0.7264,
+      "step": 572550
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 9.331522155047064e-06,
+      "loss": 0.7267,
+      "step": 572560
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 9.330374225147317e-06,
+      "loss": 0.7707,
+      "step": 572570
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 9.329226295247569e-06,
+      "loss": 0.6874,
+      "step": 572580
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 9.328078365347822e-06,
+      "loss": 0.6934,
+      "step": 572590
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 9.326930435448075e-06,
+      "loss": 0.6932,
+      "step": 572600
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 9.325782505548327e-06,
+      "loss": 0.6545,
+      "step": 572610
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 9.32463457564858e-06,
+      "loss": 0.6729,
+      "step": 572620
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 9.323486645748832e-06,
+      "loss": 0.6884,
+      "step": 572630
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 9.322338715849085e-06,
+      "loss": 0.7436,
+      "step": 572640
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 9.321190785949336e-06,
+      "loss": 0.6467,
+      "step": 572650
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 9.32004285604959e-06,
+      "loss": 0.725,
+      "step": 572660
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 9.318894926149843e-06,
+      "loss": 0.7855,
+      "step": 572670
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 9.317746996250094e-06,
+      "loss": 0.6732,
+      "step": 572680
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 9.316599066350348e-06,
+      "loss": 0.6685,
+      "step": 572690
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 9.315451136450601e-06,
+      "loss": 0.6861,
+      "step": 572700
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 9.314303206550853e-06,
+      "loss": 0.7142,
+      "step": 572710
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 9.313155276651104e-06,
+      "loss": 0.7682,
+      "step": 572720
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 9.312007346751357e-06,
+      "loss": 0.7253,
+      "step": 572730
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 9.31085941685161e-06,
+      "loss": 0.6629,
+      "step": 572740
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 9.309711486951862e-06,
+      "loss": 0.7785,
+      "step": 572750
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 9.308563557052115e-06,
+      "loss": 0.6747,
+      "step": 572760
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 9.307415627152369e-06,
+      "loss": 0.7683,
+      "step": 572770
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 9.30626769725262e-06,
+      "loss": 0.7187,
+      "step": 572780
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 9.305119767352872e-06,
+      "loss": 0.7274,
+      "step": 572790
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 9.303971837453125e-06,
+      "loss": 0.6612,
+      "step": 572800
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 9.302823907553378e-06,
+      "loss": 0.6859,
+      "step": 572810
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 9.30167597765363e-06,
+      "loss": 0.6893,
+      "step": 572820
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 9.300528047753883e-06,
+      "loss": 0.6992,
+      "step": 572830
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 9.299380117854136e-06,
+      "loss": 0.667,
+      "step": 572840
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 9.298232187954388e-06,
+      "loss": 0.7044,
+      "step": 572850
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 9.297084258054641e-06,
+      "loss": 0.7881,
+      "step": 572860
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 9.295936328154893e-06,
+      "loss": 0.7844,
+      "step": 572870
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 9.294788398255146e-06,
+      "loss": 0.745,
+      "step": 572880
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 9.293640468355399e-06,
+      "loss": 0.7617,
+      "step": 572890
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 9.29249253845565e-06,
+      "loss": 0.7504,
+      "step": 572900
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 9.291344608555904e-06,
+      "loss": 0.6636,
+      "step": 572910
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 9.290196678656157e-06,
+      "loss": 0.7466,
+      "step": 572920
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 9.289048748756409e-06,
+      "loss": 0.7055,
+      "step": 572930
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 9.28790081885666e-06,
+      "loss": 0.714,
+      "step": 572940
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 9.286752888956913e-06,
+      "loss": 0.771,
+      "step": 572950
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 9.285604959057167e-06,
+      "loss": 0.7053,
+      "step": 572960
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 9.284457029157418e-06,
+      "loss": 0.7251,
+      "step": 572970
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 9.283309099257671e-06,
+      "loss": 0.6729,
+      "step": 572980
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 9.282161169357925e-06,
+      "loss": 0.7165,
+      "step": 572990
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 9.281013239458176e-06,
+      "loss": 0.686,
+      "step": 573000
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 9.279865309558428e-06,
+      "loss": 0.7229,
+      "step": 573010
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 9.278717379658681e-06,
+      "loss": 0.6547,
+      "step": 573020
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 9.277569449758934e-06,
+      "loss": 0.6937,
+      "step": 573030
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 9.276421519859186e-06,
+      "loss": 0.6998,
+      "step": 573040
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 9.275273589959439e-06,
+      "loss": 0.7574,
+      "step": 573050
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 9.274125660059692e-06,
+      "loss": 0.6901,
+      "step": 573060
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 9.272977730159944e-06,
+      "loss": 0.7182,
+      "step": 573070
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 9.271829800260197e-06,
+      "loss": 0.7623,
+      "step": 573080
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 9.270681870360449e-06,
+      "loss": 0.702,
+      "step": 573090
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 9.269533940460702e-06,
+      "loss": 0.7355,
+      "step": 573100
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 9.268386010560955e-06,
+      "loss": 0.6129,
+      "step": 573110
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 9.267238080661207e-06,
+      "loss": 0.704,
+      "step": 573120
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 9.26609015076146e-06,
+      "loss": 0.6302,
+      "step": 573130
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 9.264942220861711e-06,
+      "loss": 0.6993,
+      "step": 573140
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 9.263794290961965e-06,
+      "loss": 0.7397,
+      "step": 573150
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 9.262646361062216e-06,
+      "loss": 0.7202,
+      "step": 573160
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 9.26149843116247e-06,
+      "loss": 0.6772,
+      "step": 573170
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 9.260350501262723e-06,
+      "loss": 0.6507,
+      "step": 573180
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 9.259202571362974e-06,
+      "loss": 0.6817,
+      "step": 573190
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 9.258054641463228e-06,
+      "loss": 0.7301,
+      "step": 573200
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 9.25690671156348e-06,
+      "loss": 0.6957,
+      "step": 573210
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 9.255758781663732e-06,
+      "loss": 0.6653,
+      "step": 573220
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 9.254610851763984e-06,
+      "loss": 0.7189,
+      "step": 573230
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 9.253462921864237e-06,
+      "loss": 0.7318,
+      "step": 573240
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 9.25231499196449e-06,
+      "loss": 0.7268,
+      "step": 573250
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 9.251167062064742e-06,
+      "loss": 0.7117,
+      "step": 573260
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 9.250019132164995e-06,
+      "loss": 0.6928,
+      "step": 573270
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 9.248871202265248e-06,
+      "loss": 0.7267,
+      "step": 573280
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 9.2477232723655e-06,
+      "loss": 0.6627,
+      "step": 573290
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 9.246575342465751e-06,
+      "loss": 0.675,
+      "step": 573300
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 9.245427412566005e-06,
+      "loss": 0.6886,
+      "step": 573310
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 9.244279482666258e-06,
+      "loss": 0.711,
+      "step": 573320
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 9.24313155276651e-06,
+      "loss": 0.7052,
+      "step": 573330
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 9.241983622866763e-06,
+      "loss": 0.6982,
+      "step": 573340
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 9.240835692967016e-06,
+      "loss": 0.7099,
+      "step": 573350
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 9.239687763067268e-06,
+      "loss": 0.6714,
+      "step": 573360
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 9.23853983316752e-06,
+      "loss": 0.7421,
+      "step": 573370
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 9.237391903267772e-06,
+      "loss": 0.6604,
+      "step": 573380
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 9.236243973368026e-06,
+      "loss": 0.7188,
+      "step": 573390
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 9.235096043468279e-06,
+      "loss": 0.6519,
+      "step": 573400
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 9.23394811356853e-06,
+      "loss": 0.7499,
+      "step": 573410
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 9.232800183668784e-06,
+      "loss": 0.7542,
+      "step": 573420
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 9.231652253769037e-06,
+      "loss": 0.6618,
+      "step": 573430
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 9.230504323869288e-06,
+      "loss": 0.7224,
+      "step": 573440
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 9.22935639396954e-06,
+      "loss": 0.6951,
+      "step": 573450
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 9.228208464069793e-06,
+      "loss": 0.6955,
+      "step": 573460
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 9.227060534170046e-06,
+      "loss": 0.7246,
+      "step": 573470
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 9.225912604270298e-06,
+      "loss": 0.7646,
+      "step": 573480
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 9.224764674370551e-06,
+      "loss": 0.7563,
+      "step": 573490
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 9.223616744470804e-06,
+      "loss": 0.7315,
+      "step": 573500
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 9.222468814571056e-06,
+      "loss": 0.712,
+      "step": 573510
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 9.221320884671308e-06,
+      "loss": 0.7385,
+      "step": 573520
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 9.22017295477156e-06,
+      "loss": 0.7088,
+      "step": 573530
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 9.219025024871814e-06,
+      "loss": 0.7414,
+      "step": 573540
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 9.217877094972066e-06,
+      "loss": 0.6802,
+      "step": 573550
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 9.216729165072319e-06,
+      "loss": 0.605,
+      "step": 573560
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 9.215581235172572e-06,
+      "loss": 0.6684,
+      "step": 573570
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 9.214433305272824e-06,
+      "loss": 0.6535,
+      "step": 573580
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 9.213285375373077e-06,
+      "loss": 0.5906,
+      "step": 573590
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 9.212137445473328e-06,
+      "loss": 0.7331,
+      "step": 573600
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 9.210989515573582e-06,
+      "loss": 0.6734,
+      "step": 573610
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 9.209841585673835e-06,
+      "loss": 0.7474,
+      "step": 573620
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 9.208693655774086e-06,
+      "loss": 0.7506,
+      "step": 573630
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 9.20754572587434e-06,
+      "loss": 0.6435,
+      "step": 573640
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 9.206397795974591e-06,
+      "loss": 0.6718,
+      "step": 573650
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 9.205249866074844e-06,
+      "loss": 0.7736,
+      "step": 573660
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 9.204101936175096e-06,
+      "loss": 0.7267,
+      "step": 573670
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 9.20295400627535e-06,
+      "loss": 0.7171,
+      "step": 573680
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 9.201806076375603e-06,
+      "loss": 0.7758,
+      "step": 573690
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 9.200658146475854e-06,
+      "loss": 0.7922,
+      "step": 573700
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 9.199510216576107e-06,
+      "loss": 0.6583,
+      "step": 573710
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 9.19836228667636e-06,
+      "loss": 0.7114,
+      "step": 573720
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 9.197214356776612e-06,
+      "loss": 0.699,
+      "step": 573730
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 9.196066426876864e-06,
+      "loss": 0.6503,
+      "step": 573740
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 9.194918496977117e-06,
+      "loss": 0.6758,
+      "step": 573750
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 9.19377056707737e-06,
+      "loss": 0.7427,
+      "step": 573760
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 9.192622637177622e-06,
+      "loss": 0.6884,
+      "step": 573770
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 9.191474707277875e-06,
+      "loss": 0.6603,
+      "step": 573780
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 9.190326777378128e-06,
+      "loss": 0.6575,
+      "step": 573790
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 9.18917884747838e-06,
+      "loss": 0.7454,
+      "step": 573800
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 9.188030917578633e-06,
+      "loss": 0.7129,
+      "step": 573810
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 9.186882987678885e-06,
+      "loss": 0.7257,
+      "step": 573820
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 9.185735057779138e-06,
+      "loss": 0.6743,
+      "step": 573830
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 9.18458712787939e-06,
+      "loss": 0.7012,
+      "step": 573840
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 9.183439197979643e-06,
+      "loss": 0.7348,
+      "step": 573850
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 9.182291268079896e-06,
+      "loss": 0.738,
+      "step": 573860
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 9.181143338180147e-06,
+      "loss": 0.6903,
+      "step": 573870
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 9.1799954082804e-06,
+      "loss": 0.7252,
+      "step": 573880
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 9.178847478380652e-06,
+      "loss": 0.7479,
+      "step": 573890
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 9.177699548480905e-06,
+      "loss": 0.7192,
+      "step": 573900
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 9.176551618581159e-06,
+      "loss": 0.7065,
+      "step": 573910
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 9.17540368868141e-06,
+      "loss": 0.6362,
+      "step": 573920
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 9.174255758781663e-06,
+      "loss": 0.7661,
+      "step": 573930
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 9.173107828881917e-06,
+      "loss": 0.7353,
+      "step": 573940
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 9.171959898982168e-06,
+      "loss": 0.7324,
+      "step": 573950
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 9.17081196908242e-06,
+      "loss": 0.8353,
+      "step": 573960
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 9.169664039182673e-06,
+      "loss": 0.7384,
+      "step": 573970
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 9.168516109282926e-06,
+      "loss": 0.6693,
+      "step": 573980
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 9.167368179383178e-06,
+      "loss": 0.6286,
+      "step": 573990
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 9.166220249483431e-06,
+      "loss": 0.6779,
+      "step": 574000
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 9.165072319583684e-06,
+      "loss": 0.6793,
+      "step": 574010
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 9.163924389683936e-06,
+      "loss": 0.6888,
+      "step": 574020
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 9.162776459784187e-06,
+      "loss": 0.7181,
+      "step": 574030
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 9.16162852988444e-06,
+      "loss": 0.7679,
+      "step": 574040
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 9.160480599984694e-06,
+      "loss": 0.7301,
+      "step": 574050
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 9.159332670084945e-06,
+      "loss": 0.6782,
+      "step": 574060
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 9.158184740185199e-06,
+      "loss": 0.6927,
+      "step": 574070
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 9.157036810285452e-06,
+      "loss": 0.6351,
+      "step": 574080
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 9.155888880385703e-06,
+      "loss": 0.7263,
+      "step": 574090
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 9.154740950485957e-06,
+      "loss": 0.7622,
+      "step": 574100
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 9.153593020586208e-06,
+      "loss": 0.764,
+      "step": 574110
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 9.152445090686461e-06,
+      "loss": 0.6502,
+      "step": 574120
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 9.151297160786715e-06,
+      "loss": 0.7661,
+      "step": 574130
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 9.150149230886966e-06,
+      "loss": 0.7,
+      "step": 574140
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 9.14900130098722e-06,
+      "loss": 0.6962,
+      "step": 574150
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 9.147853371087471e-06,
+      "loss": 0.6648,
+      "step": 574160
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 9.146705441187724e-06,
+      "loss": 0.7264,
+      "step": 574170
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 9.145557511287976e-06,
+      "loss": 0.7391,
+      "step": 574180
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 9.144409581388229e-06,
+      "loss": 0.7804,
+      "step": 574190
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 9.143261651488482e-06,
+      "loss": 0.7572,
+      "step": 574200
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 9.142113721588734e-06,
+      "loss": 0.7295,
+      "step": 574210
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 9.140965791688987e-06,
+      "loss": 0.7539,
+      "step": 574220
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 9.13981786178924e-06,
+      "loss": 0.6815,
+      "step": 574230
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 9.138669931889492e-06,
+      "loss": 0.7198,
+      "step": 574240
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 9.137522001989743e-06,
+      "loss": 0.6271,
+      "step": 574250
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 9.136374072089997e-06,
+      "loss": 0.7396,
+      "step": 574260
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 9.13522614219025e-06,
+      "loss": 0.7293,
+      "step": 574270
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 9.134078212290501e-06,
+      "loss": 0.6438,
+      "step": 574280
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 9.132930282390755e-06,
+      "loss": 0.7784,
+      "step": 574290
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 9.131782352491008e-06,
+      "loss": 0.729,
+      "step": 574300
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 9.13063442259126e-06,
+      "loss": 0.7124,
+      "step": 574310
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 9.129486492691513e-06,
+      "loss": 0.6935,
+      "step": 574320
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 9.128338562791764e-06,
+      "loss": 0.7345,
+      "step": 574330
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 9.127190632892018e-06,
+      "loss": 0.7422,
+      "step": 574340
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 9.126042702992269e-06,
+      "loss": 0.7426,
+      "step": 574350
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 9.124894773092522e-06,
+      "loss": 0.7563,
+      "step": 574360
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 9.123746843192776e-06,
+      "loss": 0.6667,
+      "step": 574370
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 9.122598913293027e-06,
+      "loss": 0.6609,
+      "step": 574380
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 9.12145098339328e-06,
+      "loss": 0.7155,
+      "step": 574390
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 9.120303053493532e-06,
+      "loss": 0.7159,
+      "step": 574400
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 9.119155123593785e-06,
+      "loss": 0.6806,
+      "step": 574410
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 9.118007193694038e-06,
+      "loss": 0.7001,
+      "step": 574420
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 9.11685926379429e-06,
+      "loss": 0.6989,
+      "step": 574430
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 9.115711333894543e-06,
+      "loss": 0.7265,
+      "step": 574440
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 9.114563403994796e-06,
+      "loss": 0.8407,
+      "step": 574450
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 9.113415474095048e-06,
+      "loss": 0.7398,
+      "step": 574460
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 9.1122675441953e-06,
+      "loss": 0.6829,
+      "step": 574470
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 9.111119614295553e-06,
+      "loss": 0.7327,
+      "step": 574480
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 9.109971684395806e-06,
+      "loss": 0.7478,
+      "step": 574490
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 9.108823754496058e-06,
+      "loss": 0.7097,
+      "step": 574500
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 9.10767582459631e-06,
+      "loss": 0.7189,
+      "step": 574510
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 9.106527894696564e-06,
+      "loss": 0.754,
+      "step": 574520
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 9.105379964796816e-06,
+      "loss": 0.6793,
+      "step": 574530
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 9.104232034897067e-06,
+      "loss": 0.7057,
+      "step": 574540
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 9.10308410499732e-06,
+      "loss": 0.6443,
+      "step": 574550
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 9.101936175097574e-06,
+      "loss": 0.7223,
+      "step": 574560
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 9.100788245197825e-06,
+      "loss": 0.7333,
+      "step": 574570
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 9.099640315298078e-06,
+      "loss": 0.7459,
+      "step": 574580
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 9.098492385398332e-06,
+      "loss": 0.6801,
+      "step": 574590
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 9.097344455498583e-06,
+      "loss": 0.793,
+      "step": 574600
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 9.096196525598836e-06,
+      "loss": 0.7437,
+      "step": 574610
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 9.095048595699088e-06,
+      "loss": 0.7312,
+      "step": 574620
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 9.093900665799341e-06,
+      "loss": 0.6506,
+      "step": 574630
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 9.092752735899594e-06,
+      "loss": 0.7414,
+      "step": 574640
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 9.091604805999846e-06,
+      "loss": 0.742,
+      "step": 574650
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 9.0904568761001e-06,
+      "loss": 0.6324,
+      "step": 574660
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 9.08930894620035e-06,
+      "loss": 0.7484,
+      "step": 574670
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 9.088161016300604e-06,
+      "loss": 0.6587,
+      "step": 574680
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 9.087013086400856e-06,
+      "loss": 0.746,
+      "step": 574690
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 9.085865156501109e-06,
+      "loss": 0.7471,
+      "step": 574700
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 9.084717226601362e-06,
+      "loss": 0.7459,
+      "step": 574710
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 9.083569296701614e-06,
+      "loss": 0.7549,
+      "step": 574720
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 9.082421366801867e-06,
+      "loss": 0.7058,
+      "step": 574730
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 9.08127343690212e-06,
+      "loss": 0.6557,
+      "step": 574740
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 9.080125507002372e-06,
+      "loss": 0.6706,
+      "step": 574750
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 9.078977577102623e-06,
+      "loss": 0.6684,
+      "step": 574760
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 9.077829647202876e-06,
+      "loss": 0.7444,
+      "step": 574770
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 9.07668171730313e-06,
+      "loss": 0.7027,
+      "step": 574780
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 9.075533787403381e-06,
+      "loss": 0.7093,
+      "step": 574790
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 9.074385857503635e-06,
+      "loss": 0.7331,
+      "step": 574800
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 9.073237927603888e-06,
+      "loss": 0.6785,
+      "step": 574810
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 9.07208999770414e-06,
+      "loss": 0.6956,
+      "step": 574820
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 9.070942067804393e-06,
+      "loss": 0.6782,
+      "step": 574830
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 9.069794137904644e-06,
+      "loss": 0.7073,
+      "step": 574840
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 9.068646208004897e-06,
+      "loss": 0.6477,
+      "step": 574850
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 9.067498278105149e-06,
+      "loss": 0.7421,
+      "step": 574860
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 9.066350348205402e-06,
+      "loss": 0.7371,
+      "step": 574870
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 9.065202418305655e-06,
+      "loss": 0.6138,
+      "step": 574880
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 9.064054488405907e-06,
+      "loss": 0.7151,
+      "step": 574890
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 9.06290655850616e-06,
+      "loss": 0.7245,
+      "step": 574900
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 9.061758628606412e-06,
+      "loss": 0.7596,
+      "step": 574910
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 9.060610698706665e-06,
+      "loss": 0.6711,
+      "step": 574920
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 9.059462768806918e-06,
+      "loss": 0.6825,
+      "step": 574930
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 9.05831483890717e-06,
+      "loss": 0.7101,
+      "step": 574940
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 9.057166909007423e-06,
+      "loss": 0.6882,
+      "step": 574950
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 9.056018979107676e-06,
+      "loss": 0.7794,
+      "step": 574960
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 9.054871049207928e-06,
+      "loss": 0.6869,
+      "step": 574970
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 9.05372311930818e-06,
+      "loss": 0.6939,
+      "step": 574980
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 9.052575189408433e-06,
+      "loss": 0.7088,
+      "step": 574990
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 9.051427259508686e-06,
+      "loss": 0.7075,
+      "step": 575000
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 9.050279329608937e-06,
+      "loss": 0.635,
+      "step": 575010
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 9.04913139970919e-06,
+      "loss": 0.7514,
+      "step": 575020
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 9.047983469809444e-06,
+      "loss": 0.7425,
+      "step": 575030
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 9.046835539909695e-06,
+      "loss": 0.7698,
+      "step": 575040
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 9.045687610009947e-06,
+      "loss": 0.7247,
+      "step": 575050
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 9.0445396801102e-06,
+      "loss": 0.6832,
+      "step": 575060
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 9.043391750210453e-06,
+      "loss": 0.7041,
+      "step": 575070
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 9.042243820310705e-06,
+      "loss": 0.7845,
+      "step": 575080
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 9.041095890410958e-06,
+      "loss": 0.6076,
+      "step": 575090
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 9.039947960511211e-06,
+      "loss": 0.7494,
+      "step": 575100
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 9.038800030611463e-06,
+      "loss": 0.7172,
+      "step": 575110
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 9.037652100711716e-06,
+      "loss": 0.692,
+      "step": 575120
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 9.036504170811968e-06,
+      "loss": 0.6244,
+      "step": 575130
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 9.035356240912221e-06,
+      "loss": 0.7236,
+      "step": 575140
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 9.034208311012474e-06,
+      "loss": 0.7008,
+      "step": 575150
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 9.033060381112726e-06,
+      "loss": 0.675,
+      "step": 575160
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 9.031912451212979e-06,
+      "loss": 0.6992,
+      "step": 575170
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 9.030764521313232e-06,
+      "loss": 0.7685,
+      "step": 575180
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 9.029616591413484e-06,
+      "loss": 0.7417,
+      "step": 575190
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 9.028468661513735e-06,
+      "loss": 0.7315,
+      "step": 575200
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 9.027320731613989e-06,
+      "loss": 0.7328,
+      "step": 575210
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 9.026172801714242e-06,
+      "loss": 0.7144,
+      "step": 575220
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 9.025024871814493e-06,
+      "loss": 0.706,
+      "step": 575230
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 9.023876941914747e-06,
+      "loss": 0.6227,
+      "step": 575240
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 9.022729012015e-06,
+      "loss": 0.6991,
+      "step": 575250
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 9.021581082115251e-06,
+      "loss": 0.6227,
+      "step": 575260
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 9.020433152215503e-06,
+      "loss": 0.6521,
+      "step": 575270
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 9.019285222315756e-06,
+      "loss": 0.7625,
+      "step": 575280
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 9.01813729241601e-06,
+      "loss": 0.7273,
+      "step": 575290
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 9.016989362516261e-06,
+      "loss": 0.6558,
+      "step": 575300
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 9.015841432616514e-06,
+      "loss": 0.7494,
+      "step": 575310
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 9.014693502716768e-06,
+      "loss": 0.6844,
+      "step": 575320
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 9.013545572817019e-06,
+      "loss": 0.6777,
+      "step": 575330
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 9.012397642917272e-06,
+      "loss": 0.6792,
+      "step": 575340
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 9.011249713017524e-06,
+      "loss": 0.6647,
+      "step": 575350
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 9.010101783117777e-06,
+      "loss": 0.7272,
+      "step": 575360
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 9.008953853218029e-06,
+      "loss": 0.782,
+      "step": 575370
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 9.007805923318282e-06,
+      "loss": 0.752,
+      "step": 575380
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 9.006657993418535e-06,
+      "loss": 0.703,
+      "step": 575390
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 9.005510063518787e-06,
+      "loss": 0.6009,
+      "step": 575400
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 9.00436213361904e-06,
+      "loss": 0.701,
+      "step": 575410
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 9.003214203719291e-06,
+      "loss": 0.7692,
+      "step": 575420
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 9.002066273819545e-06,
+      "loss": 0.7541,
+      "step": 575430
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 9.000918343919798e-06,
+      "loss": 0.6803,
+      "step": 575440
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 8.99977041402005e-06,
+      "loss": 0.6772,
+      "step": 575450
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 8.998622484120303e-06,
+      "loss": 0.6915,
+      "step": 575460
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 8.997474554220556e-06,
+      "loss": 0.7195,
+      "step": 575470
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 8.996326624320808e-06,
+      "loss": 0.7705,
+      "step": 575480
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 8.995178694421059e-06,
+      "loss": 0.7335,
+      "step": 575490
+    },
+    {
+      "epoch": 26.4,
+      "learning_rate": 8.994030764521312e-06,
+      "loss": 0.7591,
+      "step": 575500
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 8.992882834621566e-06,
+      "loss": 0.6932,
+      "step": 575510
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 8.991734904721817e-06,
+      "loss": 0.7345,
+      "step": 575520
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 8.99058697482207e-06,
+      "loss": 0.6345,
+      "step": 575530
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 8.989439044922324e-06,
+      "loss": 0.7696,
+      "step": 575540
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 8.988291115022575e-06,
+      "loss": 0.7679,
+      "step": 575550
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 8.987143185122827e-06,
+      "loss": 0.7656,
+      "step": 575560
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 8.98599525522308e-06,
+      "loss": 0.6985,
+      "step": 575570
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 8.984847325323333e-06,
+      "loss": 0.7126,
+      "step": 575580
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 8.983699395423585e-06,
+      "loss": 0.7192,
+      "step": 575590
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 8.982551465523838e-06,
+      "loss": 0.691,
+      "step": 575600
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 8.981403535624091e-06,
+      "loss": 0.7132,
+      "step": 575610
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 8.980255605724343e-06,
+      "loss": 0.7136,
+      "step": 575620
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 8.979107675824596e-06,
+      "loss": 0.694,
+      "step": 575630
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 8.977959745924848e-06,
+      "loss": 0.7346,
+      "step": 575640
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 8.9768118160251e-06,
+      "loss": 0.709,
+      "step": 575650
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 8.975663886125354e-06,
+      "loss": 0.6609,
+      "step": 575660
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 8.974515956225606e-06,
+      "loss": 0.6929,
+      "step": 575670
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 8.973368026325859e-06,
+      "loss": 0.6319,
+      "step": 575680
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 8.972220096426112e-06,
+      "loss": 0.7765,
+      "step": 575690
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 8.971072166526364e-06,
+      "loss": 0.6246,
+      "step": 575700
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 8.969924236626615e-06,
+      "loss": 0.7098,
+      "step": 575710
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 8.968776306726868e-06,
+      "loss": 0.7095,
+      "step": 575720
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 8.967628376827122e-06,
+      "loss": 0.705,
+      "step": 575730
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 8.966480446927373e-06,
+      "loss": 0.6799,
+      "step": 575740
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 8.965332517027626e-06,
+      "loss": 0.7173,
+      "step": 575750
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 8.96418458712788e-06,
+      "loss": 0.6311,
+      "step": 575760
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 8.963036657228131e-06,
+      "loss": 0.6927,
+      "step": 575770
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 8.961888727328383e-06,
+      "loss": 0.6924,
+      "step": 575780
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 8.960740797428636e-06,
+      "loss": 0.6803,
+      "step": 575790
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 8.95959286752889e-06,
+      "loss": 0.7097,
+      "step": 575800
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 8.95844493762914e-06,
+      "loss": 0.6806,
+      "step": 575810
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 8.957297007729394e-06,
+      "loss": 0.7222,
+      "step": 575820
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 8.956149077829647e-06,
+      "loss": 0.718,
+      "step": 575830
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 8.955001147929899e-06,
+      "loss": 0.7047,
+      "step": 575840
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 8.953853218030152e-06,
+      "loss": 0.6591,
+      "step": 575850
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 8.952705288130404e-06,
+      "loss": 0.7113,
+      "step": 575860
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 8.951557358230657e-06,
+      "loss": 0.7625,
+      "step": 575870
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 8.950409428330908e-06,
+      "loss": 0.6997,
+      "step": 575880
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 8.949261498431162e-06,
+      "loss": 0.7374,
+      "step": 575890
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 8.948113568531415e-06,
+      "loss": 0.692,
+      "step": 575900
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 8.946965638631666e-06,
+      "loss": 0.6636,
+      "step": 575910
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 8.94581770873192e-06,
+      "loss": 0.629,
+      "step": 575920
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 8.944669778832171e-06,
+      "loss": 0.6487,
+      "step": 575930
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 8.943521848932425e-06,
+      "loss": 0.7075,
+      "step": 575940
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 8.942373919032678e-06,
+      "loss": 0.6998,
+      "step": 575950
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 8.94122598913293e-06,
+      "loss": 0.7672,
+      "step": 575960
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 8.940078059233183e-06,
+      "loss": 0.6703,
+      "step": 575970
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 8.938930129333436e-06,
+      "loss": 0.6629,
+      "step": 575980
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 8.937782199433687e-06,
+      "loss": 0.722,
+      "step": 575990
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 8.936634269533939e-06,
+      "loss": 0.6917,
+      "step": 576000
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 8.935486339634192e-06,
+      "loss": 0.6637,
+      "step": 576010
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 8.934338409734445e-06,
+      "loss": 0.7222,
+      "step": 576020
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 8.933190479834697e-06,
+      "loss": 0.705,
+      "step": 576030
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 8.93204254993495e-06,
+      "loss": 0.7068,
+      "step": 576040
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 8.930894620035203e-06,
+      "loss": 0.6628,
+      "step": 576050
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 8.929746690135455e-06,
+      "loss": 0.6638,
+      "step": 576060
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 8.928598760235707e-06,
+      "loss": 0.6707,
+      "step": 576070
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 8.92745083033596e-06,
+      "loss": 0.6976,
+      "step": 576080
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 8.926302900436213e-06,
+      "loss": 0.6945,
+      "step": 576090
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 8.925154970536465e-06,
+      "loss": 0.7805,
+      "step": 576100
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 8.924007040636718e-06,
+      "loss": 0.6365,
+      "step": 576110
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 8.922859110736971e-06,
+      "loss": 0.7417,
+      "step": 576120
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 8.921711180837223e-06,
+      "loss": 0.6836,
+      "step": 576130
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 8.920563250937476e-06,
+      "loss": 0.7932,
+      "step": 576140
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 8.919415321037727e-06,
+      "loss": 0.7343,
+      "step": 576150
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 8.91826739113798e-06,
+      "loss": 0.6741,
+      "step": 576160
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 8.917119461238234e-06,
+      "loss": 0.7401,
+      "step": 576170
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 8.915971531338485e-06,
+      "loss": 0.6818,
+      "step": 576180
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 8.914823601438739e-06,
+      "loss": 0.7551,
+      "step": 576190
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 8.913675671538992e-06,
+      "loss": 0.6925,
+      "step": 576200
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 8.912527741639243e-06,
+      "loss": 0.7549,
+      "step": 576210
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 8.911379811739495e-06,
+      "loss": 0.6932,
+      "step": 576220
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 8.910231881839748e-06,
+      "loss": 0.6695,
+      "step": 576230
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 8.909083951940001e-06,
+      "loss": 0.7601,
+      "step": 576240
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 8.907936022040253e-06,
+      "loss": 0.5947,
+      "step": 576250
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 8.906788092140506e-06,
+      "loss": 0.7335,
+      "step": 576260
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 8.90564016224076e-06,
+      "loss": 0.6823,
+      "step": 576270
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 8.904492232341011e-06,
+      "loss": 0.7159,
+      "step": 576280
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 8.903344302441263e-06,
+      "loss": 0.7086,
+      "step": 576290
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 8.902196372541516e-06,
+      "loss": 0.677,
+      "step": 576300
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 8.901048442641769e-06,
+      "loss": 0.7244,
+      "step": 576310
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 8.89990051274202e-06,
+      "loss": 0.6731,
+      "step": 576320
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 8.898752582842274e-06,
+      "loss": 0.7203,
+      "step": 576330
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 8.897604652942527e-06,
+      "loss": 0.8148,
+      "step": 576340
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 8.896456723042779e-06,
+      "loss": 0.7078,
+      "step": 576350
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 8.895308793143032e-06,
+      "loss": 0.7296,
+      "step": 576360
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 8.894160863243283e-06,
+      "loss": 0.7183,
+      "step": 576370
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 8.893012933343537e-06,
+      "loss": 0.7556,
+      "step": 576380
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 8.891865003443788e-06,
+      "loss": 0.7621,
+      "step": 576390
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 8.890717073544041e-06,
+      "loss": 0.7414,
+      "step": 576400
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 8.889569143644295e-06,
+      "loss": 0.6957,
+      "step": 576410
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 8.888421213744546e-06,
+      "loss": 0.6641,
+      "step": 576420
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 8.8872732838448e-06,
+      "loss": 0.6731,
+      "step": 576430
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 8.886125353945051e-06,
+      "loss": 0.7136,
+      "step": 576440
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 8.884977424045304e-06,
+      "loss": 0.7098,
+      "step": 576450
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 8.883829494145558e-06,
+      "loss": 0.7299,
+      "step": 576460
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 8.882681564245809e-06,
+      "loss": 0.6775,
+      "step": 576470
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 8.881533634346062e-06,
+      "loss": 0.7883,
+      "step": 576480
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 8.880385704446316e-06,
+      "loss": 0.755,
+      "step": 576490
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 8.879237774546567e-06,
+      "loss": 0.7185,
+      "step": 576500
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 8.878089844646819e-06,
+      "loss": 0.7338,
+      "step": 576510
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 8.876941914747072e-06,
+      "loss": 0.6729,
+      "step": 576520
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 8.875793984847325e-06,
+      "loss": 0.7157,
+      "step": 576530
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 8.874646054947577e-06,
+      "loss": 0.7205,
+      "step": 576540
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 8.87349812504783e-06,
+      "loss": 0.6899,
+      "step": 576550
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 8.872350195148083e-06,
+      "loss": 0.7415,
+      "step": 576560
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 8.871202265248335e-06,
+      "loss": 0.7233,
+      "step": 576570
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 8.870054335348586e-06,
+      "loss": 0.6846,
+      "step": 576580
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 8.86890640544884e-06,
+      "loss": 0.6314,
+      "step": 576590
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 8.867758475549093e-06,
+      "loss": 0.6271,
+      "step": 576600
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 8.866610545649344e-06,
+      "loss": 0.6743,
+      "step": 576610
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 8.865462615749598e-06,
+      "loss": 0.7383,
+      "step": 576620
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 8.86431468584985e-06,
+      "loss": 0.6798,
+      "step": 576630
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 8.863166755950102e-06,
+      "loss": 0.6983,
+      "step": 576640
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 8.862018826050356e-06,
+      "loss": 0.6846,
+      "step": 576650
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 8.860870896150607e-06,
+      "loss": 0.696,
+      "step": 576660
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 8.85972296625086e-06,
+      "loss": 0.7105,
+      "step": 576670
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 8.858575036351114e-06,
+      "loss": 0.6236,
+      "step": 576680
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 8.857427106451365e-06,
+      "loss": 0.798,
+      "step": 576690
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 8.856279176551617e-06,
+      "loss": 0.6171,
+      "step": 576700
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 8.855131246651872e-06,
+      "loss": 0.6875,
+      "step": 576710
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 8.853983316752123e-06,
+      "loss": 0.6641,
+      "step": 576720
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 8.852835386852375e-06,
+      "loss": 0.6648,
+      "step": 576730
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 8.851687456952628e-06,
+      "loss": 0.6966,
+      "step": 576740
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 8.850539527052881e-06,
+      "loss": 0.7456,
+      "step": 576750
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 8.849391597153133e-06,
+      "loss": 0.7229,
+      "step": 576760
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 8.848243667253384e-06,
+      "loss": 0.6204,
+      "step": 576770
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 8.84709573735364e-06,
+      "loss": 0.7081,
+      "step": 576780
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 8.84594780745389e-06,
+      "loss": 0.6966,
+      "step": 576790
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 8.844799877554142e-06,
+      "loss": 0.6845,
+      "step": 576800
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 8.843651947654396e-06,
+      "loss": 0.7896,
+      "step": 576810
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 8.842504017754649e-06,
+      "loss": 0.7085,
+      "step": 576820
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 8.8413560878549e-06,
+      "loss": 0.7551,
+      "step": 576830
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 8.840208157955154e-06,
+      "loss": 0.7703,
+      "step": 576840
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 8.839060228055405e-06,
+      "loss": 0.6912,
+      "step": 576850
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 8.837912298155658e-06,
+      "loss": 0.6807,
+      "step": 576860
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 8.836764368255912e-06,
+      "loss": 0.7224,
+      "step": 576870
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 8.835616438356163e-06,
+      "loss": 0.744,
+      "step": 576880
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 8.834468508456416e-06,
+      "loss": 0.6506,
+      "step": 576890
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 8.833320578556668e-06,
+      "loss": 0.7122,
+      "step": 576900
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 8.832172648656921e-06,
+      "loss": 0.6316,
+      "step": 576910
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 8.831024718757173e-06,
+      "loss": 0.6916,
+      "step": 576920
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 8.829876788857426e-06,
+      "loss": 0.683,
+      "step": 576930
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 8.82872885895768e-06,
+      "loss": 0.6436,
+      "step": 576940
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 8.827580929057931e-06,
+      "loss": 0.7332,
+      "step": 576950
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 8.826432999158184e-06,
+      "loss": 0.6942,
+      "step": 576960
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 8.825285069258437e-06,
+      "loss": 0.6872,
+      "step": 576970
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 8.824137139358689e-06,
+      "loss": 0.6918,
+      "step": 576980
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 8.82298920945894e-06,
+      "loss": 0.6843,
+      "step": 576990
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 8.821841279559194e-06,
+      "loss": 0.7161,
+      "step": 577000
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 8.820693349659447e-06,
+      "loss": 0.7046,
+      "step": 577010
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 8.819545419759698e-06,
+      "loss": 0.747,
+      "step": 577020
+    },
+    {
+      "epoch": 26.47,
+      "learning_rate": 8.818397489859952e-06,
+      "loss": 0.6953,
+      "step": 577030
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 8.817249559960205e-06,
+      "loss": 0.6689,
+      "step": 577040
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 8.816101630060457e-06,
+      "loss": 0.7613,
+      "step": 577050
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 8.81495370016071e-06,
+      "loss": 0.7071,
+      "step": 577060
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 8.813805770260961e-06,
+      "loss": 0.6512,
+      "step": 577070
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 8.812657840361215e-06,
+      "loss": 0.6361,
+      "step": 577080
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 8.811509910461466e-06,
+      "loss": 0.6904,
+      "step": 577090
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 8.81036198056172e-06,
+      "loss": 0.6357,
+      "step": 577100
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 8.809214050661973e-06,
+      "loss": 0.6633,
+      "step": 577110
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 8.808066120762224e-06,
+      "loss": 0.7924,
+      "step": 577120
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 8.806918190862477e-06,
+      "loss": 0.6969,
+      "step": 577130
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 8.805770260962729e-06,
+      "loss": 0.7349,
+      "step": 577140
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 8.804622331062982e-06,
+      "loss": 0.7925,
+      "step": 577150
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 8.803474401163235e-06,
+      "loss": 0.6957,
+      "step": 577160
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 8.802326471263487e-06,
+      "loss": 0.7298,
+      "step": 577170
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 8.80117854136374e-06,
+      "loss": 0.6556,
+      "step": 577180
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 8.800030611463993e-06,
+      "loss": 0.7232,
+      "step": 577190
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 8.798882681564245e-06,
+      "loss": 0.6894,
+      "step": 577200
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 8.797734751664497e-06,
+      "loss": 0.7355,
+      "step": 577210
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 8.79658682176475e-06,
+      "loss": 0.6855,
+      "step": 577220
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 8.795438891865003e-06,
+      "loss": 0.6209,
+      "step": 577230
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 8.794290961965255e-06,
+      "loss": 0.7722,
+      "step": 577240
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 8.793143032065508e-06,
+      "loss": 0.798,
+      "step": 577250
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 8.791995102165761e-06,
+      "loss": 0.6923,
+      "step": 577260
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 8.790847172266013e-06,
+      "loss": 0.7352,
+      "step": 577270
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 8.789699242366264e-06,
+      "loss": 0.8017,
+      "step": 577280
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 8.788551312466517e-06,
+      "loss": 0.6946,
+      "step": 577290
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 8.78740338256677e-06,
+      "loss": 0.7703,
+      "step": 577300
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 8.786255452667022e-06,
+      "loss": 0.6739,
+      "step": 577310
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 8.785107522767275e-06,
+      "loss": 0.7975,
+      "step": 577320
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 8.783959592867529e-06,
+      "loss": 0.6642,
+      "step": 577330
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 8.78281166296778e-06,
+      "loss": 0.7351,
+      "step": 577340
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 8.781663733068033e-06,
+      "loss": 0.7245,
+      "step": 577350
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 8.780515803168285e-06,
+      "loss": 0.6957,
+      "step": 577360
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 8.779367873268538e-06,
+      "loss": 0.6601,
+      "step": 577370
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 8.778219943368791e-06,
+      "loss": 0.7053,
+      "step": 577380
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 8.777072013469043e-06,
+      "loss": 0.723,
+      "step": 577390
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 8.775924083569296e-06,
+      "loss": 0.6434,
+      "step": 577400
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 8.77477615366955e-06,
+      "loss": 0.7035,
+      "step": 577410
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 8.773628223769801e-06,
+      "loss": 0.6762,
+      "step": 577420
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 8.772480293870053e-06,
+      "loss": 0.6436,
+      "step": 577430
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 8.771332363970306e-06,
+      "loss": 0.7376,
+      "step": 577440
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 8.770184434070559e-06,
+      "loss": 0.706,
+      "step": 577450
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 8.76903650417081e-06,
+      "loss": 0.6713,
+      "step": 577460
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 8.767888574271064e-06,
+      "loss": 0.6776,
+      "step": 577470
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 8.766740644371317e-06,
+      "loss": 0.6574,
+      "step": 577480
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 8.765592714471569e-06,
+      "loss": 0.6219,
+      "step": 577490
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 8.76444478457182e-06,
+      "loss": 0.7516,
+      "step": 577500
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 8.763296854672073e-06,
+      "loss": 0.6612,
+      "step": 577510
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 8.762148924772327e-06,
+      "loss": 0.6834,
+      "step": 577520
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 8.761000994872578e-06,
+      "loss": 0.7281,
+      "step": 577530
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 8.759853064972831e-06,
+      "loss": 0.7292,
+      "step": 577540
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 8.758705135073085e-06,
+      "loss": 0.7037,
+      "step": 577550
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 8.757557205173336e-06,
+      "loss": 0.8321,
+      "step": 577560
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 8.75640927527359e-06,
+      "loss": 0.6774,
+      "step": 577570
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 8.755261345373841e-06,
+      "loss": 0.6737,
+      "step": 577580
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 8.754113415474094e-06,
+      "loss": 0.6628,
+      "step": 577590
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 8.752965485574346e-06,
+      "loss": 0.7193,
+      "step": 577600
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 8.751817555674599e-06,
+      "loss": 0.6629,
+      "step": 577610
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 8.750669625774852e-06,
+      "loss": 0.781,
+      "step": 577620
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 8.749521695875104e-06,
+      "loss": 0.747,
+      "step": 577630
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 8.748373765975357e-06,
+      "loss": 0.7046,
+      "step": 577640
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 8.747225836075609e-06,
+      "loss": 0.6787,
+      "step": 577650
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 8.746077906175862e-06,
+      "loss": 0.6887,
+      "step": 577660
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 8.744929976276115e-06,
+      "loss": 0.7867,
+      "step": 577670
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 8.743782046376367e-06,
+      "loss": 0.7021,
+      "step": 577680
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 8.74263411647662e-06,
+      "loss": 0.6892,
+      "step": 577690
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 8.741486186576873e-06,
+      "loss": 0.7561,
+      "step": 577700
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 8.740338256677125e-06,
+      "loss": 0.6994,
+      "step": 577710
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 8.739190326777376e-06,
+      "loss": 0.7697,
+      "step": 577720
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 8.73804239687763e-06,
+      "loss": 0.721,
+      "step": 577730
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 8.736894466977883e-06,
+      "loss": 0.7143,
+      "step": 577740
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 8.735746537078134e-06,
+      "loss": 0.6486,
+      "step": 577750
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 8.734598607178388e-06,
+      "loss": 0.6934,
+      "step": 577760
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 8.73345067727864e-06,
+      "loss": 0.6869,
+      "step": 577770
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 8.732302747378892e-06,
+      "loss": 0.6997,
+      "step": 577780
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 8.731154817479144e-06,
+      "loss": 0.6448,
+      "step": 577790
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 8.730006887579397e-06,
+      "loss": 0.7047,
+      "step": 577800
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 8.72885895767965e-06,
+      "loss": 0.6548,
+      "step": 577810
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 8.727711027779902e-06,
+      "loss": 0.6297,
+      "step": 577820
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 8.726563097880155e-06,
+      "loss": 0.6407,
+      "step": 577830
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 8.725415167980408e-06,
+      "loss": 0.7068,
+      "step": 577840
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 8.72426723808066e-06,
+      "loss": 0.6634,
+      "step": 577850
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 8.723119308180913e-06,
+      "loss": 0.7192,
+      "step": 577860
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 8.721971378281165e-06,
+      "loss": 0.6713,
+      "step": 577870
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 8.720823448381418e-06,
+      "loss": 0.6977,
+      "step": 577880
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 8.719675518481671e-06,
+      "loss": 0.671,
+      "step": 577890
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 8.718527588581923e-06,
+      "loss": 0.7363,
+      "step": 577900
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 8.717379658682176e-06,
+      "loss": 0.6743,
+      "step": 577910
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 8.71623172878243e-06,
+      "loss": 0.6931,
+      "step": 577920
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 8.715083798882681e-06,
+      "loss": 0.6312,
+      "step": 577930
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 8.713935868982932e-06,
+      "loss": 0.6394,
+      "step": 577940
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 8.712787939083186e-06,
+      "loss": 0.7789,
+      "step": 577950
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 8.711640009183439e-06,
+      "loss": 0.6931,
+      "step": 577960
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 8.71049207928369e-06,
+      "loss": 0.7629,
+      "step": 577970
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 8.709344149383944e-06,
+      "loss": 0.6869,
+      "step": 577980
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 8.708196219484197e-06,
+      "loss": 0.6374,
+      "step": 577990
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 8.707048289584448e-06,
+      "loss": 0.6509,
+      "step": 578000
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 8.7059003596847e-06,
+      "loss": 0.7675,
+      "step": 578010
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 8.704752429784953e-06,
+      "loss": 0.7275,
+      "step": 578020
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 8.703604499885206e-06,
+      "loss": 0.6943,
+      "step": 578030
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 8.702456569985458e-06,
+      "loss": 0.7084,
+      "step": 578040
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 8.701308640085711e-06,
+      "loss": 0.7518,
+      "step": 578050
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 8.700160710185965e-06,
+      "loss": 0.7366,
+      "step": 578060
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 8.699012780286216e-06,
+      "loss": 0.656,
+      "step": 578070
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 8.69786485038647e-06,
+      "loss": 0.7363,
+      "step": 578080
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 8.696716920486721e-06,
+      "loss": 0.7066,
+      "step": 578090
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 8.695568990586974e-06,
+      "loss": 0.7684,
+      "step": 578100
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 8.694421060687226e-06,
+      "loss": 0.7332,
+      "step": 578110
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 8.693273130787479e-06,
+      "loss": 0.7761,
+      "step": 578120
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 8.692125200887732e-06,
+      "loss": 0.6979,
+      "step": 578130
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 8.690977270987984e-06,
+      "loss": 0.7608,
+      "step": 578140
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 8.689829341088237e-06,
+      "loss": 0.7232,
+      "step": 578150
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 8.688681411188488e-06,
+      "loss": 0.7046,
+      "step": 578160
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 8.687533481288742e-06,
+      "loss": 0.788,
+      "step": 578170
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 8.686385551388995e-06,
+      "loss": 0.7348,
+      "step": 578180
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 8.685237621489247e-06,
+      "loss": 0.7477,
+      "step": 578190
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 8.6840896915895e-06,
+      "loss": 0.6838,
+      "step": 578200
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 8.682941761689753e-06,
+      "loss": 0.718,
+      "step": 578210
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 8.681793831790005e-06,
+      "loss": 0.6991,
+      "step": 578220
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 8.680645901890256e-06,
+      "loss": 0.658,
+      "step": 578230
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 8.67949797199051e-06,
+      "loss": 0.6772,
+      "step": 578240
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 8.678350042090763e-06,
+      "loss": 0.6829,
+      "step": 578250
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 8.677202112191014e-06,
+      "loss": 0.7651,
+      "step": 578260
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 8.676054182291267e-06,
+      "loss": 0.6282,
+      "step": 578270
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 8.67490625239152e-06,
+      "loss": 0.7469,
+      "step": 578280
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 8.673758322491772e-06,
+      "loss": 0.7311,
+      "step": 578290
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 8.672610392592024e-06,
+      "loss": 0.7225,
+      "step": 578300
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 8.671462462692277e-06,
+      "loss": 0.684,
+      "step": 578310
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 8.67031453279253e-06,
+      "loss": 0.6638,
+      "step": 578320
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 8.669166602892782e-06,
+      "loss": 0.7095,
+      "step": 578330
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 8.668018672993035e-06,
+      "loss": 0.6543,
+      "step": 578340
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 8.666870743093288e-06,
+      "loss": 0.7865,
+      "step": 578350
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 8.66572281319354e-06,
+      "loss": 0.7092,
+      "step": 578360
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 8.664574883293793e-06,
+      "loss": 0.6746,
+      "step": 578370
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 8.663426953394045e-06,
+      "loss": 0.7082,
+      "step": 578380
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 8.662279023494298e-06,
+      "loss": 0.7079,
+      "step": 578390
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 8.661131093594551e-06,
+      "loss": 0.6921,
+      "step": 578400
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 8.659983163694803e-06,
+      "loss": 0.7317,
+      "step": 578410
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 8.658835233795056e-06,
+      "loss": 0.749,
+      "step": 578420
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 8.657687303895309e-06,
+      "loss": 0.6527,
+      "step": 578430
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 8.65653937399556e-06,
+      "loss": 0.7754,
+      "step": 578440
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 8.655391444095812e-06,
+      "loss": 0.73,
+      "step": 578450
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 8.654243514196065e-06,
+      "loss": 0.6596,
+      "step": 578460
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 8.653095584296319e-06,
+      "loss": 0.7256,
+      "step": 578470
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 8.65194765439657e-06,
+      "loss": 0.699,
+      "step": 578480
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 8.650799724496823e-06,
+      "loss": 0.7455,
+      "step": 578490
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 8.649651794597077e-06,
+      "loss": 0.7212,
+      "step": 578500
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 8.648503864697328e-06,
+      "loss": 0.6801,
+      "step": 578510
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 8.64735593479758e-06,
+      "loss": 0.7431,
+      "step": 578520
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 8.646208004897833e-06,
+      "loss": 0.669,
+      "step": 578530
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 8.645060074998086e-06,
+      "loss": 0.7264,
+      "step": 578540
+    },
+    {
+      "epoch": 26.54,
+      "learning_rate": 8.643912145098338e-06,
+      "loss": 0.666,
+      "step": 578550
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 8.642764215198591e-06,
+      "loss": 0.6776,
+      "step": 578560
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 8.641616285298844e-06,
+      "loss": 0.7782,
+      "step": 578570
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 8.640468355399096e-06,
+      "loss": 0.6964,
+      "step": 578580
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 8.639320425499349e-06,
+      "loss": 0.663,
+      "step": 578590
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 8.6381724955996e-06,
+      "loss": 0.6819,
+      "step": 578600
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 8.637024565699854e-06,
+      "loss": 0.6583,
+      "step": 578610
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 8.635876635800105e-06,
+      "loss": 0.7641,
+      "step": 578620
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 8.634728705900359e-06,
+      "loss": 0.7298,
+      "step": 578630
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 8.633580776000612e-06,
+      "loss": 0.6478,
+      "step": 578640
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 8.632432846100863e-06,
+      "loss": 0.7759,
+      "step": 578650
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 8.631284916201117e-06,
+      "loss": 0.7612,
+      "step": 578660
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 8.630136986301368e-06,
+      "loss": 0.689,
+      "step": 578670
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 8.628989056401622e-06,
+      "loss": 0.6286,
+      "step": 578680
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 8.627841126501875e-06,
+      "loss": 0.7166,
+      "step": 578690
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 8.626693196602126e-06,
+      "loss": 0.5916,
+      "step": 578700
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 8.62554526670238e-06,
+      "loss": 0.6799,
+      "step": 578710
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 8.624397336802633e-06,
+      "loss": 0.7997,
+      "step": 578720
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 8.623249406902884e-06,
+      "loss": 0.7135,
+      "step": 578730
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 8.622101477003136e-06,
+      "loss": 0.7028,
+      "step": 578740
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 8.620953547103389e-06,
+      "loss": 0.6686,
+      "step": 578750
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 8.619805617203642e-06,
+      "loss": 0.7357,
+      "step": 578760
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 8.618657687303894e-06,
+      "loss": 0.7148,
+      "step": 578770
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 8.617509757404147e-06,
+      "loss": 0.741,
+      "step": 578780
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 8.6163618275044e-06,
+      "loss": 0.6371,
+      "step": 578790
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 8.615213897604652e-06,
+      "loss": 0.7441,
+      "step": 578800
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 8.614065967704904e-06,
+      "loss": 0.6523,
+      "step": 578810
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 8.612918037805157e-06,
+      "loss": 0.7119,
+      "step": 578820
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 8.61177010790541e-06,
+      "loss": 0.808,
+      "step": 578830
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 8.610622178005662e-06,
+      "loss": 0.6473,
+      "step": 578840
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 8.609474248105915e-06,
+      "loss": 0.7076,
+      "step": 578850
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 8.608326318206168e-06,
+      "loss": 0.6739,
+      "step": 578860
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 8.60717838830642e-06,
+      "loss": 0.6817,
+      "step": 578870
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 8.606030458406673e-06,
+      "loss": 0.6738,
+      "step": 578880
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 8.604882528506924e-06,
+      "loss": 0.6902,
+      "step": 578890
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 8.603734598607178e-06,
+      "loss": 0.7557,
+      "step": 578900
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 8.60258666870743e-06,
+      "loss": 0.6667,
+      "step": 578910
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 8.601438738807682e-06,
+      "loss": 0.6782,
+      "step": 578920
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 8.600290808907936e-06,
+      "loss": 0.7007,
+      "step": 578930
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 8.599142879008189e-06,
+      "loss": 0.7258,
+      "step": 578940
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 8.59799494910844e-06,
+      "loss": 0.6022,
+      "step": 578950
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 8.596847019208692e-06,
+      "loss": 0.6824,
+      "step": 578960
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 8.595699089308945e-06,
+      "loss": 0.6537,
+      "step": 578970
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 8.594551159409198e-06,
+      "loss": 0.7455,
+      "step": 578980
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 8.59340322950945e-06,
+      "loss": 0.7095,
+      "step": 578990
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 8.592255299609703e-06,
+      "loss": 0.6334,
+      "step": 579000
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 8.591107369709956e-06,
+      "loss": 0.8027,
+      "step": 579010
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 8.589959439810208e-06,
+      "loss": 0.7816,
+      "step": 579020
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 8.58881150991046e-06,
+      "loss": 0.7312,
+      "step": 579030
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 8.587663580010713e-06,
+      "loss": 0.7007,
+      "step": 579040
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 8.586515650110966e-06,
+      "loss": 0.7693,
+      "step": 579050
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 8.585367720211218e-06,
+      "loss": 0.7466,
+      "step": 579060
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 8.584219790311471e-06,
+      "loss": 0.7509,
+      "step": 579070
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 8.583071860411724e-06,
+      "loss": 0.7806,
+      "step": 579080
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 8.581923930511976e-06,
+      "loss": 0.7086,
+      "step": 579090
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 8.580776000612229e-06,
+      "loss": 0.7448,
+      "step": 579100
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 8.57962807071248e-06,
+      "loss": 0.6805,
+      "step": 579110
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 8.578480140812734e-06,
+      "loss": 0.6591,
+      "step": 579120
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 8.577332210912985e-06,
+      "loss": 0.675,
+      "step": 579130
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 8.576184281013238e-06,
+      "loss": 0.7149,
+      "step": 579140
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 8.575036351113492e-06,
+      "loss": 0.6754,
+      "step": 579150
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 8.573888421213743e-06,
+      "loss": 0.7472,
+      "step": 579160
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 8.572740491313997e-06,
+      "loss": 0.736,
+      "step": 579170
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 8.571592561414248e-06,
+      "loss": 0.6737,
+      "step": 579180
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 8.570444631514501e-06,
+      "loss": 0.6812,
+      "step": 579190
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 8.569296701614755e-06,
+      "loss": 0.5926,
+      "step": 579200
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 8.568148771715006e-06,
+      "loss": 0.8196,
+      "step": 579210
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 8.56700084181526e-06,
+      "loss": 0.6918,
+      "step": 579220
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 8.565852911915513e-06,
+      "loss": 0.7053,
+      "step": 579230
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 8.564704982015764e-06,
+      "loss": 0.7333,
+      "step": 579240
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 8.563557052116016e-06,
+      "loss": 0.8407,
+      "step": 579250
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 8.562409122216269e-06,
+      "loss": 0.677,
+      "step": 579260
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 8.561261192316522e-06,
+      "loss": 0.7669,
+      "step": 579270
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 8.560113262416774e-06,
+      "loss": 0.683,
+      "step": 579280
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 8.558965332517027e-06,
+      "loss": 0.7965,
+      "step": 579290
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 8.55781740261728e-06,
+      "loss": 0.6574,
+      "step": 579300
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 8.556669472717532e-06,
+      "loss": 0.7316,
+      "step": 579310
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 8.555521542817783e-06,
+      "loss": 0.6315,
+      "step": 579320
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 8.554373612918037e-06,
+      "loss": 0.7223,
+      "step": 579330
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 8.55322568301829e-06,
+      "loss": 0.7032,
+      "step": 579340
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 8.552077753118541e-06,
+      "loss": 0.7799,
+      "step": 579350
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 8.550929823218795e-06,
+      "loss": 0.7246,
+      "step": 579360
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 8.549781893319048e-06,
+      "loss": 0.7547,
+      "step": 579370
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 8.5486339634193e-06,
+      "loss": 0.6293,
+      "step": 579380
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 8.547486033519553e-06,
+      "loss": 0.6741,
+      "step": 579390
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 8.546338103619804e-06,
+      "loss": 0.6911,
+      "step": 579400
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 8.545190173720057e-06,
+      "loss": 0.7537,
+      "step": 579410
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 8.54404224382031e-06,
+      "loss": 0.7586,
+      "step": 579420
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 8.542894313920562e-06,
+      "loss": 0.6952,
+      "step": 579430
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 8.541746384020815e-06,
+      "loss": 0.725,
+      "step": 579440
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 8.540598454121069e-06,
+      "loss": 0.7411,
+      "step": 579450
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 8.53945052422132e-06,
+      "loss": 0.6181,
+      "step": 579460
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 8.538302594321572e-06,
+      "loss": 0.7147,
+      "step": 579470
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 8.537154664421825e-06,
+      "loss": 0.7667,
+      "step": 579480
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 8.536006734522078e-06,
+      "loss": 0.7069,
+      "step": 579490
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 8.53485880462233e-06,
+      "loss": 0.6958,
+      "step": 579500
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 8.533710874722583e-06,
+      "loss": 0.6984,
+      "step": 579510
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 8.532562944822836e-06,
+      "loss": 0.7295,
+      "step": 579520
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 8.531415014923088e-06,
+      "loss": 0.7331,
+      "step": 579530
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 8.53026708502334e-06,
+      "loss": 0.6826,
+      "step": 579540
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 8.529119155123593e-06,
+      "loss": 0.7493,
+      "step": 579550
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 8.527971225223846e-06,
+      "loss": 0.6783,
+      "step": 579560
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 8.526823295324097e-06,
+      "loss": 0.7526,
+      "step": 579570
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 8.52567536542435e-06,
+      "loss": 0.722,
+      "step": 579580
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 8.524527435524604e-06,
+      "loss": 0.7157,
+      "step": 579590
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 8.523379505624855e-06,
+      "loss": 0.7158,
+      "step": 579600
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 8.522231575725109e-06,
+      "loss": 0.8042,
+      "step": 579610
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 8.52108364582536e-06,
+      "loss": 0.742,
+      "step": 579620
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 8.519935715925613e-06,
+      "loss": 0.6556,
+      "step": 579630
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 8.518787786025867e-06,
+      "loss": 0.749,
+      "step": 579640
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 8.517639856126118e-06,
+      "loss": 0.7533,
+      "step": 579650
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 8.516491926226372e-06,
+      "loss": 0.7711,
+      "step": 579660
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 8.515343996326623e-06,
+      "loss": 0.6602,
+      "step": 579670
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 8.514196066426876e-06,
+      "loss": 0.8516,
+      "step": 579680
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 8.513048136527128e-06,
+      "loss": 0.6495,
+      "step": 579690
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 8.511900206627381e-06,
+      "loss": 0.6897,
+      "step": 579700
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 8.510752276727634e-06,
+      "loss": 0.6338,
+      "step": 579710
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 8.509604346827886e-06,
+      "loss": 0.7087,
+      "step": 579720
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 8.508456416928139e-06,
+      "loss": 0.733,
+      "step": 579730
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 8.507308487028392e-06,
+      "loss": 0.7159,
+      "step": 579740
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 8.506160557128644e-06,
+      "loss": 0.661,
+      "step": 579750
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 8.505012627228895e-06,
+      "loss": 0.68,
+      "step": 579760
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 8.503864697329149e-06,
+      "loss": 0.7732,
+      "step": 579770
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 8.502716767429402e-06,
+      "loss": 0.6929,
+      "step": 579780
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 8.501568837529653e-06,
+      "loss": 0.6813,
+      "step": 579790
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 8.500420907629907e-06,
+      "loss": 0.7279,
+      "step": 579800
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 8.49927297773016e-06,
+      "loss": 0.7141,
+      "step": 579810
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 8.498125047830412e-06,
+      "loss": 0.7024,
+      "step": 579820
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 8.496977117930663e-06,
+      "loss": 0.7966,
+      "step": 579830
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 8.495829188030916e-06,
+      "loss": 0.7367,
+      "step": 579840
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 8.49468125813117e-06,
+      "loss": 0.7084,
+      "step": 579850
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 8.493533328231421e-06,
+      "loss": 0.6912,
+      "step": 579860
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 8.492385398331674e-06,
+      "loss": 0.6999,
+      "step": 579870
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 8.491237468431928e-06,
+      "loss": 0.7758,
+      "step": 579880
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 8.490089538532179e-06,
+      "loss": 0.6576,
+      "step": 579890
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 8.488941608632432e-06,
+      "loss": 0.767,
+      "step": 579900
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 8.487793678732684e-06,
+      "loss": 0.6449,
+      "step": 579910
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 8.486645748832937e-06,
+      "loss": 0.7091,
+      "step": 579920
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 8.48549781893319e-06,
+      "loss": 0.7226,
+      "step": 579930
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 8.484349889033442e-06,
+      "loss": 0.7768,
+      "step": 579940
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 8.483201959133695e-06,
+      "loss": 0.7197,
+      "step": 579950
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 8.482054029233948e-06,
+      "loss": 0.7035,
+      "step": 579960
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 8.4809060993342e-06,
+      "loss": 0.7459,
+      "step": 579970
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 8.479758169434452e-06,
+      "loss": 0.6955,
+      "step": 579980
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 8.478610239534705e-06,
+      "loss": 0.6745,
+      "step": 579990
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 8.477462309634958e-06,
+      "loss": 0.7159,
+      "step": 580000
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 8.47631437973521e-06,
+      "loss": 0.7474,
+      "step": 580010
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 8.475166449835463e-06,
+      "loss": 0.7084,
+      "step": 580020
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 8.474018519935716e-06,
+      "loss": 0.6777,
+      "step": 580030
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 8.472870590035968e-06,
+      "loss": 0.6744,
+      "step": 580040
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 8.47172266013622e-06,
+      "loss": 0.704,
+      "step": 580050
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 8.470574730236472e-06,
+      "loss": 0.7256,
+      "step": 580060
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 8.469426800336726e-06,
+      "loss": 0.7124,
+      "step": 580070
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 8.468278870436977e-06,
+      "loss": 0.7773,
+      "step": 580080
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 8.46713094053723e-06,
+      "loss": 0.7701,
+      "step": 580090
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 8.465983010637484e-06,
+      "loss": 0.7541,
+      "step": 580100
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 8.464835080737735e-06,
+      "loss": 0.7019,
+      "step": 580110
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 8.463687150837988e-06,
+      "loss": 0.7385,
+      "step": 580120
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 8.46253922093824e-06,
+      "loss": 0.6865,
+      "step": 580130
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 8.461391291038493e-06,
+      "loss": 0.7362,
+      "step": 580140
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 8.460243361138747e-06,
+      "loss": 0.7007,
+      "step": 580150
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 8.459095431238998e-06,
+      "loss": 0.7888,
+      "step": 580160
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 8.457947501339251e-06,
+      "loss": 0.7187,
+      "step": 580170
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 8.456799571439503e-06,
+      "loss": 0.7109,
+      "step": 580180
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 8.455651641539756e-06,
+      "loss": 0.6907,
+      "step": 580190
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 8.454503711640008e-06,
+      "loss": 0.6987,
+      "step": 580200
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 8.453355781740261e-06,
+      "loss": 0.6994,
+      "step": 580210
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 8.452207851840514e-06,
+      "loss": 0.7445,
+      "step": 580220
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 8.451059921940766e-06,
+      "loss": 0.6611,
+      "step": 580230
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 8.449911992041019e-06,
+      "loss": 0.6336,
+      "step": 580240
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 8.448764062141272e-06,
+      "loss": 0.736,
+      "step": 580250
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 8.447616132241524e-06,
+      "loss": 0.7913,
+      "step": 580260
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 8.446468202341775e-06,
+      "loss": 0.6584,
+      "step": 580270
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 8.445320272442028e-06,
+      "loss": 0.7412,
+      "step": 580280
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 8.444172342542282e-06,
+      "loss": 0.6569,
+      "step": 580290
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 8.443024412642533e-06,
+      "loss": 0.6898,
+      "step": 580300
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 8.441876482742787e-06,
+      "loss": 0.7282,
+      "step": 580310
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 8.44072855284304e-06,
+      "loss": 0.6315,
+      "step": 580320
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 8.439580622943291e-06,
+      "loss": 0.6631,
+      "step": 580330
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 8.438432693043543e-06,
+      "loss": 0.6582,
+      "step": 580340
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 8.437284763143796e-06,
+      "loss": 0.7,
+      "step": 580350
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 8.43613683324405e-06,
+      "loss": 0.7126,
+      "step": 580360
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 8.434988903344301e-06,
+      "loss": 0.8043,
+      "step": 580370
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 8.433840973444554e-06,
+      "loss": 0.7119,
+      "step": 580380
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 8.432693043544807e-06,
+      "loss": 0.6494,
+      "step": 580390
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 8.431545113645059e-06,
+      "loss": 0.8046,
+      "step": 580400
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 8.430397183745312e-06,
+      "loss": 0.6939,
+      "step": 580410
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 8.429249253845564e-06,
+      "loss": 0.7198,
+      "step": 580420
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 8.428101323945817e-06,
+      "loss": 0.7137,
+      "step": 580430
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 8.42695339404607e-06,
+      "loss": 0.6689,
+      "step": 580440
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 8.425805464146322e-06,
+      "loss": 0.7348,
+      "step": 580450
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 8.424657534246575e-06,
+      "loss": 0.6604,
+      "step": 580460
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 8.423509604346828e-06,
+      "loss": 0.7782,
+      "step": 580470
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 8.42236167444708e-06,
+      "loss": 0.7829,
+      "step": 580480
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 8.421213744547331e-06,
+      "loss": 0.7849,
+      "step": 580490
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 8.420065814647585e-06,
+      "loss": 0.7299,
+      "step": 580500
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 8.418917884747838e-06,
+      "loss": 0.74,
+      "step": 580510
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 8.41776995484809e-06,
+      "loss": 0.6531,
+      "step": 580520
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 8.416622024948343e-06,
+      "loss": 0.7679,
+      "step": 580530
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 8.415474095048596e-06,
+      "loss": 0.6564,
+      "step": 580540
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 8.414326165148847e-06,
+      "loss": 0.6874,
+      "step": 580550
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 8.413178235249099e-06,
+      "loss": 0.7713,
+      "step": 580560
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 8.412030305349352e-06,
+      "loss": 0.6032,
+      "step": 580570
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 8.410882375449605e-06,
+      "loss": 0.7873,
+      "step": 580580
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 8.409734445549857e-06,
+      "loss": 0.688,
+      "step": 580590
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 8.40858651565011e-06,
+      "loss": 0.7251,
+      "step": 580600
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 8.407438585750363e-06,
+      "loss": 0.6581,
+      "step": 580610
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 8.406290655850615e-06,
+      "loss": 0.6944,
+      "step": 580620
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 8.405142725950868e-06,
+      "loss": 0.6476,
+      "step": 580630
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 8.40399479605112e-06,
+      "loss": 0.7353,
+      "step": 580640
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 8.402846866151373e-06,
+      "loss": 0.7772,
+      "step": 580650
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 8.401698936251626e-06,
+      "loss": 0.6456,
+      "step": 580660
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 8.400551006351878e-06,
+      "loss": 0.6122,
+      "step": 580670
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 8.399403076452131e-06,
+      "loss": 0.6526,
+      "step": 580680
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 8.398255146552383e-06,
+      "loss": 0.7666,
+      "step": 580690
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 8.397107216652636e-06,
+      "loss": 0.6965,
+      "step": 580700
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 8.395959286752887e-06,
+      "loss": 0.6639,
+      "step": 580710
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 8.39481135685314e-06,
+      "loss": 0.7536,
+      "step": 580720
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 8.393663426953394e-06,
+      "loss": 0.6995,
+      "step": 580730
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 8.392515497053645e-06,
+      "loss": 0.6501,
+      "step": 580740
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 8.391367567153899e-06,
+      "loss": 0.6526,
+      "step": 580750
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 8.390219637254152e-06,
+      "loss": 0.7257,
+      "step": 580760
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 8.389071707354403e-06,
+      "loss": 0.6627,
+      "step": 580770
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 8.387923777454655e-06,
+      "loss": 0.7833,
+      "step": 580780
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 8.386775847554908e-06,
+      "loss": 0.6775,
+      "step": 580790
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 8.385627917655162e-06,
+      "loss": 0.7073,
+      "step": 580800
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 8.384479987755413e-06,
+      "loss": 0.782,
+      "step": 580810
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 8.383332057855666e-06,
+      "loss": 0.6777,
+      "step": 580820
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 8.38218412795592e-06,
+      "loss": 0.6158,
+      "step": 580830
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 8.381036198056171e-06,
+      "loss": 0.7065,
+      "step": 580840
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 8.379888268156423e-06,
+      "loss": 0.7239,
+      "step": 580850
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 8.378740338256676e-06,
+      "loss": 0.6608,
+      "step": 580860
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 8.377592408356929e-06,
+      "loss": 0.7136,
+      "step": 580870
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 8.37644447845718e-06,
+      "loss": 0.7073,
+      "step": 580880
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 8.375296548557434e-06,
+      "loss": 0.6775,
+      "step": 580890
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 8.374148618657687e-06,
+      "loss": 0.6417,
+      "step": 580900
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 8.373000688757939e-06,
+      "loss": 0.7063,
+      "step": 580910
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 8.371852758858192e-06,
+      "loss": 0.6662,
+      "step": 580920
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 8.370704828958444e-06,
+      "loss": 0.6836,
+      "step": 580930
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 8.369556899058697e-06,
+      "loss": 0.6604,
+      "step": 580940
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 8.36840896915895e-06,
+      "loss": 0.7283,
+      "step": 580950
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 8.367261039259202e-06,
+      "loss": 0.7242,
+      "step": 580960
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 8.366113109359455e-06,
+      "loss": 0.7098,
+      "step": 580970
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 8.364965179459708e-06,
+      "loss": 0.6737,
+      "step": 580980
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 8.36381724955996e-06,
+      "loss": 0.719,
+      "step": 580990
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 8.362669319660211e-06,
+      "loss": 0.6975,
+      "step": 581000
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 8.361521389760464e-06,
+      "loss": 0.7038,
+      "step": 581010
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 8.360373459860718e-06,
+      "loss": 0.689,
+      "step": 581020
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 8.35922552996097e-06,
+      "loss": 0.6855,
+      "step": 581030
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 8.358077600061222e-06,
+      "loss": 0.6352,
+      "step": 581040
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 8.356929670161476e-06,
+      "loss": 0.7218,
+      "step": 581050
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 8.355781740261727e-06,
+      "loss": 0.751,
+      "step": 581060
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 8.354633810361979e-06,
+      "loss": 0.7131,
+      "step": 581070
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 8.353485880462232e-06,
+      "loss": 0.7769,
+      "step": 581080
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 8.352337950562485e-06,
+      "loss": 0.7601,
+      "step": 581090
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 8.351190020662737e-06,
+      "loss": 0.7403,
+      "step": 581100
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 8.35004209076299e-06,
+      "loss": 0.7154,
+      "step": 581110
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 8.348894160863243e-06,
+      "loss": 0.6393,
+      "step": 581120
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 8.347746230963495e-06,
+      "loss": 0.6437,
+      "step": 581130
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 8.346598301063748e-06,
+      "loss": 0.6612,
+      "step": 581140
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 8.345450371164e-06,
+      "loss": 0.7513,
+      "step": 581150
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 8.344302441264253e-06,
+      "loss": 0.6887,
+      "step": 581160
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 8.343154511364506e-06,
+      "loss": 0.7651,
+      "step": 581170
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 8.342006581464758e-06,
+      "loss": 0.6863,
+      "step": 581180
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 8.340858651565011e-06,
+      "loss": 0.7721,
+      "step": 581190
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 8.339710721665262e-06,
+      "loss": 0.7561,
+      "step": 581200
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 8.338562791765516e-06,
+      "loss": 0.7985,
+      "step": 581210
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 8.337414861865767e-06,
+      "loss": 0.6925,
+      "step": 581220
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 8.33626693196602e-06,
+      "loss": 0.7396,
+      "step": 581230
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 8.335119002066274e-06,
+      "loss": 0.6774,
+      "step": 581240
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 8.333971072166525e-06,
+      "loss": 0.7152,
+      "step": 581250
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 8.332823142266778e-06,
+      "loss": 0.7007,
+      "step": 581260
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 8.331675212367032e-06,
+      "loss": 0.6199,
+      "step": 581270
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 8.330527282467283e-06,
+      "loss": 0.7478,
+      "step": 581280
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 8.329379352567535e-06,
+      "loss": 0.7874,
+      "step": 581290
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 8.328231422667788e-06,
+      "loss": 0.6938,
+      "step": 581300
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 8.327083492768041e-06,
+      "loss": 0.6197,
+      "step": 581310
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 8.325935562868293e-06,
+      "loss": 0.7417,
+      "step": 581320
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 8.324787632968546e-06,
+      "loss": 0.6842,
+      "step": 581330
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 8.3236397030688e-06,
+      "loss": 0.772,
+      "step": 581340
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 8.322491773169051e-06,
+      "loss": 0.7348,
+      "step": 581350
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 8.321343843269304e-06,
+      "loss": 0.6921,
+      "step": 581360
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 8.320195913369556e-06,
+      "loss": 0.6862,
+      "step": 581370
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 8.319047983469809e-06,
+      "loss": 0.7,
+      "step": 581380
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 8.31790005357006e-06,
+      "loss": 0.6921,
+      "step": 581390
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 8.316752123670314e-06,
+      "loss": 0.6794,
+      "step": 581400
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 8.315604193770567e-06,
+      "loss": 0.6882,
+      "step": 581410
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 8.314456263870819e-06,
+      "loss": 0.6732,
+      "step": 581420
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 8.313308333971072e-06,
+      "loss": 0.6781,
+      "step": 581430
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 8.312160404071323e-06,
+      "loss": 0.6799,
+      "step": 581440
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 8.311012474171577e-06,
+      "loss": 0.6823,
+      "step": 581450
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 8.30986454427183e-06,
+      "loss": 0.7214,
+      "step": 581460
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 8.308716614372081e-06,
+      "loss": 0.7265,
+      "step": 581470
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 8.307568684472335e-06,
+      "loss": 0.7317,
+      "step": 581480
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 8.306420754572588e-06,
+      "loss": 0.6591,
+      "step": 581490
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 8.30527282467284e-06,
+      "loss": 0.6734,
+      "step": 581500
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 8.304124894773091e-06,
+      "loss": 0.6923,
+      "step": 581510
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 8.302976964873344e-06,
+      "loss": 0.7613,
+      "step": 581520
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 8.301829034973597e-06,
+      "loss": 0.7879,
+      "step": 581530
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 8.300681105073849e-06,
+      "loss": 0.7043,
+      "step": 581540
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 8.299533175174102e-06,
+      "loss": 0.713,
+      "step": 581550
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 8.298385245274355e-06,
+      "loss": 0.6872,
+      "step": 581560
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 8.297237315374607e-06,
+      "loss": 0.6893,
+      "step": 581570
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 8.296089385474859e-06,
+      "loss": 0.6696,
+      "step": 581580
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 8.294941455575112e-06,
+      "loss": 0.7546,
+      "step": 581590
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 8.293793525675365e-06,
+      "loss": 0.6866,
+      "step": 581600
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 8.292645595775617e-06,
+      "loss": 0.701,
+      "step": 581610
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 8.29149766587587e-06,
+      "loss": 0.6981,
+      "step": 581620
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 8.290349735976123e-06,
+      "loss": 0.7137,
+      "step": 581630
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 8.289201806076375e-06,
+      "loss": 0.7353,
+      "step": 581640
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 8.288053876176628e-06,
+      "loss": 0.7222,
+      "step": 581650
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 8.28690594627688e-06,
+      "loss": 0.678,
+      "step": 581660
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 8.285758016377133e-06,
+      "loss": 0.739,
+      "step": 581670
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 8.284610086477386e-06,
+      "loss": 0.6906,
+      "step": 581680
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 8.283462156577637e-06,
+      "loss": 0.679,
+      "step": 581690
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 8.28231422667789e-06,
+      "loss": 0.7355,
+      "step": 581700
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 8.281166296778142e-06,
+      "loss": 0.7349,
+      "step": 581710
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 8.280018366878395e-06,
+      "loss": 0.7856,
+      "step": 581720
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 8.278870436978647e-06,
+      "loss": 0.6844,
+      "step": 581730
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 8.2777225070789e-06,
+      "loss": 0.6649,
+      "step": 581740
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 8.276574577179153e-06,
+      "loss": 0.6959,
+      "step": 581750
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 8.275426647279405e-06,
+      "loss": 0.7367,
+      "step": 581760
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 8.274278717379658e-06,
+      "loss": 0.6935,
+      "step": 581770
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 8.273130787479912e-06,
+      "loss": 0.5923,
+      "step": 581780
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 8.271982857580163e-06,
+      "loss": 0.6606,
+      "step": 581790
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 8.270834927680415e-06,
+      "loss": 0.6775,
+      "step": 581800
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 8.269686997780668e-06,
+      "loss": 0.74,
+      "step": 581810
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 8.268539067880921e-06,
+      "loss": 0.7683,
+      "step": 581820
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 8.267391137981173e-06,
+      "loss": 0.7115,
+      "step": 581830
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 8.266243208081426e-06,
+      "loss": 0.6894,
+      "step": 581840
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 8.265095278181679e-06,
+      "loss": 0.7105,
+      "step": 581850
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 8.26394734828193e-06,
+      "loss": 0.7292,
+      "step": 581860
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 8.262799418382184e-06,
+      "loss": 0.6872,
+      "step": 581870
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 8.261651488482435e-06,
+      "loss": 0.7794,
+      "step": 581880
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 8.260503558582689e-06,
+      "loss": 0.7213,
+      "step": 581890
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 8.25935562868294e-06,
+      "loss": 0.7205,
+      "step": 581900
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 8.258207698783194e-06,
+      "loss": 0.7602,
+      "step": 581910
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 8.257059768883447e-06,
+      "loss": 0.7526,
+      "step": 581920
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 8.255911838983698e-06,
+      "loss": 0.7669,
+      "step": 581930
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 8.254763909083952e-06,
+      "loss": 0.7214,
+      "step": 581940
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 8.253615979184203e-06,
+      "loss": 0.6709,
+      "step": 581950
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 8.252468049284456e-06,
+      "loss": 0.7549,
+      "step": 581960
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 8.25132011938471e-06,
+      "loss": 0.6917,
+      "step": 581970
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 8.250172189484961e-06,
+      "loss": 0.6618,
+      "step": 581980
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 8.249024259585214e-06,
+      "loss": 0.6413,
+      "step": 581990
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 8.247876329685468e-06,
+      "loss": 0.7592,
+      "step": 582000
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 8.24672839978572e-06,
+      "loss": 0.6965,
+      "step": 582010
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 8.24558046988597e-06,
+      "loss": 0.7313,
+      "step": 582020
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 8.244432539986224e-06,
+      "loss": 0.745,
+      "step": 582030
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 8.243284610086477e-06,
+      "loss": 0.6787,
+      "step": 582040
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 8.242136680186729e-06,
+      "loss": 0.7082,
+      "step": 582050
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 8.240988750286982e-06,
+      "loss": 0.7015,
+      "step": 582060
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 8.239840820387235e-06,
+      "loss": 0.6942,
+      "step": 582070
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 8.238692890487487e-06,
+      "loss": 0.7215,
+      "step": 582080
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 8.237544960587738e-06,
+      "loss": 0.696,
+      "step": 582090
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 8.236397030687992e-06,
+      "loss": 0.6473,
+      "step": 582100
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 8.235249100788245e-06,
+      "loss": 0.7393,
+      "step": 582110
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 8.234101170888496e-06,
+      "loss": 0.7766,
+      "step": 582120
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 8.23295324098875e-06,
+      "loss": 0.7357,
+      "step": 582130
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 8.231805311089003e-06,
+      "loss": 0.677,
+      "step": 582140
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 8.230657381189254e-06,
+      "loss": 0.6648,
+      "step": 582150
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 8.229509451289508e-06,
+      "loss": 0.7664,
+      "step": 582160
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 8.22836152138976e-06,
+      "loss": 0.7651,
+      "step": 582170
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 8.227213591490012e-06,
+      "loss": 0.7683,
+      "step": 582180
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 8.226065661590266e-06,
+      "loss": 0.719,
+      "step": 582190
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 8.224917731690517e-06,
+      "loss": 0.7603,
+      "step": 582200
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 8.22376980179077e-06,
+      "loss": 0.6958,
+      "step": 582210
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 8.222621871891022e-06,
+      "loss": 0.6836,
+      "step": 582220
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 8.221473941991275e-06,
+      "loss": 0.7334,
+      "step": 582230
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 8.220326012091527e-06,
+      "loss": 0.6582,
+      "step": 582240
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 8.21917808219178e-06,
+      "loss": 0.7249,
+      "step": 582250
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 8.218030152292033e-06,
+      "loss": 0.7139,
+      "step": 582260
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 8.216882222392285e-06,
+      "loss": 0.6884,
+      "step": 582270
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 8.215734292492538e-06,
+      "loss": 0.7916,
+      "step": 582280
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 8.214586362592791e-06,
+      "loss": 0.7151,
+      "step": 582290
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 8.213438432693043e-06,
+      "loss": 0.6731,
+      "step": 582300
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 8.212290502793294e-06,
+      "loss": 0.7055,
+      "step": 582310
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 8.211142572893548e-06,
+      "loss": 0.7084,
+      "step": 582320
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 8.209994642993801e-06,
+      "loss": 0.6068,
+      "step": 582330
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 8.208846713094052e-06,
+      "loss": 0.7612,
+      "step": 582340
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 8.207698783194306e-06,
+      "loss": 0.7234,
+      "step": 582350
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 8.206550853294559e-06,
+      "loss": 0.7638,
+      "step": 582360
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 8.20540292339481e-06,
+      "loss": 0.6872,
+      "step": 582370
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 8.204254993495064e-06,
+      "loss": 0.7178,
+      "step": 582380
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 8.203107063595315e-06,
+      "loss": 0.7022,
+      "step": 582390
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 8.201959133695569e-06,
+      "loss": 0.6841,
+      "step": 582400
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 8.20081120379582e-06,
+      "loss": 0.7364,
+      "step": 582410
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 8.199663273896073e-06,
+      "loss": 0.7036,
+      "step": 582420
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 8.198515343996327e-06,
+      "loss": 0.7248,
+      "step": 582430
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 8.197367414096578e-06,
+      "loss": 0.7431,
+      "step": 582440
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 8.196219484196831e-06,
+      "loss": 0.6886,
+      "step": 582450
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 8.195071554297083e-06,
+      "loss": 0.7849,
+      "step": 582460
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 8.193923624397336e-06,
+      "loss": 0.7551,
+      "step": 582470
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 8.19277569449759e-06,
+      "loss": 0.7008,
+      "step": 582480
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 8.191627764597841e-06,
+      "loss": 0.7434,
+      "step": 582490
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 8.190479834698094e-06,
+      "loss": 0.6989,
+      "step": 582500
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 8.189331904798347e-06,
+      "loss": 0.7135,
+      "step": 582510
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 8.188183974898599e-06,
+      "loss": 0.7324,
+      "step": 582520
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 8.18703604499885e-06,
+      "loss": 0.7517,
+      "step": 582530
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 8.185888115099104e-06,
+      "loss": 0.6654,
+      "step": 582540
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 8.184740185199357e-06,
+      "loss": 0.6965,
+      "step": 582550
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 8.183592255299609e-06,
+      "loss": 0.7095,
+      "step": 582560
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 8.182444325399862e-06,
+      "loss": 0.7541,
+      "step": 582570
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 8.181296395500115e-06,
+      "loss": 0.6768,
+      "step": 582580
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 8.180148465600367e-06,
+      "loss": 0.6162,
+      "step": 582590
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 8.179000535700618e-06,
+      "loss": 0.7682,
+      "step": 582600
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 8.177852605800871e-06,
+      "loss": 0.6551,
+      "step": 582610
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 8.176704675901125e-06,
+      "loss": 0.6871,
+      "step": 582620
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 8.175556746001376e-06,
+      "loss": 0.6914,
+      "step": 582630
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 8.17440881610163e-06,
+      "loss": 0.6904,
+      "step": 582640
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 8.173260886201883e-06,
+      "loss": 0.6866,
+      "step": 582650
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 8.172112956302134e-06,
+      "loss": 0.6869,
+      "step": 582660
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 8.170965026402387e-06,
+      "loss": 0.7117,
+      "step": 582670
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 8.169817096502639e-06,
+      "loss": 0.7371,
+      "step": 582680
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 8.168669166602892e-06,
+      "loss": 0.7369,
+      "step": 582690
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 8.167521236703145e-06,
+      "loss": 0.6604,
+      "step": 582700
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 8.166373306803397e-06,
+      "loss": 0.7647,
+      "step": 582710
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 8.16522537690365e-06,
+      "loss": 0.6743,
+      "step": 582720
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 8.164077447003903e-06,
+      "loss": 0.6561,
+      "step": 582730
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 8.162929517104155e-06,
+      "loss": 0.7553,
+      "step": 582740
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 8.161781587204407e-06,
+      "loss": 0.6015,
+      "step": 582750
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 8.16063365730466e-06,
+      "loss": 0.6643,
+      "step": 582760
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 8.159485727404913e-06,
+      "loss": 0.7152,
+      "step": 582770
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 8.158337797505165e-06,
+      "loss": 0.7107,
+      "step": 582780
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 8.157189867605418e-06,
+      "loss": 0.7219,
+      "step": 582790
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 8.156041937705671e-06,
+      "loss": 0.7205,
+      "step": 582800
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 8.154894007805923e-06,
+      "loss": 0.7363,
+      "step": 582810
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 8.153746077906174e-06,
+      "loss": 0.7085,
+      "step": 582820
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 8.152598148006427e-06,
+      "loss": 0.6922,
+      "step": 582830
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 8.15145021810668e-06,
+      "loss": 0.6855,
+      "step": 582840
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 8.150302288206932e-06,
+      "loss": 0.7229,
+      "step": 582850
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 8.149154358307185e-06,
+      "loss": 0.7747,
+      "step": 582860
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 8.148006428407439e-06,
+      "loss": 0.6976,
+      "step": 582870
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 8.14685849850769e-06,
+      "loss": 0.7016,
+      "step": 582880
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 8.145710568607944e-06,
+      "loss": 0.7779,
+      "step": 582890
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 8.144562638708195e-06,
+      "loss": 0.722,
+      "step": 582900
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 8.143414708808448e-06,
+      "loss": 0.7591,
+      "step": 582910
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 8.1422667789087e-06,
+      "loss": 0.7638,
+      "step": 582920
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 8.141118849008953e-06,
+      "loss": 0.6823,
+      "step": 582930
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 8.139970919109206e-06,
+      "loss": 0.7573,
+      "step": 582940
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 8.138822989209458e-06,
+      "loss": 0.6972,
+      "step": 582950
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 8.137675059309711e-06,
+      "loss": 0.6525,
+      "step": 582960
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 8.136527129409963e-06,
+      "loss": 0.7255,
+      "step": 582970
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 8.135379199510216e-06,
+      "loss": 0.7068,
+      "step": 582980
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 8.134231269610469e-06,
+      "loss": 0.6784,
+      "step": 582990
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 8.13308333971072e-06,
+      "loss": 0.7401,
+      "step": 583000
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 8.131935409810974e-06,
+      "loss": 0.7216,
+      "step": 583010
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 8.130787479911227e-06,
+      "loss": 0.6704,
+      "step": 583020
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 8.129639550011479e-06,
+      "loss": 0.7289,
+      "step": 583030
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 8.12849162011173e-06,
+      "loss": 0.7429,
+      "step": 583040
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 8.127343690211984e-06,
+      "loss": 0.7053,
+      "step": 583050
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 8.126195760312237e-06,
+      "loss": 0.7324,
+      "step": 583060
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 8.125047830412488e-06,
+      "loss": 0.7038,
+      "step": 583070
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 8.123899900512742e-06,
+      "loss": 0.6852,
+      "step": 583080
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 8.122751970612995e-06,
+      "loss": 0.6803,
+      "step": 583090
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 8.121604040713246e-06,
+      "loss": 0.6512,
+      "step": 583100
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 8.120456110813498e-06,
+      "loss": 0.6783,
+      "step": 583110
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 8.119308180913751e-06,
+      "loss": 0.6695,
+      "step": 583120
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 8.118160251014004e-06,
+      "loss": 0.7473,
+      "step": 583130
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 8.117012321114256e-06,
+      "loss": 0.6544,
+      "step": 583140
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 8.11586439121451e-06,
+      "loss": 0.696,
+      "step": 583150
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 8.114716461314762e-06,
+      "loss": 0.7097,
+      "step": 583160
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 8.113568531415014e-06,
+      "loss": 0.7135,
+      "step": 583170
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 8.112420601515267e-06,
+      "loss": 0.6493,
+      "step": 583180
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 8.111272671615519e-06,
+      "loss": 0.6997,
+      "step": 583190
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 8.110124741715772e-06,
+      "loss": 0.7346,
+      "step": 583200
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 8.108976811816025e-06,
+      "loss": 0.7205,
+      "step": 583210
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 8.107828881916277e-06,
+      "loss": 0.6976,
+      "step": 583220
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 8.10668095201653e-06,
+      "loss": 0.6814,
+      "step": 583230
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 8.105533022116783e-06,
+      "loss": 0.7348,
+      "step": 583240
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 8.104385092217035e-06,
+      "loss": 0.8091,
+      "step": 583250
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 8.103237162317286e-06,
+      "loss": 0.6735,
+      "step": 583260
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 8.10208923241754e-06,
+      "loss": 0.6908,
+      "step": 583270
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 8.100941302517793e-06,
+      "loss": 0.8489,
+      "step": 583280
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 8.099793372618044e-06,
+      "loss": 0.7059,
+      "step": 583290
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 8.098645442718298e-06,
+      "loss": 0.6756,
+      "step": 583300
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 8.097497512818551e-06,
+      "loss": 0.7525,
+      "step": 583310
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 8.096349582918802e-06,
+      "loss": 0.721,
+      "step": 583320
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 8.095201653019054e-06,
+      "loss": 0.7691,
+      "step": 583330
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 8.094053723119307e-06,
+      "loss": 0.7121,
+      "step": 583340
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 8.09290579321956e-06,
+      "loss": 0.6877,
+      "step": 583350
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 8.091757863319812e-06,
+      "loss": 0.7739,
+      "step": 583360
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 8.090609933420065e-06,
+      "loss": 0.7267,
+      "step": 583370
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 8.089462003520318e-06,
+      "loss": 0.6822,
+      "step": 583380
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 8.08831407362057e-06,
+      "loss": 0.6562,
+      "step": 583390
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 8.087166143720823e-06,
+      "loss": 0.7385,
+      "step": 583400
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 8.086018213821075e-06,
+      "loss": 0.7057,
+      "step": 583410
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 8.084870283921328e-06,
+      "loss": 0.6343,
+      "step": 583420
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 8.08372235402158e-06,
+      "loss": 0.7263,
+      "step": 583430
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 8.082574424121833e-06,
+      "loss": 0.7736,
+      "step": 583440
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 8.081426494222086e-06,
+      "loss": 0.693,
+      "step": 583450
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 8.080278564322338e-06,
+      "loss": 0.7344,
+      "step": 583460
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 8.079130634422591e-06,
+      "loss": 0.7221,
+      "step": 583470
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 8.077982704522842e-06,
+      "loss": 0.7086,
+      "step": 583480
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 8.076834774623096e-06,
+      "loss": 0.7225,
+      "step": 583490
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 8.075686844723349e-06,
+      "loss": 0.7917,
+      "step": 583500
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 8.0745389148236e-06,
+      "loss": 0.7143,
+      "step": 583510
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 8.073390984923854e-06,
+      "loss": 0.735,
+      "step": 583520
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 8.072243055024107e-06,
+      "loss": 0.7165,
+      "step": 583530
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 8.071095125124359e-06,
+      "loss": 0.6891,
+      "step": 583540
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 8.06994719522461e-06,
+      "loss": 0.6675,
+      "step": 583550
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 8.068799265324863e-06,
+      "loss": 0.7343,
+      "step": 583560
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 8.067651335425117e-06,
+      "loss": 0.7657,
+      "step": 583570
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 8.066503405525368e-06,
+      "loss": 0.7051,
+      "step": 583580
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 8.065355475625621e-06,
+      "loss": 0.7209,
+      "step": 583590
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 8.064207545725875e-06,
+      "loss": 0.7237,
+      "step": 583600
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 8.063059615826126e-06,
+      "loss": 0.6434,
+      "step": 583610
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 8.061911685926378e-06,
+      "loss": 0.711,
+      "step": 583620
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 8.060763756026631e-06,
+      "loss": 0.621,
+      "step": 583630
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 8.059615826126884e-06,
+      "loss": 0.7089,
+      "step": 583640
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 8.058467896227136e-06,
+      "loss": 0.7195,
+      "step": 583650
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 8.057319966327389e-06,
+      "loss": 0.6666,
+      "step": 583660
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 8.056172036427642e-06,
+      "loss": 0.6507,
+      "step": 583670
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 8.055024106527894e-06,
+      "loss": 0.69,
+      "step": 583680
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 8.053876176628147e-06,
+      "loss": 0.7156,
+      "step": 583690
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 8.052728246728399e-06,
+      "loss": 0.684,
+      "step": 583700
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 8.051580316828652e-06,
+      "loss": 0.7138,
+      "step": 583710
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 8.050432386928905e-06,
+      "loss": 0.671,
+      "step": 583720
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 8.049284457029157e-06,
+      "loss": 0.7639,
+      "step": 583730
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 8.04813652712941e-06,
+      "loss": 0.7079,
+      "step": 583740
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 8.046988597229663e-06,
+      "loss": 0.7088,
+      "step": 583750
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 8.045840667329915e-06,
+      "loss": 0.709,
+      "step": 583760
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 8.044692737430166e-06,
+      "loss": 0.7644,
+      "step": 583770
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 8.04354480753042e-06,
+      "loss": 0.6791,
+      "step": 583780
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 8.042396877630673e-06,
+      "loss": 0.7293,
+      "step": 583790
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 8.041248947730924e-06,
+      "loss": 0.7317,
+      "step": 583800
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 8.040101017831177e-06,
+      "loss": 0.6937,
+      "step": 583810
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 8.03895308793143e-06,
+      "loss": 0.7107,
+      "step": 583820
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 8.037805158031682e-06,
+      "loss": 0.7132,
+      "step": 583830
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 8.036657228131934e-06,
+      "loss": 0.6874,
+      "step": 583840
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 8.035509298232187e-06,
+      "loss": 0.6904,
+      "step": 583850
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 8.03436136833244e-06,
+      "loss": 0.7577,
+      "step": 583860
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 8.033213438432692e-06,
+      "loss": 0.6872,
+      "step": 583870
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 8.032065508532945e-06,
+      "loss": 0.6822,
+      "step": 583880
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 8.030917578633198e-06,
+      "loss": 0.7426,
+      "step": 583890
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 8.02976964873345e-06,
+      "loss": 0.6874,
+      "step": 583900
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 8.028621718833703e-06,
+      "loss": 0.6683,
+      "step": 583910
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 8.027473788933955e-06,
+      "loss": 0.754,
+      "step": 583920
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 8.026325859034208e-06,
+      "loss": 0.7473,
+      "step": 583930
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 8.02517792913446e-06,
+      "loss": 0.777,
+      "step": 583940
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 8.024029999234713e-06,
+      "loss": 0.7173,
+      "step": 583950
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 8.022882069334966e-06,
+      "loss": 0.7038,
+      "step": 583960
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 8.021734139435217e-06,
+      "loss": 0.6919,
+      "step": 583970
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 8.02058620953547e-06,
+      "loss": 0.7549,
+      "step": 583980
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 8.019438279635722e-06,
+      "loss": 0.7222,
+      "step": 583990
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 8.018290349735975e-06,
+      "loss": 0.6402,
+      "step": 584000
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 8.017142419836229e-06,
+      "loss": 0.723,
+      "step": 584010
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 8.01599448993648e-06,
+      "loss": 0.6677,
+      "step": 584020
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 8.014846560036734e-06,
+      "loss": 0.6428,
+      "step": 584030
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 8.013698630136987e-06,
+      "loss": 0.7198,
+      "step": 584040
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 8.012550700237238e-06,
+      "loss": 0.7171,
+      "step": 584050
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 8.01140277033749e-06,
+      "loss": 0.6658,
+      "step": 584060
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 8.010254840437743e-06,
+      "loss": 0.7166,
+      "step": 584070
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 8.009106910537996e-06,
+      "loss": 0.7257,
+      "step": 584080
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 8.007958980638248e-06,
+      "loss": 0.7513,
+      "step": 584090
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 8.006811050738501e-06,
+      "loss": 0.6966,
+      "step": 584100
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 8.005663120838754e-06,
+      "loss": 0.7176,
+      "step": 584110
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 8.004515190939006e-06,
+      "loss": 0.7607,
+      "step": 584120
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 8.003367261039257e-06,
+      "loss": 0.6742,
+      "step": 584130
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 8.00221933113951e-06,
+      "loss": 0.6644,
+      "step": 584140
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 8.001071401239764e-06,
+      "loss": 0.79,
+      "step": 584150
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 7.999923471340016e-06,
+      "loss": 0.733,
+      "step": 584160
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 7.998775541440269e-06,
+      "loss": 0.7421,
+      "step": 584170
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 7.997627611540522e-06,
+      "loss": 0.6498,
+      "step": 584180
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 7.996479681640774e-06,
+      "loss": 0.6719,
+      "step": 584190
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 7.995331751741027e-06,
+      "loss": 0.7061,
+      "step": 584200
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 7.994183821841278e-06,
+      "loss": 0.6478,
+      "step": 584210
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 7.993035891941532e-06,
+      "loss": 0.7168,
+      "step": 584220
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 7.991887962041785e-06,
+      "loss": 0.7843,
+      "step": 584230
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 7.990740032142036e-06,
+      "loss": 0.66,
+      "step": 584240
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 7.98959210224229e-06,
+      "loss": 0.7434,
+      "step": 584250
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 7.988444172342543e-06,
+      "loss": 0.7233,
+      "step": 584260
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 7.987296242442794e-06,
+      "loss": 0.6671,
+      "step": 584270
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 7.986148312543046e-06,
+      "loss": 0.5861,
+      "step": 584280
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 7.9850003826433e-06,
+      "loss": 0.6311,
+      "step": 584290
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 7.983852452743552e-06,
+      "loss": 0.7278,
+      "step": 584300
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 7.982704522843804e-06,
+      "loss": 0.7067,
+      "step": 584310
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 7.981556592944057e-06,
+      "loss": 0.7709,
+      "step": 584320
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 7.98040866304431e-06,
+      "loss": 0.6474,
+      "step": 584330
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 7.979260733144562e-06,
+      "loss": 0.8053,
+      "step": 584340
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 7.978112803244814e-06,
+      "loss": 0.7264,
+      "step": 584350
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 7.976964873345067e-06,
+      "loss": 0.6824,
+      "step": 584360
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 7.97581694344532e-06,
+      "loss": 0.6782,
+      "step": 584370
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 7.974669013545572e-06,
+      "loss": 0.7447,
+      "step": 584380
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 7.973521083645825e-06,
+      "loss": 0.7328,
+      "step": 584390
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 7.972373153746078e-06,
+      "loss": 0.6952,
+      "step": 584400
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 7.97122522384633e-06,
+      "loss": 0.6813,
+      "step": 584410
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 7.970077293946583e-06,
+      "loss": 0.6558,
+      "step": 584420
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 7.968929364046834e-06,
+      "loss": 0.6497,
+      "step": 584430
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 7.967781434147088e-06,
+      "loss": 0.7089,
+      "step": 584440
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 7.96663350424734e-06,
+      "loss": 0.715,
+      "step": 584450
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 7.965485574347592e-06,
+      "loss": 0.681,
+      "step": 584460
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 7.964337644447846e-06,
+      "loss": 0.7536,
+      "step": 584470
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 7.963189714548097e-06,
+      "loss": 0.7109,
+      "step": 584480
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 7.96204178464835e-06,
+      "loss": 0.6365,
+      "step": 584490
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 7.960893854748602e-06,
+      "loss": 0.7875,
+      "step": 584500
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 7.959745924848855e-06,
+      "loss": 0.747,
+      "step": 584510
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 7.958597994949109e-06,
+      "loss": 0.6906,
+      "step": 584520
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 7.95745006504936e-06,
+      "loss": 0.751,
+      "step": 584530
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 7.956302135149613e-06,
+      "loss": 0.7442,
+      "step": 584540
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 7.955154205249867e-06,
+      "loss": 0.7688,
+      "step": 584550
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 7.954006275350118e-06,
+      "loss": 0.7128,
+      "step": 584560
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 7.95285834545037e-06,
+      "loss": 0.7138,
+      "step": 584570
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 7.951710415550623e-06,
+      "loss": 0.7316,
+      "step": 584580
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 7.950562485650876e-06,
+      "loss": 0.7262,
+      "step": 584590
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 7.949414555751128e-06,
+      "loss": 0.7067,
+      "step": 584600
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 7.948266625851381e-06,
+      "loss": 0.7017,
+      "step": 584610
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 7.947118695951634e-06,
+      "loss": 0.708,
+      "step": 584620
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 7.945970766051886e-06,
+      "loss": 0.6758,
+      "step": 584630
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 7.944822836152137e-06,
+      "loss": 0.7648,
+      "step": 584640
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 7.94367490625239e-06,
+      "loss": 0.7438,
+      "step": 584650
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 7.942526976352644e-06,
+      "loss": 0.731,
+      "step": 584660
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 7.941379046452895e-06,
+      "loss": 0.6762,
+      "step": 584670
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 7.940231116553149e-06,
+      "loss": 0.6939,
+      "step": 584680
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 7.939083186653402e-06,
+      "loss": 0.7141,
+      "step": 584690
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 7.937935256753653e-06,
+      "loss": 0.7204,
+      "step": 584700
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 7.936787326853907e-06,
+      "loss": 0.7853,
+      "step": 584710
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 7.935639396954158e-06,
+      "loss": 0.7088,
+      "step": 584720
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 7.934491467054411e-06,
+      "loss": 0.7071,
+      "step": 584730
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 7.933343537154665e-06,
+      "loss": 0.761,
+      "step": 584740
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 7.932195607254916e-06,
+      "loss": 0.6841,
+      "step": 584750
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 7.93104767735517e-06,
+      "loss": 0.7238,
+      "step": 584760
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 7.929899747455423e-06,
+      "loss": 0.6981,
+      "step": 584770
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 7.928751817555674e-06,
+      "loss": 0.7178,
+      "step": 584780
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 7.927603887655926e-06,
+      "loss": 0.7411,
+      "step": 584790
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 7.926455957756179e-06,
+      "loss": 0.668,
+      "step": 584800
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 7.925308027856432e-06,
+      "loss": 0.8024,
+      "step": 584810
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 7.924160097956684e-06,
+      "loss": 0.7409,
+      "step": 584820
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 7.923012168056937e-06,
+      "loss": 0.6186,
+      "step": 584830
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 7.92186423815719e-06,
+      "loss": 0.7094,
+      "step": 584840
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 7.920716308257442e-06,
+      "loss": 0.7393,
+      "step": 584850
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 7.919568378357693e-06,
+      "loss": 0.7077,
+      "step": 584860
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 7.918420448457947e-06,
+      "loss": 0.7303,
+      "step": 584870
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 7.9172725185582e-06,
+      "loss": 0.6928,
+      "step": 584880
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 7.916124588658451e-06,
+      "loss": 0.7196,
+      "step": 584890
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 7.914976658758705e-06,
+      "loss": 0.6741,
+      "step": 584900
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 7.913828728858958e-06,
+      "loss": 0.7671,
+      "step": 584910
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 7.91268079895921e-06,
+      "loss": 0.7597,
+      "step": 584920
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 7.911532869059463e-06,
+      "loss": 0.7379,
+      "step": 584930
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 7.910384939159714e-06,
+      "loss": 0.7597,
+      "step": 584940
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 7.909237009259967e-06,
+      "loss": 0.723,
+      "step": 584950
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 7.90808907936022e-06,
+      "loss": 0.676,
+      "step": 584960
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 7.906941149460472e-06,
+      "loss": 0.7526,
+      "step": 584970
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 7.905793219560725e-06,
+      "loss": 0.7394,
+      "step": 584980
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 7.904645289660977e-06,
+      "loss": 0.7036,
+      "step": 584990
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 7.90349735976123e-06,
+      "loss": 0.7089,
+      "step": 585000
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 7.902349429861482e-06,
+      "loss": 0.7106,
+      "step": 585010
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 7.901201499961735e-06,
+      "loss": 0.7344,
+      "step": 585020
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 7.900053570061988e-06,
+      "loss": 0.6988,
+      "step": 585030
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 7.89890564016224e-06,
+      "loss": 0.7058,
+      "step": 585040
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 7.897757710262493e-06,
+      "loss": 0.7298,
+      "step": 585050
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 7.896609780362746e-06,
+      "loss": 0.7035,
+      "step": 585060
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 7.895461850462998e-06,
+      "loss": 0.6526,
+      "step": 585070
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 7.89431392056325e-06,
+      "loss": 0.7568,
+      "step": 585080
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 7.893165990663503e-06,
+      "loss": 0.7283,
+      "step": 585090
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 7.892018060763756e-06,
+      "loss": 0.7167,
+      "step": 585100
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 7.890870130864007e-06,
+      "loss": 0.7078,
+      "step": 585110
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 7.88972220096426e-06,
+      "loss": 0.7406,
+      "step": 585120
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 7.888574271064514e-06,
+      "loss": 0.6068,
+      "step": 585130
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 7.887426341164765e-06,
+      "loss": 0.7589,
+      "step": 585140
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 7.886278411265017e-06,
+      "loss": 0.6915,
+      "step": 585150
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 7.88513048136527e-06,
+      "loss": 0.75,
+      "step": 585160
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 7.883982551465524e-06,
+      "loss": 0.716,
+      "step": 585170
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 7.882834621565775e-06,
+      "loss": 0.6211,
+      "step": 585180
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 7.881686691666028e-06,
+      "loss": 0.7315,
+      "step": 585190
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 7.880538761766282e-06,
+      "loss": 0.733,
+      "step": 585200
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 7.879390831866533e-06,
+      "loss": 0.7134,
+      "step": 585210
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 7.878242901966786e-06,
+      "loss": 0.6501,
+      "step": 585220
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 7.877094972067038e-06,
+      "loss": 0.7078,
+      "step": 585230
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 7.875947042167291e-06,
+      "loss": 0.78,
+      "step": 585240
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 7.874799112267544e-06,
+      "loss": 0.7107,
+      "step": 585250
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 7.873651182367796e-06,
+      "loss": 0.7461,
+      "step": 585260
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 7.87250325246805e-06,
+      "loss": 0.6636,
+      "step": 585270
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 7.871355322568302e-06,
+      "loss": 0.7055,
+      "step": 585280
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 7.870207392668554e-06,
+      "loss": 0.7192,
+      "step": 585290
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 7.869059462768806e-06,
+      "loss": 0.6968,
+      "step": 585300
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 7.867911532869059e-06,
+      "loss": 0.6989,
+      "step": 585310
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 7.866763602969312e-06,
+      "loss": 0.6717,
+      "step": 585320
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 7.865615673069564e-06,
+      "loss": 0.7061,
+      "step": 585330
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 7.864467743169817e-06,
+      "loss": 0.7132,
+      "step": 585340
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 7.86331981327007e-06,
+      "loss": 0.6661,
+      "step": 585350
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 7.862171883370322e-06,
+      "loss": 0.7722,
+      "step": 585360
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 7.861023953470573e-06,
+      "loss": 0.7085,
+      "step": 585370
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 7.859876023570826e-06,
+      "loss": 0.6346,
+      "step": 585380
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 7.85872809367108e-06,
+      "loss": 0.728,
+      "step": 585390
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 7.857580163771331e-06,
+      "loss": 0.6509,
+      "step": 585400
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 7.856432233871584e-06,
+      "loss": 0.718,
+      "step": 585410
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 7.855284303971838e-06,
+      "loss": 0.6457,
+      "step": 585420
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 7.85413637407209e-06,
+      "loss": 0.6819,
+      "step": 585430
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 7.852988444172342e-06,
+      "loss": 0.705,
+      "step": 585440
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 7.851840514272594e-06,
+      "loss": 0.6728,
+      "step": 585450
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 7.850692584372847e-06,
+      "loss": 0.66,
+      "step": 585460
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 7.8495446544731e-06,
+      "loss": 0.7937,
+      "step": 585470
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 7.848396724573352e-06,
+      "loss": 0.7574,
+      "step": 585480
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 7.847248794673605e-06,
+      "loss": 0.6606,
+      "step": 585490
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 7.846100864773857e-06,
+      "loss": 0.7336,
+      "step": 585500
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 7.84495293487411e-06,
+      "loss": 0.7018,
+      "step": 585510
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 7.843805004974362e-06,
+      "loss": 0.7064,
+      "step": 585520
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 7.842657075074615e-06,
+      "loss": 0.694,
+      "step": 585530
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 7.841509145174868e-06,
+      "loss": 0.6611,
+      "step": 585540
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 7.84036121527512e-06,
+      "loss": 0.7938,
+      "step": 585550
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 7.839213285375373e-06,
+      "loss": 0.7398,
+      "step": 585560
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 7.838065355475626e-06,
+      "loss": 0.7358,
+      "step": 585570
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 7.836917425575878e-06,
+      "loss": 0.6746,
+      "step": 585580
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 7.83576949567613e-06,
+      "loss": 0.7388,
+      "step": 585590
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 7.834621565776382e-06,
+      "loss": 0.7021,
+      "step": 585600
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 7.833473635876636e-06,
+      "loss": 0.6808,
+      "step": 585610
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 7.832325705976887e-06,
+      "loss": 0.7801,
+      "step": 585620
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 7.83117777607714e-06,
+      "loss": 0.6914,
+      "step": 585630
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 7.830029846177394e-06,
+      "loss": 0.6621,
+      "step": 585640
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 7.828881916277645e-06,
+      "loss": 0.7411,
+      "step": 585650
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 7.827733986377897e-06,
+      "loss": 0.728,
+      "step": 585660
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 7.82658605647815e-06,
+      "loss": 0.6999,
+      "step": 585670
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 7.825438126578403e-06,
+      "loss": 0.7268,
+      "step": 585680
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 7.824290196678655e-06,
+      "loss": 0.7698,
+      "step": 585690
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 7.823142266778908e-06,
+      "loss": 0.6755,
+      "step": 585700
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 7.821994336879161e-06,
+      "loss": 0.7454,
+      "step": 585710
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 7.820846406979413e-06,
+      "loss": 0.7315,
+      "step": 585720
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 7.819698477079666e-06,
+      "loss": 0.701,
+      "step": 585730
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 7.818550547179918e-06,
+      "loss": 0.7697,
+      "step": 585740
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 7.817402617280171e-06,
+      "loss": 0.7164,
+      "step": 585750
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 7.816254687380424e-06,
+      "loss": 0.6713,
+      "step": 585760
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 7.815106757480676e-06,
+      "loss": 0.6992,
+      "step": 585770
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 7.813958827580927e-06,
+      "loss": 0.6891,
+      "step": 585780
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 7.812810897681182e-06,
+      "loss": 0.7971,
+      "step": 585790
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 7.811662967781434e-06,
+      "loss": 0.6677,
+      "step": 585800
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 7.810515037881685e-06,
+      "loss": 0.7459,
+      "step": 585810
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 7.809367107981939e-06,
+      "loss": 0.6176,
+      "step": 585820
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 7.808219178082192e-06,
+      "loss": 0.6992,
+      "step": 585830
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 7.807071248182443e-06,
+      "loss": 0.6622,
+      "step": 585840
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 7.805923318282695e-06,
+      "loss": 0.6734,
+      "step": 585850
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 7.80477538838295e-06,
+      "loss": 0.6817,
+      "step": 585860
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 7.803627458483201e-06,
+      "loss": 0.7125,
+      "step": 585870
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 7.802479528583453e-06,
+      "loss": 0.7179,
+      "step": 585880
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 7.801331598683706e-06,
+      "loss": 0.7097,
+      "step": 585890
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 7.80018366878396e-06,
+      "loss": 0.6643,
+      "step": 585900
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 7.799035738884211e-06,
+      "loss": 0.7566,
+      "step": 585910
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 7.797887808984464e-06,
+      "loss": 0.7182,
+      "step": 585920
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 7.796739879084716e-06,
+      "loss": 0.6838,
+      "step": 585930
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 7.795591949184969e-06,
+      "loss": 0.6621,
+      "step": 585940
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 7.794444019285222e-06,
+      "loss": 0.6584,
+      "step": 585950
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 7.793296089385474e-06,
+      "loss": 0.7444,
+      "step": 585960
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 7.792148159485727e-06,
+      "loss": 0.6532,
+      "step": 585970
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 7.79100022958598e-06,
+      "loss": 0.7538,
+      "step": 585980
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 7.789852299686232e-06,
+      "loss": 0.753,
+      "step": 585990
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 7.788704369786483e-06,
+      "loss": 0.6669,
+      "step": 586000
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 7.787556439886737e-06,
+      "loss": 0.8097,
+      "step": 586010
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 7.78640850998699e-06,
+      "loss": 0.7041,
+      "step": 586020
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 7.785260580087241e-06,
+      "loss": 0.7309,
+      "step": 586030
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 7.784112650187495e-06,
+      "loss": 0.7037,
+      "step": 586040
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 7.782964720287748e-06,
+      "loss": 0.7459,
+      "step": 586050
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 7.781816790388e-06,
+      "loss": 0.6602,
+      "step": 586060
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 7.780668860488251e-06,
+      "loss": 0.7334,
+      "step": 586070
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 7.779520930588504e-06,
+      "loss": 0.7262,
+      "step": 586080
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 7.778373000688757e-06,
+      "loss": 0.6203,
+      "step": 586090
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 7.777225070789009e-06,
+      "loss": 0.7002,
+      "step": 586100
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 7.776077140889262e-06,
+      "loss": 0.6904,
+      "step": 586110
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 7.774929210989515e-06,
+      "loss": 0.693,
+      "step": 586120
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 7.773781281089767e-06,
+      "loss": 0.7543,
+      "step": 586130
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 7.77263335119002e-06,
+      "loss": 0.7244,
+      "step": 586140
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 7.771485421290272e-06,
+      "loss": 0.6674,
+      "step": 586150
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 7.770337491390525e-06,
+      "loss": 0.6888,
+      "step": 586160
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 7.769189561490777e-06,
+      "loss": 0.6844,
+      "step": 586170
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 7.76804163159103e-06,
+      "loss": 0.7376,
+      "step": 586180
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 7.766893701691283e-06,
+      "loss": 0.704,
+      "step": 586190
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 7.765745771791535e-06,
+      "loss": 0.7479,
+      "step": 586200
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 7.764597841891788e-06,
+      "loss": 0.7141,
+      "step": 586210
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 7.76344991199204e-06,
+      "loss": 0.8167,
+      "step": 586220
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 7.762301982092293e-06,
+      "loss": 0.6958,
+      "step": 586230
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 7.761154052192546e-06,
+      "loss": 0.7093,
+      "step": 586240
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 7.760006122292797e-06,
+      "loss": 0.6525,
+      "step": 586250
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 7.75885819239305e-06,
+      "loss": 0.7481,
+      "step": 586260
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 7.757710262493304e-06,
+      "loss": 0.6884,
+      "step": 586270
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 7.756562332593556e-06,
+      "loss": 0.7588,
+      "step": 586280
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 7.755414402693807e-06,
+      "loss": 0.6312,
+      "step": 586290
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 7.75426647279406e-06,
+      "loss": 0.8667,
+      "step": 586300
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 7.753118542894314e-06,
+      "loss": 0.6814,
+      "step": 586310
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 7.751970612994565e-06,
+      "loss": 0.652,
+      "step": 586320
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 7.750822683094818e-06,
+      "loss": 0.6624,
+      "step": 586330
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 7.749674753195072e-06,
+      "loss": 0.7385,
+      "step": 586340
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 7.748526823295323e-06,
+      "loss": 0.6752,
+      "step": 586350
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 7.747378893395575e-06,
+      "loss": 0.7019,
+      "step": 586360
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 7.746230963495828e-06,
+      "loss": 0.7095,
+      "step": 586370
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 7.745083033596081e-06,
+      "loss": 0.8098,
+      "step": 586380
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 7.743935103696333e-06,
+      "loss": 0.7761,
+      "step": 586390
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 7.742787173796586e-06,
+      "loss": 0.7036,
+      "step": 586400
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 7.74163924389684e-06,
+      "loss": 0.7099,
+      "step": 586410
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 7.74049131399709e-06,
+      "loss": 0.7316,
+      "step": 586420
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 7.739343384097344e-06,
+      "loss": 0.692,
+      "step": 586430
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 7.738195454197596e-06,
+      "loss": 0.7605,
+      "step": 586440
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 7.737047524297849e-06,
+      "loss": 0.6365,
+      "step": 586450
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 7.735899594398102e-06,
+      "loss": 0.7168,
+      "step": 586460
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 7.734751664498354e-06,
+      "loss": 0.6246,
+      "step": 586470
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 7.733603734598607e-06,
+      "loss": 0.7159,
+      "step": 586480
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 7.73245580469886e-06,
+      "loss": 0.6813,
+      "step": 586490
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 7.731307874799112e-06,
+      "loss": 0.7059,
+      "step": 586500
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 7.730159944899363e-06,
+      "loss": 0.6998,
+      "step": 586510
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 7.729012014999616e-06,
+      "loss": 0.6502,
+      "step": 586520
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 7.72786408509987e-06,
+      "loss": 0.696,
+      "step": 586530
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 7.726716155200121e-06,
+      "loss": 0.7334,
+      "step": 586540
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 7.725568225300374e-06,
+      "loss": 0.7647,
+      "step": 586550
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 7.724420295400628e-06,
+      "loss": 0.6654,
+      "step": 586560
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 7.72327236550088e-06,
+      "loss": 0.7197,
+      "step": 586570
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 7.72212443560113e-06,
+      "loss": 0.7737,
+      "step": 586580
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 7.720976505701384e-06,
+      "loss": 0.8019,
+      "step": 586590
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 7.719828575801637e-06,
+      "loss": 0.774,
+      "step": 586600
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 7.718680645901889e-06,
+      "loss": 0.6617,
+      "step": 586610
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 7.717532716002142e-06,
+      "loss": 0.6943,
+      "step": 586620
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 7.716384786102395e-06,
+      "loss": 0.8013,
+      "step": 586630
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 7.715236856202647e-06,
+      "loss": 0.7448,
+      "step": 586640
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 7.7140889263029e-06,
+      "loss": 0.7183,
+      "step": 586650
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 7.712940996403152e-06,
+      "loss": 0.7371,
+      "step": 586660
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 7.711793066503405e-06,
+      "loss": 0.6829,
+      "step": 586670
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 7.710645136603656e-06,
+      "loss": 0.6793,
+      "step": 586680
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 7.70949720670391e-06,
+      "loss": 0.7199,
+      "step": 586690
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 7.708349276804163e-06,
+      "loss": 0.7287,
+      "step": 586700
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 7.707201346904414e-06,
+      "loss": 0.7981,
+      "step": 586710
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 7.706053417004668e-06,
+      "loss": 0.7579,
+      "step": 586720
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 7.70490548710492e-06,
+      "loss": 0.6922,
+      "step": 586730
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 7.703757557205172e-06,
+      "loss": 0.7464,
+      "step": 586740
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 7.702609627305426e-06,
+      "loss": 0.7218,
+      "step": 586750
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 7.701461697405677e-06,
+      "loss": 0.6896,
+      "step": 586760
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 7.70031376750593e-06,
+      "loss": 0.6995,
+      "step": 586770
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 7.699165837606184e-06,
+      "loss": 0.6727,
+      "step": 586780
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 7.698017907706435e-06,
+      "loss": 0.6824,
+      "step": 586790
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 7.696869977806687e-06,
+      "loss": 0.6379,
+      "step": 586800
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 7.69572204790694e-06,
+      "loss": 0.7138,
+      "step": 586810
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 7.694574118007193e-06,
+      "loss": 0.68,
+      "step": 586820
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 7.693426188107445e-06,
+      "loss": 0.6873,
+      "step": 586830
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 7.692278258207698e-06,
+      "loss": 0.7208,
+      "step": 586840
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 7.691130328307951e-06,
+      "loss": 0.7257,
+      "step": 586850
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 7.689982398408203e-06,
+      "loss": 0.7362,
+      "step": 586860
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 7.688834468508454e-06,
+      "loss": 0.7022,
+      "step": 586870
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 7.687686538608708e-06,
+      "loss": 0.6227,
+      "step": 586880
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 7.686538608708961e-06,
+      "loss": 0.7445,
+      "step": 586890
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 7.685390678809212e-06,
+      "loss": 0.7302,
+      "step": 586900
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 7.684242748909466e-06,
+      "loss": 0.6745,
+      "step": 586910
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 7.683094819009719e-06,
+      "loss": 0.6068,
+      "step": 586920
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 7.68194688910997e-06,
+      "loss": 0.7402,
+      "step": 586930
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 7.680798959210224e-06,
+      "loss": 0.6599,
+      "step": 586940
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 7.679651029310475e-06,
+      "loss": 0.6701,
+      "step": 586950
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 7.678503099410729e-06,
+      "loss": 0.6507,
+      "step": 586960
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 7.677355169510982e-06,
+      "loss": 0.6737,
+      "step": 586970
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 7.676207239611233e-06,
+      "loss": 0.7235,
+      "step": 586980
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 7.675059309711487e-06,
+      "loss": 0.7066,
+      "step": 586990
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 7.67391137981174e-06,
+      "loss": 0.7307,
+      "step": 587000
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 7.672763449911991e-06,
+      "loss": 0.6614,
+      "step": 587010
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 7.671615520012243e-06,
+      "loss": 0.6543,
+      "step": 587020
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 7.670467590112496e-06,
+      "loss": 0.6362,
+      "step": 587030
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 7.66931966021275e-06,
+      "loss": 0.7124,
+      "step": 587040
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 7.668171730313001e-06,
+      "loss": 0.6893,
+      "step": 587050
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 7.667023800413254e-06,
+      "loss": 0.7446,
+      "step": 587060
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 7.665875870513507e-06,
+      "loss": 0.6477,
+      "step": 587070
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 7.664727940613759e-06,
+      "loss": 0.7678,
+      "step": 587080
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 7.66358001071401e-06,
+      "loss": 0.7758,
+      "step": 587090
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 7.662432080814264e-06,
+      "loss": 0.6484,
+      "step": 587100
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 7.661284150914517e-06,
+      "loss": 0.7734,
+      "step": 587110
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 7.660136221014769e-06,
+      "loss": 0.7545,
+      "step": 587120
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 7.658988291115022e-06,
+      "loss": 0.7143,
+      "step": 587130
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 7.657840361215275e-06,
+      "loss": 0.7815,
+      "step": 587140
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 7.656692431315527e-06,
+      "loss": 0.7931,
+      "step": 587150
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 7.65554450141578e-06,
+      "loss": 0.6761,
+      "step": 587160
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 7.654396571516031e-06,
+      "loss": 0.7676,
+      "step": 587170
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 7.653248641616285e-06,
+      "loss": 0.6748,
+      "step": 587180
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 7.652100711716538e-06,
+      "loss": 0.7832,
+      "step": 587190
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 7.65095278181679e-06,
+      "loss": 0.6779,
+      "step": 587200
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 7.649804851917043e-06,
+      "loss": 0.7024,
+      "step": 587210
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 7.648656922017294e-06,
+      "loss": 0.7623,
+      "step": 587220
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 7.647508992117547e-06,
+      "loss": 0.7142,
+      "step": 587230
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 7.646361062217799e-06,
+      "loss": 0.7134,
+      "step": 587240
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 7.645213132318052e-06,
+      "loss": 0.6914,
+      "step": 587250
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 7.644065202418306e-06,
+      "loss": 0.7214,
+      "step": 587260
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 7.642917272518557e-06,
+      "loss": 0.6589,
+      "step": 587270
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 7.64176934261881e-06,
+      "loss": 0.7411,
+      "step": 587280
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 7.640621412719064e-06,
+      "loss": 0.7709,
+      "step": 587290
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 7.639473482819315e-06,
+      "loss": 0.7252,
+      "step": 587300
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 7.638325552919567e-06,
+      "loss": 0.6711,
+      "step": 587310
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 7.63717762301982e-06,
+      "loss": 0.683,
+      "step": 587320
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 7.636029693120073e-06,
+      "loss": 0.646,
+      "step": 587330
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 7.634881763220325e-06,
+      "loss": 0.8028,
+      "step": 587340
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 7.633733833320578e-06,
+      "loss": 0.6811,
+      "step": 587350
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 7.632585903420831e-06,
+      "loss": 0.6449,
+      "step": 587360
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 7.631437973521083e-06,
+      "loss": 0.639,
+      "step": 587370
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 7.630290043621334e-06,
+      "loss": 0.7198,
+      "step": 587380
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 7.629142113721588e-06,
+      "loss": 0.6979,
+      "step": 587390
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 7.627994183821841e-06,
+      "loss": 0.7795,
+      "step": 587400
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 7.626846253922093e-06,
+      "loss": 0.6864,
+      "step": 587410
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 7.625698324022346e-06,
+      "loss": 0.7215,
+      "step": 587420
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 7.624550394122598e-06,
+      "loss": 0.7502,
+      "step": 587430
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 7.62340246422285e-06,
+      "loss": 0.8049,
+      "step": 587440
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 7.6222545343231036e-06,
+      "loss": 0.6482,
+      "step": 587450
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 7.621106604423356e-06,
+      "loss": 0.7401,
+      "step": 587460
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 7.619958674523608e-06,
+      "loss": 0.7407,
+      "step": 587470
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 7.618810744623862e-06,
+      "loss": 0.6863,
+      "step": 587480
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 7.617662814724114e-06,
+      "loss": 0.6824,
+      "step": 587490
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 7.6165148848243655e-06,
+      "loss": 0.8697,
+      "step": 587500
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 7.615366954924619e-06,
+      "loss": 0.7375,
+      "step": 587510
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 7.614219025024871e-06,
+      "loss": 0.7646,
+      "step": 587520
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 7.6130710951251236e-06,
+      "loss": 0.7075,
+      "step": 587530
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 7.611923165225376e-06,
+      "loss": 0.734,
+      "step": 587540
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 7.610775235325629e-06,
+      "loss": 0.7895,
+      "step": 587550
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 7.609627305425882e-06,
+      "loss": 0.6898,
+      "step": 587560
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 7.608479375526133e-06,
+      "loss": 0.6884,
+      "step": 587570
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 7.607331445626386e-06,
+      "loss": 0.7236,
+      "step": 587580
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 7.606183515726639e-06,
+      "loss": 0.6706,
+      "step": 587590
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 7.605035585826891e-06,
+      "loss": 0.6853,
+      "step": 587600
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 7.6038876559271444e-06,
+      "loss": 0.7246,
+      "step": 587610
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 7.602739726027397e-06,
+      "loss": 0.7038,
+      "step": 587620
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 7.601591796127649e-06,
+      "loss": 0.7703,
+      "step": 587630
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 7.6004438662279025e-06,
+      "loss": 0.751,
+      "step": 587640
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 7.599295936328154e-06,
+      "loss": 0.7723,
+      "step": 587650
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 7.598148006428406e-06,
+      "loss": 0.6686,
+      "step": 587660
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 7.59700007652866e-06,
+      "loss": 0.7144,
+      "step": 587670
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 7.595852146628912e-06,
+      "loss": 0.6719,
+      "step": 587680
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 7.5947042167291644e-06,
+      "loss": 0.7427,
+      "step": 587690
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 7.593556286829418e-06,
+      "loss": 0.6624,
+      "step": 587700
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 7.59240835692967e-06,
+      "loss": 0.7155,
+      "step": 587710
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 7.591260427029922e-06,
+      "loss": 0.7517,
+      "step": 587720
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 7.590112497130174e-06,
+      "loss": 0.754,
+      "step": 587730
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 7.588964567230427e-06,
+      "loss": 0.6833,
+      "step": 587740
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 7.58781663733068e-06,
+      "loss": 0.7232,
+      "step": 587750
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 7.586668707430932e-06,
+      "loss": 0.7307,
+      "step": 587760
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 7.585520777531185e-06,
+      "loss": 0.691,
+      "step": 587770
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 7.584372847631438e-06,
+      "loss": 0.7429,
+      "step": 587780
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 7.583224917731689e-06,
+      "loss": 0.7443,
+      "step": 587790
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 7.5820769878319425e-06,
+      "loss": 0.6535,
+      "step": 587800
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 7.580929057932195e-06,
+      "loss": 0.6878,
+      "step": 587810
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 7.579781128032447e-06,
+      "loss": 0.7221,
+      "step": 587820
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 7.5786331981327005e-06,
+      "loss": 0.6716,
+      "step": 587830
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 7.577485268232953e-06,
+      "loss": 0.6345,
+      "step": 587840
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 7.576337338333205e-06,
+      "loss": 0.6759,
+      "step": 587850
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 7.5751894084334586e-06,
+      "loss": 0.7208,
+      "step": 587860
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 7.57404147853371e-06,
+      "loss": 0.6897,
+      "step": 587870
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 7.5728935486339625e-06,
+      "loss": 0.6786,
+      "step": 587880
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 7.571745618734215e-06,
+      "loss": 0.7353,
+      "step": 587890
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 7.570597688834468e-06,
+      "loss": 0.7128,
+      "step": 587900
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 7.5694497589347205e-06,
+      "loss": 0.6919,
+      "step": 587910
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 7.568301829034972e-06,
+      "loss": 0.6719,
+      "step": 587920
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 7.567153899135226e-06,
+      "loss": 0.7338,
+      "step": 587930
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 7.566005969235478e-06,
+      "loss": 0.6889,
+      "step": 587940
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 7.56485803933573e-06,
+      "loss": 0.7011,
+      "step": 587950
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 7.563710109435983e-06,
+      "loss": 0.6756,
+      "step": 587960
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 7.562562179536236e-06,
+      "loss": 0.703,
+      "step": 587970
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 7.561414249636488e-06,
+      "loss": 0.6457,
+      "step": 587980
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 7.560266319736741e-06,
+      "loss": 0.7503,
+      "step": 587990
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 7.559118389836994e-06,
+      "loss": 0.6754,
+      "step": 588000
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 7.557970459937245e-06,
+      "loss": 0.7903,
+      "step": 588010
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 7.5568225300374986e-06,
+      "loss": 0.7939,
+      "step": 588020
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 7.555674600137751e-06,
+      "loss": 0.7153,
+      "step": 588030
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 7.554526670238003e-06,
+      "loss": 0.6772,
+      "step": 588040
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 7.553378740338257e-06,
+      "loss": 0.7131,
+      "step": 588050
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 7.552230810438509e-06,
+      "loss": 0.644,
+      "step": 588060
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 7.5510828805387605e-06,
+      "loss": 0.74,
+      "step": 588070
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 7.549934950639013e-06,
+      "loss": 0.6271,
+      "step": 588080
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 7.548787020739266e-06,
+      "loss": 0.7399,
+      "step": 588090
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 7.547639090839519e-06,
+      "loss": 0.6274,
+      "step": 588100
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 7.546491160939771e-06,
+      "loss": 0.741,
+      "step": 588110
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 7.545343231040024e-06,
+      "loss": 0.7556,
+      "step": 588120
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 7.544195301140277e-06,
+      "loss": 0.694,
+      "step": 588130
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 7.543047371240528e-06,
+      "loss": 0.7481,
+      "step": 588140
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 7.541899441340782e-06,
+      "loss": 0.7309,
+      "step": 588150
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 7.540751511441034e-06,
+      "loss": 0.7603,
+      "step": 588160
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 7.539603581541286e-06,
+      "loss": 0.6859,
+      "step": 588170
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 7.5384556516415394e-06,
+      "loss": 0.6836,
+      "step": 588180
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 7.537307721741792e-06,
+      "loss": 0.7185,
+      "step": 588190
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 7.536159791842044e-06,
+      "loss": 0.7588,
+      "step": 588200
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 7.5350118619422975e-06,
+      "loss": 0.6696,
+      "step": 588210
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 7.533863932042549e-06,
+      "loss": 0.7486,
+      "step": 588220
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 7.532716002142801e-06,
+      "loss": 0.7236,
+      "step": 588230
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 7.531568072243054e-06,
+      "loss": 0.6655,
+      "step": 588240
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 7.530420142343307e-06,
+      "loss": 0.6987,
+      "step": 588250
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 7.5292722124435594e-06,
+      "loss": 0.731,
+      "step": 588260
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 7.528124282543812e-06,
+      "loss": 0.6889,
+      "step": 588270
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 7.526976352644065e-06,
+      "loss": 0.7287,
+      "step": 588280
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 7.525828422744317e-06,
+      "loss": 0.7524,
+      "step": 588290
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 7.524680492844569e-06,
+      "loss": 0.7497,
+      "step": 588300
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 7.523532562944822e-06,
+      "loss": 0.7681,
+      "step": 588310
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 7.522384633045075e-06,
+      "loss": 0.8087,
+      "step": 588320
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 7.521236703145327e-06,
+      "loss": 0.68,
+      "step": 588330
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 7.52008877324558e-06,
+      "loss": 0.7846,
+      "step": 588340
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 7.518940843345833e-06,
+      "loss": 0.6214,
+      "step": 588350
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 7.517792913446084e-06,
+      "loss": 0.6785,
+      "step": 588360
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 7.5166449835463375e-06,
+      "loss": 0.6844,
+      "step": 588370
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 7.51549705364659e-06,
+      "loss": 0.7524,
+      "step": 588380
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 7.514349123746842e-06,
+      "loss": 0.7293,
+      "step": 588390
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 7.513201193847095e-06,
+      "loss": 0.7098,
+      "step": 588400
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 7.512053263947348e-06,
+      "loss": 0.7568,
+      "step": 588410
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 7.5109053340476e-06,
+      "loss": 0.7241,
+      "step": 588420
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 7.509757404147852e-06,
+      "loss": 0.7067,
+      "step": 588430
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 7.508609474248105e-06,
+      "loss": 0.7136,
+      "step": 588440
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 7.5074615443483575e-06,
+      "loss": 0.6847,
+      "step": 588450
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 7.50631361444861e-06,
+      "loss": 0.7858,
+      "step": 588460
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 7.505165684548863e-06,
+      "loss": 0.698,
+      "step": 588470
+    },
+    {
+      "epoch": 27.0,
+      "eval_cer": 0.6688680893782155,
+      "eval_em": 0.007357212003872217,
+      "eval_f1": 0.007357212003872217,
+      "eval_loss": 1.366132140159607,
+      "eval_runtime": 1296.2611,
+      "eval_samples_per_second": 7.969,
+      "eval_steps_per_second": 0.997,
+      "eval_wer": 0.9926427879961278,
+      "step": 588473
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 7.5040177546491155e-06,
+      "loss": 0.6546,
+      "step": 588480
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 7.502869824749368e-06,
+      "loss": 0.6782,
+      "step": 588490
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 7.501721894849621e-06,
+      "loss": 0.6413,
+      "step": 588500
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 7.500573964949873e-06,
+      "loss": 0.7723,
+      "step": 588510
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 7.499426035050125e-06,
+      "loss": 0.7157,
+      "step": 588520
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 7.498278105150378e-06,
+      "loss": 0.6684,
+      "step": 588530
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 7.497130175250631e-06,
+      "loss": 0.7238,
+      "step": 588540
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 7.495982245350883e-06,
+      "loss": 0.6614,
+      "step": 588550
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 7.494834315451136e-06,
+      "loss": 0.7009,
+      "step": 588560
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 7.493686385551389e-06,
+      "loss": 0.658,
+      "step": 588570
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 7.49253845565164e-06,
+      "loss": 0.7151,
+      "step": 588580
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 7.491390525751893e-06,
+      "loss": 0.6555,
+      "step": 588590
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 7.490242595852146e-06,
+      "loss": 0.6398,
+      "step": 588600
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 7.489094665952398e-06,
+      "loss": 0.6963,
+      "step": 588610
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 7.487946736052651e-06,
+      "loss": 0.6787,
+      "step": 588620
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 7.486798806152904e-06,
+      "loss": 0.7223,
+      "step": 588630
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 7.485650876253156e-06,
+      "loss": 0.6808,
+      "step": 588640
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 7.484502946353408e-06,
+      "loss": 0.6683,
+      "step": 588650
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 7.483355016453661e-06,
+      "loss": 0.7077,
+      "step": 588660
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 7.482207086553914e-06,
+      "loss": 0.6629,
+      "step": 588670
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 7.481059156654166e-06,
+      "loss": 0.6516,
+      "step": 588680
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 7.479911226754419e-06,
+      "loss": 0.6511,
+      "step": 588690
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 7.478763296854672e-06,
+      "loss": 0.7123,
+      "step": 588700
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 7.477615366954924e-06,
+      "loss": 0.7069,
+      "step": 588710
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 7.476467437055177e-06,
+      "loss": 0.747,
+      "step": 588720
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 7.475319507155429e-06,
+      "loss": 0.7034,
+      "step": 588730
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 7.474171577255681e-06,
+      "loss": 0.6349,
+      "step": 588740
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 7.473023647355934e-06,
+      "loss": 0.6965,
+      "step": 588750
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 7.471875717456187e-06,
+      "loss": 0.7031,
+      "step": 588760
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 7.470727787556439e-06,
+      "loss": 0.6919,
+      "step": 588770
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 7.469579857656692e-06,
+      "loss": 0.6933,
+      "step": 588780
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 7.468431927756945e-06,
+      "loss": 0.7025,
+      "step": 588790
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 7.467283997857196e-06,
+      "loss": 0.6369,
+      "step": 588800
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 7.466136067957449e-06,
+      "loss": 0.7343,
+      "step": 588810
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 7.464988138057702e-06,
+      "loss": 0.6347,
+      "step": 588820
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 7.4638402081579545e-06,
+      "loss": 0.6847,
+      "step": 588830
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 7.462692278258207e-06,
+      "loss": 0.6381,
+      "step": 588840
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 7.46154434835846e-06,
+      "loss": 0.7866,
+      "step": 588850
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 7.4603964184587125e-06,
+      "loss": 0.7013,
+      "step": 588860
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 7.459248488558964e-06,
+      "loss": 0.7182,
+      "step": 588870
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 7.458100558659217e-06,
+      "loss": 0.7139,
+      "step": 588880
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 7.45695262875947e-06,
+      "loss": 0.6974,
+      "step": 588890
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 7.455804698859722e-06,
+      "loss": 0.6554,
+      "step": 588900
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 7.4546567689599745e-06,
+      "loss": 0.6464,
+      "step": 588910
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 7.453508839060228e-06,
+      "loss": 0.6309,
+      "step": 588920
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 7.45236090916048e-06,
+      "loss": 0.727,
+      "step": 588930
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 7.451212979260732e-06,
+      "loss": 0.7674,
+      "step": 588940
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 7.450065049360985e-06,
+      "loss": 0.6355,
+      "step": 588950
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 7.448917119461237e-06,
+      "loss": 0.6828,
+      "step": 588960
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 7.44776918956149e-06,
+      "loss": 0.7239,
+      "step": 588970
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 7.446621259661743e-06,
+      "loss": 0.7262,
+      "step": 588980
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 7.445473329761995e-06,
+      "loss": 0.7437,
+      "step": 588990
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 7.444325399862248e-06,
+      "loss": 0.6994,
+      "step": 589000
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 7.443177469962501e-06,
+      "loss": 0.7473,
+      "step": 589010
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 7.4420295400627525e-06,
+      "loss": 0.7314,
+      "step": 589020
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 7.440881610163005e-06,
+      "loss": 0.6695,
+      "step": 589030
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 7.439733680263258e-06,
+      "loss": 0.6178,
+      "step": 589040
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 7.4385857503635105e-06,
+      "loss": 0.6837,
+      "step": 589050
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 7.437437820463763e-06,
+      "loss": 0.6368,
+      "step": 589060
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 7.436289890564016e-06,
+      "loss": 0.7435,
+      "step": 589070
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 7.435141960664269e-06,
+      "loss": 0.6935,
+      "step": 589080
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 7.43399403076452e-06,
+      "loss": 0.6552,
+      "step": 589090
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 7.4328461008647725e-06,
+      "loss": 0.7021,
+      "step": 589100
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 7.431698170965026e-06,
+      "loss": 0.662,
+      "step": 589110
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 7.430550241065278e-06,
+      "loss": 0.687,
+      "step": 589120
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 7.4294023111655306e-06,
+      "loss": 0.7366,
+      "step": 589130
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 7.428254381265784e-06,
+      "loss": 0.6248,
+      "step": 589140
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 7.427106451366036e-06,
+      "loss": 0.6462,
+      "step": 589150
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 7.425958521466288e-06,
+      "loss": 0.6663,
+      "step": 589160
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 7.424810591566541e-06,
+      "loss": 0.7011,
+      "step": 589170
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 7.423662661666793e-06,
+      "loss": 0.6946,
+      "step": 589180
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 7.422514731767046e-06,
+      "loss": 0.6341,
+      "step": 589190
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 7.421366801867299e-06,
+      "loss": 0.6742,
+      "step": 589200
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 7.420218871967551e-06,
+      "loss": 0.6718,
+      "step": 589210
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 7.419070942067804e-06,
+      "loss": 0.788,
+      "step": 589220
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 7.417923012168057e-06,
+      "loss": 0.7611,
+      "step": 589230
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 7.416775082268309e-06,
+      "loss": 0.663,
+      "step": 589240
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 7.415627152368561e-06,
+      "loss": 0.7083,
+      "step": 589250
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 7.414479222468813e-06,
+      "loss": 0.6567,
+      "step": 589260
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 7.413331292569067e-06,
+      "loss": 0.6731,
+      "step": 589270
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 7.412183362669319e-06,
+      "loss": 0.6783,
+      "step": 589280
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 7.411035432769571e-06,
+      "loss": 0.6519,
+      "step": 589290
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 7.409887502869825e-06,
+      "loss": 0.7179,
+      "step": 589300
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 7.408739572970076e-06,
+      "loss": 0.6498,
+      "step": 589310
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 7.407591643070329e-06,
+      "loss": 0.6363,
+      "step": 589320
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 7.406443713170582e-06,
+      "loss": 0.6569,
+      "step": 589330
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 7.405295783270834e-06,
+      "loss": 0.7238,
+      "step": 589340
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 7.404147853371087e-06,
+      "loss": 0.688,
+      "step": 589350
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 7.40299992347134e-06,
+      "loss": 0.6536,
+      "step": 589360
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 7.401851993571592e-06,
+      "loss": 0.7023,
+      "step": 589370
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 7.400704063671844e-06,
+      "loss": 0.6216,
+      "step": 589380
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 7.399556133772097e-06,
+      "loss": 0.6844,
+      "step": 589390
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 7.3984082038723495e-06,
+      "loss": 0.6338,
+      "step": 589400
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 7.397260273972602e-06,
+      "loss": 0.7252,
+      "step": 589410
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 7.396112344072855e-06,
+      "loss": 0.7039,
+      "step": 589420
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 7.3949644141731075e-06,
+      "loss": 0.6901,
+      "step": 589430
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 7.39381648427336e-06,
+      "loss": 0.7061,
+      "step": 589440
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 7.3926685543736114e-06,
+      "loss": 0.6667,
+      "step": 589450
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 7.391520624473865e-06,
+      "loss": 0.6461,
+      "step": 589460
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 7.390372694574117e-06,
+      "loss": 0.7044,
+      "step": 589470
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 7.3892247646743695e-06,
+      "loss": 0.6862,
+      "step": 589480
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 7.388076834774623e-06,
+      "loss": 0.6278,
+      "step": 589490
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 7.386928904874875e-06,
+      "loss": 0.6251,
+      "step": 589500
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 7.3857809749751275e-06,
+      "loss": 0.6407,
+      "step": 589510
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 7.384633045075381e-06,
+      "loss": 0.707,
+      "step": 589520
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 7.383485115175632e-06,
+      "loss": 0.7444,
+      "step": 589530
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 7.382337185275885e-06,
+      "loss": 0.6537,
+      "step": 589540
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 7.381189255376138e-06,
+      "loss": 0.694,
+      "step": 589550
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 7.38004132547639e-06,
+      "loss": 0.804,
+      "step": 589560
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 7.378893395576643e-06,
+      "loss": 0.7143,
+      "step": 589570
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 7.377745465676896e-06,
+      "loss": 0.6356,
+      "step": 589580
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 7.376597535777148e-06,
+      "loss": 0.7073,
+      "step": 589590
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 7.3754496058774e-06,
+      "loss": 0.6839,
+      "step": 589600
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 7.374301675977652e-06,
+      "loss": 0.6655,
+      "step": 589610
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 7.3731537460779056e-06,
+      "loss": 0.7256,
+      "step": 589620
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 7.372005816178158e-06,
+      "loss": 0.6706,
+      "step": 589630
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 7.37085788627841e-06,
+      "loss": 0.8035,
+      "step": 589640
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 7.369709956378664e-06,
+      "loss": 0.7203,
+      "step": 589650
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 7.368562026478916e-06,
+      "loss": 0.6745,
+      "step": 589660
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 7.3674140965791675e-06,
+      "loss": 0.7664,
+      "step": 589670
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 7.366266166679421e-06,
+      "loss": 0.7643,
+      "step": 589680
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 7.365118236779673e-06,
+      "loss": 0.5861,
+      "step": 589690
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 7.3639703068799256e-06,
+      "loss": 0.6701,
+      "step": 589700
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 7.362822376980179e-06,
+      "loss": 0.7445,
+      "step": 589710
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 7.361674447080431e-06,
+      "loss": 0.6391,
+      "step": 589720
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 7.360526517180684e-06,
+      "loss": 0.6924,
+      "step": 589730
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 7.359378587280937e-06,
+      "loss": 0.6489,
+      "step": 589740
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 7.358230657381188e-06,
+      "loss": 0.6468,
+      "step": 589750
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 7.357082727481441e-06,
+      "loss": 0.681,
+      "step": 589760
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 7.355934797581693e-06,
+      "loss": 0.6771,
+      "step": 589770
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 7.354786867681946e-06,
+      "loss": 0.7665,
+      "step": 589780
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 7.353638937782199e-06,
+      "loss": 0.6432,
+      "step": 589790
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 7.352491007882451e-06,
+      "loss": 0.6653,
+      "step": 589800
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 7.3513430779827045e-06,
+      "loss": 0.702,
+      "step": 589810
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 7.350195148082956e-06,
+      "loss": 0.7221,
+      "step": 589820
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 7.349047218183208e-06,
+      "loss": 0.6749,
+      "step": 589830
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 7.347899288283462e-06,
+      "loss": 0.6644,
+      "step": 589840
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 7.346751358383714e-06,
+      "loss": 0.721,
+      "step": 589850
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 7.3456034284839664e-06,
+      "loss": 0.6814,
+      "step": 589860
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 7.34445549858422e-06,
+      "loss": 0.6612,
+      "step": 589870
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 7.343307568684472e-06,
+      "loss": 0.7509,
+      "step": 589880
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 7.342159638784724e-06,
+      "loss": 0.6816,
+      "step": 589890
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 7.341011708884977e-06,
+      "loss": 0.6934,
+      "step": 589900
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 7.339863778985229e-06,
+      "loss": 0.7631,
+      "step": 589910
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 7.338715849085482e-06,
+      "loss": 0.6281,
+      "step": 589920
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 7.337567919185735e-06,
+      "loss": 0.6236,
+      "step": 589930
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 7.336419989285987e-06,
+      "loss": 0.727,
+      "step": 589940
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 7.33527205938624e-06,
+      "loss": 0.7247,
+      "step": 589950
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 7.334124129486491e-06,
+      "loss": 0.6461,
+      "step": 589960
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 7.3329761995867445e-06,
+      "loss": 0.6879,
+      "step": 589970
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 7.331828269686997e-06,
+      "loss": 0.6645,
+      "step": 589980
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 7.330680339787249e-06,
+      "loss": 0.7442,
+      "step": 589990
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 7.3295324098875025e-06,
+      "loss": 0.7122,
+      "step": 590000
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 7.328384479987755e-06,
+      "loss": 0.7063,
+      "step": 590010
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 7.327236550088007e-06,
+      "loss": 0.6987,
+      "step": 590020
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 7.3260886201882605e-06,
+      "loss": 0.706,
+      "step": 590030
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 7.324940690288512e-06,
+      "loss": 0.6263,
+      "step": 590040
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 7.3237927603887645e-06,
+      "loss": 0.6932,
+      "step": 590050
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 7.322644830489018e-06,
+      "loss": 0.6562,
+      "step": 590060
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 7.32149690058927e-06,
+      "loss": 0.6959,
+      "step": 590070
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 7.3203489706895225e-06,
+      "loss": 0.6794,
+      "step": 590080
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 7.319201040789776e-06,
+      "loss": 0.7359,
+      "step": 590090
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 7.318053110890028e-06,
+      "loss": 0.7617,
+      "step": 590100
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 7.31690518099028e-06,
+      "loss": 0.7056,
+      "step": 590110
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 7.315757251090532e-06,
+      "loss": 0.7156,
+      "step": 590120
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 7.314609321190785e-06,
+      "loss": 0.7251,
+      "step": 590130
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 7.313461391291038e-06,
+      "loss": 0.7044,
+      "step": 590140
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 7.31231346139129e-06,
+      "loss": 0.6976,
+      "step": 590150
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 7.311165531491543e-06,
+      "loss": 0.6655,
+      "step": 590160
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 7.310017601591796e-06,
+      "loss": 0.7197,
+      "step": 590170
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 7.308869671692047e-06,
+      "loss": 0.6719,
+      "step": 590180
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 7.3077217417923006e-06,
+      "loss": 0.7511,
+      "step": 590190
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 7.306573811892553e-06,
+      "loss": 0.6753,
+      "step": 590200
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 7.305425881992805e-06,
+      "loss": 0.6671,
+      "step": 590210
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 7.304277952093059e-06,
+      "loss": 0.7265,
+      "step": 590220
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 7.303130022193311e-06,
+      "loss": 0.6744,
+      "step": 590230
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 7.301982092293563e-06,
+      "loss": 0.6707,
+      "step": 590240
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 7.300834162393817e-06,
+      "loss": 0.7575,
+      "step": 590250
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 7.299686232494068e-06,
+      "loss": 0.7178,
+      "step": 590260
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 7.2985383025943206e-06,
+      "loss": 0.7511,
+      "step": 590270
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 7.297390372694574e-06,
+      "loss": 0.6913,
+      "step": 590280
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 7.296242442794826e-06,
+      "loss": 0.7361,
+      "step": 590290
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 7.295094512895079e-06,
+      "loss": 0.6986,
+      "step": 590300
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 7.293946582995331e-06,
+      "loss": 0.7359,
+      "step": 590310
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 7.292798653095584e-06,
+      "loss": 0.6756,
+      "step": 590320
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 7.291650723195836e-06,
+      "loss": 0.697,
+      "step": 590330
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 7.290502793296088e-06,
+      "loss": 0.743,
+      "step": 590340
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 7.2893548633963414e-06,
+      "loss": 0.7237,
+      "step": 590350
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 7.288206933496594e-06,
+      "loss": 0.7484,
+      "step": 590360
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 7.287059003596846e-06,
+      "loss": 0.7776,
+      "step": 590370
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 7.2859110736970995e-06,
+      "loss": 0.7328,
+      "step": 590380
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 7.284763143797352e-06,
+      "loss": 0.7799,
+      "step": 590390
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 7.283615213897603e-06,
+      "loss": 0.6912,
+      "step": 590400
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 7.282467283997857e-06,
+      "loss": 0.7335,
+      "step": 590410
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 7.281319354098109e-06,
+      "loss": 0.6588,
+      "step": 590420
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 7.2801714241983614e-06,
+      "loss": 0.6647,
+      "step": 590430
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 7.279023494298615e-06,
+      "loss": 0.6628,
+      "step": 590440
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 7.277875564398867e-06,
+      "loss": 0.7686,
+      "step": 590450
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 7.2767276344991195e-06,
+      "loss": 0.7225,
+      "step": 590460
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 7.275579704599371e-06,
+      "loss": 0.6652,
+      "step": 590470
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 7.274431774699624e-06,
+      "loss": 0.6479,
+      "step": 590480
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 7.273283844799877e-06,
+      "loss": 0.6537,
+      "step": 590490
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 7.272135914900129e-06,
+      "loss": 0.6479,
+      "step": 590500
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 7.270987985000382e-06,
+      "loss": 0.7228,
+      "step": 590510
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 7.269840055100635e-06,
+      "loss": 0.6773,
+      "step": 590520
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 7.268692125200887e-06,
+      "loss": 0.6758,
+      "step": 590530
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 7.26754419530114e-06,
+      "loss": 0.767,
+      "step": 590540
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 7.266396265401392e-06,
+      "loss": 0.7658,
+      "step": 590550
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 7.265248335501644e-06,
+      "loss": 0.685,
+      "step": 590560
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 7.2641004056018975e-06,
+      "loss": 0.6042,
+      "step": 590570
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 7.26295247570215e-06,
+      "loss": 0.6395,
+      "step": 590580
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 7.261804545802402e-06,
+      "loss": 0.6201,
+      "step": 590590
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 7.2606566159026555e-06,
+      "loss": 0.7223,
+      "step": 590600
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 7.259508686002908e-06,
+      "loss": 0.7005,
+      "step": 590610
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 7.2583607561031595e-06,
+      "loss": 0.6091,
+      "step": 590620
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 7.257212826203412e-06,
+      "loss": 0.637,
+      "step": 590630
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 7.256064896303665e-06,
+      "loss": 0.689,
+      "step": 590640
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 7.2549169664039175e-06,
+      "loss": 0.7381,
+      "step": 590650
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 7.25376903650417e-06,
+      "loss": 0.6822,
+      "step": 590660
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 7.252621106604423e-06,
+      "loss": 0.6884,
+      "step": 590670
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 7.2514731767046756e-06,
+      "loss": 0.7391,
+      "step": 590680
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 7.250325246804927e-06,
+      "loss": 0.6709,
+      "step": 590690
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 7.24917731690518e-06,
+      "loss": 0.6998,
+      "step": 590700
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 7.248029387005433e-06,
+      "loss": 0.7496,
+      "step": 590710
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 7.246881457105685e-06,
+      "loss": 0.7432,
+      "step": 590720
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 7.245733527205938e-06,
+      "loss": 0.7021,
+      "step": 590730
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 7.244585597306191e-06,
+      "loss": 0.747,
+      "step": 590740
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 7.243437667406443e-06,
+      "loss": 0.7185,
+      "step": 590750
+    },
+    {
+      "epoch": 27.1,
+      "learning_rate": 7.242289737506696e-06,
+      "loss": 0.7093,
+      "step": 590760
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 7.241141807606948e-06,
+      "loss": 0.6958,
+      "step": 590770
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 7.2399938777072e-06,
+      "loss": 0.6912,
+      "step": 590780
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 7.238845947807454e-06,
+      "loss": 0.6718,
+      "step": 590790
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 7.237698017907706e-06,
+      "loss": 0.628,
+      "step": 590800
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 7.236550088007958e-06,
+      "loss": 0.6967,
+      "step": 590810
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 7.235402158108211e-06,
+      "loss": 0.6776,
+      "step": 590820
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 7.234254228208464e-06,
+      "loss": 0.676,
+      "step": 590830
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 7.233106298308716e-06,
+      "loss": 0.6615,
+      "step": 590840
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 7.231958368408968e-06,
+      "loss": 0.6531,
+      "step": 590850
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 7.230810438509221e-06,
+      "loss": 0.7052,
+      "step": 590860
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 7.229662508609474e-06,
+      "loss": 0.6934,
+      "step": 590870
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 7.228514578709726e-06,
+      "loss": 0.6711,
+      "step": 590880
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 7.227366648809979e-06,
+      "loss": 0.6792,
+      "step": 590890
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 7.226218718910232e-06,
+      "loss": 0.6991,
+      "step": 590900
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 7.225070789010483e-06,
+      "loss": 0.6883,
+      "step": 590910
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 7.2239228591107364e-06,
+      "loss": 0.7781,
+      "step": 590920
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 7.222774929210989e-06,
+      "loss": 0.6328,
+      "step": 590930
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 7.221626999311241e-06,
+      "loss": 0.6153,
+      "step": 590940
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 7.2204790694114945e-06,
+      "loss": 0.6725,
+      "step": 590950
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 7.219331139511747e-06,
+      "loss": 0.8121,
+      "step": 590960
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 7.218183209611999e-06,
+      "loss": 0.7168,
+      "step": 590970
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 7.217035279712251e-06,
+      "loss": 0.7061,
+      "step": 590980
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 7.215887349812504e-06,
+      "loss": 0.6737,
+      "step": 590990
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 7.2147394199127564e-06,
+      "loss": 0.6826,
+      "step": 591000
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 7.213591490013009e-06,
+      "loss": 0.683,
+      "step": 591010
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 7.212443560113262e-06,
+      "loss": 0.6854,
+      "step": 591020
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 7.2112956302135145e-06,
+      "loss": 0.6863,
+      "step": 591030
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 7.210147700313767e-06,
+      "loss": 0.6967,
+      "step": 591040
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 7.20899977041402e-06,
+      "loss": 0.6948,
+      "step": 591050
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 7.207851840514272e-06,
+      "loss": 0.6903,
+      "step": 591060
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 7.206703910614524e-06,
+      "loss": 0.7651,
+      "step": 591070
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 7.205555980714777e-06,
+      "loss": 0.7571,
+      "step": 591080
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 7.20440805081503e-06,
+      "loss": 0.7285,
+      "step": 591090
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 7.203260120915282e-06,
+      "loss": 0.7473,
+      "step": 591100
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 7.202112191015535e-06,
+      "loss": 0.6686,
+      "step": 591110
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 7.200964261115788e-06,
+      "loss": 0.6698,
+      "step": 591120
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 7.199816331216039e-06,
+      "loss": 0.7178,
+      "step": 591130
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 7.1986684013162925e-06,
+      "loss": 0.6846,
+      "step": 591140
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 7.197520471416545e-06,
+      "loss": 0.6992,
+      "step": 591150
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 7.196372541516797e-06,
+      "loss": 0.6771,
+      "step": 591160
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 7.19522461161705e-06,
+      "loss": 0.7129,
+      "step": 591170
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 7.194076681717303e-06,
+      "loss": 0.6834,
+      "step": 591180
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 7.192928751817555e-06,
+      "loss": 0.6836,
+      "step": 591190
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 7.191780821917807e-06,
+      "loss": 0.6683,
+      "step": 591200
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 7.19063289201806e-06,
+      "loss": 0.5905,
+      "step": 591210
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 7.1894849621183125e-06,
+      "loss": 0.6873,
+      "step": 591220
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 7.188337032218565e-06,
+      "loss": 0.7711,
+      "step": 591230
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 7.187189102318818e-06,
+      "loss": 0.695,
+      "step": 591240
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 7.1860411724190706e-06,
+      "loss": 0.7156,
+      "step": 591250
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 7.184893242519323e-06,
+      "loss": 0.6716,
+      "step": 591260
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 7.183745312619576e-06,
+      "loss": 0.6762,
+      "step": 591270
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 7.182597382719828e-06,
+      "loss": 0.6019,
+      "step": 591280
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 7.18144945282008e-06,
+      "loss": 0.7615,
+      "step": 591290
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 7.180301522920333e-06,
+      "loss": 0.6967,
+      "step": 591300
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 7.179153593020586e-06,
+      "loss": 0.6768,
+      "step": 591310
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 7.178005663120838e-06,
+      "loss": 0.726,
+      "step": 591320
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 7.1768577332210906e-06,
+      "loss": 0.6455,
+      "step": 591330
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 7.175709803321344e-06,
+      "loss": 0.6285,
+      "step": 591340
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 7.174561873421595e-06,
+      "loss": 0.7829,
+      "step": 591350
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 7.173413943521848e-06,
+      "loss": 0.7014,
+      "step": 591360
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 7.172266013622101e-06,
+      "loss": 0.711,
+      "step": 591370
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 7.171118083722353e-06,
+      "loss": 0.6745,
+      "step": 591380
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 7.169970153822606e-06,
+      "loss": 0.6891,
+      "step": 591390
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 7.168822223922859e-06,
+      "loss": 0.7377,
+      "step": 591400
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 7.1676742940231114e-06,
+      "loss": 0.7174,
+      "step": 591410
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 7.166526364123363e-06,
+      "loss": 0.7044,
+      "step": 591420
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 7.165378434223616e-06,
+      "loss": 0.7531,
+      "step": 591430
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 7.164230504323869e-06,
+      "loss": 0.746,
+      "step": 591440
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 7.163082574424121e-06,
+      "loss": 0.7235,
+      "step": 591450
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 7.161934644524374e-06,
+      "loss": 0.7378,
+      "step": 591460
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 7.160786714624627e-06,
+      "loss": 0.657,
+      "step": 591470
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 7.159638784724879e-06,
+      "loss": 0.6928,
+      "step": 591480
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 7.158490854825131e-06,
+      "loss": 0.7137,
+      "step": 591490
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 7.157342924925384e-06,
+      "loss": 0.6784,
+      "step": 591500
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 7.156194995025636e-06,
+      "loss": 0.6745,
+      "step": 591510
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 7.155047065125889e-06,
+      "loss": 0.6421,
+      "step": 591520
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 7.153899135226142e-06,
+      "loss": 0.6772,
+      "step": 591530
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 7.152751205326394e-06,
+      "loss": 0.6926,
+      "step": 591540
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 7.151603275426647e-06,
+      "loss": 0.6319,
+      "step": 591550
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 7.1504553455269e-06,
+      "loss": 0.7098,
+      "step": 591560
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 7.1493074156271515e-06,
+      "loss": 0.6344,
+      "step": 591570
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 7.148159485727404e-06,
+      "loss": 0.638,
+      "step": 591580
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 7.147011555827657e-06,
+      "loss": 0.6977,
+      "step": 591590
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 7.1458636259279095e-06,
+      "loss": 0.7592,
+      "step": 591600
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 7.144715696028162e-06,
+      "loss": 0.7141,
+      "step": 591610
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 7.143567766128415e-06,
+      "loss": 0.6751,
+      "step": 591620
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 7.1424198362286675e-06,
+      "loss": 0.7358,
+      "step": 591630
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 7.141271906328919e-06,
+      "loss": 0.6898,
+      "step": 591640
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 7.140123976429172e-06,
+      "loss": 0.7192,
+      "step": 591650
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 7.138976046529425e-06,
+      "loss": 0.7555,
+      "step": 591660
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 7.137828116629677e-06,
+      "loss": 0.7278,
+      "step": 591670
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 7.1366801867299295e-06,
+      "loss": 0.7145,
+      "step": 591680
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 7.135532256830183e-06,
+      "loss": 0.7259,
+      "step": 591690
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 7.134384326930435e-06,
+      "loss": 0.6783,
+      "step": 591700
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 7.133236397030687e-06,
+      "loss": 0.771,
+      "step": 591710
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 7.13208846713094e-06,
+      "loss": 0.638,
+      "step": 591720
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 7.130940537231192e-06,
+      "loss": 0.6721,
+      "step": 591730
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 7.129792607331445e-06,
+      "loss": 0.6914,
+      "step": 591740
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 7.128644677431698e-06,
+      "loss": 0.725,
+      "step": 591750
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 7.12749674753195e-06,
+      "loss": 0.6907,
+      "step": 591760
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 7.126348817632203e-06,
+      "loss": 0.6595,
+      "step": 591770
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 7.125200887732456e-06,
+      "loss": 0.7436,
+      "step": 591780
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 7.1240529578327075e-06,
+      "loss": 0.6654,
+      "step": 591790
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 7.12290502793296e-06,
+      "loss": 0.7227,
+      "step": 591800
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 7.121757098033213e-06,
+      "loss": 0.6823,
+      "step": 591810
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 7.1206091681334656e-06,
+      "loss": 0.6799,
+      "step": 591820
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 7.119461238233718e-06,
+      "loss": 0.7744,
+      "step": 591830
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 7.11831330833397e-06,
+      "loss": 0.6138,
+      "step": 591840
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 7.117165378434224e-06,
+      "loss": 0.677,
+      "step": 591850
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 7.116017448534475e-06,
+      "loss": 0.6864,
+      "step": 591860
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 7.1148695186347275e-06,
+      "loss": 0.6867,
+      "step": 591870
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 7.113721588734981e-06,
+      "loss": 0.6613,
+      "step": 591880
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 7.112573658835233e-06,
+      "loss": 0.7269,
+      "step": 591890
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 7.111425728935486e-06,
+      "loss": 0.6794,
+      "step": 591900
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 7.110277799035739e-06,
+      "loss": 0.6879,
+      "step": 591910
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 7.109129869135991e-06,
+      "loss": 0.6766,
+      "step": 591920
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 7.107981939236243e-06,
+      "loss": 0.7104,
+      "step": 591930
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 7.106834009336496e-06,
+      "loss": 0.7331,
+      "step": 591940
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 7.105686079436748e-06,
+      "loss": 0.6708,
+      "step": 591950
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 7.104538149537001e-06,
+      "loss": 0.6432,
+      "step": 591960
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 7.103390219637254e-06,
+      "loss": 0.6572,
+      "step": 591970
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 7.1022422897375064e-06,
+      "loss": 0.6903,
+      "step": 591980
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 7.101094359837759e-06,
+      "loss": 0.7218,
+      "step": 591990
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 7.09994642993801e-06,
+      "loss": 0.6564,
+      "step": 592000
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 7.098798500038264e-06,
+      "loss": 0.7018,
+      "step": 592010
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 7.097650570138516e-06,
+      "loss": 0.6244,
+      "step": 592020
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 7.096502640238768e-06,
+      "loss": 0.6973,
+      "step": 592030
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 7.095354710339022e-06,
+      "loss": 0.6908,
+      "step": 592040
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 7.094206780439274e-06,
+      "loss": 0.6704,
+      "step": 592050
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 7.0930588505395264e-06,
+      "loss": 0.7577,
+      "step": 592060
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 7.09191092063978e-06,
+      "loss": 0.6908,
+      "step": 592070
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 7.090762990740031e-06,
+      "loss": 0.6727,
+      "step": 592080
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 7.089615060840284e-06,
+      "loss": 0.6626,
+      "step": 592090
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 7.088467130940537e-06,
+      "loss": 0.6912,
+      "step": 592100
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 7.087319201040789e-06,
+      "loss": 0.6838,
+      "step": 592110
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 7.086171271141042e-06,
+      "loss": 0.7202,
+      "step": 592120
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 7.085023341241295e-06,
+      "loss": 0.6961,
+      "step": 592130
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 7.083875411341547e-06,
+      "loss": 0.8029,
+      "step": 592140
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 7.082727481441799e-06,
+      "loss": 0.6674,
+      "step": 592150
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 7.081579551542052e-06,
+      "loss": 0.6754,
+      "step": 592160
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 7.0804316216423045e-06,
+      "loss": 0.7052,
+      "step": 592170
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 7.079283691742557e-06,
+      "loss": 0.7167,
+      "step": 592180
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 7.078135761842809e-06,
+      "loss": 0.7538,
+      "step": 592190
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 7.0769878319430625e-06,
+      "loss": 0.6528,
+      "step": 592200
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 7.075839902043315e-06,
+      "loss": 0.7585,
+      "step": 592210
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 7.0746919721435665e-06,
+      "loss": 0.7582,
+      "step": 592220
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 7.07354404224382e-06,
+      "loss": 0.6258,
+      "step": 592230
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 7.072396112344072e-06,
+      "loss": 0.6675,
+      "step": 592240
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 7.0712481824443245e-06,
+      "loss": 0.7506,
+      "step": 592250
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 7.070100252544578e-06,
+      "loss": 0.7435,
+      "step": 592260
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 7.06895232264483e-06,
+      "loss": 0.7218,
+      "step": 592270
+    },
+    {
+      "epoch": 27.17,
+      "learning_rate": 7.0678043927450825e-06,
+      "loss": 0.6945,
+      "step": 592280
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 7.066656462845336e-06,
+      "loss": 0.6685,
+      "step": 592290
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 7.065508532945587e-06,
+      "loss": 0.7415,
+      "step": 592300
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 7.06436060304584e-06,
+      "loss": 0.7419,
+      "step": 592310
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 7.063212673146093e-06,
+      "loss": 0.7133,
+      "step": 592320
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 7.062064743246345e-06,
+      "loss": 0.7708,
+      "step": 592330
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 7.060916813346598e-06,
+      "loss": 0.7207,
+      "step": 592340
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 7.059768883446849e-06,
+      "loss": 0.7211,
+      "step": 592350
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 7.058620953547103e-06,
+      "loss": 0.7253,
+      "step": 592360
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 7.057473023647355e-06,
+      "loss": 0.6575,
+      "step": 592370
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 7.056325093747607e-06,
+      "loss": 0.7008,
+      "step": 592380
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 7.055177163847861e-06,
+      "loss": 0.7307,
+      "step": 592390
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 7.054029233948113e-06,
+      "loss": 0.6505,
+      "step": 592400
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 7.052881304048365e-06,
+      "loss": 0.7277,
+      "step": 592410
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 7.051733374148619e-06,
+      "loss": 0.7065,
+      "step": 592420
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 7.050585444248871e-06,
+      "loss": 0.6847,
+      "step": 592430
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 7.0494375143491226e-06,
+      "loss": 0.7111,
+      "step": 592440
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 7.048289584449376e-06,
+      "loss": 0.6353,
+      "step": 592450
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 7.047141654549628e-06,
+      "loss": 0.6593,
+      "step": 592460
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 7.045993724649881e-06,
+      "loss": 0.7365,
+      "step": 592470
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 7.044845794750134e-06,
+      "loss": 0.7405,
+      "step": 592480
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 7.043697864850386e-06,
+      "loss": 0.707,
+      "step": 592490
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 7.042549934950638e-06,
+      "loss": 0.7501,
+      "step": 592500
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 7.041402005050892e-06,
+      "loss": 0.7264,
+      "step": 592510
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 7.040254075151143e-06,
+      "loss": 0.7362,
+      "step": 592520
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 7.039106145251396e-06,
+      "loss": 0.6485,
+      "step": 592530
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 7.037958215351648e-06,
+      "loss": 0.7504,
+      "step": 592540
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 7.0368102854519014e-06,
+      "loss": 0.6722,
+      "step": 592550
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 7.035662355552154e-06,
+      "loss": 0.7176,
+      "step": 592560
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 7.034514425652405e-06,
+      "loss": 0.6167,
+      "step": 592570
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 7.0333664957526595e-06,
+      "loss": 0.7683,
+      "step": 592580
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 7.032218565852911e-06,
+      "loss": 0.7091,
+      "step": 592590
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 7.0310706359531634e-06,
+      "loss": 0.738,
+      "step": 592600
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 7.029922706053417e-06,
+      "loss": 0.6376,
+      "step": 592610
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 7.028774776153669e-06,
+      "loss": 0.7437,
+      "step": 592620
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 7.0276268462539215e-06,
+      "loss": 0.5616,
+      "step": 592630
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 7.026478916354175e-06,
+      "loss": 0.6728,
+      "step": 592640
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 7.025330986454427e-06,
+      "loss": 0.7543,
+      "step": 592650
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 7.024183056554679e-06,
+      "loss": 0.696,
+      "step": 592660
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 7.023035126654932e-06,
+      "loss": 0.6429,
+      "step": 592670
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 7.021887196755184e-06,
+      "loss": 0.6365,
+      "step": 592680
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 7.020739266855437e-06,
+      "loss": 0.7038,
+      "step": 592690
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 7.019591336955689e-06,
+      "loss": 0.687,
+      "step": 592700
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 7.018443407055942e-06,
+      "loss": 0.6231,
+      "step": 592710
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 7.017295477156194e-06,
+      "loss": 0.676,
+      "step": 592720
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 7.016147547256446e-06,
+      "loss": 0.7137,
+      "step": 592730
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 7.0149996173566995e-06,
+      "loss": 0.792,
+      "step": 592740
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 7.013851687456952e-06,
+      "loss": 0.6467,
+      "step": 592750
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 7.012703757557204e-06,
+      "loss": 0.6596,
+      "step": 592760
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 7.0115558276574575e-06,
+      "loss": 0.6484,
+      "step": 592770
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 7.01040789775771e-06,
+      "loss": 0.6888,
+      "step": 592780
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 7.0092599678579615e-06,
+      "loss": 0.6492,
+      "step": 592790
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 7.0081120379582156e-06,
+      "loss": 0.7003,
+      "step": 592800
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 7.006964108058467e-06,
+      "loss": 0.6889,
+      "step": 592810
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 7.0058161781587195e-06,
+      "loss": 0.6733,
+      "step": 592820
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 7.004668248258973e-06,
+      "loss": 0.6838,
+      "step": 592830
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 7.003520318359225e-06,
+      "loss": 0.7137,
+      "step": 592840
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 7.0023723884594775e-06,
+      "loss": 0.6509,
+      "step": 592850
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 7.001224458559729e-06,
+      "loss": 0.6632,
+      "step": 592860
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 7.000076528659982e-06,
+      "loss": 0.6708,
+      "step": 592870
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 6.998928598760235e-06,
+      "loss": 0.6582,
+      "step": 592880
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 6.997780668860487e-06,
+      "loss": 0.7194,
+      "step": 592890
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 6.99663273896074e-06,
+      "loss": 0.6735,
+      "step": 592900
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 6.995484809060993e-06,
+      "loss": 0.7237,
+      "step": 592910
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 6.994336879161245e-06,
+      "loss": 0.7558,
+      "step": 592920
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 6.993188949261498e-06,
+      "loss": 0.6492,
+      "step": 592930
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 6.99204101936175e-06,
+      "loss": 0.6794,
+      "step": 592940
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 6.990893089462002e-06,
+      "loss": 0.6869,
+      "step": 592950
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 6.989745159562256e-06,
+      "loss": 0.663,
+      "step": 592960
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 6.988597229662508e-06,
+      "loss": 0.7382,
+      "step": 592970
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 6.98744929976276e-06,
+      "loss": 0.6958,
+      "step": 592980
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 6.986301369863014e-06,
+      "loss": 0.7092,
+      "step": 592990
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 6.985153439963266e-06,
+      "loss": 0.7972,
+      "step": 593000
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 6.9840055100635176e-06,
+      "loss": 0.7467,
+      "step": 593010
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 6.982857580163771e-06,
+      "loss": 0.6976,
+      "step": 593020
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 6.981709650264023e-06,
+      "loss": 0.7041,
+      "step": 593030
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 6.980561720364276e-06,
+      "loss": 0.7038,
+      "step": 593040
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 6.979413790464528e-06,
+      "loss": 0.7168,
+      "step": 593050
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 6.978265860564781e-06,
+      "loss": 0.6788,
+      "step": 593060
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 6.977117930665034e-06,
+      "loss": 0.7489,
+      "step": 593070
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 6.975970000765285e-06,
+      "loss": 0.6537,
+      "step": 593080
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 6.974822070865538e-06,
+      "loss": 0.5772,
+      "step": 593090
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 6.973674140965791e-06,
+      "loss": 0.6954,
+      "step": 593100
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 6.972526211066043e-06,
+      "loss": 0.6514,
+      "step": 593110
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 6.9713782811662965e-06,
+      "loss": 0.717,
+      "step": 593120
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 6.970230351266549e-06,
+      "loss": 0.6983,
+      "step": 593130
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 6.969082421366801e-06,
+      "loss": 0.6993,
+      "step": 593140
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 6.9679344914670545e-06,
+      "loss": 0.7408,
+      "step": 593150
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 6.966786561567306e-06,
+      "loss": 0.6613,
+      "step": 593160
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 6.9656386316675584e-06,
+      "loss": 0.7322,
+      "step": 593170
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 6.964490701767812e-06,
+      "loss": 0.6766,
+      "step": 593180
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 6.963342771868064e-06,
+      "loss": 0.6531,
+      "step": 593190
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 6.9621948419683165e-06,
+      "loss": 0.7104,
+      "step": 593200
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 6.961046912068569e-06,
+      "loss": 0.6803,
+      "step": 593210
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 6.959898982168822e-06,
+      "loss": 0.6795,
+      "step": 593220
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 6.958751052269074e-06,
+      "loss": 0.7471,
+      "step": 593230
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 6.957603122369326e-06,
+      "loss": 0.6701,
+      "step": 593240
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 6.956455192469579e-06,
+      "loss": 0.7319,
+      "step": 593250
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 6.955307262569832e-06,
+      "loss": 0.6366,
+      "step": 593260
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 6.954159332670084e-06,
+      "loss": 0.6793,
+      "step": 593270
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 6.953011402770337e-06,
+      "loss": 0.7561,
+      "step": 593280
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 6.95186347287059e-06,
+      "loss": 0.6738,
+      "step": 593290
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 6.950715542970841e-06,
+      "loss": 0.672,
+      "step": 593300
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 6.9495676130710945e-06,
+      "loss": 0.7142,
+      "step": 593310
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 6.948419683171347e-06,
+      "loss": 0.6826,
+      "step": 593320
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 6.947271753271599e-06,
+      "loss": 0.6915,
+      "step": 593330
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 6.9461238233718525e-06,
+      "loss": 0.7956,
+      "step": 593340
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 6.944975893472105e-06,
+      "loss": 0.6279,
+      "step": 593350
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 6.943827963572357e-06,
+      "loss": 0.6827,
+      "step": 593360
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 6.9426800336726106e-06,
+      "loss": 0.6687,
+      "step": 593370
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 6.941532103772862e-06,
+      "loss": 0.7077,
+      "step": 593380
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 6.9403841738731145e-06,
+      "loss": 0.6517,
+      "step": 593390
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 6.939236243973367e-06,
+      "loss": 0.6744,
+      "step": 593400
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 6.93808831407362e-06,
+      "loss": 0.6758,
+      "step": 593410
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 6.9369403841738726e-06,
+      "loss": 0.7081,
+      "step": 593420
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 6.935792454274125e-06,
+      "loss": 0.7315,
+      "step": 593430
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 6.934644524374378e-06,
+      "loss": 0.6997,
+      "step": 593440
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 6.93349659447463e-06,
+      "loss": 0.6946,
+      "step": 593450
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 6.932348664574882e-06,
+      "loss": 0.6537,
+      "step": 593460
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 6.931200734675135e-06,
+      "loss": 0.8245,
+      "step": 593470
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 6.930052804775388e-06,
+      "loss": 0.651,
+      "step": 593480
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 6.92890487487564e-06,
+      "loss": 0.6897,
+      "step": 593490
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 6.927756944975893e-06,
+      "loss": 0.6992,
+      "step": 593500
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 6.926609015076146e-06,
+      "loss": 0.7124,
+      "step": 593510
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 6.925461085176397e-06,
+      "loss": 0.6425,
+      "step": 593520
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 6.924313155276651e-06,
+      "loss": 0.6755,
+      "step": 593530
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 6.923165225376903e-06,
+      "loss": 0.7381,
+      "step": 593540
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 6.922017295477155e-06,
+      "loss": 0.7435,
+      "step": 593550
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 6.920869365577408e-06,
+      "loss": 0.6882,
+      "step": 593560
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 6.919721435677661e-06,
+      "loss": 0.7612,
+      "step": 593570
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 6.918573505777913e-06,
+      "loss": 0.7108,
+      "step": 593580
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 6.917425575878165e-06,
+      "loss": 0.6711,
+      "step": 593590
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 6.916277645978418e-06,
+      "loss": 0.659,
+      "step": 593600
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 6.915129716078671e-06,
+      "loss": 0.6635,
+      "step": 593610
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 6.913981786178923e-06,
+      "loss": 0.6474,
+      "step": 593620
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 6.912833856279176e-06,
+      "loss": 0.7276,
+      "step": 593630
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 6.911685926379429e-06,
+      "loss": 0.6994,
+      "step": 593640
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 6.910537996479681e-06,
+      "loss": 0.6554,
+      "step": 593650
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 6.909390066579934e-06,
+      "loss": 0.7339,
+      "step": 593660
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 6.908242136680186e-06,
+      "loss": 0.6584,
+      "step": 593670
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 6.907094206780438e-06,
+      "loss": 0.6848,
+      "step": 593680
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 6.9059462768806915e-06,
+      "loss": 0.6969,
+      "step": 593690
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 6.904798346980944e-06,
+      "loss": 0.7072,
+      "step": 593700
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 6.903650417081196e-06,
+      "loss": 0.6887,
+      "step": 593710
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 6.902502487181449e-06,
+      "loss": 0.6981,
+      "step": 593720
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 6.901354557281702e-06,
+      "loss": 0.6757,
+      "step": 593730
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 6.9002066273819534e-06,
+      "loss": 0.6477,
+      "step": 593740
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 6.899058697482206e-06,
+      "loss": 0.6812,
+      "step": 593750
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 6.897910767582459e-06,
+      "loss": 0.6794,
+      "step": 593760
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 6.8967628376827115e-06,
+      "loss": 0.6746,
+      "step": 593770
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 6.895614907782964e-06,
+      "loss": 0.6942,
+      "step": 593780
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 6.894466977883217e-06,
+      "loss": 0.7041,
+      "step": 593790
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 6.8933190479834695e-06,
+      "loss": 0.6776,
+      "step": 593800
+    },
+    {
+      "epoch": 27.24,
+      "learning_rate": 6.892171118083721e-06,
+      "loss": 0.6273,
+      "step": 593810
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 6.891023188183974e-06,
+      "loss": 0.693,
+      "step": 593820
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 6.889875258284227e-06,
+      "loss": 0.6809,
+      "step": 593830
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 6.888727328384479e-06,
+      "loss": 0.6549,
+      "step": 593840
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 6.887579398484732e-06,
+      "loss": 0.7773,
+      "step": 593850
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 6.886431468584985e-06,
+      "loss": 0.6628,
+      "step": 593860
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 6.885283538685237e-06,
+      "loss": 0.6957,
+      "step": 593870
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 6.88413560878549e-06,
+      "loss": 0.6749,
+      "step": 593880
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 6.882987678885742e-06,
+      "loss": 0.7598,
+      "step": 593890
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 6.881839748985994e-06,
+      "loss": 0.6844,
+      "step": 593900
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 6.880691819086247e-06,
+      "loss": 0.6987,
+      "step": 593910
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 6.8795438891865e-06,
+      "loss": 0.7373,
+      "step": 593920
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 6.878395959286752e-06,
+      "loss": 0.6879,
+      "step": 593930
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 6.877248029387005e-06,
+      "loss": 0.6803,
+      "step": 593940
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 6.876100099487258e-06,
+      "loss": 0.7099,
+      "step": 593950
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 6.8749521695875095e-06,
+      "loss": 0.7118,
+      "step": 593960
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 6.873804239687762e-06,
+      "loss": 0.7231,
+      "step": 593970
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 6.872656309788015e-06,
+      "loss": 0.7037,
+      "step": 593980
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 6.8715083798882676e-06,
+      "loss": 0.6369,
+      "step": 593990
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 6.87036044998852e-06,
+      "loss": 0.815,
+      "step": 594000
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 6.869212520088773e-06,
+      "loss": 0.6625,
+      "step": 594010
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 6.868064590189026e-06,
+      "loss": 0.7149,
+      "step": 594020
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 6.866916660289277e-06,
+      "loss": 0.699,
+      "step": 594030
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 6.86576873038953e-06,
+      "loss": 0.6904,
+      "step": 594040
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 6.864620800489783e-06,
+      "loss": 0.7041,
+      "step": 594050
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 6.863472870590035e-06,
+      "loss": 0.6441,
+      "step": 594060
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 6.8623249406902876e-06,
+      "loss": 0.6432,
+      "step": 594070
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 6.861177010790541e-06,
+      "loss": 0.6608,
+      "step": 594080
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 6.860029080890793e-06,
+      "loss": 0.6387,
+      "step": 594090
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 6.858881150991045e-06,
+      "loss": 0.7144,
+      "step": 594100
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 6.857733221091298e-06,
+      "loss": 0.6728,
+      "step": 594110
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 6.85658529119155e-06,
+      "loss": 0.6603,
+      "step": 594120
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 6.855437361291803e-06,
+      "loss": 0.7063,
+      "step": 594130
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 6.854289431392056e-06,
+      "loss": 0.7225,
+      "step": 594140
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 6.8531415014923084e-06,
+      "loss": 0.7162,
+      "step": 594150
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 6.851993571592561e-06,
+      "loss": 0.712,
+      "step": 594160
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 6.850845641692814e-06,
+      "loss": 0.7335,
+      "step": 594170
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 6.849697711793066e-06,
+      "loss": 0.6729,
+      "step": 594180
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 6.848549781893318e-06,
+      "loss": 0.6777,
+      "step": 594190
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 6.847401851993571e-06,
+      "loss": 0.7005,
+      "step": 594200
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 6.846253922093824e-06,
+      "loss": 0.7464,
+      "step": 594210
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 6.845105992194076e-06,
+      "loss": 0.6745,
+      "step": 594220
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 6.8439580622943284e-06,
+      "loss": 0.6923,
+      "step": 594230
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 6.842810132394582e-06,
+      "loss": 0.699,
+      "step": 594240
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 6.841662202494833e-06,
+      "loss": 0.6979,
+      "step": 594250
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 6.840514272595086e-06,
+      "loss": 0.6634,
+      "step": 594260
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 6.839366342695339e-06,
+      "loss": 0.7124,
+      "step": 594270
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 6.838218412795591e-06,
+      "loss": 0.7949,
+      "step": 594280
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 6.837070482895844e-06,
+      "loss": 0.6866,
+      "step": 594290
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 6.835922552996097e-06,
+      "loss": 0.6795,
+      "step": 594300
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 6.834774623096349e-06,
+      "loss": 0.7304,
+      "step": 594310
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 6.833626693196601e-06,
+      "loss": 0.7455,
+      "step": 594320
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 6.832478763296854e-06,
+      "loss": 0.6754,
+      "step": 594330
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 6.8313308333971065e-06,
+      "loss": 0.7528,
+      "step": 594340
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 6.830182903497359e-06,
+      "loss": 0.7306,
+      "step": 594350
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 6.829034973597612e-06,
+      "loss": 0.7525,
+      "step": 594360
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 6.8278870436978645e-06,
+      "loss": 0.6466,
+      "step": 594370
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 6.826739113798117e-06,
+      "loss": 0.6025,
+      "step": 594380
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 6.82559118389837e-06,
+      "loss": 0.6234,
+      "step": 594390
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 6.824443253998622e-06,
+      "loss": 0.6661,
+      "step": 594400
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 6.823295324098874e-06,
+      "loss": 0.6464,
+      "step": 594410
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 6.8221473941991265e-06,
+      "loss": 0.7029,
+      "step": 594420
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 6.82099946429938e-06,
+      "loss": 0.6284,
+      "step": 594430
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 6.819851534399632e-06,
+      "loss": 0.6039,
+      "step": 594440
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 6.8187036044998845e-06,
+      "loss": 0.6352,
+      "step": 594450
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 6.817555674600138e-06,
+      "loss": 0.6239,
+      "step": 594460
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 6.816407744700389e-06,
+      "loss": 0.6776,
+      "step": 594470
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 6.815259814800642e-06,
+      "loss": 0.7228,
+      "step": 594480
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 6.814111884900895e-06,
+      "loss": 0.7014,
+      "step": 594490
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 6.812963955001147e-06,
+      "loss": 0.716,
+      "step": 594500
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 6.8118160251014e-06,
+      "loss": 0.6965,
+      "step": 594510
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 6.810668095201653e-06,
+      "loss": 0.6719,
+      "step": 594520
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 6.809520165301905e-06,
+      "loss": 0.7184,
+      "step": 594530
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 6.808372235402157e-06,
+      "loss": 0.6375,
+      "step": 594540
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 6.80722430550241e-06,
+      "loss": 0.7555,
+      "step": 594550
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 6.8060763756026626e-06,
+      "loss": 0.7049,
+      "step": 594560
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 6.804928445702915e-06,
+      "loss": 0.8056,
+      "step": 594570
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 6.803780515803167e-06,
+      "loss": 0.6276,
+      "step": 594580
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 6.802632585903421e-06,
+      "loss": 0.7613,
+      "step": 594590
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 6.801484656003673e-06,
+      "loss": 0.6846,
+      "step": 594600
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 6.8003367261039245e-06,
+      "loss": 0.7195,
+      "step": 594610
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 6.799188796204178e-06,
+      "loss": 0.6386,
+      "step": 594620
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 6.79804086630443e-06,
+      "loss": 0.7717,
+      "step": 594630
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 6.796892936404683e-06,
+      "loss": 0.7123,
+      "step": 594640
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 6.795745006504936e-06,
+      "loss": 0.6946,
+      "step": 594650
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 6.794597076605188e-06,
+      "loss": 0.7049,
+      "step": 594660
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 6.793449146705441e-06,
+      "loss": 0.7381,
+      "step": 594670
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 6.792301216805694e-06,
+      "loss": 0.666,
+      "step": 594680
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 6.791153286905945e-06,
+      "loss": 0.6389,
+      "step": 594690
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 6.790005357006198e-06,
+      "loss": 0.6868,
+      "step": 594700
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 6.788857427106451e-06,
+      "loss": 0.705,
+      "step": 594710
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 6.7877094972067034e-06,
+      "loss": 0.5772,
+      "step": 594720
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 6.786561567306956e-06,
+      "loss": 0.7525,
+      "step": 594730
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 6.785413637407209e-06,
+      "loss": 0.7636,
+      "step": 594740
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 6.7842657075074615e-06,
+      "loss": 0.6917,
+      "step": 594750
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 6.783117777607713e-06,
+      "loss": 0.6705,
+      "step": 594760
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 6.781969847707965e-06,
+      "loss": 0.6354,
+      "step": 594770
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 6.780821917808219e-06,
+      "loss": 0.7384,
+      "step": 594780
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 6.779673987908471e-06,
+      "loss": 0.6907,
+      "step": 594790
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 6.7785260580087234e-06,
+      "loss": 0.6531,
+      "step": 594800
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 6.777378128108977e-06,
+      "loss": 0.6824,
+      "step": 594810
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 6.776230198209229e-06,
+      "loss": 0.6518,
+      "step": 594820
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 6.775082268309481e-06,
+      "loss": 0.7064,
+      "step": 594830
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 6.773934338409734e-06,
+      "loss": 0.657,
+      "step": 594840
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 6.772786408509986e-06,
+      "loss": 0.666,
+      "step": 594850
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 6.771638478610239e-06,
+      "loss": 0.5886,
+      "step": 594860
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 6.770490548710492e-06,
+      "loss": 0.6982,
+      "step": 594870
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 6.769342618810744e-06,
+      "loss": 0.6824,
+      "step": 594880
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 6.768194688910997e-06,
+      "loss": 0.7912,
+      "step": 594890
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 6.76704675901125e-06,
+      "loss": 0.6777,
+      "step": 594900
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 6.7658988291115015e-06,
+      "loss": 0.6686,
+      "step": 594910
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 6.764750899211754e-06,
+      "loss": 0.7338,
+      "step": 594920
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 6.763602969312006e-06,
+      "loss": 0.6647,
+      "step": 594930
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 6.7624550394122595e-06,
+      "loss": 0.6709,
+      "step": 594940
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 6.761307109512512e-06,
+      "loss": 0.6302,
+      "step": 594950
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 6.760159179612764e-06,
+      "loss": 0.6883,
+      "step": 594960
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 6.7590112497130176e-06,
+      "loss": 0.6919,
+      "step": 594970
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 6.757863319813269e-06,
+      "loss": 0.7124,
+      "step": 594980
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 6.7567153899135215e-06,
+      "loss": 0.6848,
+      "step": 594990
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 6.755567460013775e-06,
+      "loss": 0.6287,
+      "step": 595000
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 6.754419530114027e-06,
+      "loss": 0.707,
+      "step": 595010
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 6.7532716002142795e-06,
+      "loss": 0.733,
+      "step": 595020
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 6.752123670314533e-06,
+      "loss": 0.7309,
+      "step": 595030
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 6.750975740414785e-06,
+      "loss": 0.6979,
+      "step": 595040
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 6.749827810515037e-06,
+      "loss": 0.7104,
+      "step": 595050
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 6.74867988061529e-06,
+      "loss": 0.7426,
+      "step": 595060
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 6.747531950715542e-06,
+      "loss": 0.6999,
+      "step": 595070
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 6.746384020815795e-06,
+      "loss": 0.6964,
+      "step": 595080
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 6.745236090916047e-06,
+      "loss": 0.7294,
+      "step": 595090
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 6.7440881610163e-06,
+      "loss": 0.6631,
+      "step": 595100
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 6.742940231116553e-06,
+      "loss": 0.7132,
+      "step": 595110
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 6.741792301216804e-06,
+      "loss": 0.6347,
+      "step": 595120
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 6.7406443713170576e-06,
+      "loss": 0.7279,
+      "step": 595130
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 6.73949644141731e-06,
+      "loss": 0.7252,
+      "step": 595140
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 6.738348511517562e-06,
+      "loss": 0.7117,
+      "step": 595150
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 6.737200581617816e-06,
+      "loss": 0.5997,
+      "step": 595160
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 6.736052651718068e-06,
+      "loss": 0.7598,
+      "step": 595170
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 6.73490472181832e-06,
+      "loss": 0.7022,
+      "step": 595180
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 6.733756791918574e-06,
+      "loss": 0.7339,
+      "step": 595190
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 6.732608862018825e-06,
+      "loss": 0.6725,
+      "step": 595200
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 6.731460932119078e-06,
+      "loss": 0.7202,
+      "step": 595210
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 6.730313002219331e-06,
+      "loss": 0.7068,
+      "step": 595220
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 6.729165072319583e-06,
+      "loss": 0.6758,
+      "step": 595230
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 6.728017142419836e-06,
+      "loss": 0.7364,
+      "step": 595240
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 6.726869212520089e-06,
+      "loss": 0.6999,
+      "step": 595250
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 6.725721282620341e-06,
+      "loss": 0.6672,
+      "step": 595260
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 6.724573352720593e-06,
+      "loss": 0.6938,
+      "step": 595270
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 6.723425422820845e-06,
+      "loss": 0.7195,
+      "step": 595280
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 6.7222774929210984e-06,
+      "loss": 0.7304,
+      "step": 595290
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 6.721129563021351e-06,
+      "loss": 0.7343,
+      "step": 595300
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 6.719981633121603e-06,
+      "loss": 0.6662,
+      "step": 595310
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 6.7188337032218565e-06,
+      "loss": 0.6275,
+      "step": 595320
+    },
+    {
+      "epoch": 27.31,
+      "learning_rate": 6.717685773322109e-06,
+      "loss": 0.6263,
+      "step": 595330
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 6.71653784342236e-06,
+      "loss": 0.677,
+      "step": 595340
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 6.715389913522614e-06,
+      "loss": 0.768,
+      "step": 595350
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 6.714241983622866e-06,
+      "loss": 0.6355,
+      "step": 595360
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 6.7130940537231185e-06,
+      "loss": 0.7073,
+      "step": 595370
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 6.711946123823372e-06,
+      "loss": 0.6749,
+      "step": 595380
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 6.710798193923624e-06,
+      "loss": 0.7439,
+      "step": 595390
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 6.7096502640238765e-06,
+      "loss": 0.7288,
+      "step": 595400
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 6.70850233412413e-06,
+      "loss": 0.7255,
+      "step": 595410
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 6.707354404224381e-06,
+      "loss": 0.7092,
+      "step": 595420
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 6.706206474324634e-06,
+      "loss": 0.6853,
+      "step": 595430
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 6.705058544424886e-06,
+      "loss": 0.692,
+      "step": 595440
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 6.703910614525139e-06,
+      "loss": 0.769,
+      "step": 595450
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 6.702762684625392e-06,
+      "loss": 0.6493,
+      "step": 595460
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 6.701614754725644e-06,
+      "loss": 0.748,
+      "step": 595470
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 6.700466824825897e-06,
+      "loss": 0.7382,
+      "step": 595480
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 6.699318894926149e-06,
+      "loss": 0.6422,
+      "step": 595490
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 6.698170965026401e-06,
+      "loss": 0.7259,
+      "step": 595500
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 6.6970230351266545e-06,
+      "loss": 0.7461,
+      "step": 595510
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 6.695875105226907e-06,
+      "loss": 0.7231,
+      "step": 595520
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 6.694727175327159e-06,
+      "loss": 0.6708,
+      "step": 595530
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 6.6935792454274126e-06,
+      "loss": 0.7332,
+      "step": 595540
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 6.692431315527665e-06,
+      "loss": 0.7203,
+      "step": 595550
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 6.6912833856279165e-06,
+      "loss": 0.7229,
+      "step": 595560
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 6.69013545572817e-06,
+      "loss": 0.7449,
+      "step": 595570
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 6.688987525828422e-06,
+      "loss": 0.7143,
+      "step": 595580
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 6.6878395959286745e-06,
+      "loss": 0.786,
+      "step": 595590
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 6.686691666028928e-06,
+      "loss": 0.7208,
+      "step": 595600
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 6.68554373612918e-06,
+      "loss": 0.6832,
+      "step": 595610
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 6.6843958062294326e-06,
+      "loss": 0.6871,
+      "step": 595620
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 6.683247876329684e-06,
+      "loss": 0.6645,
+      "step": 595630
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 6.682099946429937e-06,
+      "loss": 0.6675,
+      "step": 595640
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 6.68095201653019e-06,
+      "loss": 0.6646,
+      "step": 595650
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 6.679804086630442e-06,
+      "loss": 0.732,
+      "step": 595660
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 6.678656156730695e-06,
+      "loss": 0.6752,
+      "step": 595670
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 6.677508226830948e-06,
+      "loss": 0.6294,
+      "step": 595680
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 6.6763602969312e-06,
+      "loss": 0.6939,
+      "step": 595690
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 6.6752123670314534e-06,
+      "loss": 0.6704,
+      "step": 595700
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 6.674064437131705e-06,
+      "loss": 0.6923,
+      "step": 595710
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 6.672916507231957e-06,
+      "loss": 0.6839,
+      "step": 595720
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 6.671768577332211e-06,
+      "loss": 0.6998,
+      "step": 595730
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 6.670620647432463e-06,
+      "loss": 0.7554,
+      "step": 595740
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 6.669472717532715e-06,
+      "loss": 0.7049,
+      "step": 595750
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 6.668324787632969e-06,
+      "loss": 0.6665,
+      "step": 595760
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 6.667176857733221e-06,
+      "loss": 0.7303,
+      "step": 595770
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 6.666028927833473e-06,
+      "loss": 0.6815,
+      "step": 595780
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 6.664880997933725e-06,
+      "loss": 0.6709,
+      "step": 595790
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 6.663733068033978e-06,
+      "loss": 0.6638,
+      "step": 595800
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 6.662585138134231e-06,
+      "loss": 0.6882,
+      "step": 595810
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 6.661437208234483e-06,
+      "loss": 0.6678,
+      "step": 595820
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 6.660289278334736e-06,
+      "loss": 0.7443,
+      "step": 595830
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 6.659141348434989e-06,
+      "loss": 0.7337,
+      "step": 595840
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 6.65799341853524e-06,
+      "loss": 0.658,
+      "step": 595850
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 6.6568454886354934e-06,
+      "loss": 0.5931,
+      "step": 595860
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 6.655697558735746e-06,
+      "loss": 0.6627,
+      "step": 595870
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 6.654549628835998e-06,
+      "loss": 0.6754,
+      "step": 595880
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 6.6534016989362515e-06,
+      "loss": 0.6335,
+      "step": 595890
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 6.652253769036504e-06,
+      "loss": 0.7397,
+      "step": 595900
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 6.651105839136756e-06,
+      "loss": 0.6822,
+      "step": 595910
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 6.6499579092370095e-06,
+      "loss": 0.6766,
+      "step": 595920
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 6.648809979337261e-06,
+      "loss": 0.7087,
+      "step": 595930
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 6.6476620494375135e-06,
+      "loss": 0.6052,
+      "step": 595940
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 6.646514119537766e-06,
+      "loss": 0.6567,
+      "step": 595950
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 6.645366189638019e-06,
+      "loss": 0.5963,
+      "step": 595960
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 6.6442182597382715e-06,
+      "loss": 0.6739,
+      "step": 595970
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 6.643070329838524e-06,
+      "loss": 0.6917,
+      "step": 595980
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 6.641922399938777e-06,
+      "loss": 0.6968,
+      "step": 595990
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 6.640774470039029e-06,
+      "loss": 0.5833,
+      "step": 596000
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 6.639626540139281e-06,
+      "loss": 0.6223,
+      "step": 596010
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 6.638478610239534e-06,
+      "loss": 0.6551,
+      "step": 596020
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 6.637330680339787e-06,
+      "loss": 0.6748,
+      "step": 596030
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 6.636182750440039e-06,
+      "loss": 0.7672,
+      "step": 596040
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 6.635034820540292e-06,
+      "loss": 0.6722,
+      "step": 596050
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 6.633886890640545e-06,
+      "loss": 0.7396,
+      "step": 596060
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 6.632738960740796e-06,
+      "loss": 0.7509,
+      "step": 596070
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 6.6315910308410495e-06,
+      "loss": 0.6243,
+      "step": 596080
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 6.630443100941302e-06,
+      "loss": 0.6017,
+      "step": 596090
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 6.629295171041554e-06,
+      "loss": 0.6565,
+      "step": 596100
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 6.6281472411418076e-06,
+      "loss": 0.6768,
+      "step": 596110
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 6.62699931124206e-06,
+      "loss": 0.6347,
+      "step": 596120
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 6.625851381342312e-06,
+      "loss": 0.6585,
+      "step": 596130
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 6.624703451442564e-06,
+      "loss": 0.7285,
+      "step": 596140
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 6.623555521542817e-06,
+      "loss": 0.6691,
+      "step": 596150
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 6.6224075916430695e-06,
+      "loss": 0.6928,
+      "step": 596160
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 6.621259661743322e-06,
+      "loss": 0.7371,
+      "step": 596170
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 6.620111731843575e-06,
+      "loss": 0.6767,
+      "step": 596180
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 6.618963801943828e-06,
+      "loss": 0.7701,
+      "step": 596190
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 6.61781587204408e-06,
+      "loss": 0.7052,
+      "step": 596200
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 6.616667942144333e-06,
+      "loss": 0.6591,
+      "step": 596210
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 6.615520012244585e-06,
+      "loss": 0.7626,
+      "step": 596220
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 6.614372082344837e-06,
+      "loss": 0.7187,
+      "step": 596230
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 6.61322415244509e-06,
+      "loss": 0.6869,
+      "step": 596240
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 6.612076222545343e-06,
+      "loss": 0.7454,
+      "step": 596250
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 6.610928292645595e-06,
+      "loss": 0.7593,
+      "step": 596260
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 6.6097803627458484e-06,
+      "loss": 0.714,
+      "step": 596270
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 6.608632432846101e-06,
+      "loss": 0.6718,
+      "step": 596280
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 6.607484502946352e-06,
+      "loss": 0.6895,
+      "step": 596290
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 6.606336573046605e-06,
+      "loss": 0.6595,
+      "step": 596300
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 6.605188643146858e-06,
+      "loss": 0.6996,
+      "step": 596310
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 6.60404071324711e-06,
+      "loss": 0.75,
+      "step": 596320
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 6.602892783347363e-06,
+      "loss": 0.6942,
+      "step": 596330
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 6.601744853447616e-06,
+      "loss": 0.771,
+      "step": 596340
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 6.6005969235478684e-06,
+      "loss": 0.705,
+      "step": 596350
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 6.59944899364812e-06,
+      "loss": 0.7593,
+      "step": 596360
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 6.598301063748373e-06,
+      "loss": 0.5989,
+      "step": 596370
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 6.597153133848626e-06,
+      "loss": 0.6615,
+      "step": 596380
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 6.596005203948878e-06,
+      "loss": 0.6966,
+      "step": 596390
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 6.594857274049131e-06,
+      "loss": 0.7171,
+      "step": 596400
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 6.593709344149384e-06,
+      "loss": 0.6856,
+      "step": 596410
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 6.592561414249636e-06,
+      "loss": 0.7639,
+      "step": 596420
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 6.591413484349889e-06,
+      "loss": 0.7174,
+      "step": 596430
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 6.590265554450141e-06,
+      "loss": 0.631,
+      "step": 596440
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 6.589117624550393e-06,
+      "loss": 0.7506,
+      "step": 596450
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 6.587969694650646e-06,
+      "loss": 0.6732,
+      "step": 596460
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 6.586821764750899e-06,
+      "loss": 0.6651,
+      "step": 596470
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 6.585673834851151e-06,
+      "loss": 0.7478,
+      "step": 596480
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 6.584525904951404e-06,
+      "loss": 0.7178,
+      "step": 596490
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 6.583377975051657e-06,
+      "loss": 0.6833,
+      "step": 596500
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 6.5822300451519085e-06,
+      "loss": 0.6666,
+      "step": 596510
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 6.581082115252161e-06,
+      "loss": 0.709,
+      "step": 596520
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 6.579934185352414e-06,
+      "loss": 0.7661,
+      "step": 596530
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 6.5787862554526665e-06,
+      "loss": 0.7018,
+      "step": 596540
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 6.577638325552919e-06,
+      "loss": 0.761,
+      "step": 596550
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 6.576490395653172e-06,
+      "loss": 0.6655,
+      "step": 596560
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 6.5753424657534245e-06,
+      "loss": 0.6832,
+      "step": 596570
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 6.574194535853676e-06,
+      "loss": 0.7224,
+      "step": 596580
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 6.573046605953929e-06,
+      "loss": 0.7452,
+      "step": 596590
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 6.571898676054182e-06,
+      "loss": 0.5983,
+      "step": 596600
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 6.570750746154434e-06,
+      "loss": 0.7267,
+      "step": 596610
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 6.569602816254687e-06,
+      "loss": 0.7138,
+      "step": 596620
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 6.56845488635494e-06,
+      "loss": 0.6432,
+      "step": 596630
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 6.567306956455192e-06,
+      "loss": 0.6462,
+      "step": 596640
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 6.566159026555444e-06,
+      "loss": 0.698,
+      "step": 596650
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 6.565011096655697e-06,
+      "loss": 0.687,
+      "step": 596660
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 6.563863166755949e-06,
+      "loss": 0.6756,
+      "step": 596670
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 6.562715236856202e-06,
+      "loss": 0.6938,
+      "step": 596680
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 6.561567306956455e-06,
+      "loss": 0.6115,
+      "step": 596690
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 6.560419377056707e-06,
+      "loss": 0.7112,
+      "step": 596700
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 6.55927144715696e-06,
+      "loss": 0.667,
+      "step": 596710
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 6.558123517257213e-06,
+      "loss": 0.6828,
+      "step": 596720
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 6.5569755873574646e-06,
+      "loss": 0.7309,
+      "step": 596730
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 6.555827657457717e-06,
+      "loss": 0.7072,
+      "step": 596740
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 6.55467972755797e-06,
+      "loss": 0.6401,
+      "step": 596750
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 6.553531797658223e-06,
+      "loss": 0.7081,
+      "step": 596760
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 6.552383867758475e-06,
+      "loss": 0.6803,
+      "step": 596770
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 6.551235937858728e-06,
+      "loss": 0.7784,
+      "step": 596780
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 6.550088007958981e-06,
+      "loss": 0.6553,
+      "step": 596790
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 6.548940078059232e-06,
+      "loss": 0.6982,
+      "step": 596800
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 6.5477921481594846e-06,
+      "loss": 0.6357,
+      "step": 596810
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 6.546644218259738e-06,
+      "loss": 0.643,
+      "step": 596820
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 6.54549628835999e-06,
+      "loss": 0.6663,
+      "step": 596830
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 6.544348358460243e-06,
+      "loss": 0.7875,
+      "step": 596840
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 6.543200428560496e-06,
+      "loss": 0.7636,
+      "step": 596850
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 6.542052498660748e-06,
+      "loss": 0.652,
+      "step": 596860
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 6.540904568761e-06,
+      "loss": 0.6862,
+      "step": 596870
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 6.539756638861253e-06,
+      "loss": 0.7144,
+      "step": 596880
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 6.538608708961505e-06,
+      "loss": 0.6855,
+      "step": 596890
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 6.537460779061758e-06,
+      "loss": 0.7255,
+      "step": 596900
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 6.536312849162011e-06,
+      "loss": 0.7042,
+      "step": 596910
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 6.5351649192622635e-06,
+      "loss": 0.6545,
+      "step": 596920
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 6.534016989362516e-06,
+      "loss": 0.6464,
+      "step": 596930
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 6.532869059462769e-06,
+      "loss": 0.7819,
+      "step": 596940
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 6.531721129563021e-06,
+      "loss": 0.7028,
+      "step": 596950
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 6.530573199663273e-06,
+      "loss": 0.709,
+      "step": 596960
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 6.529425269763526e-06,
+      "loss": 0.7028,
+      "step": 596970
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 6.528277339863779e-06,
+      "loss": 0.7055,
+      "step": 596980
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 6.527129409964031e-06,
+      "loss": 0.6932,
+      "step": 596990
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 6.525981480064283e-06,
+      "loss": 0.6631,
+      "step": 597000
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 6.524833550164537e-06,
+      "loss": 0.7778,
+      "step": 597010
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 6.523685620264788e-06,
+      "loss": 0.7609,
+      "step": 597020
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 6.522537690365041e-06,
+      "loss": 0.7986,
+      "step": 597030
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 6.521389760465294e-06,
+      "loss": 0.6631,
+      "step": 597040
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 6.520241830565546e-06,
+      "loss": 0.6876,
+      "step": 597050
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 6.519093900665799e-06,
+      "loss": 0.672,
+      "step": 597060
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 6.517945970766052e-06,
+      "loss": 0.7129,
+      "step": 597070
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 6.516798040866304e-06,
+      "loss": 0.7771,
+      "step": 597080
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 6.515650110966556e-06,
+      "loss": 0.7675,
+      "step": 597090
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 6.514502181066809e-06,
+      "loss": 0.6487,
+      "step": 597100
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 6.5133542511670615e-06,
+      "loss": 0.7342,
+      "step": 597110
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 6.512206321267314e-06,
+      "loss": 0.7519,
+      "step": 597120
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 6.511058391367567e-06,
+      "loss": 0.7406,
+      "step": 597130
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 6.5099104614678195e-06,
+      "loss": 0.673,
+      "step": 597140
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 6.508762531568071e-06,
+      "loss": 0.7289,
+      "step": 597150
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 6.5076146016683235e-06,
+      "loss": 0.7121,
+      "step": 597160
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 6.506466671768577e-06,
+      "loss": 0.7156,
+      "step": 597170
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 6.505318741868829e-06,
+      "loss": 0.7616,
+      "step": 597180
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 6.5041708119690815e-06,
+      "loss": 0.7267,
+      "step": 597190
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 6.503022882069335e-06,
+      "loss": 0.7399,
+      "step": 597200
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 6.501874952169587e-06,
+      "loss": 0.6703,
+      "step": 597210
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 6.500727022269839e-06,
+      "loss": 0.7061,
+      "step": 597220
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 6.499579092370093e-06,
+      "loss": 0.8017,
+      "step": 597230
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 6.498431162470344e-06,
+      "loss": 0.726,
+      "step": 597240
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 6.497283232570597e-06,
+      "loss": 0.6786,
+      "step": 597250
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 6.49613530267085e-06,
+      "loss": 0.7641,
+      "step": 597260
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 6.494987372771102e-06,
+      "loss": 0.7383,
+      "step": 597270
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 6.493839442871355e-06,
+      "loss": 0.7162,
+      "step": 597280
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 6.492691512971608e-06,
+      "loss": 0.7106,
+      "step": 597290
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 6.4915435830718596e-06,
+      "loss": 0.6645,
+      "step": 597300
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 6.490395653172112e-06,
+      "loss": 0.7203,
+      "step": 597310
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 6.489247723272364e-06,
+      "loss": 0.7044,
+      "step": 597320
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 6.488099793372618e-06,
+      "loss": 0.5958,
+      "step": 597330
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 6.48695186347287e-06,
+      "loss": 0.7266,
+      "step": 597340
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 6.485803933573122e-06,
+      "loss": 0.6385,
+      "step": 597350
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 6.484656003673376e-06,
+      "loss": 0.7406,
+      "step": 597360
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 6.483508073773627e-06,
+      "loss": 0.642,
+      "step": 597370
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 6.4823601438738796e-06,
+      "loss": 0.6928,
+      "step": 597380
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 6.481212213974133e-06,
+      "loss": 0.7226,
+      "step": 597390
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 6.480064284074385e-06,
+      "loss": 0.6732,
+      "step": 597400
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 6.478916354174638e-06,
+      "loss": 0.7392,
+      "step": 597410
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 6.477768424274891e-06,
+      "loss": 0.6918,
+      "step": 597420
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 6.476620494375143e-06,
+      "loss": 0.6706,
+      "step": 597430
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 6.475472564475395e-06,
+      "loss": 0.7051,
+      "step": 597440
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 6.474324634575648e-06,
+      "loss": 0.6896,
+      "step": 597450
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 6.4731767046759004e-06,
+      "loss": 0.7329,
+      "step": 597460
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 6.472028774776153e-06,
+      "loss": 0.6665,
+      "step": 597470
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 6.470880844876406e-06,
+      "loss": 0.7387,
+      "step": 597480
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 6.4697329149766585e-06,
+      "loss": 0.6682,
+      "step": 597490
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 6.468584985076911e-06,
+      "loss": 0.6817,
+      "step": 597500
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 6.467437055177162e-06,
+      "loss": 0.7216,
+      "step": 597510
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 6.466289125277416e-06,
+      "loss": 0.6433,
+      "step": 597520
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 6.465141195377668e-06,
+      "loss": 0.7131,
+      "step": 597530
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 6.4639932654779204e-06,
+      "loss": 0.7478,
+      "step": 597540
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 6.462845335578174e-06,
+      "loss": 0.7399,
+      "step": 597550
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 6.461697405678426e-06,
+      "loss": 0.7296,
+      "step": 597560
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 6.4605494757786785e-06,
+      "loss": 0.7162,
+      "step": 597570
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 6.459401545878932e-06,
+      "loss": 0.6574,
+      "step": 597580
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 6.458253615979183e-06,
+      "loss": 0.7601,
+      "step": 597590
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 6.457105686079436e-06,
+      "loss": 0.7624,
+      "step": 597600
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 6.455957756179689e-06,
+      "loss": 0.6975,
+      "step": 597610
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 6.454809826279941e-06,
+      "loss": 0.6873,
+      "step": 597620
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 6.453661896380194e-06,
+      "loss": 0.7222,
+      "step": 597630
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 6.452513966480447e-06,
+      "loss": 0.7135,
+      "step": 597640
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 6.451366036580699e-06,
+      "loss": 0.7687,
+      "step": 597650
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 6.450218106680951e-06,
+      "loss": 0.6929,
+      "step": 597660
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 6.449070176781203e-06,
+      "loss": 0.7088,
+      "step": 597670
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 6.4479222468814565e-06,
+      "loss": 0.6267,
+      "step": 597680
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 6.446774316981709e-06,
+      "loss": 0.6857,
+      "step": 597690
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 6.445626387081961e-06,
+      "loss": 0.6696,
+      "step": 597700
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 6.4444784571822145e-06,
+      "loss": 0.7167,
+      "step": 597710
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 6.443330527282467e-06,
+      "loss": 0.7056,
+      "step": 597720
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 6.4421825973827185e-06,
+      "loss": 0.6417,
+      "step": 597730
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 6.441034667482972e-06,
+      "loss": 0.6927,
+      "step": 597740
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 6.439886737583224e-06,
+      "loss": 0.7911,
+      "step": 597750
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 6.4387388076834765e-06,
+      "loss": 0.7212,
+      "step": 597760
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 6.43759087778373e-06,
+      "loss": 0.6768,
+      "step": 597770
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 6.436442947883982e-06,
+      "loss": 0.7185,
+      "step": 597780
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 6.4352950179842346e-06,
+      "loss": 0.645,
+      "step": 597790
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 6.434147088084488e-06,
+      "loss": 0.7856,
+      "step": 597800
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 6.432999158184739e-06,
+      "loss": 0.6435,
+      "step": 597810
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 6.431851228284992e-06,
+      "loss": 0.7296,
+      "step": 597820
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 6.430703298385245e-06,
+      "loss": 0.6926,
+      "step": 597830
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 6.429555368485497e-06,
+      "loss": 0.6824,
+      "step": 597840
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 6.42840743858575e-06,
+      "loss": 0.6464,
+      "step": 597850
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 6.427259508686002e-06,
+      "loss": 0.7656,
+      "step": 597860
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 6.426111578786255e-06,
+      "loss": 0.7018,
+      "step": 597870
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 6.424963648886507e-06,
+      "loss": 0.7087,
+      "step": 597880
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 6.423815718986759e-06,
+      "loss": 0.7165,
+      "step": 597890
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 6.422667789087013e-06,
+      "loss": 0.7564,
+      "step": 597900
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 6.421519859187265e-06,
+      "loss": 0.6484,
+      "step": 597910
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 6.420371929287517e-06,
+      "loss": 0.6316,
+      "step": 597920
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 6.419223999387771e-06,
+      "loss": 0.7122,
+      "step": 597930
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 6.418076069488023e-06,
+      "loss": 0.6857,
+      "step": 597940
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 6.416928139588275e-06,
+      "loss": 0.5996,
+      "step": 597950
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 6.415780209688528e-06,
+      "loss": 0.6737,
+      "step": 597960
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 6.41463227978878e-06,
+      "loss": 0.667,
+      "step": 597970
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 6.413484349889033e-06,
+      "loss": 0.665,
+      "step": 597980
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 6.412336419989286e-06,
+      "loss": 0.6583,
+      "step": 597990
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 6.411188490089538e-06,
+      "loss": 0.7068,
+      "step": 598000
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 6.410040560189791e-06,
+      "loss": 0.6745,
+      "step": 598010
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 6.408892630290042e-06,
+      "loss": 0.7013,
+      "step": 598020
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 6.4077447003902954e-06,
+      "loss": 0.7244,
+      "step": 598030
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 6.406596770490548e-06,
+      "loss": 0.7282,
+      "step": 598040
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 6.4054488405908e-06,
+      "loss": 0.7279,
+      "step": 598050
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 6.4043009106910535e-06,
+      "loss": 0.7309,
+      "step": 598060
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 6.403152980791306e-06,
+      "loss": 0.6722,
+      "step": 598070
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 6.402005050891558e-06,
+      "loss": 0.6786,
+      "step": 598080
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 6.4008571209918115e-06,
+      "loss": 0.6784,
+      "step": 598090
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 6.399709191092063e-06,
+      "loss": 0.6942,
+      "step": 598100
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 6.3985612611923154e-06,
+      "loss": 0.7116,
+      "step": 598110
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 6.397413331292569e-06,
+      "loss": 0.6527,
+      "step": 598120
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 6.396265401392821e-06,
+      "loss": 0.6587,
+      "step": 598130
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 6.3951174714930735e-06,
+      "loss": 0.7081,
+      "step": 598140
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 6.393969541593327e-06,
+      "loss": 0.6904,
+      "step": 598150
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 6.392821611693579e-06,
+      "loss": 0.7434,
+      "step": 598160
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 6.391673681793831e-06,
+      "loss": 0.6842,
+      "step": 598170
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 6.390525751894083e-06,
+      "loss": 0.6853,
+      "step": 598180
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 6.389377821994336e-06,
+      "loss": 0.7286,
+      "step": 598190
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 6.388229892094589e-06,
+      "loss": 0.6334,
+      "step": 598200
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 6.387081962194841e-06,
+      "loss": 0.7583,
+      "step": 598210
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 6.385934032295094e-06,
+      "loss": 0.6957,
+      "step": 598220
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 6.384786102395347e-06,
+      "loss": 0.603,
+      "step": 598230
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 6.383638172495598e-06,
+      "loss": 0.71,
+      "step": 598240
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 6.3824902425958515e-06,
+      "loss": 0.6883,
+      "step": 598250
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 6.381342312696104e-06,
+      "loss": 0.6966,
+      "step": 598260
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 6.380194382796356e-06,
+      "loss": 0.6204,
+      "step": 598270
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 6.3790464528966096e-06,
+      "loss": 0.7117,
+      "step": 598280
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 6.377898522996862e-06,
+      "loss": 0.6097,
+      "step": 598290
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 6.376750593097114e-06,
+      "loss": 0.7144,
+      "step": 598300
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 6.375602663197368e-06,
+      "loss": 0.6112,
+      "step": 598310
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 6.374454733297619e-06,
+      "loss": 0.7293,
+      "step": 598320
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 6.3733068033978715e-06,
+      "loss": 0.682,
+      "step": 598330
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 6.372158873498125e-06,
+      "loss": 0.7086,
+      "step": 598340
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 6.371010943598377e-06,
+      "loss": 0.7651,
+      "step": 598350
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 6.3698630136986296e-06,
+      "loss": 0.7322,
+      "step": 598360
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 6.368715083798882e-06,
+      "loss": 0.6846,
+      "step": 598370
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 6.367567153899135e-06,
+      "loss": 0.6572,
+      "step": 598380
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 6.366419223999387e-06,
+      "loss": 0.7167,
+      "step": 598390
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 6.365271294099639e-06,
+      "loss": 0.6599,
+      "step": 598400
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 6.364123364199892e-06,
+      "loss": 0.6694,
+      "step": 598410
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 6.362975434300145e-06,
+      "loss": 0.7127,
+      "step": 598420
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 6.361827504400397e-06,
+      "loss": 0.7151,
+      "step": 598430
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 6.3606795745006504e-06,
+      "loss": 0.6751,
+      "step": 598440
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 6.359531644600903e-06,
+      "loss": 0.7093,
+      "step": 598450
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 6.358383714701154e-06,
+      "loss": 0.7071,
+      "step": 598460
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 6.357235784801408e-06,
+      "loss": 0.6452,
+      "step": 598470
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 6.35608785490166e-06,
+      "loss": 0.6623,
+      "step": 598480
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 6.354939925001912e-06,
+      "loss": 0.6781,
+      "step": 598490
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 6.353791995102166e-06,
+      "loss": 0.6731,
+      "step": 598500
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 6.352644065202418e-06,
+      "loss": 0.7782,
+      "step": 598510
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 6.3514961353026704e-06,
+      "loss": 0.6822,
+      "step": 598520
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 6.350348205402922e-06,
+      "loss": 0.6617,
+      "step": 598530
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 6.349200275503175e-06,
+      "loss": 0.7582,
+      "step": 598540
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 6.348052345603428e-06,
+      "loss": 0.6713,
+      "step": 598550
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 6.34690441570368e-06,
+      "loss": 0.6942,
+      "step": 598560
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 6.345756485803933e-06,
+      "loss": 0.7447,
+      "step": 598570
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 6.344608555904186e-06,
+      "loss": 0.6756,
+      "step": 598580
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 6.343460626004438e-06,
+      "loss": 0.6389,
+      "step": 598590
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 6.342312696104691e-06,
+      "loss": 0.7359,
+      "step": 598600
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 6.341164766204943e-06,
+      "loss": 0.645,
+      "step": 598610
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 6.340016836305195e-06,
+      "loss": 0.7006,
+      "step": 598620
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 6.3388689064054485e-06,
+      "loss": 0.63,
+      "step": 598630
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 6.337720976505701e-06,
+      "loss": 0.6653,
+      "step": 598640
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 6.336573046605953e-06,
+      "loss": 0.7419,
+      "step": 598650
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 6.3354251167062065e-06,
+      "loss": 0.6826,
+      "step": 598660
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 6.334277186806459e-06,
+      "loss": 0.6714,
+      "step": 598670
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 6.3331292569067105e-06,
+      "loss": 0.6569,
+      "step": 598680
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 6.331981327006963e-06,
+      "loss": 0.6356,
+      "step": 598690
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 6.330833397107216e-06,
+      "loss": 0.7954,
+      "step": 598700
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 6.3296854672074685e-06,
+      "loss": 0.7093,
+      "step": 598710
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 6.328537537307721e-06,
+      "loss": 0.7212,
+      "step": 598720
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 6.327389607407974e-06,
+      "loss": 0.7172,
+      "step": 598730
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 6.3262416775082265e-06,
+      "loss": 0.7247,
+      "step": 598740
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 6.325093747608478e-06,
+      "loss": 0.7349,
+      "step": 598750
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 6.323945817708731e-06,
+      "loss": 0.7733,
+      "step": 598760
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 6.322797887808984e-06,
+      "loss": 0.6611,
+      "step": 598770
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 6.321649957909236e-06,
+      "loss": 0.7162,
+      "step": 598780
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 6.320502028009489e-06,
+      "loss": 0.752,
+      "step": 598790
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 6.319354098109742e-06,
+      "loss": 0.7041,
+      "step": 598800
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 6.318206168209994e-06,
+      "loss": 0.6963,
+      "step": 598810
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 6.317058238310247e-06,
+      "loss": 0.7169,
+      "step": 598820
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 6.315910308410499e-06,
+      "loss": 0.6797,
+      "step": 598830
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 6.314762378510751e-06,
+      "loss": 0.6853,
+      "step": 598840
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 6.3136144486110046e-06,
+      "loss": 0.7609,
+      "step": 598850
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 6.312466518711257e-06,
+      "loss": 0.6736,
+      "step": 598860
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 6.311318588811509e-06,
+      "loss": 0.7734,
+      "step": 598870
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 6.310170658911762e-06,
+      "loss": 0.7304,
+      "step": 598880
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 6.309022729012015e-06,
+      "loss": 0.7514,
+      "step": 598890
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 6.3078747991122665e-06,
+      "loss": 0.7661,
+      "step": 598900
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 6.306726869212519e-06,
+      "loss": 0.6981,
+      "step": 598910
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 6.305578939312772e-06,
+      "loss": 0.6177,
+      "step": 598920
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 6.3044310094130246e-06,
+      "loss": 0.6463,
+      "step": 598930
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 6.303283079513277e-06,
+      "loss": 0.677,
+      "step": 598940
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 6.30213514961353e-06,
+      "loss": 0.6188,
+      "step": 598950
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 6.300987219713783e-06,
+      "loss": 0.7349,
+      "step": 598960
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 6.299839289814034e-06,
+      "loss": 0.7152,
+      "step": 598970
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 6.298691359914287e-06,
+      "loss": 0.633,
+      "step": 598980
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 6.29754343001454e-06,
+      "loss": 0.6931,
+      "step": 598990
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 6.296395500114792e-06,
+      "loss": 0.7036,
+      "step": 599000
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 6.2952475702150454e-06,
+      "loss": 0.7338,
+      "step": 599010
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 6.294099640315298e-06,
+      "loss": 0.7202,
+      "step": 599020
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 6.29295171041555e-06,
+      "loss": 0.7564,
+      "step": 599030
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 6.291803780515802e-06,
+      "loss": 0.6234,
+      "step": 599040
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 6.290655850616055e-06,
+      "loss": 0.8031,
+      "step": 599050
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 6.289507920716307e-06,
+      "loss": 0.6832,
+      "step": 599060
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 6.28835999081656e-06,
+      "loss": 0.712,
+      "step": 599070
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 6.287212060916813e-06,
+      "loss": 0.6811,
+      "step": 599080
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 6.2860641310170654e-06,
+      "loss": 0.6939,
+      "step": 599090
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 6.284916201117318e-06,
+      "loss": 0.6904,
+      "step": 599100
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 6.283768271217571e-06,
+      "loss": 0.6802,
+      "step": 599110
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 6.282620341317823e-06,
+      "loss": 0.7187,
+      "step": 599120
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 6.281472411418075e-06,
+      "loss": 0.6647,
+      "step": 599130
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 6.280324481518328e-06,
+      "loss": 0.6667,
+      "step": 599140
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 6.279176551618581e-06,
+      "loss": 0.7309,
+      "step": 599150
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 6.278028621718833e-06,
+      "loss": 0.6858,
+      "step": 599160
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 6.276880691819086e-06,
+      "loss": 0.6305,
+      "step": 599170
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 6.275732761919339e-06,
+      "loss": 0.6981,
+      "step": 599180
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 6.27458483201959e-06,
+      "loss": 0.6983,
+      "step": 599190
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 6.2734369021198435e-06,
+      "loss": 0.6774,
+      "step": 599200
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 6.272288972220096e-06,
+      "loss": 0.7394,
+      "step": 599210
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 6.271141042320348e-06,
+      "loss": 0.7033,
+      "step": 599220
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 6.269993112420601e-06,
+      "loss": 0.6482,
+      "step": 599230
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 6.268845182520854e-06,
+      "loss": 0.7006,
+      "step": 599240
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 6.267697252621106e-06,
+      "loss": 0.7305,
+      "step": 599250
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 6.266549322721358e-06,
+      "loss": 0.6009,
+      "step": 599260
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 6.265401392821611e-06,
+      "loss": 0.6341,
+      "step": 599270
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 6.2642534629218635e-06,
+      "loss": 0.7647,
+      "step": 599280
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 6.263105533022116e-06,
+      "loss": 0.636,
+      "step": 599290
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 6.261957603122369e-06,
+      "loss": 0.7199,
+      "step": 599300
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 6.2608096732226215e-06,
+      "loss": 0.587,
+      "step": 599310
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 6.259661743322874e-06,
+      "loss": 0.7843,
+      "step": 599320
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 6.258513813423127e-06,
+      "loss": 0.7713,
+      "step": 599330
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 6.257365883523379e-06,
+      "loss": 0.6946,
+      "step": 599340
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 6.256217953623631e-06,
+      "loss": 0.7253,
+      "step": 599350
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 6.255070023723884e-06,
+      "loss": 0.7777,
+      "step": 599360
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 6.253922093824137e-06,
+      "loss": 0.7335,
+      "step": 599370
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 6.252774163924389e-06,
+      "loss": 0.6411,
+      "step": 599380
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 6.2516262340246415e-06,
+      "loss": 0.7497,
+      "step": 599390
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 6.250478304124895e-06,
+      "loss": 0.6709,
+      "step": 599400
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 6.249330374225146e-06,
+      "loss": 0.6566,
+      "step": 599410
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 6.248182444325399e-06,
+      "loss": 0.6792,
+      "step": 599420
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 6.247034514425652e-06,
+      "loss": 0.6597,
+      "step": 599430
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 6.245886584525904e-06,
+      "loss": 0.683,
+      "step": 599440
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 6.244738654626157e-06,
+      "loss": 0.7035,
+      "step": 599450
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 6.24359072472641e-06,
+      "loss": 0.6419,
+      "step": 599460
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 6.242442794826662e-06,
+      "loss": 0.7265,
+      "step": 599470
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 6.241294864926914e-06,
+      "loss": 0.6293,
+      "step": 599480
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 6.240146935027167e-06,
+      "loss": 0.7426,
+      "step": 599490
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 6.23899900512742e-06,
+      "loss": 0.733,
+      "step": 599500
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 6.237851075227672e-06,
+      "loss": 0.7485,
+      "step": 599510
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 6.236703145327925e-06,
+      "loss": 0.5899,
+      "step": 599520
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 6.235555215428178e-06,
+      "loss": 0.6812,
+      "step": 599530
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 6.23440728552843e-06,
+      "loss": 0.7361,
+      "step": 599540
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 6.2332593556286816e-06,
+      "loss": 0.6819,
+      "step": 599550
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 6.232111425728935e-06,
+      "loss": 0.6375,
+      "step": 599560
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 6.230963495829187e-06,
+      "loss": 0.6797,
+      "step": 599570
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 6.22981556592944e-06,
+      "loss": 0.7213,
+      "step": 599580
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 6.228667636029693e-06,
+      "loss": 0.6096,
+      "step": 599590
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 6.227519706129945e-06,
+      "loss": 0.6582,
+      "step": 599600
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 6.226371776230198e-06,
+      "loss": 0.6471,
+      "step": 599610
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 6.225223846330451e-06,
+      "loss": 0.7065,
+      "step": 599620
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 6.224075916430702e-06,
+      "loss": 0.648,
+      "step": 599630
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 6.222927986530955e-06,
+      "loss": 0.7239,
+      "step": 599640
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 6.221780056631208e-06,
+      "loss": 0.6943,
+      "step": 599650
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 6.2206321267314605e-06,
+      "loss": 0.6394,
+      "step": 599660
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 6.219484196831713e-06,
+      "loss": 0.6601,
+      "step": 599670
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 6.218336266931966e-06,
+      "loss": 0.746,
+      "step": 599680
+    },
+    {
+      "epoch": 27.51,
+      "learning_rate": 6.2171883370322185e-06,
+      "loss": 0.708,
+      "step": 599690
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 6.21604040713247e-06,
+      "loss": 0.6983,
+      "step": 599700
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 6.214892477232723e-06,
+      "loss": 0.6733,
+      "step": 599710
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 6.213744547332976e-06,
+      "loss": 0.7913,
+      "step": 599720
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 6.212596617433228e-06,
+      "loss": 0.7367,
+      "step": 599730
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 6.2114486875334805e-06,
+      "loss": 0.7032,
+      "step": 599740
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 6.210300757633734e-06,
+      "loss": 0.669,
+      "step": 599750
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 6.209152827733986e-06,
+      "loss": 0.6436,
+      "step": 599760
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 6.208004897834238e-06,
+      "loss": 0.7532,
+      "step": 599770
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 6.206856967934491e-06,
+      "loss": 0.7369,
+      "step": 599780
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 6.205709038034743e-06,
+      "loss": 0.7661,
+      "step": 599790
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 6.204561108134996e-06,
+      "loss": 0.6594,
+      "step": 599800
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 6.203413178235249e-06,
+      "loss": 0.6951,
+      "step": 599810
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 6.202265248335501e-06,
+      "loss": 0.709,
+      "step": 599820
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 6.201117318435754e-06,
+      "loss": 0.6324,
+      "step": 599830
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 6.199969388536007e-06,
+      "loss": 0.7263,
+      "step": 599840
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 6.1988214586362585e-06,
+      "loss": 0.7295,
+      "step": 599850
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 6.197673528736511e-06,
+      "loss": 0.6235,
+      "step": 599860
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 6.196525598836764e-06,
+      "loss": 0.6925,
+      "step": 599870
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 6.1953776689370165e-06,
+      "loss": 0.6641,
+      "step": 599880
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 6.194229739037269e-06,
+      "loss": 0.7307,
+      "step": 599890
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 6.193081809137521e-06,
+      "loss": 0.6522,
+      "step": 599900
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 6.1919338792377746e-06,
+      "loss": 0.6775,
+      "step": 599910
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 6.190785949338026e-06,
+      "loss": 0.6231,
+      "step": 599920
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 6.1896380194382785e-06,
+      "loss": 0.6662,
+      "step": 599930
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 6.188490089538532e-06,
+      "loss": 0.6847,
+      "step": 599940
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 6.187342159638784e-06,
+      "loss": 0.6749,
+      "step": 599950
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 6.1861942297390365e-06,
+      "loss": 0.7462,
+      "step": 599960
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 6.18504629983929e-06,
+      "loss": 0.6289,
+      "step": 599970
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 6.183898369939542e-06,
+      "loss": 0.678,
+      "step": 599980
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 6.182750440039794e-06,
+      "loss": 0.6759,
+      "step": 599990
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 6.181602510140047e-06,
+      "loss": 0.7205,
+      "step": 600000
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 6.180454580240299e-06,
+      "loss": 0.785,
+      "step": 600010
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 6.179306650340552e-06,
+      "loss": 0.6814,
+      "step": 600020
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 6.178158720440805e-06,
+      "loss": 0.7307,
+      "step": 600030
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 6.177010790541057e-06,
+      "loss": 0.6218,
+      "step": 600040
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 6.17586286064131e-06,
+      "loss": 0.7601,
+      "step": 600050
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 6.174714930741563e-06,
+      "loss": 0.6729,
+      "step": 600060
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 6.173567000841815e-06,
+      "loss": 0.7715,
+      "step": 600070
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 6.172419070942067e-06,
+      "loss": 0.7273,
+      "step": 600080
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 6.171271141042319e-06,
+      "loss": 0.6816,
+      "step": 600090
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 6.170123211142573e-06,
+      "loss": 0.7319,
+      "step": 600100
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 6.168975281242825e-06,
+      "loss": 0.6397,
+      "step": 600110
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 6.167827351343077e-06,
+      "loss": 0.7553,
+      "step": 600120
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 6.166679421443331e-06,
+      "loss": 0.779,
+      "step": 600130
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 6.165531491543582e-06,
+      "loss": 0.7231,
+      "step": 600140
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 6.164383561643835e-06,
+      "loss": 0.678,
+      "step": 600150
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 6.163235631744088e-06,
+      "loss": 0.6955,
+      "step": 600160
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 6.16208770184434e-06,
+      "loss": 0.745,
+      "step": 600170
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 6.160939771944593e-06,
+      "loss": 0.6609,
+      "step": 600180
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 6.159791842044846e-06,
+      "loss": 0.6569,
+      "step": 600190
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 6.158643912145098e-06,
+      "loss": 0.7427,
+      "step": 600200
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 6.15749598224535e-06,
+      "loss": 0.7273,
+      "step": 600210
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 6.156348052345603e-06,
+      "loss": 0.755,
+      "step": 600220
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 6.1552001224458555e-06,
+      "loss": 0.7487,
+      "step": 600230
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 6.154052192546108e-06,
+      "loss": 0.6277,
+      "step": 600240
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 6.15290426264636e-06,
+      "loss": 0.6789,
+      "step": 600250
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 6.1517563327466135e-06,
+      "loss": 0.6152,
+      "step": 600260
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 6.150608402846866e-06,
+      "loss": 0.7174,
+      "step": 600270
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 6.1494604729471174e-06,
+      "loss": 0.7702,
+      "step": 600280
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 6.148312543047371e-06,
+      "loss": 0.6971,
+      "step": 600290
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 6.147164613147623e-06,
+      "loss": 0.6719,
+      "step": 600300
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 6.1460166832478755e-06,
+      "loss": 0.7312,
+      "step": 600310
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 6.144868753348129e-06,
+      "loss": 0.6713,
+      "step": 600320
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 6.143720823448381e-06,
+      "loss": 0.7674,
+      "step": 600330
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 6.1425728935486335e-06,
+      "loss": 0.6827,
+      "step": 600340
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 6.141424963648887e-06,
+      "loss": 0.6639,
+      "step": 600350
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 6.140277033749138e-06,
+      "loss": 0.6708,
+      "step": 600360
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 6.139129103849391e-06,
+      "loss": 0.7144,
+      "step": 600370
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 6.137981173949644e-06,
+      "loss": 0.7501,
+      "step": 600380
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 6.136833244049896e-06,
+      "loss": 0.7081,
+      "step": 600390
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 6.135685314150149e-06,
+      "loss": 0.7025,
+      "step": 600400
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 6.134537384250401e-06,
+      "loss": 0.6884,
+      "step": 600410
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 6.133389454350654e-06,
+      "loss": 0.7719,
+      "step": 600420
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 6.132241524450906e-06,
+      "loss": 0.7163,
+      "step": 600430
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 6.131093594551158e-06,
+      "loss": 0.662,
+      "step": 600440
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 6.1299456646514115e-06,
+      "loss": 0.7278,
+      "step": 600450
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 6.128797734751664e-06,
+      "loss": 0.6448,
+      "step": 600460
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 6.127649804851916e-06,
+      "loss": 0.7109,
+      "step": 600470
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 6.12650187495217e-06,
+      "loss": 0.6365,
+      "step": 600480
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 6.125353945052422e-06,
+      "loss": 0.6763,
+      "step": 600490
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 6.1242060151526735e-06,
+      "loss": 0.6792,
+      "step": 600500
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 6.123058085252927e-06,
+      "loss": 0.6774,
+      "step": 600510
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 6.121910155353179e-06,
+      "loss": 0.7563,
+      "step": 600520
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 6.1207622254534316e-06,
+      "loss": 0.7883,
+      "step": 600530
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 6.119614295553685e-06,
+      "loss": 0.6534,
+      "step": 600540
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 6.118466365653937e-06,
+      "loss": 0.6635,
+      "step": 600550
+    },
+    {
+      "epoch": 27.55,
+      "learning_rate": 6.11731843575419e-06,
+      "loss": 0.6572,
+      "step": 600560
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 6.116170505854443e-06,
+      "loss": 0.7247,
+      "step": 600570
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 6.115022575954694e-06,
+      "loss": 0.7867,
+      "step": 600580
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 6.113874646054947e-06,
+      "loss": 0.7192,
+      "step": 600590
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 6.112726716155199e-06,
+      "loss": 0.6711,
+      "step": 600600
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 6.111578786255452e-06,
+      "loss": 0.6964,
+      "step": 600610
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 6.110430856355705e-06,
+      "loss": 0.6451,
+      "step": 600620
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 6.109282926455957e-06,
+      "loss": 0.6855,
+      "step": 600630
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 6.1081349965562104e-06,
+      "loss": 0.6941,
+      "step": 600640
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 6.106987066656462e-06,
+      "loss": 0.7291,
+      "step": 600650
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 6.105839136756714e-06,
+      "loss": 0.6401,
+      "step": 600660
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 6.104691206856968e-06,
+      "loss": 0.6256,
+      "step": 600670
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 6.10354327695722e-06,
+      "loss": 0.7414,
+      "step": 600680
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 6.102395347057472e-06,
+      "loss": 0.6898,
+      "step": 600690
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 6.101247417157726e-06,
+      "loss": 0.713,
+      "step": 600700
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 6.100099487257978e-06,
+      "loss": 0.735,
+      "step": 600710
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 6.09895155735823e-06,
+      "loss": 0.6993,
+      "step": 600720
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 6.097803627458483e-06,
+      "loss": 0.7202,
+      "step": 600730
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 6.096655697558735e-06,
+      "loss": 0.7226,
+      "step": 600740
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 6.095507767658988e-06,
+      "loss": 0.7615,
+      "step": 600750
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 6.09435983775924e-06,
+      "loss": 0.6848,
+      "step": 600760
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 6.093211907859493e-06,
+      "loss": 0.6941,
+      "step": 600770
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 6.092063977959746e-06,
+      "loss": 0.6508,
+      "step": 600780
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 6.090916048059997e-06,
+      "loss": 0.6942,
+      "step": 600790
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 6.0897681181602505e-06,
+      "loss": 0.687,
+      "step": 600800
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 6.088620188260503e-06,
+      "loss": 0.7277,
+      "step": 600810
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 6.087472258360755e-06,
+      "loss": 0.7466,
+      "step": 600820
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 6.0863243284610085e-06,
+      "loss": 0.6837,
+      "step": 600830
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 6.085176398561261e-06,
+      "loss": 0.6539,
+      "step": 600840
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 6.084028468661513e-06,
+      "loss": 0.6498,
+      "step": 600850
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 6.0828805387617665e-06,
+      "loss": 0.651,
+      "step": 600860
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 6.081732608862018e-06,
+      "loss": 0.6923,
+      "step": 600870
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 6.0805846789622705e-06,
+      "loss": 0.755,
+      "step": 600880
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 6.079436749062524e-06,
+      "loss": 0.7324,
+      "step": 600890
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 6.078288819162776e-06,
+      "loss": 0.7048,
+      "step": 600900
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 6.0771408892630285e-06,
+      "loss": 0.7263,
+      "step": 600910
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 6.075992959363281e-06,
+      "loss": 0.6887,
+      "step": 600920
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 6.074845029463534e-06,
+      "loss": 0.7215,
+      "step": 600930
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 6.073697099563786e-06,
+      "loss": 0.6928,
+      "step": 600940
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 6.072549169664038e-06,
+      "loss": 0.6954,
+      "step": 600950
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 6.071401239764291e-06,
+      "loss": 0.6494,
+      "step": 600960
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 6.070253309864544e-06,
+      "loss": 0.6773,
+      "step": 600970
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 6.069105379964796e-06,
+      "loss": 0.59,
+      "step": 600980
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 6.067957450065049e-06,
+      "loss": 0.745,
+      "step": 600990
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 6.066809520165302e-06,
+      "loss": 0.6964,
+      "step": 601000
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 6.065661590265553e-06,
+      "loss": 0.662,
+      "step": 601010
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 6.0645136603658066e-06,
+      "loss": 0.7519,
+      "step": 601020
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 6.063365730466059e-06,
+      "loss": 0.7042,
+      "step": 601030
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 6.062217800566311e-06,
+      "loss": 0.7158,
+      "step": 601040
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 6.061069870666565e-06,
+      "loss": 0.6831,
+      "step": 601050
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 6.059921940766817e-06,
+      "loss": 0.7069,
+      "step": 601060
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 6.058774010867069e-06,
+      "loss": 0.6917,
+      "step": 601070
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 6.057626080967323e-06,
+      "loss": 0.7606,
+      "step": 601080
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 6.056478151067574e-06,
+      "loss": 0.739,
+      "step": 601090
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 6.0553302211678266e-06,
+      "loss": 0.6772,
+      "step": 601100
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 6.054182291268079e-06,
+      "loss": 0.679,
+      "step": 601110
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 6.053034361368332e-06,
+      "loss": 0.7403,
+      "step": 601120
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 6.051886431468585e-06,
+      "loss": 0.7111,
+      "step": 601130
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 6.050738501568837e-06,
+      "loss": 0.6528,
+      "step": 601140
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 6.04959057166909e-06,
+      "loss": 0.69,
+      "step": 601150
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 6.048442641769342e-06,
+      "loss": 0.7659,
+      "step": 601160
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 6.047294711869594e-06,
+      "loss": 0.7684,
+      "step": 601170
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 6.046146781969847e-06,
+      "loss": 0.7034,
+      "step": 601180
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 6.0449988520701e-06,
+      "loss": 0.6763,
+      "step": 601190
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 6.043850922170352e-06,
+      "loss": 0.6935,
+      "step": 601200
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 6.0427029922706055e-06,
+      "loss": 0.6689,
+      "step": 601210
+    },
+    {
+      "epoch": 27.58,
+      "learning_rate": 6.041555062370858e-06,
+      "loss": 0.7221,
+      "step": 601220
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 6.040407132471109e-06,
+      "loss": 0.6822,
+      "step": 601230
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 6.039259202571363e-06,
+      "loss": 0.7507,
+      "step": 601240
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 6.038111272671615e-06,
+      "loss": 0.661,
+      "step": 601250
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 6.0369633427718674e-06,
+      "loss": 0.6772,
+      "step": 601260
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 6.03581541287212e-06,
+      "loss": 0.6913,
+      "step": 601270
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 6.034667482972373e-06,
+      "loss": 0.7193,
+      "step": 601280
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 6.0335195530726255e-06,
+      "loss": 0.6143,
+      "step": 601290
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 6.032371623172877e-06,
+      "loss": 0.6768,
+      "step": 601300
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 6.03122369327313e-06,
+      "loss": 0.6512,
+      "step": 601310
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 6.030075763373383e-06,
+      "loss": 0.6255,
+      "step": 601320
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 6.028927833473635e-06,
+      "loss": 0.7949,
+      "step": 601330
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 6.027779903573888e-06,
+      "loss": 0.7525,
+      "step": 601340
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 6.026631973674141e-06,
+      "loss": 0.7491,
+      "step": 601350
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 6.025484043774393e-06,
+      "loss": 0.728,
+      "step": 601360
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 6.024336113874646e-06,
+      "loss": 0.7532,
+      "step": 601370
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 6.023188183974898e-06,
+      "loss": 0.6928,
+      "step": 601380
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 6.02204025407515e-06,
+      "loss": 0.7286,
+      "step": 601390
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 6.0208923241754035e-06,
+      "loss": 0.7356,
+      "step": 601400
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 6.019744394275656e-06,
+      "loss": 0.7066,
+      "step": 601410
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 6.018596464375908e-06,
+      "loss": 0.7885,
+      "step": 601420
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 6.0174485344761615e-06,
+      "loss": 0.6649,
+      "step": 601430
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 6.016300604576414e-06,
+      "loss": 0.6388,
+      "step": 601440
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 6.0151526746766655e-06,
+      "loss": 0.7031,
+      "step": 601450
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 6.014004744776918e-06,
+      "loss": 0.7237,
+      "step": 601460
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 6.012856814877171e-06,
+      "loss": 0.7195,
+      "step": 601470
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 6.0117088849774235e-06,
+      "loss": 0.6815,
+      "step": 601480
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 6.010560955077676e-06,
+      "loss": 0.7164,
+      "step": 601490
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 6.009413025177929e-06,
+      "loss": 0.6885,
+      "step": 601500
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 6.0082650952781815e-06,
+      "loss": 0.5924,
+      "step": 601510
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 6.007117165378433e-06,
+      "loss": 0.6532,
+      "step": 601520
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 6.005969235478686e-06,
+      "loss": 0.6553,
+      "step": 601530
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 6.004821305578939e-06,
+      "loss": 0.7645,
+      "step": 601540
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 6.003673375679191e-06,
+      "loss": 0.6448,
+      "step": 601550
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 6.002525445779444e-06,
+      "loss": 0.6824,
+      "step": 601560
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 6.001377515879697e-06,
+      "loss": 0.731,
+      "step": 601570
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 6.000229585979948e-06,
+      "loss": 0.678,
+      "step": 601580
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 5.999081656080202e-06,
+      "loss": 0.6481,
+      "step": 601590
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 5.997933726180454e-06,
+      "loss": 0.7537,
+      "step": 601600
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 5.996785796280706e-06,
+      "loss": 0.69,
+      "step": 601610
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 5.995637866380959e-06,
+      "loss": 0.7367,
+      "step": 601620
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 5.994489936481212e-06,
+      "loss": 0.6436,
+      "step": 601630
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 5.993342006581464e-06,
+      "loss": 0.6489,
+      "step": 601640
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 5.992194076681716e-06,
+      "loss": 0.6925,
+      "step": 601650
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 5.99104614678197e-06,
+      "loss": 0.6882,
+      "step": 601660
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 5.9898982168822216e-06,
+      "loss": 0.7098,
+      "step": 601670
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 5.988750286982474e-06,
+      "loss": 0.7562,
+      "step": 601680
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 5.987602357082727e-06,
+      "loss": 0.6824,
+      "step": 601690
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 5.98645442718298e-06,
+      "loss": 0.7328,
+      "step": 601700
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 5.985306497283232e-06,
+      "loss": 0.7086,
+      "step": 601710
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 5.984158567383485e-06,
+      "loss": 0.6853,
+      "step": 601720
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 5.983010637483737e-06,
+      "loss": 0.658,
+      "step": 601730
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 5.981862707583989e-06,
+      "loss": 0.6063,
+      "step": 601740
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 5.9807147776842424e-06,
+      "loss": 0.7047,
+      "step": 601750
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 5.979566847784495e-06,
+      "loss": 0.6953,
+      "step": 601760
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 5.978418917884747e-06,
+      "loss": 0.7156,
+      "step": 601770
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 5.977270987985e-06,
+      "loss": 0.7539,
+      "step": 601780
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 5.976123058085253e-06,
+      "loss": 0.6661,
+      "step": 601790
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 5.974975128185504e-06,
+      "loss": 0.6979,
+      "step": 601800
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 5.973827198285757e-06,
+      "loss": 0.6636,
+      "step": 601810
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 5.97267926838601e-06,
+      "loss": 0.7261,
+      "step": 601820
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 5.9715313384862624e-06,
+      "loss": 0.7155,
+      "step": 601830
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 5.970383408586515e-06,
+      "loss": 0.7849,
+      "step": 601840
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 5.969235478686768e-06,
+      "loss": 0.6439,
+      "step": 601850
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 5.9680875487870205e-06,
+      "loss": 0.7353,
+      "step": 601860
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 5.966939618887272e-06,
+      "loss": 0.7774,
+      "step": 601870
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 5.965791688987526e-06,
+      "loss": 0.7321,
+      "step": 601880
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 5.964643759087778e-06,
+      "loss": 0.7396,
+      "step": 601890
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 5.96349582918803e-06,
+      "loss": 0.6776,
+      "step": 601900
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 5.962347899288283e-06,
+      "loss": 0.7379,
+      "step": 601910
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 5.961199969388536e-06,
+      "loss": 0.7935,
+      "step": 601920
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 5.960052039488788e-06,
+      "loss": 0.6193,
+      "step": 601930
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 5.958904109589041e-06,
+      "loss": 0.7251,
+      "step": 601940
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 5.957756179689293e-06,
+      "loss": 0.647,
+      "step": 601950
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 5.956608249789545e-06,
+      "loss": 0.7254,
+      "step": 601960
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 5.955460319889798e-06,
+      "loss": 0.6592,
+      "step": 601970
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 5.954312389990051e-06,
+      "loss": 0.7302,
+      "step": 601980
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 5.953164460090303e-06,
+      "loss": 0.6593,
+      "step": 601990
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 5.952016530190556e-06,
+      "loss": 0.7496,
+      "step": 602000
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 5.950868600290809e-06,
+      "loss": 0.6718,
+      "step": 602010
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 5.9497206703910605e-06,
+      "loss": 0.6726,
+      "step": 602020
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 5.948572740491313e-06,
+      "loss": 0.7364,
+      "step": 602030
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 5.947424810591566e-06,
+      "loss": 0.6827,
+      "step": 602040
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 5.9462768806918185e-06,
+      "loss": 0.7675,
+      "step": 602050
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 5.945128950792071e-06,
+      "loss": 0.7549,
+      "step": 602060
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 5.943981020892324e-06,
+      "loss": 0.8126,
+      "step": 602070
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 5.9428330909925766e-06,
+      "loss": 0.6692,
+      "step": 602080
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 5.941685161092828e-06,
+      "loss": 0.6901,
+      "step": 602090
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 5.940537231193081e-06,
+      "loss": 0.726,
+      "step": 602100
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 5.939389301293334e-06,
+      "loss": 0.6895,
+      "step": 602110
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 5.938241371393586e-06,
+      "loss": 0.6763,
+      "step": 602120
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 5.9370934414938385e-06,
+      "loss": 0.7882,
+      "step": 602130
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 5.935945511594092e-06,
+      "loss": 0.7771,
+      "step": 602140
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 5.934797581694344e-06,
+      "loss": 0.7709,
+      "step": 602150
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 5.933649651794596e-06,
+      "loss": 0.6749,
+      "step": 602160
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 5.932501721894849e-06,
+      "loss": 0.7011,
+      "step": 602170
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 5.931353791995101e-06,
+      "loss": 0.7156,
+      "step": 602180
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 5.930205862095354e-06,
+      "loss": 0.7706,
+      "step": 602190
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 5.929057932195607e-06,
+      "loss": 0.6773,
+      "step": 602200
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 5.927910002295859e-06,
+      "loss": 0.638,
+      "step": 602210
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 5.926762072396112e-06,
+      "loss": 0.6945,
+      "step": 602220
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 5.925614142496365e-06,
+      "loss": 0.6966,
+      "step": 602230
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 5.924466212596617e-06,
+      "loss": 0.7225,
+      "step": 602240
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 5.923318282696869e-06,
+      "loss": 0.6527,
+      "step": 602250
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 5.922170352797122e-06,
+      "loss": 0.7057,
+      "step": 602260
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 5.921022422897375e-06,
+      "loss": 0.6476,
+      "step": 602270
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 5.919874492997627e-06,
+      "loss": 0.7477,
+      "step": 602280
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 5.91872656309788e-06,
+      "loss": 0.707,
+      "step": 602290
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 5.917578633198133e-06,
+      "loss": 0.7062,
+      "step": 602300
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 5.916430703298384e-06,
+      "loss": 0.7041,
+      "step": 602310
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 5.915282773398637e-06,
+      "loss": 0.742,
+      "step": 602320
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 5.91413484349889e-06,
+      "loss": 0.6648,
+      "step": 602330
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 5.912986913599142e-06,
+      "loss": 0.6934,
+      "step": 602340
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 5.911838983699395e-06,
+      "loss": 0.7814,
+      "step": 602350
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 5.910691053799648e-06,
+      "loss": 0.7092,
+      "step": 602360
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 5.9095431238999e-06,
+      "loss": 0.6966,
+      "step": 602370
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 5.908395194000152e-06,
+      "loss": 0.7315,
+      "step": 602380
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 5.907247264100405e-06,
+      "loss": 0.6725,
+      "step": 602390
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 5.9060993342006574e-06,
+      "loss": 0.6812,
+      "step": 602400
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 5.90495140430091e-06,
+      "loss": 0.6546,
+      "step": 602410
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 5.903803474401163e-06,
+      "loss": 0.6877,
+      "step": 602420
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 5.9026555445014155e-06,
+      "loss": 0.7477,
+      "step": 602430
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 5.901507614601668e-06,
+      "loss": 0.7027,
+      "step": 602440
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 5.900359684701921e-06,
+      "loss": 0.704,
+      "step": 602450
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 5.899211754802173e-06,
+      "loss": 0.664,
+      "step": 602460
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 5.898063824902425e-06,
+      "loss": 0.7543,
+      "step": 602470
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 5.8969158950026775e-06,
+      "loss": 0.752,
+      "step": 602480
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 5.895767965102931e-06,
+      "loss": 0.7115,
+      "step": 602490
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 5.894620035203183e-06,
+      "loss": 0.7829,
+      "step": 602500
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 5.8934721053034355e-06,
+      "loss": 0.7158,
+      "step": 602510
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 5.892324175403689e-06,
+      "loss": 0.7549,
+      "step": 602520
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 5.89117624550394e-06,
+      "loss": 0.6828,
+      "step": 602530
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 5.890028315604193e-06,
+      "loss": 0.7491,
+      "step": 602540
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 5.888880385704446e-06,
+      "loss": 0.7228,
+      "step": 602550
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 5.887732455804698e-06,
+      "loss": 0.6844,
+      "step": 602560
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 5.886584525904951e-06,
+      "loss": 0.6808,
+      "step": 602570
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 5.885436596005204e-06,
+      "loss": 0.7232,
+      "step": 602580
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 5.884288666105456e-06,
+      "loss": 0.7049,
+      "step": 602590
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 5.883140736205708e-06,
+      "loss": 0.7206,
+      "step": 602600
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 5.881992806305961e-06,
+      "loss": 0.7144,
+      "step": 602610
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 5.8808448764062135e-06,
+      "loss": 0.6686,
+      "step": 602620
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 5.879696946506466e-06,
+      "loss": 0.7528,
+      "step": 602630
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 5.878549016606718e-06,
+      "loss": 0.7085,
+      "step": 602640
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 5.8774010867069716e-06,
+      "loss": 0.6932,
+      "step": 602650
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 5.876253156807224e-06,
+      "loss": 0.5988,
+      "step": 602660
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 5.8751052269074755e-06,
+      "loss": 0.7053,
+      "step": 602670
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 5.873957297007729e-06,
+      "loss": 0.6611,
+      "step": 602680
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 5.872809367107981e-06,
+      "loss": 0.7763,
+      "step": 602690
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 5.8716614372082335e-06,
+      "loss": 0.6969,
+      "step": 602700
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 5.870513507308487e-06,
+      "loss": 0.7163,
+      "step": 602710
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 5.869365577408739e-06,
+      "loss": 0.7469,
+      "step": 602720
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 5.8682176475089916e-06,
+      "loss": 0.7213,
+      "step": 602730
+    },
+    {
+      "epoch": 27.65,
+      "learning_rate": 5.867069717609245e-06,
+      "loss": 0.6375,
+      "step": 602740
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 5.865921787709496e-06,
+      "loss": 0.7105,
+      "step": 602750
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 5.864773857809749e-06,
+      "loss": 0.6999,
+      "step": 602760
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 5.863625927910002e-06,
+      "loss": 0.7438,
+      "step": 602770
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 5.862477998010254e-06,
+      "loss": 0.7065,
+      "step": 602780
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 5.861330068110507e-06,
+      "loss": 0.7099,
+      "step": 602790
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 5.86018213821076e-06,
+      "loss": 0.6854,
+      "step": 602800
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 5.8590342083110124e-06,
+      "loss": 0.7704,
+      "step": 602810
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 5.857886278411264e-06,
+      "loss": 0.7012,
+      "step": 602820
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 5.856738348511516e-06,
+      "loss": 0.6426,
+      "step": 602830
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 5.85559041861177e-06,
+      "loss": 0.6653,
+      "step": 602840
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 5.854442488712022e-06,
+      "loss": 0.6946,
+      "step": 602850
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 5.853294558812274e-06,
+      "loss": 0.7129,
+      "step": 602860
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 5.852146628912528e-06,
+      "loss": 0.7225,
+      "step": 602870
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 5.85099869901278e-06,
+      "loss": 0.6934,
+      "step": 602880
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 5.849850769113032e-06,
+      "loss": 0.7319,
+      "step": 602890
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 5.848702839213285e-06,
+      "loss": 0.7052,
+      "step": 602900
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 5.847554909313537e-06,
+      "loss": 0.6601,
+      "step": 602910
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 5.84640697941379e-06,
+      "loss": 0.6389,
+      "step": 602920
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 5.845259049514043e-06,
+      "loss": 0.7604,
+      "step": 602930
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 5.844111119614295e-06,
+      "loss": 0.6725,
+      "step": 602940
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 5.842963189714548e-06,
+      "loss": 0.7049,
+      "step": 602950
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 5.841815259814801e-06,
+      "loss": 0.7204,
+      "step": 602960
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 5.8406673299150525e-06,
+      "loss": 0.7655,
+      "step": 602970
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 5.839519400015305e-06,
+      "loss": 0.7526,
+      "step": 602980
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 5.838371470115557e-06,
+      "loss": 0.7711,
+      "step": 602990
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 5.8372235402158105e-06,
+      "loss": 0.6904,
+      "step": 603000
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 5.836075610316063e-06,
+      "loss": 0.6901,
+      "step": 603010
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 5.834927680416315e-06,
+      "loss": 0.7428,
+      "step": 603020
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 5.8337797505165685e-06,
+      "loss": 0.7328,
+      "step": 603030
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 5.83263182061682e-06,
+      "loss": 0.808,
+      "step": 603040
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 5.8314838907170725e-06,
+      "loss": 0.7414,
+      "step": 603050
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 5.830335960817326e-06,
+      "loss": 0.6838,
+      "step": 603060
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 5.829188030917578e-06,
+      "loss": 0.7606,
+      "step": 603070
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 5.8280401010178305e-06,
+      "loss": 0.6454,
+      "step": 603080
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 5.826892171118084e-06,
+      "loss": 0.7131,
+      "step": 603090
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 5.825744241218336e-06,
+      "loss": 0.6884,
+      "step": 603100
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 5.824596311318588e-06,
+      "loss": 0.6698,
+      "step": 603110
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 5.823448381418841e-06,
+      "loss": 0.7777,
+      "step": 603120
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 5.822300451519093e-06,
+      "loss": 0.7591,
+      "step": 603130
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 5.821152521619346e-06,
+      "loss": 0.6846,
+      "step": 603140
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 5.820004591719599e-06,
+      "loss": 0.7437,
+      "step": 603150
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 5.818856661819851e-06,
+      "loss": 0.6777,
+      "step": 603160
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 5.817708731920104e-06,
+      "loss": 0.701,
+      "step": 603170
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 5.816560802020355e-06,
+      "loss": 0.7187,
+      "step": 603180
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 5.8154128721206085e-06,
+      "loss": 0.735,
+      "step": 603190
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 5.814264942220861e-06,
+      "loss": 0.712,
+      "step": 603200
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 5.813117012321113e-06,
+      "loss": 0.7683,
+      "step": 603210
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 5.8119690824213666e-06,
+      "loss": 0.6972,
+      "step": 603220
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 5.810821152521619e-06,
+      "loss": 0.7457,
+      "step": 603230
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 5.809673222621871e-06,
+      "loss": 0.6549,
+      "step": 603240
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 5.808525292722125e-06,
+      "loss": 0.7606,
+      "step": 603250
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 5.807377362822376e-06,
+      "loss": 0.7569,
+      "step": 603260
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 5.8062294329226286e-06,
+      "loss": 0.702,
+      "step": 603270
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 5.805081503022882e-06,
+      "loss": 0.6646,
+      "step": 603280
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 5.803933573123134e-06,
+      "loss": 0.6503,
+      "step": 603290
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 5.802785643223387e-06,
+      "loss": 0.7389,
+      "step": 603300
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 5.80163771332364e-06,
+      "loss": 0.7207,
+      "step": 603310
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 5.800489783423892e-06,
+      "loss": 0.6488,
+      "step": 603320
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 5.799341853524144e-06,
+      "loss": 0.6747,
+      "step": 603330
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 5.798193923624396e-06,
+      "loss": 0.6962,
+      "step": 603340
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 5.797045993724649e-06,
+      "loss": 0.706,
+      "step": 603350
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 5.795898063824902e-06,
+      "loss": 0.7872,
+      "step": 603360
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 5.794750133925154e-06,
+      "loss": 0.7834,
+      "step": 603370
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 5.7936022040254074e-06,
+      "loss": 0.6661,
+      "step": 603380
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 5.79245427412566e-06,
+      "loss": 0.6831,
+      "step": 603390
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 5.791306344225911e-06,
+      "loss": 0.6673,
+      "step": 603400
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 5.790158414326165e-06,
+      "loss": 0.728,
+      "step": 603410
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 5.789010484426417e-06,
+      "loss": 0.7317,
+      "step": 603420
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 5.787862554526669e-06,
+      "loss": 0.6323,
+      "step": 603430
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 5.786714624626923e-06,
+      "loss": 0.5868,
+      "step": 603440
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 5.785566694727175e-06,
+      "loss": 0.7163,
+      "step": 603450
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 5.7844187648274275e-06,
+      "loss": 0.7814,
+      "step": 603460
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 5.783270834927681e-06,
+      "loss": 0.7116,
+      "step": 603470
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 5.782122905027932e-06,
+      "loss": 0.7011,
+      "step": 603480
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 5.780974975128185e-06,
+      "loss": 0.7243,
+      "step": 603490
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 5.779827045228437e-06,
+      "loss": 0.7676,
+      "step": 603500
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 5.77867911532869e-06,
+      "loss": 0.6194,
+      "step": 603510
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 5.777531185428943e-06,
+      "loss": 0.7191,
+      "step": 603520
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 5.776383255529195e-06,
+      "loss": 0.7077,
+      "step": 603530
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 5.775235325629448e-06,
+      "loss": 0.6226,
+      "step": 603540
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 5.7740873957297e-06,
+      "loss": 0.6999,
+      "step": 603550
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 5.772939465829952e-06,
+      "loss": 0.6476,
+      "step": 603560
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 5.7717915359302055e-06,
+      "loss": 0.7057,
+      "step": 603570
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 5.770643606030458e-06,
+      "loss": 0.6955,
+      "step": 603580
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 5.76949567613071e-06,
+      "loss": 0.7524,
+      "step": 603590
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 5.7683477462309635e-06,
+      "loss": 0.7345,
+      "step": 603600
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 5.767199816331216e-06,
+      "loss": 0.7664,
+      "step": 603610
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 5.7660518864314675e-06,
+      "loss": 0.6736,
+      "step": 603620
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 5.764903956531721e-06,
+      "loss": 0.6214,
+      "step": 603630
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 5.763756026631973e-06,
+      "loss": 0.6945,
+      "step": 603640
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 5.7626080967322255e-06,
+      "loss": 0.6966,
+      "step": 603650
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 5.761460166832479e-06,
+      "loss": 0.6532,
+      "step": 603660
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 5.760312236932731e-06,
+      "loss": 0.7283,
+      "step": 603670
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 5.7591643070329835e-06,
+      "loss": 0.6769,
+      "step": 603680
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 5.758016377133235e-06,
+      "loss": 0.6586,
+      "step": 603690
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 5.756868447233488e-06,
+      "loss": 0.7169,
+      "step": 603700
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 5.755720517333741e-06,
+      "loss": 0.6725,
+      "step": 603710
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 5.754572587433993e-06,
+      "loss": 0.6477,
+      "step": 603720
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 5.753424657534246e-06,
+      "loss": 0.7133,
+      "step": 603730
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 5.752276727634499e-06,
+      "loss": 0.7373,
+      "step": 603740
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 5.751128797734751e-06,
+      "loss": 0.6484,
+      "step": 603750
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 5.749980867835004e-06,
+      "loss": 0.7517,
+      "step": 603760
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 5.748832937935256e-06,
+      "loss": 0.7217,
+      "step": 603770
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 5.747685008035508e-06,
+      "loss": 0.7857,
+      "step": 603780
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 5.746537078135762e-06,
+      "loss": 0.6633,
+      "step": 603790
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 5.745389148236014e-06,
+      "loss": 0.7093,
+      "step": 603800
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 5.744241218336266e-06,
+      "loss": 0.6228,
+      "step": 603810
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 5.74309328843652e-06,
+      "loss": 0.6928,
+      "step": 603820
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 5.741945358536772e-06,
+      "loss": 0.6009,
+      "step": 603830
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 5.7407974286370236e-06,
+      "loss": 0.699,
+      "step": 603840
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 5.739649498737276e-06,
+      "loss": 0.6648,
+      "step": 603850
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 5.738501568837529e-06,
+      "loss": 0.7738,
+      "step": 603860
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 5.737353638937782e-06,
+      "loss": 0.702,
+      "step": 603870
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 5.736205709038034e-06,
+      "loss": 0.7345,
+      "step": 603880
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 5.735057779138287e-06,
+      "loss": 0.6474,
+      "step": 603890
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 5.73390984923854e-06,
+      "loss": 0.7173,
+      "step": 603900
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 5.732761919338791e-06,
+      "loss": 0.6204,
+      "step": 603910
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 5.731613989439044e-06,
+      "loss": 0.7094,
+      "step": 603920
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 5.730466059539297e-06,
+      "loss": 0.6891,
+      "step": 603930
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 5.729318129639549e-06,
+      "loss": 0.7306,
+      "step": 603940
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 5.7281701997398024e-06,
+      "loss": 0.7602,
+      "step": 603950
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 5.727022269840055e-06,
+      "loss": 0.6977,
+      "step": 603960
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 5.725874339940307e-06,
+      "loss": 0.6819,
+      "step": 603970
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 5.7247264100405605e-06,
+      "loss": 0.6712,
+      "step": 603980
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 5.723578480140812e-06,
+      "loss": 0.661,
+      "step": 603990
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 5.7224305502410644e-06,
+      "loss": 0.6711,
+      "step": 604000
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 5.721282620341317e-06,
+      "loss": 0.7147,
+      "step": 604010
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 5.72013469044157e-06,
+      "loss": 0.7724,
+      "step": 604020
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 5.7189867605418225e-06,
+      "loss": 0.6949,
+      "step": 604030
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 5.717838830642075e-06,
+      "loss": 0.6982,
+      "step": 604040
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 5.716690900742328e-06,
+      "loss": 0.7309,
+      "step": 604050
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 5.71554297084258e-06,
+      "loss": 0.6701,
+      "step": 604060
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 5.714395040942832e-06,
+      "loss": 0.6885,
+      "step": 604070
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 5.713247111043085e-06,
+      "loss": 0.7487,
+      "step": 604080
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 5.712099181143338e-06,
+      "loss": 0.7076,
+      "step": 604090
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 5.71095125124359e-06,
+      "loss": 0.6696,
+      "step": 604100
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 5.709803321343843e-06,
+      "loss": 0.7604,
+      "step": 604110
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 5.708655391444096e-06,
+      "loss": 0.6756,
+      "step": 604120
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 5.707507461544347e-06,
+      "loss": 0.6916,
+      "step": 604130
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 5.7063595316446005e-06,
+      "loss": 0.7195,
+      "step": 604140
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 5.705211601744853e-06,
+      "loss": 0.8115,
+      "step": 604150
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 5.704063671845105e-06,
+      "loss": 0.6635,
+      "step": 604160
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 5.7029157419453585e-06,
+      "loss": 0.6285,
+      "step": 604170
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 5.701767812045611e-06,
+      "loss": 0.6894,
+      "step": 604180
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 5.700619882145863e-06,
+      "loss": 0.7223,
+      "step": 604190
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 5.699471952246115e-06,
+      "loss": 0.7542,
+      "step": 604200
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 5.698324022346368e-06,
+      "loss": 0.6181,
+      "step": 604210
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 5.6971760924466205e-06,
+      "loss": 0.7637,
+      "step": 604220
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 5.696028162546873e-06,
+      "loss": 0.7262,
+      "step": 604230
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 5.694880232647126e-06,
+      "loss": 0.7265,
+      "step": 604240
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 5.6937323027473785e-06,
+      "loss": 0.6368,
+      "step": 604250
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 5.692584372847631e-06,
+      "loss": 0.6987,
+      "step": 604260
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 5.691436442947884e-06,
+      "loss": 0.7145,
+      "step": 604270
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 5.690288513048136e-06,
+      "loss": 0.6543,
+      "step": 604280
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 5.689140583148388e-06,
+      "loss": 0.7221,
+      "step": 604290
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 5.687992653248641e-06,
+      "loss": 0.6618,
+      "step": 604300
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 5.686844723348894e-06,
+      "loss": 0.731,
+      "step": 604310
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 5.685696793449146e-06,
+      "loss": 0.7229,
+      "step": 604320
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 5.684548863549399e-06,
+      "loss": 0.786,
+      "step": 604330
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 5.683400933649652e-06,
+      "loss": 0.6753,
+      "step": 604340
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 5.682253003749903e-06,
+      "loss": 0.6094,
+      "step": 604350
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 5.681105073850156e-06,
+      "loss": 0.6557,
+      "step": 604360
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 5.679957143950409e-06,
+      "loss": 0.7698,
+      "step": 604370
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 5.678809214050661e-06,
+      "loss": 0.606,
+      "step": 604380
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 5.677661284150914e-06,
+      "loss": 0.7015,
+      "step": 604390
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 5.676513354251167e-06,
+      "loss": 0.6754,
+      "step": 604400
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 5.675365424351419e-06,
+      "loss": 0.6557,
+      "step": 604410
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 5.674217494451671e-06,
+      "loss": 0.6972,
+      "step": 604420
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 5.673069564551924e-06,
+      "loss": 0.7211,
+      "step": 604430
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 5.671921634652177e-06,
+      "loss": 0.7145,
+      "step": 604440
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 5.670773704752429e-06,
+      "loss": 0.7288,
+      "step": 604450
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 5.669625774852682e-06,
+      "loss": 0.6647,
+      "step": 604460
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 5.668477844952935e-06,
+      "loss": 0.7847,
+      "step": 604470
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 5.667329915053187e-06,
+      "loss": 0.743,
+      "step": 604480
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 5.66618198515344e-06,
+      "loss": 0.707,
+      "step": 604490
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 5.665034055253692e-06,
+      "loss": 0.6978,
+      "step": 604500
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 5.663886125353944e-06,
+      "loss": 0.7186,
+      "step": 604510
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 5.6627381954541975e-06,
+      "loss": 0.7302,
+      "step": 604520
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 5.66159026555445e-06,
+      "loss": 0.7441,
+      "step": 604530
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 5.660442335654702e-06,
+      "loss": 0.7035,
+      "step": 604540
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 5.659294405754955e-06,
+      "loss": 0.6914,
+      "step": 604550
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 5.658146475855208e-06,
+      "loss": 0.7315,
+      "step": 604560
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 5.6569985459554594e-06,
+      "loss": 0.681,
+      "step": 604570
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 5.655850616055712e-06,
+      "loss": 0.7911,
+      "step": 604580
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 5.654702686155965e-06,
+      "loss": 0.7462,
+      "step": 604590
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 5.6535547562562175e-06,
+      "loss": 0.7148,
+      "step": 604600
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 5.65240682635647e-06,
+      "loss": 0.6723,
+      "step": 604610
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 5.651258896456723e-06,
+      "loss": 0.7257,
+      "step": 604620
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 5.6501109665569755e-06,
+      "loss": 0.7196,
+      "step": 604630
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 5.648963036657227e-06,
+      "loss": 0.7193,
+      "step": 604640
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 5.64781510675748e-06,
+      "loss": 0.6584,
+      "step": 604650
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 5.646667176857733e-06,
+      "loss": 0.738,
+      "step": 604660
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 5.645519246957985e-06,
+      "loss": 0.6798,
+      "step": 604670
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 5.644371317058238e-06,
+      "loss": 0.6541,
+      "step": 604680
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 5.643223387158491e-06,
+      "loss": 0.6946,
+      "step": 604690
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 5.642075457258743e-06,
+      "loss": 0.7315,
+      "step": 604700
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 5.640927527358995e-06,
+      "loss": 0.682,
+      "step": 604710
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 5.639779597459248e-06,
+      "loss": 0.6868,
+      "step": 604720
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 5.6386316675595e-06,
+      "loss": 0.7162,
+      "step": 604730
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 5.637483737659753e-06,
+      "loss": 0.7055,
+      "step": 604740
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 5.636335807760006e-06,
+      "loss": 0.6569,
+      "step": 604750
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 5.635187877860258e-06,
+      "loss": 0.7307,
+      "step": 604760
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 5.634039947960511e-06,
+      "loss": 0.7172,
+      "step": 604770
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 5.632892018060764e-06,
+      "loss": 0.7064,
+      "step": 604780
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 5.6317440881610155e-06,
+      "loss": 0.6767,
+      "step": 604790
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 5.630596158261268e-06,
+      "loss": 0.7082,
+      "step": 604800
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 5.629448228361521e-06,
+      "loss": 0.6921,
+      "step": 604810
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 5.6283002984617736e-06,
+      "loss": 0.7395,
+      "step": 604820
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 5.627152368562026e-06,
+      "loss": 0.7086,
+      "step": 604830
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 5.626004438662279e-06,
+      "loss": 0.7455,
+      "step": 604840
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 5.624856508762532e-06,
+      "loss": 0.7297,
+      "step": 604850
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 5.623708578862783e-06,
+      "loss": 0.7311,
+      "step": 604860
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 5.6225606489630355e-06,
+      "loss": 0.6316,
+      "step": 604870
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 5.621412719063289e-06,
+      "loss": 0.7378,
+      "step": 604880
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 5.620264789163541e-06,
+      "loss": 0.7352,
+      "step": 604890
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 5.6191168592637936e-06,
+      "loss": 0.6863,
+      "step": 604900
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 5.617968929364047e-06,
+      "loss": 0.7799,
+      "step": 604910
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 5.616820999464299e-06,
+      "loss": 0.7356,
+      "step": 604920
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 5.615673069564551e-06,
+      "loss": 0.5785,
+      "step": 604930
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 5.614525139664804e-06,
+      "loss": 0.6963,
+      "step": 604940
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 5.613377209765056e-06,
+      "loss": 0.7215,
+      "step": 604950
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 5.612229279865309e-06,
+      "loss": 0.6768,
+      "step": 604960
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 5.611081349965562e-06,
+      "loss": 0.7725,
+      "step": 604970
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 5.609933420065814e-06,
+      "loss": 0.7592,
+      "step": 604980
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 5.608785490166067e-06,
+      "loss": 0.6351,
+      "step": 604990
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 5.60763756026632e-06,
+      "loss": 0.7535,
+      "step": 605000
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 5.606489630366572e-06,
+      "loss": 0.7056,
+      "step": 605010
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 5.605341700466824e-06,
+      "loss": 0.711,
+      "step": 605020
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 5.604193770567077e-06,
+      "loss": 0.7198,
+      "step": 605030
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 5.60304584066733e-06,
+      "loss": 0.7031,
+      "step": 605040
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 5.601897910767582e-06,
+      "loss": 0.6932,
+      "step": 605050
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 5.6007499808678344e-06,
+      "loss": 0.7375,
+      "step": 605060
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 5.599602050968088e-06,
+      "loss": 0.6755,
+      "step": 605070
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 5.598454121068339e-06,
+      "loss": 0.7287,
+      "step": 605080
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 5.597306191168592e-06,
+      "loss": 0.7273,
+      "step": 605090
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 5.596158261268845e-06,
+      "loss": 0.6511,
+      "step": 605100
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 5.595010331369097e-06,
+      "loss": 0.7094,
+      "step": 605110
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 5.59386240146935e-06,
+      "loss": 0.8445,
+      "step": 605120
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 5.592714471569603e-06,
+      "loss": 0.7095,
+      "step": 605130
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 5.591566541669855e-06,
+      "loss": 0.6847,
+      "step": 605140
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 5.590418611770107e-06,
+      "loss": 0.7222,
+      "step": 605150
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 5.58927068187036e-06,
+      "loss": 0.6535,
+      "step": 605160
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 5.5881227519706125e-06,
+      "loss": 0.6952,
+      "step": 605170
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 5.586974822070865e-06,
+      "loss": 0.6346,
+      "step": 605180
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 5.585826892171118e-06,
+      "loss": 0.5964,
+      "step": 605190
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 5.5846789622713705e-06,
+      "loss": 0.6667,
+      "step": 605200
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 5.583531032371623e-06,
+      "loss": 0.8452,
+      "step": 605210
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 5.5823831024718745e-06,
+      "loss": 0.6686,
+      "step": 605220
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 5.581235172572128e-06,
+      "loss": 0.6994,
+      "step": 605230
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 5.58008724267238e-06,
+      "loss": 0.7182,
+      "step": 605240
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 5.5789393127726325e-06,
+      "loss": 0.6959,
+      "step": 605250
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 5.577791382872886e-06,
+      "loss": 0.6722,
+      "step": 605260
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 5.576643452973138e-06,
+      "loss": 0.6543,
+      "step": 605270
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 5.5754955230733905e-06,
+      "loss": 0.7145,
+      "step": 605280
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 5.574347593173644e-06,
+      "loss": 0.7291,
+      "step": 605290
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 5.573199663273895e-06,
+      "loss": 0.693,
+      "step": 605300
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 5.572051733374148e-06,
+      "loss": 0.6675,
+      "step": 605310
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 5.570903803474401e-06,
+      "loss": 0.6856,
+      "step": 605320
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 5.569755873574653e-06,
+      "loss": 0.6499,
+      "step": 605330
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 5.568607943674906e-06,
+      "loss": 0.7815,
+      "step": 605340
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 5.567460013775159e-06,
+      "loss": 0.6843,
+      "step": 605350
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 5.566312083875411e-06,
+      "loss": 0.7497,
+      "step": 605360
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 5.565164153975663e-06,
+      "loss": 0.6548,
+      "step": 605370
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 5.564016224075916e-06,
+      "loss": 0.6584,
+      "step": 605380
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 5.5628682941761686e-06,
+      "loss": 0.6757,
+      "step": 605390
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 5.561720364276421e-06,
+      "loss": 0.7572,
+      "step": 605400
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 5.560572434376673e-06,
+      "loss": 0.7345,
+      "step": 605410
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 5.559424504476927e-06,
+      "loss": 0.7045,
+      "step": 605420
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 5.558276574577179e-06,
+      "loss": 0.6756,
+      "step": 605430
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 5.5571286446774305e-06,
+      "loss": 0.7363,
+      "step": 605440
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 5.555980714777684e-06,
+      "loss": 0.7382,
+      "step": 605450
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 5.554832784877936e-06,
+      "loss": 0.7007,
+      "step": 605460
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 5.5536848549781886e-06,
+      "loss": 0.6714,
+      "step": 605470
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 5.552536925078442e-06,
+      "loss": 0.7202,
+      "step": 605480
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 5.551388995178694e-06,
+      "loss": 0.707,
+      "step": 605490
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 5.550241065278947e-06,
+      "loss": 0.6464,
+      "step": 605500
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 5.5490931353792e-06,
+      "loss": 0.7194,
+      "step": 605510
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 5.547945205479451e-06,
+      "loss": 0.7084,
+      "step": 605520
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 5.546797275579704e-06,
+      "loss": 0.6984,
+      "step": 605530
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 5.545649345679957e-06,
+      "loss": 0.77,
+      "step": 605540
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 5.5445014157802094e-06,
+      "loss": 0.6796,
+      "step": 605550
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 5.543353485880462e-06,
+      "loss": 0.6651,
+      "step": 605560
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 5.542205555980714e-06,
+      "loss": 0.7041,
+      "step": 605570
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 5.5410576260809675e-06,
+      "loss": 0.7261,
+      "step": 605580
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 5.539909696181219e-06,
+      "loss": 0.7291,
+      "step": 605590
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 5.538761766281471e-06,
+      "loss": 0.7375,
+      "step": 605600
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 5.537613836381725e-06,
+      "loss": 0.6475,
+      "step": 605610
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 5.536465906481977e-06,
+      "loss": 0.654,
+      "step": 605620
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 5.5353179765822294e-06,
+      "loss": 0.7131,
+      "step": 605630
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 5.534170046682483e-06,
+      "loss": 0.665,
+      "step": 605640
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 5.533022116782735e-06,
+      "loss": 0.7136,
+      "step": 605650
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 5.531874186882987e-06,
+      "loss": 0.668,
+      "step": 605660
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 5.53072625698324e-06,
+      "loss": 0.6931,
+      "step": 605670
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 5.529578327083492e-06,
+      "loss": 0.6853,
+      "step": 605680
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 5.528430397183745e-06,
+      "loss": 0.6799,
+      "step": 605690
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 5.527282467283998e-06,
+      "loss": 0.7428,
+      "step": 605700
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 5.52613453738425e-06,
+      "loss": 0.6364,
+      "step": 605710
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 5.524986607484503e-06,
+      "loss": 0.744,
+      "step": 605720
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 5.523838677584754e-06,
+      "loss": 0.6975,
+      "step": 605730
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 5.5226907476850075e-06,
+      "loss": 0.7117,
+      "step": 605740
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 5.52154281778526e-06,
+      "loss": 0.6962,
+      "step": 605750
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 5.520394887885512e-06,
+      "loss": 0.7588,
+      "step": 605760
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 5.5192469579857655e-06,
+      "loss": 0.6364,
+      "step": 605770
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 5.518099028086018e-06,
+      "loss": 0.6394,
+      "step": 605780
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 5.51695109818627e-06,
+      "loss": 0.6782,
+      "step": 605790
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 5.5158031682865235e-06,
+      "loss": 0.7148,
+      "step": 605800
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 5.514655238386775e-06,
+      "loss": 0.7386,
+      "step": 605810
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 5.5135073084870275e-06,
+      "loss": 0.6713,
+      "step": 605820
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 5.512359378587281e-06,
+      "loss": 0.6971,
+      "step": 605830
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 5.511211448687533e-06,
+      "loss": 0.6916,
+      "step": 605840
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 5.5100635187877855e-06,
+      "loss": 0.7176,
+      "step": 605850
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 5.508915588888039e-06,
+      "loss": 0.6714,
+      "step": 605860
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 5.507767658988291e-06,
+      "loss": 0.6107,
+      "step": 605870
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 5.506619729088543e-06,
+      "loss": 0.6719,
+      "step": 605880
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 5.505471799188796e-06,
+      "loss": 0.6406,
+      "step": 605890
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 5.504323869289048e-06,
+      "loss": 0.7189,
+      "step": 605900
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 5.503175939389301e-06,
+      "loss": 0.6148,
+      "step": 605910
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 5.502028009489553e-06,
+      "loss": 0.7108,
+      "step": 605920
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 5.500880079589806e-06,
+      "loss": 0.7689,
+      "step": 605930
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 5.499732149690059e-06,
+      "loss": 0.6653,
+      "step": 605940
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 5.49858421979031e-06,
+      "loss": 0.7284,
+      "step": 605950
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 5.4974362898905636e-06,
+      "loss": 0.6378,
+      "step": 605960
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 5.496288359990816e-06,
+      "loss": 0.7685,
+      "step": 605970
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 5.495140430091068e-06,
+      "loss": 0.6825,
+      "step": 605980
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 5.493992500191322e-06,
+      "loss": 0.7707,
+      "step": 605990
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 5.492844570291574e-06,
+      "loss": 0.6792,
+      "step": 606000
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 5.491696640391826e-06,
+      "loss": 0.6453,
+      "step": 606010
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 5.49054871049208e-06,
+      "loss": 0.7001,
+      "step": 606020
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 5.489400780592331e-06,
+      "loss": 0.72,
+      "step": 606030
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 5.488252850692584e-06,
+      "loss": 0.7483,
+      "step": 606040
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 5.487104920792837e-06,
+      "loss": 0.6887,
+      "step": 606050
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 5.485956990893089e-06,
+      "loss": 0.6901,
+      "step": 606060
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 5.484809060993342e-06,
+      "loss": 0.715,
+      "step": 606070
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 5.483661131093593e-06,
+      "loss": 0.7612,
+      "step": 606080
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 5.482513201193847e-06,
+      "loss": 0.6724,
+      "step": 606090
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 5.481365271294099e-06,
+      "loss": 0.6513,
+      "step": 606100
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 5.480217341394351e-06,
+      "loss": 0.6863,
+      "step": 606110
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 5.4790694114946044e-06,
+      "loss": 0.6916,
+      "step": 606120
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 5.477921481594857e-06,
+      "loss": 0.7098,
+      "step": 606130
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 5.476773551695109e-06,
+      "loss": 0.6457,
+      "step": 606140
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 5.4756256217953625e-06,
+      "loss": 0.5725,
+      "step": 606150
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 5.474477691895615e-06,
+      "loss": 0.8004,
+      "step": 606160
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 5.473329761995866e-06,
+      "loss": 0.7155,
+      "step": 606170
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 5.47218183209612e-06,
+      "loss": 0.6255,
+      "step": 606180
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 5.471033902196372e-06,
+      "loss": 0.7084,
+      "step": 606190
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 5.4698859722966244e-06,
+      "loss": 0.7149,
+      "step": 606200
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 5.468738042396878e-06,
+      "loss": 0.7289,
+      "step": 606210
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 5.46759011249713e-06,
+      "loss": 0.7481,
+      "step": 606220
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 5.466442182597382e-06,
+      "loss": 0.6663,
+      "step": 606230
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 5.465294252697634e-06,
+      "loss": 0.6599,
+      "step": 606240
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 5.464146322797887e-06,
+      "loss": 0.6933,
+      "step": 606250
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 5.46299839289814e-06,
+      "loss": 0.6199,
+      "step": 606260
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 5.461850462998392e-06,
+      "loss": 0.6471,
+      "step": 606270
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 5.460702533098645e-06,
+      "loss": 0.7375,
+      "step": 606280
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 5.459554603198898e-06,
+      "loss": 0.7081,
+      "step": 606290
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 5.458406673299149e-06,
+      "loss": 0.7408,
+      "step": 606300
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 5.457258743399403e-06,
+      "loss": 0.6682,
+      "step": 606310
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 5.456110813499655e-06,
+      "loss": 0.659,
+      "step": 606320
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 5.454962883599907e-06,
+      "loss": 0.7246,
+      "step": 606330
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 5.4538149537001605e-06,
+      "loss": 0.709,
+      "step": 606340
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 5.452667023800413e-06,
+      "loss": 0.6313,
+      "step": 606350
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 5.451519093900665e-06,
+      "loss": 0.6367,
+      "step": 606360
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 5.4503711640009186e-06,
+      "loss": 0.7589,
+      "step": 606370
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 5.44922323410117e-06,
+      "loss": 0.6238,
+      "step": 606380
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 5.4480753042014225e-06,
+      "loss": 0.5863,
+      "step": 606390
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 5.446927374301676e-06,
+      "loss": 0.6699,
+      "step": 606400
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 5.445779444401928e-06,
+      "loss": 0.738,
+      "step": 606410
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 5.4446315145021805e-06,
+      "loss": 0.7416,
+      "step": 606420
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 5.443483584602433e-06,
+      "loss": 0.7444,
+      "step": 606430
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 5.442335654702686e-06,
+      "loss": 0.7136,
+      "step": 606440
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 5.441187724802938e-06,
+      "loss": 0.7496,
+      "step": 606450
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 5.44003979490319e-06,
+      "loss": 0.6923,
+      "step": 606460
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 5.438891865003443e-06,
+      "loss": 0.7326,
+      "step": 606470
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 5.437743935103696e-06,
+      "loss": 0.6995,
+      "step": 606480
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 5.436596005203948e-06,
+      "loss": 0.7242,
+      "step": 606490
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 5.435448075304201e-06,
+      "loss": 0.7021,
+      "step": 606500
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 5.434300145404454e-06,
+      "loss": 0.7176,
+      "step": 606510
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 5.433152215504705e-06,
+      "loss": 0.6501,
+      "step": 606520
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 5.4320042856049586e-06,
+      "loss": 0.7068,
+      "step": 606530
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 5.430856355705211e-06,
+      "loss": 0.6529,
+      "step": 606540
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 5.429708425805463e-06,
+      "loss": 0.7091,
+      "step": 606550
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 5.428560495905717e-06,
+      "loss": 0.6049,
+      "step": 606560
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 5.427412566005969e-06,
+      "loss": 0.6849,
+      "step": 606570
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 5.426264636106221e-06,
+      "loss": 0.8038,
+      "step": 606580
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 5.425116706206473e-06,
+      "loss": 0.7008,
+      "step": 606590
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 5.423968776306726e-06,
+      "loss": 0.6871,
+      "step": 606600
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 5.422820846406979e-06,
+      "loss": 0.6598,
+      "step": 606610
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 5.421672916507231e-06,
+      "loss": 0.7946,
+      "step": 606620
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 5.420524986607484e-06,
+      "loss": 0.6933,
+      "step": 606630
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 5.419377056707737e-06,
+      "loss": 0.6516,
+      "step": 606640
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 5.418229126807989e-06,
+      "loss": 0.7529,
+      "step": 606650
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 5.417081196908242e-06,
+      "loss": 0.6387,
+      "step": 606660
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 5.415933267008494e-06,
+      "loss": 0.7364,
+      "step": 606670
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 5.414785337108746e-06,
+      "loss": 0.6734,
+      "step": 606680
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 5.4136374072089994e-06,
+      "loss": 0.7219,
+      "step": 606690
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 5.412489477309252e-06,
+      "loss": 0.7364,
+      "step": 606700
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 5.411341547409504e-06,
+      "loss": 0.6963,
+      "step": 606710
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 5.4101936175097575e-06,
+      "loss": 0.7482,
+      "step": 606720
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 5.40904568761001e-06,
+      "loss": 0.6239,
+      "step": 606730
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 5.407897757710261e-06,
+      "loss": 0.7106,
+      "step": 606740
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 5.406749827810515e-06,
+      "loss": 0.6913,
+      "step": 606750
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 5.405601897910767e-06,
+      "loss": 0.7439,
+      "step": 606760
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 5.4044539680110195e-06,
+      "loss": 0.6318,
+      "step": 606770
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 5.403306038111272e-06,
+      "loss": 0.7244,
+      "step": 606780
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 5.402158108211525e-06,
+      "loss": 0.6584,
+      "step": 606790
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 5.4010101783117775e-06,
+      "loss": 0.6394,
+      "step": 606800
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 5.399862248412029e-06,
+      "loss": 0.7125,
+      "step": 606810
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 5.398714318512282e-06,
+      "loss": 0.7092,
+      "step": 606820
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 5.397566388612535e-06,
+      "loss": 0.7854,
+      "step": 606830
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 5.396418458712787e-06,
+      "loss": 0.664,
+      "step": 606840
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 5.39527052881304e-06,
+      "loss": 0.6575,
+      "step": 606850
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 5.394122598913293e-06,
+      "loss": 0.6322,
+      "step": 606860
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 5.392974669013545e-06,
+      "loss": 0.7847,
+      "step": 606870
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 5.391826739113798e-06,
+      "loss": 0.7089,
+      "step": 606880
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 5.39067880921405e-06,
+      "loss": 0.6902,
+      "step": 606890
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 5.389530879314302e-06,
+      "loss": 0.6495,
+      "step": 606900
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 5.3883829494145555e-06,
+      "loss": 0.7008,
+      "step": 606910
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 5.387235019514808e-06,
+      "loss": 0.6959,
+      "step": 606920
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 5.38608708961506e-06,
+      "loss": 0.6708,
+      "step": 606930
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 5.384939159715313e-06,
+      "loss": 0.6887,
+      "step": 606940
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 5.383791229815566e-06,
+      "loss": 0.7158,
+      "step": 606950
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 5.3826432999158175e-06,
+      "loss": 0.6788,
+      "step": 606960
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 5.38149537001607e-06,
+      "loss": 0.6428,
+      "step": 606970
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 5.380347440116323e-06,
+      "loss": 0.672,
+      "step": 606980
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 5.3791995102165755e-06,
+      "loss": 0.7262,
+      "step": 606990
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 5.378051580316828e-06,
+      "loss": 0.7335,
+      "step": 607000
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 5.376903650417081e-06,
+      "loss": 0.6847,
+      "step": 607010
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 5.3757557205173336e-06,
+      "loss": 0.6636,
+      "step": 607020
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 5.374607790617585e-06,
+      "loss": 0.7316,
+      "step": 607030
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 5.373459860717838e-06,
+      "loss": 0.6821,
+      "step": 607040
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 5.372311930818091e-06,
+      "loss": 0.6949,
+      "step": 607050
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 5.371164000918343e-06,
+      "loss": 0.6947,
+      "step": 607060
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 5.370016071018596e-06,
+      "loss": 0.7107,
+      "step": 607070
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 5.368868141118849e-06,
+      "loss": 0.6217,
+      "step": 607080
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 5.367720211219101e-06,
+      "loss": 0.7203,
+      "step": 607090
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 5.366572281319353e-06,
+      "loss": 0.7457,
+      "step": 607100
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 5.365424351419606e-06,
+      "loss": 0.7252,
+      "step": 607110
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 5.364276421519858e-06,
+      "loss": 0.6669,
+      "step": 607120
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 5.363128491620111e-06,
+      "loss": 0.6694,
+      "step": 607130
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 5.361980561720364e-06,
+      "loss": 0.6257,
+      "step": 607140
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 5.360832631820616e-06,
+      "loss": 0.7257,
+      "step": 607150
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 5.359684701920869e-06,
+      "loss": 0.6805,
+      "step": 607160
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 5.358536772021122e-06,
+      "loss": 0.7344,
+      "step": 607170
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 5.357388842121374e-06,
+      "loss": 0.6803,
+      "step": 607180
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 5.356240912221626e-06,
+      "loss": 0.784,
+      "step": 607190
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 5.355092982321879e-06,
+      "loss": 0.6063,
+      "step": 607200
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 5.353945052422132e-06,
+      "loss": 0.7038,
+      "step": 607210
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 5.352797122522384e-06,
+      "loss": 0.8054,
+      "step": 607220
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 5.351649192622637e-06,
+      "loss": 0.6191,
+      "step": 607230
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 5.35050126272289e-06,
+      "loss": 0.7327,
+      "step": 607240
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 5.349353332823141e-06,
+      "loss": 0.642,
+      "step": 607250
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 5.3482054029233945e-06,
+      "loss": 0.72,
+      "step": 607260
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 5.347057473023647e-06,
+      "loss": 0.5911,
+      "step": 607270
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 5.345909543123899e-06,
+      "loss": 0.7351,
+      "step": 607280
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 5.344761613224152e-06,
+      "loss": 0.7159,
+      "step": 607290
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 5.343613683324405e-06,
+      "loss": 0.6836,
+      "step": 607300
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 5.342465753424657e-06,
+      "loss": 0.6465,
+      "step": 607310
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 5.341317823524909e-06,
+      "loss": 0.6943,
+      "step": 607320
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 5.340169893625162e-06,
+      "loss": 0.7248,
+      "step": 607330
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 5.3390219637254145e-06,
+      "loss": 0.7105,
+      "step": 607340
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 5.337874033825667e-06,
+      "loss": 0.7042,
+      "step": 607350
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 5.33672610392592e-06,
+      "loss": 0.67,
+      "step": 607360
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 5.3355781740261725e-06,
+      "loss": 0.668,
+      "step": 607370
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 5.334430244126425e-06,
+      "loss": 0.7725,
+      "step": 607380
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 5.333282314226678e-06,
+      "loss": 0.793,
+      "step": 607390
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 5.33213438432693e-06,
+      "loss": 0.6909,
+      "step": 607400
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 5.330986454427182e-06,
+      "loss": 0.6702,
+      "step": 607410
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 5.329838524527435e-06,
+      "loss": 0.665,
+      "step": 607420
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 5.328690594627688e-06,
+      "loss": 0.6686,
+      "step": 607430
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 5.32754266472794e-06,
+      "loss": 0.6954,
+      "step": 607440
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 5.3263947348281925e-06,
+      "loss": 0.6446,
+      "step": 607450
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 5.325246804928446e-06,
+      "loss": 0.7184,
+      "step": 607460
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 5.324098875028697e-06,
+      "loss": 0.7334,
+      "step": 607470
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 5.32295094512895e-06,
+      "loss": 0.6992,
+      "step": 607480
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 5.321803015229203e-06,
+      "loss": 0.6591,
+      "step": 607490
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 5.320655085329455e-06,
+      "loss": 0.698,
+      "step": 607500
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 5.319507155429708e-06,
+      "loss": 0.695,
+      "step": 607510
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 5.318359225529961e-06,
+      "loss": 0.7733,
+      "step": 607520
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 5.317211295630213e-06,
+      "loss": 0.7544,
+      "step": 607530
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 5.316063365730465e-06,
+      "loss": 0.6543,
+      "step": 607540
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 5.314915435830718e-06,
+      "loss": 0.7446,
+      "step": 607550
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 5.3137675059309705e-06,
+      "loss": 0.7364,
+      "step": 607560
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 5.312619576031223e-06,
+      "loss": 0.7515,
+      "step": 607570
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 5.311471646131476e-06,
+      "loss": 0.7081,
+      "step": 607580
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 5.310323716231729e-06,
+      "loss": 0.6768,
+      "step": 607590
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 5.309175786331981e-06,
+      "loss": 0.6908,
+      "step": 607600
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 5.308027856432234e-06,
+      "loss": 0.7332,
+      "step": 607610
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 5.306879926532486e-06,
+      "loss": 0.7091,
+      "step": 607620
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 5.305731996632738e-06,
+      "loss": 0.6881,
+      "step": 607630
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 5.3045840667329906e-06,
+      "loss": 0.6726,
+      "step": 607640
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 5.303436136833244e-06,
+      "loss": 0.6589,
+      "step": 607650
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 5.302288206933496e-06,
+      "loss": 0.6834,
+      "step": 607660
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 5.301140277033749e-06,
+      "loss": 0.6685,
+      "step": 607670
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 5.299992347134002e-06,
+      "loss": 0.7732,
+      "step": 607680
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 5.298844417234253e-06,
+      "loss": 0.6347,
+      "step": 607690
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 5.297696487334506e-06,
+      "loss": 0.6734,
+      "step": 607700
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 5.296548557434759e-06,
+      "loss": 0.7328,
+      "step": 607710
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 5.295400627535011e-06,
+      "loss": 0.6661,
+      "step": 607720
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 5.294252697635264e-06,
+      "loss": 0.7419,
+      "step": 607730
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 5.293104767735517e-06,
+      "loss": 0.7462,
+      "step": 607740
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 5.2919568378357694e-06,
+      "loss": 0.7761,
+      "step": 607750
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 5.290808907936021e-06,
+      "loss": 0.7753,
+      "step": 607760
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 5.289660978036274e-06,
+      "loss": 0.7173,
+      "step": 607770
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 5.288513048136527e-06,
+      "loss": 0.7276,
+      "step": 607780
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 5.287365118236779e-06,
+      "loss": 0.6115,
+      "step": 607790
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 5.2862171883370314e-06,
+      "loss": 0.801,
+      "step": 607800
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 5.285069258437285e-06,
+      "loss": 0.6512,
+      "step": 607810
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 5.283921328537537e-06,
+      "loss": 0.7402,
+      "step": 607820
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 5.282773398637789e-06,
+      "loss": 0.577,
+      "step": 607830
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 5.281625468738042e-06,
+      "loss": 0.7154,
+      "step": 607840
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 5.280477538838294e-06,
+      "loss": 0.6943,
+      "step": 607850
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 5.279329608938547e-06,
+      "loss": 0.7536,
+      "step": 607860
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 5.2781816790388e-06,
+      "loss": 0.7362,
+      "step": 607870
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 5.277033749139052e-06,
+      "loss": 0.6505,
+      "step": 607880
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 5.275885819239305e-06,
+      "loss": 0.7308,
+      "step": 607890
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 5.274737889339558e-06,
+      "loss": 0.6818,
+      "step": 607900
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 5.2735899594398095e-06,
+      "loss": 0.6601,
+      "step": 607910
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 5.272442029540062e-06,
+      "loss": 0.6257,
+      "step": 607920
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 5.271294099640315e-06,
+      "loss": 0.6736,
+      "step": 607930
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 5.2701461697405675e-06,
+      "loss": 0.6524,
+      "step": 607940
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 5.26899823984082e-06,
+      "loss": 0.7016,
+      "step": 607950
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 5.267850309941072e-06,
+      "loss": 0.7523,
+      "step": 607960
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 5.2667023800413255e-06,
+      "loss": 0.6825,
+      "step": 607970
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 5.265554450141577e-06,
+      "loss": 0.6934,
+      "step": 607980
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 5.2644065202418295e-06,
+      "loss": 0.6568,
+      "step": 607990
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 5.263258590342083e-06,
+      "loss": 0.7241,
+      "step": 608000
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 5.262110660442335e-06,
+      "loss": 0.7738,
+      "step": 608010
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 5.2609627305425875e-06,
+      "loss": 0.7093,
+      "step": 608020
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 5.259814800642841e-06,
+      "loss": 0.694,
+      "step": 608030
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 5.258666870743093e-06,
+      "loss": 0.7239,
+      "step": 608040
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 5.257518940843345e-06,
+      "loss": 0.6957,
+      "step": 608050
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 5.256371010943598e-06,
+      "loss": 0.7053,
+      "step": 608060
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 5.25522308104385e-06,
+      "loss": 0.8117,
+      "step": 608070
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 5.254075151144103e-06,
+      "loss": 0.6145,
+      "step": 608080
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 5.252927221244356e-06,
+      "loss": 0.6749,
+      "step": 608090
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 5.251779291344608e-06,
+      "loss": 0.6688,
+      "step": 608100
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 5.250631361444861e-06,
+      "loss": 0.6449,
+      "step": 608110
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 5.249483431545114e-06,
+      "loss": 0.6242,
+      "step": 608120
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 5.2483355016453656e-06,
+      "loss": 0.7057,
+      "step": 608130
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 5.247187571745618e-06,
+      "loss": 0.6772,
+      "step": 608140
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 5.24603964184587e-06,
+      "loss": 0.7756,
+      "step": 608150
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 5.244891711946124e-06,
+      "loss": 0.7764,
+      "step": 608160
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 5.243743782046376e-06,
+      "loss": 0.8162,
+      "step": 608170
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 5.242595852146628e-06,
+      "loss": 0.6773,
+      "step": 608180
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 5.241447922246882e-06,
+      "loss": 0.698,
+      "step": 608190
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 5.240299992347133e-06,
+      "loss": 0.6708,
+      "step": 608200
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 5.2391520624473856e-06,
+      "loss": 0.7127,
+      "step": 608210
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 5.238004132547639e-06,
+      "loss": 0.7208,
+      "step": 608220
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 5.236856202647891e-06,
+      "loss": 0.6796,
+      "step": 608230
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 5.235708272748144e-06,
+      "loss": 0.6715,
+      "step": 608240
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 5.234560342848397e-06,
+      "loss": 0.671,
+      "step": 608250
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 5.233412412948649e-06,
+      "loss": 0.6887,
+      "step": 608260
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 5.232264483048901e-06,
+      "loss": 0.7138,
+      "step": 608270
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 5.231116553149154e-06,
+      "loss": 0.7724,
+      "step": 608280
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 5.229968623249406e-06,
+      "loss": 0.7415,
+      "step": 608290
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 5.228820693349659e-06,
+      "loss": 0.674,
+      "step": 608300
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 5.227672763449911e-06,
+      "loss": 0.7048,
+      "step": 608310
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 5.2265248335501645e-06,
+      "loss": 0.6525,
+      "step": 608320
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 5.225376903650417e-06,
+      "loss": 0.6827,
+      "step": 608330
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 5.224228973750668e-06,
+      "loss": 0.7052,
+      "step": 608340
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 5.223081043850922e-06,
+      "loss": 0.7136,
+      "step": 608350
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 5.221933113951174e-06,
+      "loss": 0.7652,
+      "step": 608360
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 5.2207851840514264e-06,
+      "loss": 0.6626,
+      "step": 608370
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 5.21963725415168e-06,
+      "loss": 0.6342,
+      "step": 608380
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 5.218489324251932e-06,
+      "loss": 0.6795,
+      "step": 608390
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 5.2173413943521845e-06,
+      "loss": 0.5945,
+      "step": 608400
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 5.216193464452438e-06,
+      "loss": 0.6824,
+      "step": 608410
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 5.215045534552689e-06,
+      "loss": 0.6558,
+      "step": 608420
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 5.213897604652942e-06,
+      "loss": 0.8074,
+      "step": 608430
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 5.212749674753195e-06,
+      "loss": 0.7283,
+      "step": 608440
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 5.211601744853447e-06,
+      "loss": 0.6155,
+      "step": 608450
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 5.2104538149537e-06,
+      "loss": 0.6913,
+      "step": 608460
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 5.209305885053952e-06,
+      "loss": 0.6549,
+      "step": 608470
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 5.208157955154205e-06,
+      "loss": 0.6248,
+      "step": 608480
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 5.207010025254457e-06,
+      "loss": 0.6733,
+      "step": 608490
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 5.205862095354709e-06,
+      "loss": 0.7452,
+      "step": 608500
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 5.2047141654549625e-06,
+      "loss": 0.7514,
+      "step": 608510
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 5.203566235555215e-06,
+      "loss": 0.7241,
+      "step": 608520
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 5.202418305655467e-06,
+      "loss": 0.7263,
+      "step": 608530
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 5.2012703757557205e-06,
+      "loss": 0.6692,
+      "step": 608540
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 5.200122445855973e-06,
+      "loss": 0.7563,
+      "step": 608550
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 5.1989745159562245e-06,
+      "loss": 0.7472,
+      "step": 608560
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 5.197826586056478e-06,
+      "loss": 0.616,
+      "step": 608570
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 5.19667865615673e-06,
+      "loss": 0.6931,
+      "step": 608580
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 5.1955307262569825e-06,
+      "loss": 0.6573,
+      "step": 608590
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 5.194382796357236e-06,
+      "loss": 0.7257,
+      "step": 608600
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 5.193234866457488e-06,
+      "loss": 0.6862,
+      "step": 608610
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 5.1920869365577406e-06,
+      "loss": 0.6788,
+      "step": 608620
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 5.190939006657994e-06,
+      "loss": 0.7501,
+      "step": 608630
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 5.189791076758245e-06,
+      "loss": 0.6571,
+      "step": 608640
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 5.188643146858498e-06,
+      "loss": 0.718,
+      "step": 608650
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 5.18749521695875e-06,
+      "loss": 0.7208,
+      "step": 608660
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 5.186347287059003e-06,
+      "loss": 0.7391,
+      "step": 608670
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 5.185199357159256e-06,
+      "loss": 0.7197,
+      "step": 608680
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 5.184051427259508e-06,
+      "loss": 0.7277,
+      "step": 608690
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 5.182903497359761e-06,
+      "loss": 0.6815,
+      "step": 608700
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 5.181755567460013e-06,
+      "loss": 0.7072,
+      "step": 608710
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 5.180607637560265e-06,
+      "loss": 0.6945,
+      "step": 608720
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 5.179459707660519e-06,
+      "loss": 0.6668,
+      "step": 608730
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 5.178311777760771e-06,
+      "loss": 0.7309,
+      "step": 608740
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 5.177163847861023e-06,
+      "loss": 0.7478,
+      "step": 608750
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 5.176015917961277e-06,
+      "loss": 0.708,
+      "step": 608760
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 5.174867988061529e-06,
+      "loss": 0.7389,
+      "step": 608770
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 5.1737200581617806e-06,
+      "loss": 0.7576,
+      "step": 608780
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 5.172572128262034e-06,
+      "loss": 0.7237,
+      "step": 608790
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 5.171424198362286e-06,
+      "loss": 0.6785,
+      "step": 608800
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 5.170276268462539e-06,
+      "loss": 0.6531,
+      "step": 608810
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 5.169128338562791e-06,
+      "loss": 0.6803,
+      "step": 608820
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 5.167980408663044e-06,
+      "loss": 0.58,
+      "step": 608830
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 5.166832478763297e-06,
+      "loss": 0.7194,
+      "step": 608840
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 5.165684548863548e-06,
+      "loss": 0.66,
+      "step": 608850
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 5.1645366189638014e-06,
+      "loss": 0.8029,
+      "step": 608860
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 5.163388689064054e-06,
+      "loss": 0.6914,
+      "step": 608870
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 5.162240759164306e-06,
+      "loss": 0.6933,
+      "step": 608880
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 5.1610928292645595e-06,
+      "loss": 0.7587,
+      "step": 608890
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 5.159944899364812e-06,
+      "loss": 0.6362,
+      "step": 608900
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 5.158796969465064e-06,
+      "loss": 0.6553,
+      "step": 608910
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 5.1576490395653175e-06,
+      "loss": 0.7731,
+      "step": 608920
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 5.156501109665569e-06,
+      "loss": 0.6502,
+      "step": 608930
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 5.1553531797658214e-06,
+      "loss": 0.7187,
+      "step": 608940
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 5.154205249866075e-06,
+      "loss": 0.7634,
+      "step": 608950
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 5.153057319966327e-06,
+      "loss": 0.667,
+      "step": 608960
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 5.1519093900665795e-06,
+      "loss": 0.6648,
+      "step": 608970
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 5.150761460166833e-06,
+      "loss": 0.7139,
+      "step": 608980
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 5.149613530267085e-06,
+      "loss": 0.6906,
+      "step": 608990
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 5.148465600367337e-06,
+      "loss": 0.6535,
+      "step": 609000
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 5.147317670467589e-06,
+      "loss": 0.7162,
+      "step": 609010
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 5.146169740567842e-06,
+      "loss": 0.7205,
+      "step": 609020
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 5.145021810668095e-06,
+      "loss": 0.6722,
+      "step": 609030
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 5.143873880768347e-06,
+      "loss": 0.7573,
+      "step": 609040
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 5.1427259508686e-06,
+      "loss": 0.6971,
+      "step": 609050
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 5.141578020968853e-06,
+      "loss": 0.6509,
+      "step": 609060
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 5.140430091069104e-06,
+      "loss": 0.6887,
+      "step": 609070
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 5.1392821611693575e-06,
+      "loss": 0.7154,
+      "step": 609080
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 5.13813423126961e-06,
+      "loss": 0.6559,
+      "step": 609090
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 5.136986301369862e-06,
+      "loss": 0.7289,
+      "step": 609100
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 5.1358383714701156e-06,
+      "loss": 0.6977,
+      "step": 609110
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 5.134690441570368e-06,
+      "loss": 0.7164,
+      "step": 609120
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 5.13354251167062e-06,
+      "loss": 0.6983,
+      "step": 609130
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 5.132394581770874e-06,
+      "loss": 0.6895,
+      "step": 609140
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 5.131246651871125e-06,
+      "loss": 0.6279,
+      "step": 609150
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 5.1300987219713775e-06,
+      "loss": 0.6052,
+      "step": 609160
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 5.12895079207163e-06,
+      "loss": 0.6546,
+      "step": 609170
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 5.127802862171883e-06,
+      "loss": 0.6851,
+      "step": 609180
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 5.1266549322721356e-06,
+      "loss": 0.7365,
+      "step": 609190
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 5.125507002372388e-06,
+      "loss": 0.6905,
+      "step": 609200
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 5.124359072472641e-06,
+      "loss": 0.7214,
+      "step": 609210
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 5.123211142572893e-06,
+      "loss": 0.6742,
+      "step": 609220
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 5.122063212673145e-06,
+      "loss": 0.6486,
+      "step": 609230
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 5.120915282773398e-06,
+      "loss": 0.7451,
+      "step": 609240
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 5.119767352873651e-06,
+      "loss": 0.6654,
+      "step": 609250
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 5.118619422973903e-06,
+      "loss": 0.7274,
+      "step": 609260
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 5.117471493074156e-06,
+      "loss": 0.6896,
+      "step": 609270
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 5.116323563174409e-06,
+      "loss": 0.7627,
+      "step": 609280
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 5.11517563327466e-06,
+      "loss": 0.65,
+      "step": 609290
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 5.114027703374914e-06,
+      "loss": 0.6885,
+      "step": 609300
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 5.112879773475166e-06,
+      "loss": 0.6901,
+      "step": 609310
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 5.111731843575418e-06,
+      "loss": 0.6639,
+      "step": 609320
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 5.110583913675671e-06,
+      "loss": 0.6725,
+      "step": 609330
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 5.109435983775924e-06,
+      "loss": 0.561,
+      "step": 609340
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 5.1082880538761764e-06,
+      "loss": 0.6455,
+      "step": 609350
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 5.107140123976428e-06,
+      "loss": 0.6485,
+      "step": 609360
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 5.105992194076681e-06,
+      "loss": 0.6376,
+      "step": 609370
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 5.104844264176934e-06,
+      "loss": 0.6367,
+      "step": 609380
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 5.103696334277186e-06,
+      "loss": 0.7457,
+      "step": 609390
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 5.102548404377439e-06,
+      "loss": 0.7167,
+      "step": 609400
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 5.101400474477692e-06,
+      "loss": 0.617,
+      "step": 609410
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 5.100252544577944e-06,
+      "loss": 0.6687,
+      "step": 609420
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 5.099104614678197e-06,
+      "loss": 0.6999,
+      "step": 609430
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 5.097956684778449e-06,
+      "loss": 0.707,
+      "step": 609440
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 5.096808754878701e-06,
+      "loss": 0.7978,
+      "step": 609450
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 5.0956608249789545e-06,
+      "loss": 0.7436,
+      "step": 609460
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 5.094512895079207e-06,
+      "loss": 0.7272,
+      "step": 609470
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 5.093364965179459e-06,
+      "loss": 0.7807,
+      "step": 609480
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 5.0922170352797125e-06,
+      "loss": 0.6621,
+      "step": 609490
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 5.091069105379965e-06,
+      "loss": 0.7042,
+      "step": 609500
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 5.0899211754802164e-06,
+      "loss": 0.5602,
+      "step": 609510
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 5.088773245580469e-06,
+      "loss": 0.7425,
+      "step": 609520
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 5.087625315680722e-06,
+      "loss": 0.6559,
+      "step": 609530
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 5.0864773857809745e-06,
+      "loss": 0.7393,
+      "step": 609540
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 5.085329455881227e-06,
+      "loss": 0.6542,
+      "step": 609550
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 5.08418152598148e-06,
+      "loss": 0.6669,
+      "step": 609560
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 5.0830335960817325e-06,
+      "loss": 0.7421,
+      "step": 609570
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 5.081885666181984e-06,
+      "loss": 0.7806,
+      "step": 609580
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 5.080737736282237e-06,
+      "loss": 0.6529,
+      "step": 609590
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 5.07958980638249e-06,
+      "loss": 0.729,
+      "step": 609600
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 5.078441876482742e-06,
+      "loss": 0.6629,
+      "step": 609610
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 5.077293946582995e-06,
+      "loss": 0.7236,
+      "step": 609620
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 5.076146016683248e-06,
+      "loss": 0.7097,
+      "step": 609630
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 5.0749980867835e-06,
+      "loss": 0.6877,
+      "step": 609640
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 5.073850156883753e-06,
+      "loss": 0.7258,
+      "step": 609650
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 5.072702226984005e-06,
+      "loss": 0.6766,
+      "step": 609660
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 5.071554297084257e-06,
+      "loss": 0.7051,
+      "step": 609670
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 5.07040636718451e-06,
+      "loss": 0.6459,
+      "step": 609680
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 5.069258437284763e-06,
+      "loss": 0.6967,
+      "step": 609690
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 5.068110507385015e-06,
+      "loss": 0.6729,
+      "step": 609700
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 5.066962577485268e-06,
+      "loss": 0.709,
+      "step": 609710
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 5.065814647585521e-06,
+      "loss": 0.5612,
+      "step": 609720
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 5.0646667176857725e-06,
+      "loss": 0.6937,
+      "step": 609730
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 5.063518787786025e-06,
+      "loss": 0.6479,
+      "step": 609740
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 5.062370857886278e-06,
+      "loss": 0.7239,
+      "step": 609750
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 5.0612229279865306e-06,
+      "loss": 0.6705,
+      "step": 609760
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 5.060074998086783e-06,
+      "loss": 0.6756,
+      "step": 609770
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 5.058927068187036e-06,
+      "loss": 0.6965,
+      "step": 609780
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 5.057779138287289e-06,
+      "loss": 0.7038,
+      "step": 609790
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 5.05663120838754e-06,
+      "loss": 0.7674,
+      "step": 609800
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 5.055483278487793e-06,
+      "loss": 0.7831,
+      "step": 609810
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 5.054335348588046e-06,
+      "loss": 0.7218,
+      "step": 609820
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 5.053187418688298e-06,
+      "loss": 0.6356,
+      "step": 609830
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 5.0520394887885514e-06,
+      "loss": 0.7035,
+      "step": 609840
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 5.050891558888804e-06,
+      "loss": 0.6631,
+      "step": 609850
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 5.049743628989056e-06,
+      "loss": 0.7094,
+      "step": 609860
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 5.048595699089308e-06,
+      "loss": 0.749,
+      "step": 609870
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 5.047447769189561e-06,
+      "loss": 0.8079,
+      "step": 609880
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 5.046299839289813e-06,
+      "loss": 0.6773,
+      "step": 609890
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 5.045151909390066e-06,
+      "loss": 0.7562,
+      "step": 609900
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 5.044003979490319e-06,
+      "loss": 0.603,
+      "step": 609910
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 5.0428560495905714e-06,
+      "loss": 0.6744,
+      "step": 609920
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 5.041708119690824e-06,
+      "loss": 0.7103,
+      "step": 609930
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 5.040560189791077e-06,
+      "loss": 0.7209,
+      "step": 609940
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 5.039412259891329e-06,
+      "loss": 0.7418,
+      "step": 609950
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 5.038264329991581e-06,
+      "loss": 0.6324,
+      "step": 609960
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 5.037116400091834e-06,
+      "loss": 0.7027,
+      "step": 609970
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 5.035968470192087e-06,
+      "loss": 0.6772,
+      "step": 609980
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 5.034820540292339e-06,
+      "loss": 0.6756,
+      "step": 609990
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 5.033672610392592e-06,
+      "loss": 0.6562,
+      "step": 610000
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 5.032524680492845e-06,
+      "loss": 0.6284,
+      "step": 610010
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 5.031376750593096e-06,
+      "loss": 0.7298,
+      "step": 610020
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 5.030228820693349e-06,
+      "loss": 0.6551,
+      "step": 610030
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 5.029080890793602e-06,
+      "loss": 0.648,
+      "step": 610040
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 5.027932960893854e-06,
+      "loss": 0.6296,
+      "step": 610050
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 5.026785030994107e-06,
+      "loss": 0.7108,
+      "step": 610060
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 5.02563710109436e-06,
+      "loss": 0.6738,
+      "step": 610070
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 5.024489171194612e-06,
+      "loss": 0.6558,
+      "step": 610080
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 5.023341241294864e-06,
+      "loss": 0.7337,
+      "step": 610090
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 5.022193311395117e-06,
+      "loss": 0.7345,
+      "step": 610100
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 5.0210453814953695e-06,
+      "loss": 0.6451,
+      "step": 610110
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 5.019897451595622e-06,
+      "loss": 0.7991,
+      "step": 610120
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 5.018749521695875e-06,
+      "loss": 0.6738,
+      "step": 610130
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 5.0176015917961275e-06,
+      "loss": 0.7219,
+      "step": 610140
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 5.01645366189638e-06,
+      "loss": 0.6311,
+      "step": 610150
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 5.015305731996633e-06,
+      "loss": 0.6893,
+      "step": 610160
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 5.014157802096885e-06,
+      "loss": 0.6368,
+      "step": 610170
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 5.013009872197137e-06,
+      "loss": 0.6563,
+      "step": 610180
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 5.0118619422973895e-06,
+      "loss": 0.6929,
+      "step": 610190
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 5.010714012397643e-06,
+      "loss": 0.7113,
+      "step": 610200
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 5.009566082497895e-06,
+      "loss": 0.7196,
+      "step": 610210
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 5.0084181525981475e-06,
+      "loss": 0.7897,
+      "step": 610220
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 5.007270222698401e-06,
+      "loss": 0.7017,
+      "step": 610230
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 5.006122292798652e-06,
+      "loss": 0.7166,
+      "step": 610240
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 5.004974362898905e-06,
+      "loss": 0.7141,
+      "step": 610250
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 5.003826432999158e-06,
+      "loss": 0.6398,
+      "step": 610260
+    },
+    {
+      "epoch": 28.0,
+      "eval_cer": 0.6688899026728322,
+      "eval_em": 0.007357212003872217,
+      "eval_f1": 0.007357212003872217,
+      "eval_loss": 1.3739546537399292,
+      "eval_runtime": 1295.1545,
+      "eval_samples_per_second": 7.976,
+      "eval_steps_per_second": 0.998,
+      "eval_wer": 0.9926427879961278,
+      "step": 610268
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 5.00267850309941e-06,
+      "loss": 0.7052,
+      "step": 610270
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 5.001530573199663e-06,
+      "loss": 0.6289,
+      "step": 610280
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 5.000382643299916e-06,
+      "loss": 0.7626,
+      "step": 610290
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 4.999234713400168e-06,
+      "loss": 0.6617,
+      "step": 610300
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 4.99808678350042e-06,
+      "loss": 0.6697,
+      "step": 610310
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 4.996938853600673e-06,
+      "loss": 0.5867,
+      "step": 610320
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 4.9957909237009256e-06,
+      "loss": 0.7206,
+      "step": 610330
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 4.994642993801178e-06,
+      "loss": 0.7631,
+      "step": 610340
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 4.993495063901431e-06,
+      "loss": 0.6567,
+      "step": 610350
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 4.992347134001684e-06,
+      "loss": 0.6639,
+      "step": 610360
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 4.991199204101936e-06,
+      "loss": 0.675,
+      "step": 610370
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 4.9900512742021876e-06,
+      "loss": 0.7022,
+      "step": 610380
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 4.988903344302441e-06,
+      "loss": 0.7216,
+      "step": 610390
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 4.987755414402693e-06,
+      "loss": 0.6935,
+      "step": 610400
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 4.986607484502946e-06,
+      "loss": 0.6461,
+      "step": 610410
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 4.985459554603199e-06,
+      "loss": 0.6904,
+      "step": 610420
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 4.984311624703451e-06,
+      "loss": 0.7157,
+      "step": 610430
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 4.983163694803704e-06,
+      "loss": 0.6814,
+      "step": 610440
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 4.982015764903957e-06,
+      "loss": 0.7547,
+      "step": 610450
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 4.980867835004208e-06,
+      "loss": 0.6744,
+      "step": 610460
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 4.979719905104461e-06,
+      "loss": 0.6833,
+      "step": 610470
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 4.978571975204714e-06,
+      "loss": 0.6062,
+      "step": 610480
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 4.9774240453049664e-06,
+      "loss": 0.7059,
+      "step": 610490
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 4.976276115405219e-06,
+      "loss": 0.7455,
+      "step": 610500
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 4.975128185505472e-06,
+      "loss": 0.6942,
+      "step": 610510
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 4.9739802556057245e-06,
+      "loss": 0.6812,
+      "step": 610520
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 4.972832325705976e-06,
+      "loss": 0.641,
+      "step": 610530
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 4.971684395806228e-06,
+      "loss": 0.6438,
+      "step": 610540
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 4.970536465906482e-06,
+      "loss": 0.6888,
+      "step": 610550
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 4.969388536006734e-06,
+      "loss": 0.6638,
+      "step": 610560
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 4.9682406061069865e-06,
+      "loss": 0.7323,
+      "step": 610570
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 4.96709267620724e-06,
+      "loss": 0.6611,
+      "step": 610580
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 4.965944746307492e-06,
+      "loss": 0.6741,
+      "step": 610590
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 4.964796816407744e-06,
+      "loss": 0.5671,
+      "step": 610600
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 4.963648886507997e-06,
+      "loss": 0.6688,
+      "step": 610610
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 4.962500956608249e-06,
+      "loss": 0.6866,
+      "step": 610620
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 4.961353026708502e-06,
+      "loss": 0.6429,
+      "step": 610630
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 4.960205096808755e-06,
+      "loss": 0.6692,
+      "step": 610640
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 4.959057166909007e-06,
+      "loss": 0.6464,
+      "step": 610650
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 4.957909237009259e-06,
+      "loss": 0.685,
+      "step": 610660
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 4.956761307109513e-06,
+      "loss": 0.6695,
+      "step": 610670
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 4.9556133772097645e-06,
+      "loss": 0.7104,
+      "step": 610680
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 4.954465447310017e-06,
+      "loss": 0.702,
+      "step": 610690
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 4.953317517410269e-06,
+      "loss": 0.7509,
+      "step": 610700
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 4.9521695875105225e-06,
+      "loss": 0.6959,
+      "step": 610710
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 4.951021657610775e-06,
+      "loss": 0.6897,
+      "step": 610720
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 4.9498737277110265e-06,
+      "loss": 0.6245,
+      "step": 610730
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 4.9487257978112806e-06,
+      "loss": 0.6604,
+      "step": 610740
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 4.947577867911532e-06,
+      "loss": 0.7151,
+      "step": 610750
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 4.9464299380117845e-06,
+      "loss": 0.632,
+      "step": 610760
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 4.945282008112038e-06,
+      "loss": 0.7302,
+      "step": 610770
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 4.94413407821229e-06,
+      "loss": 0.6933,
+      "step": 610780
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 4.9429861483125425e-06,
+      "loss": 0.6787,
+      "step": 610790
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 4.941838218412796e-06,
+      "loss": 0.7566,
+      "step": 610800
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 4.940690288513047e-06,
+      "loss": 0.7005,
+      "step": 610810
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 4.9395423586133e-06,
+      "loss": 0.7348,
+      "step": 610820
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 4.938394428713553e-06,
+      "loss": 0.7178,
+      "step": 610830
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 4.937246498813805e-06,
+      "loss": 0.6728,
+      "step": 610840
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 4.936098568914058e-06,
+      "loss": 0.6537,
+      "step": 610850
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 4.934950639014311e-06,
+      "loss": 0.7319,
+      "step": 610860
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 4.933802709114563e-06,
+      "loss": 0.6711,
+      "step": 610870
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 4.932654779214815e-06,
+      "loss": 0.6418,
+      "step": 610880
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 4.931506849315067e-06,
+      "loss": 0.7286,
+      "step": 610890
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 4.930358919415321e-06,
+      "loss": 0.7086,
+      "step": 610900
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 4.929210989515573e-06,
+      "loss": 0.7894,
+      "step": 610910
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 4.928063059615825e-06,
+      "loss": 0.6879,
+      "step": 610920
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 4.926915129716079e-06,
+      "loss": 0.6306,
+      "step": 610930
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 4.925767199816331e-06,
+      "loss": 0.714,
+      "step": 610940
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 4.9246192699165826e-06,
+      "loss": 0.6914,
+      "step": 610950
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 4.923471340016837e-06,
+      "loss": 0.6377,
+      "step": 610960
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 4.922323410117088e-06,
+      "loss": 0.6561,
+      "step": 610970
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 4.921175480217341e-06,
+      "loss": 0.6769,
+      "step": 610980
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 4.920027550317594e-06,
+      "loss": 0.6891,
+      "step": 610990
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 4.918879620417846e-06,
+      "loss": 0.6556,
+      "step": 611000
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 4.917731690518099e-06,
+      "loss": 0.6899,
+      "step": 611010
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 4.916583760618352e-06,
+      "loss": 0.6607,
+      "step": 611020
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 4.915435830718603e-06,
+      "loss": 0.69,
+      "step": 611030
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 4.914287900818856e-06,
+      "loss": 0.7007,
+      "step": 611040
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 4.913139970919108e-06,
+      "loss": 0.6494,
+      "step": 611050
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 4.9119920410193615e-06,
+      "loss": 0.6832,
+      "step": 611060
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 4.910844111119614e-06,
+      "loss": 0.679,
+      "step": 611070
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 4.909696181219866e-06,
+      "loss": 0.7953,
+      "step": 611080
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 4.9085482513201195e-06,
+      "loss": 0.6871,
+      "step": 611090
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 4.907400321420371e-06,
+      "loss": 0.6259,
+      "step": 611100
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 4.9062523915206234e-06,
+      "loss": 0.6524,
+      "step": 611110
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 4.905104461620877e-06,
+      "loss": 0.6657,
+      "step": 611120
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 4.903956531721129e-06,
+      "loss": 0.6866,
+      "step": 611130
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 4.9028086018213815e-06,
+      "loss": 0.6163,
+      "step": 611140
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 4.901660671921635e-06,
+      "loss": 0.6704,
+      "step": 611150
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 4.900512742021887e-06,
+      "loss": 0.7915,
+      "step": 611160
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 4.899364812122139e-06,
+      "loss": 0.7953,
+      "step": 611170
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 4.898216882222392e-06,
+      "loss": 0.7282,
+      "step": 611180
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 4.897068952322644e-06,
+      "loss": 0.536,
+      "step": 611190
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 4.895921022422897e-06,
+      "loss": 0.7913,
+      "step": 611200
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 4.89477309252315e-06,
+      "loss": 0.6473,
+      "step": 611210
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 4.893625162623402e-06,
+      "loss": 0.7065,
+      "step": 611220
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 4.892477232723655e-06,
+      "loss": 0.6989,
+      "step": 611230
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 4.891329302823906e-06,
+      "loss": 0.7511,
+      "step": 611240
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 4.8901813729241595e-06,
+      "loss": 0.6431,
+      "step": 611250
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 4.889033443024412e-06,
+      "loss": 0.7155,
+      "step": 611260
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 4.887885513124664e-06,
+      "loss": 0.7557,
+      "step": 611270
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 4.8867375832249175e-06,
+      "loss": 0.6861,
+      "step": 611280
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 4.88558965332517e-06,
+      "loss": 0.7152,
+      "step": 611290
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 4.884441723425422e-06,
+      "loss": 0.7416,
+      "step": 611300
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 4.8832937935256756e-06,
+      "loss": 0.6827,
+      "step": 611310
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 4.882145863625927e-06,
+      "loss": 0.7097,
+      "step": 611320
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 4.8809979337261795e-06,
+      "loss": 0.6938,
+      "step": 611330
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 4.879850003826433e-06,
+      "loss": 0.6608,
+      "step": 611340
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 4.878702073926685e-06,
+      "loss": 0.7425,
+      "step": 611350
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 4.8775541440269375e-06,
+      "loss": 0.6873,
+      "step": 611360
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 4.876406214127191e-06,
+      "loss": 0.6479,
+      "step": 611370
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 4.875258284227443e-06,
+      "loss": 0.589,
+      "step": 611380
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 4.874110354327695e-06,
+      "loss": 0.716,
+      "step": 611390
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 4.872962424427947e-06,
+      "loss": 0.6648,
+      "step": 611400
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 4.8718144945282e-06,
+      "loss": 0.7489,
+      "step": 611410
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 4.870666564628453e-06,
+      "loss": 0.6523,
+      "step": 611420
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 4.869518634728705e-06,
+      "loss": 0.6697,
+      "step": 611430
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 4.868370704828958e-06,
+      "loss": 0.6953,
+      "step": 611440
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 4.867222774929211e-06,
+      "loss": 0.7168,
+      "step": 611450
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 4.866074845029462e-06,
+      "loss": 0.705,
+      "step": 611460
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 4.864926915129716e-06,
+      "loss": 0.713,
+      "step": 611470
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 4.863778985229968e-06,
+      "loss": 0.6721,
+      "step": 611480
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 4.86263105533022e-06,
+      "loss": 0.669,
+      "step": 611490
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 4.861483125430474e-06,
+      "loss": 0.622,
+      "step": 611500
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 4.860335195530726e-06,
+      "loss": 0.6936,
+      "step": 611510
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 4.859187265630978e-06,
+      "loss": 0.6186,
+      "step": 611520
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 4.858039335731232e-06,
+      "loss": 0.7634,
+      "step": 611530
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 4.856891405831483e-06,
+      "loss": 0.6853,
+      "step": 611540
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 4.855743475931736e-06,
+      "loss": 0.6617,
+      "step": 611550
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 4.854595546031988e-06,
+      "loss": 0.6886,
+      "step": 611560
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 4.853447616132241e-06,
+      "loss": 0.7223,
+      "step": 611570
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 4.852299686232494e-06,
+      "loss": 0.7447,
+      "step": 611580
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 4.851151756332746e-06,
+      "loss": 0.6634,
+      "step": 611590
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 4.850003826432999e-06,
+      "loss": 0.6669,
+      "step": 611600
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 4.848855896533251e-06,
+      "loss": 0.7574,
+      "step": 611610
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 4.847707966633503e-06,
+      "loss": 0.6246,
+      "step": 611620
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 4.8465600367337565e-06,
+      "loss": 0.6867,
+      "step": 611630
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 4.845412106834009e-06,
+      "loss": 0.6622,
+      "step": 611640
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 4.844264176934261e-06,
+      "loss": 0.692,
+      "step": 611650
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 4.8431162470345145e-06,
+      "loss": 0.6997,
+      "step": 611660
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 4.841968317134767e-06,
+      "loss": 0.6835,
+      "step": 611670
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 4.8408203872350184e-06,
+      "loss": 0.6939,
+      "step": 611680
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 4.839672457335272e-06,
+      "loss": 0.6789,
+      "step": 611690
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 4.838524527435524e-06,
+      "loss": 0.6458,
+      "step": 611700
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 4.8373765975357765e-06,
+      "loss": 0.7088,
+      "step": 611710
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 4.83622866763603e-06,
+      "loss": 0.7375,
+      "step": 611720
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 4.835080737736282e-06,
+      "loss": 0.6959,
+      "step": 611730
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 4.8339328078365345e-06,
+      "loss": 0.7734,
+      "step": 611740
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 4.832784877936786e-06,
+      "loss": 0.7001,
+      "step": 611750
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 4.831636948037039e-06,
+      "loss": 0.7387,
+      "step": 611760
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 4.830489018137292e-06,
+      "loss": 0.6848,
+      "step": 611770
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 4.829341088237544e-06,
+      "loss": 0.6899,
+      "step": 611780
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 4.828193158337797e-06,
+      "loss": 0.6524,
+      "step": 611790
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 4.82704522843805e-06,
+      "loss": 0.6932,
+      "step": 611800
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 4.825897298538302e-06,
+      "loss": 0.6771,
+      "step": 611810
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 4.824749368638555e-06,
+      "loss": 0.6181,
+      "step": 611820
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 4.823601438738807e-06,
+      "loss": 0.7254,
+      "step": 611830
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 4.822453508839059e-06,
+      "loss": 0.7409,
+      "step": 611840
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 4.8213055789393125e-06,
+      "loss": 0.6233,
+      "step": 611850
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 4.820157649039565e-06,
+      "loss": 0.7361,
+      "step": 611860
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 4.819009719139817e-06,
+      "loss": 0.7019,
+      "step": 611870
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 4.817861789240071e-06,
+      "loss": 0.686,
+      "step": 611880
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 4.816713859340323e-06,
+      "loss": 0.6973,
+      "step": 611890
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 4.8155659294405745e-06,
+      "loss": 0.6468,
+      "step": 611900
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 4.814417999540827e-06,
+      "loss": 0.7092,
+      "step": 611910
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 4.81327006964108e-06,
+      "loss": 0.6193,
+      "step": 611920
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 4.8121221397413326e-06,
+      "loss": 0.6441,
+      "step": 611930
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 4.810974209841585e-06,
+      "loss": 0.6062,
+      "step": 611940
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 4.809826279941838e-06,
+      "loss": 0.6947,
+      "step": 611950
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 4.808678350042091e-06,
+      "loss": 0.6367,
+      "step": 611960
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 4.807530420142342e-06,
+      "loss": 0.6895,
+      "step": 611970
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 4.806382490242595e-06,
+      "loss": 0.6901,
+      "step": 611980
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 4.805234560342848e-06,
+      "loss": 0.7553,
+      "step": 611990
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 4.8040866304431e-06,
+      "loss": 0.6502,
+      "step": 612000
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 4.802938700543353e-06,
+      "loss": 0.6281,
+      "step": 612010
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 4.801790770643606e-06,
+      "loss": 0.6895,
+      "step": 612020
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 4.800642840743858e-06,
+      "loss": 0.6871,
+      "step": 612030
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 4.7994949108441114e-06,
+      "loss": 0.7143,
+      "step": 612040
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 4.798346980944363e-06,
+      "loss": 0.6564,
+      "step": 612050
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 4.797199051044615e-06,
+      "loss": 0.6436,
+      "step": 612060
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 4.796051121144869e-06,
+      "loss": 0.6809,
+      "step": 612070
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 4.794903191245121e-06,
+      "loss": 0.6017,
+      "step": 612080
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 4.7937552613453734e-06,
+      "loss": 0.6339,
+      "step": 612090
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 4.792607331445626e-06,
+      "loss": 0.6567,
+      "step": 612100
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 4.791459401545879e-06,
+      "loss": 0.6773,
+      "step": 612110
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 4.790311471646131e-06,
+      "loss": 0.7479,
+      "step": 612120
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 4.789163541746383e-06,
+      "loss": 0.7214,
+      "step": 612130
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 4.788015611846636e-06,
+      "loss": 0.7089,
+      "step": 612140
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 4.786867681946889e-06,
+      "loss": 0.719,
+      "step": 612150
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 4.785719752047141e-06,
+      "loss": 0.7255,
+      "step": 612160
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 4.784571822147394e-06,
+      "loss": 0.688,
+      "step": 612170
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 4.783423892247647e-06,
+      "loss": 0.6788,
+      "step": 612180
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 4.782275962347898e-06,
+      "loss": 0.7484,
+      "step": 612190
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 4.7811280324481515e-06,
+      "loss": 0.6708,
+      "step": 612200
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 4.779980102548404e-06,
+      "loss": 0.7267,
+      "step": 612210
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 4.778832172648656e-06,
+      "loss": 0.6737,
+      "step": 612220
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 4.7776842427489095e-06,
+      "loss": 0.6364,
+      "step": 612230
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 4.776536312849162e-06,
+      "loss": 0.6829,
+      "step": 612240
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 4.775388382949414e-06,
+      "loss": 0.6842,
+      "step": 612250
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 4.774240453049666e-06,
+      "loss": 0.719,
+      "step": 612260
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 4.773092523149919e-06,
+      "loss": 0.7208,
+      "step": 612270
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 4.7719445932501715e-06,
+      "loss": 0.6644,
+      "step": 612280
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 4.770796663350424e-06,
+      "loss": 0.6883,
+      "step": 612290
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 4.769648733450677e-06,
+      "loss": 0.6678,
+      "step": 612300
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 4.7685008035509295e-06,
+      "loss": 0.7313,
+      "step": 612310
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 4.767352873651182e-06,
+      "loss": 0.6807,
+      "step": 612320
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 4.766204943751435e-06,
+      "loss": 0.6759,
+      "step": 612330
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 4.765057013851687e-06,
+      "loss": 0.6768,
+      "step": 612340
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 4.763909083951939e-06,
+      "loss": 0.7128,
+      "step": 612350
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 4.762761154052192e-06,
+      "loss": 0.6394,
+      "step": 612360
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 4.761613224152445e-06,
+      "loss": 0.6915,
+      "step": 612370
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 4.760465294252697e-06,
+      "loss": 0.6468,
+      "step": 612380
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 4.75931736435295e-06,
+      "loss": 0.6794,
+      "step": 612390
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 4.758169434453203e-06,
+      "loss": 0.7768,
+      "step": 612400
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 4.757021504553454e-06,
+      "loss": 0.6301,
+      "step": 612410
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 4.755873574653707e-06,
+      "loss": 0.7326,
+      "step": 612420
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 4.75472564475396e-06,
+      "loss": 0.744,
+      "step": 612430
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 4.753577714854212e-06,
+      "loss": 0.6837,
+      "step": 612440
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 4.752429784954465e-06,
+      "loss": 0.6988,
+      "step": 612450
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 4.751281855054718e-06,
+      "loss": 0.6705,
+      "step": 612460
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 4.75013392515497e-06,
+      "loss": 0.691,
+      "step": 612470
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 4.748985995255222e-06,
+      "loss": 0.6553,
+      "step": 612480
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 4.747838065355475e-06,
+      "loss": 0.7119,
+      "step": 612490
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 4.7466901354557276e-06,
+      "loss": 0.6488,
+      "step": 612500
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 4.74554220555598e-06,
+      "loss": 0.7022,
+      "step": 612510
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 4.744394275656233e-06,
+      "loss": 0.6575,
+      "step": 612520
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 4.743246345756486e-06,
+      "loss": 0.6617,
+      "step": 612530
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 4.742098415856738e-06,
+      "loss": 0.6951,
+      "step": 612540
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 4.740950485956991e-06,
+      "loss": 0.771,
+      "step": 612550
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 4.739802556057243e-06,
+      "loss": 0.7489,
+      "step": 612560
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 4.738654626157495e-06,
+      "loss": 0.7086,
+      "step": 612570
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 4.737506696257748e-06,
+      "loss": 0.6539,
+      "step": 612580
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 4.736358766358001e-06,
+      "loss": 0.7343,
+      "step": 612590
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 4.735210836458253e-06,
+      "loss": 0.6291,
+      "step": 612600
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 4.734062906558506e-06,
+      "loss": 0.6831,
+      "step": 612610
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 4.732914976658759e-06,
+      "loss": 0.5959,
+      "step": 612620
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 4.73176704675901e-06,
+      "loss": 0.7338,
+      "step": 612630
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 4.730619116859263e-06,
+      "loss": 0.6675,
+      "step": 612640
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 4.729471186959516e-06,
+      "loss": 0.6442,
+      "step": 612650
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 4.7283232570597684e-06,
+      "loss": 0.6898,
+      "step": 612660
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 4.727175327160021e-06,
+      "loss": 0.6616,
+      "step": 612670
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 4.726027397260274e-06,
+      "loss": 0.7176,
+      "step": 612680
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 4.7248794673605265e-06,
+      "loss": 0.6479,
+      "step": 612690
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 4.723731537460778e-06,
+      "loss": 0.7131,
+      "step": 612700
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 4.722583607561031e-06,
+      "loss": 0.6778,
+      "step": 612710
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 4.721435677661284e-06,
+      "loss": 0.6687,
+      "step": 612720
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 4.720287747761536e-06,
+      "loss": 0.7249,
+      "step": 612730
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 4.719139817861789e-06,
+      "loss": 0.672,
+      "step": 612740
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 4.717991887962042e-06,
+      "loss": 0.7048,
+      "step": 612750
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 4.716843958062294e-06,
+      "loss": 0.6982,
+      "step": 612760
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 4.715696028162546e-06,
+      "loss": 0.5847,
+      "step": 612770
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 4.714548098262799e-06,
+      "loss": 0.7041,
+      "step": 612780
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 4.713400168363051e-06,
+      "loss": 0.6176,
+      "step": 612790
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 4.712252238463304e-06,
+      "loss": 0.6479,
+      "step": 612800
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 4.711104308563557e-06,
+      "loss": 0.7647,
+      "step": 612810
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 4.709956378663809e-06,
+      "loss": 0.6273,
+      "step": 612820
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 4.708808448764062e-06,
+      "loss": 0.6749,
+      "step": 612830
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 4.707660518864315e-06,
+      "loss": 0.6908,
+      "step": 612840
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 4.7065125889645665e-06,
+      "loss": 0.7404,
+      "step": 612850
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 4.705364659064819e-06,
+      "loss": 0.7153,
+      "step": 612860
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 4.704216729165072e-06,
+      "loss": 0.6703,
+      "step": 612870
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 4.7030687992653245e-06,
+      "loss": 0.7218,
+      "step": 612880
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 4.701920869365577e-06,
+      "loss": 0.7018,
+      "step": 612890
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 4.70077293946583e-06,
+      "loss": 0.7251,
+      "step": 612900
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 4.6996250095660826e-06,
+      "loss": 0.6502,
+      "step": 612910
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 4.698477079666334e-06,
+      "loss": 0.7169,
+      "step": 612920
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 4.6973291497665865e-06,
+      "loss": 0.6853,
+      "step": 612930
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 4.69618121986684e-06,
+      "loss": 0.6535,
+      "step": 612940
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 4.695033289967092e-06,
+      "loss": 0.6764,
+      "step": 612950
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 4.6938853600673445e-06,
+      "loss": 0.6344,
+      "step": 612960
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 4.692737430167598e-06,
+      "loss": 0.6874,
+      "step": 612970
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 4.69158950026785e-06,
+      "loss": 0.653,
+      "step": 612980
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 4.690441570368102e-06,
+      "loss": 0.6452,
+      "step": 612990
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 4.689293640468355e-06,
+      "loss": 0.6512,
+      "step": 613000
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 4.688145710568607e-06,
+      "loss": 0.6344,
+      "step": 613010
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 4.686997780668861e-06,
+      "loss": 0.6492,
+      "step": 613020
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 4.685849850769112e-06,
+      "loss": 0.6953,
+      "step": 613030
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 4.684701920869365e-06,
+      "loss": 0.7477,
+      "step": 613040
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 4.683553990969618e-06,
+      "loss": 0.6995,
+      "step": 613050
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 4.68240606106987e-06,
+      "loss": 0.6423,
+      "step": 613060
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 4.6812581311701226e-06,
+      "loss": 0.6182,
+      "step": 613070
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 4.680110201270375e-06,
+      "loss": 0.6615,
+      "step": 613080
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 4.678962271370628e-06,
+      "loss": 0.6873,
+      "step": 613090
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 4.677814341470881e-06,
+      "loss": 0.6534,
+      "step": 613100
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 4.676666411571133e-06,
+      "loss": 0.6354,
+      "step": 613110
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 4.675518481671385e-06,
+      "loss": 0.6486,
+      "step": 613120
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 4.674370551771639e-06,
+      "loss": 0.7124,
+      "step": 613130
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 4.67322262187189e-06,
+      "loss": 0.6706,
+      "step": 613140
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 4.6720746919721434e-06,
+      "loss": 0.638,
+      "step": 613150
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 4.670926762072396e-06,
+      "loss": 0.6387,
+      "step": 613160
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 4.669778832172648e-06,
+      "loss": 0.7057,
+      "step": 613170
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 4.668630902272901e-06,
+      "loss": 0.6605,
+      "step": 613180
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 4.667482972373153e-06,
+      "loss": 0.646,
+      "step": 613190
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 4.666335042473406e-06,
+      "loss": 0.7013,
+      "step": 613200
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 4.665187112573659e-06,
+      "loss": 0.7163,
+      "step": 613210
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 4.664039182673911e-06,
+      "loss": 0.6553,
+      "step": 613220
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 4.6628912527741634e-06,
+      "loss": 0.6917,
+      "step": 613230
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 4.661743322874416e-06,
+      "loss": 0.6949,
+      "step": 613240
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 4.660595392974668e-06,
+      "loss": 0.7321,
+      "step": 613250
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 4.6594474630749215e-06,
+      "loss": 0.7538,
+      "step": 613260
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 4.658299533175174e-06,
+      "loss": 0.6554,
+      "step": 613270
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 4.657151603275426e-06,
+      "loss": 0.6431,
+      "step": 613280
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 4.656003673375679e-06,
+      "loss": 0.6846,
+      "step": 613290
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 4.654855743475931e-06,
+      "loss": 0.6842,
+      "step": 613300
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 4.653707813576184e-06,
+      "loss": 0.6709,
+      "step": 613310
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 4.652559883676436e-06,
+      "loss": 0.6818,
+      "step": 613320
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 4.651411953776689e-06,
+      "loss": 0.6857,
+      "step": 613330
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 4.6502640238769415e-06,
+      "loss": 0.6595,
+      "step": 613340
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 4.649116093977194e-06,
+      "loss": 0.6892,
+      "step": 613350
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 4.647968164077446e-06,
+      "loss": 0.6825,
+      "step": 613360
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 4.6468202341776995e-06,
+      "loss": 0.6564,
+      "step": 613370
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 4.645672304277952e-06,
+      "loss": 0.6487,
+      "step": 613380
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 4.644524374378204e-06,
+      "loss": 0.7185,
+      "step": 613390
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 4.643376444478457e-06,
+      "loss": 0.6116,
+      "step": 613400
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 4.642228514578709e-06,
+      "loss": 0.6907,
+      "step": 613410
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 4.641080584678962e-06,
+      "loss": 0.6349,
+      "step": 613420
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 4.639932654779214e-06,
+      "loss": 0.6954,
+      "step": 613430
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 4.638784724879467e-06,
+      "loss": 0.7082,
+      "step": 613440
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 4.6376367949797195e-06,
+      "loss": 0.6406,
+      "step": 613450
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 4.636488865079972e-06,
+      "loss": 0.7078,
+      "step": 613460
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 4.635340935180224e-06,
+      "loss": 0.6806,
+      "step": 613470
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 4.6341930052804776e-06,
+      "loss": 0.7303,
+      "step": 613480
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 4.63304507538073e-06,
+      "loss": 0.7246,
+      "step": 613490
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 4.631897145480982e-06,
+      "loss": 0.7065,
+      "step": 613500
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 4.630749215581235e-06,
+      "loss": 0.6597,
+      "step": 613510
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 4.629601285681487e-06,
+      "loss": 0.6453,
+      "step": 613520
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 4.62845335578174e-06,
+      "loss": 0.6682,
+      "step": 613530
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 4.627305425881992e-06,
+      "loss": 0.6994,
+      "step": 613540
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 4.626157495982245e-06,
+      "loss": 0.6873,
+      "step": 613550
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 4.6250095660824976e-06,
+      "loss": 0.6075,
+      "step": 613560
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 4.62386163618275e-06,
+      "loss": 0.6842,
+      "step": 613570
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 4.622713706283002e-06,
+      "loss": 0.6815,
+      "step": 613580
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 4.621565776383255e-06,
+      "loss": 0.77,
+      "step": 613590
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 4.620417846483508e-06,
+      "loss": 0.6825,
+      "step": 613600
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 4.61926991658376e-06,
+      "loss": 0.6552,
+      "step": 613610
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 4.618121986684013e-06,
+      "loss": 0.6648,
+      "step": 613620
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 4.616974056784265e-06,
+      "loss": 0.7132,
+      "step": 613630
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 4.6158261268845184e-06,
+      "loss": 0.6605,
+      "step": 613640
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 4.61467819698477e-06,
+      "loss": 0.6977,
+      "step": 613650
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 4.613530267085023e-06,
+      "loss": 0.7157,
+      "step": 613660
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 4.612382337185276e-06,
+      "loss": 0.7133,
+      "step": 613670
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 4.611234407285528e-06,
+      "loss": 0.6939,
+      "step": 613680
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 4.61008647738578e-06,
+      "loss": 0.6933,
+      "step": 613690
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 4.608938547486033e-06,
+      "loss": 0.7316,
+      "step": 613700
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 4.607790617586286e-06,
+      "loss": 0.6982,
+      "step": 613710
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 4.6066426876865384e-06,
+      "loss": 0.7456,
+      "step": 613720
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 4.605494757786791e-06,
+      "loss": 0.6761,
+      "step": 613730
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 4.604346827887043e-06,
+      "loss": 0.6599,
+      "step": 613740
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 4.603198897987296e-06,
+      "loss": 0.616,
+      "step": 613750
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 4.602050968087548e-06,
+      "loss": 0.7058,
+      "step": 613760
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 4.600903038187801e-06,
+      "loss": 0.6547,
+      "step": 613770
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 4.599755108288054e-06,
+      "loss": 0.716,
+      "step": 613780
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 4.598607178388306e-06,
+      "loss": 0.667,
+      "step": 613790
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 4.5974592484885584e-06,
+      "loss": 0.6616,
+      "step": 613800
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 4.596311318588811e-06,
+      "loss": 0.8211,
+      "step": 613810
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 4.595163388689064e-06,
+      "loss": 0.7789,
+      "step": 613820
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 4.5940154587893165e-06,
+      "loss": 0.7157,
+      "step": 613830
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 4.592867528889569e-06,
+      "loss": 0.6555,
+      "step": 613840
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 4.591719598989821e-06,
+      "loss": 0.7836,
+      "step": 613850
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 4.590571669090074e-06,
+      "loss": 0.6886,
+      "step": 613860
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 4.589423739190326e-06,
+      "loss": 0.657,
+      "step": 613870
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 4.588275809290579e-06,
+      "loss": 0.6957,
+      "step": 613880
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 4.587127879390832e-06,
+      "loss": 0.6861,
+      "step": 613890
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 4.585979949491084e-06,
+      "loss": 0.7585,
+      "step": 613900
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 4.5848320195913365e-06,
+      "loss": 0.7643,
+      "step": 613910
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 4.583684089691589e-06,
+      "loss": 0.7765,
+      "step": 613920
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 4.582536159791842e-06,
+      "loss": 0.6775,
+      "step": 613930
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 4.581388229892094e-06,
+      "loss": 0.693,
+      "step": 613940
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 4.580240299992347e-06,
+      "loss": 0.6674,
+      "step": 613950
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 4.579092370092599e-06,
+      "loss": 0.6785,
+      "step": 613960
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 4.577944440192852e-06,
+      "loss": 0.7663,
+      "step": 613970
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 4.576796510293104e-06,
+      "loss": 0.7181,
+      "step": 613980
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 4.575648580393357e-06,
+      "loss": 0.6824,
+      "step": 613990
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 4.57450065049361e-06,
+      "loss": 0.6973,
+      "step": 614000
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 4.573352720593862e-06,
+      "loss": 0.7272,
+      "step": 614010
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 4.5722047906941145e-06,
+      "loss": 0.7228,
+      "step": 614020
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 4.571056860794367e-06,
+      "loss": 0.7145,
+      "step": 614030
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 4.56990893089462e-06,
+      "loss": 0.6789,
+      "step": 614040
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 4.568761000994872e-06,
+      "loss": 0.7475,
+      "step": 614050
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 4.567613071095125e-06,
+      "loss": 0.6631,
+      "step": 614060
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 4.566465141195377e-06,
+      "loss": 0.7829,
+      "step": 614070
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 4.56531721129563e-06,
+      "loss": 0.6774,
+      "step": 614080
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 4.564169281395882e-06,
+      "loss": 0.6173,
+      "step": 614090
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 4.5630213514961345e-06,
+      "loss": 0.6743,
+      "step": 614100
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 4.561873421596388e-06,
+      "loss": 0.631,
+      "step": 614110
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 4.56072549169664e-06,
+      "loss": 0.7814,
+      "step": 614120
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 4.559577561796893e-06,
+      "loss": 0.7197,
+      "step": 614130
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 4.558429631897145e-06,
+      "loss": 0.6911,
+      "step": 614140
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 4.557281701997398e-06,
+      "loss": 0.6407,
+      "step": 614150
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 4.55613377209765e-06,
+      "loss": 0.6973,
+      "step": 614160
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 4.554985842197903e-06,
+      "loss": 0.6786,
+      "step": 614170
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 4.553837912298155e-06,
+      "loss": 0.7178,
+      "step": 614180
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 4.552689982398408e-06,
+      "loss": 0.5904,
+      "step": 614190
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 4.55154205249866e-06,
+      "loss": 0.708,
+      "step": 614200
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 4.550394122598913e-06,
+      "loss": 0.6859,
+      "step": 614210
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 4.549246192699166e-06,
+      "loss": 0.6338,
+      "step": 614220
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 4.548098262799418e-06,
+      "loss": 0.6676,
+      "step": 614230
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 4.546950332899671e-06,
+      "loss": 0.6373,
+      "step": 614240
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 4.545802402999923e-06,
+      "loss": 0.6081,
+      "step": 614250
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 4.544654473100175e-06,
+      "loss": 0.679,
+      "step": 614260
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 4.543506543200428e-06,
+      "loss": 0.6923,
+      "step": 614270
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 4.542358613300681e-06,
+      "loss": 0.7247,
+      "step": 614280
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 4.5412106834009334e-06,
+      "loss": 0.7603,
+      "step": 614290
+    },
+    {
+      "epoch": 28.18,
+      "learning_rate": 4.540062753501186e-06,
+      "loss": 0.6399,
+      "step": 614300
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 4.538914823601438e-06,
+      "loss": 0.6728,
+      "step": 614310
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 4.537766893701691e-06,
+      "loss": 0.7668,
+      "step": 614320
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 4.536618963801944e-06,
+      "loss": 0.6937,
+      "step": 614330
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 4.535471033902196e-06,
+      "loss": 0.6535,
+      "step": 614340
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 4.534323104002449e-06,
+      "loss": 0.6846,
+      "step": 614350
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 4.533175174102701e-06,
+      "loss": 0.686,
+      "step": 614360
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 4.5320272442029535e-06,
+      "loss": 0.7145,
+      "step": 614370
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 4.530879314303206e-06,
+      "loss": 0.6645,
+      "step": 614380
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 4.529731384403459e-06,
+      "loss": 0.7424,
+      "step": 614390
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 4.5285834545037115e-06,
+      "loss": 0.6838,
+      "step": 614400
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 4.527435524603964e-06,
+      "loss": 0.743,
+      "step": 614410
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 4.526287594704216e-06,
+      "loss": 0.6332,
+      "step": 614420
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 4.525139664804469e-06,
+      "loss": 0.7555,
+      "step": 614430
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 4.523991734904722e-06,
+      "loss": 0.6569,
+      "step": 614440
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 4.5228438050049735e-06,
+      "loss": 0.6603,
+      "step": 614450
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 4.521695875105227e-06,
+      "loss": 0.6075,
+      "step": 614460
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 4.520547945205479e-06,
+      "loss": 0.6545,
+      "step": 614470
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 4.5194000153057315e-06,
+      "loss": 0.66,
+      "step": 614480
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 4.518252085405984e-06,
+      "loss": 0.6107,
+      "step": 614490
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 4.517104155506237e-06,
+      "loss": 0.651,
+      "step": 614500
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 4.5159562256064895e-06,
+      "loss": 0.7932,
+      "step": 614510
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 4.514808295706742e-06,
+      "loss": 0.7482,
+      "step": 614520
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 4.513660365806994e-06,
+      "loss": 0.7578,
+      "step": 614530
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 4.512512435907247e-06,
+      "loss": 0.6475,
+      "step": 614540
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 4.5113645060075e-06,
+      "loss": 0.7307,
+      "step": 614550
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 4.5102165761077515e-06,
+      "loss": 0.6741,
+      "step": 614560
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 4.509068646208005e-06,
+      "loss": 0.6101,
+      "step": 614570
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 4.507920716308257e-06,
+      "loss": 0.6217,
+      "step": 614580
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 4.5067727864085095e-06,
+      "loss": 0.6587,
+      "step": 614590
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 4.505624856508762e-06,
+      "loss": 0.735,
+      "step": 614600
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 4.504476926609014e-06,
+      "loss": 0.7082,
+      "step": 614610
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 4.5033289967092676e-06,
+      "loss": 0.6441,
+      "step": 614620
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 4.50218106680952e-06,
+      "loss": 0.6298,
+      "step": 614630
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 4.501033136909772e-06,
+      "loss": 0.667,
+      "step": 614640
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 4.499885207010025e-06,
+      "loss": 0.6313,
+      "step": 614650
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 4.498737277110278e-06,
+      "loss": 0.687,
+      "step": 614660
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 4.4975893472105296e-06,
+      "loss": 0.6446,
+      "step": 614670
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 4.496441417310783e-06,
+      "loss": 0.7102,
+      "step": 614680
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 4.495293487411035e-06,
+      "loss": 0.6721,
+      "step": 614690
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 4.494145557511288e-06,
+      "loss": 0.6673,
+      "step": 614700
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 4.49299762761154e-06,
+      "loss": 0.6446,
+      "step": 614710
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 4.491849697711792e-06,
+      "loss": 0.7444,
+      "step": 614720
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 4.490701767812046e-06,
+      "loss": 0.6694,
+      "step": 614730
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 4.489553837912298e-06,
+      "loss": 0.696,
+      "step": 614740
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 4.48840590801255e-06,
+      "loss": 0.6267,
+      "step": 614750
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 4.487257978112803e-06,
+      "loss": 0.7072,
+      "step": 614760
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 4.486110048213056e-06,
+      "loss": 0.7106,
+      "step": 614770
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 4.484962118313308e-06,
+      "loss": 0.6768,
+      "step": 614780
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 4.483814188413561e-06,
+      "loss": 0.655,
+      "step": 614790
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 4.482666258513813e-06,
+      "loss": 0.7549,
+      "step": 614800
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 4.481518328614066e-06,
+      "loss": 0.6904,
+      "step": 614810
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 4.480370398714318e-06,
+      "loss": 0.6593,
+      "step": 614820
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 4.47922246881457e-06,
+      "loss": 0.6371,
+      "step": 614830
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 4.478074538914824e-06,
+      "loss": 0.6496,
+      "step": 614840
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 4.476926609015076e-06,
+      "loss": 0.7168,
+      "step": 614850
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 4.4757786791153285e-06,
+      "loss": 0.6484,
+      "step": 614860
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 4.474630749215581e-06,
+      "loss": 0.6299,
+      "step": 614870
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 4.473482819315833e-06,
+      "loss": 0.6281,
+      "step": 614880
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 4.472334889416086e-06,
+      "loss": 0.747,
+      "step": 614890
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 4.471186959516339e-06,
+      "loss": 0.621,
+      "step": 614900
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 4.470039029616591e-06,
+      "loss": 0.7,
+      "step": 614910
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 4.468891099716844e-06,
+      "loss": 0.6985,
+      "step": 614920
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 4.467743169817096e-06,
+      "loss": 0.652,
+      "step": 614930
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 4.4665952399173485e-06,
+      "loss": 0.6314,
+      "step": 614940
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 4.465447310017602e-06,
+      "loss": 0.6558,
+      "step": 614950
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 4.464299380117853e-06,
+      "loss": 0.6997,
+      "step": 614960
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 4.4631514502181065e-06,
+      "loss": 0.6803,
+      "step": 614970
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 4.462003520318359e-06,
+      "loss": 0.7136,
+      "step": 614980
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 4.460855590418611e-06,
+      "loss": 0.6234,
+      "step": 614990
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 4.459707660518864e-06,
+      "loss": 0.6951,
+      "step": 615000
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 4.458559730619117e-06,
+      "loss": 0.6982,
+      "step": 615010
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 4.457411800719369e-06,
+      "loss": 0.6948,
+      "step": 615020
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 4.456263870819622e-06,
+      "loss": 0.7207,
+      "step": 615030
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 4.455115940919874e-06,
+      "loss": 0.6701,
+      "step": 615040
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 4.4539680110201265e-06,
+      "loss": 0.7209,
+      "step": 615050
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 4.45282008112038e-06,
+      "loss": 0.736,
+      "step": 615060
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 4.451672151220631e-06,
+      "loss": 0.643,
+      "step": 615070
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 4.4505242213208845e-06,
+      "loss": 0.6156,
+      "step": 615080
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 4.449376291421137e-06,
+      "loss": 0.7401,
+      "step": 615090
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 4.448228361521389e-06,
+      "loss": 0.708,
+      "step": 615100
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 4.447080431621642e-06,
+      "loss": 0.5543,
+      "step": 615110
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 4.445932501721894e-06,
+      "loss": 0.6839,
+      "step": 615120
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 4.444784571822147e-06,
+      "loss": 0.6706,
+      "step": 615130
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 4.4436366419224e-06,
+      "loss": 0.6148,
+      "step": 615140
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 4.442488712022652e-06,
+      "loss": 0.6679,
+      "step": 615150
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 4.4413407821229045e-06,
+      "loss": 0.6761,
+      "step": 615160
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 4.440192852223158e-06,
+      "loss": 0.6722,
+      "step": 615170
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 4.439044922323409e-06,
+      "loss": 0.6577,
+      "step": 615180
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 4.437896992423663e-06,
+      "loss": 0.6838,
+      "step": 615190
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 4.436749062523915e-06,
+      "loss": 0.7387,
+      "step": 615200
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 4.435601132624167e-06,
+      "loss": 0.6514,
+      "step": 615210
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 4.43445320272442e-06,
+      "loss": 0.6391,
+      "step": 615220
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 4.433305272824672e-06,
+      "loss": 0.6515,
+      "step": 615230
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 4.432157342924925e-06,
+      "loss": 0.777,
+      "step": 615240
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 4.431009413025178e-06,
+      "loss": 0.7389,
+      "step": 615250
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 4.42986148312543e-06,
+      "loss": 0.6968,
+      "step": 615260
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 4.428713553225683e-06,
+      "loss": 0.6901,
+      "step": 615270
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 4.427565623325936e-06,
+      "loss": 0.6467,
+      "step": 615280
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 4.426417693426187e-06,
+      "loss": 0.761,
+      "step": 615290
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 4.425269763526441e-06,
+      "loss": 0.7353,
+      "step": 615300
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 4.424121833626692e-06,
+      "loss": 0.6681,
+      "step": 615310
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 4.422973903726945e-06,
+      "loss": 0.6604,
+      "step": 615320
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 4.421825973827198e-06,
+      "loss": 0.6207,
+      "step": 615330
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 4.42067804392745e-06,
+      "loss": 0.6938,
+      "step": 615340
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 4.419530114027703e-06,
+      "loss": 0.646,
+      "step": 615350
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 4.418382184127956e-06,
+      "loss": 0.7121,
+      "step": 615360
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 4.417234254228208e-06,
+      "loss": 0.7032,
+      "step": 615370
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 4.416086324328461e-06,
+      "loss": 0.6119,
+      "step": 615380
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 4.414938394428713e-06,
+      "loss": 0.6388,
+      "step": 615390
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 4.4137904645289654e-06,
+      "loss": 0.6027,
+      "step": 615400
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 4.412642534629219e-06,
+      "loss": 0.678,
+      "step": 615410
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 4.41149460472947e-06,
+      "loss": 0.6259,
+      "step": 615420
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 4.4103466748297235e-06,
+      "loss": 0.6724,
+      "step": 615430
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 4.409198744929976e-06,
+      "loss": 0.699,
+      "step": 615440
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 4.408050815030228e-06,
+      "loss": 0.6603,
+      "step": 615450
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 4.406902885130481e-06,
+      "loss": 0.6512,
+      "step": 615460
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 4.405754955230733e-06,
+      "loss": 0.7001,
+      "step": 615470
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 4.404607025330986e-06,
+      "loss": 0.6705,
+      "step": 615480
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 4.403459095431239e-06,
+      "loss": 0.6838,
+      "step": 615490
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 4.402311165531491e-06,
+      "loss": 0.7201,
+      "step": 615500
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 4.4011632356317435e-06,
+      "loss": 0.675,
+      "step": 615510
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 4.400015305731997e-06,
+      "loss": 0.6854,
+      "step": 615520
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 4.398867375832248e-06,
+      "loss": 0.7895,
+      "step": 615530
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 4.3977194459325015e-06,
+      "loss": 0.7639,
+      "step": 615540
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 4.396571516032754e-06,
+      "loss": 0.6763,
+      "step": 615550
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 4.395423586133006e-06,
+      "loss": 0.729,
+      "step": 615560
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 4.394275656233259e-06,
+      "loss": 0.6437,
+      "step": 615570
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 4.393127726333511e-06,
+      "loss": 0.7348,
+      "step": 615580
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 4.391979796433764e-06,
+      "loss": 0.713,
+      "step": 615590
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 4.390831866534017e-06,
+      "loss": 0.578,
+      "step": 615600
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 4.389683936634269e-06,
+      "loss": 0.6534,
+      "step": 615610
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 4.3885360067345215e-06,
+      "loss": 0.6501,
+      "step": 615620
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 4.387388076834775e-06,
+      "loss": 0.7105,
+      "step": 615630
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 4.386240146935026e-06,
+      "loss": 0.7765,
+      "step": 615640
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 4.3850922170352795e-06,
+      "loss": 0.7171,
+      "step": 615650
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 4.383944287135532e-06,
+      "loss": 0.6411,
+      "step": 615660
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 4.382796357235784e-06,
+      "loss": 0.7122,
+      "step": 615670
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 4.381648427336037e-06,
+      "loss": 0.714,
+      "step": 615680
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 4.380500497436289e-06,
+      "loss": 0.6918,
+      "step": 615690
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 4.379352567536542e-06,
+      "loss": 0.6985,
+      "step": 615700
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 4.378204637636795e-06,
+      "loss": 0.7146,
+      "step": 615710
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 4.377056707737047e-06,
+      "loss": 0.7409,
+      "step": 615720
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 4.3759087778372996e-06,
+      "loss": 0.6407,
+      "step": 615730
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 4.374760847937552e-06,
+      "loss": 0.6679,
+      "step": 615740
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 4.373612918037804e-06,
+      "loss": 0.6396,
+      "step": 615750
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 4.372464988138058e-06,
+      "loss": 0.6443,
+      "step": 615760
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 4.37131705823831e-06,
+      "loss": 0.7174,
+      "step": 615770
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 4.370169128338562e-06,
+      "loss": 0.7437,
+      "step": 615780
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 4.369021198438815e-06,
+      "loss": 0.6555,
+      "step": 615790
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 4.367873268539067e-06,
+      "loss": 0.7497,
+      "step": 615800
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 4.36672533863932e-06,
+      "loss": 0.7201,
+      "step": 615810
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 4.365577408739572e-06,
+      "loss": 0.6234,
+      "step": 615820
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 4.364429478839825e-06,
+      "loss": 0.6454,
+      "step": 615830
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 4.363281548940078e-06,
+      "loss": 0.7214,
+      "step": 615840
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 4.36213361904033e-06,
+      "loss": 0.6881,
+      "step": 615850
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 4.360985689140582e-06,
+      "loss": 0.6627,
+      "step": 615860
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 4.359837759240836e-06,
+      "loss": 0.7325,
+      "step": 615870
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 4.358689829341088e-06,
+      "loss": 0.7316,
+      "step": 615880
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 4.3575418994413404e-06,
+      "loss": 0.6702,
+      "step": 615890
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 4.356393969541593e-06,
+      "loss": 0.6732,
+      "step": 615900
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 4.355246039641845e-06,
+      "loss": 0.7662,
+      "step": 615910
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 4.3540981097420985e-06,
+      "loss": 0.682,
+      "step": 615920
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 4.35295017984235e-06,
+      "loss": 0.7329,
+      "step": 615930
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 4.351802249942603e-06,
+      "loss": 0.6465,
+      "step": 615940
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 4.350654320042856e-06,
+      "loss": 0.6465,
+      "step": 615950
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 4.349506390143108e-06,
+      "loss": 0.6714,
+      "step": 615960
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 4.3483584602433604e-06,
+      "loss": 0.7388,
+      "step": 615970
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 4.347210530343613e-06,
+      "loss": 0.6863,
+      "step": 615980
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 4.346062600443866e-06,
+      "loss": 0.6604,
+      "step": 615990
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 4.3449146705441185e-06,
+      "loss": 0.6735,
+      "step": 616000
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 4.343766740644371e-06,
+      "loss": 0.6923,
+      "step": 616010
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 4.342618810744623e-06,
+      "loss": 0.6629,
+      "step": 616020
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 4.3414708808448765e-06,
+      "loss": 0.6859,
+      "step": 616030
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 4.340322950945128e-06,
+      "loss": 0.644,
+      "step": 616040
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 4.339175021045381e-06,
+      "loss": 0.6934,
+      "step": 616050
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 4.338027091145634e-06,
+      "loss": 0.6825,
+      "step": 616060
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 4.336879161245886e-06,
+      "loss": 0.7076,
+      "step": 616070
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 4.3357312313461385e-06,
+      "loss": 0.6923,
+      "step": 616080
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 4.334583301446391e-06,
+      "loss": 0.6797,
+      "step": 616090
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 4.333435371546644e-06,
+      "loss": 0.7045,
+      "step": 616100
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 4.3322874416468965e-06,
+      "loss": 0.6937,
+      "step": 616110
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 4.331139511747149e-06,
+      "loss": 0.669,
+      "step": 616120
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 4.329991581847401e-06,
+      "loss": 0.7117,
+      "step": 616130
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 4.3288436519476545e-06,
+      "loss": 0.7033,
+      "step": 616140
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 4.327695722047906e-06,
+      "loss": 0.7586,
+      "step": 616150
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 4.326547792148159e-06,
+      "loss": 0.6056,
+      "step": 616160
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 4.325399862248412e-06,
+      "loss": 0.6678,
+      "step": 616170
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 4.324251932348664e-06,
+      "loss": 0.6945,
+      "step": 616180
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 4.3231040024489165e-06,
+      "loss": 0.6809,
+      "step": 616190
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 4.321956072549169e-06,
+      "loss": 0.6995,
+      "step": 616200
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 4.320808142649422e-06,
+      "loss": 0.7807,
+      "step": 616210
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 4.3196602127496746e-06,
+      "loss": 0.6975,
+      "step": 616220
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 4.318512282849927e-06,
+      "loss": 0.6383,
+      "step": 616230
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 4.317364352950179e-06,
+      "loss": 0.7953,
+      "step": 616240
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 4.316216423050432e-06,
+      "loss": 0.6782,
+      "step": 616250
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 4.315068493150684e-06,
+      "loss": 0.7044,
+      "step": 616260
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 4.313920563250937e-06,
+      "loss": 0.6785,
+      "step": 616270
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 4.31277263335119e-06,
+      "loss": 0.7123,
+      "step": 616280
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 4.311624703451442e-06,
+      "loss": 0.6762,
+      "step": 616290
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 4.3104767735516946e-06,
+      "loss": 0.6811,
+      "step": 616300
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 4.309328843651947e-06,
+      "loss": 0.7021,
+      "step": 616310
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 4.3081809137522e-06,
+      "loss": 0.7036,
+      "step": 616320
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 4.307032983852452e-06,
+      "loss": 0.7556,
+      "step": 616330
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 4.305885053952705e-06,
+      "loss": 0.7067,
+      "step": 616340
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 4.304737124052957e-06,
+      "loss": 0.6681,
+      "step": 616350
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 4.30358919415321e-06,
+      "loss": 0.7135,
+      "step": 616360
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 4.302441264253462e-06,
+      "loss": 0.6419,
+      "step": 616370
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 4.301293334353715e-06,
+      "loss": 0.6964,
+      "step": 616380
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 4.300145404453968e-06,
+      "loss": 0.6982,
+      "step": 616390
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 4.29899747455422e-06,
+      "loss": 0.7283,
+      "step": 616400
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 4.297849544654473e-06,
+      "loss": 0.6723,
+      "step": 616410
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 4.296701614754725e-06,
+      "loss": 0.7038,
+      "step": 616420
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 4.295553684854978e-06,
+      "loss": 0.6933,
+      "step": 616430
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 4.29440575495523e-06,
+      "loss": 0.7491,
+      "step": 616440
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 4.293257825055483e-06,
+      "loss": 0.6594,
+      "step": 616450
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 4.2921098951557354e-06,
+      "loss": 0.7059,
+      "step": 616460
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 4.290961965255988e-06,
+      "loss": 0.6732,
+      "step": 616470
+    },
+    {
+      "epoch": 28.28,
+      "learning_rate": 4.28981403535624e-06,
+      "loss": 0.6504,
+      "step": 616480
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 4.288666105456493e-06,
+      "loss": 0.7002,
+      "step": 616490
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 4.287518175556746e-06,
+      "loss": 0.6684,
+      "step": 616500
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 4.286370245656998e-06,
+      "loss": 0.7151,
+      "step": 616510
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 4.285222315757251e-06,
+      "loss": 0.6246,
+      "step": 616520
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 4.284074385857503e-06,
+      "loss": 0.6461,
+      "step": 616530
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 4.282926455957756e-06,
+      "loss": 0.7142,
+      "step": 616540
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 4.281778526058008e-06,
+      "loss": 0.7616,
+      "step": 616550
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 4.280630596158261e-06,
+      "loss": 0.7305,
+      "step": 616560
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 4.2794826662585135e-06,
+      "loss": 0.6983,
+      "step": 616570
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 4.278334736358766e-06,
+      "loss": 0.6672,
+      "step": 616580
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 4.277186806459018e-06,
+      "loss": 0.611,
+      "step": 616590
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 4.276038876559271e-06,
+      "loss": 0.7397,
+      "step": 616600
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 4.274890946659524e-06,
+      "loss": 0.7082,
+      "step": 616610
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 4.273743016759776e-06,
+      "loss": 0.7333,
+      "step": 616620
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 4.272595086860029e-06,
+      "loss": 0.6842,
+      "step": 616630
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 4.271447156960281e-06,
+      "loss": 0.7349,
+      "step": 616640
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 4.270299227060534e-06,
+      "loss": 0.7395,
+      "step": 616650
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 4.269151297160786e-06,
+      "loss": 0.6828,
+      "step": 616660
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 4.268003367261039e-06,
+      "loss": 0.771,
+      "step": 616670
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 4.2668554373612915e-06,
+      "loss": 0.6759,
+      "step": 616680
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 4.265707507461544e-06,
+      "loss": 0.6953,
+      "step": 616690
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 4.264559577561796e-06,
+      "loss": 0.756,
+      "step": 616700
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 4.263411647662049e-06,
+      "loss": 0.7139,
+      "step": 616710
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 4.262263717762302e-06,
+      "loss": 0.6669,
+      "step": 616720
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 4.261115787862554e-06,
+      "loss": 0.675,
+      "step": 616730
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 4.259967857962807e-06,
+      "loss": 0.6869,
+      "step": 616740
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 4.258819928063059e-06,
+      "loss": 0.6803,
+      "step": 616750
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 4.2576719981633115e-06,
+      "loss": 0.6369,
+      "step": 616760
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 4.256524068263564e-06,
+      "loss": 0.6641,
+      "step": 616770
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 4.255376138363817e-06,
+      "loss": 0.5969,
+      "step": 616780
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 4.2542282084640696e-06,
+      "loss": 0.7001,
+      "step": 616790
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 4.253080278564322e-06,
+      "loss": 0.7269,
+      "step": 616800
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 4.251932348664574e-06,
+      "loss": 0.732,
+      "step": 616810
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 4.250784418764827e-06,
+      "loss": 0.716,
+      "step": 616820
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 4.24963648886508e-06,
+      "loss": 0.734,
+      "step": 616830
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 4.2484885589653315e-06,
+      "loss": 0.7146,
+      "step": 616840
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 4.247340629065585e-06,
+      "loss": 0.7103,
+      "step": 616850
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 4.246192699165837e-06,
+      "loss": 0.7502,
+      "step": 616860
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 4.2450447692660896e-06,
+      "loss": 0.7826,
+      "step": 616870
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 4.243896839366342e-06,
+      "loss": 0.6735,
+      "step": 616880
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 4.242748909466595e-06,
+      "loss": 0.7831,
+      "step": 616890
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 4.241600979566848e-06,
+      "loss": 0.653,
+      "step": 616900
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 4.2404530496671e-06,
+      "loss": 0.6418,
+      "step": 616910
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 4.239305119767352e-06,
+      "loss": 0.7207,
+      "step": 616920
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 4.238157189867605e-06,
+      "loss": 0.7585,
+      "step": 616930
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 4.237009259967858e-06,
+      "loss": 0.7073,
+      "step": 616940
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 4.23586133006811e-06,
+      "loss": 0.6646,
+      "step": 616950
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 4.234713400168363e-06,
+      "loss": 0.7207,
+      "step": 616960
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 4.233565470268615e-06,
+      "loss": 0.6373,
+      "step": 616970
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 4.232417540368868e-06,
+      "loss": 0.6453,
+      "step": 616980
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 4.23126961046912e-06,
+      "loss": 0.6498,
+      "step": 616990
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 4.230121680569373e-06,
+      "loss": 0.6643,
+      "step": 617000
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 4.228973750669626e-06,
+      "loss": 0.6924,
+      "step": 617010
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 4.227825820769878e-06,
+      "loss": 0.7286,
+      "step": 617020
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 4.2266778908701304e-06,
+      "loss": 0.7844,
+      "step": 617030
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 4.225529960970383e-06,
+      "loss": 0.6437,
+      "step": 617040
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 4.224382031070636e-06,
+      "loss": 0.6783,
+      "step": 617050
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 4.223234101170888e-06,
+      "loss": 0.6632,
+      "step": 617060
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 4.222086171271141e-06,
+      "loss": 0.7441,
+      "step": 617070
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 4.220938241371393e-06,
+      "loss": 0.6958,
+      "step": 617080
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 4.219790311471646e-06,
+      "loss": 0.7228,
+      "step": 617090
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 4.218642381571898e-06,
+      "loss": 0.6732,
+      "step": 617100
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 4.2174944516721504e-06,
+      "loss": 0.6432,
+      "step": 617110
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 4.216346521772404e-06,
+      "loss": 0.643,
+      "step": 617120
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 4.215198591872656e-06,
+      "loss": 0.6608,
+      "step": 617130
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 4.2140506619729085e-06,
+      "loss": 0.6977,
+      "step": 617140
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 4.212902732073161e-06,
+      "loss": 0.5967,
+      "step": 617150
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 4.211754802173414e-06,
+      "loss": 0.6329,
+      "step": 617160
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 4.210606872273666e-06,
+      "loss": 0.6563,
+      "step": 617170
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 4.209458942373919e-06,
+      "loss": 0.6691,
+      "step": 617180
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 4.208311012474171e-06,
+      "loss": 0.7208,
+      "step": 617190
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 4.207163082574424e-06,
+      "loss": 0.6907,
+      "step": 617200
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 4.206015152674676e-06,
+      "loss": 0.6068,
+      "step": 617210
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 4.2048672227749285e-06,
+      "loss": 0.6711,
+      "step": 617220
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 4.203719292875182e-06,
+      "loss": 0.7114,
+      "step": 617230
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 4.202571362975434e-06,
+      "loss": 0.7483,
+      "step": 617240
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 4.2014234330756865e-06,
+      "loss": 0.6967,
+      "step": 617250
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 4.200275503175939e-06,
+      "loss": 0.6724,
+      "step": 617260
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 4.199127573276191e-06,
+      "loss": 0.6419,
+      "step": 617270
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 4.197979643376444e-06,
+      "loss": 0.6503,
+      "step": 617280
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 4.196831713476697e-06,
+      "loss": 0.7361,
+      "step": 617290
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 4.195683783576949e-06,
+      "loss": 0.7011,
+      "step": 617300
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 4.194535853677202e-06,
+      "loss": 0.6792,
+      "step": 617310
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 4.193387923777454e-06,
+      "loss": 0.6533,
+      "step": 617320
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 4.1922399938777065e-06,
+      "loss": 0.7242,
+      "step": 617330
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 4.19109206397796e-06,
+      "loss": 0.6287,
+      "step": 617340
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 4.189944134078211e-06,
+      "loss": 0.6904,
+      "step": 617350
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 4.1887962041784646e-06,
+      "loss": 0.6739,
+      "step": 617360
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 4.187648274278717e-06,
+      "loss": 0.668,
+      "step": 617370
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 4.186500344378969e-06,
+      "loss": 0.6945,
+      "step": 617380
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 4.185352414479222e-06,
+      "loss": 0.724,
+      "step": 617390
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 4.184204484579475e-06,
+      "loss": 0.6949,
+      "step": 617400
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 4.183056554679727e-06,
+      "loss": 0.7217,
+      "step": 617410
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 4.18190862477998e-06,
+      "loss": 0.7023,
+      "step": 617420
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 4.180760694880232e-06,
+      "loss": 0.6066,
+      "step": 617430
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 4.179612764980485e-06,
+      "loss": 0.7539,
+      "step": 617440
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 4.178464835080738e-06,
+      "loss": 0.6706,
+      "step": 617450
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 4.177316905180989e-06,
+      "loss": 0.6632,
+      "step": 617460
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 4.176168975281243e-06,
+      "loss": 0.6701,
+      "step": 617470
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 4.175021045381495e-06,
+      "loss": 0.6852,
+      "step": 617480
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 4.173873115481747e-06,
+      "loss": 0.7495,
+      "step": 617490
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 4.172725185582e-06,
+      "loss": 0.6956,
+      "step": 617500
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 4.171577255682253e-06,
+      "loss": 0.7166,
+      "step": 617510
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 4.1704293257825054e-06,
+      "loss": 0.6899,
+      "step": 617520
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 4.169281395882758e-06,
+      "loss": 0.6668,
+      "step": 617530
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 4.16813346598301e-06,
+      "loss": 0.6407,
+      "step": 617540
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 4.166985536083263e-06,
+      "loss": 0.715,
+      "step": 617550
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 4.165837606183516e-06,
+      "loss": 0.6548,
+      "step": 617560
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 4.164689676283767e-06,
+      "loss": 0.7395,
+      "step": 617570
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 4.163541746384021e-06,
+      "loss": 0.7062,
+      "step": 617580
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 4.162393816484273e-06,
+      "loss": 0.6868,
+      "step": 617590
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 4.1612458865845254e-06,
+      "loss": 0.7026,
+      "step": 617600
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 4.160097956684778e-06,
+      "loss": 0.7061,
+      "step": 617610
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 4.15895002678503e-06,
+      "loss": 0.6109,
+      "step": 617620
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 4.1578020968852835e-06,
+      "loss": 0.6444,
+      "step": 617630
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 4.156654166985536e-06,
+      "loss": 0.7366,
+      "step": 617640
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 4.155506237085788e-06,
+      "loss": 0.6778,
+      "step": 617650
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 4.154358307186041e-06,
+      "loss": 0.6125,
+      "step": 617660
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 4.153210377286294e-06,
+      "loss": 0.6186,
+      "step": 617670
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 4.1520624473865455e-06,
+      "loss": 0.7087,
+      "step": 617680
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 4.150914517486799e-06,
+      "loss": 0.7256,
+      "step": 617690
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 4.149766587587051e-06,
+      "loss": 0.6763,
+      "step": 617700
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 4.1486186576873035e-06,
+      "loss": 0.7399,
+      "step": 617710
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 4.147470727787556e-06,
+      "loss": 0.7779,
+      "step": 617720
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 4.146322797887808e-06,
+      "loss": 0.6882,
+      "step": 617730
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 4.1451748679880615e-06,
+      "loss": 0.7095,
+      "step": 617740
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 4.144026938088314e-06,
+      "loss": 0.6919,
+      "step": 617750
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 4.142879008188566e-06,
+      "loss": 0.6422,
+      "step": 617760
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 4.141731078288819e-06,
+      "loss": 0.6339,
+      "step": 617770
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 4.140583148389071e-06,
+      "loss": 0.6655,
+      "step": 617780
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 4.1394352184893235e-06,
+      "loss": 0.5993,
+      "step": 617790
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 4.138287288589577e-06,
+      "loss": 0.6865,
+      "step": 617800
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 4.137139358689829e-06,
+      "loss": 0.7418,
+      "step": 617810
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 4.1359914287900815e-06,
+      "loss": 0.6572,
+      "step": 617820
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 4.134843498890334e-06,
+      "loss": 0.6592,
+      "step": 617830
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 4.133695568990586e-06,
+      "loss": 0.6685,
+      "step": 617840
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 4.1325476390908396e-06,
+      "loss": 0.6202,
+      "step": 617850
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 4.131399709191092e-06,
+      "loss": 0.6945,
+      "step": 617860
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 4.130251779291344e-06,
+      "loss": 0.6452,
+      "step": 617870
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 4.129103849391597e-06,
+      "loss": 0.7276,
+      "step": 617880
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 4.127955919491849e-06,
+      "loss": 0.7357,
+      "step": 617890
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 4.1268079895921015e-06,
+      "loss": 0.6526,
+      "step": 617900
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 4.125660059692355e-06,
+      "loss": 0.6402,
+      "step": 617910
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 4.124512129792607e-06,
+      "loss": 0.697,
+      "step": 617920
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 4.12336419989286e-06,
+      "loss": 0.7193,
+      "step": 617930
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 4.122216269993112e-06,
+      "loss": 0.6791,
+      "step": 617940
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 4.121068340093364e-06,
+      "loss": 0.6902,
+      "step": 617950
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 4.119920410193618e-06,
+      "loss": 0.7582,
+      "step": 617960
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 4.118772480293869e-06,
+      "loss": 0.6348,
+      "step": 617970
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 4.117624550394122e-06,
+      "loss": 0.7385,
+      "step": 617980
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 4.116476620494375e-06,
+      "loss": 0.7203,
+      "step": 617990
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 4.115328690594627e-06,
+      "loss": 0.6716,
+      "step": 618000
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 4.11418076069488e-06,
+      "loss": 0.6618,
+      "step": 618010
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 4.113032830795133e-06,
+      "loss": 0.6738,
+      "step": 618020
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 4.111884900895385e-06,
+      "loss": 0.7661,
+      "step": 618030
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 4.110736970995638e-06,
+      "loss": 0.6802,
+      "step": 618040
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 4.10958904109589e-06,
+      "loss": 0.6986,
+      "step": 618050
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 4.108441111196142e-06,
+      "loss": 0.6727,
+      "step": 618060
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 4.107293181296396e-06,
+      "loss": 0.6902,
+      "step": 618070
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 4.106145251396647e-06,
+      "loss": 0.6782,
+      "step": 618080
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 4.1049973214969004e-06,
+      "loss": 0.6684,
+      "step": 618090
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 4.103849391597153e-06,
+      "loss": 0.682,
+      "step": 618100
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 4.102701461697405e-06,
+      "loss": 0.6967,
+      "step": 618110
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 4.101553531797658e-06,
+      "loss": 0.7156,
+      "step": 618120
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 4.10040560189791e-06,
+      "loss": 0.7503,
+      "step": 618130
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 4.099257671998163e-06,
+      "loss": 0.6582,
+      "step": 618140
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 4.098109742098416e-06,
+      "loss": 0.6719,
+      "step": 618150
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 4.096961812198668e-06,
+      "loss": 0.7569,
+      "step": 618160
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 4.0958138822989205e-06,
+      "loss": 0.7003,
+      "step": 618170
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 4.094665952399174e-06,
+      "loss": 0.6587,
+      "step": 618180
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 4.093518022499425e-06,
+      "loss": 0.6458,
+      "step": 618190
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 4.0923700925996785e-06,
+      "loss": 0.6531,
+      "step": 618200
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 4.091222162699931e-06,
+      "loss": 0.7654,
+      "step": 618210
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 4.090074232800183e-06,
+      "loss": 0.5956,
+      "step": 618220
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 4.088926302900436e-06,
+      "loss": 0.6652,
+      "step": 618230
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 4.087778373000688e-06,
+      "loss": 0.7201,
+      "step": 618240
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 4.086630443100941e-06,
+      "loss": 0.6979,
+      "step": 618250
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 4.085482513201194e-06,
+      "loss": 0.6676,
+      "step": 618260
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 4.084334583301446e-06,
+      "loss": 0.697,
+      "step": 618270
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 4.0831866534016985e-06,
+      "loss": 0.7448,
+      "step": 618280
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 4.082038723501952e-06,
+      "loss": 0.766,
+      "step": 618290
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 4.080890793602203e-06,
+      "loss": 0.6823,
+      "step": 618300
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 4.0797428637024565e-06,
+      "loss": 0.742,
+      "step": 618310
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 4.078594933802709e-06,
+      "loss": 0.6996,
+      "step": 618320
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 4.077447003902961e-06,
+      "loss": 0.728,
+      "step": 618330
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 4.076299074003214e-06,
+      "loss": 0.6796,
+      "step": 618340
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 4.075151144103466e-06,
+      "loss": 0.7054,
+      "step": 618350
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 4.074003214203719e-06,
+      "loss": 0.7357,
+      "step": 618360
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 4.072855284303972e-06,
+      "loss": 0.7579,
+      "step": 618370
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 4.071707354404224e-06,
+      "loss": 0.6937,
+      "step": 618380
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 4.0705594245044765e-06,
+      "loss": 0.6612,
+      "step": 618390
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 4.069411494604729e-06,
+      "loss": 0.7272,
+      "step": 618400
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 4.068263564704981e-06,
+      "loss": 0.6604,
+      "step": 618410
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 4.0671156348052346e-06,
+      "loss": 0.7414,
+      "step": 618420
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 4.065967704905487e-06,
+      "loss": 0.7017,
+      "step": 618430
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 4.064819775005739e-06,
+      "loss": 0.7734,
+      "step": 618440
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 4.063671845105992e-06,
+      "loss": 0.6139,
+      "step": 618450
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 4.062523915206244e-06,
+      "loss": 0.6644,
+      "step": 618460
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 4.061375985306497e-06,
+      "loss": 0.6481,
+      "step": 618470
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 4.060228055406749e-06,
+      "loss": 0.6147,
+      "step": 618480
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 4.059080125507002e-06,
+      "loss": 0.7312,
+      "step": 618490
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 4.057932195607255e-06,
+      "loss": 0.7324,
+      "step": 618500
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 4.056784265707507e-06,
+      "loss": 0.713,
+      "step": 618510
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 4.055636335807759e-06,
+      "loss": 0.6583,
+      "step": 618520
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 4.054488405908013e-06,
+      "loss": 0.6649,
+      "step": 618530
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 4.053340476008265e-06,
+      "loss": 0.715,
+      "step": 618540
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 4.052192546108517e-06,
+      "loss": 0.7267,
+      "step": 618550
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 4.05104461620877e-06,
+      "loss": 0.7233,
+      "step": 618560
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 4.049896686309022e-06,
+      "loss": 0.6947,
+      "step": 618570
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 4.0487487564092754e-06,
+      "loss": 0.6888,
+      "step": 618580
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 4.047600826509527e-06,
+      "loss": 0.6628,
+      "step": 618590
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 4.04645289660978e-06,
+      "loss": 0.658,
+      "step": 618600
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 4.045304966710033e-06,
+      "loss": 0.7606,
+      "step": 618610
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 4.044157036810285e-06,
+      "loss": 0.7388,
+      "step": 618620
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 4.043009106910537e-06,
+      "loss": 0.7631,
+      "step": 618630
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 4.04186117701079e-06,
+      "loss": 0.7949,
+      "step": 618640
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 4.040713247111043e-06,
+      "loss": 0.6622,
+      "step": 618650
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 4.0395653172112955e-06,
+      "loss": 0.6947,
+      "step": 618660
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 4.038417387311548e-06,
+      "loss": 0.6667,
+      "step": 618670
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 4.0372694574118e-06,
+      "loss": 0.6687,
+      "step": 618680
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 4.0361215275120535e-06,
+      "loss": 0.7272,
+      "step": 618690
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 4.034973597612305e-06,
+      "loss": 0.6686,
+      "step": 618700
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 4.033825667712558e-06,
+      "loss": 0.7345,
+      "step": 618710
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 4.032677737812811e-06,
+      "loss": 0.717,
+      "step": 618720
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 4.031529807913063e-06,
+      "loss": 0.7318,
+      "step": 618730
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 4.0303818780133155e-06,
+      "loss": 0.6396,
+      "step": 618740
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 4.029233948113568e-06,
+      "loss": 0.6729,
+      "step": 618750
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 4.028086018213821e-06,
+      "loss": 0.6996,
+      "step": 618760
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 4.0269380883140735e-06,
+      "loss": 0.7042,
+      "step": 618770
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 4.025790158414326e-06,
+      "loss": 0.6612,
+      "step": 618780
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 4.024642228514578e-06,
+      "loss": 0.7157,
+      "step": 618790
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 4.0234942986148315e-06,
+      "loss": 0.6678,
+      "step": 618800
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 4.022346368715083e-06,
+      "loss": 0.6579,
+      "step": 618810
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 4.021198438815336e-06,
+      "loss": 0.7014,
+      "step": 618820
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 4.020050508915589e-06,
+      "loss": 0.6626,
+      "step": 618830
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 4.018902579015841e-06,
+      "loss": 0.7375,
+      "step": 618840
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 4.0177546491160935e-06,
+      "loss": 0.6735,
+      "step": 618850
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 4.016606719216346e-06,
+      "loss": 0.7099,
+      "step": 618860
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 4.015458789316599e-06,
+      "loss": 0.6647,
+      "step": 618870
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 4.0143108594168515e-06,
+      "loss": 0.6358,
+      "step": 618880
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 4.013162929517104e-06,
+      "loss": 0.6947,
+      "step": 618890
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 4.012014999617356e-06,
+      "loss": 0.7058,
+      "step": 618900
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 4.010867069717609e-06,
+      "loss": 0.6161,
+      "step": 618910
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 4.009719139817861e-06,
+      "loss": 0.6774,
+      "step": 618920
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 4.008571209918114e-06,
+      "loss": 0.6911,
+      "step": 618930
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 4.007423280018367e-06,
+      "loss": 0.6741,
+      "step": 618940
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 4.006275350118619e-06,
+      "loss": 0.6958,
+      "step": 618950
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 4.0051274202188715e-06,
+      "loss": 0.6465,
+      "step": 618960
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 4.003979490319124e-06,
+      "loss": 0.612,
+      "step": 618970
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 4.002831560419377e-06,
+      "loss": 0.6695,
+      "step": 618980
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 4.001683630519629e-06,
+      "loss": 0.701,
+      "step": 618990
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 4.000535700619882e-06,
+      "loss": 0.642,
+      "step": 619000
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 3.999387770720134e-06,
+      "loss": 0.6874,
+      "step": 619010
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 3.998239840820387e-06,
+      "loss": 0.7317,
+      "step": 619020
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 3.997091910920639e-06,
+      "loss": 0.6808,
+      "step": 619030
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 3.995943981020892e-06,
+      "loss": 0.6835,
+      "step": 619040
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 3.994796051121145e-06,
+      "loss": 0.736,
+      "step": 619050
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 3.993648121221397e-06,
+      "loss": 0.721,
+      "step": 619060
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 3.99250019132165e-06,
+      "loss": 0.6488,
+      "step": 619070
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 3.991352261421902e-06,
+      "loss": 0.6982,
+      "step": 619080
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 3.990204331522155e-06,
+      "loss": 0.6461,
+      "step": 619090
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 3.989056401622407e-06,
+      "loss": 0.7104,
+      "step": 619100
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 3.98790847172266e-06,
+      "loss": 0.7479,
+      "step": 619110
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 3.986760541822912e-06,
+      "loss": 0.6668,
+      "step": 619120
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 3.985612611923165e-06,
+      "loss": 0.7828,
+      "step": 619130
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 3.984464682023417e-06,
+      "loss": 0.6868,
+      "step": 619140
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 3.98331675212367e-06,
+      "loss": 0.7137,
+      "step": 619150
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 3.982168822223923e-06,
+      "loss": 0.7754,
+      "step": 619160
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 3.981020892324175e-06,
+      "loss": 0.6834,
+      "step": 619170
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 3.979872962424428e-06,
+      "loss": 0.6387,
+      "step": 619180
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 3.97872503252468e-06,
+      "loss": 0.6243,
+      "step": 619190
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 3.977577102624933e-06,
+      "loss": 0.6523,
+      "step": 619200
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 3.976429172725185e-06,
+      "loss": 0.7041,
+      "step": 619210
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 3.975281242825438e-06,
+      "loss": 0.6434,
+      "step": 619220
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 3.9741333129256905e-06,
+      "loss": 0.7459,
+      "step": 619230
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 3.972985383025943e-06,
+      "loss": 0.668,
+      "step": 619240
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 3.971837453126195e-06,
+      "loss": 0.6137,
+      "step": 619250
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 3.970689523226448e-06,
+      "loss": 0.6629,
+      "step": 619260
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 3.969541593326701e-06,
+      "loss": 0.6156,
+      "step": 619270
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 3.968393663426953e-06,
+      "loss": 0.7055,
+      "step": 619280
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 3.967245733527206e-06,
+      "loss": 0.6293,
+      "step": 619290
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 3.966097803627458e-06,
+      "loss": 0.6152,
+      "step": 619300
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 3.964949873727711e-06,
+      "loss": 0.6547,
+      "step": 619310
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 3.963801943827963e-06,
+      "loss": 0.6715,
+      "step": 619320
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 3.962654013928216e-06,
+      "loss": 0.6958,
+      "step": 619330
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 3.9615060840284685e-06,
+      "loss": 0.6607,
+      "step": 619340
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 3.960358154128721e-06,
+      "loss": 0.7418,
+      "step": 619350
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 3.959210224228973e-06,
+      "loss": 0.7536,
+      "step": 619360
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 3.958062294329226e-06,
+      "loss": 0.6424,
+      "step": 619370
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 3.956914364429479e-06,
+      "loss": 0.6775,
+      "step": 619380
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 3.955766434529731e-06,
+      "loss": 0.7284,
+      "step": 619390
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 3.954618504629984e-06,
+      "loss": 0.7398,
+      "step": 619400
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 3.953470574730236e-06,
+      "loss": 0.6768,
+      "step": 619410
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 3.9523226448304885e-06,
+      "loss": 0.6606,
+      "step": 619420
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 3.951174714930741e-06,
+      "loss": 0.7348,
+      "step": 619430
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 3.950026785030994e-06,
+      "loss": 0.6784,
+      "step": 619440
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 3.9488788551312465e-06,
+      "loss": 0.7135,
+      "step": 619450
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 3.947730925231499e-06,
+      "loss": 0.6212,
+      "step": 619460
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 3.946582995331751e-06,
+      "loss": 0.7918,
+      "step": 619470
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 3.945435065432004e-06,
+      "loss": 0.7419,
+      "step": 619480
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 3.944287135532257e-06,
+      "loss": 0.6472,
+      "step": 619490
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 3.9431392056325085e-06,
+      "loss": 0.6589,
+      "step": 619500
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 3.941991275732762e-06,
+      "loss": 0.658,
+      "step": 619510
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 3.940843345833014e-06,
+      "loss": 0.7043,
+      "step": 619520
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 3.9396954159332666e-06,
+      "loss": 0.7234,
+      "step": 619530
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 3.938547486033519e-06,
+      "loss": 0.73,
+      "step": 619540
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 3.937399556133772e-06,
+      "loss": 0.7,
+      "step": 619550
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 3.936251626234025e-06,
+      "loss": 0.6991,
+      "step": 619560
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 3.935103696334277e-06,
+      "loss": 0.7429,
+      "step": 619570
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 3.933955766434529e-06,
+      "loss": 0.6324,
+      "step": 619580
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 3.932807836534782e-06,
+      "loss": 0.7175,
+      "step": 619590
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 3.931659906635035e-06,
+      "loss": 0.7009,
+      "step": 619600
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 3.9305119767352866e-06,
+      "loss": 0.7115,
+      "step": 619610
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 3.92936404683554e-06,
+      "loss": 0.6792,
+      "step": 619620
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 3.928216116935792e-06,
+      "loss": 0.6306,
+      "step": 619630
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 3.927068187036045e-06,
+      "loss": 0.6531,
+      "step": 619640
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 3.925920257136297e-06,
+      "loss": 0.691,
+      "step": 619650
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 3.92477232723655e-06,
+      "loss": 0.6567,
+      "step": 619660
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 3.923624397336803e-06,
+      "loss": 0.6945,
+      "step": 619670
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 3.922476467437055e-06,
+      "loss": 0.6571,
+      "step": 619680
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 3.9213285375373074e-06,
+      "loss": 0.7009,
+      "step": 619690
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 3.92018060763756e-06,
+      "loss": 0.7113,
+      "step": 619700
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 3.919032677737813e-06,
+      "loss": 0.689,
+      "step": 619710
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 3.917884747838065e-06,
+      "loss": 0.7036,
+      "step": 619720
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 3.916736817938318e-06,
+      "loss": 0.7307,
+      "step": 619730
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 3.91558888803857e-06,
+      "loss": 0.6183,
+      "step": 619740
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 3.914440958138823e-06,
+      "loss": 0.6636,
+      "step": 619750
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 3.913293028239075e-06,
+      "loss": 0.6731,
+      "step": 619760
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 3.9121450983393274e-06,
+      "loss": 0.6594,
+      "step": 619770
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 3.910997168439581e-06,
+      "loss": 0.6928,
+      "step": 619780
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 3.909849238539833e-06,
+      "loss": 0.6664,
+      "step": 619790
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 3.9087013086400855e-06,
+      "loss": 0.7188,
+      "step": 619800
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 3.907553378740338e-06,
+      "loss": 0.6357,
+      "step": 619810
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 3.906405448840591e-06,
+      "loss": 0.6723,
+      "step": 619820
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 3.905257518940843e-06,
+      "loss": 0.7222,
+      "step": 619830
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 3.904109589041096e-06,
+      "loss": 0.6697,
+      "step": 619840
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 3.9029616591413474e-06,
+      "loss": 0.6609,
+      "step": 619850
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 3.901813729241601e-06,
+      "loss": 0.6237,
+      "step": 619860
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 3.900665799341853e-06,
+      "loss": 0.6141,
+      "step": 619870
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 3.8995178694421055e-06,
+      "loss": 0.6683,
+      "step": 619880
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 3.898369939542358e-06,
+      "loss": 0.7195,
+      "step": 619890
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 3.897222009642611e-06,
+      "loss": 0.6728,
+      "step": 619900
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 3.8960740797428635e-06,
+      "loss": 0.71,
+      "step": 619910
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 3.894926149843116e-06,
+      "loss": 0.7165,
+      "step": 619920
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 3.893778219943368e-06,
+      "loss": 0.6786,
+      "step": 619930
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 3.892630290043621e-06,
+      "loss": 0.7399,
+      "step": 619940
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 3.891482360143874e-06,
+      "loss": 0.7563,
+      "step": 619950
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 3.8903344302441255e-06,
+      "loss": 0.7276,
+      "step": 619960
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 3.889186500344379e-06,
+      "loss": 0.7121,
+      "step": 619970
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 3.888038570444631e-06,
+      "loss": 0.7202,
+      "step": 619980
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 3.8868906405448835e-06,
+      "loss": 0.6406,
+      "step": 619990
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 3.885742710645136e-06,
+      "loss": 0.6946,
+      "step": 620000
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 3.884594780745388e-06,
+      "loss": 0.6449,
+      "step": 620010
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 3.8834468508456416e-06,
+      "loss": 0.7297,
+      "step": 620020
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 3.882298920945894e-06,
+      "loss": 0.6879,
+      "step": 620030
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 3.881150991046146e-06,
+      "loss": 0.6915,
+      "step": 620040
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 3.880003061146399e-06,
+      "loss": 0.7357,
+      "step": 620050
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 3.878855131246652e-06,
+      "loss": 0.6178,
+      "step": 620060
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 3.8777072013469035e-06,
+      "loss": 0.6034,
+      "step": 620070
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 3.876559271447157e-06,
+      "loss": 0.6619,
+      "step": 620080
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 3.875411341547409e-06,
+      "loss": 0.753,
+      "step": 620090
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 3.8742634116476616e-06,
+      "loss": 0.6957,
+      "step": 620100
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 3.873115481747914e-06,
+      "loss": 0.6022,
+      "step": 620110
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 3.871967551848166e-06,
+      "loss": 0.7167,
+      "step": 620120
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 3.87081962194842e-06,
+      "loss": 0.7119,
+      "step": 620130
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 3.869671692048672e-06,
+      "loss": 0.776,
+      "step": 620140
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 3.868523762148924e-06,
+      "loss": 0.665,
+      "step": 620150
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 3.867375832249177e-06,
+      "loss": 0.7106,
+      "step": 620160
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 3.86622790234943e-06,
+      "loss": 0.6644,
+      "step": 620170
+    },
+    {
+      "epoch": 28.45,
+      "learning_rate": 3.8650799724496816e-06,
+      "loss": 0.5991,
+      "step": 620180
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 3.863932042549935e-06,
+      "loss": 0.6539,
+      "step": 620190
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 3.862784112650187e-06,
+      "loss": 0.6763,
+      "step": 620200
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 3.86163618275044e-06,
+      "loss": 0.67,
+      "step": 620210
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 3.860488252850692e-06,
+      "loss": 0.6698,
+      "step": 620220
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 3.859340322950944e-06,
+      "loss": 0.6725,
+      "step": 620230
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 3.858192393051198e-06,
+      "loss": 0.6419,
+      "step": 620240
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 3.85704446315145e-06,
+      "loss": 0.7078,
+      "step": 620250
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 3.8558965332517024e-06,
+      "loss": 0.7388,
+      "step": 620260
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 3.854748603351955e-06,
+      "loss": 0.6841,
+      "step": 620270
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 3.853600673452207e-06,
+      "loss": 0.6047,
+      "step": 620280
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 3.85245274355246e-06,
+      "loss": 0.7571,
+      "step": 620290
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 3.851304813652713e-06,
+      "loss": 0.6832,
+      "step": 620300
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 3.850156883752965e-06,
+      "loss": 0.6895,
+      "step": 620310
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 3.849008953853218e-06,
+      "loss": 0.7049,
+      "step": 620320
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 3.84786102395347e-06,
+      "loss": 0.6198,
+      "step": 620330
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 3.8467130940537224e-06,
+      "loss": 0.7253,
+      "step": 620340
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 3.845565164153976e-06,
+      "loss": 0.6907,
+      "step": 620350
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 3.844417234254227e-06,
+      "loss": 0.68,
+      "step": 620360
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 3.8432693043544805e-06,
+      "loss": 0.6422,
+      "step": 620370
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 3.842121374454733e-06,
+      "loss": 0.6709,
+      "step": 620380
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 3.840973444554985e-06,
+      "loss": 0.7104,
+      "step": 620390
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 3.839825514655238e-06,
+      "loss": 0.7313,
+      "step": 620400
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 3.838677584755491e-06,
+      "loss": 0.6784,
+      "step": 620410
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 3.837529654855743e-06,
+      "loss": 0.6875,
+      "step": 620420
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 3.836381724955996e-06,
+      "loss": 0.7284,
+      "step": 620430
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 3.835233795056248e-06,
+      "loss": 0.6381,
+      "step": 620440
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 3.8340858651565005e-06,
+      "loss": 0.6677,
+      "step": 620450
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 3.832937935256754e-06,
+      "loss": 0.5847,
+      "step": 620460
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 3.831790005357005e-06,
+      "loss": 0.6463,
+      "step": 620470
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 3.8306420754572585e-06,
+      "loss": 0.6794,
+      "step": 620480
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 3.829494145557511e-06,
+      "loss": 0.702,
+      "step": 620490
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 3.828346215657763e-06,
+      "loss": 0.7024,
+      "step": 620500
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 3.827198285758016e-06,
+      "loss": 0.7201,
+      "step": 620510
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 3.826050355858269e-06,
+      "loss": 0.7015,
+      "step": 620520
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 3.824902425958521e-06,
+      "loss": 0.6404,
+      "step": 620530
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 3.823754496058774e-06,
+      "loss": 0.6222,
+      "step": 620540
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 3.822606566159026e-06,
+      "loss": 0.7255,
+      "step": 620550
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 3.8214586362592785e-06,
+      "loss": 0.6488,
+      "step": 620560
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 3.820310706359532e-06,
+      "loss": 0.6176,
+      "step": 620570
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 3.819162776459783e-06,
+      "loss": 0.6281,
+      "step": 620580
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 3.8180148465600366e-06,
+      "loss": 0.6875,
+      "step": 620590
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 3.816866916660289e-06,
+      "loss": 0.6799,
+      "step": 620600
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 3.815718986760541e-06,
+      "loss": 0.6966,
+      "step": 620610
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 3.814571056860794e-06,
+      "loss": 0.6846,
+      "step": 620620
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 3.8134231269610466e-06,
+      "loss": 0.722,
+      "step": 620630
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 3.812275197061299e-06,
+      "loss": 0.6746,
+      "step": 620640
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 3.8111272671615518e-06,
+      "loss": 0.7476,
+      "step": 620650
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 3.809979337261804e-06,
+      "loss": 0.6724,
+      "step": 620660
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 3.808831407362057e-06,
+      "loss": 0.7746,
+      "step": 620670
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 3.8076834774623094e-06,
+      "loss": 0.7007,
+      "step": 620680
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 3.8065355475625618e-06,
+      "loss": 0.7728,
+      "step": 620690
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 3.8053876176628146e-06,
+      "loss": 0.7127,
+      "step": 620700
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 3.8042396877630666e-06,
+      "loss": 0.6226,
+      "step": 620710
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 3.8030917578633194e-06,
+      "loss": 0.6667,
+      "step": 620720
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 3.8019438279635722e-06,
+      "loss": 0.6796,
+      "step": 620730
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 3.8007958980638246e-06,
+      "loss": 0.6792,
+      "step": 620740
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 3.799647968164077e-06,
+      "loss": 0.6341,
+      "step": 620750
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 3.79850003826433e-06,
+      "loss": 0.7028,
+      "step": 620760
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 3.7973521083645822e-06,
+      "loss": 0.6597,
+      "step": 620770
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 3.796204178464835e-06,
+      "loss": 0.7156,
+      "step": 620780
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 3.795056248565087e-06,
+      "loss": 0.7137,
+      "step": 620790
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 3.79390831866534e-06,
+      "loss": 0.6772,
+      "step": 620800
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 3.7927603887655926e-06,
+      "loss": 0.6112,
+      "step": 620810
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 3.7916124588658446e-06,
+      "loss": 0.6961,
+      "step": 620820
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 3.7904645289660974e-06,
+      "loss": 0.7373,
+      "step": 620830
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 3.7893165990663503e-06,
+      "loss": 0.6868,
+      "step": 620840
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 3.7881686691666027e-06,
+      "loss": 0.7152,
+      "step": 620850
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 3.787020739266855e-06,
+      "loss": 0.7213,
+      "step": 620860
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 3.7858728093671074e-06,
+      "loss": 0.6705,
+      "step": 620870
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 3.7847248794673603e-06,
+      "loss": 0.7069,
+      "step": 620880
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 3.783576949567613e-06,
+      "loss": 0.6558,
+      "step": 620890
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 3.782429019667865e-06,
+      "loss": 0.6812,
+      "step": 620900
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 3.781281089768118e-06,
+      "loss": 0.6914,
+      "step": 620910
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 3.7801331598683707e-06,
+      "loss": 0.7165,
+      "step": 620920
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 3.7789852299686227e-06,
+      "loss": 0.74,
+      "step": 620930
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 3.7778373000688755e-06,
+      "loss": 0.6394,
+      "step": 620940
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 3.7766893701691283e-06,
+      "loss": 0.7607,
+      "step": 620950
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 3.7755414402693803e-06,
+      "loss": 0.6797,
+      "step": 620960
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 3.774393510369633e-06,
+      "loss": 0.6935,
+      "step": 620970
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 3.7732455804698855e-06,
+      "loss": 0.7133,
+      "step": 620980
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 3.7720976505701383e-06,
+      "loss": 0.6819,
+      "step": 620990
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 3.770949720670391e-06,
+      "loss": 0.6748,
+      "step": 621000
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 3.769801790770643e-06,
+      "loss": 0.6742,
+      "step": 621010
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 3.768653860870896e-06,
+      "loss": 0.6848,
+      "step": 621020
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 3.7675059309711487e-06,
+      "loss": 0.7222,
+      "step": 621030
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 3.7663580010714007e-06,
+      "loss": 0.6192,
+      "step": 621040
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 3.7652100711716535e-06,
+      "loss": 0.6895,
+      "step": 621050
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 3.764062141271906e-06,
+      "loss": 0.632,
+      "step": 621060
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 3.7629142113721583e-06,
+      "loss": 0.5808,
+      "step": 621070
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 3.761766281472411e-06,
+      "loss": 0.7071,
+      "step": 621080
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 3.7606183515726635e-06,
+      "loss": 0.6958,
+      "step": 621090
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 3.7594704216729164e-06,
+      "loss": 0.7014,
+      "step": 621100
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 3.7583224917731687e-06,
+      "loss": 0.6578,
+      "step": 621110
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 3.757174561873421e-06,
+      "loss": 0.7193,
+      "step": 621120
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 3.756026631973674e-06,
+      "loss": 0.6916,
+      "step": 621130
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 3.754878702073926e-06,
+      "loss": 0.7098,
+      "step": 621140
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 3.7537307721741788e-06,
+      "loss": 0.632,
+      "step": 621150
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 3.7525828422744316e-06,
+      "loss": 0.7184,
+      "step": 621160
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 3.751434912374684e-06,
+      "loss": 0.7265,
+      "step": 621170
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 3.7502869824749364e-06,
+      "loss": 0.6629,
+      "step": 621180
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 3.749139052575189e-06,
+      "loss": 0.7043,
+      "step": 621190
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 3.7479911226754416e-06,
+      "loss": 0.6361,
+      "step": 621200
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 3.7468431927756944e-06,
+      "loss": 0.7167,
+      "step": 621210
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 3.7456952628759464e-06,
+      "loss": 0.6207,
+      "step": 621220
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 3.744547332976199e-06,
+      "loss": 0.7079,
+      "step": 621230
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 3.743399403076452e-06,
+      "loss": 0.6611,
+      "step": 621240
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 3.742251473176704e-06,
+      "loss": 0.6775,
+      "step": 621250
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 3.741103543276957e-06,
+      "loss": 0.6782,
+      "step": 621260
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 3.7399556133772096e-06,
+      "loss": 0.7685,
+      "step": 621270
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 3.738807683477462e-06,
+      "loss": 0.6681,
+      "step": 621280
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 3.7376597535777144e-06,
+      "loss": 0.5815,
+      "step": 621290
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 3.736511823677967e-06,
+      "loss": 0.7731,
+      "step": 621300
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 3.7353638937782196e-06,
+      "loss": 0.6462,
+      "step": 621310
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 3.7342159638784724e-06,
+      "loss": 0.7104,
+      "step": 621320
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 3.7330680339787244e-06,
+      "loss": 0.6763,
+      "step": 621330
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 3.7319201040789772e-06,
+      "loss": 0.6391,
+      "step": 621340
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 3.73077217417923e-06,
+      "loss": 0.6595,
+      "step": 621350
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 3.729624244279482e-06,
+      "loss": 0.6718,
+      "step": 621360
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 3.728476314379735e-06,
+      "loss": 0.6967,
+      "step": 621370
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 3.7273283844799872e-06,
+      "loss": 0.6063,
+      "step": 621380
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 3.72618045458024e-06,
+      "loss": 0.6842,
+      "step": 621390
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 3.7250325246804924e-06,
+      "loss": 0.7059,
+      "step": 621400
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 3.723884594780745e-06,
+      "loss": 0.6588,
+      "step": 621410
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 3.7227366648809977e-06,
+      "loss": 0.7017,
+      "step": 621420
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 3.7215887349812505e-06,
+      "loss": 0.6819,
+      "step": 621430
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 3.7204408050815025e-06,
+      "loss": 0.6631,
+      "step": 621440
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 3.7192928751817553e-06,
+      "loss": 0.6996,
+      "step": 621450
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 3.718144945282008e-06,
+      "loss": 0.6976,
+      "step": 621460
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 3.71699701538226e-06,
+      "loss": 0.7391,
+      "step": 621470
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 3.715849085482513e-06,
+      "loss": 0.7157,
+      "step": 621480
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 3.7147011555827653e-06,
+      "loss": 0.7131,
+      "step": 621490
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 3.713553225683018e-06,
+      "loss": 0.5989,
+      "step": 621500
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 3.7124052957832705e-06,
+      "loss": 0.6257,
+      "step": 621510
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 3.711257365883523e-06,
+      "loss": 0.696,
+      "step": 621520
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 3.7101094359837757e-06,
+      "loss": 0.6924,
+      "step": 621530
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 3.7089615060840285e-06,
+      "loss": 0.7458,
+      "step": 621540
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 3.7078135761842805e-06,
+      "loss": 0.7292,
+      "step": 621550
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 3.7066656462845333e-06,
+      "loss": 0.6244,
+      "step": 621560
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 3.7055177163847857e-06,
+      "loss": 0.6726,
+      "step": 621570
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 3.704369786485038e-06,
+      "loss": 0.6986,
+      "step": 621580
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 3.703221856585291e-06,
+      "loss": 0.7001,
+      "step": 621590
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 3.7020739266855433e-06,
+      "loss": 0.6735,
+      "step": 621600
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 3.700925996785796e-06,
+      "loss": 0.7217,
+      "step": 621610
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 3.6997780668860485e-06,
+      "loss": 0.7244,
+      "step": 621620
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 3.698630136986301e-06,
+      "loss": 0.7128,
+      "step": 621630
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 3.6974822070865537e-06,
+      "loss": 0.6801,
+      "step": 621640
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 3.6963342771868057e-06,
+      "loss": 0.7789,
+      "step": 621650
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 3.6951863472870585e-06,
+      "loss": 0.6528,
+      "step": 621660
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 3.6940384173873114e-06,
+      "loss": 0.7271,
+      "step": 621670
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 3.6928904874875638e-06,
+      "loss": 0.7571,
+      "step": 621680
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 3.691742557587816e-06,
+      "loss": 0.653,
+      "step": 621690
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 3.690594627688069e-06,
+      "loss": 0.6894,
+      "step": 621700
+    },
+    {
+      "epoch": 28.52,
+      "learning_rate": 3.6894466977883214e-06,
+      "loss": 0.7203,
+      "step": 621710
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 3.688298767888574e-06,
+      "loss": 0.6858,
+      "step": 621720
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 3.687150837988826e-06,
+      "loss": 0.8172,
+      "step": 621730
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 3.686002908089079e-06,
+      "loss": 0.6428,
+      "step": 621740
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 3.684854978189332e-06,
+      "loss": 0.6382,
+      "step": 621750
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 3.6837070482895838e-06,
+      "loss": 0.7278,
+      "step": 621760
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 3.6825591183898366e-06,
+      "loss": 0.696,
+      "step": 621770
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 3.6814111884900894e-06,
+      "loss": 0.6628,
+      "step": 621780
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 3.680263258590342e-06,
+      "loss": 0.6817,
+      "step": 621790
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 3.679115328690594e-06,
+      "loss": 0.7429,
+      "step": 621800
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 3.6779673987908466e-06,
+      "loss": 0.6635,
+      "step": 621810
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 3.6768194688910994e-06,
+      "loss": 0.7028,
+      "step": 621820
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 3.6756715389913522e-06,
+      "loss": 0.6743,
+      "step": 621830
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 3.674523609091604e-06,
+      "loss": 0.6738,
+      "step": 621840
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 3.673375679191857e-06,
+      "loss": 0.6048,
+      "step": 621850
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 3.67222774929211e-06,
+      "loss": 0.8186,
+      "step": 621860
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 3.671079819392362e-06,
+      "loss": 0.6624,
+      "step": 621870
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 3.6699318894926146e-06,
+      "loss": 0.6652,
+      "step": 621880
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 3.6687839595928674e-06,
+      "loss": 0.6039,
+      "step": 621890
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 3.66763602969312e-06,
+      "loss": 0.7296,
+      "step": 621900
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 3.6664880997933722e-06,
+      "loss": 0.7195,
+      "step": 621910
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 3.6653401698936246e-06,
+      "loss": 0.6569,
+      "step": 621920
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 3.6641922399938775e-06,
+      "loss": 0.6627,
+      "step": 621930
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 3.6630443100941303e-06,
+      "loss": 0.7353,
+      "step": 621940
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 3.6618963801943822e-06,
+      "loss": 0.7047,
+      "step": 621950
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 3.660748450294635e-06,
+      "loss": 0.6903,
+      "step": 621960
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 3.659600520394888e-06,
+      "loss": 0.7449,
+      "step": 621970
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 3.65845259049514e-06,
+      "loss": 0.7295,
+      "step": 621980
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 3.6573046605953927e-06,
+      "loss": 0.7171,
+      "step": 621990
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 3.656156730695645e-06,
+      "loss": 0.6999,
+      "step": 622000
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 3.655008800795898e-06,
+      "loss": 0.6837,
+      "step": 622010
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 3.6538608708961503e-06,
+      "loss": 0.7043,
+      "step": 622020
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 3.6527129409964027e-06,
+      "loss": 0.6917,
+      "step": 622030
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 3.6515650110966555e-06,
+      "loss": 0.6261,
+      "step": 622040
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 3.6504170811969083e-06,
+      "loss": 0.7251,
+      "step": 622050
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 3.6492691512971603e-06,
+      "loss": 0.7016,
+      "step": 622060
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 3.648121221397413e-06,
+      "loss": 0.6802,
+      "step": 622070
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 3.6469732914976655e-06,
+      "loss": 0.7383,
+      "step": 622080
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 3.645825361597918e-06,
+      "loss": 0.5784,
+      "step": 622090
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 3.6446774316981707e-06,
+      "loss": 0.6337,
+      "step": 622100
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 3.643529501798423e-06,
+      "loss": 0.6903,
+      "step": 622110
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 3.642381571898676e-06,
+      "loss": 0.6653,
+      "step": 622120
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 3.6412336419989283e-06,
+      "loss": 0.7213,
+      "step": 622130
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 3.6400857120991807e-06,
+      "loss": 0.7946,
+      "step": 622140
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 3.6389377821994335e-06,
+      "loss": 0.638,
+      "step": 622150
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 3.6377898522996855e-06,
+      "loss": 0.6944,
+      "step": 622160
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 3.6366419223999383e-06,
+      "loss": 0.7054,
+      "step": 622170
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 3.635493992500191e-06,
+      "loss": 0.6946,
+      "step": 622180
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 3.6343460626004435e-06,
+      "loss": 0.6827,
+      "step": 622190
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 3.633198132700696e-06,
+      "loss": 0.6548,
+      "step": 622200
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 3.6320502028009488e-06,
+      "loss": 0.7173,
+      "step": 622210
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 3.630902272901201e-06,
+      "loss": 0.6238,
+      "step": 622220
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 3.629754343001454e-06,
+      "loss": 0.6921,
+      "step": 622230
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 3.628606413101706e-06,
+      "loss": 0.7033,
+      "step": 622240
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 3.6274584832019588e-06,
+      "loss": 0.6778,
+      "step": 622250
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 3.6263105533022116e-06,
+      "loss": 0.6862,
+      "step": 622260
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 3.6251626234024636e-06,
+      "loss": 0.701,
+      "step": 622270
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 3.6240146935027164e-06,
+      "loss": 0.6751,
+      "step": 622280
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 3.622866763602969e-06,
+      "loss": 0.679,
+      "step": 622290
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 3.6217188337032216e-06,
+      "loss": 0.683,
+      "step": 622300
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 3.620570903803474e-06,
+      "loss": 0.7193,
+      "step": 622310
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 3.619422973903727e-06,
+      "loss": 0.6374,
+      "step": 622320
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 3.618275044003979e-06,
+      "loss": 0.6297,
+      "step": 622330
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 3.617127114104232e-06,
+      "loss": 0.68,
+      "step": 622340
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 3.615979184204484e-06,
+      "loss": 0.6923,
+      "step": 622350
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 3.614831254304737e-06,
+      "loss": 0.6573,
+      "step": 622360
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 3.6136833244049896e-06,
+      "loss": 0.6305,
+      "step": 622370
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 3.6125353945052416e-06,
+      "loss": 0.6537,
+      "step": 622380
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 3.6113874646054944e-06,
+      "loss": 0.6648,
+      "step": 622390
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 3.6102395347057472e-06,
+      "loss": 0.7169,
+      "step": 622400
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 3.6090916048059996e-06,
+      "loss": 0.7015,
+      "step": 622410
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 3.607943674906252e-06,
+      "loss": 0.6869,
+      "step": 622420
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 3.6067957450065044e-06,
+      "loss": 0.7182,
+      "step": 622430
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 3.6056478151067572e-06,
+      "loss": 0.6696,
+      "step": 622440
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 3.60449988520701e-06,
+      "loss": 0.6728,
+      "step": 622450
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 3.603351955307262e-06,
+      "loss": 0.6117,
+      "step": 622460
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 3.602204025407515e-06,
+      "loss": 0.6622,
+      "step": 622470
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 3.6010560955077677e-06,
+      "loss": 0.6741,
+      "step": 622480
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 3.5999081656080196e-06,
+      "loss": 0.7041,
+      "step": 622490
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 3.5987602357082725e-06,
+      "loss": 0.7225,
+      "step": 622500
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 3.597612305808525e-06,
+      "loss": 0.6448,
+      "step": 622510
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 3.5964643759087777e-06,
+      "loss": 0.7146,
+      "step": 622520
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 3.59531644600903e-06,
+      "loss": 0.6769,
+      "step": 622530
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 3.5941685161092825e-06,
+      "loss": 0.701,
+      "step": 622540
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 3.5930205862095353e-06,
+      "loss": 0.6047,
+      "step": 622550
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 3.591872656309788e-06,
+      "loss": 0.7538,
+      "step": 622560
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 3.59072472641004e-06,
+      "loss": 0.6497,
+      "step": 622570
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 3.589576796510293e-06,
+      "loss": 0.7062,
+      "step": 622580
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 3.5884288666105453e-06,
+      "loss": 0.6294,
+      "step": 622590
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 3.5872809367107977e-06,
+      "loss": 0.7404,
+      "step": 622600
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 3.5861330068110505e-06,
+      "loss": 0.6755,
+      "step": 622610
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 3.584985076911303e-06,
+      "loss": 0.6977,
+      "step": 622620
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 3.5838371470115557e-06,
+      "loss": 0.7454,
+      "step": 622630
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 3.582689217111808e-06,
+      "loss": 0.6595,
+      "step": 622640
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 3.5815412872120605e-06,
+      "loss": 0.6395,
+      "step": 622650
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 3.5803933573123133e-06,
+      "loss": 0.683,
+      "step": 622660
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 3.5792454274125653e-06,
+      "loss": 0.6816,
+      "step": 622670
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 3.578097497512818e-06,
+      "loss": 0.7253,
+      "step": 622680
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 3.576949567613071e-06,
+      "loss": 0.7026,
+      "step": 622690
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 3.5758016377133233e-06,
+      "loss": 0.7306,
+      "step": 622700
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 3.5746537078135757e-06,
+      "loss": 0.6612,
+      "step": 622710
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 3.5735057779138285e-06,
+      "loss": 0.641,
+      "step": 622720
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 3.572357848014081e-06,
+      "loss": 0.6016,
+      "step": 622730
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 3.5712099181143338e-06,
+      "loss": 0.6355,
+      "step": 622740
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 3.570061988214586e-06,
+      "loss": 0.7632,
+      "step": 622750
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 3.5689140583148385e-06,
+      "loss": 0.6982,
+      "step": 622760
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 3.5677661284150914e-06,
+      "loss": 0.7676,
+      "step": 622770
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 3.5666181985153433e-06,
+      "loss": 0.7144,
+      "step": 622780
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 3.565470268615596e-06,
+      "loss": 0.7639,
+      "step": 622790
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 3.564322338715849e-06,
+      "loss": 0.602,
+      "step": 622800
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 3.5631744088161014e-06,
+      "loss": 0.6701,
+      "step": 622810
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 3.5620264789163538e-06,
+      "loss": 0.7063,
+      "step": 622820
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 3.5608785490166066e-06,
+      "loss": 0.6386,
+      "step": 622830
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 3.559730619116859e-06,
+      "loss": 0.7126,
+      "step": 622840
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 3.558582689217112e-06,
+      "loss": 0.6732,
+      "step": 622850
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 3.5574347593173638e-06,
+      "loss": 0.6606,
+      "step": 622860
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 3.5562868294176166e-06,
+      "loss": 0.6784,
+      "step": 622870
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 3.5551388995178694e-06,
+      "loss": 0.6966,
+      "step": 622880
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 3.5539909696181214e-06,
+      "loss": 0.6973,
+      "step": 622890
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 3.552843039718374e-06,
+      "loss": 0.6356,
+      "step": 622900
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 3.551695109818627e-06,
+      "loss": 0.6841,
+      "step": 622910
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 3.5505471799188794e-06,
+      "loss": 0.6522,
+      "step": 622920
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 3.549399250019132e-06,
+      "loss": 0.6502,
+      "step": 622930
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 3.548251320119384e-06,
+      "loss": 0.7179,
+      "step": 622940
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 3.547103390219637e-06,
+      "loss": 0.7361,
+      "step": 622950
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 3.54595546031989e-06,
+      "loss": 0.6439,
+      "step": 622960
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 3.544807530420142e-06,
+      "loss": 0.6395,
+      "step": 622970
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 3.5436596005203946e-06,
+      "loss": 0.6757,
+      "step": 622980
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 3.5425116706206475e-06,
+      "loss": 0.6713,
+      "step": 622990
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 3.5413637407208994e-06,
+      "loss": 0.6986,
+      "step": 623000
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 3.5402158108211522e-06,
+      "loss": 0.6306,
+      "step": 623010
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 3.5390678809214046e-06,
+      "loss": 0.7321,
+      "step": 623020
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 3.5379199510216575e-06,
+      "loss": 0.6978,
+      "step": 623030
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 3.53677202112191e-06,
+      "loss": 0.7133,
+      "step": 623040
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 3.5356240912221623e-06,
+      "loss": 0.7297,
+      "step": 623050
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 3.534476161322415e-06,
+      "loss": 0.793,
+      "step": 623060
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 3.533328231422668e-06,
+      "loss": 0.7113,
+      "step": 623070
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 3.53218030152292e-06,
+      "loss": 0.7877,
+      "step": 623080
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 3.5310323716231727e-06,
+      "loss": 0.7438,
+      "step": 623090
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 3.5298844417234247e-06,
+      "loss": 0.6924,
+      "step": 623100
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 3.5287365118236775e-06,
+      "loss": 0.6901,
+      "step": 623110
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 3.5275885819239303e-06,
+      "loss": 0.7324,
+      "step": 623120
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 3.5264406520241827e-06,
+      "loss": 0.7175,
+      "step": 623130
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 3.5252927221244355e-06,
+      "loss": 0.6597,
+      "step": 623140
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 3.524144792224688e-06,
+      "loss": 0.6993,
+      "step": 623150
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 3.5229968623249403e-06,
+      "loss": 0.7015,
+      "step": 623160
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 3.521848932425193e-06,
+      "loss": 0.671,
+      "step": 623170
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 3.520701002525446e-06,
+      "loss": 0.7097,
+      "step": 623180
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 3.519553072625698e-06,
+      "loss": 0.6133,
+      "step": 623190
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 3.5184051427259507e-06,
+      "loss": 0.7377,
+      "step": 623200
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 3.5172572128262027e-06,
+      "loss": 0.7113,
+      "step": 623210
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 3.5161092829264555e-06,
+      "loss": 0.6926,
+      "step": 623220
+    },
+    {
+      "epoch": 28.59,
+      "learning_rate": 3.5149613530267083e-06,
+      "loss": 0.6276,
+      "step": 623230
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 3.5138134231269607e-06,
+      "loss": 0.7421,
+      "step": 623240
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 3.5126654932272135e-06,
+      "loss": 0.6463,
+      "step": 623250
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 3.511517563327466e-06,
+      "loss": 0.6714,
+      "step": 623260
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 3.5103696334277183e-06,
+      "loss": 0.695,
+      "step": 623270
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 3.509221703527971e-06,
+      "loss": 0.7131,
+      "step": 623280
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 3.508073773628223e-06,
+      "loss": 0.6572,
+      "step": 623290
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 3.506925843728476e-06,
+      "loss": 0.704,
+      "step": 623300
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 3.5057779138287288e-06,
+      "loss": 0.6582,
+      "step": 623310
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 3.5046299839289807e-06,
+      "loss": 0.7169,
+      "step": 623320
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 3.5034820540292336e-06,
+      "loss": 0.6467,
+      "step": 623330
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 3.5023341241294864e-06,
+      "loss": 0.7183,
+      "step": 623340
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 3.5011861942297388e-06,
+      "loss": 0.6447,
+      "step": 623350
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 3.500038264329991e-06,
+      "loss": 0.6966,
+      "step": 623360
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 3.4988903344302436e-06,
+      "loss": 0.6581,
+      "step": 623370
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 3.4977424045304964e-06,
+      "loss": 0.6979,
+      "step": 623380
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 3.496594474630749e-06,
+      "loss": 0.6865,
+      "step": 623390
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 3.495446544731001e-06,
+      "loss": 0.7027,
+      "step": 623400
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 3.494298614831254e-06,
+      "loss": 0.7027,
+      "step": 623410
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 3.493150684931507e-06,
+      "loss": 0.7825,
+      "step": 623420
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 3.4920027550317588e-06,
+      "loss": 0.6559,
+      "step": 623430
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 3.4908548251320116e-06,
+      "loss": 0.6882,
+      "step": 623440
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 3.489706895232264e-06,
+      "loss": 0.6144,
+      "step": 623450
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 3.488558965332517e-06,
+      "loss": 0.7122,
+      "step": 623460
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 3.487411035432769e-06,
+      "loss": 0.7067,
+      "step": 623470
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 3.4862631055330216e-06,
+      "loss": 0.6905,
+      "step": 623480
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 3.4851151756332744e-06,
+      "loss": 0.6712,
+      "step": 623490
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 3.4839672457335272e-06,
+      "loss": 0.7366,
+      "step": 623500
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 3.4828193158337792e-06,
+      "loss": 0.6706,
+      "step": 623510
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 3.481671385934032e-06,
+      "loss": 0.5696,
+      "step": 623520
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 3.4805234560342844e-06,
+      "loss": 0.5938,
+      "step": 623530
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 3.479375526134537e-06,
+      "loss": 0.6185,
+      "step": 623540
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 3.4782275962347896e-06,
+      "loss": 0.6397,
+      "step": 623550
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 3.477079666335042e-06,
+      "loss": 0.6982,
+      "step": 623560
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 3.475931736435295e-06,
+      "loss": 0.6734,
+      "step": 623570
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 3.4747838065355473e-06,
+      "loss": 0.7582,
+      "step": 623580
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 3.4736358766357996e-06,
+      "loss": 0.7465,
+      "step": 623590
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 3.4724879467360525e-06,
+      "loss": 0.674,
+      "step": 623600
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 3.4713400168363053e-06,
+      "loss": 0.6963,
+      "step": 623610
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 3.4701920869365573e-06,
+      "loss": 0.7248,
+      "step": 623620
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 3.46904415703681e-06,
+      "loss": 0.668,
+      "step": 623630
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 3.4678962271370625e-06,
+      "loss": 0.666,
+      "step": 623640
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 3.466748297237315e-06,
+      "loss": 0.6523,
+      "step": 623650
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 3.4656003673375677e-06,
+      "loss": 0.6822,
+      "step": 623660
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 3.46445243743782e-06,
+      "loss": 0.703,
+      "step": 623670
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 3.463304507538073e-06,
+      "loss": 0.6814,
+      "step": 623680
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 3.4621565776383253e-06,
+      "loss": 0.6439,
+      "step": 623690
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 3.4610086477385777e-06,
+      "loss": 0.7199,
+      "step": 623700
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 3.4598607178388305e-06,
+      "loss": 0.686,
+      "step": 623710
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 3.4587127879390825e-06,
+      "loss": 0.6987,
+      "step": 623720
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 3.4575648580393353e-06,
+      "loss": 0.7152,
+      "step": 623730
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 3.456416928139588e-06,
+      "loss": 0.7732,
+      "step": 623740
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 3.4552689982398405e-06,
+      "loss": 0.695,
+      "step": 623750
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 3.454121068340093e-06,
+      "loss": 0.7564,
+      "step": 623760
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 3.4529731384403457e-06,
+      "loss": 0.6459,
+      "step": 623770
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 3.451825208540598e-06,
+      "loss": 0.6928,
+      "step": 623780
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 3.450677278640851e-06,
+      "loss": 0.6891,
+      "step": 623790
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 3.449529348741103e-06,
+      "loss": 0.7065,
+      "step": 623800
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 3.4483814188413557e-06,
+      "loss": 0.6664,
+      "step": 623810
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 3.4472334889416086e-06,
+      "loss": 0.6723,
+      "step": 623820
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 3.4460855590418605e-06,
+      "loss": 0.6147,
+      "step": 623830
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 3.4449376291421133e-06,
+      "loss": 0.7393,
+      "step": 623840
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 3.443789699242366e-06,
+      "loss": 0.673,
+      "step": 623850
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 3.4426417693426186e-06,
+      "loss": 0.7014,
+      "step": 623860
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 3.441493839442871e-06,
+      "loss": 0.66,
+      "step": 623870
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 3.4403459095431234e-06,
+      "loss": 0.6782,
+      "step": 623880
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 3.439197979643376e-06,
+      "loss": 0.7406,
+      "step": 623890
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 3.438050049743629e-06,
+      "loss": 0.7353,
+      "step": 623900
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 3.436902119843881e-06,
+      "loss": 0.6502,
+      "step": 623910
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 3.4357541899441338e-06,
+      "loss": 0.6437,
+      "step": 623920
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 3.4346062600443866e-06,
+      "loss": 0.679,
+      "step": 623930
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 3.4334583301446386e-06,
+      "loss": 0.7894,
+      "step": 623940
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 3.4323104002448914e-06,
+      "loss": 0.6631,
+      "step": 623950
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 3.4311624703451438e-06,
+      "loss": 0.6366,
+      "step": 623960
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 3.4300145404453966e-06,
+      "loss": 0.7058,
+      "step": 623970
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 3.428866610545649e-06,
+      "loss": 0.66,
+      "step": 623980
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 3.4277186806459014e-06,
+      "loss": 0.674,
+      "step": 623990
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 3.4265707507461542e-06,
+      "loss": 0.6632,
+      "step": 624000
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 3.425422820846407e-06,
+      "loss": 0.6467,
+      "step": 624010
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 3.424274890946659e-06,
+      "loss": 0.7203,
+      "step": 624020
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 3.423126961046912e-06,
+      "loss": 0.7061,
+      "step": 624030
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 3.4219790311471642e-06,
+      "loss": 0.6176,
+      "step": 624040
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 3.4208311012474166e-06,
+      "loss": 0.6866,
+      "step": 624050
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 3.4196831713476694e-06,
+      "loss": 0.6112,
+      "step": 624060
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 3.418535241447922e-06,
+      "loss": 0.7212,
+      "step": 624070
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 3.4173873115481746e-06,
+      "loss": 0.8134,
+      "step": 624080
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 3.416239381648427e-06,
+      "loss": 0.7438,
+      "step": 624090
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 3.4150914517486794e-06,
+      "loss": 0.7057,
+      "step": 624100
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 3.4139435218489323e-06,
+      "loss": 0.672,
+      "step": 624110
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 3.412795591949185e-06,
+      "loss": 0.6602,
+      "step": 624120
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 3.411647662049437e-06,
+      "loss": 0.7128,
+      "step": 624130
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 3.41049973214969e-06,
+      "loss": 0.6634,
+      "step": 624140
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 3.4093518022499423e-06,
+      "loss": 0.7577,
+      "step": 624150
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 3.4082038723501947e-06,
+      "loss": 0.7035,
+      "step": 624160
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 3.4070559424504475e-06,
+      "loss": 0.6977,
+      "step": 624170
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 3.4059080125507e-06,
+      "loss": 0.7376,
+      "step": 624180
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 3.4047600826509527e-06,
+      "loss": 0.7113,
+      "step": 624190
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 3.403612152751205e-06,
+      "loss": 0.7001,
+      "step": 624200
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 3.4024642228514575e-06,
+      "loss": 0.7306,
+      "step": 624210
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 3.4013162929517103e-06,
+      "loss": 0.7543,
+      "step": 624220
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 3.4001683630519623e-06,
+      "loss": 0.6332,
+      "step": 624230
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 3.399020433152215e-06,
+      "loss": 0.6178,
+      "step": 624240
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 3.397872503252468e-06,
+      "loss": 0.7535,
+      "step": 624250
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 3.3967245733527203e-06,
+      "loss": 0.7625,
+      "step": 624260
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 3.3955766434529727e-06,
+      "loss": 0.7063,
+      "step": 624270
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 3.3944287135532255e-06,
+      "loss": 0.6995,
+      "step": 624280
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 3.393280783653478e-06,
+      "loss": 0.7843,
+      "step": 624290
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 3.3921328537537307e-06,
+      "loss": 0.7048,
+      "step": 624300
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 3.3909849238539827e-06,
+      "loss": 0.6956,
+      "step": 624310
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 3.3898369939542355e-06,
+      "loss": 0.7191,
+      "step": 624320
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 3.3886890640544883e-06,
+      "loss": 0.6696,
+      "step": 624330
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 3.3875411341547403e-06,
+      "loss": 0.6667,
+      "step": 624340
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 3.386393204254993e-06,
+      "loss": 0.7672,
+      "step": 624350
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 3.385245274355246e-06,
+      "loss": 0.6677,
+      "step": 624360
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 3.3840973444554983e-06,
+      "loss": 0.6441,
+      "step": 624370
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 3.3829494145557507e-06,
+      "loss": 0.6485,
+      "step": 624380
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 3.381801484656003e-06,
+      "loss": 0.7425,
+      "step": 624390
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 3.380653554756256e-06,
+      "loss": 0.6728,
+      "step": 624400
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 3.3795056248565088e-06,
+      "loss": 0.6629,
+      "step": 624410
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 3.3783576949567607e-06,
+      "loss": 0.685,
+      "step": 624420
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 3.3772097650570136e-06,
+      "loss": 0.6261,
+      "step": 624430
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 3.3760618351572664e-06,
+      "loss": 0.6594,
+      "step": 624440
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 3.3749139052575184e-06,
+      "loss": 0.6072,
+      "step": 624450
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 3.373765975357771e-06,
+      "loss": 0.6729,
+      "step": 624460
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 3.3726180454580236e-06,
+      "loss": 0.6495,
+      "step": 624470
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 3.3714701155582764e-06,
+      "loss": 0.7389,
+      "step": 624480
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 3.3703221856585288e-06,
+      "loss": 0.6706,
+      "step": 624490
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 3.369174255758781e-06,
+      "loss": 0.6441,
+      "step": 624500
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 3.368026325859034e-06,
+      "loss": 0.6183,
+      "step": 624510
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 3.366878395959287e-06,
+      "loss": 0.6796,
+      "step": 624520
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 3.365730466059539e-06,
+      "loss": 0.6383,
+      "step": 624530
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 3.3645825361597916e-06,
+      "loss": 0.659,
+      "step": 624540
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 3.3634346062600444e-06,
+      "loss": 0.6333,
+      "step": 624550
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 3.3622866763602964e-06,
+      "loss": 0.6042,
+      "step": 624560
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 3.3611387464605492e-06,
+      "loss": 0.5753,
+      "step": 624570
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 3.3599908165608016e-06,
+      "loss": 0.7142,
+      "step": 624580
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 3.3588428866610544e-06,
+      "loss": 0.6652,
+      "step": 624590
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 3.357694956761307e-06,
+      "loss": 0.7001,
+      "step": 624600
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 3.3565470268615592e-06,
+      "loss": 0.6201,
+      "step": 624610
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 3.355399096961812e-06,
+      "loss": 0.6732,
+      "step": 624620
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 3.354251167062065e-06,
+      "loss": 0.6203,
+      "step": 624630
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 3.353103237162317e-06,
+      "loss": 0.5977,
+      "step": 624640
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 3.3519553072625697e-06,
+      "loss": 0.6557,
+      "step": 624650
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 3.350807377362822e-06,
+      "loss": 0.6712,
+      "step": 624660
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 3.3496594474630744e-06,
+      "loss": 0.7057,
+      "step": 624670
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 3.3485115175633273e-06,
+      "loss": 0.6489,
+      "step": 624680
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 3.3473635876635797e-06,
+      "loss": 0.7208,
+      "step": 624690
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 3.3462156577638325e-06,
+      "loss": 0.6809,
+      "step": 624700
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 3.345067727864085e-06,
+      "loss": 0.7699,
+      "step": 624710
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 3.3439197979643373e-06,
+      "loss": 0.6976,
+      "step": 624720
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 3.34277186806459e-06,
+      "loss": 0.6012,
+      "step": 624730
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 3.341623938164842e-06,
+      "loss": 0.6964,
+      "step": 624740
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 3.340476008265095e-06,
+      "loss": 0.6882,
+      "step": 624750
+    },
+    {
+      "epoch": 28.66,
+      "learning_rate": 3.3393280783653477e-06,
+      "loss": 0.6959,
+      "step": 624760
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 3.3381801484656e-06,
+      "loss": 0.7243,
+      "step": 624770
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 3.3370322185658525e-06,
+      "loss": 0.5538,
+      "step": 624780
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 3.3358842886661053e-06,
+      "loss": 0.7017,
+      "step": 624790
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 3.3347363587663577e-06,
+      "loss": 0.6842,
+      "step": 624800
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 3.3335884288666105e-06,
+      "loss": 0.6862,
+      "step": 624810
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 3.3324404989668625e-06,
+      "loss": 0.7764,
+      "step": 624820
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 3.3312925690671153e-06,
+      "loss": 0.7443,
+      "step": 624830
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 3.330144639167368e-06,
+      "loss": 0.6823,
+      "step": 624840
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 3.32899670926762e-06,
+      "loss": 0.673,
+      "step": 624850
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 3.327848779367873e-06,
+      "loss": 0.6501,
+      "step": 624860
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 3.3267008494681257e-06,
+      "loss": 0.7254,
+      "step": 624870
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 3.325552919568378e-06,
+      "loss": 0.742,
+      "step": 624880
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 3.3244049896686305e-06,
+      "loss": 0.7478,
+      "step": 624890
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 3.323257059768883e-06,
+      "loss": 0.6805,
+      "step": 624900
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 3.3221091298691357e-06,
+      "loss": 0.6825,
+      "step": 624910
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 3.3209611999693886e-06,
+      "loss": 0.6404,
+      "step": 624920
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 3.3198132700696405e-06,
+      "loss": 0.6817,
+      "step": 624930
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 3.3186653401698934e-06,
+      "loss": 0.6706,
+      "step": 624940
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 3.317517410270146e-06,
+      "loss": 0.6537,
+      "step": 624950
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 3.316369480370398e-06,
+      "loss": 0.6681,
+      "step": 624960
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 3.315221550470651e-06,
+      "loss": 0.6756,
+      "step": 624970
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 3.3140736205709038e-06,
+      "loss": 0.69,
+      "step": 624980
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 3.312925690671156e-06,
+      "loss": 0.6435,
+      "step": 624990
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 3.3117777607714086e-06,
+      "loss": 0.6929,
+      "step": 625000
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 3.310629830871661e-06,
+      "loss": 0.7077,
+      "step": 625010
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 3.309481900971914e-06,
+      "loss": 0.7853,
+      "step": 625020
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 3.3083339710721666e-06,
+      "loss": 0.6878,
+      "step": 625030
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 3.3071860411724186e-06,
+      "loss": 0.8115,
+      "step": 625040
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 3.3060381112726714e-06,
+      "loss": 0.7343,
+      "step": 625050
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 3.3048901813729242e-06,
+      "loss": 0.7227,
+      "step": 625060
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 3.303742251473176e-06,
+      "loss": 0.6947,
+      "step": 625070
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 3.302594321573429e-06,
+      "loss": 0.6695,
+      "step": 625080
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 3.3014463916736814e-06,
+      "loss": 0.646,
+      "step": 625090
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 3.3002984617739342e-06,
+      "loss": 0.6282,
+      "step": 625100
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 3.2991505318741866e-06,
+      "loss": 0.6717,
+      "step": 625110
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 3.298002601974439e-06,
+      "loss": 0.6677,
+      "step": 625120
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 3.296854672074692e-06,
+      "loss": 0.7093,
+      "step": 625130
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 3.2957067421749447e-06,
+      "loss": 0.7132,
+      "step": 625140
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 3.2945588122751966e-06,
+      "loss": 0.6497,
+      "step": 625150
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 3.2934108823754494e-06,
+      "loss": 0.6879,
+      "step": 625160
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 3.292262952475702e-06,
+      "loss": 0.6674,
+      "step": 625170
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 3.2911150225759542e-06,
+      "loss": 0.6594,
+      "step": 625180
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 3.289967092676207e-06,
+      "loss": 0.7022,
+      "step": 625190
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 3.2888191627764594e-06,
+      "loss": 0.6971,
+      "step": 625200
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 3.2876712328767123e-06,
+      "loss": 0.6756,
+      "step": 625210
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 3.2865233029769647e-06,
+      "loss": 0.6285,
+      "step": 625220
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 3.285375373077217e-06,
+      "loss": 0.6246,
+      "step": 625230
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 3.28422744317747e-06,
+      "loss": 0.725,
+      "step": 625240
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 3.283079513277722e-06,
+      "loss": 0.6927,
+      "step": 625250
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 3.2819315833779747e-06,
+      "loss": 0.747,
+      "step": 625260
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 3.2807836534782275e-06,
+      "loss": 0.6288,
+      "step": 625270
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 3.27963572357848e-06,
+      "loss": 0.7049,
+      "step": 625280
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 3.2784877936787323e-06,
+      "loss": 0.6562,
+      "step": 625290
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 3.277339863778985e-06,
+      "loss": 0.6791,
+      "step": 625300
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 3.2761919338792375e-06,
+      "loss": 0.7602,
+      "step": 625310
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 3.2750440039794903e-06,
+      "loss": 0.6929,
+      "step": 625320
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 3.2738960740797423e-06,
+      "loss": 0.7429,
+      "step": 625330
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 3.272748144179995e-06,
+      "loss": 0.7362,
+      "step": 625340
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 3.271600214280248e-06,
+      "loss": 0.6981,
+      "step": 625350
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 3.2704522843805e-06,
+      "loss": 0.7003,
+      "step": 625360
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 3.2693043544807527e-06,
+      "loss": 0.6158,
+      "step": 625370
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 3.2681564245810055e-06,
+      "loss": 0.6267,
+      "step": 625380
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 3.267008494681258e-06,
+      "loss": 0.6541,
+      "step": 625390
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 3.2658605647815103e-06,
+      "loss": 0.6519,
+      "step": 625400
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 3.264712634881763e-06,
+      "loss": 0.6774,
+      "step": 625410
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 3.2635647049820155e-06,
+      "loss": 0.6895,
+      "step": 625420
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 3.2624167750822684e-06,
+      "loss": 0.6561,
+      "step": 625430
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 3.2612688451825203e-06,
+      "loss": 0.6268,
+      "step": 625440
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 3.260120915282773e-06,
+      "loss": 0.6627,
+      "step": 625450
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 3.258972985383026e-06,
+      "loss": 0.6747,
+      "step": 625460
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 3.257825055483278e-06,
+      "loss": 0.7095,
+      "step": 625470
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 3.2566771255835308e-06,
+      "loss": 0.7154,
+      "step": 625480
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 3.2555291956837836e-06,
+      "loss": 0.6069,
+      "step": 625490
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 3.2543812657840355e-06,
+      "loss": 0.6608,
+      "step": 625500
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 3.2532333358842884e-06,
+      "loss": 0.6267,
+      "step": 625510
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 3.2520854059845408e-06,
+      "loss": 0.6691,
+      "step": 625520
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 3.2509374760847936e-06,
+      "loss": 0.6061,
+      "step": 625530
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 3.2497895461850464e-06,
+      "loss": 0.8046,
+      "step": 625540
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 3.2486416162852984e-06,
+      "loss": 0.7124,
+      "step": 625550
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 3.247493686385551e-06,
+      "loss": 0.6075,
+      "step": 625560
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 3.246345756485804e-06,
+      "loss": 0.7663,
+      "step": 625570
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 3.245197826586056e-06,
+      "loss": 0.6258,
+      "step": 625580
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 3.244049896686309e-06,
+      "loss": 0.641,
+      "step": 625590
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 3.242901966786561e-06,
+      "loss": 0.6815,
+      "step": 625600
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 3.2417540368868136e-06,
+      "loss": 0.7758,
+      "step": 625610
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 3.2406061069870664e-06,
+      "loss": 0.6951,
+      "step": 625620
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 3.239458177087319e-06,
+      "loss": 0.6982,
+      "step": 625630
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 3.2383102471875716e-06,
+      "loss": 0.7122,
+      "step": 625640
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 3.237162317287824e-06,
+      "loss": 0.7363,
+      "step": 625650
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 3.2360143873880764e-06,
+      "loss": 0.7547,
+      "step": 625660
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 3.2348664574883292e-06,
+      "loss": 0.7741,
+      "step": 625670
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 3.233718527588581e-06,
+      "loss": 0.6398,
+      "step": 625680
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 3.232570597688834e-06,
+      "loss": 0.6838,
+      "step": 625690
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 3.231422667789087e-06,
+      "loss": 0.6712,
+      "step": 625700
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 3.2302747378893392e-06,
+      "loss": 0.6609,
+      "step": 625710
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 3.2291268079895916e-06,
+      "loss": 0.7622,
+      "step": 625720
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 3.2279788780898445e-06,
+      "loss": 0.6732,
+      "step": 625730
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 3.226830948190097e-06,
+      "loss": 0.7742,
+      "step": 625740
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 3.2256830182903497e-06,
+      "loss": 0.7012,
+      "step": 625750
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 3.2245350883906016e-06,
+      "loss": 0.6468,
+      "step": 625760
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 3.2233871584908545e-06,
+      "loss": 0.7596,
+      "step": 625770
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 3.2222392285911073e-06,
+      "loss": 0.6801,
+      "step": 625780
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 3.2210912986913592e-06,
+      "loss": 0.6845,
+      "step": 625790
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 3.219943368791612e-06,
+      "loss": 0.719,
+      "step": 625800
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 3.218795438891865e-06,
+      "loss": 0.6383,
+      "step": 625810
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 3.2176475089921173e-06,
+      "loss": 0.7253,
+      "step": 625820
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 3.2164995790923697e-06,
+      "loss": 0.692,
+      "step": 625830
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 3.2153516491926225e-06,
+      "loss": 0.7166,
+      "step": 625840
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 3.214203719292875e-06,
+      "loss": 0.6577,
+      "step": 625850
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 3.2130557893931277e-06,
+      "loss": 0.7235,
+      "step": 625860
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 3.2119078594933797e-06,
+      "loss": 0.6899,
+      "step": 625870
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 3.2107599295936325e-06,
+      "loss": 0.7053,
+      "step": 625880
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 3.2096119996938853e-06,
+      "loss": 0.6891,
+      "step": 625890
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 3.2084640697941373e-06,
+      "loss": 0.7264,
+      "step": 625900
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 3.20731613989439e-06,
+      "loss": 0.5967,
+      "step": 625910
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 3.206168209994643e-06,
+      "loss": 0.6661,
+      "step": 625920
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 3.2050202800948953e-06,
+      "loss": 0.7481,
+      "step": 625930
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 3.2038723501951477e-06,
+      "loss": 0.6603,
+      "step": 625940
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 3.2027244202954e-06,
+      "loss": 0.6129,
+      "step": 625950
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 3.201576490395653e-06,
+      "loss": 0.6704,
+      "step": 625960
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 3.2004285604959058e-06,
+      "loss": 0.6647,
+      "step": 625970
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 3.1992806305961577e-06,
+      "loss": 0.6779,
+      "step": 625980
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 3.1981327006964105e-06,
+      "loss": 0.6518,
+      "step": 625990
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 3.1969847707966634e-06,
+      "loss": 0.7026,
+      "step": 626000
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 3.1958368408969153e-06,
+      "loss": 0.7361,
+      "step": 626010
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 3.194688910997168e-06,
+      "loss": 0.6823,
+      "step": 626020
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 3.1935409810974205e-06,
+      "loss": 0.7439,
+      "step": 626030
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 3.1923930511976734e-06,
+      "loss": 0.6582,
+      "step": 626040
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 3.1912451212979258e-06,
+      "loss": 0.7142,
+      "step": 626050
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 3.190097191398178e-06,
+      "loss": 0.6357,
+      "step": 626060
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 3.188949261498431e-06,
+      "loss": 0.6934,
+      "step": 626070
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 3.187801331598684e-06,
+      "loss": 0.7018,
+      "step": 626080
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 3.1866534016989358e-06,
+      "loss": 0.6842,
+      "step": 626090
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 3.1855054717991886e-06,
+      "loss": 0.6703,
+      "step": 626100
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 3.184357541899441e-06,
+      "loss": 0.6928,
+      "step": 626110
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 3.1832096119996934e-06,
+      "loss": 0.7538,
+      "step": 626120
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 3.182061682099946e-06,
+      "loss": 0.6728,
+      "step": 626130
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 3.1809137522001986e-06,
+      "loss": 0.6972,
+      "step": 626140
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 3.1797658223004514e-06,
+      "loss": 0.734,
+      "step": 626150
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 3.178617892400704e-06,
+      "loss": 0.7154,
+      "step": 626160
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 3.177469962500956e-06,
+      "loss": 0.6668,
+      "step": 626170
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 3.176322032601209e-06,
+      "loss": 0.6431,
+      "step": 626180
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 3.175174102701461e-06,
+      "loss": 0.7748,
+      "step": 626190
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 3.174026172801714e-06,
+      "loss": 0.7479,
+      "step": 626200
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 3.1728782429019666e-06,
+      "loss": 0.6482,
+      "step": 626210
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 3.171730313002219e-06,
+      "loss": 0.7459,
+      "step": 626220
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 3.1705823831024714e-06,
+      "loss": 0.7195,
+      "step": 626230
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 3.1694344532027242e-06,
+      "loss": 0.8124,
+      "step": 626240
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 3.1682865233029766e-06,
+      "loss": 0.6445,
+      "step": 626250
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 3.1671385934032295e-06,
+      "loss": 0.5779,
+      "step": 626260
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 3.1659906635034814e-06,
+      "loss": 0.6935,
+      "step": 626270
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 3.1648427336037342e-06,
+      "loss": 0.7388,
+      "step": 626280
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 3.163694803703987e-06,
+      "loss": 0.6254,
+      "step": 626290
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 3.162546873804239e-06,
+      "loss": 0.7128,
+      "step": 626300
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 3.161398943904492e-06,
+      "loss": 0.6841,
+      "step": 626310
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 3.1602510140047447e-06,
+      "loss": 0.6972,
+      "step": 626320
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 3.159103084104997e-06,
+      "loss": 0.6708,
+      "step": 626330
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 3.1579551542052495e-06,
+      "loss": 0.7109,
+      "step": 626340
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 3.1568072243055023e-06,
+      "loss": 0.732,
+      "step": 626350
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 3.1556592944057547e-06,
+      "loss": 0.6526,
+      "step": 626360
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 3.1545113645060075e-06,
+      "loss": 0.6218,
+      "step": 626370
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 3.1533634346062595e-06,
+      "loss": 0.6445,
+      "step": 626380
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 3.1522155047065123e-06,
+      "loss": 0.7016,
+      "step": 626390
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 3.151067574806765e-06,
+      "loss": 0.7014,
+      "step": 626400
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 3.149919644907017e-06,
+      "loss": 0.6719,
+      "step": 626410
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 3.14877171500727e-06,
+      "loss": 0.6813,
+      "step": 626420
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 3.1476237851075227e-06,
+      "loss": 0.5934,
+      "step": 626430
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 3.146475855207775e-06,
+      "loss": 0.6375,
+      "step": 626440
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 3.1453279253080275e-06,
+      "loss": 0.6971,
+      "step": 626450
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 3.14417999540828e-06,
+      "loss": 0.667,
+      "step": 626460
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 3.1430320655085327e-06,
+      "loss": 0.7597,
+      "step": 626470
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 3.1418841356087855e-06,
+      "loss": 0.6502,
+      "step": 626480
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 3.1407362057090375e-06,
+      "loss": 0.6738,
+      "step": 626490
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 3.1395882758092903e-06,
+      "loss": 0.6867,
+      "step": 626500
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 3.138440345909543e-06,
+      "loss": 0.6294,
+      "step": 626510
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 3.137292416009795e-06,
+      "loss": 0.6907,
+      "step": 626520
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 3.136144486110048e-06,
+      "loss": 0.6922,
+      "step": 626530
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 3.1349965562103003e-06,
+      "loss": 0.6888,
+      "step": 626540
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 3.133848626310553e-06,
+      "loss": 0.6722,
+      "step": 626550
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 3.1327006964108056e-06,
+      "loss": 0.6791,
+      "step": 626560
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 3.131552766511058e-06,
+      "loss": 0.6626,
+      "step": 626570
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 3.1304048366113108e-06,
+      "loss": 0.6741,
+      "step": 626580
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 3.1292569067115636e-06,
+      "loss": 0.7025,
+      "step": 626590
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 3.1281089768118156e-06,
+      "loss": 0.6971,
+      "step": 626600
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 3.1269610469120684e-06,
+      "loss": 0.7338,
+      "step": 626610
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 3.1258131170123208e-06,
+      "loss": 0.6321,
+      "step": 626620
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 3.124665187112573e-06,
+      "loss": 0.7632,
+      "step": 626630
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 3.123517257212826e-06,
+      "loss": 0.7058,
+      "step": 626640
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 3.1223693273130784e-06,
+      "loss": 0.7126,
+      "step": 626650
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 3.121221397413331e-06,
+      "loss": 0.6472,
+      "step": 626660
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 3.1200734675135836e-06,
+      "loss": 0.8114,
+      "step": 626670
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 3.118925537613836e-06,
+      "loss": 0.6859,
+      "step": 626680
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 3.117777607714089e-06,
+      "loss": 0.7066,
+      "step": 626690
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 3.1166296778143408e-06,
+      "loss": 0.6085,
+      "step": 626700
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 3.1154817479145936e-06,
+      "loss": 0.713,
+      "step": 626710
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 3.1143338180148464e-06,
+      "loss": 0.7284,
+      "step": 626720
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 3.113185888115099e-06,
+      "loss": 0.7102,
+      "step": 626730
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 3.112037958215351e-06,
+      "loss": 0.6668,
+      "step": 626740
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 3.110890028315604e-06,
+      "loss": 0.678,
+      "step": 626750
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 3.1097420984158564e-06,
+      "loss": 0.68,
+      "step": 626760
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 3.1085941685161092e-06,
+      "loss": 0.6944,
+      "step": 626770
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 3.1074462386163616e-06,
+      "loss": 0.6908,
+      "step": 626780
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 3.106298308716614e-06,
+      "loss": 0.6038,
+      "step": 626790
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 3.105150378816867e-06,
+      "loss": 0.7127,
+      "step": 626800
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 3.104002448917119e-06,
+      "loss": 0.682,
+      "step": 626810
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 3.1028545190173716e-06,
+      "loss": 0.6637,
+      "step": 626820
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 3.1017065891176245e-06,
+      "loss": 0.6706,
+      "step": 626830
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 3.100558659217877e-06,
+      "loss": 0.7357,
+      "step": 626840
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 3.0994107293181293e-06,
+      "loss": 0.6705,
+      "step": 626850
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 3.098262799418382e-06,
+      "loss": 0.6394,
+      "step": 626860
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 3.0971148695186345e-06,
+      "loss": 0.6979,
+      "step": 626870
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 3.0959669396188873e-06,
+      "loss": 0.7164,
+      "step": 626880
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 3.0948190097191393e-06,
+      "loss": 0.6549,
+      "step": 626890
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 3.093671079819392e-06,
+      "loss": 0.671,
+      "step": 626900
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 3.092523149919645e-06,
+      "loss": 0.6388,
+      "step": 626910
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 3.091375220019897e-06,
+      "loss": 0.671,
+      "step": 626920
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 3.0902272901201497e-06,
+      "loss": 0.7961,
+      "step": 626930
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 3.0890793602204025e-06,
+      "loss": 0.6535,
+      "step": 626940
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 3.087931430320655e-06,
+      "loss": 0.6907,
+      "step": 626950
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 3.0867835004209073e-06,
+      "loss": 0.6804,
+      "step": 626960
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 3.0856355705211597e-06,
+      "loss": 0.659,
+      "step": 626970
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 3.0844876406214125e-06,
+      "loss": 0.6755,
+      "step": 626980
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 3.0833397107216653e-06,
+      "loss": 0.7719,
+      "step": 626990
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 3.0821917808219173e-06,
+      "loss": 0.6245,
+      "step": 627000
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 3.08104385092217e-06,
+      "loss": 0.6453,
+      "step": 627010
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 3.079895921022423e-06,
+      "loss": 0.7155,
+      "step": 627020
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 3.078747991122675e-06,
+      "loss": 0.7163,
+      "step": 627030
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 3.0776000612229277e-06,
+      "loss": 0.6777,
+      "step": 627040
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 3.07645213132318e-06,
+      "loss": 0.6218,
+      "step": 627050
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 3.075304201423433e-06,
+      "loss": 0.6935,
+      "step": 627060
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 3.0741562715236853e-06,
+      "loss": 0.7145,
+      "step": 627070
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 3.0730083416239377e-06,
+      "loss": 0.6996,
+      "step": 627080
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 3.0718604117241906e-06,
+      "loss": 0.6442,
+      "step": 627090
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 3.0707124818244434e-06,
+      "loss": 0.717,
+      "step": 627100
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 3.0695645519246953e-06,
+      "loss": 0.7366,
+      "step": 627110
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 3.068416622024948e-06,
+      "loss": 0.7485,
+      "step": 627120
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 3.0672686921252006e-06,
+      "loss": 0.7055,
+      "step": 627130
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 3.066120762225453e-06,
+      "loss": 0.7475,
+      "step": 627140
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 3.0649728323257058e-06,
+      "loss": 0.6286,
+      "step": 627150
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 3.063824902425958e-06,
+      "loss": 0.6692,
+      "step": 627160
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 3.062676972526211e-06,
+      "loss": 0.7651,
+      "step": 627170
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 3.0615290426264634e-06,
+      "loss": 0.7152,
+      "step": 627180
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 3.0603811127267158e-06,
+      "loss": 0.6679,
+      "step": 627190
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 3.0592331828269686e-06,
+      "loss": 0.6785,
+      "step": 627200
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 3.0580852529272214e-06,
+      "loss": 0.6719,
+      "step": 627210
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 3.0569373230274734e-06,
+      "loss": 0.6897,
+      "step": 627220
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 3.055789393127726e-06,
+      "loss": 0.6813,
+      "step": 627230
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 3.0546414632279786e-06,
+      "loss": 0.6869,
+      "step": 627240
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 3.053493533328231e-06,
+      "loss": 0.8385,
+      "step": 627250
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 3.052345603428484e-06,
+      "loss": 0.633,
+      "step": 627260
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 3.051197673528736e-06,
+      "loss": 0.7069,
+      "step": 627270
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 3.050049743628989e-06,
+      "loss": 0.7881,
+      "step": 627280
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 3.0489018137292414e-06,
+      "loss": 0.613,
+      "step": 627290
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 3.047753883829494e-06,
+      "loss": 0.7109,
+      "step": 627300
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 3.0466059539297466e-06,
+      "loss": 0.6264,
+      "step": 627310
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 3.0454580240299986e-06,
+      "loss": 0.6552,
+      "step": 627320
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 3.0443100941302514e-06,
+      "loss": 0.6958,
+      "step": 627330
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 3.0431621642305042e-06,
+      "loss": 0.6936,
+      "step": 627340
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 3.0420142343307566e-06,
+      "loss": 0.6672,
+      "step": 627350
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 3.040866304431009e-06,
+      "loss": 0.6858,
+      "step": 627360
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 3.039718374531262e-06,
+      "loss": 0.7031,
+      "step": 627370
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 3.0385704446315143e-06,
+      "loss": 0.6678,
+      "step": 627380
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 3.037422514731767e-06,
+      "loss": 0.6242,
+      "step": 627390
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 3.036274584832019e-06,
+      "loss": 0.6985,
+      "step": 627400
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 3.035126654932272e-06,
+      "loss": 0.8219,
+      "step": 627410
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 3.0339787250325247e-06,
+      "loss": 0.6804,
+      "step": 627420
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 3.0328307951327767e-06,
+      "loss": 0.7159,
+      "step": 627430
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 3.0316828652330295e-06,
+      "loss": 0.6942,
+      "step": 627440
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 3.0305349353332823e-06,
+      "loss": 0.7596,
+      "step": 627450
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 3.0293870054335347e-06,
+      "loss": 0.668,
+      "step": 627460
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 3.028239075533787e-06,
+      "loss": 0.628,
+      "step": 627470
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 3.0270911456340395e-06,
+      "loss": 0.6952,
+      "step": 627480
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 3.0259432157342923e-06,
+      "loss": 0.7473,
+      "step": 627490
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 3.024795285834545e-06,
+      "loss": 0.6688,
+      "step": 627500
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 3.023647355934797e-06,
+      "loss": 0.663,
+      "step": 627510
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 3.02249942603505e-06,
+      "loss": 0.6416,
+      "step": 627520
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 3.0213514961353027e-06,
+      "loss": 0.6822,
+      "step": 627530
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 3.0202035662355547e-06,
+      "loss": 0.716,
+      "step": 627540
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 3.0190556363358075e-06,
+      "loss": 0.7042,
+      "step": 627550
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 3.01790770643606e-06,
+      "loss": 0.6715,
+      "step": 627560
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 3.0167597765363127e-06,
+      "loss": 0.7385,
+      "step": 627570
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 3.015611846636565e-06,
+      "loss": 0.5775,
+      "step": 627580
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 3.0144639167368175e-06,
+      "loss": 0.7474,
+      "step": 627590
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 3.0133159868370703e-06,
+      "loss": 0.6117,
+      "step": 627600
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 3.012168056937323e-06,
+      "loss": 0.6722,
+      "step": 627610
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 3.011020127037575e-06,
+      "loss": 0.7025,
+      "step": 627620
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 3.009872197137828e-06,
+      "loss": 0.657,
+      "step": 627630
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 3.0087242672380808e-06,
+      "loss": 0.6856,
+      "step": 627640
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 3.0075763373383327e-06,
+      "loss": 0.679,
+      "step": 627650
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 3.0064284074385856e-06,
+      "loss": 0.6678,
+      "step": 627660
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 3.005280477538838e-06,
+      "loss": 0.6178,
+      "step": 627670
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 3.0041325476390908e-06,
+      "loss": 0.719,
+      "step": 627680
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 3.002984617739343e-06,
+      "loss": 0.695,
+      "step": 627690
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 3.0018366878395956e-06,
+      "loss": 0.7301,
+      "step": 627700
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 3.0006887579398484e-06,
+      "loss": 0.7192,
+      "step": 627710
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 2.999540828040101e-06,
+      "loss": 0.663,
+      "step": 627720
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 2.998392898140353e-06,
+      "loss": 0.7051,
+      "step": 627730
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 2.997244968240606e-06,
+      "loss": 0.6896,
+      "step": 627740
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 2.996097038340858e-06,
+      "loss": 0.7209,
+      "step": 627750
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 2.9949491084411108e-06,
+      "loss": 0.6621,
+      "step": 627760
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 2.9938011785413636e-06,
+      "loss": 0.7066,
+      "step": 627770
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 2.992653248641616e-06,
+      "loss": 0.7272,
+      "step": 627780
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 2.9915053187418684e-06,
+      "loss": 0.6235,
+      "step": 627790
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 2.9903573888421212e-06,
+      "loss": 0.6329,
+      "step": 627800
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 2.9892094589423736e-06,
+      "loss": 0.7543,
+      "step": 627810
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 2.9880615290426264e-06,
+      "loss": 0.7512,
+      "step": 627820
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 2.9869135991428784e-06,
+      "loss": 0.7333,
+      "step": 627830
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 2.9857656692431312e-06,
+      "loss": 0.6559,
+      "step": 627840
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 2.984617739343384e-06,
+      "loss": 0.7269,
+      "step": 627850
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 2.983469809443636e-06,
+      "loss": 0.7338,
+      "step": 627860
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 2.982321879543889e-06,
+      "loss": 0.7141,
+      "step": 627870
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 2.9811739496441416e-06,
+      "loss": 0.7112,
+      "step": 627880
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 2.980026019744394e-06,
+      "loss": 0.6922,
+      "step": 627890
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 2.9788780898446464e-06,
+      "loss": 0.7519,
+      "step": 627900
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 2.977730159944899e-06,
+      "loss": 0.6762,
+      "step": 627910
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 2.9765822300451517e-06,
+      "loss": 0.6725,
+      "step": 627920
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 2.9754343001454045e-06,
+      "loss": 0.6192,
+      "step": 627930
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 2.9742863702456564e-06,
+      "loss": 0.7103,
+      "step": 627940
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 2.9731384403459093e-06,
+      "loss": 0.7831,
+      "step": 627950
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 2.971990510446162e-06,
+      "loss": 0.6389,
+      "step": 627960
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 2.970842580546414e-06,
+      "loss": 0.7425,
+      "step": 627970
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 2.969694650646667e-06,
+      "loss": 0.6361,
+      "step": 627980
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 2.9685467207469193e-06,
+      "loss": 0.6606,
+      "step": 627990
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 2.967398790847172e-06,
+      "loss": 0.7395,
+      "step": 628000
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 2.9662508609474245e-06,
+      "loss": 0.6277,
+      "step": 628010
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 2.965102931047677e-06,
+      "loss": 0.7164,
+      "step": 628020
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 2.9639550011479297e-06,
+      "loss": 0.7411,
+      "step": 628030
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 2.9628070712481825e-06,
+      "loss": 0.6889,
+      "step": 628040
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 2.9616591413484345e-06,
+      "loss": 0.636,
+      "step": 628050
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 2.9605112114486873e-06,
+      "loss": 0.6568,
+      "step": 628060
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 2.95936328154894e-06,
+      "loss": 0.6276,
+      "step": 628070
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 2.958215351649192e-06,
+      "loss": 0.7421,
+      "step": 628080
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 2.957067421749445e-06,
+      "loss": 0.7287,
+      "step": 628090
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 2.9559194918496973e-06,
+      "loss": 0.7902,
+      "step": 628100
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 2.95477156194995e-06,
+      "loss": 0.7598,
+      "step": 628110
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 2.9536236320502025e-06,
+      "loss": 0.7057,
+      "step": 628120
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 2.952475702150455e-06,
+      "loss": 0.6264,
+      "step": 628130
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 2.9513277722507077e-06,
+      "loss": 0.7107,
+      "step": 628140
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 2.9501798423509606e-06,
+      "loss": 0.747,
+      "step": 628150
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 2.9490319124512125e-06,
+      "loss": 0.7092,
+      "step": 628160
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 2.9478839825514653e-06,
+      "loss": 0.7561,
+      "step": 628170
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 2.9467360526517177e-06,
+      "loss": 0.713,
+      "step": 628180
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 2.94558812275197e-06,
+      "loss": 0.6797,
+      "step": 628190
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 2.944440192852223e-06,
+      "loss": 0.664,
+      "step": 628200
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 2.9432922629524754e-06,
+      "loss": 0.6905,
+      "step": 628210
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 2.942144333052728e-06,
+      "loss": 0.7155,
+      "step": 628220
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 2.9409964031529806e-06,
+      "loss": 0.7266,
+      "step": 628230
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 2.939848473253233e-06,
+      "loss": 0.7128,
+      "step": 628240
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 2.9387005433534858e-06,
+      "loss": 0.6611,
+      "step": 628250
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 2.9375526134537378e-06,
+      "loss": 0.6458,
+      "step": 628260
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 2.9364046835539906e-06,
+      "loss": 0.6922,
+      "step": 628270
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 2.9352567536542434e-06,
+      "loss": 0.6943,
+      "step": 628280
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 2.9341088237544958e-06,
+      "loss": 0.6213,
+      "step": 628290
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 2.932960893854748e-06,
+      "loss": 0.6526,
+      "step": 628300
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 2.931812963955001e-06,
+      "loss": 0.6796,
+      "step": 628310
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 2.9306650340552534e-06,
+      "loss": 0.6645,
+      "step": 628320
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 2.9295171041555062e-06,
+      "loss": 0.6645,
+      "step": 628330
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 2.928369174255758e-06,
+      "loss": 0.6527,
+      "step": 628340
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 2.927221244356011e-06,
+      "loss": 0.7668,
+      "step": 628350
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 2.926073314456264e-06,
+      "loss": 0.6556,
+      "step": 628360
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 2.924925384556516e-06,
+      "loss": 0.7581,
+      "step": 628370
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 2.9237774546567686e-06,
+      "loss": 0.6453,
+      "step": 628380
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 2.9226295247570214e-06,
+      "loss": 0.643,
+      "step": 628390
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 2.921481594857274e-06,
+      "loss": 0.6803,
+      "step": 628400
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 2.9203336649575262e-06,
+      "loss": 0.6679,
+      "step": 628410
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 2.9191857350577786e-06,
+      "loss": 0.6796,
+      "step": 628420
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 2.9180378051580314e-06,
+      "loss": 0.7098,
+      "step": 628430
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 2.9168898752582843e-06,
+      "loss": 0.5788,
+      "step": 628440
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 2.9157419453585362e-06,
+      "loss": 0.6219,
+      "step": 628450
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 2.914594015458789e-06,
+      "loss": 0.7031,
+      "step": 628460
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 2.913446085559042e-06,
+      "loss": 0.7102,
+      "step": 628470
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 2.912298155659294e-06,
+      "loss": 0.6945,
+      "step": 628480
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 2.9111502257595467e-06,
+      "loss": 0.8014,
+      "step": 628490
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 2.9100022958597995e-06,
+      "loss": 0.6178,
+      "step": 628500
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 2.908854365960052e-06,
+      "loss": 0.7326,
+      "step": 628510
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 2.9077064360603043e-06,
+      "loss": 0.5821,
+      "step": 628520
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 2.9065585061605567e-06,
+      "loss": 0.6586,
+      "step": 628530
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 2.9054105762608095e-06,
+      "loss": 0.7477,
+      "step": 628540
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 2.9042626463610623e-06,
+      "loss": 0.6231,
+      "step": 628550
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 2.9031147164613143e-06,
+      "loss": 0.6682,
+      "step": 628560
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 2.901966786561567e-06,
+      "loss": 0.6735,
+      "step": 628570
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 2.90081885666182e-06,
+      "loss": 0.6796,
+      "step": 628580
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 2.899670926762072e-06,
+      "loss": 0.739,
+      "step": 628590
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 2.8985229968623247e-06,
+      "loss": 0.6966,
+      "step": 628600
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 2.897375066962577e-06,
+      "loss": 0.7269,
+      "step": 628610
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 2.89622713706283e-06,
+      "loss": 0.6731,
+      "step": 628620
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 2.8950792071630823e-06,
+      "loss": 0.6931,
+      "step": 628630
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 2.8939312772633347e-06,
+      "loss": 0.6608,
+      "step": 628640
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 2.8927833473635875e-06,
+      "loss": 0.678,
+      "step": 628650
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 2.8916354174638403e-06,
+      "loss": 0.7213,
+      "step": 628660
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 2.8904874875640923e-06,
+      "loss": 0.7119,
+      "step": 628670
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 2.889339557664345e-06,
+      "loss": 0.6798,
+      "step": 628680
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 2.8881916277645975e-06,
+      "loss": 0.6844,
+      "step": 628690
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 2.88704369786485e-06,
+      "loss": 0.65,
+      "step": 628700
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 2.8858957679651027e-06,
+      "loss": 0.6768,
+      "step": 628710
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 2.884747838065355e-06,
+      "loss": 0.693,
+      "step": 628720
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 2.883599908165608e-06,
+      "loss": 0.7769,
+      "step": 628730
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 2.8824519782658604e-06,
+      "loss": 0.6788,
+      "step": 628740
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 2.8813040483661128e-06,
+      "loss": 0.5812,
+      "step": 628750
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 2.8801561184663656e-06,
+      "loss": 0.7173,
+      "step": 628760
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 2.8790081885666175e-06,
+      "loss": 0.6684,
+      "step": 628770
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 2.8778602586668704e-06,
+      "loss": 0.7994,
+      "step": 628780
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 2.876712328767123e-06,
+      "loss": 0.692,
+      "step": 628790
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 2.8755643988673756e-06,
+      "loss": 0.7021,
+      "step": 628800
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 2.874416468967628e-06,
+      "loss": 0.7761,
+      "step": 628810
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 2.873268539067881e-06,
+      "loss": 0.7523,
+      "step": 628820
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 2.872120609168133e-06,
+      "loss": 0.6867,
+      "step": 628830
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 2.870972679268386e-06,
+      "loss": 0.7061,
+      "step": 628840
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 2.869824749368638e-06,
+      "loss": 0.6982,
+      "step": 628850
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 2.868676819468891e-06,
+      "loss": 0.6364,
+      "step": 628860
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 2.8675288895691436e-06,
+      "loss": 0.6954,
+      "step": 628870
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 2.8663809596693956e-06,
+      "loss": 0.6734,
+      "step": 628880
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 2.8652330297696484e-06,
+      "loss": 0.7361,
+      "step": 628890
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 2.8640850998699012e-06,
+      "loss": 0.6971,
+      "step": 628900
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 2.8629371699701536e-06,
+      "loss": 0.6775,
+      "step": 628910
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 2.861789240070406e-06,
+      "loss": 0.7406,
+      "step": 628920
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 2.8606413101706584e-06,
+      "loss": 0.6694,
+      "step": 628930
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 2.8594933802709112e-06,
+      "loss": 0.7326,
+      "step": 628940
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 2.858345450371164e-06,
+      "loss": 0.6887,
+      "step": 628950
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 2.857197520471416e-06,
+      "loss": 0.6559,
+      "step": 628960
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 2.856049590571669e-06,
+      "loss": 0.6719,
+      "step": 628970
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 2.8549016606719217e-06,
+      "loss": 0.6829,
+      "step": 628980
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 2.8537537307721736e-06,
+      "loss": 0.7,
+      "step": 628990
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 2.8526058008724264e-06,
+      "loss": 0.6794,
+      "step": 629000
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 2.8514578709726793e-06,
+      "loss": 0.6892,
+      "step": 629010
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 2.8503099410729317e-06,
+      "loss": 0.6045,
+      "step": 629020
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 2.849162011173184e-06,
+      "loss": 0.7235,
+      "step": 629030
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 2.8480140812734365e-06,
+      "loss": 0.6836,
+      "step": 629040
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 2.8468661513736893e-06,
+      "loss": 0.6514,
+      "step": 629050
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 2.845718221473942e-06,
+      "loss": 0.6997,
+      "step": 629060
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 2.844570291574194e-06,
+      "loss": 0.6416,
+      "step": 629070
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 2.843422361674447e-06,
+      "loss": 0.7045,
+      "step": 629080
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 2.8422744317746997e-06,
+      "loss": 0.7597,
+      "step": 629090
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 2.8411265018749517e-06,
+      "loss": 0.693,
+      "step": 629100
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 2.8399785719752045e-06,
+      "loss": 0.6882,
+      "step": 629110
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 2.838830642075457e-06,
+      "loss": 0.622,
+      "step": 629120
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 2.8376827121757097e-06,
+      "loss": 0.7308,
+      "step": 629130
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 2.836534782275962e-06,
+      "loss": 0.6918,
+      "step": 629140
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 2.8353868523762145e-06,
+      "loss": 0.7803,
+      "step": 629150
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 2.8342389224764673e-06,
+      "loss": 0.7513,
+      "step": 629160
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 2.83309099257672e-06,
+      "loss": 0.7526,
+      "step": 629170
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 2.831943062676972e-06,
+      "loss": 0.7479,
+      "step": 629180
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 2.830795132777225e-06,
+      "loss": 0.6885,
+      "step": 629190
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 2.8296472028774773e-06,
+      "loss": 0.6992,
+      "step": 629200
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 2.8284992729777297e-06,
+      "loss": 0.6688,
+      "step": 629210
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 2.8273513430779825e-06,
+      "loss": 0.7764,
+      "step": 629220
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 2.826203413178235e-06,
+      "loss": 0.6673,
+      "step": 629230
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 2.8250554832784877e-06,
+      "loss": 0.7413,
+      "step": 629240
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 2.82390755337874e-06,
+      "loss": 0.6816,
+      "step": 629250
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 2.8227596234789925e-06,
+      "loss": 0.6786,
+      "step": 629260
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 2.8216116935792454e-06,
+      "loss": 0.6891,
+      "step": 629270
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 2.8204637636794973e-06,
+      "loss": 0.6251,
+      "step": 629280
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 2.81931583377975e-06,
+      "loss": 0.588,
+      "step": 629290
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 2.818167903880003e-06,
+      "loss": 0.6472,
+      "step": 629300
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 2.8170199739802554e-06,
+      "loss": 0.6553,
+      "step": 629310
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 2.8158720440805078e-06,
+      "loss": 0.7237,
+      "step": 629320
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 2.8147241141807606e-06,
+      "loss": 0.71,
+      "step": 629330
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 2.813576184281013e-06,
+      "loss": 0.7102,
+      "step": 629340
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 2.812428254381266e-06,
+      "loss": 0.7405,
+      "step": 629350
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 2.8112803244815178e-06,
+      "loss": 0.657,
+      "step": 629360
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 2.8101323945817706e-06,
+      "loss": 0.7217,
+      "step": 629370
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 2.8089844646820234e-06,
+      "loss": 0.6885,
+      "step": 629380
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 2.8078365347822754e-06,
+      "loss": 0.6829,
+      "step": 629390
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 2.806688604882528e-06,
+      "loss": 0.7101,
+      "step": 629400
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 2.805540674982781e-06,
+      "loss": 0.7794,
+      "step": 629410
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 2.8043927450830334e-06,
+      "loss": 0.6162,
+      "step": 629420
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 2.803244815183286e-06,
+      "loss": 0.7436,
+      "step": 629430
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 2.8020968852835386e-06,
+      "loss": 0.6466,
+      "step": 629440
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 2.800948955383791e-06,
+      "loss": 0.696,
+      "step": 629450
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 2.799801025484044e-06,
+      "loss": 0.7212,
+      "step": 629460
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 2.798653095584296e-06,
+      "loss": 0.6809,
+      "step": 629470
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 2.7975051656845486e-06,
+      "loss": 0.7132,
+      "step": 629480
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 2.7963572357848014e-06,
+      "loss": 0.6405,
+      "step": 629490
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 2.7952093058850534e-06,
+      "loss": 0.6702,
+      "step": 629500
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 2.7940613759853062e-06,
+      "loss": 0.6901,
+      "step": 629510
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 2.792913446085559e-06,
+      "loss": 0.7066,
+      "step": 629520
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 2.7917655161858115e-06,
+      "loss": 0.6452,
+      "step": 629530
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 2.790617586286064e-06,
+      "loss": 0.6615,
+      "step": 629540
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 2.7894696563863162e-06,
+      "loss": 0.7214,
+      "step": 629550
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 2.788321726486569e-06,
+      "loss": 0.6813,
+      "step": 629560
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 2.787173796586822e-06,
+      "loss": 0.677,
+      "step": 629570
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 2.786025866687074e-06,
+      "loss": 0.6625,
+      "step": 629580
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 2.7848779367873267e-06,
+      "loss": 0.6974,
+      "step": 629590
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 2.7837300068875795e-06,
+      "loss": 0.7115,
+      "step": 629600
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 2.7825820769878315e-06,
+      "loss": 0.7119,
+      "step": 629610
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 2.7814341470880843e-06,
+      "loss": 0.6814,
+      "step": 629620
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 2.7802862171883367e-06,
+      "loss": 0.7547,
+      "step": 629630
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 2.7791382872885895e-06,
+      "loss": 0.6178,
+      "step": 629640
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 2.777990357388842e-06,
+      "loss": 0.7206,
+      "step": 629650
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 2.7768424274890943e-06,
+      "loss": 0.6867,
+      "step": 629660
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 2.775694497589347e-06,
+      "loss": 0.7041,
+      "step": 629670
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 2.7745465676896e-06,
+      "loss": 0.7193,
+      "step": 629680
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 2.773398637789852e-06,
+      "loss": 0.7179,
+      "step": 629690
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 2.7722507078901047e-06,
+      "loss": 0.6937,
+      "step": 629700
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 2.771102777990357e-06,
+      "loss": 0.7214,
+      "step": 629710
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 2.7699548480906095e-06,
+      "loss": 0.6623,
+      "step": 629720
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 2.7688069181908623e-06,
+      "loss": 0.714,
+      "step": 629730
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 2.7676589882911147e-06,
+      "loss": 0.6887,
+      "step": 629740
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 2.7665110583913675e-06,
+      "loss": 0.6338,
+      "step": 629750
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 2.76536312849162e-06,
+      "loss": 0.6841,
+      "step": 629760
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 2.7642151985918723e-06,
+      "loss": 0.6896,
+      "step": 629770
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 2.763067268692125e-06,
+      "loss": 0.608,
+      "step": 629780
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 2.761919338792377e-06,
+      "loss": 0.7556,
+      "step": 629790
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 2.76077140889263e-06,
+      "loss": 0.634,
+      "step": 629800
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 2.7596234789928828e-06,
+      "loss": 0.5994,
+      "step": 629810
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 2.758475549093135e-06,
+      "loss": 0.6894,
+      "step": 629820
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 2.7573276191933875e-06,
+      "loss": 0.6909,
+      "step": 629830
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 2.7561796892936404e-06,
+      "loss": 0.7036,
+      "step": 629840
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 2.7550317593938928e-06,
+      "loss": 0.7555,
+      "step": 629850
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 2.7538838294941456e-06,
+      "loss": 0.7205,
+      "step": 629860
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 2.752735899594398e-06,
+      "loss": 0.7417,
+      "step": 629870
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 2.7515879696946504e-06,
+      "loss": 0.6953,
+      "step": 629880
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 2.750440039794903e-06,
+      "loss": 0.7246,
+      "step": 629890
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 2.749292109895155e-06,
+      "loss": 0.6731,
+      "step": 629900
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 2.748144179995408e-06,
+      "loss": 0.6147,
+      "step": 629910
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 2.746996250095661e-06,
+      "loss": 0.7406,
+      "step": 629920
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 2.745848320195913e-06,
+      "loss": 0.7226,
+      "step": 629930
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 2.7447003902961656e-06,
+      "loss": 0.6606,
+      "step": 629940
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 2.7435524603964184e-06,
+      "loss": 0.6861,
+      "step": 629950
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 2.742404530496671e-06,
+      "loss": 0.6257,
+      "step": 629960
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 2.7412566005969236e-06,
+      "loss": 0.6659,
+      "step": 629970
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 2.7401086706971756e-06,
+      "loss": 0.6917,
+      "step": 629980
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 2.7389607407974284e-06,
+      "loss": 0.5789,
+      "step": 629990
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 2.7378128108976812e-06,
+      "loss": 0.6755,
+      "step": 630000
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 2.736664880997933e-06,
+      "loss": 0.6813,
+      "step": 630010
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 2.735516951098186e-06,
+      "loss": 0.6802,
+      "step": 630020
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 2.734369021198439e-06,
+      "loss": 0.6696,
+      "step": 630030
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 2.733221091298691e-06,
+      "loss": 0.611,
+      "step": 630040
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 2.7320731613989436e-06,
+      "loss": 0.6593,
+      "step": 630050
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 2.730925231499196e-06,
+      "loss": 0.7146,
+      "step": 630060
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 2.729777301599449e-06,
+      "loss": 0.5547,
+      "step": 630070
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 2.7286293716997017e-06,
+      "loss": 0.6106,
+      "step": 630080
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 2.7274814417999536e-06,
+      "loss": 0.6515,
+      "step": 630090
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 2.7263335119002065e-06,
+      "loss": 0.6524,
+      "step": 630100
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 2.7251855820004593e-06,
+      "loss": 0.6607,
+      "step": 630110
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 2.7240376521007112e-06,
+      "loss": 0.7405,
+      "step": 630120
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 2.722889722200964e-06,
+      "loss": 0.6966,
+      "step": 630130
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 2.7217417923012165e-06,
+      "loss": 0.7183,
+      "step": 630140
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 2.720593862401469e-06,
+      "loss": 0.7082,
+      "step": 630150
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 2.7194459325017217e-06,
+      "loss": 0.6591,
+      "step": 630160
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 2.718298002601974e-06,
+      "loss": 0.7347,
+      "step": 630170
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 2.717150072702227e-06,
+      "loss": 0.6774,
+      "step": 630180
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 2.7160021428024793e-06,
+      "loss": 0.7278,
+      "step": 630190
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 2.7148542129027317e-06,
+      "loss": 0.7613,
+      "step": 630200
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 2.7137062830029845e-06,
+      "loss": 0.6895,
+      "step": 630210
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 2.7125583531032365e-06,
+      "loss": 0.7311,
+      "step": 630220
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 2.7114104232034893e-06,
+      "loss": 0.6027,
+      "step": 630230
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 2.710262493303742e-06,
+      "loss": 0.6606,
+      "step": 630240
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 2.7091145634039945e-06,
+      "loss": 0.6769,
+      "step": 630250
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 2.707966633504247e-06,
+      "loss": 0.6388,
+      "step": 630260
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 2.7068187036044997e-06,
+      "loss": 0.6456,
+      "step": 630270
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 2.705670773704752e-06,
+      "loss": 0.6936,
+      "step": 630280
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 2.704522843805005e-06,
+      "loss": 0.6763,
+      "step": 630290
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 2.7033749139052573e-06,
+      "loss": 0.6962,
+      "step": 630300
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 2.7022269840055097e-06,
+      "loss": 0.7916,
+      "step": 630310
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 2.7010790541057625e-06,
+      "loss": 0.7095,
+      "step": 630320
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 2.6999311242060145e-06,
+      "loss": 0.7033,
+      "step": 630330
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 2.6987831943062673e-06,
+      "loss": 0.6536,
+      "step": 630340
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 2.69763526440652e-06,
+      "loss": 0.6886,
+      "step": 630350
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 2.6964873345067726e-06,
+      "loss": 0.6612,
+      "step": 630360
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 2.695339404607025e-06,
+      "loss": 0.7422,
+      "step": 630370
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 2.6941914747072778e-06,
+      "loss": 0.667,
+      "step": 630380
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 2.69304354480753e-06,
+      "loss": 0.7508,
+      "step": 630390
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 2.691895614907783e-06,
+      "loss": 0.6112,
+      "step": 630400
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 2.690747685008035e-06,
+      "loss": 0.7051,
+      "step": 630410
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 2.6895997551082878e-06,
+      "loss": 0.7241,
+      "step": 630420
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 2.6884518252085406e-06,
+      "loss": 0.6705,
+      "step": 630430
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 2.6873038953087926e-06,
+      "loss": 0.7812,
+      "step": 630440
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 2.6861559654090454e-06,
+      "loss": 0.7067,
+      "step": 630450
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 2.685008035509298e-06,
+      "loss": 0.6905,
+      "step": 630460
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 2.6838601056095506e-06,
+      "loss": 0.7366,
+      "step": 630470
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 2.682712175709803e-06,
+      "loss": 0.643,
+      "step": 630480
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 2.6815642458100554e-06,
+      "loss": 0.7259,
+      "step": 630490
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 2.680416315910308e-06,
+      "loss": 0.6938,
+      "step": 630500
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 2.679268386010561e-06,
+      "loss": 0.7189,
+      "step": 630510
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 2.678120456110813e-06,
+      "loss": 0.7141,
+      "step": 630520
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 2.676972526211066e-06,
+      "loss": 0.6288,
+      "step": 630530
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 2.6758245963113186e-06,
+      "loss": 0.7337,
+      "step": 630540
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 2.6746766664115706e-06,
+      "loss": 0.7469,
+      "step": 630550
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 2.6735287365118234e-06,
+      "loss": 0.6716,
+      "step": 630560
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 2.672380806612076e-06,
+      "loss": 0.6454,
+      "step": 630570
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 2.6712328767123286e-06,
+      "loss": 0.7115,
+      "step": 630580
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 2.670084946812581e-06,
+      "loss": 0.7226,
+      "step": 630590
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 2.6689370169128334e-06,
+      "loss": 0.7134,
+      "step": 630600
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 2.6677890870130862e-06,
+      "loss": 0.6962,
+      "step": 630610
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 2.666641157113339e-06,
+      "loss": 0.6729,
+      "step": 630620
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 2.665493227213591e-06,
+      "loss": 0.7682,
+      "step": 630630
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 2.664345297313844e-06,
+      "loss": 0.6984,
+      "step": 630640
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 2.6631973674140963e-06,
+      "loss": 0.6525,
+      "step": 630650
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 2.6620494375143486e-06,
+      "loss": 0.7114,
+      "step": 630660
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 2.6609015076146015e-06,
+      "loss": 0.735,
+      "step": 630670
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 2.659753577714854e-06,
+      "loss": 0.638,
+      "step": 630680
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 2.6586056478151067e-06,
+      "loss": 0.7384,
+      "step": 630690
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 2.657457717915359e-06,
+      "loss": 0.695,
+      "step": 630700
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 2.6563097880156115e-06,
+      "loss": 0.6648,
+      "step": 630710
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 2.6551618581158643e-06,
+      "loss": 0.7062,
+      "step": 630720
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 2.654013928216117e-06,
+      "loss": 0.7433,
+      "step": 630730
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 2.652865998316369e-06,
+      "loss": 0.7204,
+      "step": 630740
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 2.651718068416622e-06,
+      "loss": 0.7468,
+      "step": 630750
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 2.6505701385168743e-06,
+      "loss": 0.7116,
+      "step": 630760
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 2.6494222086171267e-06,
+      "loss": 0.758,
+      "step": 630770
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 2.6482742787173795e-06,
+      "loss": 0.6525,
+      "step": 630780
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 2.647126348817632e-06,
+      "loss": 0.6825,
+      "step": 630790
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 2.6459784189178847e-06,
+      "loss": 0.7689,
+      "step": 630800
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 2.644830489018137e-06,
+      "loss": 0.7379,
+      "step": 630810
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 2.6436825591183895e-06,
+      "loss": 0.6531,
+      "step": 630820
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 2.6425346292186423e-06,
+      "loss": 0.6859,
+      "step": 630830
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 2.6413866993188943e-06,
+      "loss": 0.6382,
+      "step": 630840
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 2.640238769419147e-06,
+      "loss": 0.7315,
+      "step": 630850
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 2.6390908395194e-06,
+      "loss": 0.6825,
+      "step": 630860
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 2.6379429096196523e-06,
+      "loss": 0.7296,
+      "step": 630870
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 2.6367949797199047e-06,
+      "loss": 0.6695,
+      "step": 630880
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 2.6356470498201576e-06,
+      "loss": 0.7205,
+      "step": 630890
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 2.63449911992041e-06,
+      "loss": 0.6192,
+      "step": 630900
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 2.6333511900206628e-06,
+      "loss": 0.7254,
+      "step": 630910
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 2.6322032601209147e-06,
+      "loss": 0.6753,
+      "step": 630920
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 2.6310553302211676e-06,
+      "loss": 0.6717,
+      "step": 630930
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 2.6299074003214204e-06,
+      "loss": 0.7294,
+      "step": 630940
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 2.6287594704216723e-06,
+      "loss": 0.6835,
+      "step": 630950
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 2.627611540521925e-06,
+      "loss": 0.6365,
+      "step": 630960
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 2.626463610622178e-06,
+      "loss": 0.7415,
+      "step": 630970
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 2.6253156807224304e-06,
+      "loss": 0.6839,
+      "step": 630980
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 2.6241677508226828e-06,
+      "loss": 0.6691,
+      "step": 630990
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 2.623019820922935e-06,
+      "loss": 0.6986,
+      "step": 631000
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 2.621871891023188e-06,
+      "loss": 0.7586,
+      "step": 631010
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 2.620723961123441e-06,
+      "loss": 0.7142,
+      "step": 631020
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 2.6195760312236928e-06,
+      "loss": 0.7556,
+      "step": 631030
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 2.6184281013239456e-06,
+      "loss": 0.7392,
+      "step": 631040
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 2.6172801714241984e-06,
+      "loss": 0.7886,
+      "step": 631050
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 2.6161322415244504e-06,
+      "loss": 0.759,
+      "step": 631060
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 2.614984311624703e-06,
+      "loss": 0.6964,
+      "step": 631070
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 2.6138363817249556e-06,
+      "loss": 0.7803,
+      "step": 631080
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 2.6126884518252084e-06,
+      "loss": 0.7032,
+      "step": 631090
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 2.611540521925461e-06,
+      "loss": 0.674,
+      "step": 631100
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 2.6103925920257132e-06,
+      "loss": 0.6398,
+      "step": 631110
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 2.609244662125966e-06,
+      "loss": 0.7339,
+      "step": 631120
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 2.608096732226219e-06,
+      "loss": 0.782,
+      "step": 631130
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 2.606948802326471e-06,
+      "loss": 0.7041,
+      "step": 631140
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 2.6058008724267236e-06,
+      "loss": 0.7092,
+      "step": 631150
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 2.604652942526976e-06,
+      "loss": 0.7182,
+      "step": 631160
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 2.6035050126272284e-06,
+      "loss": 0.7343,
+      "step": 631170
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 2.6023570827274813e-06,
+      "loss": 0.6757,
+      "step": 631180
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 2.6012091528277337e-06,
+      "loss": 0.7074,
+      "step": 631190
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 2.6000612229279865e-06,
+      "loss": 0.671,
+      "step": 631200
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 2.598913293028239e-06,
+      "loss": 0.6833,
+      "step": 631210
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 2.5977653631284913e-06,
+      "loss": 0.6737,
+      "step": 631220
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 2.596617433228744e-06,
+      "loss": 0.6498,
+      "step": 631230
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 2.595469503328997e-06,
+      "loss": 0.7068,
+      "step": 631240
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 2.594321573429249e-06,
+      "loss": 0.7262,
+      "step": 631250
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 2.5931736435295017e-06,
+      "loss": 0.7353,
+      "step": 631260
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 2.592025713629754e-06,
+      "loss": 0.7715,
+      "step": 631270
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 2.5908777837300065e-06,
+      "loss": 0.6707,
+      "step": 631280
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 2.5897298538302593e-06,
+      "loss": 0.6854,
+      "step": 631290
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 2.5885819239305117e-06,
+      "loss": 0.7095,
+      "step": 631300
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 2.5874339940307645e-06,
+      "loss": 0.5895,
+      "step": 631310
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 2.586286064131017e-06,
+      "loss": 0.7003,
+      "step": 631320
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 2.5851381342312693e-06,
+      "loss": 0.69,
+      "step": 631330
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 2.583990204331522e-06,
+      "loss": 0.7269,
+      "step": 631340
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 2.582842274431774e-06,
+      "loss": 0.6837,
+      "step": 631350
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 2.581694344532027e-06,
+      "loss": 0.6528,
+      "step": 631360
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 2.5805464146322797e-06,
+      "loss": 0.7614,
+      "step": 631370
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 2.579398484732532e-06,
+      "loss": 0.6919,
+      "step": 631380
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 2.5782505548327845e-06,
+      "loss": 0.7186,
+      "step": 631390
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 2.5771026249330373e-06,
+      "loss": 0.6673,
+      "step": 631400
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 2.5759546950332897e-06,
+      "loss": 0.7491,
+      "step": 631410
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 2.5748067651335426e-06,
+      "loss": 0.6323,
+      "step": 631420
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 2.5736588352337945e-06,
+      "loss": 0.7982,
+      "step": 631430
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 2.5725109053340473e-06,
+      "loss": 0.6799,
+      "step": 631440
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 2.5713629754343e-06,
+      "loss": 0.727,
+      "step": 631450
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 2.570215045534552e-06,
+      "loss": 0.689,
+      "step": 631460
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 2.569067115634805e-06,
+      "loss": 0.7078,
+      "step": 631470
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 2.5679191857350578e-06,
+      "loss": 0.7118,
+      "step": 631480
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 2.56677125583531e-06,
+      "loss": 0.7468,
+      "step": 631490
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 2.5656233259355626e-06,
+      "loss": 0.7086,
+      "step": 631500
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 2.564475396035815e-06,
+      "loss": 0.7229,
+      "step": 631510
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 2.5633274661360678e-06,
+      "loss": 0.7552,
+      "step": 631520
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 2.5621795362363206e-06,
+      "loss": 0.7079,
+      "step": 631530
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 2.5610316063365726e-06,
+      "loss": 0.658,
+      "step": 631540
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 2.5598836764368254e-06,
+      "loss": 0.7066,
+      "step": 631550
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 2.558735746537078e-06,
+      "loss": 0.6384,
+      "step": 631560
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 2.55758781663733e-06,
+      "loss": 0.6254,
+      "step": 631570
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 2.556439886737583e-06,
+      "loss": 0.6831,
+      "step": 631580
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 2.5552919568378354e-06,
+      "loss": 0.7697,
+      "step": 631590
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 2.5541440269380882e-06,
+      "loss": 0.7571,
+      "step": 631600
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 2.5529960970383406e-06,
+      "loss": 0.6872,
+      "step": 631610
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 2.551848167138593e-06,
+      "loss": 0.7151,
+      "step": 631620
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 2.550700237238846e-06,
+      "loss": 0.659,
+      "step": 631630
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 2.5495523073390986e-06,
+      "loss": 0.7165,
+      "step": 631640
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 2.5484043774393506e-06,
+      "loss": 0.7452,
+      "step": 631650
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 2.5472564475396034e-06,
+      "loss": 0.7179,
+      "step": 631660
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 2.5461085176398563e-06,
+      "loss": 0.6457,
+      "step": 631670
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 2.5449605877401082e-06,
+      "loss": 0.7117,
+      "step": 631680
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 2.543812657840361e-06,
+      "loss": 0.7074,
+      "step": 631690
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 2.5426647279406134e-06,
+      "loss": 0.7201,
+      "step": 631700
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 2.5415167980408663e-06,
+      "loss": 0.6525,
+      "step": 631710
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 2.5403688681411187e-06,
+      "loss": 0.6237,
+      "step": 631720
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 2.539220938241371e-06,
+      "loss": 0.6504,
+      "step": 631730
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 2.538073008341624e-06,
+      "loss": 0.6946,
+      "step": 631740
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 2.5369250784418767e-06,
+      "loss": 0.6126,
+      "step": 631750
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 2.5357771485421287e-06,
+      "loss": 0.7192,
+      "step": 631760
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 2.5346292186423815e-06,
+      "loss": 0.6556,
+      "step": 631770
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 2.533481288742634e-06,
+      "loss": 0.6745,
+      "step": 631780
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 2.5323333588428863e-06,
+      "loss": 0.6678,
+      "step": 631790
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 2.531185428943139e-06,
+      "loss": 0.7225,
+      "step": 631800
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 2.5300374990433915e-06,
+      "loss": 0.6966,
+      "step": 631810
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 2.5288895691436443e-06,
+      "loss": 0.6948,
+      "step": 631820
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 2.5277416392438967e-06,
+      "loss": 0.6995,
+      "step": 631830
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 2.526593709344149e-06,
+      "loss": 0.7356,
+      "step": 631840
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 2.525445779444402e-06,
+      "loss": 0.6763,
+      "step": 631850
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 2.524297849544654e-06,
+      "loss": 0.7087,
+      "step": 631860
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 2.5231499196449067e-06,
+      "loss": 0.6788,
+      "step": 631870
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 2.5220019897451595e-06,
+      "loss": 0.692,
+      "step": 631880
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 2.520854059845412e-06,
+      "loss": 0.6205,
+      "step": 631890
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 2.5197061299456643e-06,
+      "loss": 0.6396,
+      "step": 631900
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 2.518558200045917e-06,
+      "loss": 0.6858,
+      "step": 631910
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 2.5174102701461695e-06,
+      "loss": 0.6879,
+      "step": 631920
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 2.5162623402464223e-06,
+      "loss": 0.7915,
+      "step": 631930
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 2.5151144103466743e-06,
+      "loss": 0.8258,
+      "step": 631940
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 2.513966480446927e-06,
+      "loss": 0.674,
+      "step": 631950
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 2.51281855054718e-06,
+      "loss": 0.6615,
+      "step": 631960
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 2.511670620647432e-06,
+      "loss": 0.7353,
+      "step": 631970
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 2.5105226907476847e-06,
+      "loss": 0.7669,
+      "step": 631980
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 2.5093747608479376e-06,
+      "loss": 0.7912,
+      "step": 631990
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 2.50822683094819e-06,
+      "loss": 0.7236,
+      "step": 632000
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 2.5070789010484424e-06,
+      "loss": 0.7258,
+      "step": 632010
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 2.5059309711486947e-06,
+      "loss": 0.6866,
+      "step": 632020
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 2.5047830412489476e-06,
+      "loss": 0.7432,
+      "step": 632030
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 2.5036351113492004e-06,
+      "loss": 0.6457,
+      "step": 632040
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 2.5024871814494524e-06,
+      "loss": 0.7043,
+      "step": 632050
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 2.501339251549705e-06,
+      "loss": 0.6411,
+      "step": 632060
+    },
+    {
+      "epoch": 29.0,
+      "eval_cer": 0.668770515931328,
+      "eval_em": 0.007454017424975799,
+      "eval_f1": 0.007454017424975799,
+      "eval_loss": 1.3774735927581787,
+      "eval_runtime": 1293.7276,
+      "eval_samples_per_second": 7.985,
+      "eval_steps_per_second": 0.999,
+      "eval_wer": 0.9925459825750242,
+      "step": 632064
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 2.500191321649958e-06,
+      "loss": 0.6089,
+      "step": 632070
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 2.49904339175021e-06,
+      "loss": 0.7033,
+      "step": 632080
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 2.4978954618504628e-06,
+      "loss": 0.5975,
+      "step": 632090
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 2.4967475319507156e-06,
+      "loss": 0.6455,
+      "step": 632100
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 2.495599602050968e-06,
+      "loss": 0.7337,
+      "step": 632110
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 2.4944516721512204e-06,
+      "loss": 0.67,
+      "step": 632120
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 2.493303742251473e-06,
+      "loss": 0.6843,
+      "step": 632130
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 2.4921558123517256e-06,
+      "loss": 0.7108,
+      "step": 632140
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 2.4910078824519784e-06,
+      "loss": 0.6438,
+      "step": 632150
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 2.4898599525522304e-06,
+      "loss": 0.6123,
+      "step": 632160
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 2.4887120226524832e-06,
+      "loss": 0.6501,
+      "step": 632170
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 2.487564092752736e-06,
+      "loss": 0.7203,
+      "step": 632180
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 2.486416162852988e-06,
+      "loss": 0.6776,
+      "step": 632190
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 2.485268232953241e-06,
+      "loss": 0.6728,
+      "step": 632200
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 2.4841203030534932e-06,
+      "loss": 0.7566,
+      "step": 632210
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 2.482972373153746e-06,
+      "loss": 0.6775,
+      "step": 632220
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 2.4818244432539984e-06,
+      "loss": 0.6384,
+      "step": 632230
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 2.480676513354251e-06,
+      "loss": 0.6495,
+      "step": 632240
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 2.4795285834545037e-06,
+      "loss": 0.6617,
+      "step": 632250
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 2.4783806535547565e-06,
+      "loss": 0.7089,
+      "step": 632260
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 2.4772327236550084e-06,
+      "loss": 0.6814,
+      "step": 632270
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 2.4760847937552613e-06,
+      "loss": 0.7455,
+      "step": 632280
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 2.4749368638555132e-06,
+      "loss": 0.6446,
+      "step": 632290
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 2.473788933955766e-06,
+      "loss": 0.6419,
+      "step": 632300
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 2.472641004056019e-06,
+      "loss": 0.6597,
+      "step": 632310
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 2.4714930741562713e-06,
+      "loss": 0.692,
+      "step": 632320
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 2.4703451442565237e-06,
+      "loss": 0.7094,
+      "step": 632330
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 2.4691972143567765e-06,
+      "loss": 0.6895,
+      "step": 632340
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 2.468049284457029e-06,
+      "loss": 0.6503,
+      "step": 632350
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 2.4669013545572817e-06,
+      "loss": 0.7553,
+      "step": 632360
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 2.4657534246575337e-06,
+      "loss": 0.6481,
+      "step": 632370
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 2.4646054947577865e-06,
+      "loss": 0.7194,
+      "step": 632380
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 2.4634575648580393e-06,
+      "loss": 0.6459,
+      "step": 632390
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 2.4623096349582913e-06,
+      "loss": 0.7022,
+      "step": 632400
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 2.461161705058544e-06,
+      "loss": 0.7145,
+      "step": 632410
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 2.460013775158797e-06,
+      "loss": 0.6557,
+      "step": 632420
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 2.4588658452590493e-06,
+      "loss": 0.6653,
+      "step": 632430
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 2.4577179153593017e-06,
+      "loss": 0.5996,
+      "step": 632440
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 2.456569985459554e-06,
+      "loss": 0.6908,
+      "step": 632450
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 2.455422055559807e-06,
+      "loss": 0.7727,
+      "step": 632460
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 2.4542741256600597e-06,
+      "loss": 0.6835,
+      "step": 632470
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 2.4531261957603117e-06,
+      "loss": 0.6194,
+      "step": 632480
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 2.4519782658605645e-06,
+      "loss": 0.6917,
+      "step": 632490
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 2.4508303359608174e-06,
+      "loss": 0.6089,
+      "step": 632500
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 2.4496824060610693e-06,
+      "loss": 0.6266,
+      "step": 632510
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 2.448534476161322e-06,
+      "loss": 0.7089,
+      "step": 632520
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 2.447386546261575e-06,
+      "loss": 0.5908,
+      "step": 632530
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 2.4462386163618274e-06,
+      "loss": 0.6644,
+      "step": 632540
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 2.4450906864620798e-06,
+      "loss": 0.6397,
+      "step": 632550
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 2.443942756562332e-06,
+      "loss": 0.7082,
+      "step": 632560
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 2.442794826662585e-06,
+      "loss": 0.6607,
+      "step": 632570
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 2.4416468967628378e-06,
+      "loss": 0.7887,
+      "step": 632580
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 2.4404989668630898e-06,
+      "loss": 0.6734,
+      "step": 632590
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 2.4393510369633426e-06,
+      "loss": 0.7181,
+      "step": 632600
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 2.4382031070635954e-06,
+      "loss": 0.6944,
+      "step": 632610
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 2.4370551771638474e-06,
+      "loss": 0.6413,
+      "step": 632620
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 2.4359072472641e-06,
+      "loss": 0.6834,
+      "step": 632630
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 2.4347593173643526e-06,
+      "loss": 0.6409,
+      "step": 632640
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 2.4336113874646054e-06,
+      "loss": 0.634,
+      "step": 632650
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 2.432463457564858e-06,
+      "loss": 0.596,
+      "step": 632660
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 2.43131552766511e-06,
+      "loss": 0.7172,
+      "step": 632670
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 2.430167597765363e-06,
+      "loss": 0.639,
+      "step": 632680
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 2.429019667865616e-06,
+      "loss": 0.6932,
+      "step": 632690
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 2.427871737965868e-06,
+      "loss": 0.6713,
+      "step": 632700
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 2.4267238080661206e-06,
+      "loss": 0.6473,
+      "step": 632710
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 2.425575878166373e-06,
+      "loss": 0.6792,
+      "step": 632720
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 2.4244279482666254e-06,
+      "loss": 0.6517,
+      "step": 632730
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 2.4232800183668782e-06,
+      "loss": 0.663,
+      "step": 632740
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 2.4221320884671306e-06,
+      "loss": 0.6244,
+      "step": 632750
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 2.4209841585673834e-06,
+      "loss": 0.6282,
+      "step": 632760
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 2.419836228667636e-06,
+      "loss": 0.7018,
+      "step": 632770
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 2.4186882987678882e-06,
+      "loss": 0.6645,
+      "step": 632780
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 2.417540368868141e-06,
+      "loss": 0.6679,
+      "step": 632790
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 2.416392438968393e-06,
+      "loss": 0.705,
+      "step": 632800
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 2.415244509068646e-06,
+      "loss": 0.6974,
+      "step": 632810
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 2.4140965791688987e-06,
+      "loss": 0.711,
+      "step": 632820
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 2.412948649269151e-06,
+      "loss": 0.6464,
+      "step": 632830
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 2.4118007193694035e-06,
+      "loss": 0.6579,
+      "step": 632840
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 2.4106527894696563e-06,
+      "loss": 0.6667,
+      "step": 632850
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 2.4095048595699087e-06,
+      "loss": 0.691,
+      "step": 632860
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 2.4083569296701615e-06,
+      "loss": 0.7038,
+      "step": 632870
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 2.4072089997704135e-06,
+      "loss": 0.732,
+      "step": 632880
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 2.4060610698706663e-06,
+      "loss": 0.6201,
+      "step": 632890
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 2.404913139970919e-06,
+      "loss": 0.6417,
+      "step": 632900
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 2.403765210071171e-06,
+      "loss": 0.6751,
+      "step": 632910
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 2.402617280171424e-06,
+      "loss": 0.6821,
+      "step": 632920
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 2.4014693502716767e-06,
+      "loss": 0.7421,
+      "step": 632930
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 2.400321420371929e-06,
+      "loss": 0.726,
+      "step": 632940
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 2.3991734904721815e-06,
+      "loss": 0.7271,
+      "step": 632950
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 2.3980255605724343e-06,
+      "loss": 0.7137,
+      "step": 632960
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 2.3968776306726867e-06,
+      "loss": 0.6607,
+      "step": 632970
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 2.3957297007729395e-06,
+      "loss": 0.652,
+      "step": 632980
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 2.3945817708731915e-06,
+      "loss": 0.6784,
+      "step": 632990
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 2.3934338409734443e-06,
+      "loss": 0.6379,
+      "step": 633000
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 2.392285911073697e-06,
+      "loss": 0.6617,
+      "step": 633010
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 2.391137981173949e-06,
+      "loss": 0.6527,
+      "step": 633020
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 2.389990051274202e-06,
+      "loss": 0.7055,
+      "step": 633030
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 2.3888421213744548e-06,
+      "loss": 0.7315,
+      "step": 633040
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 2.387694191474707e-06,
+      "loss": 0.5947,
+      "step": 633050
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 2.3865462615749595e-06,
+      "loss": 0.6508,
+      "step": 633060
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 2.385398331675212e-06,
+      "loss": 0.6898,
+      "step": 633070
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 2.3842504017754648e-06,
+      "loss": 0.6809,
+      "step": 633080
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 2.3831024718757176e-06,
+      "loss": 0.7406,
+      "step": 633090
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 2.3819545419759695e-06,
+      "loss": 0.653,
+      "step": 633100
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 2.3808066120762224e-06,
+      "loss": 0.7018,
+      "step": 633110
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 2.379658682176475e-06,
+      "loss": 0.6195,
+      "step": 633120
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 2.378510752276727e-06,
+      "loss": 0.6189,
+      "step": 633130
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 2.37736282237698e-06,
+      "loss": 0.7227,
+      "step": 633140
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 2.3762148924772324e-06,
+      "loss": 0.6481,
+      "step": 633150
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 2.375066962577485e-06,
+      "loss": 0.6499,
+      "step": 633160
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 2.3739190326777376e-06,
+      "loss": 0.6601,
+      "step": 633170
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 2.37277110277799e-06,
+      "loss": 0.6653,
+      "step": 633180
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 2.371623172878243e-06,
+      "loss": 0.6353,
+      "step": 633190
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 2.3704752429784956e-06,
+      "loss": 0.6739,
+      "step": 633200
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 2.3693273130787476e-06,
+      "loss": 0.6738,
+      "step": 633210
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 2.3681793831790004e-06,
+      "loss": 0.735,
+      "step": 633220
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 2.367031453279253e-06,
+      "loss": 0.7844,
+      "step": 633230
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 2.365883523379505e-06,
+      "loss": 0.7277,
+      "step": 633240
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 2.364735593479758e-06,
+      "loss": 0.6679,
+      "step": 633250
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 2.3635876635800104e-06,
+      "loss": 0.7833,
+      "step": 633260
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 2.3624397336802632e-06,
+      "loss": 0.7047,
+      "step": 633270
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 2.3612918037805156e-06,
+      "loss": 0.6655,
+      "step": 633280
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 2.360143873880768e-06,
+      "loss": 0.6704,
+      "step": 633290
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 2.358995943981021e-06,
+      "loss": 0.7152,
+      "step": 633300
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 2.357848014081273e-06,
+      "loss": 0.75,
+      "step": 633310
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 2.3567000841815256e-06,
+      "loss": 0.7586,
+      "step": 633320
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 2.3555521542817785e-06,
+      "loss": 0.7521,
+      "step": 633330
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 2.354404224382031e-06,
+      "loss": 0.7265,
+      "step": 633340
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 2.3532562944822832e-06,
+      "loss": 0.7002,
+      "step": 633350
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 2.352108364582536e-06,
+      "loss": 0.6333,
+      "step": 633360
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 2.3509604346827885e-06,
+      "loss": 0.6898,
+      "step": 633370
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 2.3498125047830413e-06,
+      "loss": 0.6847,
+      "step": 633380
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 2.3486645748832932e-06,
+      "loss": 0.6699,
+      "step": 633390
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 2.347516644983546e-06,
+      "loss": 0.6741,
+      "step": 633400
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 2.346368715083799e-06,
+      "loss": 0.6624,
+      "step": 633410
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 2.345220785184051e-06,
+      "loss": 0.6205,
+      "step": 633420
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 2.3440728552843037e-06,
+      "loss": 0.6939,
+      "step": 633430
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 2.342924925384556e-06,
+      "loss": 0.7777,
+      "step": 633440
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 2.341776995484809e-06,
+      "loss": 0.6674,
+      "step": 633450
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 2.3406290655850613e-06,
+      "loss": 0.6489,
+      "step": 633460
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 2.339481135685314e-06,
+      "loss": 0.6151,
+      "step": 633470
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 2.3383332057855665e-06,
+      "loss": 0.6722,
+      "step": 633480
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 2.3371852758858193e-06,
+      "loss": 0.7196,
+      "step": 633490
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 2.3360373459860717e-06,
+      "loss": 0.7182,
+      "step": 633500
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 2.334889416086324e-06,
+      "loss": 0.6418,
+      "step": 633510
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 2.3337414861865765e-06,
+      "loss": 0.6341,
+      "step": 633520
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 2.3325935562868293e-06,
+      "loss": 0.6413,
+      "step": 633530
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 2.3314456263870817e-06,
+      "loss": 0.678,
+      "step": 633540
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 2.330297696487334e-06,
+      "loss": 0.6761,
+      "step": 633550
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 2.329149766587587e-06,
+      "loss": 0.6958,
+      "step": 633560
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 2.3280018366878393e-06,
+      "loss": 0.6564,
+      "step": 633570
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 2.326853906788092e-06,
+      "loss": 0.6727,
+      "step": 633580
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 2.3257059768883445e-06,
+      "loss": 0.6588,
+      "step": 633590
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 2.324558046988597e-06,
+      "loss": 0.6391,
+      "step": 633600
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 2.3234101170888498e-06,
+      "loss": 0.7076,
+      "step": 633610
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 2.322262187189102e-06,
+      "loss": 0.7522,
+      "step": 633620
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 2.3211142572893545e-06,
+      "loss": 0.689,
+      "step": 633630
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 2.319966327389607e-06,
+      "loss": 0.7097,
+      "step": 633640
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 2.3188183974898598e-06,
+      "loss": 0.6595,
+      "step": 633650
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 2.317670467590112e-06,
+      "loss": 0.6704,
+      "step": 633660
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 2.316522537690365e-06,
+      "loss": 0.7178,
+      "step": 633670
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 2.3153746077906174e-06,
+      "loss": 0.7266,
+      "step": 633680
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 2.31422667789087e-06,
+      "loss": 0.7491,
+      "step": 633690
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 2.3130787479911226e-06,
+      "loss": 0.6447,
+      "step": 633700
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 2.311930818091375e-06,
+      "loss": 0.6987,
+      "step": 633710
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 2.3107828881916274e-06,
+      "loss": 0.6971,
+      "step": 633720
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 2.30963495829188e-06,
+      "loss": 0.7972,
+      "step": 633730
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 2.3084870283921326e-06,
+      "loss": 0.6202,
+      "step": 633740
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 2.307339098492385e-06,
+      "loss": 0.6563,
+      "step": 633750
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 2.306191168592638e-06,
+      "loss": 0.7028,
+      "step": 633760
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 2.30504323869289e-06,
+      "loss": 0.7123,
+      "step": 633770
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 2.303895308793143e-06,
+      "loss": 0.7528,
+      "step": 633780
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 2.3027473788933954e-06,
+      "loss": 0.6668,
+      "step": 633790
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 2.301599448993648e-06,
+      "loss": 0.7038,
+      "step": 633800
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 2.3004515190939006e-06,
+      "loss": 0.6599,
+      "step": 633810
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 2.299303589194153e-06,
+      "loss": 0.6554,
+      "step": 633820
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 2.2981556592944054e-06,
+      "loss": 0.6468,
+      "step": 633830
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 2.2970077293946582e-06,
+      "loss": 0.6688,
+      "step": 633840
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 2.2958597994949106e-06,
+      "loss": 0.6707,
+      "step": 633850
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 2.294711869595163e-06,
+      "loss": 0.7226,
+      "step": 633860
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 2.293563939695416e-06,
+      "loss": 0.7033,
+      "step": 633870
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 2.2924160097956682e-06,
+      "loss": 0.6567,
+      "step": 633880
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 2.291268079895921e-06,
+      "loss": 0.6493,
+      "step": 633890
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 2.2901201499961735e-06,
+      "loss": 0.7195,
+      "step": 633900
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 2.288972220096426e-06,
+      "loss": 0.6908,
+      "step": 633910
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 2.2878242901966787e-06,
+      "loss": 0.6945,
+      "step": 633920
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 2.286676360296931e-06,
+      "loss": 0.7179,
+      "step": 633930
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 2.2855284303971835e-06,
+      "loss": 0.7136,
+      "step": 633940
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 2.284380500497436e-06,
+      "loss": 0.6548,
+      "step": 633950
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 2.2832325705976887e-06,
+      "loss": 0.7322,
+      "step": 633960
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 2.282084640697941e-06,
+      "loss": 0.7061,
+      "step": 633970
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 2.280936710798194e-06,
+      "loss": 0.7272,
+      "step": 633980
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 2.2797887808984463e-06,
+      "loss": 0.6073,
+      "step": 633990
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 2.278640850998699e-06,
+      "loss": 0.7128,
+      "step": 634000
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 2.2774929210989515e-06,
+      "loss": 0.6918,
+      "step": 634010
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 2.276344991199204e-06,
+      "loss": 0.718,
+      "step": 634020
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 2.2751970612994563e-06,
+      "loss": 0.6658,
+      "step": 634030
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 2.274049131399709e-06,
+      "loss": 0.7429,
+      "step": 634040
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 2.2729012014999615e-06,
+      "loss": 0.6994,
+      "step": 634050
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 2.271753271600214e-06,
+      "loss": 0.7221,
+      "step": 634060
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 2.2706053417004667e-06,
+      "loss": 0.6554,
+      "step": 634070
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 2.269457411800719e-06,
+      "loss": 0.6359,
+      "step": 634080
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 2.268309481900972e-06,
+      "loss": 0.706,
+      "step": 634090
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 2.2671615520012243e-06,
+      "loss": 0.6786,
+      "step": 634100
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 2.2660136221014767e-06,
+      "loss": 0.7277,
+      "step": 634110
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 2.2648656922017295e-06,
+      "loss": 0.6227,
+      "step": 634120
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 2.263717762301982e-06,
+      "loss": 0.6824,
+      "step": 634130
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 2.2625698324022343e-06,
+      "loss": 0.7593,
+      "step": 634140
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 2.2614219025024867e-06,
+      "loss": 0.6049,
+      "step": 634150
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 2.2602739726027396e-06,
+      "loss": 0.6648,
+      "step": 634160
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 2.259126042702992e-06,
+      "loss": 0.6438,
+      "step": 634170
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 2.2579781128032448e-06,
+      "loss": 0.7633,
+      "step": 634180
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 2.256830182903497e-06,
+      "loss": 0.7382,
+      "step": 634190
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 2.25568225300375e-06,
+      "loss": 0.6579,
+      "step": 634200
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 2.2545343231040024e-06,
+      "loss": 0.6291,
+      "step": 634210
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 2.2533863932042548e-06,
+      "loss": 0.7128,
+      "step": 634220
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 2.252238463304507e-06,
+      "loss": 0.753,
+      "step": 634230
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 2.25109053340476e-06,
+      "loss": 0.6981,
+      "step": 634240
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 2.2499426035050124e-06,
+      "loss": 0.6356,
+      "step": 634250
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 2.2487946736052648e-06,
+      "loss": 0.6981,
+      "step": 634260
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 2.2476467437055176e-06,
+      "loss": 0.6582,
+      "step": 634270
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 2.24649881380577e-06,
+      "loss": 0.6195,
+      "step": 634280
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 2.245350883906023e-06,
+      "loss": 0.6469,
+      "step": 634290
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 2.244202954006275e-06,
+      "loss": 0.6279,
+      "step": 634300
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 2.243055024106528e-06,
+      "loss": 0.6754,
+      "step": 634310
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 2.2419070942067804e-06,
+      "loss": 0.6685,
+      "step": 634320
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 2.240759164307033e-06,
+      "loss": 0.6834,
+      "step": 634330
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 2.239611234407285e-06,
+      "loss": 0.6599,
+      "step": 634340
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 2.238463304507538e-06,
+      "loss": 0.739,
+      "step": 634350
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 2.2373153746077904e-06,
+      "loss": 0.6506,
+      "step": 634360
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 2.236167444708043e-06,
+      "loss": 0.6959,
+      "step": 634370
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 2.2350195148082956e-06,
+      "loss": 0.6856,
+      "step": 634380
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 2.233871584908548e-06,
+      "loss": 0.6586,
+      "step": 634390
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 2.232723655008801e-06,
+      "loss": 0.692,
+      "step": 634400
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 2.2315757251090532e-06,
+      "loss": 0.6636,
+      "step": 634410
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 2.2304277952093056e-06,
+      "loss": 0.7572,
+      "step": 634420
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 2.2292798653095585e-06,
+      "loss": 0.6199,
+      "step": 634430
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 2.228131935409811e-06,
+      "loss": 0.629,
+      "step": 634440
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 2.2269840055100633e-06,
+      "loss": 0.6515,
+      "step": 634450
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 2.2258360756103156e-06,
+      "loss": 0.6315,
+      "step": 634460
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 2.2246881457105685e-06,
+      "loss": 0.6955,
+      "step": 634470
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 2.223540215810821e-06,
+      "loss": 0.6517,
+      "step": 634480
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 2.2223922859110737e-06,
+      "loss": 0.7077,
+      "step": 634490
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 2.221244356011326e-06,
+      "loss": 0.6329,
+      "step": 634500
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 2.220096426111579e-06,
+      "loss": 0.6556,
+      "step": 634510
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 2.2189484962118313e-06,
+      "loss": 0.7234,
+      "step": 634520
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 2.2178005663120837e-06,
+      "loss": 0.6363,
+      "step": 634530
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 2.216652636412336e-06,
+      "loss": 0.7152,
+      "step": 634540
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 2.215504706512589e-06,
+      "loss": 0.6449,
+      "step": 634550
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 2.2143567766128413e-06,
+      "loss": 0.685,
+      "step": 634560
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 2.2132088467130937e-06,
+      "loss": 0.6624,
+      "step": 634570
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 2.212060916813346e-06,
+      "loss": 0.705,
+      "step": 634580
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 2.210912986913599e-06,
+      "loss": 0.6919,
+      "step": 634590
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 2.2097650570138513e-06,
+      "loss": 0.7285,
+      "step": 634600
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 2.208617127114104e-06,
+      "loss": 0.7725,
+      "step": 634610
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 2.2074691972143565e-06,
+      "loss": 0.6069,
+      "step": 634620
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 2.2063212673146093e-06,
+      "loss": 0.5871,
+      "step": 634630
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 2.2051733374148617e-06,
+      "loss": 0.7419,
+      "step": 634640
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 2.204025407515114e-06,
+      "loss": 0.7102,
+      "step": 634650
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 2.2028774776153665e-06,
+      "loss": 0.7362,
+      "step": 634660
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 2.2017295477156193e-06,
+      "loss": 0.6548,
+      "step": 634670
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 2.2005816178158717e-06,
+      "loss": 0.766,
+      "step": 634680
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 2.199433687916124e-06,
+      "loss": 0.7127,
+      "step": 634690
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 2.198285758016377e-06,
+      "loss": 0.7621,
+      "step": 634700
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 2.1971378281166293e-06,
+      "loss": 0.6661,
+      "step": 634710
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 2.195989898216882e-06,
+      "loss": 0.6519,
+      "step": 634720
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 2.1948419683171346e-06,
+      "loss": 0.685,
+      "step": 634730
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 2.1936940384173874e-06,
+      "loss": 0.6497,
+      "step": 634740
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 2.1925461085176398e-06,
+      "loss": 0.6246,
+      "step": 634750
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 2.191398178617892e-06,
+      "loss": 0.7214,
+      "step": 634760
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 2.1902502487181446e-06,
+      "loss": 0.728,
+      "step": 634770
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 2.1891023188183974e-06,
+      "loss": 0.6907,
+      "step": 634780
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 2.1879543889186498e-06,
+      "loss": 0.6663,
+      "step": 634790
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 2.186806459018902e-06,
+      "loss": 0.6917,
+      "step": 634800
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 2.185658529119155e-06,
+      "loss": 0.6246,
+      "step": 634810
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 2.1845105992194074e-06,
+      "loss": 0.6684,
+      "step": 634820
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 2.18336266931966e-06,
+      "loss": 0.6321,
+      "step": 634830
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 2.1822147394199126e-06,
+      "loss": 0.6642,
+      "step": 634840
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 2.181066809520165e-06,
+      "loss": 0.6667,
+      "step": 634850
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 2.179918879620418e-06,
+      "loss": 0.7814,
+      "step": 634860
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 2.1787709497206702e-06,
+      "loss": 0.6539,
+      "step": 634870
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 2.1776230198209226e-06,
+      "loss": 0.6601,
+      "step": 634880
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 2.176475089921175e-06,
+      "loss": 0.6874,
+      "step": 634890
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 2.175327160021428e-06,
+      "loss": 0.7141,
+      "step": 634900
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 2.1741792301216802e-06,
+      "loss": 0.6766,
+      "step": 634910
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 2.173031300221933e-06,
+      "loss": 0.6536,
+      "step": 634920
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 2.1718833703221854e-06,
+      "loss": 0.7103,
+      "step": 634930
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 2.1707354404224383e-06,
+      "loss": 0.6377,
+      "step": 634940
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 2.1695875105226906e-06,
+      "loss": 0.6337,
+      "step": 634950
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 2.168439580622943e-06,
+      "loss": 0.6572,
+      "step": 634960
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 2.1672916507231954e-06,
+      "loss": 0.7131,
+      "step": 634970
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 2.1661437208234483e-06,
+      "loss": 0.637,
+      "step": 634980
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 2.1649957909237007e-06,
+      "loss": 0.7487,
+      "step": 634990
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 2.163847861023953e-06,
+      "loss": 0.6587,
+      "step": 635000
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 2.162699931124206e-06,
+      "loss": 0.7555,
+      "step": 635010
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 2.1615520012244583e-06,
+      "loss": 0.7782,
+      "step": 635020
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 2.160404071324711e-06,
+      "loss": 0.6151,
+      "step": 635030
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 2.1592561414249635e-06,
+      "loss": 0.7187,
+      "step": 635040
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 2.158108211525216e-06,
+      "loss": 0.6191,
+      "step": 635050
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 2.1569602816254687e-06,
+      "loss": 0.6771,
+      "step": 635060
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 2.155812351725721e-06,
+      "loss": 0.6177,
+      "step": 635070
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 2.1546644218259735e-06,
+      "loss": 0.7067,
+      "step": 635080
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 2.153516491926226e-06,
+      "loss": 0.7285,
+      "step": 635090
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 2.1523685620264787e-06,
+      "loss": 0.7165,
+      "step": 635100
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 2.151220632126731e-06,
+      "loss": 0.7558,
+      "step": 635110
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 2.150072702226984e-06,
+      "loss": 0.6608,
+      "step": 635120
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 2.1489247723272363e-06,
+      "loss": 0.6841,
+      "step": 635130
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 2.147776842427489e-06,
+      "loss": 0.631,
+      "step": 635140
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 2.1466289125277415e-06,
+      "loss": 0.7299,
+      "step": 635150
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 2.145480982627994e-06,
+      "loss": 0.7235,
+      "step": 635160
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 2.1443330527282463e-06,
+      "loss": 0.6342,
+      "step": 635170
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 2.143185122828499e-06,
+      "loss": 0.6266,
+      "step": 635180
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 2.1420371929287515e-06,
+      "loss": 0.5981,
+      "step": 635190
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 2.140889263029004e-06,
+      "loss": 0.7075,
+      "step": 635200
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 2.1397413331292567e-06,
+      "loss": 0.6909,
+      "step": 635210
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 2.138593403229509e-06,
+      "loss": 0.7172,
+      "step": 635220
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 2.137445473329762e-06,
+      "loss": 0.6331,
+      "step": 635230
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 2.1362975434300143e-06,
+      "loss": 0.6833,
+      "step": 635240
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 2.135149613530267e-06,
+      "loss": 0.7087,
+      "step": 635250
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 2.1340016836305196e-06,
+      "loss": 0.7195,
+      "step": 635260
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 2.132853753730772e-06,
+      "loss": 0.6481,
+      "step": 635270
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 2.1317058238310244e-06,
+      "loss": 0.6704,
+      "step": 635280
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 2.130557893931277e-06,
+      "loss": 0.6345,
+      "step": 635290
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 2.1294099640315296e-06,
+      "loss": 0.7421,
+      "step": 635300
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 2.128262034131782e-06,
+      "loss": 0.7072,
+      "step": 635310
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 2.1271141042320348e-06,
+      "loss": 0.6704,
+      "step": 635320
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 2.125966174332287e-06,
+      "loss": 0.7204,
+      "step": 635330
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 2.12481824443254e-06,
+      "loss": 0.6478,
+      "step": 635340
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 2.1236703145327924e-06,
+      "loss": 0.7249,
+      "step": 635350
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 2.1225223846330448e-06,
+      "loss": 0.6923,
+      "step": 635360
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 2.1213744547332976e-06,
+      "loss": 0.7326,
+      "step": 635370
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 2.12022652483355e-06,
+      "loss": 0.7154,
+      "step": 635380
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 2.1190785949338024e-06,
+      "loss": 0.6365,
+      "step": 635390
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 2.117930665034055e-06,
+      "loss": 0.6564,
+      "step": 635400
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 2.1167827351343076e-06,
+      "loss": 0.6536,
+      "step": 635410
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 2.11563480523456e-06,
+      "loss": 0.648,
+      "step": 635420
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 2.114486875334813e-06,
+      "loss": 0.7373,
+      "step": 635430
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 2.1133389454350652e-06,
+      "loss": 0.6342,
+      "step": 635440
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 2.112191015535318e-06,
+      "loss": 0.6478,
+      "step": 635450
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 2.1110430856355704e-06,
+      "loss": 0.6883,
+      "step": 635460
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 2.109895155735823e-06,
+      "loss": 0.6933,
+      "step": 635470
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 2.1087472258360752e-06,
+      "loss": 0.732,
+      "step": 635480
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 2.107599295936328e-06,
+      "loss": 0.7042,
+      "step": 635490
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 2.1064513660365804e-06,
+      "loss": 0.6501,
+      "step": 635500
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 2.105303436136833e-06,
+      "loss": 0.6598,
+      "step": 635510
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 2.1041555062370857e-06,
+      "loss": 0.6855,
+      "step": 635520
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 2.103007576337338e-06,
+      "loss": 0.6789,
+      "step": 635530
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 2.101859646437591e-06,
+      "loss": 0.7155,
+      "step": 635540
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 2.1007117165378433e-06,
+      "loss": 0.6767,
+      "step": 635550
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 2.0995637866380957e-06,
+      "loss": 0.7328,
+      "step": 635560
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 2.0984158567383485e-06,
+      "loss": 0.6739,
+      "step": 635570
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 2.097267926838601e-06,
+      "loss": 0.6735,
+      "step": 635580
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 2.0961199969388533e-06,
+      "loss": 0.7033,
+      "step": 635590
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 2.0949720670391057e-06,
+      "loss": 0.7023,
+      "step": 635600
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 2.0938241371393585e-06,
+      "loss": 0.6804,
+      "step": 635610
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 2.092676207239611e-06,
+      "loss": 0.6574,
+      "step": 635620
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 2.0915282773398637e-06,
+      "loss": 0.6525,
+      "step": 635630
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 2.090380347440116e-06,
+      "loss": 0.63,
+      "step": 635640
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 2.089232417540369e-06,
+      "loss": 0.6758,
+      "step": 635650
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 2.0880844876406213e-06,
+      "loss": 0.7535,
+      "step": 635660
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 2.0869365577408737e-06,
+      "loss": 0.6249,
+      "step": 635670
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 2.0857886278411265e-06,
+      "loss": 0.7161,
+      "step": 635680
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 2.084640697941379e-06,
+      "loss": 0.7346,
+      "step": 635690
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 2.0834927680416313e-06,
+      "loss": 0.6543,
+      "step": 635700
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 2.0823448381418837e-06,
+      "loss": 0.735,
+      "step": 635710
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 2.0811969082421365e-06,
+      "loss": 0.7291,
+      "step": 635720
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 2.080048978342389e-06,
+      "loss": 0.6565,
+      "step": 635730
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 2.0789010484426417e-06,
+      "loss": 0.5888,
+      "step": 635740
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 2.077753118542894e-06,
+      "loss": 0.6667,
+      "step": 635750
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 2.076605188643147e-06,
+      "loss": 0.7333,
+      "step": 635760
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 2.0754572587433993e-06,
+      "loss": 0.6762,
+      "step": 635770
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 2.0743093288436517e-06,
+      "loss": 0.7037,
+      "step": 635780
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 2.073161398943904e-06,
+      "loss": 0.6929,
+      "step": 635790
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 2.072013469044157e-06,
+      "loss": 0.6389,
+      "step": 635800
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 2.0708655391444094e-06,
+      "loss": 0.6385,
+      "step": 635810
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 2.0697176092446618e-06,
+      "loss": 0.6196,
+      "step": 635820
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 2.0685696793449146e-06,
+      "loss": 0.76,
+      "step": 635830
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 2.067421749445167e-06,
+      "loss": 0.7218,
+      "step": 635840
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 2.0662738195454198e-06,
+      "loss": 0.7211,
+      "step": 635850
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 2.065125889645672e-06,
+      "loss": 0.6329,
+      "step": 635860
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 2.0639779597459246e-06,
+      "loss": 0.7042,
+      "step": 635870
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 2.0628300298461774e-06,
+      "loss": 0.7164,
+      "step": 635880
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 2.06168209994643e-06,
+      "loss": 0.6753,
+      "step": 635890
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 2.060534170046682e-06,
+      "loss": 0.6415,
+      "step": 635900
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 2.0593862401469346e-06,
+      "loss": 0.6664,
+      "step": 635910
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 2.0582383102471874e-06,
+      "loss": 0.6864,
+      "step": 635920
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 2.05709038034744e-06,
+      "loss": 0.67,
+      "step": 635930
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 2.0559424504476926e-06,
+      "loss": 0.652,
+      "step": 635940
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 2.054794520547945e-06,
+      "loss": 0.7277,
+      "step": 635950
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 2.053646590648198e-06,
+      "loss": 0.6736,
+      "step": 635960
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 2.0524986607484502e-06,
+      "loss": 0.7313,
+      "step": 635970
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 2.0513507308487026e-06,
+      "loss": 0.6707,
+      "step": 635980
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 2.050202800948955e-06,
+      "loss": 0.713,
+      "step": 635990
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 2.049054871049208e-06,
+      "loss": 0.7137,
+      "step": 636000
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 2.0479069411494602e-06,
+      "loss": 0.6153,
+      "step": 636010
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 2.0467590112497126e-06,
+      "loss": 0.6883,
+      "step": 636020
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 2.0456110813499654e-06,
+      "loss": 0.6814,
+      "step": 636030
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 2.044463151450218e-06,
+      "loss": 0.6585,
+      "step": 636040
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 2.0433152215504707e-06,
+      "loss": 0.6837,
+      "step": 636050
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 2.042167291650723e-06,
+      "loss": 0.6449,
+      "step": 636060
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 2.041019361750976e-06,
+      "loss": 0.6779,
+      "step": 636070
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 2.0398714318512283e-06,
+      "loss": 0.6416,
+      "step": 636080
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 2.0387235019514807e-06,
+      "loss": 0.5175,
+      "step": 636090
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 2.037575572051733e-06,
+      "loss": 0.546,
+      "step": 636100
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 2.036427642151986e-06,
+      "loss": 0.7721,
+      "step": 636110
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 2.0352797122522383e-06,
+      "loss": 0.6944,
+      "step": 636120
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 2.0341317823524907e-06,
+      "loss": 0.5908,
+      "step": 636130
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 2.0329838524527435e-06,
+      "loss": 0.6485,
+      "step": 636140
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 2.031835922552996e-06,
+      "loss": 0.6732,
+      "step": 636150
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 2.0306879926532487e-06,
+      "loss": 0.6655,
+      "step": 636160
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 2.029540062753501e-06,
+      "loss": 0.7017,
+      "step": 636170
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 2.0283921328537535e-06,
+      "loss": 0.7569,
+      "step": 636180
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 2.0272442029540063e-06,
+      "loss": 0.6915,
+      "step": 636190
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 2.0260962730542587e-06,
+      "loss": 0.6785,
+      "step": 636200
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 2.024948343154511e-06,
+      "loss": 0.7016,
+      "step": 636210
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 2.0238004132547635e-06,
+      "loss": 0.7126,
+      "step": 636220
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 2.0226524833550163e-06,
+      "loss": 0.6689,
+      "step": 636230
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 2.0215045534552687e-06,
+      "loss": 0.6632,
+      "step": 636240
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 2.0203566235555215e-06,
+      "loss": 0.6618,
+      "step": 636250
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 2.019208693655774e-06,
+      "loss": 0.7196,
+      "step": 636260
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 2.0180607637560267e-06,
+      "loss": 0.674,
+      "step": 636270
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 2.016912833856279e-06,
+      "loss": 0.723,
+      "step": 636280
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 2.0157649039565315e-06,
+      "loss": 0.6775,
+      "step": 636290
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 2.014616974056784e-06,
+      "loss": 0.6892,
+      "step": 636300
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 2.0134690441570367e-06,
+      "loss": 0.7609,
+      "step": 636310
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 2.012321114257289e-06,
+      "loss": 0.6932,
+      "step": 636320
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 2.0111731843575415e-06,
+      "loss": 0.6754,
+      "step": 636330
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 2.0100252544577944e-06,
+      "loss": 0.6211,
+      "step": 636340
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 2.0088773245580468e-06,
+      "loss": 0.6624,
+      "step": 636350
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 2.0077293946582996e-06,
+      "loss": 0.7314,
+      "step": 636360
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 2.006581464758552e-06,
+      "loss": 0.6887,
+      "step": 636370
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 2.0054335348588044e-06,
+      "loss": 0.7152,
+      "step": 636380
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 2.004285604959057e-06,
+      "loss": 0.7572,
+      "step": 636390
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 2.0031376750593096e-06,
+      "loss": 0.6568,
+      "step": 636400
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 2.001989745159562e-06,
+      "loss": 0.731,
+      "step": 636410
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 2.0008418152598144e-06,
+      "loss": 0.6788,
+      "step": 636420
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 1.999693885360067e-06,
+      "loss": 0.7058,
+      "step": 636430
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 1.9985459554603196e-06,
+      "loss": 0.6648,
+      "step": 636440
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 1.9973980255605724e-06,
+      "loss": 0.6226,
+      "step": 636450
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 1.996250095660825e-06,
+      "loss": 0.7613,
+      "step": 636460
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 1.9951021657610776e-06,
+      "loss": 0.6368,
+      "step": 636470
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 1.99395423586133e-06,
+      "loss": 0.6505,
+      "step": 636480
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 1.9928063059615824e-06,
+      "loss": 0.6114,
+      "step": 636490
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 1.991658376061835e-06,
+      "loss": 0.6824,
+      "step": 636500
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 1.9905104461620876e-06,
+      "loss": 0.6493,
+      "step": 636510
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 1.98936251626234e-06,
+      "loss": 0.7242,
+      "step": 636520
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 1.9882145863625924e-06,
+      "loss": 0.7459,
+      "step": 636530
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 1.9870666564628452e-06,
+      "loss": 0.6908,
+      "step": 636540
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 1.9859187265630976e-06,
+      "loss": 0.7131,
+      "step": 636550
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 1.9847707966633504e-06,
+      "loss": 0.6485,
+      "step": 636560
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 1.983622866763603e-06,
+      "loss": 0.7479,
+      "step": 636570
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 1.9824749368638557e-06,
+      "loss": 0.6071,
+      "step": 636580
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 1.981327006964108e-06,
+      "loss": 0.7006,
+      "step": 636590
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 1.9801790770643604e-06,
+      "loss": 0.6968,
+      "step": 636600
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 1.979031147164613e-06,
+      "loss": 0.7173,
+      "step": 636610
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 1.9778832172648657e-06,
+      "loss": 0.6543,
+      "step": 636620
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 1.976735287365118e-06,
+      "loss": 0.6285,
+      "step": 636630
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 1.9755873574653705e-06,
+      "loss": 0.5978,
+      "step": 636640
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 1.9744394275656233e-06,
+      "loss": 0.6598,
+      "step": 636650
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 1.9732914976658757e-06,
+      "loss": 0.6419,
+      "step": 636660
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 1.9721435677661285e-06,
+      "loss": 0.6573,
+      "step": 636670
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 1.970995637866381e-06,
+      "loss": 0.6596,
+      "step": 636680
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 1.9698477079666333e-06,
+      "loss": 0.6631,
+      "step": 636690
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 1.968699778066886e-06,
+      "loss": 0.682,
+      "step": 636700
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 1.9675518481671385e-06,
+      "loss": 0.65,
+      "step": 636710
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 1.966403918267391e-06,
+      "loss": 0.6927,
+      "step": 636720
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 1.9652559883676433e-06,
+      "loss": 0.7496,
+      "step": 636730
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 1.964108058467896e-06,
+      "loss": 0.7376,
+      "step": 636740
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 1.9629601285681485e-06,
+      "loss": 0.6657,
+      "step": 636750
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 1.9618121986684013e-06,
+      "loss": 0.7129,
+      "step": 636760
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 1.9606642687686537e-06,
+      "loss": 0.7424,
+      "step": 636770
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 1.9595163388689065e-06,
+      "loss": 0.7151,
+      "step": 636780
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 1.958368408969159e-06,
+      "loss": 0.6392,
+      "step": 636790
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 1.9572204790694113e-06,
+      "loss": 0.6465,
+      "step": 636800
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 1.9560725491696637e-06,
+      "loss": 0.6498,
+      "step": 636810
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 1.9549246192699165e-06,
+      "loss": 0.6533,
+      "step": 636820
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 1.953776689370169e-06,
+      "loss": 0.6828,
+      "step": 636830
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 1.9526287594704213e-06,
+      "loss": 0.6468,
+      "step": 636840
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 1.9514808295706737e-06,
+      "loss": 0.6814,
+      "step": 636850
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 1.9503328996709265e-06,
+      "loss": 0.6476,
+      "step": 636860
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 1.949184969771179e-06,
+      "loss": 0.7454,
+      "step": 636870
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 1.9480370398714318e-06,
+      "loss": 0.7267,
+      "step": 636880
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 1.946889109971684e-06,
+      "loss": 0.6692,
+      "step": 636890
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 1.945741180071937e-06,
+      "loss": 0.6756,
+      "step": 636900
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 1.9445932501721894e-06,
+      "loss": 0.6699,
+      "step": 636910
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 1.9434453202724418e-06,
+      "loss": 0.6836,
+      "step": 636920
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 1.942297390372694e-06,
+      "loss": 0.7148,
+      "step": 636930
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 1.941149460472947e-06,
+      "loss": 0.639,
+      "step": 636940
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 1.9400015305731994e-06,
+      "loss": 0.6488,
+      "step": 636950
+    },
+    {
+      "epoch": 29.22,
+      "learning_rate": 1.9388536006734518e-06,
+      "loss": 0.7152,
+      "step": 636960
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 1.9377056707737046e-06,
+      "loss": 0.713,
+      "step": 636970
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 1.936557740873957e-06,
+      "loss": 0.7335,
+      "step": 636980
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 1.93540981097421e-06,
+      "loss": 0.7471,
+      "step": 636990
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 1.934261881074462e-06,
+      "loss": 0.5992,
+      "step": 637000
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 1.933113951174715e-06,
+      "loss": 0.6842,
+      "step": 637010
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 1.9319660212749674e-06,
+      "loss": 0.6716,
+      "step": 637020
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 1.93081809137522e-06,
+      "loss": 0.6477,
+      "step": 637030
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 1.929670161475472e-06,
+      "loss": 0.5966,
+      "step": 637040
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 1.928522231575725e-06,
+      "loss": 0.7989,
+      "step": 637050
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 1.9273743016759774e-06,
+      "loss": 0.6566,
+      "step": 637060
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 1.92622637177623e-06,
+      "loss": 0.6704,
+      "step": 637070
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 1.9250784418764826e-06,
+      "loss": 0.6435,
+      "step": 637080
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 1.923930511976735e-06,
+      "loss": 0.6765,
+      "step": 637090
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 1.922782582076988e-06,
+      "loss": 0.6201,
+      "step": 637100
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 1.9216346521772402e-06,
+      "loss": 0.6478,
+      "step": 637110
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 1.9204867222774926e-06,
+      "loss": 0.6602,
+      "step": 637120
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 1.9193387923777455e-06,
+      "loss": 0.7017,
+      "step": 637130
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 1.918190862477998e-06,
+      "loss": 0.6554,
+      "step": 637140
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 1.9170429325782502e-06,
+      "loss": 0.6574,
+      "step": 637150
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 1.9158950026785026e-06,
+      "loss": 0.6752,
+      "step": 637160
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 1.9147470727787555e-06,
+      "loss": 0.6919,
+      "step": 637170
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 1.913599142879008e-06,
+      "loss": 0.7245,
+      "step": 637180
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 1.9124512129792607e-06,
+      "loss": 0.7293,
+      "step": 637190
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 1.911303283079513e-06,
+      "loss": 0.6734,
+      "step": 637200
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 1.910155353179766e-06,
+      "loss": 0.6651,
+      "step": 637210
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 1.9090074232800183e-06,
+      "loss": 0.6143,
+      "step": 637220
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 1.9078594933802707e-06,
+      "loss": 0.7574,
+      "step": 637230
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 1.9067115634805233e-06,
+      "loss": 0.7389,
+      "step": 637240
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 1.9055636335807759e-06,
+      "loss": 0.6177,
+      "step": 637250
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 1.9044157036810285e-06,
+      "loss": 0.639,
+      "step": 637260
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 1.9032677737812809e-06,
+      "loss": 0.6958,
+      "step": 637270
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 1.9021198438815333e-06,
+      "loss": 0.6872,
+      "step": 637280
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 1.9009719139817861e-06,
+      "loss": 0.7294,
+      "step": 637290
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 1.8998239840820385e-06,
+      "loss": 0.6514,
+      "step": 637300
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 1.8986760541822911e-06,
+      "loss": 0.7458,
+      "step": 637310
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 1.8975281242825435e-06,
+      "loss": 0.6126,
+      "step": 637320
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 1.8963801943827963e-06,
+      "loss": 0.6386,
+      "step": 637330
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 1.8952322644830487e-06,
+      "loss": 0.7325,
+      "step": 637340
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 1.8940843345833013e-06,
+      "loss": 0.7816,
+      "step": 637350
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 1.8929364046835537e-06,
+      "loss": 0.678,
+      "step": 637360
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 1.8917884747838065e-06,
+      "loss": 0.7204,
+      "step": 637370
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 1.890640544884059e-06,
+      "loss": 0.6407,
+      "step": 637380
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 1.8894926149843113e-06,
+      "loss": 0.6655,
+      "step": 637390
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 1.8883446850845642e-06,
+      "loss": 0.6859,
+      "step": 637400
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 1.8871967551848165e-06,
+      "loss": 0.7373,
+      "step": 637410
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 1.8860488252850692e-06,
+      "loss": 0.7133,
+      "step": 637420
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 1.8849008953853215e-06,
+      "loss": 0.6878,
+      "step": 637430
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 1.8837529654855744e-06,
+      "loss": 0.6478,
+      "step": 637440
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 1.8826050355858268e-06,
+      "loss": 0.7031,
+      "step": 637450
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 1.8814571056860792e-06,
+      "loss": 0.6789,
+      "step": 637460
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 1.8803091757863318e-06,
+      "loss": 0.7402,
+      "step": 637470
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 1.8791612458865844e-06,
+      "loss": 0.6286,
+      "step": 637480
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 1.878013315986837e-06,
+      "loss": 0.6658,
+      "step": 637490
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 1.8768653860870894e-06,
+      "loss": 0.6661,
+      "step": 637500
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 1.875717456187342e-06,
+      "loss": 0.6542,
+      "step": 637510
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 1.8745695262875946e-06,
+      "loss": 0.7193,
+      "step": 637520
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 1.8734215963878472e-06,
+      "loss": 0.7048,
+      "step": 637530
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 1.8722736664880996e-06,
+      "loss": 0.721,
+      "step": 637540
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 1.871125736588352e-06,
+      "loss": 0.6794,
+      "step": 637550
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 1.8699778066886048e-06,
+      "loss": 0.6819,
+      "step": 637560
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 1.8688298767888572e-06,
+      "loss": 0.6691,
+      "step": 637570
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 1.8676819468891098e-06,
+      "loss": 0.5886,
+      "step": 637580
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 1.8665340169893622e-06,
+      "loss": 0.6654,
+      "step": 637590
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 1.865386087089615e-06,
+      "loss": 0.6751,
+      "step": 637600
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 1.8642381571898674e-06,
+      "loss": 0.6623,
+      "step": 637610
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 1.86309022729012e-06,
+      "loss": 0.6752,
+      "step": 637620
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 1.8619422973903724e-06,
+      "loss": 0.7029,
+      "step": 637630
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 1.8607943674906252e-06,
+      "loss": 0.6385,
+      "step": 637640
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 1.8596464375908776e-06,
+      "loss": 0.6978,
+      "step": 637650
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 1.85849850769113e-06,
+      "loss": 0.6338,
+      "step": 637660
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 1.8573505777913826e-06,
+      "loss": 0.6729,
+      "step": 637670
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 1.8562026478916352e-06,
+      "loss": 0.6382,
+      "step": 637680
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 1.8550547179918879e-06,
+      "loss": 0.6182,
+      "step": 637690
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 1.8539067880921402e-06,
+      "loss": 0.6302,
+      "step": 637700
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 1.8527588581923929e-06,
+      "loss": 0.7433,
+      "step": 637710
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 1.8516109282926455e-06,
+      "loss": 0.672,
+      "step": 637720
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 1.850462998392898e-06,
+      "loss": 0.6786,
+      "step": 637730
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 1.8493150684931505e-06,
+      "loss": 0.6818,
+      "step": 637740
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 1.8481671385934029e-06,
+      "loss": 0.7107,
+      "step": 637750
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 1.8470192086936557e-06,
+      "loss": 0.6656,
+      "step": 637760
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 1.845871278793908e-06,
+      "loss": 0.6946,
+      "step": 637770
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 1.8447233488941607e-06,
+      "loss": 0.728,
+      "step": 637780
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 1.843575418994413e-06,
+      "loss": 0.7521,
+      "step": 637790
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 1.842427489094666e-06,
+      "loss": 0.6668,
+      "step": 637800
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 1.8412795591949183e-06,
+      "loss": 0.6876,
+      "step": 637810
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 1.840131629295171e-06,
+      "loss": 0.6695,
+      "step": 637820
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 1.8389836993954233e-06,
+      "loss": 0.6517,
+      "step": 637830
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 1.8378357694956761e-06,
+      "loss": 0.6679,
+      "step": 637840
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 1.8366878395959285e-06,
+      "loss": 0.6488,
+      "step": 637850
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 1.835539909696181e-06,
+      "loss": 0.6777,
+      "step": 637860
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 1.8343919797964337e-06,
+      "loss": 0.7148,
+      "step": 637870
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 1.8332440498966861e-06,
+      "loss": 0.6565,
+      "step": 637880
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 1.8320961199969387e-06,
+      "loss": 0.6753,
+      "step": 637890
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 1.8309481900971911e-06,
+      "loss": 0.6725,
+      "step": 637900
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 1.829800260197444e-06,
+      "loss": 0.6787,
+      "step": 637910
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 1.8286523302976963e-06,
+      "loss": 0.724,
+      "step": 637920
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 1.827504400397949e-06,
+      "loss": 0.7052,
+      "step": 637930
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 1.8263564704982013e-06,
+      "loss": 0.6713,
+      "step": 637940
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 1.8252085405984542e-06,
+      "loss": 0.6529,
+      "step": 637950
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 1.8240606106987066e-06,
+      "loss": 0.7073,
+      "step": 637960
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 1.822912680798959e-06,
+      "loss": 0.7135,
+      "step": 637970
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 1.8217647508992116e-06,
+      "loss": 0.7012,
+      "step": 637980
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 1.8206168209994642e-06,
+      "loss": 0.6019,
+      "step": 637990
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 1.8194688910997168e-06,
+      "loss": 0.6822,
+      "step": 638000
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 1.8183209611999692e-06,
+      "loss": 0.6863,
+      "step": 638010
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 1.8171730313002218e-06,
+      "loss": 0.7132,
+      "step": 638020
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 1.8160251014004744e-06,
+      "loss": 0.6395,
+      "step": 638030
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 1.814877171500727e-06,
+      "loss": 0.6933,
+      "step": 638040
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 1.8137292416009794e-06,
+      "loss": 0.7268,
+      "step": 638050
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 1.8125813117012318e-06,
+      "loss": 0.6462,
+      "step": 638060
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 1.8114333818014846e-06,
+      "loss": 0.61,
+      "step": 638070
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 1.810285451901737e-06,
+      "loss": 0.7365,
+      "step": 638080
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 1.8091375220019896e-06,
+      "loss": 0.6704,
+      "step": 638090
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 1.807989592102242e-06,
+      "loss": 0.7699,
+      "step": 638100
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 1.8068416622024948e-06,
+      "loss": 0.7827,
+      "step": 638110
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 1.8056937323027472e-06,
+      "loss": 0.7692,
+      "step": 638120
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 1.8045458024029998e-06,
+      "loss": 0.6951,
+      "step": 638130
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 1.8033978725032522e-06,
+      "loss": 0.5985,
+      "step": 638140
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 1.802249942603505e-06,
+      "loss": 0.7934,
+      "step": 638150
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 1.8011020127037574e-06,
+      "loss": 0.7903,
+      "step": 638160
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 1.7999540828040098e-06,
+      "loss": 0.6491,
+      "step": 638170
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 1.7988061529042624e-06,
+      "loss": 0.6645,
+      "step": 638180
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 1.797658223004515e-06,
+      "loss": 0.614,
+      "step": 638190
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 1.7965102931047676e-06,
+      "loss": 0.7077,
+      "step": 638200
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 1.79536236320502e-06,
+      "loss": 0.6739,
+      "step": 638210
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 1.7942144333052726e-06,
+      "loss": 0.6676,
+      "step": 638220
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 1.7930665034055253e-06,
+      "loss": 0.622,
+      "step": 638230
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 1.7919185735057779e-06,
+      "loss": 0.6101,
+      "step": 638240
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 1.7907706436060303e-06,
+      "loss": 0.6048,
+      "step": 638250
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 1.7896227137062826e-06,
+      "loss": 0.6474,
+      "step": 638260
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 1.7884747838065355e-06,
+      "loss": 0.7523,
+      "step": 638270
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 1.7873268539067879e-06,
+      "loss": 0.6686,
+      "step": 638280
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 1.7861789240070405e-06,
+      "loss": 0.6656,
+      "step": 638290
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 1.785030994107293e-06,
+      "loss": 0.7161,
+      "step": 638300
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 1.7838830642075457e-06,
+      "loss": 0.7407,
+      "step": 638310
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 1.782735134307798e-06,
+      "loss": 0.605,
+      "step": 638320
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 1.7815872044080507e-06,
+      "loss": 0.6918,
+      "step": 638330
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 1.7804392745083033e-06,
+      "loss": 0.6469,
+      "step": 638340
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 1.779291344608556e-06,
+      "loss": 0.64,
+      "step": 638350
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 1.7781434147088083e-06,
+      "loss": 0.7145,
+      "step": 638360
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 1.7769954848090607e-06,
+      "loss": 0.6343,
+      "step": 638370
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 1.7758475549093135e-06,
+      "loss": 0.641,
+      "step": 638380
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 1.774699625009566e-06,
+      "loss": 0.7024,
+      "step": 638390
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 1.7735516951098185e-06,
+      "loss": 0.7225,
+      "step": 638400
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 1.772403765210071e-06,
+      "loss": 0.6959,
+      "step": 638410
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 1.7712558353103237e-06,
+      "loss": 0.6895,
+      "step": 638420
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 1.7701079054105761e-06,
+      "loss": 0.5873,
+      "step": 638430
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 1.7689599755108287e-06,
+      "loss": 0.6845,
+      "step": 638440
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 1.7678120456110811e-06,
+      "loss": 0.7187,
+      "step": 638450
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 1.766664115711334e-06,
+      "loss": 0.6924,
+      "step": 638460
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 1.7655161858115863e-06,
+      "loss": 0.6567,
+      "step": 638470
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 1.7643682559118387e-06,
+      "loss": 0.6624,
+      "step": 638480
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 1.7632203260120913e-06,
+      "loss": 0.6601,
+      "step": 638490
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 1.762072396112344e-06,
+      "loss": 0.7626,
+      "step": 638500
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 1.7609244662125966e-06,
+      "loss": 0.6412,
+      "step": 638510
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 1.759776536312849e-06,
+      "loss": 0.7171,
+      "step": 638520
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 1.7586286064131013e-06,
+      "loss": 0.6577,
+      "step": 638530
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 1.7574806765133542e-06,
+      "loss": 0.6392,
+      "step": 638540
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 1.7563327466136068e-06,
+      "loss": 0.6351,
+      "step": 638550
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 1.7551848167138592e-06,
+      "loss": 0.6456,
+      "step": 638560
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 1.7540368868141116e-06,
+      "loss": 0.7156,
+      "step": 638570
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 1.7528889569143644e-06,
+      "loss": 0.6901,
+      "step": 638580
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 1.7517410270146168e-06,
+      "loss": 0.6726,
+      "step": 638590
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 1.7505930971148694e-06,
+      "loss": 0.6262,
+      "step": 638600
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 1.7494451672151218e-06,
+      "loss": 0.7397,
+      "step": 638610
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 1.7482972373153746e-06,
+      "loss": 0.6155,
+      "step": 638620
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 1.747149307415627e-06,
+      "loss": 0.6977,
+      "step": 638630
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 1.7460013775158794e-06,
+      "loss": 0.6994,
+      "step": 638640
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 1.744853447616132e-06,
+      "loss": 0.7299,
+      "step": 638650
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 1.7437055177163846e-06,
+      "loss": 0.6769,
+      "step": 638660
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 1.7425575878166372e-06,
+      "loss": 0.7137,
+      "step": 638670
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 1.7414096579168896e-06,
+      "loss": 0.6662,
+      "step": 638680
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 1.7402617280171422e-06,
+      "loss": 0.7135,
+      "step": 638690
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 1.7391137981173948e-06,
+      "loss": 0.7023,
+      "step": 638700
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 1.7379658682176474e-06,
+      "loss": 0.6705,
+      "step": 638710
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 1.7368179383178998e-06,
+      "loss": 0.7019,
+      "step": 638720
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 1.7356700084181526e-06,
+      "loss": 0.6611,
+      "step": 638730
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 1.734522078518405e-06,
+      "loss": 0.6976,
+      "step": 638740
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 1.7333741486186574e-06,
+      "loss": 0.6856,
+      "step": 638750
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 1.73222621871891e-06,
+      "loss": 0.7271,
+      "step": 638760
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 1.7310782888191626e-06,
+      "loss": 0.6737,
+      "step": 638770
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 1.7299303589194153e-06,
+      "loss": 0.63,
+      "step": 638780
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 1.7287824290196677e-06,
+      "loss": 0.6696,
+      "step": 638790
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 1.7276344991199203e-06,
+      "loss": 0.6552,
+      "step": 638800
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 1.7264865692201729e-06,
+      "loss": 0.7069,
+      "step": 638810
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 1.7253386393204255e-06,
+      "loss": 0.785,
+      "step": 638820
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 1.7241907094206779e-06,
+      "loss": 0.63,
+      "step": 638830
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 1.7230427795209303e-06,
+      "loss": 0.7282,
+      "step": 638840
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 1.721894849621183e-06,
+      "loss": 0.6693,
+      "step": 638850
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 1.7207469197214355e-06,
+      "loss": 0.651,
+      "step": 638860
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 1.719598989821688e-06,
+      "loss": 0.7871,
+      "step": 638870
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 1.7184510599219405e-06,
+      "loss": 0.7078,
+      "step": 638880
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 1.7173031300221933e-06,
+      "loss": 0.7584,
+      "step": 638890
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 1.7161552001224457e-06,
+      "loss": 0.739,
+      "step": 638900
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 1.7150072702226983e-06,
+      "loss": 0.7293,
+      "step": 638910
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 1.7138593403229507e-06,
+      "loss": 0.6634,
+      "step": 638920
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 1.7127114104232035e-06,
+      "loss": 0.6526,
+      "step": 638930
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 1.711563480523456e-06,
+      "loss": 0.7231,
+      "step": 638940
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 1.7104155506237083e-06,
+      "loss": 0.7118,
+      "step": 638950
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 1.709267620723961e-06,
+      "loss": 0.5896,
+      "step": 638960
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 1.7081196908242135e-06,
+      "loss": 0.691,
+      "step": 638970
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 1.7069717609244661e-06,
+      "loss": 0.6939,
+      "step": 638980
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 1.7058238310247185e-06,
+      "loss": 0.7792,
+      "step": 638990
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 1.7046759011249711e-06,
+      "loss": 0.6501,
+      "step": 639000
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 1.7035279712252237e-06,
+      "loss": 0.6285,
+      "step": 639010
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 1.7023800413254763e-06,
+      "loss": 0.701,
+      "step": 639020
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 1.7012321114257287e-06,
+      "loss": 0.6699,
+      "step": 639030
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 1.7000841815259811e-06,
+      "loss": 0.7181,
+      "step": 639040
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 1.698936251626234e-06,
+      "loss": 0.6433,
+      "step": 639050
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 1.6977883217264864e-06,
+      "loss": 0.6953,
+      "step": 639060
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 1.696640391826739e-06,
+      "loss": 0.7338,
+      "step": 639070
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 1.6954924619269914e-06,
+      "loss": 0.7485,
+      "step": 639080
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 1.6943445320272442e-06,
+      "loss": 0.6553,
+      "step": 639090
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 1.6931966021274966e-06,
+      "loss": 0.7329,
+      "step": 639100
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 1.6920486722277492e-06,
+      "loss": 0.6473,
+      "step": 639110
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 1.6909007423280016e-06,
+      "loss": 0.6742,
+      "step": 639120
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 1.6897528124282544e-06,
+      "loss": 0.7662,
+      "step": 639130
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 1.6886048825285068e-06,
+      "loss": 0.6819,
+      "step": 639140
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 1.6874569526287592e-06,
+      "loss": 0.6987,
+      "step": 639150
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 1.6863090227290118e-06,
+      "loss": 0.6768,
+      "step": 639160
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 1.6851610928292644e-06,
+      "loss": 0.6286,
+      "step": 639170
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 1.684013162929517e-06,
+      "loss": 0.6672,
+      "step": 639180
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 1.6828652330297694e-06,
+      "loss": 0.6525,
+      "step": 639190
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 1.6817173031300222e-06,
+      "loss": 0.6114,
+      "step": 639200
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 1.6805693732302746e-06,
+      "loss": 0.662,
+      "step": 639210
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 1.6794214433305272e-06,
+      "loss": 0.7379,
+      "step": 639220
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 1.6782735134307796e-06,
+      "loss": 0.6628,
+      "step": 639230
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 1.6771255835310324e-06,
+      "loss": 0.7547,
+      "step": 639240
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 1.6759776536312848e-06,
+      "loss": 0.6824,
+      "step": 639250
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 1.6748297237315372e-06,
+      "loss": 0.7047,
+      "step": 639260
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 1.6736817938317898e-06,
+      "loss": 0.6725,
+      "step": 639270
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 1.6725338639320424e-06,
+      "loss": 0.608,
+      "step": 639280
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 1.671385934032295e-06,
+      "loss": 0.6808,
+      "step": 639290
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 1.6702380041325474e-06,
+      "loss": 0.7103,
+      "step": 639300
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 1.6690900742328e-06,
+      "loss": 0.7293,
+      "step": 639310
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 1.6679421443330527e-06,
+      "loss": 0.7079,
+      "step": 639320
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 1.6667942144333053e-06,
+      "loss": 0.6926,
+      "step": 639330
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 1.6656462845335577e-06,
+      "loss": 0.6598,
+      "step": 639340
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 1.66449835463381e-06,
+      "loss": 0.6832,
+      "step": 639350
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 1.6633504247340629e-06,
+      "loss": 0.6876,
+      "step": 639360
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 1.6622024948343153e-06,
+      "loss": 0.6446,
+      "step": 639370
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 1.6610545649345679e-06,
+      "loss": 0.7178,
+      "step": 639380
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 1.6599066350348203e-06,
+      "loss": 0.6904,
+      "step": 639390
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 1.658758705135073e-06,
+      "loss": 0.7091,
+      "step": 639400
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 1.6576107752353255e-06,
+      "loss": 0.6238,
+      "step": 639410
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 1.656462845335578e-06,
+      "loss": 0.6639,
+      "step": 639420
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 1.6553149154358305e-06,
+      "loss": 0.6424,
+      "step": 639430
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 1.6541669855360833e-06,
+      "loss": 0.6932,
+      "step": 639440
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 1.6530190556363357e-06,
+      "loss": 0.726,
+      "step": 639450
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 1.651871125736588e-06,
+      "loss": 0.7232,
+      "step": 639460
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 1.6507231958368407e-06,
+      "loss": 0.6927,
+      "step": 639470
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 1.6495752659370933e-06,
+      "loss": 0.7086,
+      "step": 639480
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 1.648427336037346e-06,
+      "loss": 0.6484,
+      "step": 639490
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 1.6472794061375983e-06,
+      "loss": 0.6813,
+      "step": 639500
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 1.646131476237851e-06,
+      "loss": 0.6669,
+      "step": 639510
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 1.6449835463381035e-06,
+      "loss": 0.6054,
+      "step": 639520
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 1.6438356164383561e-06,
+      "loss": 0.7031,
+      "step": 639530
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 1.6426876865386085e-06,
+      "loss": 0.7252,
+      "step": 639540
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 1.641539756638861e-06,
+      "loss": 0.6563,
+      "step": 639550
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 1.6403918267391137e-06,
+      "loss": 0.6932,
+      "step": 639560
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 1.6392438968393661e-06,
+      "loss": 0.6773,
+      "step": 639570
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 1.6380959669396187e-06,
+      "loss": 0.6932,
+      "step": 639580
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 1.6369480370398711e-06,
+      "loss": 0.6832,
+      "step": 639590
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 1.635800107140124e-06,
+      "loss": 0.6411,
+      "step": 639600
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 1.6346521772403764e-06,
+      "loss": 0.6486,
+      "step": 639610
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 1.633504247340629e-06,
+      "loss": 0.6994,
+      "step": 639620
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 1.6323563174408816e-06,
+      "loss": 0.6846,
+      "step": 639630
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 1.6312083875411342e-06,
+      "loss": 0.6865,
+      "step": 639640
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 1.6300604576413866e-06,
+      "loss": 0.7041,
+      "step": 639650
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 1.628912527741639e-06,
+      "loss": 0.6849,
+      "step": 639660
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 1.6277645978418918e-06,
+      "loss": 0.6552,
+      "step": 639670
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 1.6266166679421442e-06,
+      "loss": 0.6524,
+      "step": 639680
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 1.6254687380423968e-06,
+      "loss": 0.7119,
+      "step": 639690
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 1.6243208081426492e-06,
+      "loss": 0.7286,
+      "step": 639700
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 1.623172878242902e-06,
+      "loss": 0.605,
+      "step": 639710
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 1.6220249483431544e-06,
+      "loss": 0.6773,
+      "step": 639720
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 1.6208770184434068e-06,
+      "loss": 0.663,
+      "step": 639730
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 1.6197290885436594e-06,
+      "loss": 0.659,
+      "step": 639740
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 1.618581158643912e-06,
+      "loss": 0.6798,
+      "step": 639750
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 1.6174332287441646e-06,
+      "loss": 0.6924,
+      "step": 639760
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 1.616285298844417e-06,
+      "loss": 0.6858,
+      "step": 639770
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 1.6151373689446696e-06,
+      "loss": 0.6479,
+      "step": 639780
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 1.6139894390449222e-06,
+      "loss": 0.7102,
+      "step": 639790
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 1.6128415091451748e-06,
+      "loss": 0.6569,
+      "step": 639800
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 1.6116935792454272e-06,
+      "loss": 0.7575,
+      "step": 639810
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 1.6105456493456796e-06,
+      "loss": 0.6515,
+      "step": 639820
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 1.6093977194459324e-06,
+      "loss": 0.7064,
+      "step": 639830
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 1.6082497895461848e-06,
+      "loss": 0.6139,
+      "step": 639840
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 1.6071018596464374e-06,
+      "loss": 0.6716,
+      "step": 639850
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 1.6059539297466898e-06,
+      "loss": 0.69,
+      "step": 639860
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 1.6048059998469427e-06,
+      "loss": 0.6412,
+      "step": 639870
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 1.603658069947195e-06,
+      "loss": 0.6883,
+      "step": 639880
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 1.6025101400474477e-06,
+      "loss": 0.654,
+      "step": 639890
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 1.6013622101477e-06,
+      "loss": 0.7288,
+      "step": 639900
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 1.6002142802479529e-06,
+      "loss": 0.6515,
+      "step": 639910
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 1.5990663503482053e-06,
+      "loss": 0.6812,
+      "step": 639920
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 1.5979184204484577e-06,
+      "loss": 0.7584,
+      "step": 639930
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 1.5967704905487103e-06,
+      "loss": 0.6727,
+      "step": 639940
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 1.5956225606489629e-06,
+      "loss": 0.6159,
+      "step": 639950
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 1.5944746307492155e-06,
+      "loss": 0.6771,
+      "step": 639960
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 1.5933267008494679e-06,
+      "loss": 0.7128,
+      "step": 639970
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 1.5921787709497205e-06,
+      "loss": 0.7214,
+      "step": 639980
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 1.591030841049973e-06,
+      "loss": 0.6796,
+      "step": 639990
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 1.5898829111502257e-06,
+      "loss": 0.6273,
+      "step": 640000
+    },
+    {
+      "epoch": 29.36,
+      "learning_rate": 1.588734981250478e-06,
+      "loss": 0.7379,
+      "step": 640010
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 1.5875870513507305e-06,
+      "loss": 0.7271,
+      "step": 640020
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 1.5864391214509833e-06,
+      "loss": 0.6996,
+      "step": 640030
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 1.5852911915512357e-06,
+      "loss": 0.6973,
+      "step": 640040
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 1.5841432616514883e-06,
+      "loss": 0.6623,
+      "step": 640050
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 1.5829953317517407e-06,
+      "loss": 0.6853,
+      "step": 640060
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 1.5818474018519935e-06,
+      "loss": 0.6468,
+      "step": 640070
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 1.580699471952246e-06,
+      "loss": 0.6712,
+      "step": 640080
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 1.5795515420524985e-06,
+      "loss": 0.6517,
+      "step": 640090
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 1.5784036121527511e-06,
+      "loss": 0.7153,
+      "step": 640100
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 1.5772556822530037e-06,
+      "loss": 0.6515,
+      "step": 640110
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 1.5761077523532561e-06,
+      "loss": 0.6361,
+      "step": 640120
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 1.5749598224535085e-06,
+      "loss": 0.7175,
+      "step": 640130
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 1.5738118925537614e-06,
+      "loss": 0.6751,
+      "step": 640140
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 1.5726639626540138e-06,
+      "loss": 0.6941,
+      "step": 640150
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 1.5715160327542664e-06,
+      "loss": 0.7124,
+      "step": 640160
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 1.5703681028545188e-06,
+      "loss": 0.6764,
+      "step": 640170
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 1.5692201729547716e-06,
+      "loss": 0.6649,
+      "step": 640180
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 1.568072243055024e-06,
+      "loss": 0.6916,
+      "step": 640190
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 1.5669243131552766e-06,
+      "loss": 0.7079,
+      "step": 640200
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 1.565776383255529e-06,
+      "loss": 0.6242,
+      "step": 640210
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 1.5646284533557818e-06,
+      "loss": 0.6995,
+      "step": 640220
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 1.5634805234560342e-06,
+      "loss": 0.6993,
+      "step": 640230
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 1.5623325935562866e-06,
+      "loss": 0.6536,
+      "step": 640240
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 1.5611846636565392e-06,
+      "loss": 0.6754,
+      "step": 640250
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 1.5600367337567918e-06,
+      "loss": 0.6577,
+      "step": 640260
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 1.5588888038570444e-06,
+      "loss": 0.6991,
+      "step": 640270
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 1.5577408739572968e-06,
+      "loss": 0.7764,
+      "step": 640280
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 1.5565929440575494e-06,
+      "loss": 0.646,
+      "step": 640290
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 1.555445014157802e-06,
+      "loss": 0.6943,
+      "step": 640300
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 1.5542970842580546e-06,
+      "loss": 0.6864,
+      "step": 640310
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 1.553149154358307e-06,
+      "loss": 0.7284,
+      "step": 640320
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 1.5520012244585594e-06,
+      "loss": 0.6553,
+      "step": 640330
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 1.5508532945588122e-06,
+      "loss": 0.7177,
+      "step": 640340
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 1.5497053646590646e-06,
+      "loss": 0.7029,
+      "step": 640350
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 1.5485574347593172e-06,
+      "loss": 0.7122,
+      "step": 640360
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 1.5474095048595696e-06,
+      "loss": 0.6777,
+      "step": 640370
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 1.5462615749598224e-06,
+      "loss": 0.6387,
+      "step": 640380
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 1.5451136450600748e-06,
+      "loss": 0.6909,
+      "step": 640390
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 1.5439657151603274e-06,
+      "loss": 0.6417,
+      "step": 640400
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 1.5428177852605798e-06,
+      "loss": 0.703,
+      "step": 640410
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 1.5416698553608327e-06,
+      "loss": 0.7188,
+      "step": 640420
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 1.540521925461085e-06,
+      "loss": 0.6222,
+      "step": 640430
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 1.5393739955613375e-06,
+      "loss": 0.6537,
+      "step": 640440
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 1.53822606566159e-06,
+      "loss": 0.7128,
+      "step": 640450
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 1.5370781357618427e-06,
+      "loss": 0.6178,
+      "step": 640460
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 1.5359302058620953e-06,
+      "loss": 0.71,
+      "step": 640470
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 1.5347822759623477e-06,
+      "loss": 0.7447,
+      "step": 640480
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 1.5336343460626003e-06,
+      "loss": 0.7298,
+      "step": 640490
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 1.5324864161628529e-06,
+      "loss": 0.6569,
+      "step": 640500
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 1.5313384862631055e-06,
+      "loss": 0.6721,
+      "step": 640510
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 1.5301905563633579e-06,
+      "loss": 0.6907,
+      "step": 640520
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 1.5290426264636107e-06,
+      "loss": 0.6184,
+      "step": 640530
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 1.527894696563863e-06,
+      "loss": 0.7145,
+      "step": 640540
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 1.5267467666641155e-06,
+      "loss": 0.6758,
+      "step": 640550
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 1.525598836764368e-06,
+      "loss": 0.6586,
+      "step": 640560
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 1.5244509068646207e-06,
+      "loss": 0.6907,
+      "step": 640570
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 1.5233029769648733e-06,
+      "loss": 0.6948,
+      "step": 640580
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 1.5221550470651257e-06,
+      "loss": 0.7318,
+      "step": 640590
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 1.5210071171653783e-06,
+      "loss": 0.6575,
+      "step": 640600
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 1.519859187265631e-06,
+      "loss": 0.6348,
+      "step": 640610
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 1.5187112573658835e-06,
+      "loss": 0.7074,
+      "step": 640620
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 1.517563327466136e-06,
+      "loss": 0.6274,
+      "step": 640630
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 1.5164153975663883e-06,
+      "loss": 0.7126,
+      "step": 640640
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 1.5152674676666411e-06,
+      "loss": 0.6058,
+      "step": 640650
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 1.5141195377668935e-06,
+      "loss": 0.6773,
+      "step": 640660
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 1.5129716078671461e-06,
+      "loss": 0.658,
+      "step": 640670
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 1.5118236779673985e-06,
+      "loss": 0.6703,
+      "step": 640680
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 1.5106757480676514e-06,
+      "loss": 0.6622,
+      "step": 640690
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 1.5095278181679038e-06,
+      "loss": 0.6467,
+      "step": 640700
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 1.5083798882681564e-06,
+      "loss": 0.7198,
+      "step": 640710
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 1.5072319583684088e-06,
+      "loss": 0.6749,
+      "step": 640720
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 1.5060840284686616e-06,
+      "loss": 0.6949,
+      "step": 640730
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 1.504936098568914e-06,
+      "loss": 0.6009,
+      "step": 640740
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 1.5037881686691664e-06,
+      "loss": 0.6801,
+      "step": 640750
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 1.502640238769419e-06,
+      "loss": 0.7024,
+      "step": 640760
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 1.5014923088696716e-06,
+      "loss": 0.8326,
+      "step": 640770
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 1.5003443789699242e-06,
+      "loss": 0.6609,
+      "step": 640780
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 1.4991964490701766e-06,
+      "loss": 0.6116,
+      "step": 640790
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 1.498048519170429e-06,
+      "loss": 0.6434,
+      "step": 640800
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 1.4969005892706818e-06,
+      "loss": 0.7268,
+      "step": 640810
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 1.4957526593709342e-06,
+      "loss": 0.6602,
+      "step": 640820
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 1.4946047294711868e-06,
+      "loss": 0.6196,
+      "step": 640830
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 1.4934567995714392e-06,
+      "loss": 0.7031,
+      "step": 640840
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 1.492308869671692e-06,
+      "loss": 0.6846,
+      "step": 640850
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 1.4911609397719444e-06,
+      "loss": 0.6884,
+      "step": 640860
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 1.490013009872197e-06,
+      "loss": 0.6931,
+      "step": 640870
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 1.4888650799724494e-06,
+      "loss": 0.6241,
+      "step": 640880
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 1.4877171500727022e-06,
+      "loss": 0.6263,
+      "step": 640890
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 1.4865692201729546e-06,
+      "loss": 0.6588,
+      "step": 640900
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 1.485421290273207e-06,
+      "loss": 0.6682,
+      "step": 640910
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 1.4842733603734596e-06,
+      "loss": 0.6573,
+      "step": 640920
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 1.4831254304737122e-06,
+      "loss": 0.6143,
+      "step": 640930
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 1.4819775005739648e-06,
+      "loss": 0.6136,
+      "step": 640940
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 1.4808295706742172e-06,
+      "loss": 0.6388,
+      "step": 640950
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 1.47968164077447e-06,
+      "loss": 0.6774,
+      "step": 640960
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 1.4785337108747225e-06,
+      "loss": 0.6554,
+      "step": 640970
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 1.477385780974975e-06,
+      "loss": 0.7489,
+      "step": 640980
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 1.4762378510752275e-06,
+      "loss": 0.7249,
+      "step": 640990
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 1.4750899211754803e-06,
+      "loss": 0.6709,
+      "step": 641000
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 1.4739419912757327e-06,
+      "loss": 0.6345,
+      "step": 641010
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 1.472794061375985e-06,
+      "loss": 0.6952,
+      "step": 641020
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 1.4716461314762377e-06,
+      "loss": 0.7125,
+      "step": 641030
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 1.4704982015764903e-06,
+      "loss": 0.7462,
+      "step": 641040
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 1.4693502716767429e-06,
+      "loss": 0.7317,
+      "step": 641050
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 1.4682023417769953e-06,
+      "loss": 0.628,
+      "step": 641060
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 1.4670544118772479e-06,
+      "loss": 0.7217,
+      "step": 641070
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 1.4659064819775005e-06,
+      "loss": 0.6885,
+      "step": 641080
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 1.4647585520777531e-06,
+      "loss": 0.6383,
+      "step": 641090
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 1.4636106221780055e-06,
+      "loss": 0.6407,
+      "step": 641100
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 1.462462692278258e-06,
+      "loss": 0.6567,
+      "step": 641110
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 1.4613147623785107e-06,
+      "loss": 0.7436,
+      "step": 641120
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 1.4601668324787631e-06,
+      "loss": 0.6553,
+      "step": 641130
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 1.4590189025790157e-06,
+      "loss": 0.7523,
+      "step": 641140
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 1.4578709726792681e-06,
+      "loss": 0.6683,
+      "step": 641150
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 1.456723042779521e-06,
+      "loss": 0.6297,
+      "step": 641160
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 1.4555751128797733e-06,
+      "loss": 0.6539,
+      "step": 641170
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 1.454427182980026e-06,
+      "loss": 0.6602,
+      "step": 641180
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 1.4532792530802783e-06,
+      "loss": 0.7058,
+      "step": 641190
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 1.4521313231805312e-06,
+      "loss": 0.7002,
+      "step": 641200
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 1.4509833932807835e-06,
+      "loss": 0.6945,
+      "step": 641210
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 1.449835463381036e-06,
+      "loss": 0.6497,
+      "step": 641220
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 1.4486875334812885e-06,
+      "loss": 0.717,
+      "step": 641230
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 1.4475396035815412e-06,
+      "loss": 0.7146,
+      "step": 641240
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 1.4463916736817938e-06,
+      "loss": 0.6637,
+      "step": 641250
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 1.4452437437820462e-06,
+      "loss": 0.6651,
+      "step": 641260
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 1.4440958138822988e-06,
+      "loss": 0.6607,
+      "step": 641270
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 1.4429478839825514e-06,
+      "loss": 0.6426,
+      "step": 641280
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 1.441799954082804e-06,
+      "loss": 0.7048,
+      "step": 641290
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 1.4406520241830564e-06,
+      "loss": 0.7455,
+      "step": 641300
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 1.4395040942833088e-06,
+      "loss": 0.6912,
+      "step": 641310
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 1.4383561643835616e-06,
+      "loss": 0.6782,
+      "step": 641320
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 1.437208234483814e-06,
+      "loss": 0.675,
+      "step": 641330
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 1.4360603045840666e-06,
+      "loss": 0.6706,
+      "step": 641340
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 1.434912374684319e-06,
+      "loss": 0.5955,
+      "step": 641350
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 1.4337644447845718e-06,
+      "loss": 0.6748,
+      "step": 641360
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 1.4326165148848242e-06,
+      "loss": 0.7019,
+      "step": 641370
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 1.4314685849850768e-06,
+      "loss": 0.7164,
+      "step": 641380
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 1.4303206550853292e-06,
+      "loss": 0.7325,
+      "step": 641390
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 1.429172725185582e-06,
+      "loss": 0.6452,
+      "step": 641400
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 1.4280247952858344e-06,
+      "loss": 0.6875,
+      "step": 641410
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 1.4268768653860868e-06,
+      "loss": 0.725,
+      "step": 641420
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 1.4257289354863396e-06,
+      "loss": 0.721,
+      "step": 641430
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 1.424581005586592e-06,
+      "loss": 0.6582,
+      "step": 641440
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 1.4234330756868446e-06,
+      "loss": 0.6764,
+      "step": 641450
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 1.422285145787097e-06,
+      "loss": 0.5469,
+      "step": 641460
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 1.4211372158873499e-06,
+      "loss": 0.7101,
+      "step": 641470
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 1.4199892859876022e-06,
+      "loss": 0.6676,
+      "step": 641480
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 1.4188413560878549e-06,
+      "loss": 0.6289,
+      "step": 641490
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 1.4176934261881072e-06,
+      "loss": 0.6506,
+      "step": 641500
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 1.41654549628836e-06,
+      "loss": 0.6707,
+      "step": 641510
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 1.4153975663886125e-06,
+      "loss": 0.6439,
+      "step": 641520
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 1.4142496364888649e-06,
+      "loss": 0.6312,
+      "step": 641530
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 1.4131017065891175e-06,
+      "loss": 0.7043,
+      "step": 641540
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 1.41195377668937e-06,
+      "loss": 0.7017,
+      "step": 641550
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 1.4108058467896227e-06,
+      "loss": 0.6926,
+      "step": 641560
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 1.409657916889875e-06,
+      "loss": 0.6542,
+      "step": 641570
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 1.4085099869901277e-06,
+      "loss": 0.7158,
+      "step": 641580
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 1.4073620570903803e-06,
+      "loss": 0.6619,
+      "step": 641590
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 1.406214127190633e-06,
+      "loss": 0.6144,
+      "step": 641600
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 1.4050661972908853e-06,
+      "loss": 0.7181,
+      "step": 641610
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 1.4039182673911377e-06,
+      "loss": 0.6858,
+      "step": 641620
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 1.4027703374913905e-06,
+      "loss": 0.6586,
+      "step": 641630
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 1.401622407591643e-06,
+      "loss": 0.6616,
+      "step": 641640
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 1.4004744776918955e-06,
+      "loss": 0.6559,
+      "step": 641650
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 1.399326547792148e-06,
+      "loss": 0.702,
+      "step": 641660
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 1.3981786178924007e-06,
+      "loss": 0.6914,
+      "step": 641670
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 1.3970306879926531e-06,
+      "loss": 0.7398,
+      "step": 641680
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 1.3958827580929057e-06,
+      "loss": 0.6946,
+      "step": 641690
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 1.3947348281931581e-06,
+      "loss": 0.7213,
+      "step": 641700
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 1.393586898293411e-06,
+      "loss": 0.7084,
+      "step": 641710
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 1.3924389683936633e-06,
+      "loss": 0.6581,
+      "step": 641720
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 1.3912910384939157e-06,
+      "loss": 0.5777,
+      "step": 641730
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 1.3901431085941683e-06,
+      "loss": 0.6548,
+      "step": 641740
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 1.388995178694421e-06,
+      "loss": 0.6626,
+      "step": 641750
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 1.3878472487946736e-06,
+      "loss": 0.6923,
+      "step": 641760
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 1.386699318894926e-06,
+      "loss": 0.722,
+      "step": 641770
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 1.3855513889951786e-06,
+      "loss": 0.6444,
+      "step": 641780
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 1.3844034590954312e-06,
+      "loss": 0.6436,
+      "step": 641790
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 1.3832555291956838e-06,
+      "loss": 0.6458,
+      "step": 641800
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 1.3821075992959362e-06,
+      "loss": 0.6683,
+      "step": 641810
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 1.3809596693961886e-06,
+      "loss": 0.6613,
+      "step": 641820
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 1.3798117394964414e-06,
+      "loss": 0.6816,
+      "step": 641830
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 1.3786638095966938e-06,
+      "loss": 0.7204,
+      "step": 641840
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 1.3775158796969464e-06,
+      "loss": 0.6574,
+      "step": 641850
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 1.376367949797199e-06,
+      "loss": 0.7556,
+      "step": 641860
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 1.3752200198974516e-06,
+      "loss": 0.7275,
+      "step": 641870
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 1.374072089997704e-06,
+      "loss": 0.6372,
+      "step": 641880
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 1.3729241600979566e-06,
+      "loss": 0.6307,
+      "step": 641890
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 1.3717762301982092e-06,
+      "loss": 0.6798,
+      "step": 641900
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 1.3706283002984618e-06,
+      "loss": 0.6448,
+      "step": 641910
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 1.3694803703987142e-06,
+      "loss": 0.7344,
+      "step": 641920
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 1.3683324404989666e-06,
+      "loss": 0.6287,
+      "step": 641930
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 1.3671845105992194e-06,
+      "loss": 0.6733,
+      "step": 641940
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 1.3660365806994718e-06,
+      "loss": 0.6929,
+      "step": 641950
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 1.3648886507997244e-06,
+      "loss": 0.683,
+      "step": 641960
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 1.3637407208999768e-06,
+      "loss": 0.663,
+      "step": 641970
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 1.3625927910002296e-06,
+      "loss": 0.68,
+      "step": 641980
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 1.361444861100482e-06,
+      "loss": 0.598,
+      "step": 641990
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 1.3602969312007344e-06,
+      "loss": 0.7056,
+      "step": 642000
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 1.359149001300987e-06,
+      "loss": 0.7114,
+      "step": 642010
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 1.3580010714012396e-06,
+      "loss": 0.6627,
+      "step": 642020
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 1.3568531415014923e-06,
+      "loss": 0.6882,
+      "step": 642030
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 1.3557052116017446e-06,
+      "loss": 0.6849,
+      "step": 642040
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 1.3545572817019973e-06,
+      "loss": 0.7154,
+      "step": 642050
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 1.3534093518022499e-06,
+      "loss": 0.6566,
+      "step": 642060
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 1.3522614219025025e-06,
+      "loss": 0.6948,
+      "step": 642070
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 1.3511134920027549e-06,
+      "loss": 0.7322,
+      "step": 642080
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 1.3499655621030073e-06,
+      "loss": 0.6444,
+      "step": 642090
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 1.34881763220326e-06,
+      "loss": 0.7424,
+      "step": 642100
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 1.3476697023035125e-06,
+      "loss": 0.6759,
+      "step": 642110
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 1.346521772403765e-06,
+      "loss": 0.6566,
+      "step": 642120
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 1.3453738425040175e-06,
+      "loss": 0.6691,
+      "step": 642130
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 1.3442259126042703e-06,
+      "loss": 0.7663,
+      "step": 642140
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 1.3430779827045227e-06,
+      "loss": 0.636,
+      "step": 642150
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 1.3419300528047753e-06,
+      "loss": 0.7642,
+      "step": 642160
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 1.3407821229050277e-06,
+      "loss": 0.7148,
+      "step": 642170
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 1.3396341930052805e-06,
+      "loss": 0.643,
+      "step": 642180
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 1.338486263105533e-06,
+      "loss": 0.6252,
+      "step": 642190
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 1.3373383332057853e-06,
+      "loss": 0.6779,
+      "step": 642200
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 1.336190403306038e-06,
+      "loss": 0.6859,
+      "step": 642210
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 1.3350424734062905e-06,
+      "loss": 0.6568,
+      "step": 642220
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 1.3338945435065431e-06,
+      "loss": 0.7354,
+      "step": 642230
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 1.3327466136067955e-06,
+      "loss": 0.6978,
+      "step": 642240
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 1.3315986837070481e-06,
+      "loss": 0.7489,
+      "step": 642250
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 1.3304507538073007e-06,
+      "loss": 0.7,
+      "step": 642260
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 1.3293028239075533e-06,
+      "loss": 0.7166,
+      "step": 642270
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 1.3281548940078057e-06,
+      "loss": 0.6797,
+      "step": 642280
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 1.3270069641080586e-06,
+      "loss": 0.6995,
+      "step": 642290
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 1.325859034208311e-06,
+      "loss": 0.7167,
+      "step": 642300
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 1.3247111043085633e-06,
+      "loss": 0.6877,
+      "step": 642310
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 1.323563174408816e-06,
+      "loss": 0.6532,
+      "step": 642320
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 1.3224152445090686e-06,
+      "loss": 0.682,
+      "step": 642330
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 1.3212673146093212e-06,
+      "loss": 0.6588,
+      "step": 642340
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 1.3201193847095736e-06,
+      "loss": 0.6847,
+      "step": 642350
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 1.3189714548098262e-06,
+      "loss": 0.6433,
+      "step": 642360
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 1.3178235249100788e-06,
+      "loss": 0.7441,
+      "step": 642370
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 1.3166755950103314e-06,
+      "loss": 0.7134,
+      "step": 642380
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 1.3155276651105838e-06,
+      "loss": 0.6768,
+      "step": 642390
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 1.3143797352108362e-06,
+      "loss": 0.6755,
+      "step": 642400
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 1.313231805311089e-06,
+      "loss": 0.6178,
+      "step": 642410
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 1.3120838754113414e-06,
+      "loss": 0.6662,
+      "step": 642420
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 1.310935945511594e-06,
+      "loss": 0.7191,
+      "step": 642430
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 1.3097880156118464e-06,
+      "loss": 0.6819,
+      "step": 642440
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 1.3086400857120992e-06,
+      "loss": 0.63,
+      "step": 642450
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 1.3074921558123516e-06,
+      "loss": 0.5955,
+      "step": 642460
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 1.3063442259126042e-06,
+      "loss": 0.6108,
+      "step": 642470
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 1.3051962960128566e-06,
+      "loss": 0.6558,
+      "step": 642480
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 1.3040483661131094e-06,
+      "loss": 0.6653,
+      "step": 642490
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 1.3029004362133618e-06,
+      "loss": 0.701,
+      "step": 642500
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 1.3017525063136142e-06,
+      "loss": 0.6735,
+      "step": 642510
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 1.3006045764138668e-06,
+      "loss": 0.7323,
+      "step": 642520
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 1.2994566465141194e-06,
+      "loss": 0.626,
+      "step": 642530
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 1.298308716614372e-06,
+      "loss": 0.7276,
+      "step": 642540
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 1.2971607867146244e-06,
+      "loss": 0.7025,
+      "step": 642550
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 1.296012856814877e-06,
+      "loss": 0.6667,
+      "step": 642560
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 1.2948649269151296e-06,
+      "loss": 0.6933,
+      "step": 642570
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 1.2937169970153823e-06,
+      "loss": 0.7451,
+      "step": 642580
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 1.2925690671156347e-06,
+      "loss": 0.6796,
+      "step": 642590
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 1.291421137215887e-06,
+      "loss": 0.7241,
+      "step": 642600
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 1.2902732073161399e-06,
+      "loss": 0.7593,
+      "step": 642610
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 1.2891252774163923e-06,
+      "loss": 0.6413,
+      "step": 642620
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 1.2879773475166449e-06,
+      "loss": 0.6769,
+      "step": 642630
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 1.2868294176168973e-06,
+      "loss": 0.6913,
+      "step": 642640
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 1.28568148771715e-06,
+      "loss": 0.6597,
+      "step": 642650
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 1.2845335578174025e-06,
+      "loss": 0.6782,
+      "step": 642660
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 1.283385627917655e-06,
+      "loss": 0.6551,
+      "step": 642670
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 1.2822376980179075e-06,
+      "loss": 0.6545,
+      "step": 642680
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 1.2810897681181603e-06,
+      "loss": 0.7399,
+      "step": 642690
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 1.2799418382184127e-06,
+      "loss": 0.7756,
+      "step": 642700
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 1.278793908318665e-06,
+      "loss": 0.6763,
+      "step": 642710
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 1.2776459784189177e-06,
+      "loss": 0.6467,
+      "step": 642720
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 1.2764980485191703e-06,
+      "loss": 0.6507,
+      "step": 642730
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 1.275350118619423e-06,
+      "loss": 0.7504,
+      "step": 642740
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 1.2742021887196753e-06,
+      "loss": 0.6789,
+      "step": 642750
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 1.2730542588199281e-06,
+      "loss": 0.6468,
+      "step": 642760
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 1.2719063289201805e-06,
+      "loss": 0.6376,
+      "step": 642770
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 1.2707583990204331e-06,
+      "loss": 0.6502,
+      "step": 642780
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 1.2696104691206855e-06,
+      "loss": 0.7769,
+      "step": 642790
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 1.2684625392209383e-06,
+      "loss": 0.6475,
+      "step": 642800
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 1.2673146093211907e-06,
+      "loss": 0.7112,
+      "step": 642810
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 1.2661666794214431e-06,
+      "loss": 0.7147,
+      "step": 642820
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 1.2650187495216957e-06,
+      "loss": 0.5987,
+      "step": 642830
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 1.2638708196219483e-06,
+      "loss": 0.6775,
+      "step": 642840
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 1.262722889722201e-06,
+      "loss": 0.692,
+      "step": 642850
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 1.2615749598224534e-06,
+      "loss": 0.6969,
+      "step": 642860
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 1.260427029922706e-06,
+      "loss": 0.7089,
+      "step": 642870
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 1.2592791000229586e-06,
+      "loss": 0.6332,
+      "step": 642880
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 1.2581311701232112e-06,
+      "loss": 0.6584,
+      "step": 642890
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 1.2569832402234636e-06,
+      "loss": 0.7233,
+      "step": 642900
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 1.255835310323716e-06,
+      "loss": 0.6735,
+      "step": 642910
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 1.2546873804239688e-06,
+      "loss": 0.6111,
+      "step": 642920
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 1.2535394505242212e-06,
+      "loss": 0.6775,
+      "step": 642930
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 1.2523915206244738e-06,
+      "loss": 0.6441,
+      "step": 642940
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 1.2512435907247262e-06,
+      "loss": 0.706,
+      "step": 642950
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 1.250095660824979e-06,
+      "loss": 0.6715,
+      "step": 642960
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 1.2489477309252314e-06,
+      "loss": 0.7368,
+      "step": 642970
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 1.247799801025484e-06,
+      "loss": 0.6786,
+      "step": 642980
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 1.2466518711257364e-06,
+      "loss": 0.6878,
+      "step": 642990
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 1.2455039412259892e-06,
+      "loss": 0.6945,
+      "step": 643000
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 1.2443560113262416e-06,
+      "loss": 0.7584,
+      "step": 643010
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 1.243208081426494e-06,
+      "loss": 0.6993,
+      "step": 643020
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 1.2420601515267466e-06,
+      "loss": 0.7966,
+      "step": 643030
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 1.2409122216269992e-06,
+      "loss": 0.621,
+      "step": 643040
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 1.2397642917272518e-06,
+      "loss": 0.6879,
+      "step": 643050
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 1.2386163618275042e-06,
+      "loss": 0.6785,
+      "step": 643060
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 1.2374684319277566e-06,
+      "loss": 0.7081,
+      "step": 643070
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 1.2363205020280094e-06,
+      "loss": 0.6729,
+      "step": 643080
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 1.2351725721282618e-06,
+      "loss": 0.6074,
+      "step": 643090
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 1.2340246422285144e-06,
+      "loss": 0.6758,
+      "step": 643100
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 1.2328767123287668e-06,
+      "loss": 0.6843,
+      "step": 643110
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 1.2317287824290197e-06,
+      "loss": 0.6636,
+      "step": 643120
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 1.230580852529272e-06,
+      "loss": 0.6846,
+      "step": 643130
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 1.2294329226295247e-06,
+      "loss": 0.7506,
+      "step": 643140
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 1.228284992729777e-06,
+      "loss": 0.7132,
+      "step": 643150
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 1.2271370628300299e-06,
+      "loss": 0.6813,
+      "step": 643160
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 1.2259891329302823e-06,
+      "loss": 0.6625,
+      "step": 643170
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 1.2248412030305347e-06,
+      "loss": 0.672,
+      "step": 643180
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 1.2236932731307875e-06,
+      "loss": 0.6814,
+      "step": 643190
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 1.2225453432310399e-06,
+      "loss": 0.7008,
+      "step": 643200
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 1.2213974133312925e-06,
+      "loss": 0.6838,
+      "step": 643210
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 1.2202494834315449e-06,
+      "loss": 0.6774,
+      "step": 643220
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 1.2191015535317977e-06,
+      "loss": 0.6836,
+      "step": 643230
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 1.21795362363205e-06,
+      "loss": 0.6842,
+      "step": 643240
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 1.2168056937323027e-06,
+      "loss": 0.6898,
+      "step": 643250
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 1.215657763832555e-06,
+      "loss": 0.6886,
+      "step": 643260
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 1.214509833932808e-06,
+      "loss": 0.6944,
+      "step": 643270
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 1.2133619040330603e-06,
+      "loss": 0.681,
+      "step": 643280
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 1.2122139741333127e-06,
+      "loss": 0.6502,
+      "step": 643290
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 1.2110660442335653e-06,
+      "loss": 0.5848,
+      "step": 643300
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 1.209918114333818e-06,
+      "loss": 0.6078,
+      "step": 643310
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 1.2087701844340705e-06,
+      "loss": 0.6488,
+      "step": 643320
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 1.207622254534323e-06,
+      "loss": 0.7249,
+      "step": 643330
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 1.2064743246345755e-06,
+      "loss": 0.66,
+      "step": 643340
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 1.2053263947348281e-06,
+      "loss": 0.7275,
+      "step": 643350
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 1.2041784648350807e-06,
+      "loss": 0.6393,
+      "step": 643360
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 1.2030305349353331e-06,
+      "loss": 0.6942,
+      "step": 643370
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 1.2018826050355855e-06,
+      "loss": 0.7289,
+      "step": 643380
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 1.2007346751358384e-06,
+      "loss": 0.6733,
+      "step": 643390
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 1.1995867452360907e-06,
+      "loss": 0.7225,
+      "step": 643400
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 1.1984388153363434e-06,
+      "loss": 0.6689,
+      "step": 643410
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 1.1972908854365958e-06,
+      "loss": 0.6956,
+      "step": 643420
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 1.1961429555368486e-06,
+      "loss": 0.7195,
+      "step": 643430
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 1.194995025637101e-06,
+      "loss": 0.7261,
+      "step": 643440
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 1.1938470957373536e-06,
+      "loss": 0.707,
+      "step": 643450
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 1.192699165837606e-06,
+      "loss": 0.6312,
+      "step": 643460
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 1.1915512359378588e-06,
+      "loss": 0.6348,
+      "step": 643470
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 1.1904033060381112e-06,
+      "loss": 0.6822,
+      "step": 643480
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 1.1892553761383636e-06,
+      "loss": 0.6951,
+      "step": 643490
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 1.1881074462386162e-06,
+      "loss": 0.652,
+      "step": 643500
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 1.1869595163388688e-06,
+      "loss": 0.7117,
+      "step": 643510
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 1.1858115864391214e-06,
+      "loss": 0.6516,
+      "step": 643520
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 1.1846636565393738e-06,
+      "loss": 0.7228,
+      "step": 643530
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 1.1835157266396264e-06,
+      "loss": 0.6942,
+      "step": 643540
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 1.182367796739879e-06,
+      "loss": 0.5747,
+      "step": 643550
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 1.1812198668401316e-06,
+      "loss": 0.6878,
+      "step": 643560
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 1.180071936940384e-06,
+      "loss": 0.646,
+      "step": 643570
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 1.1789240070406364e-06,
+      "loss": 0.637,
+      "step": 643580
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 1.1777760771408892e-06,
+      "loss": 0.6582,
+      "step": 643590
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 1.1766281472411416e-06,
+      "loss": 0.6263,
+      "step": 643600
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 1.1754802173413942e-06,
+      "loss": 0.6405,
+      "step": 643610
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 1.1743322874416466e-06,
+      "loss": 0.6456,
+      "step": 643620
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 1.1731843575418994e-06,
+      "loss": 0.6756,
+      "step": 643630
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 1.1720364276421518e-06,
+      "loss": 0.6947,
+      "step": 643640
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 1.1708884977424044e-06,
+      "loss": 0.6446,
+      "step": 643650
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 1.169740567842657e-06,
+      "loss": 0.6936,
+      "step": 643660
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 1.1685926379429097e-06,
+      "loss": 0.6812,
+      "step": 643670
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 1.167444708043162e-06,
+      "loss": 0.6996,
+      "step": 643680
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 1.1662967781434147e-06,
+      "loss": 0.698,
+      "step": 643690
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 1.165148848243667e-06,
+      "loss": 0.634,
+      "step": 643700
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 1.1640009183439197e-06,
+      "loss": 0.5995,
+      "step": 643710
+    },
+    {
+      "epoch": 29.53,
+      "learning_rate": 1.1628529884441723e-06,
+      "loss": 0.6758,
+      "step": 643720
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 1.1617050585444249e-06,
+      "loss": 0.6243,
+      "step": 643730
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 1.1605571286446773e-06,
+      "loss": 0.5942,
+      "step": 643740
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 1.1594091987449299e-06,
+      "loss": 0.6628,
+      "step": 643750
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 1.1582612688451825e-06,
+      "loss": 0.7532,
+      "step": 643760
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 1.157113338945435e-06,
+      "loss": 0.6676,
+      "step": 643770
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 1.1559654090456875e-06,
+      "loss": 0.7065,
+      "step": 643780
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 1.15481747914594e-06,
+      "loss": 0.7179,
+      "step": 643790
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 1.1536695492461925e-06,
+      "loss": 0.6549,
+      "step": 643800
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 1.152521619346445e-06,
+      "loss": 0.6559,
+      "step": 643810
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 1.1513736894466977e-06,
+      "loss": 0.6435,
+      "step": 643820
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 1.1502257595469503e-06,
+      "loss": 0.6492,
+      "step": 643830
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 1.1490778296472027e-06,
+      "loss": 0.7107,
+      "step": 643840
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 1.1479298997474553e-06,
+      "loss": 0.6959,
+      "step": 643850
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 1.146781969847708e-06,
+      "loss": 0.6758,
+      "step": 643860
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 1.1456340399479605e-06,
+      "loss": 0.7083,
+      "step": 643870
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 1.144486110048213e-06,
+      "loss": 0.668,
+      "step": 643880
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 1.1433381801484655e-06,
+      "loss": 0.7379,
+      "step": 643890
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 1.142190250248718e-06,
+      "loss": 0.6121,
+      "step": 643900
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 1.1410423203489705e-06,
+      "loss": 0.7004,
+      "step": 643910
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 1.1398943904492231e-06,
+      "loss": 0.6957,
+      "step": 643920
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 1.1387464605494758e-06,
+      "loss": 0.7098,
+      "step": 643930
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 1.1375985306497281e-06,
+      "loss": 0.6299,
+      "step": 643940
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 1.1364506007499808e-06,
+      "loss": 0.7064,
+      "step": 643950
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 1.1353026708502334e-06,
+      "loss": 0.683,
+      "step": 643960
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 1.134154740950486e-06,
+      "loss": 0.6862,
+      "step": 643970
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 1.1330068110507384e-06,
+      "loss": 0.7146,
+      "step": 643980
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 1.131858881150991e-06,
+      "loss": 0.7458,
+      "step": 643990
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 1.1307109512512434e-06,
+      "loss": 0.6935,
+      "step": 644000
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 1.129563021351496e-06,
+      "loss": 0.6172,
+      "step": 644010
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 1.1284150914517486e-06,
+      "loss": 0.6709,
+      "step": 644020
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 1.1272671615520012e-06,
+      "loss": 0.7029,
+      "step": 644030
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 1.1261192316522536e-06,
+      "loss": 0.6815,
+      "step": 644040
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 1.1249713017525062e-06,
+      "loss": 0.7283,
+      "step": 644050
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 1.1238233718527588e-06,
+      "loss": 0.685,
+      "step": 644060
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 1.1226754419530114e-06,
+      "loss": 0.6565,
+      "step": 644070
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 1.121527512053264e-06,
+      "loss": 0.6153,
+      "step": 644080
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 1.1203795821535164e-06,
+      "loss": 0.6061,
+      "step": 644090
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 1.119231652253769e-06,
+      "loss": 0.6338,
+      "step": 644100
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 1.1180837223540214e-06,
+      "loss": 0.735,
+      "step": 644110
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 1.116935792454274e-06,
+      "loss": 0.71,
+      "step": 644120
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 1.1157878625545266e-06,
+      "loss": 0.6989,
+      "step": 644130
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 1.1146399326547792e-06,
+      "loss": 0.7129,
+      "step": 644140
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 1.1134920027550316e-06,
+      "loss": 0.6984,
+      "step": 644150
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 1.1123440728552842e-06,
+      "loss": 0.7003,
+      "step": 644160
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 1.1111961429555368e-06,
+      "loss": 0.6933,
+      "step": 644170
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 1.1100482130557894e-06,
+      "loss": 0.6616,
+      "step": 644180
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 1.1089002831560418e-06,
+      "loss": 0.6688,
+      "step": 644190
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 1.1077523532562945e-06,
+      "loss": 0.7683,
+      "step": 644200
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 1.1066044233565468e-06,
+      "loss": 0.7339,
+      "step": 644210
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 1.1054564934567995e-06,
+      "loss": 0.6462,
+      "step": 644220
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 1.104308563557052e-06,
+      "loss": 0.7168,
+      "step": 644230
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 1.1031606336573047e-06,
+      "loss": 0.6875,
+      "step": 644240
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 1.102012703757557e-06,
+      "loss": 0.6353,
+      "step": 644250
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 1.1008647738578097e-06,
+      "loss": 0.664,
+      "step": 644260
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 1.099716843958062e-06,
+      "loss": 0.5989,
+      "step": 644270
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 1.0985689140583147e-06,
+      "loss": 0.7104,
+      "step": 644280
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 1.0974209841585673e-06,
+      "loss": 0.7209,
+      "step": 644290
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 1.0962730542588199e-06,
+      "loss": 0.6676,
+      "step": 644300
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 1.0951251243590723e-06,
+      "loss": 0.6371,
+      "step": 644310
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 1.0939771944593249e-06,
+      "loss": 0.651,
+      "step": 644320
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 1.0928292645595775e-06,
+      "loss": 0.6892,
+      "step": 644330
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 1.09168133465983e-06,
+      "loss": 0.7463,
+      "step": 644340
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 1.0905334047600825e-06,
+      "loss": 0.6798,
+      "step": 644350
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 1.0893854748603351e-06,
+      "loss": 0.7286,
+      "step": 644360
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 1.0882375449605875e-06,
+      "loss": 0.6994,
+      "step": 644370
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 1.0870896150608401e-06,
+      "loss": 0.6846,
+      "step": 644380
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 1.0859416851610927e-06,
+      "loss": 0.6708,
+      "step": 644390
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 1.0847937552613453e-06,
+      "loss": 0.662,
+      "step": 644400
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 1.0836458253615977e-06,
+      "loss": 0.6389,
+      "step": 644410
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 1.0824978954618503e-06,
+      "loss": 0.695,
+      "step": 644420
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 1.081349965562103e-06,
+      "loss": 0.6803,
+      "step": 644430
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 1.0802020356623555e-06,
+      "loss": 0.6578,
+      "step": 644440
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 1.079054105762608e-06,
+      "loss": 0.6294,
+      "step": 644450
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 1.0779061758628605e-06,
+      "loss": 0.6863,
+      "step": 644460
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 1.076758245963113e-06,
+      "loss": 0.6397,
+      "step": 644470
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 1.0756103160633655e-06,
+      "loss": 0.633,
+      "step": 644480
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 1.0744623861636182e-06,
+      "loss": 0.6399,
+      "step": 644490
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 1.0733144562638708e-06,
+      "loss": 0.7501,
+      "step": 644500
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 1.0721665263641232e-06,
+      "loss": 0.7359,
+      "step": 644510
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 1.0710185964643758e-06,
+      "loss": 0.6832,
+      "step": 644520
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 1.0698706665646284e-06,
+      "loss": 0.6861,
+      "step": 644530
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 1.068722736664881e-06,
+      "loss": 0.6546,
+      "step": 644540
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 1.0675748067651336e-06,
+      "loss": 0.6658,
+      "step": 644550
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 1.066426876865386e-06,
+      "loss": 0.6649,
+      "step": 644560
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 1.0652789469656386e-06,
+      "loss": 0.7428,
+      "step": 644570
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 1.064131017065891e-06,
+      "loss": 0.6945,
+      "step": 644580
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 1.0629830871661436e-06,
+      "loss": 0.6904,
+      "step": 644590
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 1.0618351572663962e-06,
+      "loss": 0.7258,
+      "step": 644600
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 1.0606872273666488e-06,
+      "loss": 0.6697,
+      "step": 644610
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 1.0595392974669012e-06,
+      "loss": 0.6119,
+      "step": 644620
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 1.0583913675671538e-06,
+      "loss": 0.6945,
+      "step": 644630
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 1.0572434376674064e-06,
+      "loss": 0.7151,
+      "step": 644640
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 1.056095507767659e-06,
+      "loss": 0.6594,
+      "step": 644650
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 1.0549475778679114e-06,
+      "loss": 0.6764,
+      "step": 644660
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 1.053799647968164e-06,
+      "loss": 0.6837,
+      "step": 644670
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 1.0526517180684164e-06,
+      "loss": 0.6826,
+      "step": 644680
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 1.051503788168669e-06,
+      "loss": 0.6752,
+      "step": 644690
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 1.0503558582689216e-06,
+      "loss": 0.6959,
+      "step": 644700
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 1.0492079283691742e-06,
+      "loss": 0.6615,
+      "step": 644710
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 1.0480599984694266e-06,
+      "loss": 0.6997,
+      "step": 644720
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 1.0469120685696792e-06,
+      "loss": 0.7154,
+      "step": 644730
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 1.0457641386699318e-06,
+      "loss": 0.6302,
+      "step": 644740
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 1.0446162087701845e-06,
+      "loss": 0.6959,
+      "step": 644750
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 1.0434682788704369e-06,
+      "loss": 0.6459,
+      "step": 644760
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 1.0423203489706895e-06,
+      "loss": 0.6098,
+      "step": 644770
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 1.0411724190709419e-06,
+      "loss": 0.6926,
+      "step": 644780
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 1.0400244891711945e-06,
+      "loss": 0.6871,
+      "step": 644790
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 1.038876559271447e-06,
+      "loss": 0.6587,
+      "step": 644800
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 1.0377286293716997e-06,
+      "loss": 0.6641,
+      "step": 644810
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 1.036580699471952e-06,
+      "loss": 0.7233,
+      "step": 644820
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 1.0354327695722047e-06,
+      "loss": 0.712,
+      "step": 644830
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 1.0342848396724573e-06,
+      "loss": 0.6925,
+      "step": 644840
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 1.0331369097727099e-06,
+      "loss": 0.7277,
+      "step": 644850
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 1.0319889798729623e-06,
+      "loss": 0.6436,
+      "step": 644860
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 1.030841049973215e-06,
+      "loss": 0.6468,
+      "step": 644870
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 1.0296931200734673e-06,
+      "loss": 0.594,
+      "step": 644880
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 1.02854519017372e-06,
+      "loss": 0.6955,
+      "step": 644890
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 1.0273972602739725e-06,
+      "loss": 0.7002,
+      "step": 644900
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 1.0262493303742251e-06,
+      "loss": 0.6405,
+      "step": 644910
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 1.0251014004744775e-06,
+      "loss": 0.6899,
+      "step": 644920
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 1.0239534705747301e-06,
+      "loss": 0.6639,
+      "step": 644930
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 1.0228055406749827e-06,
+      "loss": 0.6613,
+      "step": 644940
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 1.0216576107752353e-06,
+      "loss": 0.6584,
+      "step": 644950
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 1.020509680875488e-06,
+      "loss": 0.7189,
+      "step": 644960
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 1.0193617509757403e-06,
+      "loss": 0.7339,
+      "step": 644970
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 1.018213821075993e-06,
+      "loss": 0.7965,
+      "step": 644980
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 1.0170658911762453e-06,
+      "loss": 0.6876,
+      "step": 644990
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 1.015917961276498e-06,
+      "loss": 0.6253,
+      "step": 645000
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 1.0147700313767505e-06,
+      "loss": 0.6427,
+      "step": 645010
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 1.0136221014770032e-06,
+      "loss": 0.7055,
+      "step": 645020
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 1.0124741715772555e-06,
+      "loss": 0.6751,
+      "step": 645030
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 1.0113262416775082e-06,
+      "loss": 0.6987,
+      "step": 645040
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 1.0101783117777608e-06,
+      "loss": 0.6227,
+      "step": 645050
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 1.0090303818780134e-06,
+      "loss": 0.6753,
+      "step": 645060
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 1.0078824519782658e-06,
+      "loss": 0.7356,
+      "step": 645070
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 1.0067345220785184e-06,
+      "loss": 0.6916,
+      "step": 645080
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 1.0055865921787708e-06,
+      "loss": 0.7223,
+      "step": 645090
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 1.0044386622790234e-06,
+      "loss": 0.6031,
+      "step": 645100
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 1.003290732379276e-06,
+      "loss": 0.691,
+      "step": 645110
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 1.0021428024795286e-06,
+      "loss": 0.6984,
+      "step": 645120
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 1.000994872579781e-06,
+      "loss": 0.6793,
+      "step": 645130
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 9.998469426800336e-07,
+      "loss": 0.6776,
+      "step": 645140
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 9.986990127802862e-07,
+      "loss": 0.706,
+      "step": 645150
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 9.975510828805388e-07,
+      "loss": 0.6725,
+      "step": 645160
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 9.964031529807912e-07,
+      "loss": 0.6649,
+      "step": 645170
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 9.952552230810438e-07,
+      "loss": 0.6907,
+      "step": 645180
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 9.941072931812962e-07,
+      "loss": 0.6998,
+      "step": 645190
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 9.929593632815488e-07,
+      "loss": 0.6881,
+      "step": 645200
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 9.918114333818014e-07,
+      "loss": 0.6619,
+      "step": 645210
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 9.90663503482054e-07,
+      "loss": 0.6456,
+      "step": 645220
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 9.895155735823064e-07,
+      "loss": 0.6913,
+      "step": 645230
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 9.88367643682559e-07,
+      "loss": 0.7188,
+      "step": 645240
+    },
+    {
+      "epoch": 29.6,
+      "learning_rate": 9.872197137828116e-07,
+      "loss": 0.7075,
+      "step": 645250
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 9.860717838830642e-07,
+      "loss": 0.7447,
+      "step": 645260
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 9.849238539833166e-07,
+      "loss": 0.7052,
+      "step": 645270
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 9.837759240835692e-07,
+      "loss": 0.6252,
+      "step": 645280
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 9.826279941838216e-07,
+      "loss": 0.6387,
+      "step": 645290
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 9.814800642840742e-07,
+      "loss": 0.6237,
+      "step": 645300
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 9.803321343843269e-07,
+      "loss": 0.5503,
+      "step": 645310
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 9.791842044845795e-07,
+      "loss": 0.6445,
+      "step": 645320
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 9.780362745848319e-07,
+      "loss": 0.6701,
+      "step": 645330
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 9.768883446850845e-07,
+      "loss": 0.6913,
+      "step": 645340
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 9.757404147853369e-07,
+      "loss": 0.7408,
+      "step": 645350
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 9.745924848855895e-07,
+      "loss": 0.7207,
+      "step": 645360
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 9.73444554985842e-07,
+      "loss": 0.6861,
+      "step": 645370
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 9.722966250860947e-07,
+      "loss": 0.5942,
+      "step": 645380
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 9.71148695186347e-07,
+      "loss": 0.6725,
+      "step": 645390
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 9.700007652865997e-07,
+      "loss": 0.6989,
+      "step": 645400
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 9.688528353868523e-07,
+      "loss": 0.6283,
+      "step": 645410
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 9.67704905487105e-07,
+      "loss": 0.7416,
+      "step": 645420
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 9.665569755873575e-07,
+      "loss": 0.778,
+      "step": 645430
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 9.6540904568761e-07,
+      "loss": 0.7166,
+      "step": 645440
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 9.642611157878625e-07,
+      "loss": 0.6521,
+      "step": 645450
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 9.63113185888115e-07,
+      "loss": 0.7151,
+      "step": 645460
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 9.619652559883675e-07,
+      "loss": 0.6029,
+      "step": 645470
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 9.608173260886201e-07,
+      "loss": 0.6292,
+      "step": 645480
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 9.596693961888727e-07,
+      "loss": 0.7144,
+      "step": 645490
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 9.585214662891251e-07,
+      "loss": 0.7322,
+      "step": 645500
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 9.573735363893777e-07,
+      "loss": 0.6939,
+      "step": 645510
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 9.562256064896303e-07,
+      "loss": 0.7378,
+      "step": 645520
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 9.55077676589883e-07,
+      "loss": 0.6921,
+      "step": 645530
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 9.539297466901353e-07,
+      "loss": 0.7931,
+      "step": 645540
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 9.527818167903879e-07,
+      "loss": 0.6754,
+      "step": 645550
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 9.516338868906404e-07,
+      "loss": 0.653,
+      "step": 645560
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 9.504859569908931e-07,
+      "loss": 0.6872,
+      "step": 645570
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 9.493380270911456e-07,
+      "loss": 0.6818,
+      "step": 645580
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 9.481900971913982e-07,
+      "loss": 0.6252,
+      "step": 645590
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 9.470421672916507e-07,
+      "loss": 0.7465,
+      "step": 645600
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 9.458942373919033e-07,
+      "loss": 0.6107,
+      "step": 645610
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 9.447463074921557e-07,
+      "loss": 0.6624,
+      "step": 645620
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 9.435983775924083e-07,
+      "loss": 0.7586,
+      "step": 645630
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 9.424504476926608e-07,
+      "loss": 0.7108,
+      "step": 645640
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 9.413025177929134e-07,
+      "loss": 0.6749,
+      "step": 645650
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 9.401545878931659e-07,
+      "loss": 0.6826,
+      "step": 645660
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 9.390066579934185e-07,
+      "loss": 0.6085,
+      "step": 645670
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 9.37858728093671e-07,
+      "loss": 0.7368,
+      "step": 645680
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 9.367107981939236e-07,
+      "loss": 0.6537,
+      "step": 645690
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 9.35562868294176e-07,
+      "loss": 0.784,
+      "step": 645700
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 9.344149383944286e-07,
+      "loss": 0.6921,
+      "step": 645710
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 9.332670084946811e-07,
+      "loss": 0.6316,
+      "step": 645720
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 9.321190785949337e-07,
+      "loss": 0.7449,
+      "step": 645730
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 9.309711486951862e-07,
+      "loss": 0.7607,
+      "step": 645740
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 9.298232187954388e-07,
+      "loss": 0.665,
+      "step": 645750
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 9.286752888956913e-07,
+      "loss": 0.6977,
+      "step": 645760
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 9.275273589959439e-07,
+      "loss": 0.625,
+      "step": 645770
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 9.263794290961964e-07,
+      "loss": 0.7177,
+      "step": 645780
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 9.25231499196449e-07,
+      "loss": 0.7098,
+      "step": 645790
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 9.240835692967014e-07,
+      "loss": 0.6462,
+      "step": 645800
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 9.22935639396954e-07,
+      "loss": 0.6221,
+      "step": 645810
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 9.217877094972065e-07,
+      "loss": 0.6537,
+      "step": 645820
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 9.206397795974591e-07,
+      "loss": 0.7232,
+      "step": 645830
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 9.194918496977116e-07,
+      "loss": 0.6499,
+      "step": 645840
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 9.183439197979643e-07,
+      "loss": 0.6793,
+      "step": 645850
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 9.171959898982169e-07,
+      "loss": 0.7317,
+      "step": 645860
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 9.160480599984694e-07,
+      "loss": 0.682,
+      "step": 645870
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 9.14900130098722e-07,
+      "loss": 0.6614,
+      "step": 645880
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 9.137522001989745e-07,
+      "loss": 0.7191,
+      "step": 645890
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 9.126042702992271e-07,
+      "loss": 0.654,
+      "step": 645900
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 9.114563403994795e-07,
+      "loss": 0.739,
+      "step": 645910
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 9.103084104997321e-07,
+      "loss": 0.6647,
+      "step": 645920
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 9.091604805999846e-07,
+      "loss": 0.729,
+      "step": 645930
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 9.080125507002372e-07,
+      "loss": 0.6322,
+      "step": 645940
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 9.068646208004897e-07,
+      "loss": 0.6882,
+      "step": 645950
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 9.057166909007423e-07,
+      "loss": 0.6655,
+      "step": 645960
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 9.045687610009948e-07,
+      "loss": 0.6951,
+      "step": 645970
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 9.034208311012474e-07,
+      "loss": 0.6947,
+      "step": 645980
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 9.022729012014999e-07,
+      "loss": 0.7428,
+      "step": 645990
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 9.011249713017525e-07,
+      "loss": 0.6956,
+      "step": 646000
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 8.999770414020049e-07,
+      "loss": 0.6772,
+      "step": 646010
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 8.988291115022575e-07,
+      "loss": 0.6828,
+      "step": 646020
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 8.9768118160251e-07,
+      "loss": 0.7018,
+      "step": 646030
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 8.965332517027626e-07,
+      "loss": 0.6568,
+      "step": 646040
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 8.953853218030151e-07,
+      "loss": 0.6828,
+      "step": 646050
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 8.942373919032677e-07,
+      "loss": 0.6467,
+      "step": 646060
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 8.930894620035202e-07,
+      "loss": 0.7274,
+      "step": 646070
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 8.919415321037728e-07,
+      "loss": 0.6635,
+      "step": 646080
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 8.907936022040253e-07,
+      "loss": 0.6861,
+      "step": 646090
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 8.89645672304278e-07,
+      "loss": 0.6812,
+      "step": 646100
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 8.884977424045303e-07,
+      "loss": 0.7028,
+      "step": 646110
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 8.87349812504783e-07,
+      "loss": 0.6719,
+      "step": 646120
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 8.862018826050355e-07,
+      "loss": 0.7221,
+      "step": 646130
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 8.850539527052881e-07,
+      "loss": 0.6275,
+      "step": 646140
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 8.839060228055406e-07,
+      "loss": 0.7414,
+      "step": 646150
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 8.827580929057932e-07,
+      "loss": 0.7252,
+      "step": 646160
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 8.816101630060457e-07,
+      "loss": 0.6974,
+      "step": 646170
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 8.804622331062983e-07,
+      "loss": 0.6434,
+      "step": 646180
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 8.793143032065507e-07,
+      "loss": 0.673,
+      "step": 646190
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 8.781663733068034e-07,
+      "loss": 0.7996,
+      "step": 646200
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 8.770184434070558e-07,
+      "loss": 0.6612,
+      "step": 646210
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 8.758705135073084e-07,
+      "loss": 0.6869,
+      "step": 646220
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 8.747225836075609e-07,
+      "loss": 0.6467,
+      "step": 646230
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 8.735746537078135e-07,
+      "loss": 0.642,
+      "step": 646240
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 8.72426723808066e-07,
+      "loss": 0.6592,
+      "step": 646250
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 8.712787939083186e-07,
+      "loss": 0.6792,
+      "step": 646260
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 8.701308640085711e-07,
+      "loss": 0.6262,
+      "step": 646270
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 8.689829341088237e-07,
+      "loss": 0.7136,
+      "step": 646280
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 8.678350042090763e-07,
+      "loss": 0.7022,
+      "step": 646290
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 8.666870743093287e-07,
+      "loss": 0.6949,
+      "step": 646300
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 8.655391444095813e-07,
+      "loss": 0.6437,
+      "step": 646310
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 8.643912145098338e-07,
+      "loss": 0.7049,
+      "step": 646320
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 8.632432846100864e-07,
+      "loss": 0.7026,
+      "step": 646330
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 8.620953547103389e-07,
+      "loss": 0.7441,
+      "step": 646340
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 8.609474248105915e-07,
+      "loss": 0.7508,
+      "step": 646350
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 8.59799494910844e-07,
+      "loss": 0.6886,
+      "step": 646360
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 8.586515650110966e-07,
+      "loss": 0.7653,
+      "step": 646370
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 8.575036351113492e-07,
+      "loss": 0.7174,
+      "step": 646380
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 8.563557052116018e-07,
+      "loss": 0.5505,
+      "step": 646390
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 8.552077753118542e-07,
+      "loss": 0.7098,
+      "step": 646400
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 8.540598454121068e-07,
+      "loss": 0.6363,
+      "step": 646410
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 8.529119155123593e-07,
+      "loss": 0.6828,
+      "step": 646420
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 8.517639856126119e-07,
+      "loss": 0.6398,
+      "step": 646430
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 8.506160557128644e-07,
+      "loss": 0.6993,
+      "step": 646440
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 8.49468125813117e-07,
+      "loss": 0.73,
+      "step": 646450
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 8.483201959133695e-07,
+      "loss": 0.6843,
+      "step": 646460
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 8.471722660136221e-07,
+      "loss": 0.6539,
+      "step": 646470
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 8.460243361138746e-07,
+      "loss": 0.6675,
+      "step": 646480
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 8.448764062141272e-07,
+      "loss": 0.7255,
+      "step": 646490
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 8.437284763143796e-07,
+      "loss": 0.6341,
+      "step": 646500
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 8.425805464146322e-07,
+      "loss": 0.778,
+      "step": 646510
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 8.414326165148847e-07,
+      "loss": 0.7304,
+      "step": 646520
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 8.402846866151373e-07,
+      "loss": 0.6615,
+      "step": 646530
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 8.391367567153898e-07,
+      "loss": 0.6058,
+      "step": 646540
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 8.379888268156424e-07,
+      "loss": 0.609,
+      "step": 646550
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 8.368408969158949e-07,
+      "loss": 0.6643,
+      "step": 646560
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 8.356929670161475e-07,
+      "loss": 0.6366,
+      "step": 646570
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 8.345450371164e-07,
+      "loss": 0.7453,
+      "step": 646580
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 8.333971072166526e-07,
+      "loss": 0.7027,
+      "step": 646590
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 8.32249177316905e-07,
+      "loss": 0.6928,
+      "step": 646600
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 8.311012474171576e-07,
+      "loss": 0.6775,
+      "step": 646610
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 8.299533175174101e-07,
+      "loss": 0.6255,
+      "step": 646620
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 8.288053876176627e-07,
+      "loss": 0.6865,
+      "step": 646630
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 8.276574577179152e-07,
+      "loss": 0.635,
+      "step": 646640
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 8.265095278181678e-07,
+      "loss": 0.7174,
+      "step": 646650
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 8.253615979184204e-07,
+      "loss": 0.7241,
+      "step": 646660
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 8.24213668018673e-07,
+      "loss": 0.736,
+      "step": 646670
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 8.230657381189255e-07,
+      "loss": 0.6785,
+      "step": 646680
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 8.219178082191781e-07,
+      "loss": 0.7449,
+      "step": 646690
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 8.207698783194305e-07,
+      "loss": 0.6175,
+      "step": 646700
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 8.196219484196831e-07,
+      "loss": 0.7076,
+      "step": 646710
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 8.184740185199356e-07,
+      "loss": 0.6935,
+      "step": 646720
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 8.173260886201882e-07,
+      "loss": 0.6578,
+      "step": 646730
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 8.161781587204408e-07,
+      "loss": 0.6289,
+      "step": 646740
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 8.150302288206933e-07,
+      "loss": 0.6865,
+      "step": 646750
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 8.138822989209459e-07,
+      "loss": 0.6763,
+      "step": 646760
+    },
+    {
+      "epoch": 29.67,
+      "learning_rate": 8.127343690211984e-07,
+      "loss": 0.741,
+      "step": 646770
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 8.11586439121451e-07,
+      "loss": 0.6638,
+      "step": 646780
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 8.104385092217034e-07,
+      "loss": 0.6471,
+      "step": 646790
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 8.09290579321956e-07,
+      "loss": 0.7195,
+      "step": 646800
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 8.081426494222085e-07,
+      "loss": 0.6809,
+      "step": 646810
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 8.069947195224611e-07,
+      "loss": 0.6923,
+      "step": 646820
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 8.058467896227136e-07,
+      "loss": 0.6164,
+      "step": 646830
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 8.046988597229662e-07,
+      "loss": 0.6777,
+      "step": 646840
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 8.035509298232187e-07,
+      "loss": 0.654,
+      "step": 646850
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 8.024029999234713e-07,
+      "loss": 0.6668,
+      "step": 646860
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 8.012550700237238e-07,
+      "loss": 0.6614,
+      "step": 646870
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 8.001071401239764e-07,
+      "loss": 0.696,
+      "step": 646880
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 7.989592102242288e-07,
+      "loss": 0.6481,
+      "step": 646890
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 7.978112803244814e-07,
+      "loss": 0.6926,
+      "step": 646900
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 7.966633504247339e-07,
+      "loss": 0.6607,
+      "step": 646910
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 7.955154205249865e-07,
+      "loss": 0.6475,
+      "step": 646920
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 7.94367490625239e-07,
+      "loss": 0.6851,
+      "step": 646930
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 7.932195607254917e-07,
+      "loss": 0.6646,
+      "step": 646940
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 7.920716308257442e-07,
+      "loss": 0.7041,
+      "step": 646950
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 7.909237009259968e-07,
+      "loss": 0.6997,
+      "step": 646960
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 7.897757710262493e-07,
+      "loss": 0.638,
+      "step": 646970
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 7.886278411265019e-07,
+      "loss": 0.7066,
+      "step": 646980
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 7.874799112267543e-07,
+      "loss": 0.706,
+      "step": 646990
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 7.863319813270069e-07,
+      "loss": 0.7482,
+      "step": 647000
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 7.851840514272594e-07,
+      "loss": 0.7309,
+      "step": 647010
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 7.84036121527512e-07,
+      "loss": 0.668,
+      "step": 647020
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 7.828881916277645e-07,
+      "loss": 0.7269,
+      "step": 647030
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 7.817402617280171e-07,
+      "loss": 0.6916,
+      "step": 647040
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 7.805923318282696e-07,
+      "loss": 0.6865,
+      "step": 647050
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 7.794444019285222e-07,
+      "loss": 0.6392,
+      "step": 647060
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 7.782964720287747e-07,
+      "loss": 0.7408,
+      "step": 647070
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 7.771485421290273e-07,
+      "loss": 0.6663,
+      "step": 647080
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 7.760006122292797e-07,
+      "loss": 0.6615,
+      "step": 647090
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 7.748526823295323e-07,
+      "loss": 0.8007,
+      "step": 647100
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 7.737047524297848e-07,
+      "loss": 0.6241,
+      "step": 647110
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 7.725568225300374e-07,
+      "loss": 0.6647,
+      "step": 647120
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 7.714088926302899e-07,
+      "loss": 0.6528,
+      "step": 647130
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 7.702609627305425e-07,
+      "loss": 0.7205,
+      "step": 647140
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 7.69113032830795e-07,
+      "loss": 0.6638,
+      "step": 647150
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 7.679651029310476e-07,
+      "loss": 0.7243,
+      "step": 647160
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 7.668171730313001e-07,
+      "loss": 0.6215,
+      "step": 647170
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 7.656692431315527e-07,
+      "loss": 0.7027,
+      "step": 647180
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 7.645213132318054e-07,
+      "loss": 0.7722,
+      "step": 647190
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 7.633733833320577e-07,
+      "loss": 0.7739,
+      "step": 647200
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 7.622254534323104e-07,
+      "loss": 0.6742,
+      "step": 647210
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 7.610775235325629e-07,
+      "loss": 0.6924,
+      "step": 647220
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 7.599295936328155e-07,
+      "loss": 0.6697,
+      "step": 647230
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 7.58781663733068e-07,
+      "loss": 0.667,
+      "step": 647240
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 7.576337338333206e-07,
+      "loss": 0.708,
+      "step": 647250
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 7.564858039335731e-07,
+      "loss": 0.6554,
+      "step": 647260
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 7.553378740338257e-07,
+      "loss": 0.6526,
+      "step": 647270
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 7.541899441340782e-07,
+      "loss": 0.7755,
+      "step": 647280
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 7.530420142343308e-07,
+      "loss": 0.7245,
+      "step": 647290
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 7.518940843345832e-07,
+      "loss": 0.7027,
+      "step": 647300
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 7.507461544348358e-07,
+      "loss": 0.6581,
+      "step": 647310
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 7.495982245350883e-07,
+      "loss": 0.8335,
+      "step": 647320
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 7.484502946353409e-07,
+      "loss": 0.6518,
+      "step": 647330
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 7.473023647355934e-07,
+      "loss": 0.6443,
+      "step": 647340
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 7.46154434835846e-07,
+      "loss": 0.7222,
+      "step": 647350
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 7.450065049360985e-07,
+      "loss": 0.7207,
+      "step": 647360
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 7.438585750363511e-07,
+      "loss": 0.6235,
+      "step": 647370
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 7.427106451366035e-07,
+      "loss": 0.6827,
+      "step": 647380
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 7.415627152368561e-07,
+      "loss": 0.6665,
+      "step": 647390
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 7.404147853371086e-07,
+      "loss": 0.7088,
+      "step": 647400
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 7.392668554373612e-07,
+      "loss": 0.6706,
+      "step": 647410
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 7.381189255376137e-07,
+      "loss": 0.6336,
+      "step": 647420
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 7.369709956378663e-07,
+      "loss": 0.7211,
+      "step": 647430
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 7.358230657381188e-07,
+      "loss": 0.614,
+      "step": 647440
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 7.346751358383714e-07,
+      "loss": 0.7858,
+      "step": 647450
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 7.335272059386239e-07,
+      "loss": 0.682,
+      "step": 647460
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 7.323792760388766e-07,
+      "loss": 0.6801,
+      "step": 647470
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 7.31231346139129e-07,
+      "loss": 0.7157,
+      "step": 647480
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 7.300834162393816e-07,
+      "loss": 0.5917,
+      "step": 647490
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 7.289354863396341e-07,
+      "loss": 0.7439,
+      "step": 647500
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 7.277875564398867e-07,
+      "loss": 0.7116,
+      "step": 647510
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 7.266396265401392e-07,
+      "loss": 0.7763,
+      "step": 647520
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 7.254916966403918e-07,
+      "loss": 0.636,
+      "step": 647530
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 7.243437667406443e-07,
+      "loss": 0.6825,
+      "step": 647540
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 7.231958368408969e-07,
+      "loss": 0.6377,
+      "step": 647550
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 7.220479069411494e-07,
+      "loss": 0.7084,
+      "step": 647560
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 7.20899977041402e-07,
+      "loss": 0.6417,
+      "step": 647570
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 7.197520471416544e-07,
+      "loss": 0.6751,
+      "step": 647580
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 7.18604117241907e-07,
+      "loss": 0.6524,
+      "step": 647590
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 7.174561873421595e-07,
+      "loss": 0.7551,
+      "step": 647600
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 7.163082574424121e-07,
+      "loss": 0.6948,
+      "step": 647610
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 7.151603275426646e-07,
+      "loss": 0.68,
+      "step": 647620
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 7.140123976429172e-07,
+      "loss": 0.7217,
+      "step": 647630
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 7.128644677431698e-07,
+      "loss": 0.6291,
+      "step": 647640
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 7.117165378434223e-07,
+      "loss": 0.6867,
+      "step": 647650
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 7.105686079436749e-07,
+      "loss": 0.7773,
+      "step": 647660
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 7.094206780439274e-07,
+      "loss": 0.7246,
+      "step": 647670
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 7.0827274814418e-07,
+      "loss": 0.6724,
+      "step": 647680
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 7.071248182444324e-07,
+      "loss": 0.7293,
+      "step": 647690
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 7.05976888344685e-07,
+      "loss": 0.7283,
+      "step": 647700
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 7.048289584449375e-07,
+      "loss": 0.6961,
+      "step": 647710
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 7.036810285451901e-07,
+      "loss": 0.6817,
+      "step": 647720
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 7.025330986454426e-07,
+      "loss": 0.7207,
+      "step": 647730
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 7.013851687456953e-07,
+      "loss": 0.6157,
+      "step": 647740
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 7.002372388459478e-07,
+      "loss": 0.6573,
+      "step": 647750
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 6.990893089462004e-07,
+      "loss": 0.7174,
+      "step": 647760
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 6.979413790464529e-07,
+      "loss": 0.6322,
+      "step": 647770
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 6.967934491467055e-07,
+      "loss": 0.6965,
+      "step": 647780
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 6.956455192469579e-07,
+      "loss": 0.684,
+      "step": 647790
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 6.944975893472105e-07,
+      "loss": 0.6583,
+      "step": 647800
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 6.93349659447463e-07,
+      "loss": 0.7349,
+      "step": 647810
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 6.922017295477156e-07,
+      "loss": 0.7243,
+      "step": 647820
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 6.910537996479681e-07,
+      "loss": 0.6141,
+      "step": 647830
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 6.899058697482207e-07,
+      "loss": 0.6956,
+      "step": 647840
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 6.887579398484732e-07,
+      "loss": 0.7086,
+      "step": 647850
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 6.876100099487258e-07,
+      "loss": 0.7376,
+      "step": 647860
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 6.864620800489783e-07,
+      "loss": 0.6139,
+      "step": 647870
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 6.853141501492309e-07,
+      "loss": 0.6468,
+      "step": 647880
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 6.841662202494833e-07,
+      "loss": 0.6803,
+      "step": 647890
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 6.830182903497359e-07,
+      "loss": 0.6368,
+      "step": 647900
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 6.818703604499884e-07,
+      "loss": 0.6886,
+      "step": 647910
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 6.80722430550241e-07,
+      "loss": 0.687,
+      "step": 647920
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 6.795745006504935e-07,
+      "loss": 0.6379,
+      "step": 647930
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 6.784265707507461e-07,
+      "loss": 0.6578,
+      "step": 647940
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 6.772786408509986e-07,
+      "loss": 0.7021,
+      "step": 647950
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 6.761307109512512e-07,
+      "loss": 0.6753,
+      "step": 647960
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 6.749827810515036e-07,
+      "loss": 0.665,
+      "step": 647970
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 6.738348511517562e-07,
+      "loss": 0.6542,
+      "step": 647980
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 6.726869212520087e-07,
+      "loss": 0.6685,
+      "step": 647990
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 6.715389913522613e-07,
+      "loss": 0.6892,
+      "step": 648000
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 6.703910614525138e-07,
+      "loss": 0.6978,
+      "step": 648010
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 6.692431315527665e-07,
+      "loss": 0.6529,
+      "step": 648020
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 6.68095201653019e-07,
+      "loss": 0.7616,
+      "step": 648030
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 6.669472717532716e-07,
+      "loss": 0.6302,
+      "step": 648040
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 6.657993418535241e-07,
+      "loss": 0.6657,
+      "step": 648050
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 6.646514119537767e-07,
+      "loss": 0.7095,
+      "step": 648060
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 6.635034820540293e-07,
+      "loss": 0.6739,
+      "step": 648070
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 6.623555521542817e-07,
+      "loss": 0.5739,
+      "step": 648080
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 6.612076222545343e-07,
+      "loss": 0.7525,
+      "step": 648090
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 6.600596923547868e-07,
+      "loss": 0.6838,
+      "step": 648100
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 6.589117624550394e-07,
+      "loss": 0.726,
+      "step": 648110
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 6.577638325552919e-07,
+      "loss": 0.7091,
+      "step": 648120
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 6.566159026555445e-07,
+      "loss": 0.6609,
+      "step": 648130
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 6.55467972755797e-07,
+      "loss": 0.6869,
+      "step": 648140
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 6.543200428560496e-07,
+      "loss": 0.6882,
+      "step": 648150
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 6.531721129563021e-07,
+      "loss": 0.6737,
+      "step": 648160
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 6.520241830565547e-07,
+      "loss": 0.6824,
+      "step": 648170
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 6.508762531568071e-07,
+      "loss": 0.6302,
+      "step": 648180
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 6.497283232570597e-07,
+      "loss": 0.6706,
+      "step": 648190
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 6.485803933573122e-07,
+      "loss": 0.6918,
+      "step": 648200
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 6.474324634575648e-07,
+      "loss": 0.723,
+      "step": 648210
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 6.462845335578173e-07,
+      "loss": 0.6799,
+      "step": 648220
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 6.451366036580699e-07,
+      "loss": 0.6157,
+      "step": 648230
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 6.439886737583224e-07,
+      "loss": 0.6459,
+      "step": 648240
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 6.42840743858575e-07,
+      "loss": 0.7188,
+      "step": 648250
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 6.416928139588275e-07,
+      "loss": 0.7119,
+      "step": 648260
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 6.405448840590801e-07,
+      "loss": 0.6264,
+      "step": 648270
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 6.393969541593325e-07,
+      "loss": 0.617,
+      "step": 648280
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 6.382490242595852e-07,
+      "loss": 0.6901,
+      "step": 648290
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 6.371010943598377e-07,
+      "loss": 0.7063,
+      "step": 648300
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 6.359531644600903e-07,
+      "loss": 0.6829,
+      "step": 648310
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 6.348052345603428e-07,
+      "loss": 0.7,
+      "step": 648320
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 6.336573046605954e-07,
+      "loss": 0.7086,
+      "step": 648330
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 6.325093747608479e-07,
+      "loss": 0.6821,
+      "step": 648340
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 6.313614448611005e-07,
+      "loss": 0.5785,
+      "step": 648350
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 6.30213514961353e-07,
+      "loss": 0.6644,
+      "step": 648360
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 6.290655850616056e-07,
+      "loss": 0.6071,
+      "step": 648370
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 6.27917655161858e-07,
+      "loss": 0.6863,
+      "step": 648380
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 6.267697252621106e-07,
+      "loss": 0.6204,
+      "step": 648390
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 6.256217953623631e-07,
+      "loss": 0.74,
+      "step": 648400
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 6.244738654626157e-07,
+      "loss": 0.6915,
+      "step": 648410
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 6.233259355628682e-07,
+      "loss": 0.7123,
+      "step": 648420
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 6.221780056631208e-07,
+      "loss": 0.6845,
+      "step": 648430
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 6.210300757633733e-07,
+      "loss": 0.6868,
+      "step": 648440
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 6.198821458636259e-07,
+      "loss": 0.6866,
+      "step": 648450
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 6.187342159638783e-07,
+      "loss": 0.6983,
+      "step": 648460
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 6.175862860641309e-07,
+      "loss": 0.6687,
+      "step": 648470
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 6.164383561643834e-07,
+      "loss": 0.6806,
+      "step": 648480
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 6.15290426264636e-07,
+      "loss": 0.701,
+      "step": 648490
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 6.141424963648885e-07,
+      "loss": 0.651,
+      "step": 648500
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 6.129945664651411e-07,
+      "loss": 0.631,
+      "step": 648510
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 6.118466365653937e-07,
+      "loss": 0.7405,
+      "step": 648520
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 6.106987066656462e-07,
+      "loss": 0.8089,
+      "step": 648530
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 6.095507767658988e-07,
+      "loss": 0.6404,
+      "step": 648540
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 6.084028468661514e-07,
+      "loss": 0.6357,
+      "step": 648550
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 6.07254916966404e-07,
+      "loss": 0.6121,
+      "step": 648560
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 6.061069870666564e-07,
+      "loss": 0.6458,
+      "step": 648570
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 6.04959057166909e-07,
+      "loss": 0.6163,
+      "step": 648580
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 6.038111272671615e-07,
+      "loss": 0.6777,
+      "step": 648590
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 6.026631973674141e-07,
+      "loss": 0.7168,
+      "step": 648600
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 6.015152674676666e-07,
+      "loss": 0.7216,
+      "step": 648610
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 6.003673375679192e-07,
+      "loss": 0.7251,
+      "step": 648620
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 5.992194076681717e-07,
+      "loss": 0.5926,
+      "step": 648630
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 5.980714777684243e-07,
+      "loss": 0.6557,
+      "step": 648640
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 5.969235478686768e-07,
+      "loss": 0.7355,
+      "step": 648650
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 5.957756179689294e-07,
+      "loss": 0.7599,
+      "step": 648660
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 5.946276880691818e-07,
+      "loss": 0.7019,
+      "step": 648670
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 5.934797581694344e-07,
+      "loss": 0.6551,
+      "step": 648680
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 5.923318282696869e-07,
+      "loss": 0.5765,
+      "step": 648690
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 5.911838983699395e-07,
+      "loss": 0.7006,
+      "step": 648700
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 5.90035968470192e-07,
+      "loss": 0.6816,
+      "step": 648710
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 5.888880385704446e-07,
+      "loss": 0.7592,
+      "step": 648720
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 5.877401086706971e-07,
+      "loss": 0.644,
+      "step": 648730
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 5.865921787709497e-07,
+      "loss": 0.6497,
+      "step": 648740
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 5.854442488712022e-07,
+      "loss": 0.6602,
+      "step": 648750
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 5.842963189714548e-07,
+      "loss": 0.623,
+      "step": 648760
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 5.831483890717073e-07,
+      "loss": 0.6625,
+      "step": 648770
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 5.820004591719598e-07,
+      "loss": 0.7024,
+      "step": 648780
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 5.808525292722124e-07,
+      "loss": 0.6244,
+      "step": 648790
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 5.797045993724649e-07,
+      "loss": 0.7448,
+      "step": 648800
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 5.785566694727175e-07,
+      "loss": 0.6773,
+      "step": 648810
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 5.7740873957297e-07,
+      "loss": 0.7215,
+      "step": 648820
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 5.762608096732226e-07,
+      "loss": 0.738,
+      "step": 648830
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 5.751128797734752e-07,
+      "loss": 0.704,
+      "step": 648840
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 5.739649498737277e-07,
+      "loss": 0.6561,
+      "step": 648850
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 5.728170199739803e-07,
+      "loss": 0.7087,
+      "step": 648860
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 5.716690900742328e-07,
+      "loss": 0.7102,
+      "step": 648870
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 5.705211601744853e-07,
+      "loss": 0.65,
+      "step": 648880
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 5.693732302747379e-07,
+      "loss": 0.6913,
+      "step": 648890
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 5.682253003749904e-07,
+      "loss": 0.6634,
+      "step": 648900
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 5.67077370475243e-07,
+      "loss": 0.6754,
+      "step": 648910
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 5.659294405754955e-07,
+      "loss": 0.7032,
+      "step": 648920
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 5.64781510675748e-07,
+      "loss": 0.6666,
+      "step": 648930
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 5.636335807760006e-07,
+      "loss": 0.6388,
+      "step": 648940
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 5.624856508762531e-07,
+      "loss": 0.6675,
+      "step": 648950
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 5.613377209765057e-07,
+      "loss": 0.7469,
+      "step": 648960
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 5.601897910767582e-07,
+      "loss": 0.6755,
+      "step": 648970
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 5.590418611770107e-07,
+      "loss": 0.7118,
+      "step": 648980
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 5.578939312772633e-07,
+      "loss": 0.6796,
+      "step": 648990
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 5.567460013775158e-07,
+      "loss": 0.5907,
+      "step": 649000
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 5.555980714777684e-07,
+      "loss": 0.6564,
+      "step": 649010
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 5.544501415780209e-07,
+      "loss": 0.69,
+      "step": 649020
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 5.533022116782734e-07,
+      "loss": 0.694,
+      "step": 649030
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 5.52154281778526e-07,
+      "loss": 0.6244,
+      "step": 649040
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 5.510063518787785e-07,
+      "loss": 0.6096,
+      "step": 649050
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 5.49858421979031e-07,
+      "loss": 0.6647,
+      "step": 649060
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 5.487104920792836e-07,
+      "loss": 0.724,
+      "step": 649070
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 5.475625621795361e-07,
+      "loss": 0.6295,
+      "step": 649080
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 5.464146322797887e-07,
+      "loss": 0.7335,
+      "step": 649090
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 5.452667023800412e-07,
+      "loss": 0.6692,
+      "step": 649100
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 5.441187724802938e-07,
+      "loss": 0.7541,
+      "step": 649110
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 5.429708425805464e-07,
+      "loss": 0.6977,
+      "step": 649120
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 5.418229126807989e-07,
+      "loss": 0.6969,
+      "step": 649130
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 5.406749827810515e-07,
+      "loss": 0.6512,
+      "step": 649140
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 5.39527052881304e-07,
+      "loss": 0.6973,
+      "step": 649150
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 5.383791229815565e-07,
+      "loss": 0.7224,
+      "step": 649160
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 5.372311930818091e-07,
+      "loss": 0.6529,
+      "step": 649170
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 5.360832631820616e-07,
+      "loss": 0.608,
+      "step": 649180
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 5.349353332823142e-07,
+      "loss": 0.7223,
+      "step": 649190
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 5.337874033825668e-07,
+      "loss": 0.7435,
+      "step": 649200
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 5.326394734828193e-07,
+      "loss": 0.761,
+      "step": 649210
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 5.314915435830718e-07,
+      "loss": 0.7288,
+      "step": 649220
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 5.303436136833244e-07,
+      "loss": 0.6869,
+      "step": 649230
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 5.291956837835769e-07,
+      "loss": 0.5899,
+      "step": 649240
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 5.280477538838295e-07,
+      "loss": 0.602,
+      "step": 649250
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 5.26899823984082e-07,
+      "loss": 0.6868,
+      "step": 649260
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 5.257518940843345e-07,
+      "loss": 0.7064,
+      "step": 649270
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 5.246039641845871e-07,
+      "loss": 0.749,
+      "step": 649280
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 5.234560342848396e-07,
+      "loss": 0.6571,
+      "step": 649290
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 5.223081043850922e-07,
+      "loss": 0.6859,
+      "step": 649300
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 5.211601744853447e-07,
+      "loss": 0.6688,
+      "step": 649310
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 5.200122445855972e-07,
+      "loss": 0.6236,
+      "step": 649320
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 5.188643146858498e-07,
+      "loss": 0.6308,
+      "step": 649330
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 5.177163847861023e-07,
+      "loss": 0.7607,
+      "step": 649340
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 5.165684548863549e-07,
+      "loss": 0.6488,
+      "step": 649350
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 5.154205249866074e-07,
+      "loss": 0.6343,
+      "step": 649360
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 5.1427259508686e-07,
+      "loss": 0.6699,
+      "step": 649370
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 5.131246651871126e-07,
+      "loss": 0.7081,
+      "step": 649380
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 5.119767352873651e-07,
+      "loss": 0.7093,
+      "step": 649390
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 5.108288053876177e-07,
+      "loss": 0.7494,
+      "step": 649400
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 5.096808754878702e-07,
+      "loss": 0.6752,
+      "step": 649410
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 5.085329455881227e-07,
+      "loss": 0.7711,
+      "step": 649420
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 5.073850156883753e-07,
+      "loss": 0.6742,
+      "step": 649430
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 5.062370857886278e-07,
+      "loss": 0.7527,
+      "step": 649440
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 5.050891558888804e-07,
+      "loss": 0.677,
+      "step": 649450
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 5.039412259891329e-07,
+      "loss": 0.7025,
+      "step": 649460
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 5.027932960893854e-07,
+      "loss": 0.6557,
+      "step": 649470
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 5.01645366189638e-07,
+      "loss": 0.6651,
+      "step": 649480
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 5.004974362898905e-07,
+      "loss": 0.6897,
+      "step": 649490
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 4.993495063901431e-07,
+      "loss": 0.667,
+      "step": 649500
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 4.982015764903956e-07,
+      "loss": 0.6273,
+      "step": 649510
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 4.970536465906481e-07,
+      "loss": 0.6575,
+      "step": 649520
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 4.959057166909007e-07,
+      "loss": 0.6877,
+      "step": 649530
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 4.947577867911532e-07,
+      "loss": 0.6673,
+      "step": 649540
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 4.936098568914058e-07,
+      "loss": 0.6941,
+      "step": 649550
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 4.924619269916583e-07,
+      "loss": 0.6999,
+      "step": 649560
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 4.913139970919108e-07,
+      "loss": 0.7344,
+      "step": 649570
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 4.901660671921634e-07,
+      "loss": 0.611,
+      "step": 649580
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 4.890181372924159e-07,
+      "loss": 0.6539,
+      "step": 649590
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 4.878702073926684e-07,
+      "loss": 0.6529,
+      "step": 649600
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 4.86722277492921e-07,
+      "loss": 0.6866,
+      "step": 649610
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 4.855743475931735e-07,
+      "loss": 0.6342,
+      "step": 649620
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 4.844264176934261e-07,
+      "loss": 0.6568,
+      "step": 649630
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 4.832784877936788e-07,
+      "loss": 0.6991,
+      "step": 649640
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 4.821305578939313e-07,
+      "loss": 0.7446,
+      "step": 649650
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 4.809826279941838e-07,
+      "loss": 0.6687,
+      "step": 649660
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 4.798346980944364e-07,
+      "loss": 0.6772,
+      "step": 649670
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 4.786867681946889e-07,
+      "loss": 0.7015,
+      "step": 649680
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 4.775388382949415e-07,
+      "loss": 0.7073,
+      "step": 649690
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 4.7639090839519397e-07,
+      "loss": 0.7091,
+      "step": 649700
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 4.7524297849544653e-07,
+      "loss": 0.6172,
+      "step": 649710
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 4.740950485956991e-07,
+      "loss": 0.6822,
+      "step": 649720
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 4.7294711869595164e-07,
+      "loss": 0.6165,
+      "step": 649730
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 4.7179918879620414e-07,
+      "loss": 0.6558,
+      "step": 649740
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 4.706512588964567e-07,
+      "loss": 0.6637,
+      "step": 649750
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 4.6950332899670925e-07,
+      "loss": 0.7124,
+      "step": 649760
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 4.683553990969618e-07,
+      "loss": 0.7001,
+      "step": 649770
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 4.672074691972143e-07,
+      "loss": 0.6858,
+      "step": 649780
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 4.6605953929746685e-07,
+      "loss": 0.6986,
+      "step": 649790
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 4.649116093977194e-07,
+      "loss": 0.6812,
+      "step": 649800
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 4.6376367949797196e-07,
+      "loss": 0.7223,
+      "step": 649810
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 4.626157495982245e-07,
+      "loss": 0.6747,
+      "step": 649820
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 4.61467819698477e-07,
+      "loss": 0.7217,
+      "step": 649830
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 4.6031988979872957e-07,
+      "loss": 0.6582,
+      "step": 649840
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 4.5917195989898213e-07,
+      "loss": 0.7161,
+      "step": 649850
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 4.580240299992347e-07,
+      "loss": 0.65,
+      "step": 649860
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 4.5687610009948724e-07,
+      "loss": 0.6929,
+      "step": 649870
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 4.5572817019973974e-07,
+      "loss": 0.6786,
+      "step": 649880
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 4.545802402999923e-07,
+      "loss": 0.669,
+      "step": 649890
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 4.5343231040024485e-07,
+      "loss": 0.6799,
+      "step": 649900
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 4.522843805004974e-07,
+      "loss": 0.6575,
+      "step": 649910
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 4.5113645060074995e-07,
+      "loss": 0.737,
+      "step": 649920
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 4.4998852070100245e-07,
+      "loss": 0.7382,
+      "step": 649930
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 4.48840590801255e-07,
+      "loss": 0.668,
+      "step": 649940
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 4.4769266090150756e-07,
+      "loss": 0.7135,
+      "step": 649950
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 4.465447310017601e-07,
+      "loss": 0.7777,
+      "step": 649960
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 4.4539680110201267e-07,
+      "loss": 0.6293,
+      "step": 649970
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 4.4424887120226517e-07,
+      "loss": 0.6923,
+      "step": 649980
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 4.4310094130251773e-07,
+      "loss": 0.6977,
+      "step": 649990
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 4.419530114027703e-07,
+      "loss": 0.7291,
+      "step": 650000
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 4.4080508150302284e-07,
+      "loss": 0.7319,
+      "step": 650010
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 4.3965715160327534e-07,
+      "loss": 0.6164,
+      "step": 650020
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 4.385092217035279e-07,
+      "loss": 0.7075,
+      "step": 650030
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 4.3736129180378045e-07,
+      "loss": 0.7292,
+      "step": 650040
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 4.36213361904033e-07,
+      "loss": 0.74,
+      "step": 650050
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 4.3506543200428555e-07,
+      "loss": 0.6476,
+      "step": 650060
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 4.3391750210453816e-07,
+      "loss": 0.71,
+      "step": 650070
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 4.3276957220479066e-07,
+      "loss": 0.7152,
+      "step": 650080
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 4.316216423050432e-07,
+      "loss": 0.7131,
+      "step": 650090
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 4.3047371240529577e-07,
+      "loss": 0.7383,
+      "step": 650100
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 4.293257825055483e-07,
+      "loss": 0.6179,
+      "step": 650110
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 4.281778526058009e-07,
+      "loss": 0.6948,
+      "step": 650120
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 4.270299227060534e-07,
+      "loss": 0.7607,
+      "step": 650130
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 4.2588199280630593e-07,
+      "loss": 0.6425,
+      "step": 650140
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 4.247340629065585e-07,
+      "loss": 0.7015,
+      "step": 650150
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 4.2358613300681104e-07,
+      "loss": 0.6615,
+      "step": 650160
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 4.224382031070636e-07,
+      "loss": 0.5979,
+      "step": 650170
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 4.212902732073161e-07,
+      "loss": 0.6519,
+      "step": 650180
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 4.2014234330756865e-07,
+      "loss": 0.6675,
+      "step": 650190
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 4.189944134078212e-07,
+      "loss": 0.7214,
+      "step": 650200
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 4.1784648350807376e-07,
+      "loss": 0.6895,
+      "step": 650210
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 4.166985536083263e-07,
+      "loss": 0.6393,
+      "step": 650220
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 4.155506237085788e-07,
+      "loss": 0.6722,
+      "step": 650230
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 4.1440269380883137e-07,
+      "loss": 0.7155,
+      "step": 650240
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 4.132547639090839e-07,
+      "loss": 0.7162,
+      "step": 650250
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 4.121068340093365e-07,
+      "loss": 0.6735,
+      "step": 650260
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 4.1095890410958903e-07,
+      "loss": 0.6708,
+      "step": 650270
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 4.0981097420984153e-07,
+      "loss": 0.7158,
+      "step": 650280
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 4.086630443100941e-07,
+      "loss": 0.752,
+      "step": 650290
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 4.0751511441034664e-07,
+      "loss": 0.6479,
+      "step": 650300
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 4.063671845105992e-07,
+      "loss": 0.5791,
+      "step": 650310
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 4.052192546108517e-07,
+      "loss": 0.752,
+      "step": 650320
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 4.0407132471110425e-07,
+      "loss": 0.631,
+      "step": 650330
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 4.029233948113568e-07,
+      "loss": 0.6353,
+      "step": 650340
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 4.0177546491160936e-07,
+      "loss": 0.6413,
+      "step": 650350
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 4.006275350118619e-07,
+      "loss": 0.7294,
+      "step": 650360
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 3.994796051121144e-07,
+      "loss": 0.688,
+      "step": 650370
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 3.9833167521236697e-07,
+      "loss": 0.7184,
+      "step": 650380
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 3.971837453126195e-07,
+      "loss": 0.6001,
+      "step": 650390
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 3.960358154128721e-07,
+      "loss": 0.6069,
+      "step": 650400
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 3.9488788551312463e-07,
+      "loss": 0.7364,
+      "step": 650410
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 3.9373995561337713e-07,
+      "loss": 0.7256,
+      "step": 650420
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 3.925920257136297e-07,
+      "loss": 0.7485,
+      "step": 650430
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 3.9144409581388224e-07,
+      "loss": 0.6383,
+      "step": 650440
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 3.902961659141348e-07,
+      "loss": 0.6893,
+      "step": 650450
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 3.8914823601438735e-07,
+      "loss": 0.6892,
+      "step": 650460
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 3.8800030611463985e-07,
+      "loss": 0.7157,
+      "step": 650470
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 3.868523762148924e-07,
+      "loss": 0.7392,
+      "step": 650480
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 3.8570444631514496e-07,
+      "loss": 0.6246,
+      "step": 650490
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 3.845565164153975e-07,
+      "loss": 0.6491,
+      "step": 650500
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 3.8340858651565007e-07,
+      "loss": 0.6473,
+      "step": 650510
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 3.822606566159027e-07,
+      "loss": 0.7257,
+      "step": 650520
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 3.811127267161552e-07,
+      "loss": 0.6242,
+      "step": 650530
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 3.7996479681640773e-07,
+      "loss": 0.5992,
+      "step": 650540
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 3.788168669166603e-07,
+      "loss": 0.6574,
+      "step": 650550
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 3.7766893701691284e-07,
+      "loss": 0.7773,
+      "step": 650560
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 3.765210071171654e-07,
+      "loss": 0.6614,
+      "step": 650570
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 3.753730772174179e-07,
+      "loss": 0.711,
+      "step": 650580
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 3.7422514731767045e-07,
+      "loss": 0.6471,
+      "step": 650590
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 3.73077217417923e-07,
+      "loss": 0.6608,
+      "step": 650600
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 3.7192928751817556e-07,
+      "loss": 0.6592,
+      "step": 650610
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 3.7078135761842806e-07,
+      "loss": 0.6417,
+      "step": 650620
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 3.696334277186806e-07,
+      "loss": 0.6324,
+      "step": 650630
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 3.6848549781893317e-07,
+      "loss": 0.7034,
+      "step": 650640
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 3.673375679191857e-07,
+      "loss": 0.6876,
+      "step": 650650
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 3.661896380194383e-07,
+      "loss": 0.7353,
+      "step": 650660
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 3.650417081196908e-07,
+      "loss": 0.6842,
+      "step": 650670
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 3.6389377821994333e-07,
+      "loss": 0.6828,
+      "step": 650680
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 3.627458483201959e-07,
+      "loss": 0.6952,
+      "step": 650690
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 3.6159791842044844e-07,
+      "loss": 0.7504,
+      "step": 650700
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 3.60449988520701e-07,
+      "loss": 0.7121,
+      "step": 650710
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 3.593020586209535e-07,
+      "loss": 0.6521,
+      "step": 650720
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 3.5815412872120605e-07,
+      "loss": 0.6552,
+      "step": 650730
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 3.570061988214586e-07,
+      "loss": 0.6413,
+      "step": 650740
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 3.5585826892171116e-07,
+      "loss": 0.6283,
+      "step": 650750
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 3.547103390219637e-07,
+      "loss": 0.6658,
+      "step": 650760
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 3.535624091222162e-07,
+      "loss": 0.7243,
+      "step": 650770
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 3.5241447922246877e-07,
+      "loss": 0.6852,
+      "step": 650780
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 3.512665493227213e-07,
+      "loss": 0.7613,
+      "step": 650790
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 3.501186194229739e-07,
+      "loss": 0.6723,
+      "step": 650800
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 3.4897068952322643e-07,
+      "loss": 0.7086,
+      "step": 650810
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 3.4782275962347893e-07,
+      "loss": 0.6335,
+      "step": 650820
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 3.466748297237315e-07,
+      "loss": 0.6378,
+      "step": 650830
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 3.4552689982398404e-07,
+      "loss": 0.672,
+      "step": 650840
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 3.443789699242366e-07,
+      "loss": 0.7184,
+      "step": 650850
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 3.4323104002448915e-07,
+      "loss": 0.7305,
+      "step": 650860
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 3.4208311012474165e-07,
+      "loss": 0.6416,
+      "step": 650870
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 3.409351802249942e-07,
+      "loss": 0.7587,
+      "step": 650880
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 3.3978725032524676e-07,
+      "loss": 0.7707,
+      "step": 650890
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 3.386393204254993e-07,
+      "loss": 0.6922,
+      "step": 650900
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 3.374913905257518e-07,
+      "loss": 0.6868,
+      "step": 650910
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 3.3634346062600437e-07,
+      "loss": 0.6552,
+      "step": 650920
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 3.351955307262569e-07,
+      "loss": 0.6857,
+      "step": 650930
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 3.340476008265095e-07,
+      "loss": 0.6787,
+      "step": 650940
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 3.3289967092676203e-07,
+      "loss": 0.6753,
+      "step": 650950
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 3.3175174102701464e-07,
+      "loss": 0.7278,
+      "step": 650960
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 3.3060381112726714e-07,
+      "loss": 0.6973,
+      "step": 650970
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 3.294558812275197e-07,
+      "loss": 0.7096,
+      "step": 650980
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 3.2830795132777225e-07,
+      "loss": 0.6749,
+      "step": 650990
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 3.271600214280248e-07,
+      "loss": 0.7096,
+      "step": 651000
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 3.2601209152827736e-07,
+      "loss": 0.677,
+      "step": 651010
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 3.2486416162852986e-07,
+      "loss": 0.6884,
+      "step": 651020
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 3.237162317287824e-07,
+      "loss": 0.6675,
+      "step": 651030
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 3.2256830182903497e-07,
+      "loss": 0.6468,
+      "step": 651040
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 3.214203719292875e-07,
+      "loss": 0.7063,
+      "step": 651050
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 3.202724420295401e-07,
+      "loss": 0.73,
+      "step": 651060
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 3.191245121297926e-07,
+      "loss": 0.7124,
+      "step": 651070
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 3.1797658223004513e-07,
+      "loss": 0.6274,
+      "step": 651080
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 3.168286523302977e-07,
+      "loss": 0.6299,
+      "step": 651090
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 3.1568072243055024e-07,
+      "loss": 0.6928,
+      "step": 651100
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 3.145327925308028e-07,
+      "loss": 0.6035,
+      "step": 651110
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 3.133848626310553e-07,
+      "loss": 0.7069,
+      "step": 651120
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 3.1223693273130785e-07,
+      "loss": 0.6834,
+      "step": 651130
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 3.110890028315604e-07,
+      "loss": 0.6813,
+      "step": 651140
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 3.0994107293181296e-07,
+      "loss": 0.6805,
+      "step": 651150
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 3.0879314303206546e-07,
+      "loss": 0.7463,
+      "step": 651160
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 3.07645213132318e-07,
+      "loss": 0.71,
+      "step": 651170
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 3.0649728323257057e-07,
+      "loss": 0.6452,
+      "step": 651180
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 3.053493533328231e-07,
+      "loss": 0.6824,
+      "step": 651190
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 3.042014234330757e-07,
+      "loss": 0.7215,
+      "step": 651200
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 3.030534935333282e-07,
+      "loss": 0.7829,
+      "step": 651210
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 3.0190556363358073e-07,
+      "loss": 0.6524,
+      "step": 651220
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 3.007576337338333e-07,
+      "loss": 0.6627,
+      "step": 651230
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 2.9960970383408584e-07,
+      "loss": 0.6728,
+      "step": 651240
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 2.984617739343384e-07,
+      "loss": 0.6582,
+      "step": 651250
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 2.973138440345909e-07,
+      "loss": 0.7766,
+      "step": 651260
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 2.9616591413484345e-07,
+      "loss": 0.6963,
+      "step": 651270
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 2.95017984235096e-07,
+      "loss": 0.6741,
+      "step": 651280
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 2.9387005433534856e-07,
+      "loss": 0.6346,
+      "step": 651290
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 2.927221244356011e-07,
+      "loss": 0.6661,
+      "step": 651300
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 2.9157419453585367e-07,
+      "loss": 0.7333,
+      "step": 651310
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 2.904262646361062e-07,
+      "loss": 0.6258,
+      "step": 651320
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 2.892783347363588e-07,
+      "loss": 0.7764,
+      "step": 651330
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 2.881304048366113e-07,
+      "loss": 0.7513,
+      "step": 651340
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 2.8698247493686383e-07,
+      "loss": 0.625,
+      "step": 651350
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 2.858345450371164e-07,
+      "loss": 0.6539,
+      "step": 651360
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 2.8468661513736894e-07,
+      "loss": 0.7236,
+      "step": 651370
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 2.835386852376215e-07,
+      "loss": 0.6411,
+      "step": 651380
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 2.82390755337874e-07,
+      "loss": 0.6887,
+      "step": 651390
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 2.8124282543812655e-07,
+      "loss": 0.6697,
+      "step": 651400
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 2.800948955383791e-07,
+      "loss": 0.6069,
+      "step": 651410
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 2.7894696563863166e-07,
+      "loss": 0.6377,
+      "step": 651420
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 2.777990357388842e-07,
+      "loss": 0.79,
+      "step": 651430
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 2.766511058391367e-07,
+      "loss": 0.6536,
+      "step": 651440
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 2.7550317593938927e-07,
+      "loss": 0.6089,
+      "step": 651450
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 2.743552460396418e-07,
+      "loss": 0.6364,
+      "step": 651460
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 2.732073161398944e-07,
+      "loss": 0.6183,
+      "step": 651470
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 2.720593862401469e-07,
+      "loss": 0.6859,
+      "step": 651480
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 2.7091145634039943e-07,
+      "loss": 0.6445,
+      "step": 651490
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 2.69763526440652e-07,
+      "loss": 0.7051,
+      "step": 651500
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 2.6861559654090454e-07,
+      "loss": 0.6283,
+      "step": 651510
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 2.674676666411571e-07,
+      "loss": 0.7263,
+      "step": 651520
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 2.6631973674140965e-07,
+      "loss": 0.642,
+      "step": 651530
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 2.651718068416622e-07,
+      "loss": 0.6829,
+      "step": 651540
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 2.6402387694191475e-07,
+      "loss": 0.6888,
+      "step": 651550
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 2.6287594704216726e-07,
+      "loss": 0.7098,
+      "step": 651560
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 2.617280171424198e-07,
+      "loss": 0.6913,
+      "step": 651570
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 2.6058008724267236e-07,
+      "loss": 0.7251,
+      "step": 651580
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 2.594321573429249e-07,
+      "loss": 0.6821,
+      "step": 651590
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 2.5828422744317747e-07,
+      "loss": 0.6447,
+      "step": 651600
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 2.5713629754343e-07,
+      "loss": 0.6595,
+      "step": 651610
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 2.5598836764368253e-07,
+      "loss": 0.6899,
+      "step": 651620
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 2.548404377439351e-07,
+      "loss": 0.7551,
+      "step": 651630
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 2.5369250784418764e-07,
+      "loss": 0.6832,
+      "step": 651640
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 2.525445779444402e-07,
+      "loss": 0.7423,
+      "step": 651650
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 2.513966480446927e-07,
+      "loss": 0.6676,
+      "step": 651660
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 2.5024871814494525e-07,
+      "loss": 0.6948,
+      "step": 651670
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 2.491007882451978e-07,
+      "loss": 0.6576,
+      "step": 651680
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 2.4795285834545035e-07,
+      "loss": 0.6448,
+      "step": 651690
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 2.468049284457029e-07,
+      "loss": 0.6539,
+      "step": 651700
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 2.456569985459554e-07,
+      "loss": 0.6772,
+      "step": 651710
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 2.4450906864620796e-07,
+      "loss": 0.6743,
+      "step": 651720
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 2.433611387464605e-07,
+      "loss": 0.6267,
+      "step": 651730
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 2.4221320884671307e-07,
+      "loss": 0.6959,
+      "step": 651740
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 2.4106527894696563e-07,
+      "loss": 0.738,
+      "step": 651750
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 2.399173490472182e-07,
+      "loss": 0.7734,
+      "step": 651760
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 2.3876941914747074e-07,
+      "loss": 0.6766,
+      "step": 651770
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 2.3762148924772326e-07,
+      "loss": 0.6766,
+      "step": 651780
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 2.3647355934797582e-07,
+      "loss": 0.7182,
+      "step": 651790
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 2.3532562944822835e-07,
+      "loss": 0.741,
+      "step": 651800
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 2.341776995484809e-07,
+      "loss": 0.6209,
+      "step": 651810
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 2.3302976964873343e-07,
+      "loss": 0.6969,
+      "step": 651820
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 2.3188183974898598e-07,
+      "loss": 0.674,
+      "step": 651830
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 2.307339098492385e-07,
+      "loss": 0.7212,
+      "step": 651840
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 2.2958597994949106e-07,
+      "loss": 0.7125,
+      "step": 651850
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 2.2843805004974362e-07,
+      "loss": 0.7223,
+      "step": 651860
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 2.2729012014999615e-07,
+      "loss": 0.7289,
+      "step": 651870
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 2.261421902502487e-07,
+      "loss": 0.6489,
+      "step": 651880
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 2.2499426035050123e-07,
+      "loss": 0.743,
+      "step": 651890
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 2.2384633045075378e-07,
+      "loss": 0.6129,
+      "step": 651900
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 2.2269840055100634e-07,
+      "loss": 0.6596,
+      "step": 651910
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 2.2155047065125886e-07,
+      "loss": 0.6832,
+      "step": 651920
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 2.2040254075151142e-07,
+      "loss": 0.7235,
+      "step": 651930
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 2.1925461085176395e-07,
+      "loss": 0.7359,
+      "step": 651940
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 2.181066809520165e-07,
+      "loss": 0.6673,
+      "step": 651950
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 2.1695875105226908e-07,
+      "loss": 0.6612,
+      "step": 651960
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 2.158108211525216e-07,
+      "loss": 0.6886,
+      "step": 651970
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 2.1466289125277416e-07,
+      "loss": 0.6577,
+      "step": 651980
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 2.135149613530267e-07,
+      "loss": 0.6821,
+      "step": 651990
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 2.1236703145327924e-07,
+      "loss": 0.6264,
+      "step": 652000
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 2.112191015535318e-07,
+      "loss": 0.698,
+      "step": 652010
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 2.1007117165378433e-07,
+      "loss": 0.5996,
+      "step": 652020
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 2.0892324175403688e-07,
+      "loss": 0.6683,
+      "step": 652030
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 2.077753118542894e-07,
+      "loss": 0.6498,
+      "step": 652040
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 2.0662738195454196e-07,
+      "loss": 0.7087,
+      "step": 652050
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 2.0547945205479452e-07,
+      "loss": 0.6589,
+      "step": 652060
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 2.0433152215504704e-07,
+      "loss": 0.68,
+      "step": 652070
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 2.031835922552996e-07,
+      "loss": 0.6785,
+      "step": 652080
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 2.0203566235555213e-07,
+      "loss": 0.7436,
+      "step": 652090
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 2.0088773245580468e-07,
+      "loss": 0.6642,
+      "step": 652100
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 1.997398025560572e-07,
+      "loss": 0.6843,
+      "step": 652110
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 1.9859187265630976e-07,
+      "loss": 0.749,
+      "step": 652120
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 1.9744394275656232e-07,
+      "loss": 0.6748,
+      "step": 652130
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 1.9629601285681484e-07,
+      "loss": 0.7222,
+      "step": 652140
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 1.951480829570674e-07,
+      "loss": 0.6026,
+      "step": 652150
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 1.9400015305731993e-07,
+      "loss": 0.7487,
+      "step": 652160
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 1.9285222315757248e-07,
+      "loss": 0.7524,
+      "step": 652170
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 1.9170429325782503e-07,
+      "loss": 0.7151,
+      "step": 652180
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 1.905563633580776e-07,
+      "loss": 0.6578,
+      "step": 652190
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 1.8940843345833014e-07,
+      "loss": 0.6471,
+      "step": 652200
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 1.882605035585827e-07,
+      "loss": 0.582,
+      "step": 652210
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 1.8711257365883523e-07,
+      "loss": 0.6209,
+      "step": 652220
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 1.8596464375908778e-07,
+      "loss": 0.7089,
+      "step": 652230
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 1.848167138593403e-07,
+      "loss": 0.6788,
+      "step": 652240
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 1.8366878395959286e-07,
+      "loss": 0.7114,
+      "step": 652250
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 1.825208540598454e-07,
+      "loss": 0.5941,
+      "step": 652260
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 1.8137292416009794e-07,
+      "loss": 0.6919,
+      "step": 652270
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 1.802249942603505e-07,
+      "loss": 0.6458,
+      "step": 652280
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 1.7907706436060303e-07,
+      "loss": 0.6714,
+      "step": 652290
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 1.7792913446085558e-07,
+      "loss": 0.6052,
+      "step": 652300
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 1.767812045611081e-07,
+      "loss": 0.6337,
+      "step": 652310
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 1.7563327466136066e-07,
+      "loss": 0.7229,
+      "step": 652320
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 1.7448534476161322e-07,
+      "loss": 0.6241,
+      "step": 652330
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 1.7333741486186574e-07,
+      "loss": 0.703,
+      "step": 652340
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 1.721894849621183e-07,
+      "loss": 0.7633,
+      "step": 652350
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 1.7104155506237083e-07,
+      "loss": 0.7002,
+      "step": 652360
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 1.6989362516262338e-07,
+      "loss": 0.6744,
+      "step": 652370
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 1.687456952628759e-07,
+      "loss": 0.7012,
+      "step": 652380
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 1.6759776536312846e-07,
+      "loss": 0.6807,
+      "step": 652390
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 1.6644983546338102e-07,
+      "loss": 0.696,
+      "step": 652400
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 1.6530190556363357e-07,
+      "loss": 0.6357,
+      "step": 652410
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 1.6415397566388612e-07,
+      "loss": 0.7296,
+      "step": 652420
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 1.6300604576413868e-07,
+      "loss": 0.6471,
+      "step": 652430
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 1.618581158643912e-07,
+      "loss": 0.7089,
+      "step": 652440
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 1.6071018596464376e-07,
+      "loss": 0.6982,
+      "step": 652450
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 1.595622560648963e-07,
+      "loss": 0.674,
+      "step": 652460
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 1.5841432616514884e-07,
+      "loss": 0.6238,
+      "step": 652470
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 1.572663962654014e-07,
+      "loss": 0.7189,
+      "step": 652480
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 1.5611846636565392e-07,
+      "loss": 0.6512,
+      "step": 652490
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 1.5497053646590648e-07,
+      "loss": 0.6083,
+      "step": 652500
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 1.53822606566159e-07,
+      "loss": 0.7096,
+      "step": 652510
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 1.5267467666641156e-07,
+      "loss": 0.6573,
+      "step": 652520
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 1.515267467666641e-07,
+      "loss": 0.7408,
+      "step": 652530
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 1.5037881686691664e-07,
+      "loss": 0.6335,
+      "step": 652540
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 1.492308869671692e-07,
+      "loss": 0.6788,
+      "step": 652550
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 1.4808295706742172e-07,
+      "loss": 0.7095,
+      "step": 652560
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 1.4693502716767428e-07,
+      "loss": 0.6494,
+      "step": 652570
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 1.4578709726792683e-07,
+      "loss": 0.6785,
+      "step": 652580
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 1.446391673681794e-07,
+      "loss": 0.6825,
+      "step": 652590
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 1.4349123746843191e-07,
+      "loss": 0.6482,
+      "step": 652600
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 1.4234330756868447e-07,
+      "loss": 0.7108,
+      "step": 652610
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 1.41195377668937e-07,
+      "loss": 0.7668,
+      "step": 652620
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 1.4004744776918955e-07,
+      "loss": 0.6794,
+      "step": 652630
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 1.388995178694421e-07,
+      "loss": 0.6685,
+      "step": 652640
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 1.3775158796969463e-07,
+      "loss": 0.6409,
+      "step": 652650
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 1.366036580699472e-07,
+      "loss": 0.7042,
+      "step": 652660
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 1.3545572817019971e-07,
+      "loss": 0.6564,
+      "step": 652670
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 1.3430779827045227e-07,
+      "loss": 0.668,
+      "step": 652680
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 1.3315986837070482e-07,
+      "loss": 0.6494,
+      "step": 652690
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 1.3201193847095738e-07,
+      "loss": 0.7334,
+      "step": 652700
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 1.308640085712099e-07,
+      "loss": 0.7281,
+      "step": 652710
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 1.2971607867146246e-07,
+      "loss": 0.7697,
+      "step": 652720
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 1.28568148771715e-07,
+      "loss": 0.7439,
+      "step": 652730
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 1.2742021887196754e-07,
+      "loss": 0.7265,
+      "step": 652740
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 1.262722889722201e-07,
+      "loss": 0.6919,
+      "step": 652750
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 1.2512435907247262e-07,
+      "loss": 0.7483,
+      "step": 652760
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 1.2397642917272518e-07,
+      "loss": 0.7166,
+      "step": 652770
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 1.228284992729777e-07,
+      "loss": 0.6791,
+      "step": 652780
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 1.2168056937323026e-07,
+      "loss": 0.6736,
+      "step": 652790
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 1.2053263947348281e-07,
+      "loss": 0.6704,
+      "step": 652800
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 1.1938470957373537e-07,
+      "loss": 0.7162,
+      "step": 652810
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 1.1823677967398791e-07,
+      "loss": 0.6657,
+      "step": 652820
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 1.1708884977424045e-07,
+      "loss": 0.7043,
+      "step": 652830
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 1.1594091987449299e-07,
+      "loss": 0.7032,
+      "step": 652840
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 1.1479298997474553e-07,
+      "loss": 0.6924,
+      "step": 652850
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 1.1364506007499807e-07,
+      "loss": 0.7246,
+      "step": 652860
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 1.1249713017525061e-07,
+      "loss": 0.6942,
+      "step": 652870
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 1.1134920027550317e-07,
+      "loss": 0.6624,
+      "step": 652880
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 1.1020127037575571e-07,
+      "loss": 0.7109,
+      "step": 652890
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 1.0905334047600825e-07,
+      "loss": 0.7064,
+      "step": 652900
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 1.079054105762608e-07,
+      "loss": 0.6884,
+      "step": 652910
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 1.0675748067651335e-07,
+      "loss": 0.7325,
+      "step": 652920
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 1.056095507767659e-07,
+      "loss": 0.7424,
+      "step": 652930
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 1.0446162087701844e-07,
+      "loss": 0.6828,
+      "step": 652940
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 1.0331369097727098e-07,
+      "loss": 0.6826,
+      "step": 652950
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 1.0216576107752352e-07,
+      "loss": 0.6567,
+      "step": 652960
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 1.0101783117777606e-07,
+      "loss": 0.7232,
+      "step": 652970
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 9.98699012780286e-08,
+      "loss": 0.6767,
+      "step": 652980
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 9.872197137828116e-08,
+      "loss": 0.6886,
+      "step": 652990
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 9.75740414785337e-08,
+      "loss": 0.638,
+      "step": 653000
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 9.642611157878624e-08,
+      "loss": 0.6775,
+      "step": 653010
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 9.52781816790388e-08,
+      "loss": 0.7008,
+      "step": 653020
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 9.413025177929135e-08,
+      "loss": 0.6272,
+      "step": 653030
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 9.298232187954389e-08,
+      "loss": 0.6892,
+      "step": 653040
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 9.183439197979643e-08,
+      "loss": 0.6633,
+      "step": 653050
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 9.068646208004897e-08,
+      "loss": 0.6922,
+      "step": 653060
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 8.953853218030151e-08,
+      "loss": 0.6993,
+      "step": 653070
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 8.839060228055405e-08,
+      "loss": 0.6454,
+      "step": 653080
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 8.724267238080661e-08,
+      "loss": 0.7578,
+      "step": 653090
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 8.609474248105915e-08,
+      "loss": 0.6355,
+      "step": 653100
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 8.494681258131169e-08,
+      "loss": 0.7066,
+      "step": 653110
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 8.379888268156423e-08,
+      "loss": 0.6652,
+      "step": 653120
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 8.265095278181678e-08,
+      "loss": 0.6954,
+      "step": 653130
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 8.150302288206934e-08,
+      "loss": 0.7123,
+      "step": 653140
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 8.035509298232188e-08,
+      "loss": 0.6797,
+      "step": 653150
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 7.920716308257442e-08,
+      "loss": 0.6463,
+      "step": 653160
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 7.805923318282696e-08,
+      "loss": 0.785,
+      "step": 653170
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 7.69113032830795e-08,
+      "loss": 0.7283,
+      "step": 653180
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 7.576337338333204e-08,
+      "loss": 0.7485,
+      "step": 653190
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 7.46154434835846e-08,
+      "loss": 0.7008,
+      "step": 653200
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 7.346751358383714e-08,
+      "loss": 0.6524,
+      "step": 653210
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 7.23195836840897e-08,
+      "loss": 0.5988,
+      "step": 653220
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 7.117165378434223e-08,
+      "loss": 0.6933,
+      "step": 653230
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 7.002372388459478e-08,
+      "loss": 0.7053,
+      "step": 653240
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 6.887579398484732e-08,
+      "loss": 0.6735,
+      "step": 653250
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 6.772786408509986e-08,
+      "loss": 0.6946,
+      "step": 653260
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 6.657993418535241e-08,
+      "loss": 0.6776,
+      "step": 653270
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 6.543200428560495e-08,
+      "loss": 0.7093,
+      "step": 653280
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 6.42840743858575e-08,
+      "loss": 0.7018,
+      "step": 653290
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 6.313614448611005e-08,
+      "loss": 0.6737,
+      "step": 653300
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 6.198821458636259e-08,
+      "loss": 0.6894,
+      "step": 653310
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 6.084028468661513e-08,
+      "loss": 0.5847,
+      "step": 653320
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 5.969235478686768e-08,
+      "loss": 0.6464,
+      "step": 653330
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 5.8544424887120225e-08,
+      "loss": 0.7501,
+      "step": 653340
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 5.7396494987372766e-08,
+      "loss": 0.6856,
+      "step": 653350
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 5.624856508762531e-08,
+      "loss": 0.6619,
+      "step": 653360
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 5.5100635187877854e-08,
+      "loss": 0.6575,
+      "step": 653370
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 5.39527052881304e-08,
+      "loss": 0.5954,
+      "step": 653380
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 5.280477538838295e-08,
+      "loss": 0.6871,
+      "step": 653390
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 5.165684548863549e-08,
+      "loss": 0.6596,
+      "step": 653400
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 5.050891558888803e-08,
+      "loss": 0.6962,
+      "step": 653410
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 4.936098568914058e-08,
+      "loss": 0.6327,
+      "step": 653420
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 4.821305578939312e-08,
+      "loss": 0.7225,
+      "step": 653430
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 4.7065125889645674e-08,
+      "loss": 0.6134,
+      "step": 653440
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 4.5917195989898215e-08,
+      "loss": 0.7605,
+      "step": 653450
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 4.4769266090150756e-08,
+      "loss": 0.6531,
+      "step": 653460
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 4.3621336190403304e-08,
+      "loss": 0.6854,
+      "step": 653470
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 4.2473406290655845e-08,
+      "loss": 0.7148,
+      "step": 653480
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 4.132547639090839e-08,
+      "loss": 0.6486,
+      "step": 653490
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 4.017754649116094e-08,
+      "loss": 0.7072,
+      "step": 653500
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 3.902961659141348e-08,
+      "loss": 0.6071,
+      "step": 653510
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 3.788168669166602e-08,
+      "loss": 0.7078,
+      "step": 653520
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 3.673375679191857e-08,
+      "loss": 0.6146,
+      "step": 653530
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 3.558582689217112e-08,
+      "loss": 0.7221,
+      "step": 653540
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 3.443789699242366e-08,
+      "loss": 0.7192,
+      "step": 653550
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 3.3289967092676206e-08,
+      "loss": 0.6829,
+      "step": 653560
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 3.214203719292875e-08,
+      "loss": 0.6503,
+      "step": 653570
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 3.0994107293181294e-08,
+      "loss": 0.6731,
+      "step": 653580
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 2.984617739343384e-08,
+      "loss": 0.7112,
+      "step": 653590
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 2.8698247493686383e-08,
+      "loss": 0.6741,
+      "step": 653600
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 2.7550317593938927e-08,
+      "loss": 0.7036,
+      "step": 653610
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 2.6402387694191475e-08,
+      "loss": 0.7093,
+      "step": 653620
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 2.5254457794444016e-08,
+      "loss": 0.6504,
+      "step": 653630
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 2.410652789469656e-08,
+      "loss": 0.7715,
+      "step": 653640
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 2.2958597994949108e-08,
+      "loss": 0.7108,
+      "step": 653650
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 2.1810668095201652e-08,
+      "loss": 0.6637,
+      "step": 653660
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 2.0662738195454196e-08,
+      "loss": 0.6375,
+      "step": 653670
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 1.951480829570674e-08,
+      "loss": 0.7637,
+      "step": 653680
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 1.8366878395959285e-08,
+      "loss": 0.7009,
+      "step": 653690
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 1.721894849621183e-08,
+      "loss": 0.6336,
+      "step": 653700
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 1.6071018596464373e-08,
+      "loss": 0.6524,
+      "step": 653710
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 1.492308869671692e-08,
+      "loss": 0.7522,
+      "step": 653720
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 1.3775158796969464e-08,
+      "loss": 0.7113,
+      "step": 653730
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 1.2627228897222008e-08,
+      "loss": 0.6302,
+      "step": 653740
+    },
+    {
+      "epoch": 29.99,
+      "learning_rate": 1.1479298997474554e-08,
+      "loss": 0.6697,
+      "step": 653750
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 1.0331369097727098e-08,
+      "loss": 0.6567,
+      "step": 653760
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 9.183439197979642e-09,
+      "loss": 0.6753,
+      "step": 653770
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 8.035509298232187e-09,
+      "loss": 0.6838,
+      "step": 653780
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 6.887579398484732e-09,
+      "loss": 0.7249,
+      "step": 653790
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 5.739649498737277e-09,
+      "loss": 0.6522,
+      "step": 653800
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 4.591719598989821e-09,
+      "loss": 0.713,
+      "step": 653810
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 3.443789699242366e-09,
+      "loss": 0.8018,
+      "step": 653820
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 2.2958597994949106e-09,
+      "loss": 0.6048,
+      "step": 653830
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 1.1479298997474553e-09,
+      "loss": 0.7216,
+      "step": 653840
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 0.0,
+      "loss": 0.6503,
+      "step": 653850
+    },
+    {
+      "epoch": 30.0,
+      "eval_cer": 0.6689201598234296,
+      "eval_em": 0.007454017424975799,
+      "eval_f1": 0.007454017424975799,
+      "eval_loss": 1.3820773363113403,
+      "eval_runtime": 1315.155,
+      "eval_samples_per_second": 7.855,
+      "eval_steps_per_second": 0.982,
+      "eval_wer": 0.9925459825750242,
+      "step": 653850
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 653850,
+  "num_train_epochs": 30,
+  "save_steps": 500,
+  "total_flos": 5.821301102716957e+18,
+  "trial_name": null,
+  "trial_params": null
+}