diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,30025 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.4222323358743885,
+  "global_step": 50000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999e-05,
+      "loss": 4.3259,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9980000000000006e-05,
+      "loss": 4.2788,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.997e-05,
+      "loss": 4.2004,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996e-05,
+      "loss": 4.1964,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.995e-05,
+      "loss": 4.2148,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9940000000000006e-05,
+      "loss": 4.0366,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9930000000000005e-05,
+      "loss": 4.1655,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.992e-05,
+      "loss": 3.924,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.991e-05,
+      "loss": 3.9823,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99e-05,
+      "loss": 3.7808,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9890000000000005e-05,
+      "loss": 3.8206,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9880000000000004e-05,
+      "loss": 3.9016,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.987e-05,
+      "loss": 4.2323,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.986e-05,
+      "loss": 4.0438,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9850000000000006e-05,
+      "loss": 3.9466,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9840000000000004e-05,
+      "loss": 4.1152,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.983e-05,
+      "loss": 3.9074,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.982e-05,
+      "loss": 3.8318,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.981e-05,
+      "loss": 3.6682,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 3.9341,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.979e-05,
+      "loss": 3.9676,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.978e-05,
+      "loss": 3.9058,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.977e-05,
+      "loss": 3.7062,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.976e-05,
+      "loss": 3.7872,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.975e-05,
+      "loss": 3.7812,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.974e-05,
+      "loss": 3.7666,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.973000000000001e-05,
+      "loss": 3.7976,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.972e-05,
+      "loss": 4.0788,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9710000000000003e-05,
+      "loss": 3.8427,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.97e-05,
+      "loss": 3.5896,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.969e-05,
+      "loss": 4.0309,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9680000000000005e-05,
+      "loss": 3.7941,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.967e-05,
+      "loss": 3.7299,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.966e-05,
+      "loss": 3.618,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.965e-05,
+      "loss": 3.8804,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9640000000000006e-05,
+      "loss": 3.7064,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9630000000000004e-05,
+      "loss": 3.7241,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.962e-05,
+      "loss": 3.9084,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.961e-05,
+      "loss": 3.8277,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.96e-05,
+      "loss": 4.06,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9590000000000005e-05,
+      "loss": 3.7065,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.958e-05,
+      "loss": 3.9011,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.957e-05,
+      "loss": 3.7923,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.956e-05,
+      "loss": 3.8609,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9550000000000005e-05,
+      "loss": 3.5528,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9540000000000003e-05,
+      "loss": 3.8945,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.953e-05,
+      "loss": 3.6211,
+      "step": 470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.952e-05,
+      "loss": 3.8777,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.951e-05,
+      "loss": 4.0037,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9500000000000004e-05,
+      "loss": 3.8309,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.949e-05,
+      "loss": 3.5665,
+      "step": 510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.948000000000001e-05,
+      "loss": 4.1132,
+      "step": 520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.947e-05,
+      "loss": 3.7971,
+      "step": 530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.946e-05,
+      "loss": 3.6497,
+      "step": 540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.945e-05,
+      "loss": 3.8993,
+      "step": 550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.944e-05,
+      "loss": 3.9588,
+      "step": 560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9430000000000006e-05,
+      "loss": 3.7,
+      "step": 570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.942e-05,
+      "loss": 3.9216,
+      "step": 580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.941e-05,
+      "loss": 3.946,
+      "step": 590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.94e-05,
+      "loss": 4.1132,
+      "step": 600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.939e-05,
+      "loss": 3.867,
+      "step": 610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9380000000000005e-05,
+      "loss": 3.7977,
+      "step": 620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.937e-05,
+      "loss": 4.0146,
+      "step": 630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.936e-05,
+      "loss": 3.7132,
+      "step": 640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.935e-05,
+      "loss": 3.7761,
+      "step": 650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9340000000000005e-05,
+      "loss": 3.9205,
+      "step": 660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9330000000000004e-05,
+      "loss": 3.753,
+      "step": 670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.932e-05,
+      "loss": 3.6907,
+      "step": 680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.931e-05,
+      "loss": 3.7411,
+      "step": 690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.93e-05,
+      "loss": 4.011,
+      "step": 700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9290000000000004e-05,
+      "loss": 3.7086,
+      "step": 710
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.928e-05,
+      "loss": 3.729,
+      "step": 720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.927000000000001e-05,
+      "loss": 3.9183,
+      "step": 730
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.926e-05,
+      "loss": 3.9488,
+      "step": 740
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9250000000000004e-05,
+      "loss": 3.7448,
+      "step": 750
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.924e-05,
+      "loss": 3.754,
+      "step": 760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.923e-05,
+      "loss": 3.6775,
+      "step": 770
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9220000000000006e-05,
+      "loss": 3.455,
+      "step": 780
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.921e-05,
+      "loss": 3.6346,
+      "step": 790
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.92e-05,
+      "loss": 3.6869,
+      "step": 800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.919e-05,
+      "loss": 3.9867,
+      "step": 810
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.918000000000001e-05,
+      "loss": 3.5628,
+      "step": 820
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9170000000000005e-05,
+      "loss": 3.9801,
+      "step": 830
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9160000000000004e-05,
+      "loss": 3.8769,
+      "step": 840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.915e-05,
+      "loss": 3.7076,
+      "step": 850
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.914e-05,
+      "loss": 3.752,
+      "step": 860
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9130000000000006e-05,
+      "loss": 3.7699,
+      "step": 870
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9120000000000004e-05,
+      "loss": 3.7779,
+      "step": 880
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.911e-05,
+      "loss": 3.9206,
+      "step": 890
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.91e-05,
+      "loss": 3.8448,
+      "step": 900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9090000000000006e-05,
+      "loss": 3.737,
+      "step": 910
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9080000000000004e-05,
+      "loss": 3.9557,
+      "step": 920
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.907e-05,
+      "loss": 3.6594,
+      "step": 930
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.906e-05,
+      "loss": 3.7355,
+      "step": 940
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.905e-05,
+      "loss": 3.7564,
+      "step": 950
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9040000000000005e-05,
+      "loss": 3.7111,
+      "step": 960
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.903e-05,
+      "loss": 3.5212,
+      "step": 970
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.902e-05,
+      "loss": 3.5224,
+      "step": 980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.901e-05,
+      "loss": 3.6404,
+      "step": 990
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9e-05,
+      "loss": 3.8373,
+      "step": 1000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8990000000000004e-05,
+      "loss": 3.5095,
+      "step": 1010
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.898e-05,
+      "loss": 3.5473,
+      "step": 1020
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.897000000000001e-05,
+      "loss": 3.4195,
+      "step": 1030
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.896e-05,
+      "loss": 3.7702,
+      "step": 1040
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8950000000000004e-05,
+      "loss": 3.7989,
+      "step": 1050
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.894e-05,
+      "loss": 3.8276,
+      "step": 1060
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.893e-05,
+      "loss": 3.5593,
+      "step": 1070
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8920000000000006e-05,
+      "loss": 3.644,
+      "step": 1080
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.891e-05,
+      "loss": 3.4157,
+      "step": 1090
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.89e-05,
+      "loss": 3.6898,
+      "step": 1100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.889e-05,
+      "loss": 3.7184,
+      "step": 1110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8880000000000006e-05,
+      "loss": 3.6377,
+      "step": 1120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8870000000000005e-05,
+      "loss": 3.7491,
+      "step": 1130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.886e-05,
+      "loss": 3.8688,
+      "step": 1140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.885e-05,
+      "loss": 3.6841,
+      "step": 1150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.884e-05,
+      "loss": 3.7557,
+      "step": 1160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8830000000000005e-05,
+      "loss": 3.6131,
+      "step": 1170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8820000000000004e-05,
+      "loss": 3.8026,
+      "step": 1180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.881e-05,
+      "loss": 3.6739,
+      "step": 1190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.88e-05,
+      "loss": 3.8314,
+      "step": 1200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8790000000000006e-05,
+      "loss": 3.7403,
+      "step": 1210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8780000000000004e-05,
+      "loss": 3.4544,
+      "step": 1220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.877e-05,
+      "loss": 3.6875,
+      "step": 1230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.876e-05,
+      "loss": 3.7621,
+      "step": 1240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.875e-05,
+      "loss": 3.6873,
+      "step": 1250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8740000000000004e-05,
+      "loss": 3.8111,
+      "step": 1260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.873e-05,
+      "loss": 3.7419,
+      "step": 1270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.872000000000001e-05,
+      "loss": 3.8183,
+      "step": 1280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.871e-05,
+      "loss": 3.7553,
+      "step": 1290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.87e-05,
+      "loss": 3.796,
+      "step": 1300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.869e-05,
+      "loss": 3.8271,
+      "step": 1310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.868e-05,
+      "loss": 3.639,
+      "step": 1320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.867000000000001e-05,
+      "loss": 3.9647,
+      "step": 1330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.866e-05,
+      "loss": 3.6117,
+      "step": 1340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8650000000000003e-05,
+      "loss": 3.3536,
+      "step": 1350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.864e-05,
+      "loss": 3.5784,
+      "step": 1360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.863e-05,
+      "loss": 3.7616,
+      "step": 1370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8620000000000005e-05,
+      "loss": 3.6273,
+      "step": 1380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.861e-05,
+      "loss": 3.8242,
+      "step": 1390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.86e-05,
+      "loss": 3.7375,
+      "step": 1400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.859e-05,
+      "loss": 3.5997,
+      "step": 1410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8580000000000006e-05,
+      "loss": 3.6069,
+      "step": 1420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8570000000000004e-05,
+      "loss": 3.6156,
+      "step": 1430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.856e-05,
+      "loss": 3.7375,
+      "step": 1440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.855e-05,
+      "loss": 3.768,
+      "step": 1450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.854e-05,
+      "loss": 3.7438,
+      "step": 1460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8530000000000005e-05,
+      "loss": 3.9872,
+      "step": 1470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.852e-05,
+      "loss": 3.6242,
+      "step": 1480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.851e-05,
+      "loss": 3.7022,
+      "step": 1490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.85e-05,
+      "loss": 3.8314,
+      "step": 1500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8490000000000005e-05,
+      "loss": 3.7473,
+      "step": 1510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8480000000000003e-05,
+      "loss": 3.5398,
+      "step": 1520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.847e-05,
+      "loss": 3.9378,
+      "step": 1530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.846e-05,
+      "loss": 3.6402,
+      "step": 1540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.845e-05,
+      "loss": 3.5014,
+      "step": 1550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8440000000000004e-05,
+      "loss": 3.7056,
+      "step": 1560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.843e-05,
+      "loss": 3.7061,
+      "step": 1570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.842000000000001e-05,
+      "loss": 3.8977,
+      "step": 1580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.841e-05,
+      "loss": 3.345,
+      "step": 1590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8400000000000004e-05,
+      "loss": 3.9044,
+      "step": 1600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.839e-05,
+      "loss": 3.6748,
+      "step": 1610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.838e-05,
+      "loss": 3.6044,
+      "step": 1620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8370000000000006e-05,
+      "loss": 3.4527,
+      "step": 1630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.836e-05,
+      "loss": 3.5471,
+      "step": 1640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.835e-05,
+      "loss": 3.5705,
+      "step": 1650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.834e-05,
+      "loss": 3.7876,
+      "step": 1660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.833e-05,
+      "loss": 3.4821,
+      "step": 1670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8320000000000005e-05,
+      "loss": 3.4002,
+      "step": 1680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8309999999999997e-05,
+      "loss": 3.3432,
+      "step": 1690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.83e-05,
+      "loss": 3.3522,
+      "step": 1700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.829e-05,
+      "loss": 3.7299,
+      "step": 1710
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8280000000000005e-05,
+      "loss": 3.5696,
+      "step": 1720
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8270000000000004e-05,
+      "loss": 3.6591,
+      "step": 1730
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.826e-05,
+      "loss": 3.5362,
+      "step": 1740
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.825e-05,
+      "loss": 3.6701,
+      "step": 1750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.824e-05,
+      "loss": 3.6528,
+      "step": 1760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8230000000000004e-05,
+      "loss": 3.6795,
+      "step": 1770
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.822e-05,
+      "loss": 3.8493,
+      "step": 1780
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.821e-05,
+      "loss": 3.7579,
+      "step": 1790
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.82e-05,
+      "loss": 3.6223,
+      "step": 1800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8190000000000004e-05,
+      "loss": 3.7029,
+      "step": 1810
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.818e-05,
+      "loss": 3.7146,
+      "step": 1820
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.817e-05,
+      "loss": 3.9603,
+      "step": 1830
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.816e-05,
+      "loss": 3.642,
+      "step": 1840
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.815e-05,
+      "loss": 3.7575,
+      "step": 1850
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.814e-05,
+      "loss": 3.6183,
+      "step": 1860
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.813e-05,
+      "loss": 3.5998,
+      "step": 1870
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.812000000000001e-05,
+      "loss": 3.7677,
+      "step": 1880
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8110000000000005e-05,
+      "loss": 3.6834,
+      "step": 1890
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8100000000000004e-05,
+      "loss": 3.5415,
+      "step": 1900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.809e-05,
+      "loss": 3.4683,
+      "step": 1910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.808e-05,
+      "loss": 3.7761,
+      "step": 1920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8070000000000006e-05,
+      "loss": 3.3624,
+      "step": 1930
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8060000000000004e-05,
+      "loss": 3.6062,
+      "step": 1940
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.805e-05,
+      "loss": 3.6708,
+      "step": 1950
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.804e-05,
+      "loss": 3.659,
+      "step": 1960
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8030000000000006e-05,
+      "loss": 3.6432,
+      "step": 1970
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8020000000000004e-05,
+      "loss": 3.7559,
+      "step": 1980
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.801e-05,
+      "loss": 3.6808,
+      "step": 1990
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8e-05,
+      "loss": 3.5562,
+      "step": 2000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.799e-05,
+      "loss": 3.6808,
+      "step": 2010
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7980000000000005e-05,
+      "loss": 3.7421,
+      "step": 2020
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.797e-05,
+      "loss": 3.6611,
+      "step": 2030
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.796e-05,
+      "loss": 3.4509,
+      "step": 2040
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.795e-05,
+      "loss": 3.7911,
+      "step": 2050
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.794e-05,
+      "loss": 3.8066,
+      "step": 2060
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7930000000000004e-05,
+      "loss": 3.543,
+      "step": 2070
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.792e-05,
+      "loss": 3.632,
+      "step": 2080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.791000000000001e-05,
+      "loss": 3.6286,
+      "step": 2090
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.79e-05,
+      "loss": 3.5805,
+      "step": 2100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7890000000000004e-05,
+      "loss": 3.8786,
+      "step": 2110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.788e-05,
+      "loss": 3.4437,
+      "step": 2120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.787e-05,
+      "loss": 3.5751,
+      "step": 2130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7860000000000006e-05,
+      "loss": 3.8495,
+      "step": 2140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.785e-05,
+      "loss": 3.3502,
+      "step": 2150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.784e-05,
+      "loss": 3.5337,
+      "step": 2160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.783e-05,
+      "loss": 3.3342,
+      "step": 2170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7820000000000006e-05,
+      "loss": 3.4367,
+      "step": 2180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7810000000000005e-05,
+      "loss": 3.475,
+      "step": 2190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.78e-05,
+      "loss": 3.5047,
+      "step": 2200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.779e-05,
+      "loss": 3.5839,
+      "step": 2210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.778e-05,
+      "loss": 3.5478,
+      "step": 2220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7770000000000005e-05,
+      "loss": 3.5976,
+      "step": 2230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7760000000000004e-05,
+      "loss": 3.5672,
+      "step": 2240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.775e-05,
+      "loss": 3.631,
+      "step": 2250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.774e-05,
+      "loss": 3.3537,
+      "step": 2260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7730000000000005e-05,
+      "loss": 3.4634,
+      "step": 2270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7720000000000004e-05,
+      "loss": 3.6188,
+      "step": 2280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.771e-05,
+      "loss": 3.6358,
+      "step": 2290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.77e-05,
+      "loss": 3.6207,
+      "step": 2300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.769e-05,
+      "loss": 3.4692,
+      "step": 2310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7680000000000004e-05,
+      "loss": 3.2103,
+      "step": 2320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.767e-05,
+      "loss": 3.4731,
+      "step": 2330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.766000000000001e-05,
+      "loss": 3.5369,
+      "step": 2340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.765e-05,
+      "loss": 3.6837,
+      "step": 2350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7640000000000005e-05,
+      "loss": 3.3719,
+      "step": 2360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.763e-05,
+      "loss": 3.6863,
+      "step": 2370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.762e-05,
+      "loss": 3.5715,
+      "step": 2380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.761000000000001e-05,
+      "loss": 3.3855,
+      "step": 2390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.76e-05,
+      "loss": 3.4901,
+      "step": 2400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7590000000000003e-05,
+      "loss": 3.5449,
+      "step": 2410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.758e-05,
+      "loss": 3.4904,
+      "step": 2420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.757e-05,
+      "loss": 3.607,
+      "step": 2430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7560000000000005e-05,
+      "loss": 3.4393,
+      "step": 2440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.755e-05,
+      "loss": 3.5945,
+      "step": 2450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.754e-05,
+      "loss": 3.418,
+      "step": 2460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.753e-05,
+      "loss": 3.6086,
+      "step": 2470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7520000000000006e-05,
+      "loss": 3.4392,
+      "step": 2480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7510000000000004e-05,
+      "loss": 3.4812,
+      "step": 2490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.75e-05,
+      "loss": 3.449,
+      "step": 2500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.749e-05,
+      "loss": 3.4597,
+      "step": 2510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.748e-05,
+      "loss": 3.4045,
+      "step": 2520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7470000000000005e-05,
+      "loss": 3.5779,
+      "step": 2530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.746e-05,
+      "loss": 3.3592,
+      "step": 2540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.745e-05,
+      "loss": 3.4763,
+      "step": 2550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.744e-05,
+      "loss": 3.4167,
+      "step": 2560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7430000000000005e-05,
+      "loss": 3.7355,
+      "step": 2570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.742e-05,
+      "loss": 3.3503,
+      "step": 2580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.741e-05,
+      "loss": 3.4657,
+      "step": 2590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.74e-05,
+      "loss": 3.4138,
+      "step": 2600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.739e-05,
+      "loss": 3.6395,
+      "step": 2610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7380000000000004e-05,
+      "loss": 3.5497,
+      "step": 2620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.737e-05,
+      "loss": 3.6791,
+      "step": 2630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.736000000000001e-05,
+      "loss": 3.5455,
+      "step": 2640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.735e-05,
+      "loss": 3.3831,
+      "step": 2650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7340000000000004e-05,
+      "loss": 3.5829,
+      "step": 2660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.733e-05,
+      "loss": 3.537,
+      "step": 2670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.732e-05,
+      "loss": 3.5842,
+      "step": 2680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7310000000000006e-05,
+      "loss": 3.5003,
+      "step": 2690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.73e-05,
+      "loss": 3.2071,
+      "step": 2700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.729e-05,
+      "loss": 3.6639,
+      "step": 2710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.728e-05,
+      "loss": 3.557,
+      "step": 2720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7270000000000007e-05,
+      "loss": 3.6231,
+      "step": 2730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7260000000000005e-05,
+      "loss": 3.2699,
+      "step": 2740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7249999999999997e-05,
+      "loss": 3.4872,
+      "step": 2750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.724e-05,
+      "loss": 3.6028,
+      "step": 2760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.723e-05,
+      "loss": 3.4034,
+      "step": 2770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7220000000000005e-05,
+      "loss": 3.7685,
+      "step": 2780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7210000000000004e-05,
+      "loss": 3.4174,
+      "step": 2790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.72e-05,
+      "loss": 3.7549,
+      "step": 2800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.719e-05,
+      "loss": 3.3195,
+      "step": 2810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.718e-05,
+      "loss": 3.5254,
+      "step": 2820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7170000000000004e-05,
+      "loss": 3.4303,
+      "step": 2830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.716e-05,
+      "loss": 3.6177,
+      "step": 2840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.715e-05,
+      "loss": 3.308,
+      "step": 2850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.714e-05,
+      "loss": 3.3523,
+      "step": 2860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7130000000000004e-05,
+      "loss": 3.6198,
+      "step": 2870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.712e-05,
+      "loss": 3.6589,
+      "step": 2880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.711e-05,
+      "loss": 3.6859,
+      "step": 2890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.71e-05,
+      "loss": 3.1658,
+      "step": 2900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.709e-05,
+      "loss": 3.5507,
+      "step": 2910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.708e-05,
+      "loss": 3.5265,
+      "step": 2920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.707e-05,
+      "loss": 3.4948,
+      "step": 2930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.706000000000001e-05,
+      "loss": 3.311,
+      "step": 2940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.705e-05,
+      "loss": 3.5113,
+      "step": 2950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7040000000000004e-05,
+      "loss": 3.3193,
+      "step": 2960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.703e-05,
+      "loss": 3.3185,
+      "step": 2970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.702e-05,
+      "loss": 3.4786,
+      "step": 2980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.7010000000000006e-05,
+      "loss": 3.6599,
+      "step": 2990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.7e-05,
+      "loss": 3.4794,
+      "step": 3000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.699e-05,
+      "loss": 3.7727,
+      "step": 3010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.698e-05,
+      "loss": 3.6649,
+      "step": 3020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6970000000000006e-05,
+      "loss": 3.3802,
+      "step": 3030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6960000000000004e-05,
+      "loss": 3.4594,
+      "step": 3040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.695e-05,
+      "loss": 3.5647,
+      "step": 3050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.694e-05,
+      "loss": 3.3488,
+      "step": 3060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.693e-05,
+      "loss": 3.6628,
+      "step": 3070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6920000000000005e-05,
+      "loss": 3.3488,
+      "step": 3080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.691e-05,
+      "loss": 3.3116,
+      "step": 3090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.69e-05,
+      "loss": 3.533,
+      "step": 3100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.689e-05,
+      "loss": 3.5471,
+      "step": 3110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.688e-05,
+      "loss": 3.6178,
+      "step": 3120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6870000000000004e-05,
+      "loss": 3.3414,
+      "step": 3130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.686e-05,
+      "loss": 3.4021,
+      "step": 3140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.685000000000001e-05,
+      "loss": 3.462,
+      "step": 3150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.684e-05,
+      "loss": 3.3365,
+      "step": 3160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6830000000000004e-05,
+      "loss": 3.5816,
+      "step": 3170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.682e-05,
+      "loss": 3.7455,
+      "step": 3180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.681e-05,
+      "loss": 3.5124,
+      "step": 3190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6800000000000006e-05,
+      "loss": 3.3641,
+      "step": 3200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.679e-05,
+      "loss": 3.5211,
+      "step": 3210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.678e-05,
+      "loss": 3.4137,
+      "step": 3220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.677e-05,
+      "loss": 3.6448,
+      "step": 3230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6760000000000006e-05,
+      "loss": 3.6163,
+      "step": 3240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6750000000000005e-05,
+      "loss": 3.496,
+      "step": 3250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.674e-05,
+      "loss": 3.6562,
+      "step": 3260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.673e-05,
+      "loss": 3.3876,
+      "step": 3270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.672e-05,
+      "loss": 3.447,
+      "step": 3280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6710000000000005e-05,
+      "loss": 3.3306,
+      "step": 3290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6700000000000003e-05,
+      "loss": 3.3389,
+      "step": 3300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.669e-05,
+      "loss": 3.367,
+      "step": 3310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.668e-05,
+      "loss": 3.7478,
+      "step": 3320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6670000000000005e-05,
+      "loss": 3.2181,
+      "step": 3330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6660000000000004e-05,
+      "loss": 3.6435,
+      "step": 3340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.665e-05,
+      "loss": 3.5119,
+      "step": 3350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.664e-05,
+      "loss": 3.4484,
+      "step": 3360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.663e-05,
+      "loss": 3.5205,
+      "step": 3370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6620000000000004e-05,
+      "loss": 3.7501,
+      "step": 3380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.661e-05,
+      "loss": 3.3655,
+      "step": 3390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.660000000000001e-05,
+      "loss": 3.659,
+      "step": 3400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.659e-05,
+      "loss": 3.4518,
+      "step": 3410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6580000000000005e-05,
+      "loss": 3.319,
+      "step": 3420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.657e-05,
+      "loss": 3.4247,
+      "step": 3430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.656e-05,
+      "loss": 3.2857,
+      "step": 3440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.655000000000001e-05,
+      "loss": 3.5347,
+      "step": 3450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.654e-05,
+      "loss": 3.4765,
+      "step": 3460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6530000000000003e-05,
+      "loss": 3.4449,
+      "step": 3470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.652e-05,
+      "loss": 3.7257,
+      "step": 3480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.651e-05,
+      "loss": 3.5354,
+      "step": 3490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6500000000000005e-05,
+      "loss": 3.0273,
+      "step": 3500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.649e-05,
+      "loss": 3.1105,
+      "step": 3510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.648e-05,
+      "loss": 3.587,
+      "step": 3520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.647e-05,
+      "loss": 3.4661,
+      "step": 3530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6460000000000006e-05,
+      "loss": 3.474,
+      "step": 3540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6450000000000004e-05,
+      "loss": 3.3273,
+      "step": 3550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.644e-05,
+      "loss": 3.3607,
+      "step": 3560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.643e-05,
+      "loss": 3.3567,
+      "step": 3570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.642e-05,
+      "loss": 3.2668,
+      "step": 3580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6410000000000005e-05,
+      "loss": 3.3724,
+      "step": 3590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.64e-05,
+      "loss": 3.3521,
+      "step": 3600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.639e-05,
+      "loss": 3.7417,
+      "step": 3610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.638e-05,
+      "loss": 3.3845,
+      "step": 3620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6370000000000005e-05,
+      "loss": 3.3729,
+      "step": 3630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.636e-05,
+      "loss": 3.4532,
+      "step": 3640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.635e-05,
+      "loss": 3.147,
+      "step": 3650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.634e-05,
+      "loss": 3.2608,
+      "step": 3660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.633e-05,
+      "loss": 3.5124,
+      "step": 3670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6320000000000004e-05,
+      "loss": 3.4326,
+      "step": 3680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.631e-05,
+      "loss": 3.4291,
+      "step": 3690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.630000000000001e-05,
+      "loss": 3.5528,
+      "step": 3700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.629e-05,
+      "loss": 3.5479,
+      "step": 3710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6280000000000004e-05,
+      "loss": 3.474,
+      "step": 3720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.627e-05,
+      "loss": 3.6121,
+      "step": 3730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.626e-05,
+      "loss": 3.5069,
+      "step": 3740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6250000000000006e-05,
+      "loss": 3.1404,
+      "step": 3750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.624e-05,
+      "loss": 3.5408,
+      "step": 3760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.623e-05,
+      "loss": 3.3988,
+      "step": 3770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.622e-05,
+      "loss": 3.4223,
+      "step": 3780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6210000000000006e-05,
+      "loss": 3.3476,
+      "step": 3790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6200000000000005e-05,
+      "loss": 3.2544,
+      "step": 3800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.619e-05,
+      "loss": 3.2888,
+      "step": 3810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.618e-05,
+      "loss": 3.0995,
+      "step": 3820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.617e-05,
+      "loss": 3.3316,
+      "step": 3830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6160000000000005e-05,
+      "loss": 3.2685,
+      "step": 3840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6150000000000004e-05,
+      "loss": 3.3612,
+      "step": 3850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.614e-05,
+      "loss": 3.236,
+      "step": 3860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.613e-05,
+      "loss": 3.229,
+      "step": 3870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.612e-05,
+      "loss": 3.6302,
+      "step": 3880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6110000000000004e-05,
+      "loss": 3.4187,
+      "step": 3890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.61e-05,
+      "loss": 3.4789,
+      "step": 3900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.609e-05,
+      "loss": 3.4351,
+      "step": 3910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.608e-05,
+      "loss": 3.5173,
+      "step": 3920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6070000000000004e-05,
+      "loss": 3.2549,
+      "step": 3930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.606e-05,
+      "loss": 3.5009,
+      "step": 3940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.605e-05,
+      "loss": 3.4557,
+      "step": 3950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.604e-05,
+      "loss": 3.518,
+      "step": 3960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.603e-05,
+      "loss": 3.433,
+      "step": 3970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.602e-05,
+      "loss": 3.3028,
+      "step": 3980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.601e-05,
+      "loss": 3.3163,
+      "step": 3990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 3.3874,
+      "step": 4000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.599e-05,
+      "loss": 3.6262,
+      "step": 4010
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.5980000000000004e-05,
+      "loss": 3.4417,
+      "step": 4020
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.597e-05,
+      "loss": 3.1665,
+      "step": 4030
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.596e-05,
+      "loss": 3.3696,
+      "step": 4040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.5950000000000006e-05,
+      "loss": 3.4642,
+      "step": 4050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.594e-05,
+      "loss": 3.4854,
+      "step": 4060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.593e-05,
+      "loss": 3.5345,
+      "step": 4070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.592e-05,
+      "loss": 3.5544,
+      "step": 4080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.5910000000000006e-05,
+      "loss": 3.6669,
+      "step": 4090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.5900000000000004e-05,
+      "loss": 3.2138,
+      "step": 4100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.589e-05,
+      "loss": 3.4442,
+      "step": 4110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.588e-05,
+      "loss": 3.1168,
+      "step": 4120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.587e-05,
+      "loss": 3.4121,
+      "step": 4130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.5860000000000005e-05,
+      "loss": 3.508,
+      "step": 4140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.585e-05,
+      "loss": 3.4415,
+      "step": 4150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.584e-05,
+      "loss": 3.6142,
+      "step": 4160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.583e-05,
+      "loss": 3.4301,
+      "step": 4170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.5820000000000005e-05,
+      "loss": 3.4218,
+      "step": 4180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.5810000000000004e-05,
+      "loss": 3.5095,
+      "step": 4190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.58e-05,
+      "loss": 3.3872,
+      "step": 4200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.579e-05,
+      "loss": 3.2319,
+      "step": 4210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.578e-05,
+      "loss": 3.6069,
+      "step": 4220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.5770000000000004e-05,
+      "loss": 3.2633,
+      "step": 4230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.576e-05,
+      "loss": 3.3327,
+      "step": 4240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.575e-05,
+      "loss": 3.191,
+      "step": 4250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.574e-05,
+      "loss": 3.2526,
+      "step": 4260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.573e-05,
+      "loss": 3.4598,
+      "step": 4270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.572e-05,
+      "loss": 3.437,
+      "step": 4280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.571e-05,
+      "loss": 3.202,
+      "step": 4290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.5700000000000006e-05,
+      "loss": 3.5076,
+      "step": 4300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.569e-05,
+      "loss": 3.4542,
+      "step": 4310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.568e-05,
+      "loss": 3.6039,
+      "step": 4320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.567e-05,
+      "loss": 3.3303,
+      "step": 4330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.566e-05,
+      "loss": 3.2707,
+      "step": 4340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.5650000000000005e-05,
+      "loss": 3.2776,
+      "step": 4350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.564e-05,
+      "loss": 3.235,
+      "step": 4360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.563e-05,
+      "loss": 3.3518,
+      "step": 4370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.562e-05,
+      "loss": 3.1763,
+      "step": 4380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.5610000000000005e-05,
+      "loss": 3.4638,
+      "step": 4390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.5600000000000004e-05,
+      "loss": 3.5596,
+      "step": 4400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.559e-05,
+      "loss": 3.2105,
+      "step": 4410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.558e-05,
+      "loss": 3.257,
+      "step": 4420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.557e-05,
+      "loss": 3.3486,
+      "step": 4430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.5560000000000004e-05,
+      "loss": 3.131,
+      "step": 4440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.555e-05,
+      "loss": 3.2593,
+      "step": 4450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.554000000000001e-05,
+      "loss": 3.3469,
+      "step": 4460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.553e-05,
+      "loss": 3.2903,
+      "step": 4470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.5520000000000005e-05,
+      "loss": 3.2924,
+      "step": 4480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.551e-05,
+      "loss": 3.3393,
+      "step": 4490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.55e-05,
+      "loss": 3.411,
+      "step": 4500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.549000000000001e-05,
+      "loss": 3.4837,
+      "step": 4510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.548e-05,
+      "loss": 3.2569,
+      "step": 4520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.5470000000000003e-05,
+      "loss": 3.4656,
+      "step": 4530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.546e-05,
+      "loss": 3.0857,
+      "step": 4540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.545000000000001e-05,
+      "loss": 3.3217,
+      "step": 4550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.5440000000000005e-05,
+      "loss": 3.2881,
+      "step": 4560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.543e-05,
+      "loss": 3.1974,
+      "step": 4570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.542e-05,
+      "loss": 3.2016,
+      "step": 4580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.541e-05,
+      "loss": 3.2432,
+      "step": 4590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.5400000000000006e-05,
+      "loss": 3.1732,
+      "step": 4600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.5390000000000004e-05,
+      "loss": 3.3628,
+      "step": 4610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.538e-05,
+      "loss": 3.3314,
+      "step": 4620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.537e-05,
+      "loss": 3.4407,
+      "step": 4630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.536e-05,
+      "loss": 3.2983,
+      "step": 4640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.5350000000000005e-05,
+      "loss": 3.1479,
+      "step": 4650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.534e-05,
+      "loss": 3.3486,
+      "step": 4660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.533e-05,
+      "loss": 3.1898,
+      "step": 4670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.532e-05,
+      "loss": 3.2745,
+      "step": 4680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.5310000000000005e-05,
+      "loss": 3.2287,
+      "step": 4690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.53e-05,
+      "loss": 3.1701,
+      "step": 4700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.529e-05,
+      "loss": 3.3459,
+      "step": 4710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.528e-05,
+      "loss": 3.5383,
+      "step": 4720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.527e-05,
+      "loss": 3.1829,
+      "step": 4730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.5260000000000004e-05,
+      "loss": 3.1035,
+      "step": 4740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.525e-05,
+      "loss": 3.1199,
+      "step": 4750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.524000000000001e-05,
+      "loss": 3.1871,
+      "step": 4760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.523e-05,
+      "loss": 3.3412,
+      "step": 4770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5220000000000004e-05,
+      "loss": 3.1824,
+      "step": 4780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.521e-05,
+      "loss": 3.2989,
+      "step": 4790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.52e-05,
+      "loss": 3.4504,
+      "step": 4800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5190000000000006e-05,
+      "loss": 3.3349,
+      "step": 4810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.518e-05,
+      "loss": 3.3956,
+      "step": 4820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.517e-05,
+      "loss": 3.2992,
+      "step": 4830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.516e-05,
+      "loss": 3.4262,
+      "step": 4840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5150000000000006e-05,
+      "loss": 3.1954,
+      "step": 4850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5140000000000005e-05,
+      "loss": 3.2248,
+      "step": 4860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.513e-05,
+      "loss": 3.4866,
+      "step": 4870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.512e-05,
+      "loss": 3.3418,
+      "step": 4880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.511e-05,
+      "loss": 3.0721,
+      "step": 4890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5100000000000005e-05,
+      "loss": 3.3774,
+      "step": 4900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5090000000000004e-05,
+      "loss": 3.3656,
+      "step": 4910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.508e-05,
+      "loss": 3.1737,
+      "step": 4920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.507e-05,
+      "loss": 3.2064,
+      "step": 4930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.506e-05,
+      "loss": 3.4258,
+      "step": 4940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5050000000000004e-05,
+      "loss": 3.2721,
+      "step": 4950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.504e-05,
+      "loss": 3.282,
+      "step": 4960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.503e-05,
+      "loss": 3.122,
+      "step": 4970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.502e-05,
+      "loss": 3.4015,
+      "step": 4980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5010000000000004e-05,
+      "loss": 3.3525,
+      "step": 4990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5e-05,
+      "loss": 3.149,
+      "step": 5000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.499e-05,
+      "loss": 3.2453,
+      "step": 5010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.498e-05,
+      "loss": 3.299,
+      "step": 5020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.497e-05,
+      "loss": 3.4224,
+      "step": 5030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.496e-05,
+      "loss": 3.0839,
+      "step": 5040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.495e-05,
+      "loss": 3.5246,
+      "step": 5050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.494000000000001e-05,
+      "loss": 3.4662,
+      "step": 5060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.493e-05,
+      "loss": 3.3234,
+      "step": 5070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.4920000000000004e-05,
+      "loss": 3.2707,
+      "step": 5080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.491e-05,
+      "loss": 3.4139,
+      "step": 5090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.49e-05,
+      "loss": 3.2635,
+      "step": 5100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.4890000000000006e-05,
+      "loss": 3.5254,
+      "step": 5110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.488e-05,
+      "loss": 3.1986,
+      "step": 5120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.487e-05,
+      "loss": 3.1394,
+      "step": 5130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.486e-05,
+      "loss": 3.1707,
+      "step": 5140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.4850000000000006e-05,
+      "loss": 3.0617,
+      "step": 5150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.4840000000000004e-05,
+      "loss": 3.2352,
+      "step": 5160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.483e-05,
+      "loss": 3.1787,
+      "step": 5170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.482e-05,
+      "loss": 3.3874,
+      "step": 5180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.481e-05,
+      "loss": 3.0142,
+      "step": 5190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.4800000000000005e-05,
+      "loss": 3.2601,
+      "step": 5200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.479e-05,
+      "loss": 3.4422,
+      "step": 5210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.478e-05,
+      "loss": 3.0343,
+      "step": 5220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.477e-05,
+      "loss": 3.3087,
+      "step": 5230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.4760000000000005e-05,
+      "loss": 3.2712,
+      "step": 5240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.4750000000000004e-05,
+      "loss": 3.3796,
+      "step": 5250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.474e-05,
+      "loss": 3.0198,
+      "step": 5260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.473e-05,
+      "loss": 3.0114,
+      "step": 5270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.472e-05,
+      "loss": 3.2613,
+      "step": 5280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.4710000000000004e-05,
+      "loss": 3.006,
+      "step": 5290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.47e-05,
+      "loss": 3.3925,
+      "step": 5300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.469e-05,
+      "loss": 3.3835,
+      "step": 5310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.468e-05,
+      "loss": 3.3226,
+      "step": 5320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.467e-05,
+      "loss": 3.2786,
+      "step": 5330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.466e-05,
+      "loss": 3.4058,
+      "step": 5340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.465e-05,
+      "loss": 3.0447,
+      "step": 5350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.4640000000000006e-05,
+      "loss": 3.2119,
+      "step": 5360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.463e-05,
+      "loss": 3.2743,
+      "step": 5370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.462e-05,
+      "loss": 3.4929,
+      "step": 5380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.461e-05,
+      "loss": 3.1102,
+      "step": 5390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.46e-05,
+      "loss": 3.562,
+      "step": 5400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.4590000000000005e-05,
+      "loss": 3.0491,
+      "step": 5410
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.458e-05,
+      "loss": 3.3934,
+      "step": 5420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.457e-05,
+      "loss": 3.4381,
+      "step": 5430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.456e-05,
+      "loss": 3.4174,
+      "step": 5440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.4550000000000005e-05,
+      "loss": 3.1057,
+      "step": 5450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.4540000000000004e-05,
+      "loss": 3.1365,
+      "step": 5460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.453e-05,
+      "loss": 3.1697,
+      "step": 5470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.452e-05,
+      "loss": 3.0941,
+      "step": 5480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.451e-05,
+      "loss": 3.2025,
+      "step": 5490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.4500000000000004e-05,
+      "loss": 3.2804,
+      "step": 5500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.449e-05,
+      "loss": 3.372,
+      "step": 5510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.448e-05,
+      "loss": 3.0677,
+      "step": 5520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.447e-05,
+      "loss": 3.4813,
+      "step": 5530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.4460000000000005e-05,
+      "loss": 3.1439,
+      "step": 5540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.445e-05,
+      "loss": 3.181,
+      "step": 5550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.444e-05,
+      "loss": 3.3403,
+      "step": 5560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.443e-05,
+      "loss": 3.2583,
+      "step": 5570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.442e-05,
+      "loss": 3.048,
+      "step": 5580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.4410000000000003e-05,
+      "loss": 3.3493,
+      "step": 5590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.44e-05,
+      "loss": 3.2835,
+      "step": 5600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.439000000000001e-05,
+      "loss": 3.18,
+      "step": 5610
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.438e-05,
+      "loss": 3.2323,
+      "step": 5620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.4370000000000004e-05,
+      "loss": 3.3205,
+      "step": 5630
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.436e-05,
+      "loss": 3.4412,
+      "step": 5640
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.435e-05,
+      "loss": 3.1678,
+      "step": 5650
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.4340000000000006e-05,
+      "loss": 3.0126,
+      "step": 5660
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.4330000000000004e-05,
+      "loss": 2.9894,
+      "step": 5670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.432e-05,
+      "loss": 2.9588,
+      "step": 5680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.431e-05,
+      "loss": 3.2089,
+      "step": 5690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.43e-05,
+      "loss": 3.3925,
+      "step": 5700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.4290000000000005e-05,
+      "loss": 3.2344,
+      "step": 5710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.428e-05,
+      "loss": 2.9364,
+      "step": 5720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.427e-05,
+      "loss": 3.4431,
+      "step": 5730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.426e-05,
+      "loss": 3.1455,
+      "step": 5740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.4250000000000005e-05,
+      "loss": 3.1432,
+      "step": 5750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.424e-05,
+      "loss": 3.3195,
+      "step": 5760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.423e-05,
+      "loss": 3.1551,
+      "step": 5770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.422e-05,
+      "loss": 3.1103,
+      "step": 5780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.421e-05,
+      "loss": 3.2512,
+      "step": 5790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.4200000000000004e-05,
+      "loss": 3.1682,
+      "step": 5800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.419e-05,
+      "loss": 3.3619,
+      "step": 5810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.418000000000001e-05,
+      "loss": 3.1158,
+      "step": 5820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.417e-05,
+      "loss": 3.0808,
+      "step": 5830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.4160000000000004e-05,
+      "loss": 3.344,
+      "step": 5840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.415e-05,
+      "loss": 3.1639,
+      "step": 5850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.414e-05,
+      "loss": 3.3826,
+      "step": 5860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.4130000000000006e-05,
+      "loss": 3.1063,
+      "step": 5870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.412e-05,
+      "loss": 3.216,
+      "step": 5880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.411e-05,
+      "loss": 3.2587,
+      "step": 5890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.41e-05,
+      "loss": 2.7709,
+      "step": 5900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.4090000000000006e-05,
+      "loss": 3.1192,
+      "step": 5910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.4080000000000005e-05,
+      "loss": 3.2508,
+      "step": 5920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.407e-05,
+      "loss": 3.2943,
+      "step": 5930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.406e-05,
+      "loss": 3.1809,
+      "step": 5940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.405e-05,
+      "loss": 3.2429,
+      "step": 5950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.4040000000000005e-05,
+      "loss": 3.329,
+      "step": 5960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.4030000000000004e-05,
+      "loss": 3.3533,
+      "step": 5970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.402e-05,
+      "loss": 3.0325,
+      "step": 5980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.401e-05,
+      "loss": 3.2711,
+      "step": 5990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 3.191,
+      "step": 6000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.3990000000000004e-05,
+      "loss": 3.3888,
+      "step": 6010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.398e-05,
+      "loss": 3.2033,
+      "step": 6020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.397e-05,
+      "loss": 3.1532,
+      "step": 6030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.396e-05,
+      "loss": 3.0797,
+      "step": 6040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.3950000000000004e-05,
+      "loss": 3.2978,
+      "step": 6050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.394e-05,
+      "loss": 3.143,
+      "step": 6060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.393e-05,
+      "loss": 3.1399,
+      "step": 6070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.392e-05,
+      "loss": 3.0769,
+      "step": 6080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.391e-05,
+      "loss": 2.9954,
+      "step": 6090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.39e-05,
+      "loss": 3.0037,
+      "step": 6100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.389e-05,
+      "loss": 3.0514,
+      "step": 6110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.388000000000001e-05,
+      "loss": 2.9742,
+      "step": 6120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.387e-05,
+      "loss": 3.1006,
+      "step": 6130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.3860000000000004e-05,
+      "loss": 3.2109,
+      "step": 6140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.385e-05,
+      "loss": 3.0101,
+      "step": 6150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.384e-05,
+      "loss": 2.9207,
+      "step": 6160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.3830000000000006e-05,
+      "loss": 3.2267,
+      "step": 6170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.382e-05,
+      "loss": 3.1295,
+      "step": 6180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.381e-05,
+      "loss": 2.9679,
+      "step": 6190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.38e-05,
+      "loss": 2.7938,
+      "step": 6200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.3790000000000006e-05,
+      "loss": 3.0195,
+      "step": 6210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.3780000000000004e-05,
+      "loss": 2.9921,
+      "step": 6220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.377e-05,
+      "loss": 2.9958,
+      "step": 6230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.376e-05,
+      "loss": 3.339,
+      "step": 6240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.375e-05,
+      "loss": 3.2248,
+      "step": 6250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.3740000000000005e-05,
+      "loss": 3.0739,
+      "step": 6260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.373e-05,
+      "loss": 3.1813,
+      "step": 6270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.372e-05,
+      "loss": 3.2701,
+      "step": 6280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.371e-05,
+      "loss": 3.067,
+      "step": 6290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.3700000000000005e-05,
+      "loss": 3.1613,
+      "step": 6300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.3690000000000004e-05,
+      "loss": 3.1088,
+      "step": 6310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.368e-05,
+      "loss": 3.083,
+      "step": 6320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.367e-05,
+      "loss": 3.2792,
+      "step": 6330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.366e-05,
+      "loss": 3.0194,
+      "step": 6340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.3650000000000004e-05,
+      "loss": 3.0761,
+      "step": 6350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.364e-05,
+      "loss": 3.0719,
+      "step": 6360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.363000000000001e-05,
+      "loss": 3.2211,
+      "step": 6370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.362e-05,
+      "loss": 3.1511,
+      "step": 6380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.361e-05,
+      "loss": 3.1341,
+      "step": 6390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.36e-05,
+      "loss": 3.23,
+      "step": 6400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.359e-05,
+      "loss": 2.9773,
+      "step": 6410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.3580000000000006e-05,
+      "loss": 3.0157,
+      "step": 6420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.357e-05,
+      "loss": 3.2114,
+      "step": 6430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.356e-05,
+      "loss": 3.1846,
+      "step": 6440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.355e-05,
+      "loss": 2.9138,
+      "step": 6450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.354e-05,
+      "loss": 2.9442,
+      "step": 6460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.3530000000000005e-05,
+      "loss": 2.8412,
+      "step": 6470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.352e-05,
+      "loss": 3.1785,
+      "step": 6480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.351e-05,
+      "loss": 3.2634,
+      "step": 6490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.35e-05,
+      "loss": 3.3384,
+      "step": 6500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.3490000000000005e-05,
+      "loss": 3.2171,
+      "step": 6510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.3480000000000004e-05,
+      "loss": 2.8692,
+      "step": 6520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.347e-05,
+      "loss": 3.0287,
+      "step": 6530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.346e-05,
+      "loss": 2.9838,
+      "step": 6540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.345e-05,
+      "loss": 3.2502,
+      "step": 6550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.3440000000000004e-05,
+      "loss": 3.1976,
+      "step": 6560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.343e-05,
+      "loss": 3.1596,
+      "step": 6570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.342e-05,
+      "loss": 3.0616,
+      "step": 6580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.341e-05,
+      "loss": 3.0666,
+      "step": 6590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.3400000000000005e-05,
+      "loss": 3.4082,
+      "step": 6600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.339e-05,
+      "loss": 3.2049,
+      "step": 6610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.338e-05,
+      "loss": 3.0718,
+      "step": 6620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.337e-05,
+      "loss": 3.2982,
+      "step": 6630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.336e-05,
+      "loss": 3.2257,
+      "step": 6640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.335e-05,
+      "loss": 3.1157,
+      "step": 6650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.334e-05,
+      "loss": 3.3634,
+      "step": 6660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.333000000000001e-05,
+      "loss": 3.1452,
+      "step": 6670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.332e-05,
+      "loss": 3.0224,
+      "step": 6680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.3310000000000004e-05,
+      "loss": 2.7654,
+      "step": 6690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.33e-05,
+      "loss": 3.229,
+      "step": 6700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.329e-05,
+      "loss": 3.1844,
+      "step": 6710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.3280000000000006e-05,
+      "loss": 3.2128,
+      "step": 6720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.327e-05,
+      "loss": 3.0261,
+      "step": 6730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.326e-05,
+      "loss": 3.1222,
+      "step": 6740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.325e-05,
+      "loss": 3.089,
+      "step": 6750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.324e-05,
+      "loss": 3.0884,
+      "step": 6760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.3230000000000005e-05,
+      "loss": 3.2092,
+      "step": 6770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.3219999999999996e-05,
+      "loss": 2.9955,
+      "step": 6780
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.321e-05,
+      "loss": 3.0103,
+      "step": 6790
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.32e-05,
+      "loss": 2.9306,
+      "step": 6800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.3190000000000005e-05,
+      "loss": 2.8493,
+      "step": 6810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.318e-05,
+      "loss": 3.1847,
+      "step": 6820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.317e-05,
+      "loss": 3.267,
+      "step": 6830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.316e-05,
+      "loss": 3.0126,
+      "step": 6840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.315e-05,
+      "loss": 3.014,
+      "step": 6850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.3140000000000004e-05,
+      "loss": 3.3542,
+      "step": 6860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.313e-05,
+      "loss": 3.431,
+      "step": 6870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.312000000000001e-05,
+      "loss": 3.1132,
+      "step": 6880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.311e-05,
+      "loss": 3.3333,
+      "step": 6890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.3100000000000004e-05,
+      "loss": 3.0463,
+      "step": 6900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.309e-05,
+      "loss": 2.8804,
+      "step": 6910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.308e-05,
+      "loss": 3.127,
+      "step": 6920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.3070000000000006e-05,
+      "loss": 3.1801,
+      "step": 6930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.306e-05,
+      "loss": 3.154,
+      "step": 6940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.305e-05,
+      "loss": 3.14,
+      "step": 6950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.304e-05,
+      "loss": 3.1792,
+      "step": 6960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.3030000000000006e-05,
+      "loss": 3.1981,
+      "step": 6970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.3020000000000005e-05,
+      "loss": 3.2747,
+      "step": 6980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.301e-05,
+      "loss": 3.1662,
+      "step": 6990
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.3e-05,
+      "loss": 2.8414,
+      "step": 7000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.299e-05,
+      "loss": 3.074,
+      "step": 7010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.2980000000000005e-05,
+      "loss": 3.2187,
+      "step": 7020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.2970000000000004e-05,
+      "loss": 3.3183,
+      "step": 7030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.296e-05,
+      "loss": 2.8145,
+      "step": 7040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.295e-05,
+      "loss": 3.0797,
+      "step": 7050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.2940000000000006e-05,
+      "loss": 3.2178,
+      "step": 7060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.2930000000000004e-05,
+      "loss": 3.0168,
+      "step": 7070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.292e-05,
+      "loss": 2.9879,
+      "step": 7080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.291e-05,
+      "loss": 3.0367,
+      "step": 7090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.29e-05,
+      "loss": 3.095,
+      "step": 7100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.2890000000000004e-05,
+      "loss": 3.0779,
+      "step": 7110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.288e-05,
+      "loss": 3.2029,
+      "step": 7120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.287000000000001e-05,
+      "loss": 3.0887,
+      "step": 7130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.286e-05,
+      "loss": 3.1174,
+      "step": 7140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.285e-05,
+      "loss": 3.0846,
+      "step": 7150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.284e-05,
+      "loss": 3.0313,
+      "step": 7160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.283e-05,
+      "loss": 3.1364,
+      "step": 7170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.282000000000001e-05,
+      "loss": 3.2634,
+      "step": 7180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.281e-05,
+      "loss": 2.8806,
+      "step": 7190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.2800000000000004e-05,
+      "loss": 3.2121,
+      "step": 7200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.279e-05,
+      "loss": 3.1878,
+      "step": 7210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.278e-05,
+      "loss": 3.1154,
+      "step": 7220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.2770000000000006e-05,
+      "loss": 3.0217,
+      "step": 7230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.276e-05,
+      "loss": 2.9985,
+      "step": 7240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.275e-05,
+      "loss": 3.0441,
+      "step": 7250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.274e-05,
+      "loss": 3.2285,
+      "step": 7260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.2730000000000006e-05,
+      "loss": 2.9866,
+      "step": 7270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.2720000000000004e-05,
+      "loss": 3.0957,
+      "step": 7280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.271e-05,
+      "loss": 3.1427,
+      "step": 7290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.27e-05,
+      "loss": 3.1098,
+      "step": 7300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.269e-05,
+      "loss": 2.9373,
+      "step": 7310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.2680000000000005e-05,
+      "loss": 2.9383,
+      "step": 7320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.267e-05,
+      "loss": 3.1315,
+      "step": 7330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.266e-05,
+      "loss": 3.0393,
+      "step": 7340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.265e-05,
+      "loss": 3.0831,
+      "step": 7350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.2640000000000005e-05,
+      "loss": 2.917,
+      "step": 7360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.2630000000000004e-05,
+      "loss": 2.9646,
+      "step": 7370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.262e-05,
+      "loss": 3.2483,
+      "step": 7380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.261e-05,
+      "loss": 3.1851,
+      "step": 7390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.26e-05,
+      "loss": 2.9896,
+      "step": 7400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.2590000000000004e-05,
+      "loss": 3.1845,
+      "step": 7410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.258e-05,
+      "loss": 3.042,
+      "step": 7420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.257000000000001e-05,
+      "loss": 2.7334,
+      "step": 7430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.256e-05,
+      "loss": 3.0986,
+      "step": 7440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.2550000000000004e-05,
+      "loss": 3.1178,
+      "step": 7450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.254e-05,
+      "loss": 3.1692,
+      "step": 7460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.253e-05,
+      "loss": 3.0565,
+      "step": 7470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.2520000000000006e-05,
+      "loss": 3.159,
+      "step": 7480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.251e-05,
+      "loss": 3.309,
+      "step": 7490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.25e-05,
+      "loss": 3.1642,
+      "step": 7500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.249e-05,
+      "loss": 3.1963,
+      "step": 7510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.248e-05,
+      "loss": 3.1287,
+      "step": 7520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.2470000000000005e-05,
+      "loss": 3.0931,
+      "step": 7530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.246e-05,
+      "loss": 2.8858,
+      "step": 7540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.245e-05,
+      "loss": 2.9546,
+      "step": 7550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.244e-05,
+      "loss": 3.1842,
+      "step": 7560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.2430000000000005e-05,
+      "loss": 3.1039,
+      "step": 7570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.2420000000000004e-05,
+      "loss": 2.904,
+      "step": 7580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.241e-05,
+      "loss": 2.9024,
+      "step": 7590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.24e-05,
+      "loss": 2.8572,
+      "step": 7600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.239e-05,
+      "loss": 3.0006,
+      "step": 7610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.2380000000000004e-05,
+      "loss": 2.9087,
+      "step": 7620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.237e-05,
+      "loss": 2.7505,
+      "step": 7630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.236e-05,
+      "loss": 3.1397,
+      "step": 7640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.235e-05,
+      "loss": 3.212,
+      "step": 7650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.2340000000000005e-05,
+      "loss": 2.8834,
+      "step": 7660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.233e-05,
+      "loss": 3.0359,
+      "step": 7670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.232e-05,
+      "loss": 3.3123,
+      "step": 7680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.231e-05,
+      "loss": 2.9365,
+      "step": 7690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.23e-05,
+      "loss": 3.1102,
+      "step": 7700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.229e-05,
+      "loss": 3.1042,
+      "step": 7710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.228e-05,
+      "loss": 3.2393,
+      "step": 7720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.227000000000001e-05,
+      "loss": 3.134,
+      "step": 7730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.226e-05,
+      "loss": 2.6955,
+      "step": 7740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.2250000000000004e-05,
+      "loss": 3.225,
+      "step": 7750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.224e-05,
+      "loss": 3.1435,
+      "step": 7760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.223e-05,
+      "loss": 2.9459,
+      "step": 7770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.2220000000000006e-05,
+      "loss": 3.177,
+      "step": 7780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.221e-05,
+      "loss": 2.8804,
+      "step": 7790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.22e-05,
+      "loss": 2.8398,
+      "step": 7800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.219e-05,
+      "loss": 3.1209,
+      "step": 7810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.2180000000000006e-05,
+      "loss": 2.8864,
+      "step": 7820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.2170000000000005e-05,
+      "loss": 2.8861,
+      "step": 7830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.2159999999999996e-05,
+      "loss": 3.1585,
+      "step": 7840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.215e-05,
+      "loss": 3.1513,
+      "step": 7850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.214e-05,
+      "loss": 3.1592,
+      "step": 7860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.2130000000000005e-05,
+      "loss": 2.9661,
+      "step": 7870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.212e-05,
+      "loss": 3.3416,
+      "step": 7880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.211e-05,
+      "loss": 2.9189,
+      "step": 7890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.21e-05,
+      "loss": 3.118,
+      "step": 7900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.209e-05,
+      "loss": 3.1984,
+      "step": 7910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.2080000000000004e-05,
+      "loss": 2.9185,
+      "step": 7920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.207e-05,
+      "loss": 3.0439,
+      "step": 7930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.206e-05,
+      "loss": 3.199,
+      "step": 7940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.205e-05,
+      "loss": 3.0051,
+      "step": 7950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.2040000000000004e-05,
+      "loss": 3.0338,
+      "step": 7960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.203e-05,
+      "loss": 3.0044,
+      "step": 7970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.202e-05,
+      "loss": 3.0226,
+      "step": 7980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.201e-05,
+      "loss": 2.9914,
+      "step": 7990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.2e-05,
+      "loss": 2.8446,
+      "step": 8000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.199e-05,
+      "loss": 2.9577,
+      "step": 8010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.198e-05,
+      "loss": 3.1902,
+      "step": 8020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.1970000000000006e-05,
+      "loss": 2.936,
+      "step": 8030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.196e-05,
+      "loss": 2.8726,
+      "step": 8040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.195e-05,
+      "loss": 2.9582,
+      "step": 8050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.194e-05,
+      "loss": 3.0746,
+      "step": 8060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.193e-05,
+      "loss": 3.1021,
+      "step": 8070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.1920000000000005e-05,
+      "loss": 2.9772,
+      "step": 8080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.191e-05,
+      "loss": 2.8608,
+      "step": 8090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.19e-05,
+      "loss": 3.0292,
+      "step": 8100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.189e-05,
+      "loss": 3.0353,
+      "step": 8110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.1880000000000006e-05,
+      "loss": 2.9829,
+      "step": 8120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.1870000000000004e-05,
+      "loss": 3.1405,
+      "step": 8130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.186e-05,
+      "loss": 2.9741,
+      "step": 8140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.185e-05,
+      "loss": 3.0733,
+      "step": 8150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.184e-05,
+      "loss": 3.011,
+      "step": 8160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.1830000000000004e-05,
+      "loss": 3.0209,
+      "step": 8170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.182e-05,
+      "loss": 2.8827,
+      "step": 8180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.181000000000001e-05,
+      "loss": 3.0681,
+      "step": 8190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.18e-05,
+      "loss": 2.9051,
+      "step": 8200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.179e-05,
+      "loss": 3.1408,
+      "step": 8210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.178e-05,
+      "loss": 3.0148,
+      "step": 8220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.177e-05,
+      "loss": 2.9948,
+      "step": 8230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.176000000000001e-05,
+      "loss": 3.0307,
+      "step": 8240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.175e-05,
+      "loss": 2.9831,
+      "step": 8250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.1740000000000004e-05,
+      "loss": 2.8361,
+      "step": 8260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.173e-05,
+      "loss": 2.9719,
+      "step": 8270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.172e-05,
+      "loss": 3.2532,
+      "step": 8280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.1710000000000006e-05,
+      "loss": 3.0276,
+      "step": 8290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.17e-05,
+      "loss": 3.0773,
+      "step": 8300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.169e-05,
+      "loss": 3.2642,
+      "step": 8310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.168e-05,
+      "loss": 3.1946,
+      "step": 8320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.1670000000000006e-05,
+      "loss": 2.9617,
+      "step": 8330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.1660000000000004e-05,
+      "loss": 2.7249,
+      "step": 8340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.165e-05,
+      "loss": 2.9111,
+      "step": 8350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.164e-05,
+      "loss": 3.0262,
+      "step": 8360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.163e-05,
+      "loss": 3.0568,
+      "step": 8370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.1620000000000005e-05,
+      "loss": 3.0012,
+      "step": 8380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.161e-05,
+      "loss": 3.0039,
+      "step": 8390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.16e-05,
+      "loss": 3.1146,
+      "step": 8400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.159e-05,
+      "loss": 2.7789,
+      "step": 8410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.1580000000000005e-05,
+      "loss": 3.18,
+      "step": 8420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.1570000000000003e-05,
+      "loss": 3.0653,
+      "step": 8430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.156e-05,
+      "loss": 3.034,
+      "step": 8440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.155e-05,
+      "loss": 3.0619,
+      "step": 8450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.154e-05,
+      "loss": 2.9076,
+      "step": 8460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.1530000000000004e-05,
+      "loss": 2.8999,
+      "step": 8470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.152e-05,
+      "loss": 2.9422,
+      "step": 8480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.151000000000001e-05,
+      "loss": 3.0256,
+      "step": 8490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.15e-05,
+      "loss": 2.8945,
+      "step": 8500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.1490000000000004e-05,
+      "loss": 2.9759,
+      "step": 8510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.148e-05,
+      "loss": 2.953,
+      "step": 8520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.147e-05,
+      "loss": 3.0842,
+      "step": 8530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.1460000000000006e-05,
+      "loss": 2.985,
+      "step": 8540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.145e-05,
+      "loss": 2.8859,
+      "step": 8550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.144e-05,
+      "loss": 2.9175,
+      "step": 8560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.143e-05,
+      "loss": 2.8667,
+      "step": 8570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.142000000000001e-05,
+      "loss": 3.212,
+      "step": 8580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.1410000000000005e-05,
+      "loss": 3.0047,
+      "step": 8590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.14e-05,
+      "loss": 3.0591,
+      "step": 8600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.139e-05,
+      "loss": 3.1836,
+      "step": 8610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.138e-05,
+      "loss": 3.098,
+      "step": 8620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.1370000000000005e-05,
+      "loss": 3.2656,
+      "step": 8630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.1360000000000004e-05,
+      "loss": 2.7611,
+      "step": 8640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.135e-05,
+      "loss": 2.9352,
+      "step": 8650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.134e-05,
+      "loss": 3.0456,
+      "step": 8660
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.133e-05,
+      "loss": 2.9891,
+      "step": 8670
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.1320000000000004e-05,
+      "loss": 2.8309,
+      "step": 8680
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.131e-05,
+      "loss": 2.9986,
+      "step": 8690
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.13e-05,
+      "loss": 3.0206,
+      "step": 8700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.129e-05,
+      "loss": 3.0124,
+      "step": 8710
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.1280000000000005e-05,
+      "loss": 2.8834,
+      "step": 8720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.127e-05,
+      "loss": 3.0414,
+      "step": 8730
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.126e-05,
+      "loss": 3.026,
+      "step": 8740
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.125e-05,
+      "loss": 2.8843,
+      "step": 8750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.124e-05,
+      "loss": 2.9531,
+      "step": 8760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.123e-05,
+      "loss": 2.8876,
+      "step": 8770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.122e-05,
+      "loss": 2.8246,
+      "step": 8780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.121000000000001e-05,
+      "loss": 2.9301,
+      "step": 8790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.12e-05,
+      "loss": 2.9412,
+      "step": 8800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.1190000000000004e-05,
+      "loss": 2.9879,
+      "step": 8810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.118e-05,
+      "loss": 2.8749,
+      "step": 8820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.117e-05,
+      "loss": 2.9561,
+      "step": 8830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.1160000000000006e-05,
+      "loss": 3.0098,
+      "step": 8840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.115e-05,
+      "loss": 3.0702,
+      "step": 8850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.114e-05,
+      "loss": 2.779,
+      "step": 8860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.113e-05,
+      "loss": 2.8641,
+      "step": 8870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.1120000000000006e-05,
+      "loss": 2.9109,
+      "step": 8880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.1110000000000005e-05,
+      "loss": 2.8169,
+      "step": 8890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.11e-05,
+      "loss": 3.0296,
+      "step": 8900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.109e-05,
+      "loss": 2.8616,
+      "step": 8910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.108e-05,
+      "loss": 2.856,
+      "step": 8920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.1070000000000005e-05,
+      "loss": 2.8322,
+      "step": 8930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.106e-05,
+      "loss": 2.9511,
+      "step": 8940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.105e-05,
+      "loss": 2.9421,
+      "step": 8950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.104e-05,
+      "loss": 3.0371,
+      "step": 8960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.103e-05,
+      "loss": 3.1391,
+      "step": 8970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.1020000000000004e-05,
+      "loss": 2.8942,
+      "step": 8980
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.101e-05,
+      "loss": 2.9707,
+      "step": 8990
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.1e-05,
+      "loss": 2.8912,
+      "step": 9000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.099e-05,
+      "loss": 2.9291,
+      "step": 9010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.0980000000000004e-05,
+      "loss": 2.959,
+      "step": 9020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.097e-05,
+      "loss": 2.9408,
+      "step": 9030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.096e-05,
+      "loss": 2.9719,
+      "step": 9040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.095e-05,
+      "loss": 2.906,
+      "step": 9050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.094e-05,
+      "loss": 2.9058,
+      "step": 9060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.093e-05,
+      "loss": 2.9658,
+      "step": 9070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.092e-05,
+      "loss": 2.8658,
+      "step": 9080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.0910000000000006e-05,
+      "loss": 2.9146,
+      "step": 9090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.09e-05,
+      "loss": 2.7098,
+      "step": 9100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.089e-05,
+      "loss": 2.9795,
+      "step": 9110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.088e-05,
+      "loss": 2.8176,
+      "step": 9120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.087e-05,
+      "loss": 3.0667,
+      "step": 9130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.0860000000000005e-05,
+      "loss": 2.7731,
+      "step": 9140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.085e-05,
+      "loss": 2.8874,
+      "step": 9150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.084e-05,
+      "loss": 3.1121,
+      "step": 9160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.083e-05,
+      "loss": 2.9211,
+      "step": 9170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.0820000000000006e-05,
+      "loss": 2.7703,
+      "step": 9180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.0810000000000004e-05,
+      "loss": 3.0616,
+      "step": 9190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.08e-05,
+      "loss": 3.0278,
+      "step": 9200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.079e-05,
+      "loss": 2.8582,
+      "step": 9210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.078e-05,
+      "loss": 3.0005,
+      "step": 9220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.0770000000000004e-05,
+      "loss": 2.6651,
+      "step": 9230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.076e-05,
+      "loss": 3.0119,
+      "step": 9240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.075e-05,
+      "loss": 2.9333,
+      "step": 9250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.074e-05,
+      "loss": 3.1117,
+      "step": 9260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.0730000000000005e-05,
+      "loss": 2.9148,
+      "step": 9270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.072e-05,
+      "loss": 3.1094,
+      "step": 9280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.071e-05,
+      "loss": 2.814,
+      "step": 9290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.07e-05,
+      "loss": 2.937,
+      "step": 9300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.069e-05,
+      "loss": 3.0629,
+      "step": 9310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.0680000000000004e-05,
+      "loss": 2.8566,
+      "step": 9320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.067e-05,
+      "loss": 2.9136,
+      "step": 9330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.066e-05,
+      "loss": 2.9148,
+      "step": 9340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.065e-05,
+      "loss": 2.7988,
+      "step": 9350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.064e-05,
+      "loss": 3.0697,
+      "step": 9360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.063e-05,
+      "loss": 2.9867,
+      "step": 9370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.062e-05,
+      "loss": 3.0646,
+      "step": 9380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.0610000000000006e-05,
+      "loss": 3.0656,
+      "step": 9390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.0600000000000004e-05,
+      "loss": 2.8774,
+      "step": 9400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.059e-05,
+      "loss": 3.0456,
+      "step": 9410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.058e-05,
+      "loss": 2.8044,
+      "step": 9420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.057e-05,
+      "loss": 3.0464,
+      "step": 9430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.0560000000000005e-05,
+      "loss": 3.1577,
+      "step": 9440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.055e-05,
+      "loss": 2.8804,
+      "step": 9450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.054e-05,
+      "loss": 2.8044,
+      "step": 9460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.053e-05,
+      "loss": 2.8347,
+      "step": 9470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.0520000000000005e-05,
+      "loss": 2.8557,
+      "step": 9480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.0510000000000003e-05,
+      "loss": 2.7614,
+      "step": 9490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.05e-05,
+      "loss": 2.9492,
+      "step": 9500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.049e-05,
+      "loss": 2.843,
+      "step": 9510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.048e-05,
+      "loss": 2.9671,
+      "step": 9520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.0470000000000004e-05,
+      "loss": 2.8892,
+      "step": 9530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.046e-05,
+      "loss": 2.9188,
+      "step": 9540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.045000000000001e-05,
+      "loss": 2.8911,
+      "step": 9550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.044e-05,
+      "loss": 2.8575,
+      "step": 9560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.0430000000000004e-05,
+      "loss": 2.9141,
+      "step": 9570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.042e-05,
+      "loss": 2.8411,
+      "step": 9580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.041e-05,
+      "loss": 2.9864,
+      "step": 9590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.0400000000000006e-05,
+      "loss": 2.7169,
+      "step": 9600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.039e-05,
+      "loss": 2.9778,
+      "step": 9610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.038e-05,
+      "loss": 2.8953,
+      "step": 9620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.037e-05,
+      "loss": 3.1031,
+      "step": 9630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.0360000000000007e-05,
+      "loss": 2.87,
+      "step": 9640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.0350000000000005e-05,
+      "loss": 2.8458,
+      "step": 9650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.034e-05,
+      "loss": 2.8736,
+      "step": 9660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.033e-05,
+      "loss": 2.6443,
+      "step": 9670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.032e-05,
+      "loss": 2.7202,
+      "step": 9680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0310000000000005e-05,
+      "loss": 2.9445,
+      "step": 9690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0300000000000004e-05,
+      "loss": 3.1389,
+      "step": 9700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.029e-05,
+      "loss": 2.9316,
+      "step": 9710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.028e-05,
+      "loss": 2.8796,
+      "step": 9720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.027e-05,
+      "loss": 2.6658,
+      "step": 9730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0260000000000004e-05,
+      "loss": 2.9459,
+      "step": 9740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.025e-05,
+      "loss": 3.0945,
+      "step": 9750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.024e-05,
+      "loss": 2.9478,
+      "step": 9760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.023e-05,
+      "loss": 2.9303,
+      "step": 9770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0220000000000005e-05,
+      "loss": 2.8952,
+      "step": 9780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.021e-05,
+      "loss": 2.8146,
+      "step": 9790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.02e-05,
+      "loss": 2.9645,
+      "step": 9800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.019e-05,
+      "loss": 3.1053,
+      "step": 9810
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.018e-05,
+      "loss": 2.9772,
+      "step": 9820
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.017e-05,
+      "loss": 2.9059,
+      "step": 9830
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.016e-05,
+      "loss": 2.9527,
+      "step": 9840
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.015000000000001e-05,
+      "loss": 2.9311,
+      "step": 9850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.014e-05,
+      "loss": 2.7273,
+      "step": 9860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0130000000000004e-05,
+      "loss": 2.9279,
+      "step": 9870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.012e-05,
+      "loss": 2.7953,
+      "step": 9880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.011e-05,
+      "loss": 2.7064,
+      "step": 9890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0100000000000006e-05,
+      "loss": 2.9889,
+      "step": 9900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.009e-05,
+      "loss": 2.8456,
+      "step": 9910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.008e-05,
+      "loss": 2.8898,
+      "step": 9920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.007e-05,
+      "loss": 2.8237,
+      "step": 9930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0060000000000006e-05,
+      "loss": 2.9552,
+      "step": 9940
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0050000000000004e-05,
+      "loss": 2.8919,
+      "step": 9950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.004e-05,
+      "loss": 3.1478,
+      "step": 9960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.003e-05,
+      "loss": 2.988,
+      "step": 9970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.002e-05,
+      "loss": 2.9377,
+      "step": 9980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0010000000000005e-05,
+      "loss": 2.8547,
+      "step": 9990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4e-05,
+      "loss": 2.726,
+      "step": 10000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.999e-05,
+      "loss": 3.0365,
+      "step": 10010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.998e-05,
+      "loss": 2.8287,
+      "step": 10020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.9970000000000005e-05,
+      "loss": 2.7793,
+      "step": 10030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.9960000000000004e-05,
+      "loss": 2.994,
+      "step": 10040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.995e-05,
+      "loss": 2.8193,
+      "step": 10050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.994e-05,
+      "loss": 2.6963,
+      "step": 10060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.993e-05,
+      "loss": 3.1446,
+      "step": 10070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.9920000000000004e-05,
+      "loss": 3.0178,
+      "step": 10080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.991e-05,
+      "loss": 2.9954,
+      "step": 10090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.99e-05,
+      "loss": 2.7289,
+      "step": 10100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.989e-05,
+      "loss": 2.8975,
+      "step": 10110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.988e-05,
+      "loss": 2.8215,
+      "step": 10120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.987e-05,
+      "loss": 3.2782,
+      "step": 10130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.986e-05,
+      "loss": 2.9152,
+      "step": 10140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.9850000000000006e-05,
+      "loss": 2.8904,
+      "step": 10150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.984e-05,
+      "loss": 2.7347,
+      "step": 10160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.983e-05,
+      "loss": 2.9805,
+      "step": 10170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.982e-05,
+      "loss": 2.5942,
+      "step": 10180
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.981e-05,
+      "loss": 3.0049,
+      "step": 10190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 2.9099,
+      "step": 10200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.979e-05,
+      "loss": 2.6679,
+      "step": 10210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.978e-05,
+      "loss": 2.9058,
+      "step": 10220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.977e-05,
+      "loss": 3.0467,
+      "step": 10230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.9760000000000006e-05,
+      "loss": 2.7013,
+      "step": 10240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.9750000000000004e-05,
+      "loss": 2.9317,
+      "step": 10250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.974e-05,
+      "loss": 2.7981,
+      "step": 10260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.973e-05,
+      "loss": 2.7123,
+      "step": 10270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.972e-05,
+      "loss": 2.8828,
+      "step": 10280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.9710000000000004e-05,
+      "loss": 3.0578,
+      "step": 10290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.97e-05,
+      "loss": 2.7758,
+      "step": 10300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.969e-05,
+      "loss": 2.766,
+      "step": 10310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.968e-05,
+      "loss": 2.8502,
+      "step": 10320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.9670000000000005e-05,
+      "loss": 2.7009,
+      "step": 10330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.966e-05,
+      "loss": 2.9943,
+      "step": 10340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.965e-05,
+      "loss": 3.0213,
+      "step": 10350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.964e-05,
+      "loss": 2.7339,
+      "step": 10360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.963e-05,
+      "loss": 2.7819,
+      "step": 10370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9620000000000004e-05,
+      "loss": 2.8516,
+      "step": 10380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.961e-05,
+      "loss": 2.8846,
+      "step": 10390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.960000000000001e-05,
+      "loss": 2.7053,
+      "step": 10400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.959e-05,
+      "loss": 2.7228,
+      "step": 10410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.958e-05,
+      "loss": 2.7249,
+      "step": 10420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.957e-05,
+      "loss": 2.6788,
+      "step": 10430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.956e-05,
+      "loss": 2.7033,
+      "step": 10440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9550000000000006e-05,
+      "loss": 2.7985,
+      "step": 10450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.954e-05,
+      "loss": 2.7543,
+      "step": 10460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.953e-05,
+      "loss": 2.7816,
+      "step": 10470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.952e-05,
+      "loss": 2.832,
+      "step": 10480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.951e-05,
+      "loss": 2.9134,
+      "step": 10490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9500000000000005e-05,
+      "loss": 2.9084,
+      "step": 10500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9489999999999996e-05,
+      "loss": 2.8395,
+      "step": 10510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.948e-05,
+      "loss": 2.8368,
+      "step": 10520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.947e-05,
+      "loss": 2.9039,
+      "step": 10530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9460000000000005e-05,
+      "loss": 2.8992,
+      "step": 10540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9450000000000003e-05,
+      "loss": 2.8237,
+      "step": 10550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.944e-05,
+      "loss": 2.9338,
+      "step": 10560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.943e-05,
+      "loss": 2.8491,
+      "step": 10570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.942e-05,
+      "loss": 2.6541,
+      "step": 10580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9410000000000004e-05,
+      "loss": 2.6683,
+      "step": 10590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.94e-05,
+      "loss": 2.8077,
+      "step": 10600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.939e-05,
+      "loss": 2.6507,
+      "step": 10610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.938e-05,
+      "loss": 2.8952,
+      "step": 10620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9370000000000004e-05,
+      "loss": 2.7977,
+      "step": 10630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.936e-05,
+      "loss": 2.8023,
+      "step": 10640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.935e-05,
+      "loss": 2.9838,
+      "step": 10650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9340000000000006e-05,
+      "loss": 2.5821,
+      "step": 10660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.933e-05,
+      "loss": 2.7993,
+      "step": 10670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.932e-05,
+      "loss": 2.785,
+      "step": 10680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.931e-05,
+      "loss": 2.8028,
+      "step": 10690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9300000000000007e-05,
+      "loss": 2.9749,
+      "step": 10700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9290000000000005e-05,
+      "loss": 2.8898,
+      "step": 10710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9280000000000003e-05,
+      "loss": 2.7794,
+      "step": 10720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.927e-05,
+      "loss": 3.0213,
+      "step": 10730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.926e-05,
+      "loss": 2.7023,
+      "step": 10740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9250000000000005e-05,
+      "loss": 2.8605,
+      "step": 10750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9240000000000004e-05,
+      "loss": 2.8549,
+      "step": 10760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.923e-05,
+      "loss": 2.8901,
+      "step": 10770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.922e-05,
+      "loss": 2.7848,
+      "step": 10780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.921e-05,
+      "loss": 2.8249,
+      "step": 10790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9200000000000004e-05,
+      "loss": 2.8632,
+      "step": 10800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.919e-05,
+      "loss": 2.7656,
+      "step": 10810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.918e-05,
+      "loss": 2.6846,
+      "step": 10820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.917e-05,
+      "loss": 2.8072,
+      "step": 10830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9160000000000005e-05,
+      "loss": 2.8691,
+      "step": 10840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.915e-05,
+      "loss": 2.7525,
+      "step": 10850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.914e-05,
+      "loss": 2.8316,
+      "step": 10860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.913e-05,
+      "loss": 2.726,
+      "step": 10870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.912e-05,
+      "loss": 2.9107,
+      "step": 10880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.911e-05,
+      "loss": 2.7756,
+      "step": 10890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.91e-05,
+      "loss": 2.8286,
+      "step": 10900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.909000000000001e-05,
+      "loss": 2.7699,
+      "step": 10910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.908e-05,
+      "loss": 2.8121,
+      "step": 10920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9070000000000004e-05,
+      "loss": 2.7864,
+      "step": 10930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.906e-05,
+      "loss": 2.8602,
+      "step": 10940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.905e-05,
+      "loss": 2.7015,
+      "step": 10950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9040000000000006e-05,
+      "loss": 2.7896,
+      "step": 10960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.903e-05,
+      "loss": 2.8724,
+      "step": 10970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.902e-05,
+      "loss": 2.976,
+      "step": 10980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.901e-05,
+      "loss": 2.7864,
+      "step": 10990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 3.0275,
+      "step": 11000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8990000000000004e-05,
+      "loss": 2.971,
+      "step": 11010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.898e-05,
+      "loss": 2.854,
+      "step": 11020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.897e-05,
+      "loss": 2.7263,
+      "step": 11030
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.896e-05,
+      "loss": 2.8803,
+      "step": 11040
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8950000000000005e-05,
+      "loss": 3.0067,
+      "step": 11050
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.894e-05,
+      "loss": 2.9497,
+      "step": 11060
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.893e-05,
+      "loss": 2.7826,
+      "step": 11070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.892e-05,
+      "loss": 2.9559,
+      "step": 11080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8910000000000005e-05,
+      "loss": 2.683,
+      "step": 11090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8900000000000004e-05,
+      "loss": 2.7658,
+      "step": 11100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.889e-05,
+      "loss": 2.9627,
+      "step": 11110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.888e-05,
+      "loss": 2.8375,
+      "step": 11120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.887e-05,
+      "loss": 2.6929,
+      "step": 11130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8860000000000004e-05,
+      "loss": 2.7728,
+      "step": 11140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.885e-05,
+      "loss": 2.7428,
+      "step": 11150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.884e-05,
+      "loss": 2.7527,
+      "step": 11160
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.883e-05,
+      "loss": 2.8053,
+      "step": 11170
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.882e-05,
+      "loss": 2.8631,
+      "step": 11180
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.881e-05,
+      "loss": 2.7828,
+      "step": 11190
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.88e-05,
+      "loss": 2.7482,
+      "step": 11200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8790000000000006e-05,
+      "loss": 2.8255,
+      "step": 11210
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.878e-05,
+      "loss": 2.6559,
+      "step": 11220
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.877e-05,
+      "loss": 2.8019,
+      "step": 11230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.876e-05,
+      "loss": 2.709,
+      "step": 11240
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.875e-05,
+      "loss": 2.7974,
+      "step": 11250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8740000000000005e-05,
+      "loss": 2.8337,
+      "step": 11260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.873e-05,
+      "loss": 2.7359,
+      "step": 11270
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.872e-05,
+      "loss": 2.6966,
+      "step": 11280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.871e-05,
+      "loss": 2.955,
+      "step": 11290
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8700000000000006e-05,
+      "loss": 2.8268,
+      "step": 11300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8690000000000004e-05,
+      "loss": 2.8004,
+      "step": 11310
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.868e-05,
+      "loss": 2.8104,
+      "step": 11320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.867e-05,
+      "loss": 2.8751,
+      "step": 11330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.866e-05,
+      "loss": 2.8125,
+      "step": 11340
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8650000000000004e-05,
+      "loss": 2.6957,
+      "step": 11350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.864e-05,
+      "loss": 2.9878,
+      "step": 11360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.863e-05,
+      "loss": 3.0071,
+      "step": 11370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.862e-05,
+      "loss": 2.5459,
+      "step": 11380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8610000000000005e-05,
+      "loss": 2.8858,
+      "step": 11390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.86e-05,
+      "loss": 2.5167,
+      "step": 11400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.859e-05,
+      "loss": 2.9057,
+      "step": 11410
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.858e-05,
+      "loss": 2.7498,
+      "step": 11420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.857e-05,
+      "loss": 2.82,
+      "step": 11430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8560000000000004e-05,
+      "loss": 2.7653,
+      "step": 11440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.855e-05,
+      "loss": 2.8641,
+      "step": 11450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.854000000000001e-05,
+      "loss": 2.8057,
+      "step": 11460
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.853e-05,
+      "loss": 2.7536,
+      "step": 11470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8520000000000004e-05,
+      "loss": 2.6604,
+      "step": 11480
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.851e-05,
+      "loss": 2.8227,
+      "step": 11490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.85e-05,
+      "loss": 2.8693,
+      "step": 11500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8490000000000006e-05,
+      "loss": 2.6229,
+      "step": 11510
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.848e-05,
+      "loss": 2.7132,
+      "step": 11520
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.847e-05,
+      "loss": 2.7811,
+      "step": 11530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.846e-05,
+      "loss": 2.7038,
+      "step": 11540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.845e-05,
+      "loss": 2.6311,
+      "step": 11550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8440000000000005e-05,
+      "loss": 2.8697,
+      "step": 11560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8429999999999996e-05,
+      "loss": 2.8141,
+      "step": 11570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.842e-05,
+      "loss": 2.5407,
+      "step": 11580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.841e-05,
+      "loss": 2.9019,
+      "step": 11590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8400000000000005e-05,
+      "loss": 2.8707,
+      "step": 11600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8390000000000003e-05,
+      "loss": 2.6427,
+      "step": 11610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.838e-05,
+      "loss": 2.7426,
+      "step": 11620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.837e-05,
+      "loss": 2.7663,
+      "step": 11630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.836e-05,
+      "loss": 2.7092,
+      "step": 11640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8350000000000004e-05,
+      "loss": 2.8337,
+      "step": 11650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.834e-05,
+      "loss": 2.8943,
+      "step": 11660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.833e-05,
+      "loss": 2.8854,
+      "step": 11670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.832e-05,
+      "loss": 2.5257,
+      "step": 11680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8310000000000004e-05,
+      "loss": 2.5287,
+      "step": 11690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.83e-05,
+      "loss": 2.6224,
+      "step": 11700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.829e-05,
+      "loss": 2.718,
+      "step": 11710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.828e-05,
+      "loss": 2.7853,
+      "step": 11720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.827e-05,
+      "loss": 2.7141,
+      "step": 11730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.826e-05,
+      "loss": 2.888,
+      "step": 11740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.825e-05,
+      "loss": 2.6863,
+      "step": 11750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8240000000000007e-05,
+      "loss": 2.6107,
+      "step": 11760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.823e-05,
+      "loss": 2.7796,
+      "step": 11770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.822e-05,
+      "loss": 2.9233,
+      "step": 11780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.821e-05,
+      "loss": 2.5865,
+      "step": 11790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.82e-05,
+      "loss": 2.792,
+      "step": 11800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8190000000000005e-05,
+      "loss": 2.8733,
+      "step": 11810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.818e-05,
+      "loss": 2.825,
+      "step": 11820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.817e-05,
+      "loss": 3.013,
+      "step": 11830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.816e-05,
+      "loss": 2.9865,
+      "step": 11840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8150000000000006e-05,
+      "loss": 2.9172,
+      "step": 11850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8140000000000004e-05,
+      "loss": 2.5708,
+      "step": 11860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8129999999999996e-05,
+      "loss": 2.6821,
+      "step": 11870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.812e-05,
+      "loss": 2.715,
+      "step": 11880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.811e-05,
+      "loss": 2.6825,
+      "step": 11890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8100000000000005e-05,
+      "loss": 2.7224,
+      "step": 11900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.809e-05,
+      "loss": 2.8003,
+      "step": 11910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.808e-05,
+      "loss": 2.5917,
+      "step": 11920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.807e-05,
+      "loss": 2.7149,
+      "step": 11930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.806e-05,
+      "loss": 2.5951,
+      "step": 11940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.805e-05,
+      "loss": 2.5518,
+      "step": 11950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.804e-05,
+      "loss": 2.6437,
+      "step": 11960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.803000000000001e-05,
+      "loss": 2.8274,
+      "step": 11970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.802e-05,
+      "loss": 2.7567,
+      "step": 11980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8010000000000004e-05,
+      "loss": 2.4552,
+      "step": 11990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8e-05,
+      "loss": 2.6057,
+      "step": 12000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.799e-05,
+      "loss": 2.7214,
+      "step": 12010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7980000000000006e-05,
+      "loss": 2.6975,
+      "step": 12020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.797e-05,
+      "loss": 2.8462,
+      "step": 12030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.796e-05,
+      "loss": 2.6298,
+      "step": 12040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.795e-05,
+      "loss": 2.6702,
+      "step": 12050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7940000000000006e-05,
+      "loss": 2.5996,
+      "step": 12060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7930000000000004e-05,
+      "loss": 2.6786,
+      "step": 12070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.792e-05,
+      "loss": 2.7512,
+      "step": 12080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.791e-05,
+      "loss": 2.8113,
+      "step": 12090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.79e-05,
+      "loss": 2.5713,
+      "step": 12100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7890000000000005e-05,
+      "loss": 2.6323,
+      "step": 12110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.788e-05,
+      "loss": 2.7707,
+      "step": 12120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.787e-05,
+      "loss": 2.7979,
+      "step": 12130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.786e-05,
+      "loss": 2.785,
+      "step": 12140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7850000000000005e-05,
+      "loss": 2.6186,
+      "step": 12150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7840000000000004e-05,
+      "loss": 2.8311,
+      "step": 12160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.783e-05,
+      "loss": 2.5887,
+      "step": 12170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.782e-05,
+      "loss": 2.7045,
+      "step": 12180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.781e-05,
+      "loss": 2.5806,
+      "step": 12190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 2.5752,
+      "step": 12200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.779e-05,
+      "loss": 2.5024,
+      "step": 12210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.778000000000001e-05,
+      "loss": 2.4028,
+      "step": 12220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.777e-05,
+      "loss": 2.8412,
+      "step": 12230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.776e-05,
+      "loss": 2.5392,
+      "step": 12240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.775e-05,
+      "loss": 2.8256,
+      "step": 12250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.774e-05,
+      "loss": 2.6206,
+      "step": 12260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7730000000000006e-05,
+      "loss": 2.6976,
+      "step": 12270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.772e-05,
+      "loss": 2.8831,
+      "step": 12280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.771e-05,
+      "loss": 2.9425,
+      "step": 12290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.77e-05,
+      "loss": 2.8803,
+      "step": 12300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.769e-05,
+      "loss": 2.7195,
+      "step": 12310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7680000000000005e-05,
+      "loss": 2.7102,
+      "step": 12320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.767e-05,
+      "loss": 2.8099,
+      "step": 12330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.766e-05,
+      "loss": 2.7494,
+      "step": 12340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.765e-05,
+      "loss": 2.7274,
+      "step": 12350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7640000000000006e-05,
+      "loss": 2.5808,
+      "step": 12360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7630000000000004e-05,
+      "loss": 2.6225,
+      "step": 12370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.762e-05,
+      "loss": 2.7648,
+      "step": 12380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.761e-05,
+      "loss": 2.9792,
+      "step": 12390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.76e-05,
+      "loss": 2.9321,
+      "step": 12400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7590000000000004e-05,
+      "loss": 2.7436,
+      "step": 12410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.758e-05,
+      "loss": 2.5353,
+      "step": 12420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.757e-05,
+      "loss": 2.7439,
+      "step": 12430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.756e-05,
+      "loss": 2.6791,
+      "step": 12440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7550000000000005e-05,
+      "loss": 2.5963,
+      "step": 12450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.754e-05,
+      "loss": 2.5298,
+      "step": 12460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.753e-05,
+      "loss": 2.5764,
+      "step": 12470
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.752e-05,
+      "loss": 2.7176,
+      "step": 12480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.751e-05,
+      "loss": 2.6067,
+      "step": 12490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 2.6043,
+      "step": 12500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.749e-05,
+      "loss": 2.7942,
+      "step": 12510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.748000000000001e-05,
+      "loss": 2.5782,
+      "step": 12520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.747e-05,
+      "loss": 2.6083,
+      "step": 12530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7460000000000004e-05,
+      "loss": 2.5607,
+      "step": 12540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.745e-05,
+      "loss": 2.7342,
+      "step": 12550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.744e-05,
+      "loss": 2.6714,
+      "step": 12560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7430000000000006e-05,
+      "loss": 2.3737,
+      "step": 12570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.742e-05,
+      "loss": 2.7433,
+      "step": 12580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.741e-05,
+      "loss": 2.5314,
+      "step": 12590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.74e-05,
+      "loss": 2.7101,
+      "step": 12600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.739e-05,
+      "loss": 2.483,
+      "step": 12610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7380000000000005e-05,
+      "loss": 2.5809,
+      "step": 12620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7369999999999996e-05,
+      "loss": 2.584,
+      "step": 12630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.736e-05,
+      "loss": 2.6354,
+      "step": 12640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.735e-05,
+      "loss": 2.7126,
+      "step": 12650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7340000000000005e-05,
+      "loss": 2.7396,
+      "step": 12660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7330000000000003e-05,
+      "loss": 2.5299,
+      "step": 12670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.732e-05,
+      "loss": 2.7111,
+      "step": 12680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.731e-05,
+      "loss": 2.5097,
+      "step": 12690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.73e-05,
+      "loss": 2.6153,
+      "step": 12700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7290000000000004e-05,
+      "loss": 2.5833,
+      "step": 12710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.728e-05,
+      "loss": 2.7183,
+      "step": 12720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.727e-05,
+      "loss": 2.4377,
+      "step": 12730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.726e-05,
+      "loss": 2.7027,
+      "step": 12740
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7250000000000004e-05,
+      "loss": 2.8283,
+      "step": 12750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.724e-05,
+      "loss": 2.6935,
+      "step": 12760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.723e-05,
+      "loss": 2.6993,
+      "step": 12770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.722e-05,
+      "loss": 2.6528,
+      "step": 12780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.721e-05,
+      "loss": 2.772,
+      "step": 12790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.72e-05,
+      "loss": 2.7675,
+      "step": 12800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.719e-05,
+      "loss": 2.6969,
+      "step": 12810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7180000000000007e-05,
+      "loss": 2.624,
+      "step": 12820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.717e-05,
+      "loss": 2.7552,
+      "step": 12830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.716e-05,
+      "loss": 2.6592,
+      "step": 12840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.715e-05,
+      "loss": 2.4314,
+      "step": 12850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.714e-05,
+      "loss": 2.5263,
+      "step": 12860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.7130000000000005e-05,
+      "loss": 2.683,
+      "step": 12870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.712e-05,
+      "loss": 2.5565,
+      "step": 12880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.711e-05,
+      "loss": 2.574,
+      "step": 12890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.71e-05,
+      "loss": 2.6472,
+      "step": 12900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.7090000000000006e-05,
+      "loss": 2.6184,
+      "step": 12910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.7080000000000004e-05,
+      "loss": 2.7343,
+      "step": 12920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.707e-05,
+      "loss": 2.4546,
+      "step": 12930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.706e-05,
+      "loss": 2.7048,
+      "step": 12940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.705e-05,
+      "loss": 2.6889,
+      "step": 12950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.7040000000000005e-05,
+      "loss": 2.785,
+      "step": 12960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.703e-05,
+      "loss": 2.6469,
+      "step": 12970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.702e-05,
+      "loss": 2.7096,
+      "step": 12980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.701e-05,
+      "loss": 2.7048,
+      "step": 12990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.7e-05,
+      "loss": 2.433,
+      "step": 13000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.699e-05,
+      "loss": 2.721,
+      "step": 13010
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.698e-05,
+      "loss": 2.5083,
+      "step": 13020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.697e-05,
+      "loss": 2.5648,
+      "step": 13030
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.696e-05,
+      "loss": 2.724,
+      "step": 13040
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6950000000000004e-05,
+      "loss": 2.656,
+      "step": 13050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.694e-05,
+      "loss": 2.7811,
+      "step": 13060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.693e-05,
+      "loss": 2.7614,
+      "step": 13070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.692e-05,
+      "loss": 2.7435,
+      "step": 13080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.691e-05,
+      "loss": 2.5415,
+      "step": 13090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.69e-05,
+      "loss": 2.6825,
+      "step": 13100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.689e-05,
+      "loss": 2.5756,
+      "step": 13110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6880000000000006e-05,
+      "loss": 2.7057,
+      "step": 13120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6870000000000004e-05,
+      "loss": 2.5918,
+      "step": 13130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.686e-05,
+      "loss": 2.5374,
+      "step": 13140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.685e-05,
+      "loss": 2.4937,
+      "step": 13150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.684e-05,
+      "loss": 2.7188,
+      "step": 13160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6830000000000005e-05,
+      "loss": 2.506,
+      "step": 13170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.682e-05,
+      "loss": 2.6311,
+      "step": 13180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.681e-05,
+      "loss": 2.7619,
+      "step": 13190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.68e-05,
+      "loss": 2.7618,
+      "step": 13200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.6790000000000005e-05,
+      "loss": 2.8238,
+      "step": 13210
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.6780000000000004e-05,
+      "loss": 2.5226,
+      "step": 13220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.677e-05,
+      "loss": 2.3743,
+      "step": 13230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.676e-05,
+      "loss": 2.7828,
+      "step": 13240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.675e-05,
+      "loss": 2.7767,
+      "step": 13250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.6740000000000004e-05,
+      "loss": 2.7773,
+      "step": 13260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.673e-05,
+      "loss": 2.7058,
+      "step": 13270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.672000000000001e-05,
+      "loss": 2.5345,
+      "step": 13280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.671e-05,
+      "loss": 2.7118,
+      "step": 13290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.6700000000000004e-05,
+      "loss": 2.6941,
+      "step": 13300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.669e-05,
+      "loss": 2.6781,
+      "step": 13310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.668e-05,
+      "loss": 2.5794,
+      "step": 13320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.6670000000000006e-05,
+      "loss": 2.6094,
+      "step": 13330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.666e-05,
+      "loss": 2.6299,
+      "step": 13340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.665e-05,
+      "loss": 2.7205,
+      "step": 13350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.664e-05,
+      "loss": 2.6013,
+      "step": 13360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.663e-05,
+      "loss": 2.5503,
+      "step": 13370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.6620000000000005e-05,
+      "loss": 2.7625,
+      "step": 13380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.661e-05,
+      "loss": 2.752,
+      "step": 13390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.66e-05,
+      "loss": 2.6976,
+      "step": 13400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.659e-05,
+      "loss": 2.4048,
+      "step": 13410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.6580000000000006e-05,
+      "loss": 2.7544,
+      "step": 13420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.6570000000000004e-05,
+      "loss": 2.5601,
+      "step": 13430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.656e-05,
+      "loss": 2.638,
+      "step": 13440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.655e-05,
+      "loss": 2.4951,
+      "step": 13450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.654e-05,
+      "loss": 2.5843,
+      "step": 13460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.6530000000000004e-05,
+      "loss": 2.6631,
+      "step": 13470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.652e-05,
+      "loss": 2.7554,
+      "step": 13480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.651e-05,
+      "loss": 2.6108,
+      "step": 13490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.65e-05,
+      "loss": 2.7536,
+      "step": 13500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.6490000000000005e-05,
+      "loss": 2.8123,
+      "step": 13510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.648e-05,
+      "loss": 2.5569,
+      "step": 13520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.647e-05,
+      "loss": 2.5828,
+      "step": 13530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.646e-05,
+      "loss": 2.6888,
+      "step": 13540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.645e-05,
+      "loss": 2.72,
+      "step": 13550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.6440000000000003e-05,
+      "loss": 2.3896,
+      "step": 13560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.643e-05,
+      "loss": 2.4704,
+      "step": 13570
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.642000000000001e-05,
+      "loss": 2.5637,
+      "step": 13580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.641e-05,
+      "loss": 2.5671,
+      "step": 13590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.6400000000000004e-05,
+      "loss": 2.7944,
+      "step": 13600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.639e-05,
+      "loss": 2.9435,
+      "step": 13610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.638e-05,
+      "loss": 2.5653,
+      "step": 13620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.6370000000000006e-05,
+      "loss": 2.5742,
+      "step": 13630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.636e-05,
+      "loss": 2.6917,
+      "step": 13640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.635e-05,
+      "loss": 2.5824,
+      "step": 13650
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.634e-05,
+      "loss": 2.5287,
+      "step": 13660
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.6330000000000006e-05,
+      "loss": 2.5605,
+      "step": 13670
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.6320000000000005e-05,
+      "loss": 2.4991,
+      "step": 13680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.6309999999999996e-05,
+      "loss": 2.6383,
+      "step": 13690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.63e-05,
+      "loss": 2.6967,
+      "step": 13700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.629e-05,
+      "loss": 2.6952,
+      "step": 13710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.6280000000000005e-05,
+      "loss": 2.8692,
+      "step": 13720
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.6270000000000003e-05,
+      "loss": 2.8015,
+      "step": 13730
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.626e-05,
+      "loss": 2.5962,
+      "step": 13740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.625e-05,
+      "loss": 2.6648,
+      "step": 13750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.624e-05,
+      "loss": 2.5614,
+      "step": 13760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.6230000000000004e-05,
+      "loss": 2.747,
+      "step": 13770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.622e-05,
+      "loss": 2.6372,
+      "step": 13780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.621e-05,
+      "loss": 2.6186,
+      "step": 13790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.62e-05,
+      "loss": 2.596,
+      "step": 13800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.6190000000000004e-05,
+      "loss": 2.4255,
+      "step": 13810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.618e-05,
+      "loss": 2.5903,
+      "step": 13820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.617e-05,
+      "loss": 2.4543,
+      "step": 13830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.616e-05,
+      "loss": 2.779,
+      "step": 13840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.615e-05,
+      "loss": 2.5935,
+      "step": 13850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.614e-05,
+      "loss": 2.6253,
+      "step": 13860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.613e-05,
+      "loss": 2.6635,
+      "step": 13870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.6120000000000007e-05,
+      "loss": 2.6575,
+      "step": 13880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.611e-05,
+      "loss": 2.6639,
+      "step": 13890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.61e-05,
+      "loss": 2.4902,
+      "step": 13900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.609e-05,
+      "loss": 2.5646,
+      "step": 13910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.608e-05,
+      "loss": 2.4878,
+      "step": 13920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.6070000000000005e-05,
+      "loss": 2.439,
+      "step": 13930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.606e-05,
+      "loss": 2.3793,
+      "step": 13940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.605e-05,
+      "loss": 2.511,
+      "step": 13950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.604e-05,
+      "loss": 2.6061,
+      "step": 13960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.6030000000000006e-05,
+      "loss": 2.6459,
+      "step": 13970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.6020000000000004e-05,
+      "loss": 2.5779,
+      "step": 13980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.601e-05,
+      "loss": 2.7534,
+      "step": 13990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.6e-05,
+      "loss": 2.4975,
+      "step": 14000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.599e-05,
+      "loss": 2.5948,
+      "step": 14010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.5980000000000004e-05,
+      "loss": 2.6036,
+      "step": 14020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.597e-05,
+      "loss": 2.6794,
+      "step": 14030
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.596e-05,
+      "loss": 2.4862,
+      "step": 14040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.595e-05,
+      "loss": 2.5064,
+      "step": 14050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.594e-05,
+      "loss": 2.5664,
+      "step": 14060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.593e-05,
+      "loss": 2.5933,
+      "step": 14070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.592e-05,
+      "loss": 2.3969,
+      "step": 14080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.591e-05,
+      "loss": 2.3456,
+      "step": 14090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.59e-05,
+      "loss": 2.4281,
+      "step": 14100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.5890000000000004e-05,
+      "loss": 2.6433,
+      "step": 14110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.588e-05,
+      "loss": 2.496,
+      "step": 14120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.587e-05,
+      "loss": 2.7259,
+      "step": 14130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.586e-05,
+      "loss": 2.6197,
+      "step": 14140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.585e-05,
+      "loss": 2.5849,
+      "step": 14150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.584e-05,
+      "loss": 2.4639,
+      "step": 14160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.583e-05,
+      "loss": 2.4608,
+      "step": 14170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.5820000000000006e-05,
+      "loss": 2.628,
+      "step": 14180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.581e-05,
+      "loss": 2.5823,
+      "step": 14190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.58e-05,
+      "loss": 2.5831,
+      "step": 14200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.579e-05,
+      "loss": 2.5613,
+      "step": 14210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.578e-05,
+      "loss": 2.6,
+      "step": 14220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.5770000000000005e-05,
+      "loss": 2.5232,
+      "step": 14230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.5759999999999996e-05,
+      "loss": 2.5788,
+      "step": 14240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.575e-05,
+      "loss": 2.4726,
+      "step": 14250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.574e-05,
+      "loss": 2.5007,
+      "step": 14260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.5730000000000005e-05,
+      "loss": 2.3794,
+      "step": 14270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.5720000000000004e-05,
+      "loss": 2.5953,
+      "step": 14280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.571e-05,
+      "loss": 2.5358,
+      "step": 14290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.57e-05,
+      "loss": 2.4394,
+      "step": 14300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.569e-05,
+      "loss": 2.4275,
+      "step": 14310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.5680000000000004e-05,
+      "loss": 2.5395,
+      "step": 14320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.567e-05,
+      "loss": 2.733,
+      "step": 14330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.566e-05,
+      "loss": 2.6425,
+      "step": 14340
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.565e-05,
+      "loss": 2.4429,
+      "step": 14350
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.5640000000000004e-05,
+      "loss": 2.4657,
+      "step": 14360
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.563e-05,
+      "loss": 2.3837,
+      "step": 14370
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.562e-05,
+      "loss": 2.5623,
+      "step": 14380
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.5610000000000006e-05,
+      "loss": 2.5704,
+      "step": 14390
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.56e-05,
+      "loss": 2.5566,
+      "step": 14400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.559e-05,
+      "loss": 2.6516,
+      "step": 14410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.558e-05,
+      "loss": 2.2844,
+      "step": 14420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.557e-05,
+      "loss": 2.5483,
+      "step": 14430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.5560000000000005e-05,
+      "loss": 2.5478,
+      "step": 14440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.555e-05,
+      "loss": 2.5289,
+      "step": 14450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.554e-05,
+      "loss": 2.6042,
+      "step": 14460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.553e-05,
+      "loss": 2.4913,
+      "step": 14470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.5520000000000006e-05,
+      "loss": 2.5877,
+      "step": 14480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.5510000000000004e-05,
+      "loss": 2.7788,
+      "step": 14490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.55e-05,
+      "loss": 2.7201,
+      "step": 14500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.549e-05,
+      "loss": 2.5263,
+      "step": 14510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.548e-05,
+      "loss": 2.623,
+      "step": 14520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.5470000000000004e-05,
+      "loss": 2.5747,
+      "step": 14530
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.546e-05,
+      "loss": 2.4484,
+      "step": 14540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.545e-05,
+      "loss": 2.5042,
+      "step": 14550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.544e-05,
+      "loss": 2.4662,
+      "step": 14560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.5430000000000005e-05,
+      "loss": 2.5222,
+      "step": 14570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.542e-05,
+      "loss": 2.6546,
+      "step": 14580
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.541e-05,
+      "loss": 2.634,
+      "step": 14590
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.54e-05,
+      "loss": 2.4217,
+      "step": 14600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.539e-05,
+      "loss": 2.5515,
+      "step": 14610
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5380000000000003e-05,
+      "loss": 2.4434,
+      "step": 14620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.537e-05,
+      "loss": 2.5039,
+      "step": 14630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.536000000000001e-05,
+      "loss": 2.4352,
+      "step": 14640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.535e-05,
+      "loss": 2.3898,
+      "step": 14650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5340000000000004e-05,
+      "loss": 2.7212,
+      "step": 14660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.533e-05,
+      "loss": 2.5355,
+      "step": 14670
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.532e-05,
+      "loss": 2.717,
+      "step": 14680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5310000000000006e-05,
+      "loss": 2.4337,
+      "step": 14690
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.53e-05,
+      "loss": 2.4829,
+      "step": 14700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.529e-05,
+      "loss": 2.4421,
+      "step": 14710
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.528e-05,
+      "loss": 2.4276,
+      "step": 14720
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5270000000000006e-05,
+      "loss": 2.6666,
+      "step": 14730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5260000000000005e-05,
+      "loss": 2.518,
+      "step": 14740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.525e-05,
+      "loss": 2.5606,
+      "step": 14750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.524e-05,
+      "loss": 2.5667,
+      "step": 14760
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.523e-05,
+      "loss": 2.4442,
+      "step": 14770
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5220000000000005e-05,
+      "loss": 2.5122,
+      "step": 14780
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5210000000000003e-05,
+      "loss": 2.6278,
+      "step": 14790
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.52e-05,
+      "loss": 2.3534,
+      "step": 14800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.519e-05,
+      "loss": 2.5814,
+      "step": 14810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.518e-05,
+      "loss": 2.4766,
+      "step": 14820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5170000000000004e-05,
+      "loss": 2.4365,
+      "step": 14830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.516e-05,
+      "loss": 2.6753,
+      "step": 14840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.515e-05,
+      "loss": 2.6517,
+      "step": 14850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.514e-05,
+      "loss": 2.4637,
+      "step": 14860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5130000000000004e-05,
+      "loss": 2.6287,
+      "step": 14870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.512e-05,
+      "loss": 2.5007,
+      "step": 14880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.511e-05,
+      "loss": 2.6098,
+      "step": 14890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.51e-05,
+      "loss": 2.3769,
+      "step": 14900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.509e-05,
+      "loss": 2.3857,
+      "step": 14910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.508e-05,
+      "loss": 2.3594,
+      "step": 14920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.507e-05,
+      "loss": 2.306,
+      "step": 14930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.5060000000000007e-05,
+      "loss": 2.5282,
+      "step": 14940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.505e-05,
+      "loss": 2.4343,
+      "step": 14950
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.504e-05,
+      "loss": 2.4153,
+      "step": 14960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.503e-05,
+      "loss": 2.3135,
+      "step": 14970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.502e-05,
+      "loss": 2.5415,
+      "step": 14980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5010000000000005e-05,
+      "loss": 2.489,
+      "step": 14990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.5e-05,
+      "loss": 2.629,
+      "step": 15000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.499e-05,
+      "loss": 2.4217,
+      "step": 15010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.498e-05,
+      "loss": 2.354,
+      "step": 15020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4970000000000006e-05,
+      "loss": 2.4704,
+      "step": 15030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4960000000000004e-05,
+      "loss": 2.4545,
+      "step": 15040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.495e-05,
+      "loss": 2.2453,
+      "step": 15050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.494e-05,
+      "loss": 2.446,
+      "step": 15060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.493e-05,
+      "loss": 2.4564,
+      "step": 15070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4920000000000004e-05,
+      "loss": 2.3244,
+      "step": 15080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.491e-05,
+      "loss": 2.5621,
+      "step": 15090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.49e-05,
+      "loss": 2.5686,
+      "step": 15100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.489e-05,
+      "loss": 2.3068,
+      "step": 15110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4880000000000005e-05,
+      "loss": 2.3365,
+      "step": 15120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.487e-05,
+      "loss": 2.3542,
+      "step": 15130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.486e-05,
+      "loss": 2.5003,
+      "step": 15140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.485e-05,
+      "loss": 2.4588,
+      "step": 15150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.484e-05,
+      "loss": 2.4098,
+      "step": 15160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4830000000000004e-05,
+      "loss": 2.397,
+      "step": 15170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.482e-05,
+      "loss": 2.3901,
+      "step": 15180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.481e-05,
+      "loss": 2.3694,
+      "step": 15190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.48e-05,
+      "loss": 2.4854,
+      "step": 15200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.479e-05,
+      "loss": 2.3339,
+      "step": 15210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.478e-05,
+      "loss": 2.365,
+      "step": 15220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.477e-05,
+      "loss": 2.3803,
+      "step": 15230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4760000000000006e-05,
+      "loss": 2.3439,
+      "step": 15240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.475e-05,
+      "loss": 2.5294,
+      "step": 15250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.474e-05,
+      "loss": 2.391,
+      "step": 15260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.473e-05,
+      "loss": 2.5305,
+      "step": 15270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.472e-05,
+      "loss": 2.215,
+      "step": 15280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.4710000000000005e-05,
+      "loss": 2.4003,
+      "step": 15290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4699999999999996e-05,
+      "loss": 2.3719,
+      "step": 15300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.469e-05,
+      "loss": 2.5319,
+      "step": 15310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.468e-05,
+      "loss": 2.4811,
+      "step": 15320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4670000000000005e-05,
+      "loss": 2.4521,
+      "step": 15330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4660000000000004e-05,
+      "loss": 2.6166,
+      "step": 15340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.465e-05,
+      "loss": 2.602,
+      "step": 15350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.464e-05,
+      "loss": 2.4606,
+      "step": 15360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.463e-05,
+      "loss": 2.4517,
+      "step": 15370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4620000000000004e-05,
+      "loss": 2.6067,
+      "step": 15380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.461e-05,
+      "loss": 2.4625,
+      "step": 15390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.46e-05,
+      "loss": 2.5576,
+      "step": 15400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.459e-05,
+      "loss": 2.2396,
+      "step": 15410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4580000000000004e-05,
+      "loss": 2.4437,
+      "step": 15420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.457e-05,
+      "loss": 2.2068,
+      "step": 15430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.456e-05,
+      "loss": 2.3614,
+      "step": 15440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.455e-05,
+      "loss": 2.767,
+      "step": 15450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.454e-05,
+      "loss": 2.3554,
+      "step": 15460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.453e-05,
+      "loss": 2.4391,
+      "step": 15470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.452e-05,
+      "loss": 2.4812,
+      "step": 15480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.451000000000001e-05,
+      "loss": 2.4431,
+      "step": 15490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.45e-05,
+      "loss": 2.3748,
+      "step": 15500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.449e-05,
+      "loss": 2.2781,
+      "step": 15510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.448e-05,
+      "loss": 2.5714,
+      "step": 15520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.447e-05,
+      "loss": 2.3346,
+      "step": 15530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4460000000000005e-05,
+      "loss": 2.3102,
+      "step": 15540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.445e-05,
+      "loss": 2.35,
+      "step": 15550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.444e-05,
+      "loss": 2.2049,
+      "step": 15560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.443e-05,
+      "loss": 2.5271,
+      "step": 15570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.442e-05,
+      "loss": 2.3706,
+      "step": 15580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4410000000000004e-05,
+      "loss": 2.439,
+      "step": 15590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4399999999999996e-05,
+      "loss": 2.4348,
+      "step": 15600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.439e-05,
+      "loss": 2.3294,
+      "step": 15610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.438e-05,
+      "loss": 2.3488,
+      "step": 15620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4370000000000005e-05,
+      "loss": 2.495,
+      "step": 15630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.436e-05,
+      "loss": 2.5397,
+      "step": 15640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.435e-05,
+      "loss": 2.3677,
+      "step": 15650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.434e-05,
+      "loss": 2.5199,
+      "step": 15660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.433e-05,
+      "loss": 2.5308,
+      "step": 15670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.4320000000000003e-05,
+      "loss": 2.3485,
+      "step": 15680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.431e-05,
+      "loss": 2.6083,
+      "step": 15690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.430000000000001e-05,
+      "loss": 2.4095,
+      "step": 15700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.429e-05,
+      "loss": 2.3732,
+      "step": 15710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.4280000000000004e-05,
+      "loss": 2.4285,
+      "step": 15720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.427e-05,
+      "loss": 2.541,
+      "step": 15730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.426e-05,
+      "loss": 2.3,
+      "step": 15740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.4250000000000006e-05,
+      "loss": 2.4785,
+      "step": 15750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.424e-05,
+      "loss": 2.3457,
+      "step": 15760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.423e-05,
+      "loss": 2.6211,
+      "step": 15770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.422e-05,
+      "loss": 2.3889,
+      "step": 15780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.4210000000000006e-05,
+      "loss": 2.447,
+      "step": 15790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.4200000000000005e-05,
+      "loss": 2.4569,
+      "step": 15800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.419e-05,
+      "loss": 2.3407,
+      "step": 15810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.418e-05,
+      "loss": 2.4357,
+      "step": 15820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.417e-05,
+      "loss": 2.3959,
+      "step": 15830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.4160000000000005e-05,
+      "loss": 2.4728,
+      "step": 15840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.415e-05,
+      "loss": 2.2934,
+      "step": 15850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.414e-05,
+      "loss": 2.457,
+      "step": 15860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.413e-05,
+      "loss": 2.4873,
+      "step": 15870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.412e-05,
+      "loss": 2.3062,
+      "step": 15880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.4110000000000004e-05,
+      "loss": 2.4393,
+      "step": 15890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.41e-05,
+      "loss": 2.4724,
+      "step": 15900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.409e-05,
+      "loss": 2.2744,
+      "step": 15910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.408e-05,
+      "loss": 2.5205,
+      "step": 15920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.4070000000000004e-05,
+      "loss": 2.4898,
+      "step": 15930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.406e-05,
+      "loss": 2.3401,
+      "step": 15940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.405e-05,
+      "loss": 2.1853,
+      "step": 15950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.404e-05,
+      "loss": 2.4002,
+      "step": 15960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.403e-05,
+      "loss": 2.4955,
+      "step": 15970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.402e-05,
+      "loss": 2.3759,
+      "step": 15980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.401e-05,
+      "loss": 2.3783,
+      "step": 15990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 2.5848,
+      "step": 16000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.399e-05,
+      "loss": 2.261,
+      "step": 16010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.398e-05,
+      "loss": 2.4567,
+      "step": 16020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.397e-05,
+      "loss": 2.251,
+      "step": 16030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.396e-05,
+      "loss": 2.3596,
+      "step": 16040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.3950000000000005e-05,
+      "loss": 2.5022,
+      "step": 16050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.394e-05,
+      "loss": 2.2384,
+      "step": 16060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.393e-05,
+      "loss": 2.4885,
+      "step": 16070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.392e-05,
+      "loss": 2.3548,
+      "step": 16080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.3910000000000006e-05,
+      "loss": 2.4365,
+      "step": 16090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.3900000000000004e-05,
+      "loss": 2.5252,
+      "step": 16100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.389e-05,
+      "loss": 2.4402,
+      "step": 16110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.388e-05,
+      "loss": 2.4608,
+      "step": 16120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.387e-05,
+      "loss": 2.4777,
+      "step": 16130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.3860000000000004e-05,
+      "loss": 2.3566,
+      "step": 16140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.385e-05,
+      "loss": 2.2542,
+      "step": 16150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.384e-05,
+      "loss": 2.4082,
+      "step": 16160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.383e-05,
+      "loss": 2.3047,
+      "step": 16170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.3820000000000005e-05,
+      "loss": 2.4877,
+      "step": 16180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.381e-05,
+      "loss": 2.4721,
+      "step": 16190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.38e-05,
+      "loss": 2.3086,
+      "step": 16200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.379e-05,
+      "loss": 2.3835,
+      "step": 16210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.378e-05,
+      "loss": 2.2365,
+      "step": 16220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.3770000000000004e-05,
+      "loss": 2.3181,
+      "step": 16230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.376e-05,
+      "loss": 2.3747,
+      "step": 16240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.375000000000001e-05,
+      "loss": 2.4445,
+      "step": 16250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.374e-05,
+      "loss": 2.4439,
+      "step": 16260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.373e-05,
+      "loss": 2.3841,
+      "step": 16270
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.372e-05,
+      "loss": 2.262,
+      "step": 16280
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.371e-05,
+      "loss": 2.3319,
+      "step": 16290
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.3700000000000006e-05,
+      "loss": 2.5897,
+      "step": 16300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.369e-05,
+      "loss": 2.3816,
+      "step": 16310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.368e-05,
+      "loss": 2.4306,
+      "step": 16320
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.367e-05,
+      "loss": 2.5015,
+      "step": 16330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.366e-05,
+      "loss": 2.3789,
+      "step": 16340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.3650000000000005e-05,
+      "loss": 2.315,
+      "step": 16350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.3639999999999996e-05,
+      "loss": 2.264,
+      "step": 16360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.363e-05,
+      "loss": 2.3271,
+      "step": 16370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.362e-05,
+      "loss": 2.3605,
+      "step": 16380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.3610000000000005e-05,
+      "loss": 2.2937,
+      "step": 16390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.3600000000000004e-05,
+      "loss": 2.4009,
+      "step": 16400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.359e-05,
+      "loss": 2.2928,
+      "step": 16410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.358e-05,
+      "loss": 2.4688,
+      "step": 16420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.357e-05,
+      "loss": 2.2971,
+      "step": 16430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.3560000000000004e-05,
+      "loss": 2.4,
+      "step": 16440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.355e-05,
+      "loss": 2.4584,
+      "step": 16450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.354e-05,
+      "loss": 2.2586,
+      "step": 16460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.353e-05,
+      "loss": 2.2278,
+      "step": 16470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.3520000000000004e-05,
+      "loss": 2.3663,
+      "step": 16480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.351e-05,
+      "loss": 2.1613,
+      "step": 16490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.35e-05,
+      "loss": 2.2759,
+      "step": 16500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.349e-05,
+      "loss": 2.0191,
+      "step": 16510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.348e-05,
+      "loss": 2.2429,
+      "step": 16520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.347e-05,
+      "loss": 2.4235,
+      "step": 16530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.346e-05,
+      "loss": 2.3868,
+      "step": 16540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.345000000000001e-05,
+      "loss": 2.3624,
+      "step": 16550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.344e-05,
+      "loss": 2.3687,
+      "step": 16560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.3430000000000003e-05,
+      "loss": 2.1474,
+      "step": 16570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.342e-05,
+      "loss": 2.3532,
+      "step": 16580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.341e-05,
+      "loss": 2.3707,
+      "step": 16590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.3400000000000005e-05,
+      "loss": 2.2385,
+      "step": 16600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.339e-05,
+      "loss": 2.2935,
+      "step": 16610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.338e-05,
+      "loss": 2.284,
+      "step": 16620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.337e-05,
+      "loss": 2.1678,
+      "step": 16630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.336e-05,
+      "loss": 2.4149,
+      "step": 16640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.3350000000000004e-05,
+      "loss": 2.3175,
+      "step": 16650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.3339999999999996e-05,
+      "loss": 2.3596,
+      "step": 16660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.333e-05,
+      "loss": 2.4324,
+      "step": 16670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.332e-05,
+      "loss": 2.1555,
+      "step": 16680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.3310000000000005e-05,
+      "loss": 2.4658,
+      "step": 16690
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.33e-05,
+      "loss": 2.3866,
+      "step": 16700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.329e-05,
+      "loss": 2.3832,
+      "step": 16710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.328e-05,
+      "loss": 2.2385,
+      "step": 16720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.327e-05,
+      "loss": 2.3336,
+      "step": 16730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.3260000000000003e-05,
+      "loss": 2.4136,
+      "step": 16740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.325e-05,
+      "loss": 2.3729,
+      "step": 16750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.324e-05,
+      "loss": 2.3012,
+      "step": 16760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.323e-05,
+      "loss": 2.3478,
+      "step": 16770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.3220000000000004e-05,
+      "loss": 2.2283,
+      "step": 16780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.321e-05,
+      "loss": 2.4687,
+      "step": 16790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.32e-05,
+      "loss": 2.2569,
+      "step": 16800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.319e-05,
+      "loss": 2.3569,
+      "step": 16810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.318e-05,
+      "loss": 2.5009,
+      "step": 16820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.317e-05,
+      "loss": 2.2943,
+      "step": 16830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.316e-05,
+      "loss": 2.2707,
+      "step": 16840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.3150000000000006e-05,
+      "loss": 2.3881,
+      "step": 16850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.314e-05,
+      "loss": 2.3448,
+      "step": 16860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.313e-05,
+      "loss": 2.3951,
+      "step": 16870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.312e-05,
+      "loss": 2.2029,
+      "step": 16880
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.311e-05,
+      "loss": 2.3396,
+      "step": 16890
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.3100000000000005e-05,
+      "loss": 2.4099,
+      "step": 16900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.309e-05,
+      "loss": 2.3602,
+      "step": 16910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.308e-05,
+      "loss": 2.2748,
+      "step": 16920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.307e-05,
+      "loss": 2.2737,
+      "step": 16930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.3060000000000005e-05,
+      "loss": 2.4477,
+      "step": 16940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.3050000000000004e-05,
+      "loss": 2.4816,
+      "step": 16950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.304e-05,
+      "loss": 2.3719,
+      "step": 16960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.303e-05,
+      "loss": 2.3088,
+      "step": 16970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.302e-05,
+      "loss": 2.3919,
+      "step": 16980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.3010000000000004e-05,
+      "loss": 2.417,
+      "step": 16990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.3e-05,
+      "loss": 2.1801,
+      "step": 17000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.299e-05,
+      "loss": 2.3389,
+      "step": 17010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.298e-05,
+      "loss": 2.3909,
+      "step": 17020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.297e-05,
+      "loss": 2.3705,
+      "step": 17030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.296e-05,
+      "loss": 2.3833,
+      "step": 17040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.295e-05,
+      "loss": 2.2904,
+      "step": 17050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.2940000000000006e-05,
+      "loss": 2.382,
+      "step": 17060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.293e-05,
+      "loss": 2.1668,
+      "step": 17070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.292e-05,
+      "loss": 2.1872,
+      "step": 17080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.291e-05,
+      "loss": 2.4307,
+      "step": 17090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.29e-05,
+      "loss": 2.4058,
+      "step": 17100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.2890000000000005e-05,
+      "loss": 2.3001,
+      "step": 17110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.288e-05,
+      "loss": 2.291,
+      "step": 17120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.287e-05,
+      "loss": 2.1656,
+      "step": 17130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.286e-05,
+      "loss": 2.4873,
+      "step": 17140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.2850000000000006e-05,
+      "loss": 2.1937,
+      "step": 17150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.2840000000000004e-05,
+      "loss": 2.4061,
+      "step": 17160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.283e-05,
+      "loss": 2.4479,
+      "step": 17170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.282e-05,
+      "loss": 2.3261,
+      "step": 17180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.281e-05,
+      "loss": 2.147,
+      "step": 17190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.2800000000000004e-05,
+      "loss": 2.32,
+      "step": 17200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.279e-05,
+      "loss": 2.3351,
+      "step": 17210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.278e-05,
+      "loss": 2.3287,
+      "step": 17220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.277e-05,
+      "loss": 2.173,
+      "step": 17230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.2760000000000005e-05,
+      "loss": 2.2279,
+      "step": 17240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.275e-05,
+      "loss": 2.3007,
+      "step": 17250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.274e-05,
+      "loss": 2.1808,
+      "step": 17260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.273e-05,
+      "loss": 2.153,
+      "step": 17270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.272e-05,
+      "loss": 2.1991,
+      "step": 17280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.2710000000000004e-05,
+      "loss": 2.2158,
+      "step": 17290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.27e-05,
+      "loss": 2.3353,
+      "step": 17300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.269000000000001e-05,
+      "loss": 2.3362,
+      "step": 17310
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.268e-05,
+      "loss": 2.3612,
+      "step": 17320
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.267e-05,
+      "loss": 2.2792,
+      "step": 17330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.266e-05,
+      "loss": 2.4084,
+      "step": 17340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.265e-05,
+      "loss": 2.3105,
+      "step": 17350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.2640000000000006e-05,
+      "loss": 2.2666,
+      "step": 17360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.263e-05,
+      "loss": 2.4154,
+      "step": 17370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.262e-05,
+      "loss": 2.2347,
+      "step": 17380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.261e-05,
+      "loss": 2.3233,
+      "step": 17390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.26e-05,
+      "loss": 2.2788,
+      "step": 17400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.2590000000000005e-05,
+      "loss": 2.2282,
+      "step": 17410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.2579999999999996e-05,
+      "loss": 2.3992,
+      "step": 17420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.257e-05,
+      "loss": 2.432,
+      "step": 17430
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.256e-05,
+      "loss": 2.1232,
+      "step": 17440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.2550000000000005e-05,
+      "loss": 2.5051,
+      "step": 17450
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.2540000000000004e-05,
+      "loss": 2.2906,
+      "step": 17460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.253e-05,
+      "loss": 2.4099,
+      "step": 17470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.252e-05,
+      "loss": 2.3253,
+      "step": 17480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.251e-05,
+      "loss": 2.2278,
+      "step": 17490
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 2.2402,
+      "step": 17500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.249e-05,
+      "loss": 2.181,
+      "step": 17510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.248e-05,
+      "loss": 2.235,
+      "step": 17520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.247e-05,
+      "loss": 2.4027,
+      "step": 17530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.2460000000000004e-05,
+      "loss": 2.2085,
+      "step": 17540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.245e-05,
+      "loss": 2.2006,
+      "step": 17550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.244e-05,
+      "loss": 2.2804,
+      "step": 17560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.243e-05,
+      "loss": 2.3271,
+      "step": 17570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.242e-05,
+      "loss": 2.2326,
+      "step": 17580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.241e-05,
+      "loss": 2.334,
+      "step": 17590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.24e-05,
+      "loss": 2.1472,
+      "step": 17600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.239000000000001e-05,
+      "loss": 2.1343,
+      "step": 17610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.238e-05,
+      "loss": 2.2088,
+      "step": 17620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.2370000000000003e-05,
+      "loss": 2.2009,
+      "step": 17630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.236e-05,
+      "loss": 2.1441,
+      "step": 17640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.235e-05,
+      "loss": 2.2498,
+      "step": 17650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.2340000000000005e-05,
+      "loss": 2.2777,
+      "step": 17660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.233e-05,
+      "loss": 2.3353,
+      "step": 17670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.232e-05,
+      "loss": 2.1884,
+      "step": 17680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.231e-05,
+      "loss": 2.1649,
+      "step": 17690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.2300000000000006e-05,
+      "loss": 2.1889,
+      "step": 17700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.2290000000000004e-05,
+      "loss": 2.1771,
+      "step": 17710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.2279999999999996e-05,
+      "loss": 2.2834,
+      "step": 17720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.227e-05,
+      "loss": 2.3113,
+      "step": 17730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.226e-05,
+      "loss": 2.269,
+      "step": 17740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.2250000000000005e-05,
+      "loss": 2.2981,
+      "step": 17750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.224e-05,
+      "loss": 2.2942,
+      "step": 17760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.223e-05,
+      "loss": 2.2297,
+      "step": 17770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.222e-05,
+      "loss": 2.2011,
+      "step": 17780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.221e-05,
+      "loss": 1.9458,
+      "step": 17790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.2200000000000003e-05,
+      "loss": 2.4189,
+      "step": 17800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.219e-05,
+      "loss": 2.3158,
+      "step": 17810
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.218e-05,
+      "loss": 2.0334,
+      "step": 17820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.217e-05,
+      "loss": 2.3181,
+      "step": 17830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.2160000000000004e-05,
+      "loss": 2.285,
+      "step": 17840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.215e-05,
+      "loss": 2.2372,
+      "step": 17850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.214e-05,
+      "loss": 2.2422,
+      "step": 17860
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.213e-05,
+      "loss": 2.2028,
+      "step": 17870
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.212e-05,
+      "loss": 2.2367,
+      "step": 17880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.211e-05,
+      "loss": 2.1921,
+      "step": 17890
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.21e-05,
+      "loss": 2.3523,
+      "step": 17900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.2090000000000006e-05,
+      "loss": 2.3996,
+      "step": 17910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.208e-05,
+      "loss": 2.2247,
+      "step": 17920
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.207e-05,
+      "loss": 2.2546,
+      "step": 17930
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.206e-05,
+      "loss": 2.0049,
+      "step": 17940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.205e-05,
+      "loss": 2.324,
+      "step": 17950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.2040000000000005e-05,
+      "loss": 2.3109,
+      "step": 17960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.2029999999999997e-05,
+      "loss": 2.4441,
+      "step": 17970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.202e-05,
+      "loss": 2.2172,
+      "step": 17980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.201e-05,
+      "loss": 2.3533,
+      "step": 17990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 2.1671,
+      "step": 18000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.1990000000000004e-05,
+      "loss": 2.3543,
+      "step": 18010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.198e-05,
+      "loss": 2.1031,
+      "step": 18020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.197e-05,
+      "loss": 2.5012,
+      "step": 18030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.196e-05,
+      "loss": 2.4286,
+      "step": 18040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.1950000000000004e-05,
+      "loss": 2.2222,
+      "step": 18050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.194e-05,
+      "loss": 2.4215,
+      "step": 18060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.193e-05,
+      "loss": 2.3948,
+      "step": 18070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.192e-05,
+      "loss": 2.2038,
+      "step": 18080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.191e-05,
+      "loss": 2.2063,
+      "step": 18090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.19e-05,
+      "loss": 2.1409,
+      "step": 18100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.189e-05,
+      "loss": 2.177,
+      "step": 18110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.188e-05,
+      "loss": 2.1898,
+      "step": 18120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.187e-05,
+      "loss": 2.2849,
+      "step": 18130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.186e-05,
+      "loss": 2.103,
+      "step": 18140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.185e-05,
+      "loss": 2.3184,
+      "step": 18150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.184e-05,
+      "loss": 2.1983,
+      "step": 18160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.1830000000000005e-05,
+      "loss": 2.1582,
+      "step": 18170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.182e-05,
+      "loss": 2.2457,
+      "step": 18180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.181e-05,
+      "loss": 2.3134,
+      "step": 18190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.18e-05,
+      "loss": 2.1658,
+      "step": 18200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.1790000000000006e-05,
+      "loss": 2.3098,
+      "step": 18210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.1780000000000004e-05,
+      "loss": 2.1136,
+      "step": 18220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.177e-05,
+      "loss": 2.1031,
+      "step": 18230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.176e-05,
+      "loss": 2.0208,
+      "step": 18240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.175e-05,
+      "loss": 2.0896,
+      "step": 18250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.1740000000000004e-05,
+      "loss": 2.1916,
+      "step": 18260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.173e-05,
+      "loss": 2.2739,
+      "step": 18270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.172e-05,
+      "loss": 2.1036,
+      "step": 18280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.171e-05,
+      "loss": 2.1644,
+      "step": 18290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.1700000000000005e-05,
+      "loss": 2.1241,
+      "step": 18300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.169e-05,
+      "loss": 2.2575,
+      "step": 18310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.168e-05,
+      "loss": 2.1842,
+      "step": 18320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.167e-05,
+      "loss": 2.21,
+      "step": 18330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.166e-05,
+      "loss": 2.3518,
+      "step": 18340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.1650000000000004e-05,
+      "loss": 2.1046,
+      "step": 18350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.164e-05,
+      "loss": 2.3081,
+      "step": 18360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.163000000000001e-05,
+      "loss": 2.1158,
+      "step": 18370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.162e-05,
+      "loss": 2.2141,
+      "step": 18380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.1610000000000004e-05,
+      "loss": 2.143,
+      "step": 18390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.16e-05,
+      "loss": 2.1211,
+      "step": 18400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.159e-05,
+      "loss": 2.2086,
+      "step": 18410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.1580000000000006e-05,
+      "loss": 2.2434,
+      "step": 18420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.157e-05,
+      "loss": 2.2972,
+      "step": 18430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.156e-05,
+      "loss": 2.2417,
+      "step": 18440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.155e-05,
+      "loss": 2.089,
+      "step": 18450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.154e-05,
+      "loss": 2.22,
+      "step": 18460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.1530000000000005e-05,
+      "loss": 2.5553,
+      "step": 18470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.1519999999999996e-05,
+      "loss": 2.1298,
+      "step": 18480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.151e-05,
+      "loss": 2.1073,
+      "step": 18490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.15e-05,
+      "loss": 2.2038,
+      "step": 18500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.1490000000000005e-05,
+      "loss": 2.2216,
+      "step": 18510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.1480000000000004e-05,
+      "loss": 2.2125,
+      "step": 18520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.147e-05,
+      "loss": 2.0848,
+      "step": 18530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.146e-05,
+      "loss": 2.1348,
+      "step": 18540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.145e-05,
+      "loss": 2.2875,
+      "step": 18550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.1440000000000004e-05,
+      "loss": 2.3036,
+      "step": 18560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.143e-05,
+      "loss": 2.1276,
+      "step": 18570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.142e-05,
+      "loss": 2.1005,
+      "step": 18580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.141e-05,
+      "loss": 2.0729,
+      "step": 18590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.1400000000000004e-05,
+      "loss": 2.3177,
+      "step": 18600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.139e-05,
+      "loss": 2.0566,
+      "step": 18610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.138e-05,
+      "loss": 2.3056,
+      "step": 18620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.137e-05,
+      "loss": 2.2063,
+      "step": 18630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.136e-05,
+      "loss": 2.2933,
+      "step": 18640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.135e-05,
+      "loss": 2.1966,
+      "step": 18650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.134e-05,
+      "loss": 2.1442,
+      "step": 18660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.133000000000001e-05,
+      "loss": 2.1928,
+      "step": 18670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.132e-05,
+      "loss": 2.1911,
+      "step": 18680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.1310000000000003e-05,
+      "loss": 2.2593,
+      "step": 18690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.13e-05,
+      "loss": 2.3174,
+      "step": 18700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.129e-05,
+      "loss": 2.3918,
+      "step": 18710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.1280000000000005e-05,
+      "loss": 2.188,
+      "step": 18720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.127e-05,
+      "loss": 2.234,
+      "step": 18730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.126e-05,
+      "loss": 2.2984,
+      "step": 18740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.125e-05,
+      "loss": 2.1079,
+      "step": 18750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.1240000000000006e-05,
+      "loss": 2.2948,
+      "step": 18760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.1230000000000004e-05,
+      "loss": 2.1106,
+      "step": 18770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.122e-05,
+      "loss": 2.2245,
+      "step": 18780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.121e-05,
+      "loss": 2.1758,
+      "step": 18790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.12e-05,
+      "loss": 2.2385,
+      "step": 18800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.1190000000000005e-05,
+      "loss": 2.3847,
+      "step": 18810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.118e-05,
+      "loss": 2.0903,
+      "step": 18820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.117e-05,
+      "loss": 2.243,
+      "step": 18830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.116e-05,
+      "loss": 2.1863,
+      "step": 18840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.115e-05,
+      "loss": 2.1447,
+      "step": 18850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.1140000000000003e-05,
+      "loss": 2.2886,
+      "step": 18860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.113e-05,
+      "loss": 2.3176,
+      "step": 18870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.112e-05,
+      "loss": 2.2827,
+      "step": 18880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.111e-05,
+      "loss": 2.0539,
+      "step": 18890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.1100000000000004e-05,
+      "loss": 2.2043,
+      "step": 18900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.109e-05,
+      "loss": 1.8242,
+      "step": 18910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.108e-05,
+      "loss": 2.3362,
+      "step": 18920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.107e-05,
+      "loss": 2.1801,
+      "step": 18930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.106e-05,
+      "loss": 1.9841,
+      "step": 18940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.105e-05,
+      "loss": 2.0929,
+      "step": 18950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.104e-05,
+      "loss": 2.0365,
+      "step": 18960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.1030000000000006e-05,
+      "loss": 2.1283,
+      "step": 18970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.102e-05,
+      "loss": 2.161,
+      "step": 18980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.101e-05,
+      "loss": 2.1855,
+      "step": 18990
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.1e-05,
+      "loss": 2.0725,
+      "step": 19000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.099e-05,
+      "loss": 2.1629,
+      "step": 19010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.0980000000000005e-05,
+      "loss": 2.0964,
+      "step": 19020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.0969999999999997e-05,
+      "loss": 2.0764,
+      "step": 19030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.096e-05,
+      "loss": 2.2399,
+      "step": 19040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.095e-05,
+      "loss": 1.9927,
+      "step": 19050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.0940000000000005e-05,
+      "loss": 2.3781,
+      "step": 19060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.0930000000000004e-05,
+      "loss": 2.1865,
+      "step": 19070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.092e-05,
+      "loss": 2.2101,
+      "step": 19080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.091e-05,
+      "loss": 2.1825,
+      "step": 19090
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.09e-05,
+      "loss": 2.0107,
+      "step": 19100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.0890000000000004e-05,
+      "loss": 2.2567,
+      "step": 19110
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.088e-05,
+      "loss": 2.0498,
+      "step": 19120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.087e-05,
+      "loss": 2.0916,
+      "step": 19130
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.086e-05,
+      "loss": 2.12,
+      "step": 19140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.0850000000000004e-05,
+      "loss": 2.1778,
+      "step": 19150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.084e-05,
+      "loss": 2.2404,
+      "step": 19160
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.083e-05,
+      "loss": 2.1303,
+      "step": 19170
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.082e-05,
+      "loss": 1.9885,
+      "step": 19180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.081e-05,
+      "loss": 2.2172,
+      "step": 19190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.08e-05,
+      "loss": 2.107,
+      "step": 19200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.079e-05,
+      "loss": 2.2707,
+      "step": 19210
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.078e-05,
+      "loss": 2.1249,
+      "step": 19220
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.077e-05,
+      "loss": 2.2427,
+      "step": 19230
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.076e-05,
+      "loss": 2.1726,
+      "step": 19240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.075e-05,
+      "loss": 2.1146,
+      "step": 19250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.074e-05,
+      "loss": 2.0914,
+      "step": 19260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.0730000000000006e-05,
+      "loss": 2.2738,
+      "step": 19270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.072e-05,
+      "loss": 2.0885,
+      "step": 19280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.071e-05,
+      "loss": 2.1149,
+      "step": 19290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.07e-05,
+      "loss": 2.0864,
+      "step": 19300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.069e-05,
+      "loss": 2.1798,
+      "step": 19310
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.0680000000000004e-05,
+      "loss": 2.15,
+      "step": 19320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.0669999999999996e-05,
+      "loss": 2.0721,
+      "step": 19330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.066e-05,
+      "loss": 2.2543,
+      "step": 19340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.065e-05,
+      "loss": 2.0739,
+      "step": 19350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.0640000000000005e-05,
+      "loss": 2.0931,
+      "step": 19360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.063e-05,
+      "loss": 2.3337,
+      "step": 19370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.062e-05,
+      "loss": 2.4076,
+      "step": 19380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.061e-05,
+      "loss": 2.1718,
+      "step": 19390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.06e-05,
+      "loss": 2.4298,
+      "step": 19400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.0590000000000004e-05,
+      "loss": 2.1661,
+      "step": 19410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.058e-05,
+      "loss": 2.0859,
+      "step": 19420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.057000000000001e-05,
+      "loss": 2.3036,
+      "step": 19430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.056e-05,
+      "loss": 2.0243,
+      "step": 19440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.0550000000000004e-05,
+      "loss": 2.154,
+      "step": 19450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.054e-05,
+      "loss": 2.0845,
+      "step": 19460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.053e-05,
+      "loss": 1.9999,
+      "step": 19470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.0520000000000006e-05,
+      "loss": 2.2655,
+      "step": 19480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.051e-05,
+      "loss": 2.2372,
+      "step": 19490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.05e-05,
+      "loss": 2.0083,
+      "step": 19500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.049e-05,
+      "loss": 2.1551,
+      "step": 19510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.0480000000000003e-05,
+      "loss": 2.1037,
+      "step": 19520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.0470000000000005e-05,
+      "loss": 2.208,
+      "step": 19530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.046e-05,
+      "loss": 2.1545,
+      "step": 19540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.045e-05,
+      "loss": 2.1536,
+      "step": 19550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.0440000000000003e-05,
+      "loss": 2.1808,
+      "step": 19560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.0430000000000002e-05,
+      "loss": 2.1891,
+      "step": 19570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.0420000000000004e-05,
+      "loss": 1.9063,
+      "step": 19580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.041e-05,
+      "loss": 2.3341,
+      "step": 19590
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.04e-05,
+      "loss": 1.9917,
+      "step": 19600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.0390000000000002e-05,
+      "loss": 2.1782,
+      "step": 19610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.0380000000000004e-05,
+      "loss": 2.394,
+      "step": 19620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.0370000000000006e-05,
+      "loss": 2.0456,
+      "step": 19630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.036e-05,
+      "loss": 2.319,
+      "step": 19640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.035e-05,
+      "loss": 2.0731,
+      "step": 19650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.034e-05,
+      "loss": 2.2288,
+      "step": 19660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.0330000000000003e-05,
+      "loss": 2.0102,
+      "step": 19670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.0320000000000004e-05,
+      "loss": 2.1863,
+      "step": 19680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.031e-05,
+      "loss": 2.1622,
+      "step": 19690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.03e-05,
+      "loss": 2.4279,
+      "step": 19700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.0290000000000003e-05,
+      "loss": 2.1661,
+      "step": 19710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.028e-05,
+      "loss": 1.9711,
+      "step": 19720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.0270000000000003e-05,
+      "loss": 2.083,
+      "step": 19730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.0259999999999998e-05,
+      "loss": 2.0068,
+      "step": 19740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.025e-05,
+      "loss": 2.097,
+      "step": 19750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.0240000000000002e-05,
+      "loss": 2.2404,
+      "step": 19760
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.0230000000000004e-05,
+      "loss": 2.0192,
+      "step": 19770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.0220000000000005e-05,
+      "loss": 1.9505,
+      "step": 19780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.021e-05,
+      "loss": 1.9316,
+      "step": 19790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.02e-05,
+      "loss": 2.1547,
+      "step": 19800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.019e-05,
+      "loss": 2.1136,
+      "step": 19810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.0180000000000002e-05,
+      "loss": 1.9443,
+      "step": 19820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.0170000000000004e-05,
+      "loss": 2.1931,
+      "step": 19830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.016e-05,
+      "loss": 1.9617,
+      "step": 19840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.015e-05,
+      "loss": 2.2501,
+      "step": 19850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.0140000000000003e-05,
+      "loss": 2.0372,
+      "step": 19860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.013e-05,
+      "loss": 2.0448,
+      "step": 19870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.0120000000000003e-05,
+      "loss": 2.1395,
+      "step": 19880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.0109999999999998e-05,
+      "loss": 2.1271,
+      "step": 19890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.01e-05,
+      "loss": 2.1552,
+      "step": 19900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.009e-05,
+      "loss": 2.2011,
+      "step": 19910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.0080000000000003e-05,
+      "loss": 2.0491,
+      "step": 19920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.0070000000000005e-05,
+      "loss": 2.1524,
+      "step": 19930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.006e-05,
+      "loss": 2.173,
+      "step": 19940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.0050000000000002e-05,
+      "loss": 2.0837,
+      "step": 19950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.004e-05,
+      "loss": 2.0221,
+      "step": 19960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.0030000000000002e-05,
+      "loss": 2.234,
+      "step": 19970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.0020000000000004e-05,
+      "loss": 1.9607,
+      "step": 19980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.001e-05,
+      "loss": 2.1321,
+      "step": 19990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3e-05,
+      "loss": 2.0996,
+      "step": 20000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9990000000000003e-05,
+      "loss": 1.9221,
+      "step": 20010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.998e-05,
+      "loss": 2.2381,
+      "step": 20020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9970000000000003e-05,
+      "loss": 2.1012,
+      "step": 20030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9959999999999998e-05,
+      "loss": 2.2092,
+      "step": 20040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.995e-05,
+      "loss": 2.341,
+      "step": 20050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.994e-05,
+      "loss": 2.1217,
+      "step": 20060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9930000000000003e-05,
+      "loss": 2.1826,
+      "step": 20070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9920000000000005e-05,
+      "loss": 2.0826,
+      "step": 20080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.991e-05,
+      "loss": 2.0444,
+      "step": 20090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9900000000000002e-05,
+      "loss": 2.2184,
+      "step": 20100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.989e-05,
+      "loss": 2.0548,
+      "step": 20110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9880000000000002e-05,
+      "loss": 1.9752,
+      "step": 20120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9870000000000004e-05,
+      "loss": 2.0799,
+      "step": 20130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.986e-05,
+      "loss": 2.1595,
+      "step": 20140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.985e-05,
+      "loss": 2.0756,
+      "step": 20150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9840000000000002e-05,
+      "loss": 2.1021,
+      "step": 20160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9830000000000004e-05,
+      "loss": 2.239,
+      "step": 20170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9820000000000002e-05,
+      "loss": 2.1428,
+      "step": 20180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9809999999999997e-05,
+      "loss": 2.0463,
+      "step": 20190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.98e-05,
+      "loss": 1.8987,
+      "step": 20200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.979e-05,
+      "loss": 2.0579,
+      "step": 20210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9780000000000003e-05,
+      "loss": 2.0816,
+      "step": 20220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9770000000000005e-05,
+      "loss": 1.9726,
+      "step": 20230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.976e-05,
+      "loss": 2.0625,
+      "step": 20240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.975e-05,
+      "loss": 1.9287,
+      "step": 20250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.974e-05,
+      "loss": 1.9617,
+      "step": 20260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.973e-05,
+      "loss": 2.0335,
+      "step": 20270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9720000000000003e-05,
+      "loss": 1.866,
+      "step": 20280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.971e-05,
+      "loss": 1.9324,
+      "step": 20290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.97e-05,
+      "loss": 2.0179,
+      "step": 20300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9690000000000002e-05,
+      "loss": 2.1233,
+      "step": 20310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9680000000000004e-05,
+      "loss": 2.1894,
+      "step": 20320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9670000000000002e-05,
+      "loss": 2.0785,
+      "step": 20330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9659999999999997e-05,
+      "loss": 1.9827,
+      "step": 20340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.965e-05,
+      "loss": 1.9859,
+      "step": 20350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.964e-05,
+      "loss": 1.9137,
+      "step": 20360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9630000000000003e-05,
+      "loss": 2.062,
+      "step": 20370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9620000000000004e-05,
+      "loss": 2.1605,
+      "step": 20380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.961e-05,
+      "loss": 2.2647,
+      "step": 20390
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.96e-05,
+      "loss": 1.8769,
+      "step": 20400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.959e-05,
+      "loss": 2.1613,
+      "step": 20410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.958e-05,
+      "loss": 2.116,
+      "step": 20420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9570000000000003e-05,
+      "loss": 2.1098,
+      "step": 20430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9559999999999998e-05,
+      "loss": 2.0938,
+      "step": 20440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.955e-05,
+      "loss": 2.0711,
+      "step": 20450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9540000000000002e-05,
+      "loss": 2.1345,
+      "step": 20460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9530000000000004e-05,
+      "loss": 1.9431,
+      "step": 20470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9520000000000002e-05,
+      "loss": 2.144,
+      "step": 20480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.951e-05,
+      "loss": 2.1519,
+      "step": 20490
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.95e-05,
+      "loss": 2.174,
+      "step": 20500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.949e-05,
+      "loss": 1.9324,
+      "step": 20510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9480000000000002e-05,
+      "loss": 2.0487,
+      "step": 20520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9470000000000004e-05,
+      "loss": 2.0358,
+      "step": 20530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.946e-05,
+      "loss": 1.9931,
+      "step": 20540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.945e-05,
+      "loss": 1.9814,
+      "step": 20550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.944e-05,
+      "loss": 2.0278,
+      "step": 20560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.943e-05,
+      "loss": 2.136,
+      "step": 20570
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9420000000000003e-05,
+      "loss": 2.0136,
+      "step": 20580
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9409999999999998e-05,
+      "loss": 2.0381,
+      "step": 20590
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.94e-05,
+      "loss": 2.1817,
+      "step": 20600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.939e-05,
+      "loss": 2.1116,
+      "step": 20610
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9380000000000003e-05,
+      "loss": 1.965,
+      "step": 20620
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9370000000000002e-05,
+      "loss": 1.9431,
+      "step": 20630
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9360000000000003e-05,
+      "loss": 2.0865,
+      "step": 20640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.935e-05,
+      "loss": 2.089,
+      "step": 20650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.934e-05,
+      "loss": 2.1841,
+      "step": 20660
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9330000000000002e-05,
+      "loss": 2.0591,
+      "step": 20670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9320000000000004e-05,
+      "loss": 2.1019,
+      "step": 20680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9310000000000006e-05,
+      "loss": 2.1368,
+      "step": 20690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.93e-05,
+      "loss": 2.1164,
+      "step": 20700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.929e-05,
+      "loss": 2.0619,
+      "step": 20710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.928e-05,
+      "loss": 1.9016,
+      "step": 20720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9270000000000003e-05,
+      "loss": 1.8915,
+      "step": 20730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9260000000000004e-05,
+      "loss": 2.0365,
+      "step": 20740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.925e-05,
+      "loss": 2.1895,
+      "step": 20750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.924e-05,
+      "loss": 2.1428,
+      "step": 20760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9230000000000003e-05,
+      "loss": 2.0439,
+      "step": 20770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.922e-05,
+      "loss": 2.0534,
+      "step": 20780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9210000000000003e-05,
+      "loss": 2.0161,
+      "step": 20790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9199999999999998e-05,
+      "loss": 2.0914,
+      "step": 20800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.919e-05,
+      "loss": 1.9198,
+      "step": 20810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9180000000000002e-05,
+      "loss": 1.9583,
+      "step": 20820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9170000000000004e-05,
+      "loss": 2.0848,
+      "step": 20830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9160000000000005e-05,
+      "loss": 2.1121,
+      "step": 20840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.915e-05,
+      "loss": 2.1883,
+      "step": 20850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9140000000000002e-05,
+      "loss": 2.151,
+      "step": 20860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.913e-05,
+      "loss": 2.1055,
+      "step": 20870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9120000000000002e-05,
+      "loss": 1.9464,
+      "step": 20880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9110000000000004e-05,
+      "loss": 1.9324,
+      "step": 20890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.91e-05,
+      "loss": 2.0604,
+      "step": 20900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.909e-05,
+      "loss": 2.1614,
+      "step": 20910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9080000000000003e-05,
+      "loss": 2.0018,
+      "step": 20920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.907e-05,
+      "loss": 2.1461,
+      "step": 20930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9060000000000003e-05,
+      "loss": 2.0087,
+      "step": 20940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9049999999999998e-05,
+      "loss": 2.1528,
+      "step": 20950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.904e-05,
+      "loss": 2.1206,
+      "step": 20960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.903e-05,
+      "loss": 2.0626,
+      "step": 20970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9020000000000003e-05,
+      "loss": 2.0507,
+      "step": 20980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9010000000000005e-05,
+      "loss": 1.9523,
+      "step": 20990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9e-05,
+      "loss": 1.9959,
+      "step": 21000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8990000000000002e-05,
+      "loss": 2.1358,
+      "step": 21010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.898e-05,
+      "loss": 2.1269,
+      "step": 21020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8970000000000002e-05,
+      "loss": 1.9518,
+      "step": 21030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8960000000000004e-05,
+      "loss": 2.1912,
+      "step": 21040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.895e-05,
+      "loss": 1.9376,
+      "step": 21050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.894e-05,
+      "loss": 2.2169,
+      "step": 21060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8930000000000003e-05,
+      "loss": 2.0769,
+      "step": 21070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8920000000000004e-05,
+      "loss": 2.0517,
+      "step": 21080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8910000000000003e-05,
+      "loss": 2.0517,
+      "step": 21090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8899999999999998e-05,
+      "loss": 2.0265,
+      "step": 21100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.889e-05,
+      "loss": 1.9224,
+      "step": 21110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.888e-05,
+      "loss": 2.0432,
+      "step": 21120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8870000000000003e-05,
+      "loss": 1.9318,
+      "step": 21130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8860000000000005e-05,
+      "loss": 1.9575,
+      "step": 21140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.885e-05,
+      "loss": 1.9542,
+      "step": 21150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8840000000000002e-05,
+      "loss": 1.897,
+      "step": 21160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.883e-05,
+      "loss": 2.1811,
+      "step": 21170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8820000000000002e-05,
+      "loss": 1.9352,
+      "step": 21180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8810000000000004e-05,
+      "loss": 1.9073,
+      "step": 21190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.88e-05,
+      "loss": 1.9286,
+      "step": 21200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.879e-05,
+      "loss": 1.9638,
+      "step": 21210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8780000000000002e-05,
+      "loss": 1.9789,
+      "step": 21220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8770000000000004e-05,
+      "loss": 2.2062,
+      "step": 21230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8760000000000002e-05,
+      "loss": 2.0664,
+      "step": 21240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8749999999999997e-05,
+      "loss": 2.0684,
+      "step": 21250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.874e-05,
+      "loss": 1.9747,
+      "step": 21260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.873e-05,
+      "loss": 2.1395,
+      "step": 21270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8720000000000003e-05,
+      "loss": 1.8555,
+      "step": 21280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8710000000000005e-05,
+      "loss": 1.8957,
+      "step": 21290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.87e-05,
+      "loss": 1.9464,
+      "step": 21300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.869e-05,
+      "loss": 2.0719,
+      "step": 21310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.868e-05,
+      "loss": 1.7902,
+      "step": 21320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.867e-05,
+      "loss": 2.1629,
+      "step": 21330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8660000000000003e-05,
+      "loss": 1.9715,
+      "step": 21340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.865e-05,
+      "loss": 1.8976,
+      "step": 21350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.864e-05,
+      "loss": 2.028,
+      "step": 21360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8630000000000002e-05,
+      "loss": 1.9904,
+      "step": 21370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8620000000000004e-05,
+      "loss": 1.9989,
+      "step": 21380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8610000000000002e-05,
+      "loss": 2.1064,
+      "step": 21390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.86e-05,
+      "loss": 2.0839,
+      "step": 21400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.859e-05,
+      "loss": 2.0304,
+      "step": 21410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.858e-05,
+      "loss": 2.1473,
+      "step": 21420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8570000000000003e-05,
+      "loss": 1.9157,
+      "step": 21430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8560000000000004e-05,
+      "loss": 2.1178,
+      "step": 21440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.855e-05,
+      "loss": 2.1562,
+      "step": 21450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.854e-05,
+      "loss": 1.9399,
+      "step": 21460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.853e-05,
+      "loss": 1.9897,
+      "step": 21470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.852e-05,
+      "loss": 2.1736,
+      "step": 21480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8510000000000003e-05,
+      "loss": 2.0185,
+      "step": 21490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 1.916,
+      "step": 21500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.849e-05,
+      "loss": 2.0558,
+      "step": 21510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8480000000000002e-05,
+      "loss": 2.0693,
+      "step": 21520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8470000000000004e-05,
+      "loss": 2.0783,
+      "step": 21530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8460000000000002e-05,
+      "loss": 2.0608,
+      "step": 21540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.845e-05,
+      "loss": 2.117,
+      "step": 21550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.844e-05,
+      "loss": 2.1641,
+      "step": 21560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.843e-05,
+      "loss": 2.0588,
+      "step": 21570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8420000000000002e-05,
+      "loss": 1.9024,
+      "step": 21580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8410000000000004e-05,
+      "loss": 2.0206,
+      "step": 21590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.84e-05,
+      "loss": 2.0151,
+      "step": 21600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.839e-05,
+      "loss": 1.8427,
+      "step": 21610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.8380000000000003e-05,
+      "loss": 1.9605,
+      "step": 21620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.837e-05,
+      "loss": 2.0543,
+      "step": 21630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8360000000000003e-05,
+      "loss": 2.2256,
+      "step": 21640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8349999999999998e-05,
+      "loss": 1.881,
+      "step": 21650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.834e-05,
+      "loss": 2.2879,
+      "step": 21660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.833e-05,
+      "loss": 2.037,
+      "step": 21670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8320000000000003e-05,
+      "loss": 2.0062,
+      "step": 21680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8310000000000002e-05,
+      "loss": 2.155,
+      "step": 21690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.83e-05,
+      "loss": 1.9498,
+      "step": 21700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.829e-05,
+      "loss": 2.1058,
+      "step": 21710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.828e-05,
+      "loss": 2.0479,
+      "step": 21720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8270000000000002e-05,
+      "loss": 1.9271,
+      "step": 21730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8260000000000004e-05,
+      "loss": 2.0364,
+      "step": 21740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.825e-05,
+      "loss": 1.8948,
+      "step": 21750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.824e-05,
+      "loss": 1.8888,
+      "step": 21760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8230000000000002e-05,
+      "loss": 1.9076,
+      "step": 21770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.822e-05,
+      "loss": 2.0122,
+      "step": 21780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8210000000000003e-05,
+      "loss": 1.9441,
+      "step": 21790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 2.0632,
+      "step": 21800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.819e-05,
+      "loss": 1.9832,
+      "step": 21810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.818e-05,
+      "loss": 2.0993,
+      "step": 21820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8170000000000003e-05,
+      "loss": 1.9623,
+      "step": 21830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.816e-05,
+      "loss": 1.92,
+      "step": 21840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.815e-05,
+      "loss": 2.0533,
+      "step": 21850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8139999999999998e-05,
+      "loss": 1.9431,
+      "step": 21860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.813e-05,
+      "loss": 1.8863,
+      "step": 21870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8120000000000002e-05,
+      "loss": 1.8342,
+      "step": 21880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8110000000000004e-05,
+      "loss": 1.9229,
+      "step": 21890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8100000000000005e-05,
+      "loss": 1.7602,
+      "step": 21900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.809e-05,
+      "loss": 1.8696,
+      "step": 21910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8080000000000002e-05,
+      "loss": 2.1201,
+      "step": 21920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.807e-05,
+      "loss": 2.0018,
+      "step": 21930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8060000000000002e-05,
+      "loss": 2.0119,
+      "step": 21940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.8050000000000004e-05,
+      "loss": 2.0666,
+      "step": 21950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.804e-05,
+      "loss": 2.1068,
+      "step": 21960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.803e-05,
+      "loss": 1.9606,
+      "step": 21970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.8020000000000003e-05,
+      "loss": 2.1597,
+      "step": 21980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.8010000000000005e-05,
+      "loss": 1.9593,
+      "step": 21990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 1.937,
+      "step": 22000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7989999999999998e-05,
+      "loss": 1.9132,
+      "step": 22010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.798e-05,
+      "loss": 1.9143,
+      "step": 22020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.797e-05,
+      "loss": 1.9628,
+      "step": 22030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7960000000000003e-05,
+      "loss": 1.9912,
+      "step": 22040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7950000000000005e-05,
+      "loss": 1.7827,
+      "step": 22050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.794e-05,
+      "loss": 1.9289,
+      "step": 22060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7930000000000002e-05,
+      "loss": 2.0143,
+      "step": 22070
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.792e-05,
+      "loss": 2.076,
+      "step": 22080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7910000000000002e-05,
+      "loss": 1.8994,
+      "step": 22090
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7900000000000004e-05,
+      "loss": 1.9853,
+      "step": 22100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.789e-05,
+      "loss": 2.0674,
+      "step": 22110
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.788e-05,
+      "loss": 1.9628,
+      "step": 22120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7870000000000003e-05,
+      "loss": 1.8327,
+      "step": 22130
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7860000000000004e-05,
+      "loss": 2.1159,
+      "step": 22140
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7850000000000003e-05,
+      "loss": 2.0384,
+      "step": 22150
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7839999999999998e-05,
+      "loss": 2.0206,
+      "step": 22160
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.783e-05,
+      "loss": 1.9639,
+      "step": 22170
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.782e-05,
+      "loss": 1.9604,
+      "step": 22180
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7810000000000003e-05,
+      "loss": 1.8903,
+      "step": 22190
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 1.8147,
+      "step": 22200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.779e-05,
+      "loss": 2.0224,
+      "step": 22210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.778e-05,
+      "loss": 1.8057,
+      "step": 22220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.777e-05,
+      "loss": 1.907,
+      "step": 22230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7760000000000002e-05,
+      "loss": 2.0711,
+      "step": 22240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7750000000000004e-05,
+      "loss": 1.9523,
+      "step": 22250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.774e-05,
+      "loss": 1.9777,
+      "step": 22260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.773e-05,
+      "loss": 1.9884,
+      "step": 22270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7720000000000002e-05,
+      "loss": 2.036,
+      "step": 22280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7710000000000004e-05,
+      "loss": 1.7448,
+      "step": 22290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7700000000000002e-05,
+      "loss": 1.9634,
+      "step": 22300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.769e-05,
+      "loss": 1.8195,
+      "step": 22310
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.768e-05,
+      "loss": 2.0636,
+      "step": 22320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.767e-05,
+      "loss": 2.087,
+      "step": 22330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.7660000000000003e-05,
+      "loss": 2.1348,
+      "step": 22340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.7650000000000005e-05,
+      "loss": 1.9633,
+      "step": 22350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.764e-05,
+      "loss": 1.9784,
+      "step": 22360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.763e-05,
+      "loss": 1.9472,
+      "step": 22370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.762e-05,
+      "loss": 1.9415,
+      "step": 22380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.761e-05,
+      "loss": 2.0605,
+      "step": 22390
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 1.9945,
+      "step": 22400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.759e-05,
+      "loss": 1.8807,
+      "step": 22410
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.758e-05,
+      "loss": 1.9,
+      "step": 22420
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.7570000000000002e-05,
+      "loss": 1.8796,
+      "step": 22430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.7560000000000004e-05,
+      "loss": 2.0359,
+      "step": 22440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.7550000000000002e-05,
+      "loss": 1.9024,
+      "step": 22450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.754e-05,
+      "loss": 1.8889,
+      "step": 22460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.753e-05,
+      "loss": 1.7773,
+      "step": 22470
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.752e-05,
+      "loss": 1.9873,
+      "step": 22480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.7510000000000003e-05,
+      "loss": 1.7095,
+      "step": 22490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 2.0201,
+      "step": 22500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.749e-05,
+      "loss": 2.0369,
+      "step": 22510
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.748e-05,
+      "loss": 1.9524,
+      "step": 22520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.7470000000000003e-05,
+      "loss": 2.093,
+      "step": 22530
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.746e-05,
+      "loss": 1.8307,
+      "step": 22540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.7450000000000003e-05,
+      "loss": 2.1173,
+      "step": 22550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.7439999999999998e-05,
+      "loss": 1.8519,
+      "step": 22560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.743e-05,
+      "loss": 1.7809,
+      "step": 22570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.7420000000000002e-05,
+      "loss": 1.8855,
+      "step": 22580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.7410000000000004e-05,
+      "loss": 1.865,
+      "step": 22590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.7400000000000002e-05,
+      "loss": 2.1297,
+      "step": 22600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.739e-05,
+      "loss": 1.9396,
+      "step": 22610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.738e-05,
+      "loss": 1.9349,
+      "step": 22620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.737e-05,
+      "loss": 1.9292,
+      "step": 22630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.7360000000000002e-05,
+      "loss": 2.0672,
+      "step": 22640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.7350000000000004e-05,
+      "loss": 1.8596,
+      "step": 22650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.734e-05,
+      "loss": 2.0637,
+      "step": 22660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.733e-05,
+      "loss": 2.1113,
+      "step": 22670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7320000000000003e-05,
+      "loss": 1.8967,
+      "step": 22680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.731e-05,
+      "loss": 1.905,
+      "step": 22690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7300000000000003e-05,
+      "loss": 1.8637,
+      "step": 22700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7289999999999998e-05,
+      "loss": 1.8102,
+      "step": 22710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.728e-05,
+      "loss": 2.0004,
+      "step": 22720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.727e-05,
+      "loss": 1.8657,
+      "step": 22730
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7260000000000003e-05,
+      "loss": 1.9581,
+      "step": 22740
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.725e-05,
+      "loss": 1.8546,
+      "step": 22750
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.724e-05,
+      "loss": 1.9141,
+      "step": 22760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.723e-05,
+      "loss": 1.9375,
+      "step": 22770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.722e-05,
+      "loss": 1.8125,
+      "step": 22780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7210000000000002e-05,
+      "loss": 2.0981,
+      "step": 22790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7200000000000004e-05,
+      "loss": 1.9965,
+      "step": 22800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.719e-05,
+      "loss": 1.8367,
+      "step": 22810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.718e-05,
+      "loss": 1.9609,
+      "step": 22820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7170000000000002e-05,
+      "loss": 1.9268,
+      "step": 22830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.716e-05,
+      "loss": 1.9978,
+      "step": 22840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7150000000000003e-05,
+      "loss": 1.8377,
+      "step": 22850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7139999999999998e-05,
+      "loss": 1.951,
+      "step": 22860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.713e-05,
+      "loss": 1.8229,
+      "step": 22870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.712e-05,
+      "loss": 1.9864,
+      "step": 22880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7110000000000003e-05,
+      "loss": 1.8898,
+      "step": 22890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7100000000000005e-05,
+      "loss": 1.8216,
+      "step": 22900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.709e-05,
+      "loss": 2.0068,
+      "step": 22910
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7079999999999998e-05,
+      "loss": 1.9504,
+      "step": 22920
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.707e-05,
+      "loss": 1.9206,
+      "step": 22930
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7060000000000002e-05,
+      "loss": 1.9056,
+      "step": 22940
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7050000000000004e-05,
+      "loss": 1.938,
+      "step": 22950
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.704e-05,
+      "loss": 1.7936,
+      "step": 22960
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.703e-05,
+      "loss": 1.7133,
+      "step": 22970
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7020000000000002e-05,
+      "loss": 1.7792,
+      "step": 22980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.701e-05,
+      "loss": 2.0009,
+      "step": 22990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 1.84,
+      "step": 23000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.6989999999999997e-05,
+      "loss": 2.0544,
+      "step": 23010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.698e-05,
+      "loss": 1.8767,
+      "step": 23020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.697e-05,
+      "loss": 1.7787,
+      "step": 23030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6960000000000003e-05,
+      "loss": 1.8894,
+      "step": 23040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6950000000000005e-05,
+      "loss": 1.8188,
+      "step": 23050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.694e-05,
+      "loss": 1.9366,
+      "step": 23060
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.693e-05,
+      "loss": 2.049,
+      "step": 23070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.692e-05,
+      "loss": 1.8622,
+      "step": 23080
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.691e-05,
+      "loss": 1.8488,
+      "step": 23090
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6900000000000003e-05,
+      "loss": 1.9878,
+      "step": 23100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.689e-05,
+      "loss": 2.0913,
+      "step": 23110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.688e-05,
+      "loss": 1.8725,
+      "step": 23120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6870000000000002e-05,
+      "loss": 1.7917,
+      "step": 23130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.686e-05,
+      "loss": 2.0304,
+      "step": 23140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6850000000000002e-05,
+      "loss": 1.9223,
+      "step": 23150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6840000000000004e-05,
+      "loss": 1.7804,
+      "step": 23160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.683e-05,
+      "loss": 1.9748,
+      "step": 23170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.682e-05,
+      "loss": 1.9598,
+      "step": 23180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6810000000000003e-05,
+      "loss": 1.806,
+      "step": 23190
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6800000000000004e-05,
+      "loss": 1.9846,
+      "step": 23200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6790000000000003e-05,
+      "loss": 1.9217,
+      "step": 23210
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.678e-05,
+      "loss": 2.0258,
+      "step": 23220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.677e-05,
+      "loss": 1.8087,
+      "step": 23230
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.676e-05,
+      "loss": 1.919,
+      "step": 23240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6750000000000003e-05,
+      "loss": 1.9171,
+      "step": 23250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6740000000000005e-05,
+      "loss": 1.9868,
+      "step": 23260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.673e-05,
+      "loss": 1.7897,
+      "step": 23270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.672e-05,
+      "loss": 1.8976,
+      "step": 23280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.671e-05,
+      "loss": 1.8855,
+      "step": 23290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6700000000000002e-05,
+      "loss": 1.744,
+      "step": 23300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6690000000000004e-05,
+      "loss": 2.051,
+      "step": 23310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.668e-05,
+      "loss": 1.9919,
+      "step": 23320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.667e-05,
+      "loss": 1.8828,
+      "step": 23330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6660000000000002e-05,
+      "loss": 1.8529,
+      "step": 23340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6650000000000004e-05,
+      "loss": 1.9193,
+      "step": 23350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6640000000000002e-05,
+      "loss": 2.1031,
+      "step": 23360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.663e-05,
+      "loss": 1.8116,
+      "step": 23370
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.662e-05,
+      "loss": 1.9605,
+      "step": 23380
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.661e-05,
+      "loss": 2.0027,
+      "step": 23390
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6600000000000003e-05,
+      "loss": 1.7882,
+      "step": 23400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6590000000000005e-05,
+      "loss": 1.8453,
+      "step": 23410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.658e-05,
+      "loss": 1.8334,
+      "step": 23420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.657e-05,
+      "loss": 1.9542,
+      "step": 23430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6560000000000003e-05,
+      "loss": 1.9597,
+      "step": 23440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.655e-05,
+      "loss": 2.009,
+      "step": 23450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6540000000000003e-05,
+      "loss": 2.0994,
+      "step": 23460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.653e-05,
+      "loss": 2.0972,
+      "step": 23470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.652e-05,
+      "loss": 1.865,
+      "step": 23480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6510000000000002e-05,
+      "loss": 1.9362,
+      "step": 23490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6500000000000004e-05,
+      "loss": 1.9063,
+      "step": 23500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6490000000000002e-05,
+      "loss": 1.8962,
+      "step": 23510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.648e-05,
+      "loss": 1.7648,
+      "step": 23520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.647e-05,
+      "loss": 1.9337,
+      "step": 23530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.646e-05,
+      "loss": 1.9174,
+      "step": 23540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6450000000000003e-05,
+      "loss": 1.9719,
+      "step": 23550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6440000000000004e-05,
+      "loss": 1.8917,
+      "step": 23560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.643e-05,
+      "loss": 1.8602,
+      "step": 23570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.642e-05,
+      "loss": 2.1528,
+      "step": 23580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6410000000000003e-05,
+      "loss": 1.9392,
+      "step": 23590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.64e-05,
+      "loss": 1.8787,
+      "step": 23600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6390000000000003e-05,
+      "loss": 1.9979,
+      "step": 23610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6379999999999998e-05,
+      "loss": 1.8217,
+      "step": 23620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.637e-05,
+      "loss": 1.9567,
+      "step": 23630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6360000000000002e-05,
+      "loss": 1.9461,
+      "step": 23640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6350000000000004e-05,
+      "loss": 1.9861,
+      "step": 23650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6340000000000002e-05,
+      "loss": 2.0402,
+      "step": 23660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.633e-05,
+      "loss": 1.9585,
+      "step": 23670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.632e-05,
+      "loss": 1.824,
+      "step": 23680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.631e-05,
+      "loss": 1.8987,
+      "step": 23690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6300000000000002e-05,
+      "loss": 1.969,
+      "step": 23700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6290000000000004e-05,
+      "loss": 1.9008,
+      "step": 23710
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.628e-05,
+      "loss": 1.8174,
+      "step": 23720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.627e-05,
+      "loss": 1.833,
+      "step": 23730
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6260000000000003e-05,
+      "loss": 2.2275,
+      "step": 23740
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.625e-05,
+      "loss": 2.023,
+      "step": 23750
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6240000000000003e-05,
+      "loss": 1.729,
+      "step": 23760
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6229999999999998e-05,
+      "loss": 1.842,
+      "step": 23770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.622e-05,
+      "loss": 1.9329,
+      "step": 23780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.621e-05,
+      "loss": 1.9335,
+      "step": 23790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6200000000000003e-05,
+      "loss": 1.8261,
+      "step": 23800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6190000000000005e-05,
+      "loss": 1.9717,
+      "step": 23810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.618e-05,
+      "loss": 1.8059,
+      "step": 23820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.617e-05,
+      "loss": 1.871,
+      "step": 23830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.616e-05,
+      "loss": 1.7294,
+      "step": 23840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6150000000000002e-05,
+      "loss": 1.9741,
+      "step": 23850
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6140000000000004e-05,
+      "loss": 1.9679,
+      "step": 23860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.613e-05,
+      "loss": 1.913,
+      "step": 23870
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.612e-05,
+      "loss": 1.9218,
+      "step": 23880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6110000000000002e-05,
+      "loss": 1.848,
+      "step": 23890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.61e-05,
+      "loss": 2.0244,
+      "step": 23900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6090000000000003e-05,
+      "loss": 1.9494,
+      "step": 23910
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6079999999999998e-05,
+      "loss": 1.8097,
+      "step": 23920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.607e-05,
+      "loss": 1.7749,
+      "step": 23930
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.606e-05,
+      "loss": 1.9053,
+      "step": 23940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6050000000000003e-05,
+      "loss": 2.0055,
+      "step": 23950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6040000000000005e-05,
+      "loss": 1.8792,
+      "step": 23960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.603e-05,
+      "loss": 1.7969,
+      "step": 23970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.602e-05,
+      "loss": 1.9211,
+      "step": 23980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.601e-05,
+      "loss": 1.7649,
+      "step": 23990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 1.7959,
+      "step": 24000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5990000000000004e-05,
+      "loss": 1.7857,
+      "step": 24010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.598e-05,
+      "loss": 1.8776,
+      "step": 24020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.597e-05,
+      "loss": 1.926,
+      "step": 24030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5960000000000002e-05,
+      "loss": 1.7728,
+      "step": 24040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.595e-05,
+      "loss": 1.7953,
+      "step": 24050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5940000000000002e-05,
+      "loss": 1.9283,
+      "step": 24060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5929999999999997e-05,
+      "loss": 1.9165,
+      "step": 24070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.592e-05,
+      "loss": 1.851,
+      "step": 24080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.591e-05,
+      "loss": 1.9194,
+      "step": 24090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5900000000000003e-05,
+      "loss": 1.9092,
+      "step": 24100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5890000000000005e-05,
+      "loss": 2.0591,
+      "step": 24110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.588e-05,
+      "loss": 1.8228,
+      "step": 24120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.587e-05,
+      "loss": 1.9732,
+      "step": 24130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.586e-05,
+      "loss": 1.879,
+      "step": 24140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.585e-05,
+      "loss": 1.9908,
+      "step": 24150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5840000000000003e-05,
+      "loss": 1.8793,
+      "step": 24160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.583e-05,
+      "loss": 1.9682,
+      "step": 24170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.582e-05,
+      "loss": 1.8261,
+      "step": 24180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5810000000000002e-05,
+      "loss": 1.946,
+      "step": 24190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.58e-05,
+      "loss": 1.978,
+      "step": 24200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5790000000000002e-05,
+      "loss": 1.8383,
+      "step": 24210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5779999999999997e-05,
+      "loss": 1.7986,
+      "step": 24220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.577e-05,
+      "loss": 1.7376,
+      "step": 24230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.576e-05,
+      "loss": 1.7892,
+      "step": 24240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5750000000000002e-05,
+      "loss": 1.9143,
+      "step": 24250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5740000000000004e-05,
+      "loss": 1.8697,
+      "step": 24260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.573e-05,
+      "loss": 1.9092,
+      "step": 24270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.572e-05,
+      "loss": 1.909,
+      "step": 24280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.571e-05,
+      "loss": 2.1078,
+      "step": 24290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.57e-05,
+      "loss": 1.9573,
+      "step": 24300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5690000000000003e-05,
+      "loss": 1.8182,
+      "step": 24310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5679999999999998e-05,
+      "loss": 1.9251,
+      "step": 24320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.567e-05,
+      "loss": 1.9277,
+      "step": 24330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.566e-05,
+      "loss": 1.8644,
+      "step": 24340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5650000000000003e-05,
+      "loss": 1.9754,
+      "step": 24350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5640000000000002e-05,
+      "loss": 1.8713,
+      "step": 24360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5629999999999997e-05,
+      "loss": 1.9566,
+      "step": 24370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.562e-05,
+      "loss": 1.7931,
+      "step": 24380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.561e-05,
+      "loss": 1.8206,
+      "step": 24390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5600000000000002e-05,
+      "loss": 1.8788,
+      "step": 24400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5590000000000004e-05,
+      "loss": 1.9586,
+      "step": 24410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5580000000000002e-05,
+      "loss": 1.8801,
+      "step": 24420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.557e-05,
+      "loss": 1.8035,
+      "step": 24430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.556e-05,
+      "loss": 1.9266,
+      "step": 24440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.555e-05,
+      "loss": 1.8962,
+      "step": 24450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5540000000000003e-05,
+      "loss": 1.9312,
+      "step": 24460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5530000000000005e-05,
+      "loss": 1.9815,
+      "step": 24470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.552e-05,
+      "loss": 1.9039,
+      "step": 24480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.551e-05,
+      "loss": 1.8257,
+      "step": 24490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5500000000000003e-05,
+      "loss": 1.7961,
+      "step": 24500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.549e-05,
+      "loss": 1.8982,
+      "step": 24510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5480000000000003e-05,
+      "loss": 1.8905,
+      "step": 24520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.547e-05,
+      "loss": 1.8735,
+      "step": 24530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.546e-05,
+      "loss": 1.7892,
+      "step": 24540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5450000000000002e-05,
+      "loss": 1.9198,
+      "step": 24550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5440000000000004e-05,
+      "loss": 1.9207,
+      "step": 24560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5430000000000002e-05,
+      "loss": 1.8124,
+      "step": 24570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.542e-05,
+      "loss": 1.8251,
+      "step": 24580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.541e-05,
+      "loss": 1.8191,
+      "step": 24590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.54e-05,
+      "loss": 1.9495,
+      "step": 24600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5390000000000003e-05,
+      "loss": 1.8665,
+      "step": 24610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5380000000000004e-05,
+      "loss": 1.6979,
+      "step": 24620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.537e-05,
+      "loss": 1.8731,
+      "step": 24630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.536e-05,
+      "loss": 1.974,
+      "step": 24640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5350000000000003e-05,
+      "loss": 1.9183,
+      "step": 24650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.534e-05,
+      "loss": 1.8335,
+      "step": 24660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5330000000000003e-05,
+      "loss": 1.9919,
+      "step": 24670
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5319999999999998e-05,
+      "loss": 1.7645,
+      "step": 24680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.531e-05,
+      "loss": 1.8771,
+      "step": 24690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5300000000000002e-05,
+      "loss": 1.8775,
+      "step": 24700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5290000000000004e-05,
+      "loss": 1.974,
+      "step": 24710
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5280000000000005e-05,
+      "loss": 1.7585,
+      "step": 24720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.527e-05,
+      "loss": 1.8347,
+      "step": 24730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.526e-05,
+      "loss": 1.6438,
+      "step": 24740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.525e-05,
+      "loss": 1.7589,
+      "step": 24750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5240000000000002e-05,
+      "loss": 1.8466,
+      "step": 24760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5230000000000004e-05,
+      "loss": 1.8337,
+      "step": 24770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.522e-05,
+      "loss": 1.7146,
+      "step": 24780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.521e-05,
+      "loss": 1.852,
+      "step": 24790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.5200000000000003e-05,
+      "loss": 1.9475,
+      "step": 24800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.519e-05,
+      "loss": 1.8589,
+      "step": 24810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.5180000000000003e-05,
+      "loss": 1.8138,
+      "step": 24820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.5169999999999998e-05,
+      "loss": 1.88,
+      "step": 24830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.516e-05,
+      "loss": 1.8692,
+      "step": 24840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.515e-05,
+      "loss": 1.8182,
+      "step": 24850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.5140000000000003e-05,
+      "loss": 1.7634,
+      "step": 24860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.5130000000000005e-05,
+      "loss": 2.0104,
+      "step": 24870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.512e-05,
+      "loss": 1.7987,
+      "step": 24880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.5110000000000002e-05,
+      "loss": 1.7854,
+      "step": 24890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.51e-05,
+      "loss": 1.8427,
+      "step": 24900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.5090000000000002e-05,
+      "loss": 1.7979,
+      "step": 24910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.5080000000000004e-05,
+      "loss": 1.9186,
+      "step": 24920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.507e-05,
+      "loss": 1.9194,
+      "step": 24930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.506e-05,
+      "loss": 1.7619,
+      "step": 24940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.5050000000000002e-05,
+      "loss": 1.9539,
+      "step": 24950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.504e-05,
+      "loss": 1.7937,
+      "step": 24960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.5030000000000003e-05,
+      "loss": 1.9167,
+      "step": 24970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.5019999999999998e-05,
+      "loss": 1.7095,
+      "step": 24980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.501e-05,
+      "loss": 1.7876,
+      "step": 24990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.5e-05,
+      "loss": 1.7894,
+      "step": 25000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4990000000000003e-05,
+      "loss": 1.7879,
+      "step": 25010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.498e-05,
+      "loss": 1.7613,
+      "step": 25020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4970000000000003e-05,
+      "loss": 1.809,
+      "step": 25030
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.496e-05,
+      "loss": 1.8781,
+      "step": 25040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.495e-05,
+      "loss": 1.8388,
+      "step": 25050
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4940000000000002e-05,
+      "loss": 1.721,
+      "step": 25060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.493e-05,
+      "loss": 1.9334,
+      "step": 25070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4920000000000002e-05,
+      "loss": 1.8039,
+      "step": 25080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.491e-05,
+      "loss": 1.7778,
+      "step": 25090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4900000000000002e-05,
+      "loss": 1.9545,
+      "step": 25100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.489e-05,
+      "loss": 1.6829,
+      "step": 25110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.488e-05,
+      "loss": 1.7736,
+      "step": 25120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.487e-05,
+      "loss": 1.7387,
+      "step": 25130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.486e-05,
+      "loss": 1.9489,
+      "step": 25140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.485e-05,
+      "loss": 2.0753,
+      "step": 25150
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.4840000000000003e-05,
+      "loss": 1.7101,
+      "step": 25160
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.483e-05,
+      "loss": 1.8474,
+      "step": 25170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.4820000000000003e-05,
+      "loss": 1.783,
+      "step": 25180
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.481e-05,
+      "loss": 1.8592,
+      "step": 25190
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.48e-05,
+      "loss": 1.7366,
+      "step": 25200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.479e-05,
+      "loss": 1.6961,
+      "step": 25210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.478e-05,
+      "loss": 1.8309,
+      "step": 25220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.4770000000000002e-05,
+      "loss": 1.7767,
+      "step": 25230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.476e-05,
+      "loss": 1.6928,
+      "step": 25240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.4750000000000002e-05,
+      "loss": 1.8192,
+      "step": 25250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.4740000000000004e-05,
+      "loss": 1.8214,
+      "step": 25260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.473e-05,
+      "loss": 1.7114,
+      "step": 25270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.472e-05,
+      "loss": 1.6498,
+      "step": 25280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.471e-05,
+      "loss": 1.822,
+      "step": 25290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.47e-05,
+      "loss": 1.862,
+      "step": 25300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.4690000000000002e-05,
+      "loss": 1.8796,
+      "step": 25310
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.468e-05,
+      "loss": 2.0371,
+      "step": 25320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.4670000000000003e-05,
+      "loss": 1.8458,
+      "step": 25330
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.466e-05,
+      "loss": 1.8226,
+      "step": 25340
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.465e-05,
+      "loss": 1.915,
+      "step": 25350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.464e-05,
+      "loss": 1.945,
+      "step": 25360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.463e-05,
+      "loss": 1.9106,
+      "step": 25370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.462e-05,
+      "loss": 1.7997,
+      "step": 25380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.4610000000000003e-05,
+      "loss": 1.8991,
+      "step": 25390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.46e-05,
+      "loss": 1.8014,
+      "step": 25400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.4590000000000003e-05,
+      "loss": 1.6916,
+      "step": 25410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.4580000000000002e-05,
+      "loss": 1.8835,
+      "step": 25420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.457e-05,
+      "loss": 1.6374,
+      "step": 25430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.4560000000000002e-05,
+      "loss": 2.0773,
+      "step": 25440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.455e-05,
+      "loss": 1.8998,
+      "step": 25450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.4540000000000002e-05,
+      "loss": 1.7857,
+      "step": 25460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.453e-05,
+      "loss": 1.6332,
+      "step": 25470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.4520000000000002e-05,
+      "loss": 1.7787,
+      "step": 25480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.451e-05,
+      "loss": 1.8527,
+      "step": 25490
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.45e-05,
+      "loss": 1.7389,
+      "step": 25500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.449e-05,
+      "loss": 1.8534,
+      "step": 25510
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.448e-05,
+      "loss": 1.9705,
+      "step": 25520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.447e-05,
+      "loss": 1.777,
+      "step": 25530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.4460000000000003e-05,
+      "loss": 1.7511,
+      "step": 25540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.445e-05,
+      "loss": 1.8265,
+      "step": 25550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.4440000000000003e-05,
+      "loss": 1.719,
+      "step": 25560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.443e-05,
+      "loss": 1.6619,
+      "step": 25570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.442e-05,
+      "loss": 1.9589,
+      "step": 25580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.4410000000000002e-05,
+      "loss": 1.8832,
+      "step": 25590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.44e-05,
+      "loss": 1.9509,
+      "step": 25600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.4390000000000002e-05,
+      "loss": 1.788,
+      "step": 25610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.438e-05,
+      "loss": 1.7224,
+      "step": 25620
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.4370000000000002e-05,
+      "loss": 1.7956,
+      "step": 25630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.4360000000000004e-05,
+      "loss": 1.8978,
+      "step": 25640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.435e-05,
+      "loss": 1.8134,
+      "step": 25650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.434e-05,
+      "loss": 1.8233,
+      "step": 25660
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.433e-05,
+      "loss": 1.8165,
+      "step": 25670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.432e-05,
+      "loss": 1.8629,
+      "step": 25680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.4310000000000003e-05,
+      "loss": 1.9276,
+      "step": 25690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.43e-05,
+      "loss": 1.8054,
+      "step": 25700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.4290000000000003e-05,
+      "loss": 1.7782,
+      "step": 25710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.428e-05,
+      "loss": 1.7506,
+      "step": 25720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.427e-05,
+      "loss": 1.8919,
+      "step": 25730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.426e-05,
+      "loss": 1.8512,
+      "step": 25740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.425e-05,
+      "loss": 1.7165,
+      "step": 25750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.4240000000000002e-05,
+      "loss": 1.7994,
+      "step": 25760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.423e-05,
+      "loss": 1.7985,
+      "step": 25770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.4220000000000002e-05,
+      "loss": 1.8908,
+      "step": 25780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.4210000000000004e-05,
+      "loss": 1.7429,
+      "step": 25790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.4200000000000002e-05,
+      "loss": 1.7719,
+      "step": 25800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.419e-05,
+      "loss": 1.8459,
+      "step": 25810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.418e-05,
+      "loss": 1.8964,
+      "step": 25820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.417e-05,
+      "loss": 2.0585,
+      "step": 25830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4160000000000002e-05,
+      "loss": 1.8753,
+      "step": 25840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.415e-05,
+      "loss": 1.9391,
+      "step": 25850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4140000000000003e-05,
+      "loss": 1.943,
+      "step": 25860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.413e-05,
+      "loss": 1.5552,
+      "step": 25870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.412e-05,
+      "loss": 1.8022,
+      "step": 25880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.411e-05,
+      "loss": 1.8724,
+      "step": 25890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.41e-05,
+      "loss": 1.7815,
+      "step": 25900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.409e-05,
+      "loss": 1.7216,
+      "step": 25910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.408e-05,
+      "loss": 1.8501,
+      "step": 25920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.407e-05,
+      "loss": 1.7748,
+      "step": 25930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4060000000000003e-05,
+      "loss": 1.8374,
+      "step": 25940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4050000000000002e-05,
+      "loss": 1.869,
+      "step": 25950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.404e-05,
+      "loss": 1.9012,
+      "step": 25960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4030000000000002e-05,
+      "loss": 1.7007,
+      "step": 25970
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.402e-05,
+      "loss": 1.6651,
+      "step": 25980
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4010000000000002e-05,
+      "loss": 1.7174,
+      "step": 25990
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.4e-05,
+      "loss": 1.9503,
+      "step": 26000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3990000000000002e-05,
+      "loss": 1.7992,
+      "step": 26010
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.398e-05,
+      "loss": 1.786,
+      "step": 26020
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.397e-05,
+      "loss": 1.687,
+      "step": 26030
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.396e-05,
+      "loss": 1.5983,
+      "step": 26040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.395e-05,
+      "loss": 1.6539,
+      "step": 26050
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.394e-05,
+      "loss": 1.8533,
+      "step": 26060
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3930000000000003e-05,
+      "loss": 1.8195,
+      "step": 26070
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.392e-05,
+      "loss": 1.9003,
+      "step": 26080
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3910000000000003e-05,
+      "loss": 1.9106,
+      "step": 26090
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.39e-05,
+      "loss": 1.6891,
+      "step": 26100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.389e-05,
+      "loss": 1.81,
+      "step": 26110
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3880000000000002e-05,
+      "loss": 1.7865,
+      "step": 26120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.387e-05,
+      "loss": 1.9366,
+      "step": 26130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3860000000000002e-05,
+      "loss": 2.017,
+      "step": 26140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.385e-05,
+      "loss": 1.8984,
+      "step": 26150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3840000000000002e-05,
+      "loss": 1.8163,
+      "step": 26160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3830000000000004e-05,
+      "loss": 1.7372,
+      "step": 26170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.3820000000000002e-05,
+      "loss": 1.7623,
+      "step": 26180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.381e-05,
+      "loss": 1.7759,
+      "step": 26190
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.38e-05,
+      "loss": 1.7681,
+      "step": 26200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.379e-05,
+      "loss": 1.8053,
+      "step": 26210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3780000000000003e-05,
+      "loss": 1.8326,
+      "step": 26220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.377e-05,
+      "loss": 1.7458,
+      "step": 26230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3760000000000003e-05,
+      "loss": 1.751,
+      "step": 26240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.375e-05,
+      "loss": 1.7357,
+      "step": 26250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.374e-05,
+      "loss": 1.706,
+      "step": 26260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.373e-05,
+      "loss": 1.6767,
+      "step": 26270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.372e-05,
+      "loss": 1.8655,
+      "step": 26280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.371e-05,
+      "loss": 1.7361,
+      "step": 26290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.37e-05,
+      "loss": 1.5933,
+      "step": 26300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3690000000000002e-05,
+      "loss": 1.8392,
+      "step": 26310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3680000000000004e-05,
+      "loss": 1.7855,
+      "step": 26320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3670000000000002e-05,
+      "loss": 1.7963,
+      "step": 26330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.366e-05,
+      "loss": 1.7501,
+      "step": 26340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.365e-05,
+      "loss": 1.8133,
+      "step": 26350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.364e-05,
+      "loss": 1.617,
+      "step": 26360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3630000000000002e-05,
+      "loss": 1.7403,
+      "step": 26370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.362e-05,
+      "loss": 1.7942,
+      "step": 26380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3610000000000003e-05,
+      "loss": 1.8563,
+      "step": 26390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.36e-05,
+      "loss": 1.8231,
+      "step": 26400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.359e-05,
+      "loss": 1.8679,
+      "step": 26410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.358e-05,
+      "loss": 1.7753,
+      "step": 26420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.357e-05,
+      "loss": 1.7047,
+      "step": 26430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.356e-05,
+      "loss": 1.9255,
+      "step": 26440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.355e-05,
+      "loss": 1.694,
+      "step": 26450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.354e-05,
+      "loss": 1.695,
+      "step": 26460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3530000000000003e-05,
+      "loss": 1.664,
+      "step": 26470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3520000000000002e-05,
+      "loss": 1.6987,
+      "step": 26480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.351e-05,
+      "loss": 1.8534,
+      "step": 26490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.35e-05,
+      "loss": 1.7977,
+      "step": 26500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.349e-05,
+      "loss": 1.7898,
+      "step": 26510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3480000000000002e-05,
+      "loss": 1.6716,
+      "step": 26520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.347e-05,
+      "loss": 1.8509,
+      "step": 26530
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.3460000000000002e-05,
+      "loss": 1.7883,
+      "step": 26540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.345e-05,
+      "loss": 1.6583,
+      "step": 26550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.344e-05,
+      "loss": 1.7734,
+      "step": 26560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.343e-05,
+      "loss": 1.8128,
+      "step": 26570
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.342e-05,
+      "loss": 1.9391,
+      "step": 26580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.341e-05,
+      "loss": 1.9014,
+      "step": 26590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 1.8231,
+      "step": 26600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.339e-05,
+      "loss": 1.6392,
+      "step": 26610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3380000000000003e-05,
+      "loss": 1.7348,
+      "step": 26620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.337e-05,
+      "loss": 1.7437,
+      "step": 26630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.336e-05,
+      "loss": 1.9593,
+      "step": 26640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3350000000000002e-05,
+      "loss": 1.7122,
+      "step": 26650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.334e-05,
+      "loss": 1.8559,
+      "step": 26660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3330000000000002e-05,
+      "loss": 1.7135,
+      "step": 26670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.332e-05,
+      "loss": 1.7933,
+      "step": 26680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3310000000000002e-05,
+      "loss": 1.8359,
+      "step": 26690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3300000000000004e-05,
+      "loss": 1.6993,
+      "step": 26700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3290000000000002e-05,
+      "loss": 1.81,
+      "step": 26710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.328e-05,
+      "loss": 1.6277,
+      "step": 26720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.327e-05,
+      "loss": 1.7365,
+      "step": 26730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.326e-05,
+      "loss": 1.8539,
+      "step": 26740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3250000000000003e-05,
+      "loss": 1.7655,
+      "step": 26750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.324e-05,
+      "loss": 1.7625,
+      "step": 26760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3230000000000003e-05,
+      "loss": 1.7708,
+      "step": 26770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.322e-05,
+      "loss": 1.7143,
+      "step": 26780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.321e-05,
+      "loss": 1.6412,
+      "step": 26790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.32e-05,
+      "loss": 1.7817,
+      "step": 26800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.319e-05,
+      "loss": 1.8891,
+      "step": 26810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.318e-05,
+      "loss": 1.8717,
+      "step": 26820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.317e-05,
+      "loss": 1.6799,
+      "step": 26830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3160000000000002e-05,
+      "loss": 1.7527,
+      "step": 26840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3150000000000004e-05,
+      "loss": 1.8477,
+      "step": 26850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.3140000000000002e-05,
+      "loss": 1.8854,
+      "step": 26860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.313e-05,
+      "loss": 1.8073,
+      "step": 26870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.312e-05,
+      "loss": 1.6824,
+      "step": 26880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.311e-05,
+      "loss": 1.7503,
+      "step": 26890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3100000000000002e-05,
+      "loss": 1.799,
+      "step": 26900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.309e-05,
+      "loss": 1.6581,
+      "step": 26910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3080000000000003e-05,
+      "loss": 1.7352,
+      "step": 26920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.307e-05,
+      "loss": 1.8112,
+      "step": 26930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.306e-05,
+      "loss": 1.8097,
+      "step": 26940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.305e-05,
+      "loss": 1.7859,
+      "step": 26950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.304e-05,
+      "loss": 1.6707,
+      "step": 26960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.303e-05,
+      "loss": 1.7402,
+      "step": 26970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.302e-05,
+      "loss": 1.7898,
+      "step": 26980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.301e-05,
+      "loss": 1.7345,
+      "step": 26990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 1.9801,
+      "step": 27000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2990000000000002e-05,
+      "loss": 1.7532,
+      "step": 27010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.298e-05,
+      "loss": 1.7272,
+      "step": 27020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.297e-05,
+      "loss": 1.8078,
+      "step": 27030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.296e-05,
+      "loss": 1.8798,
+      "step": 27040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2950000000000002e-05,
+      "loss": 1.7551,
+      "step": 27050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.294e-05,
+      "loss": 1.6913,
+      "step": 27060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2930000000000002e-05,
+      "loss": 1.802,
+      "step": 27070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.292e-05,
+      "loss": 1.5824,
+      "step": 27080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2910000000000003e-05,
+      "loss": 1.7564,
+      "step": 27090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.29e-05,
+      "loss": 1.8328,
+      "step": 27100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.289e-05,
+      "loss": 1.8683,
+      "step": 27110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.288e-05,
+      "loss": 1.7756,
+      "step": 27120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.287e-05,
+      "loss": 1.6826,
+      "step": 27130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.286e-05,
+      "loss": 1.6758,
+      "step": 27140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2850000000000003e-05,
+      "loss": 1.8069,
+      "step": 27150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.284e-05,
+      "loss": 1.8181,
+      "step": 27160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.283e-05,
+      "loss": 1.8222,
+      "step": 27170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.282e-05,
+      "loss": 1.819,
+      "step": 27180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.281e-05,
+      "loss": 1.8714,
+      "step": 27190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 1.8758,
+      "step": 27200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.279e-05,
+      "loss": 1.9029,
+      "step": 27210
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2780000000000002e-05,
+      "loss": 1.7858,
+      "step": 27220
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2770000000000004e-05,
+      "loss": 1.6941,
+      "step": 27230
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2760000000000002e-05,
+      "loss": 1.649,
+      "step": 27240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.275e-05,
+      "loss": 1.6204,
+      "step": 27250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.274e-05,
+      "loss": 1.7316,
+      "step": 27260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.273e-05,
+      "loss": 1.7371,
+      "step": 27270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2720000000000003e-05,
+      "loss": 1.9391,
+      "step": 27280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.271e-05,
+      "loss": 1.7544,
+      "step": 27290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2700000000000003e-05,
+      "loss": 1.7482,
+      "step": 27300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.269e-05,
+      "loss": 1.8188,
+      "step": 27310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.268e-05,
+      "loss": 1.6825,
+      "step": 27320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.267e-05,
+      "loss": 1.685,
+      "step": 27330
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.266e-05,
+      "loss": 1.6029,
+      "step": 27340
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.265e-05,
+      "loss": 1.752,
+      "step": 27350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.264e-05,
+      "loss": 1.6974,
+      "step": 27360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2630000000000002e-05,
+      "loss": 1.7889,
+      "step": 27370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2620000000000004e-05,
+      "loss": 1.7815,
+      "step": 27380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2610000000000002e-05,
+      "loss": 1.6904,
+      "step": 27390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.26e-05,
+      "loss": 1.6955,
+      "step": 27400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.259e-05,
+      "loss": 1.9031,
+      "step": 27410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.258e-05,
+      "loss": 1.7787,
+      "step": 27420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2570000000000002e-05,
+      "loss": 1.66,
+      "step": 27430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.256e-05,
+      "loss": 1.7933,
+      "step": 27440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2550000000000003e-05,
+      "loss": 1.752,
+      "step": 27450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.254e-05,
+      "loss": 1.638,
+      "step": 27460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.253e-05,
+      "loss": 1.7469,
+      "step": 27470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.252e-05,
+      "loss": 1.5794,
+      "step": 27480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.251e-05,
+      "loss": 1.7614,
+      "step": 27490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.25e-05,
+      "loss": 1.7772,
+      "step": 27500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.249e-05,
+      "loss": 1.5684,
+      "step": 27510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.248e-05,
+      "loss": 1.7281,
+      "step": 27520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2470000000000003e-05,
+      "loss": 1.6606,
+      "step": 27530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2460000000000002e-05,
+      "loss": 1.8255,
+      "step": 27540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.245e-05,
+      "loss": 1.821,
+      "step": 27550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.244e-05,
+      "loss": 1.7818,
+      "step": 27560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.243e-05,
+      "loss": 1.6604,
+      "step": 27570
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2420000000000002e-05,
+      "loss": 1.5789,
+      "step": 27580
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.241e-05,
+      "loss": 1.933,
+      "step": 27590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2400000000000002e-05,
+      "loss": 1.8337,
+      "step": 27600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.239e-05,
+      "loss": 1.7513,
+      "step": 27610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2380000000000003e-05,
+      "loss": 1.7788,
+      "step": 27620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.237e-05,
+      "loss": 1.7813,
+      "step": 27630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.236e-05,
+      "loss": 1.8456,
+      "step": 27640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.235e-05,
+      "loss": 1.8513,
+      "step": 27650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.234e-05,
+      "loss": 1.6633,
+      "step": 27660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.233e-05,
+      "loss": 1.6163,
+      "step": 27670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2320000000000003e-05,
+      "loss": 1.7243,
+      "step": 27680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.231e-05,
+      "loss": 1.727,
+      "step": 27690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.23e-05,
+      "loss": 1.7903,
+      "step": 27700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.229e-05,
+      "loss": 1.664,
+      "step": 27710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.228e-05,
+      "loss": 1.7081,
+      "step": 27720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2270000000000002e-05,
+      "loss": 1.5651,
+      "step": 27730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.226e-05,
+      "loss": 1.7577,
+      "step": 27740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2250000000000002e-05,
+      "loss": 1.6718,
+      "step": 27750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.224e-05,
+      "loss": 1.7158,
+      "step": 27760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2230000000000002e-05,
+      "loss": 1.9132,
+      "step": 27770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.222e-05,
+      "loss": 1.6729,
+      "step": 27780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.221e-05,
+      "loss": 1.7463,
+      "step": 27790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.22e-05,
+      "loss": 1.7122,
+      "step": 27800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.219e-05,
+      "loss": 1.6008,
+      "step": 27810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.218e-05,
+      "loss": 1.7445,
+      "step": 27820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2170000000000003e-05,
+      "loss": 1.802,
+      "step": 27830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.216e-05,
+      "loss": 1.8212,
+      "step": 27840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.215e-05,
+      "loss": 1.6896,
+      "step": 27850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.214e-05,
+      "loss": 1.7449,
+      "step": 27860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.213e-05,
+      "loss": 1.867,
+      "step": 27870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.212e-05,
+      "loss": 1.7685,
+      "step": 27880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.211e-05,
+      "loss": 1.7142,
+      "step": 27890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2100000000000002e-05,
+      "loss": 1.7243,
+      "step": 27900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2090000000000004e-05,
+      "loss": 1.8258,
+      "step": 27910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2080000000000002e-05,
+      "loss": 1.7739,
+      "step": 27920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.207e-05,
+      "loss": 1.7094,
+      "step": 27930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.206e-05,
+      "loss": 1.8315,
+      "step": 27940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.205e-05,
+      "loss": 1.8049,
+      "step": 27950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2040000000000002e-05,
+      "loss": 1.8919,
+      "step": 27960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.203e-05,
+      "loss": 1.7214,
+      "step": 27970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2020000000000003e-05,
+      "loss": 1.8363,
+      "step": 27980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.201e-05,
+      "loss": 1.6905,
+      "step": 27990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 1.8609,
+      "step": 28000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.199e-05,
+      "loss": 1.6889,
+      "step": 28010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.198e-05,
+      "loss": 1.6315,
+      "step": 28020
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.197e-05,
+      "loss": 1.6235,
+      "step": 28030
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.196e-05,
+      "loss": 1.653,
+      "step": 28040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.195e-05,
+      "loss": 1.6484,
+      "step": 28050
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1940000000000003e-05,
+      "loss": 1.6448,
+      "step": 28060
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1930000000000002e-05,
+      "loss": 1.8275,
+      "step": 28070
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.192e-05,
+      "loss": 1.7648,
+      "step": 28080
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.191e-05,
+      "loss": 1.7795,
+      "step": 28090
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.19e-05,
+      "loss": 1.7407,
+      "step": 28100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1890000000000002e-05,
+      "loss": 1.8232,
+      "step": 28110
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.188e-05,
+      "loss": 1.7157,
+      "step": 28120
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1870000000000002e-05,
+      "loss": 1.6182,
+      "step": 28130
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.186e-05,
+      "loss": 1.742,
+      "step": 28140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1850000000000003e-05,
+      "loss": 1.7386,
+      "step": 28150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.184e-05,
+      "loss": 1.7879,
+      "step": 28160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.183e-05,
+      "loss": 1.5792,
+      "step": 28170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.182e-05,
+      "loss": 1.7692,
+      "step": 28180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.181e-05,
+      "loss": 1.7697,
+      "step": 28190
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.18e-05,
+      "loss": 1.7417,
+      "step": 28200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1790000000000003e-05,
+      "loss": 1.8308,
+      "step": 28210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.178e-05,
+      "loss": 1.5906,
+      "step": 28220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.177e-05,
+      "loss": 1.7572,
+      "step": 28230
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.176e-05,
+      "loss": 1.7159,
+      "step": 28240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.175e-05,
+      "loss": 1.7134,
+      "step": 28250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1740000000000002e-05,
+      "loss": 1.7472,
+      "step": 28260
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.173e-05,
+      "loss": 1.9046,
+      "step": 28270
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1720000000000002e-05,
+      "loss": 1.6921,
+      "step": 28280
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.171e-05,
+      "loss": 1.825,
+      "step": 28290
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1700000000000002e-05,
+      "loss": 1.753,
+      "step": 28300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.169e-05,
+      "loss": 1.692,
+      "step": 28310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.168e-05,
+      "loss": 1.7054,
+      "step": 28320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.167e-05,
+      "loss": 1.634,
+      "step": 28330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.166e-05,
+      "loss": 1.6136,
+      "step": 28340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.165e-05,
+      "loss": 1.6965,
+      "step": 28350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1640000000000003e-05,
+      "loss": 1.6957,
+      "step": 28360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.163e-05,
+      "loss": 1.7422,
+      "step": 28370
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.162e-05,
+      "loss": 1.7393,
+      "step": 28380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1609999999999998e-05,
+      "loss": 1.7829,
+      "step": 28390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.16e-05,
+      "loss": 1.5845,
+      "step": 28400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.159e-05,
+      "loss": 1.6879,
+      "step": 28410
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.158e-05,
+      "loss": 1.8627,
+      "step": 28420
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1570000000000002e-05,
+      "loss": 1.6167,
+      "step": 28430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1560000000000004e-05,
+      "loss": 1.9048,
+      "step": 28440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1550000000000002e-05,
+      "loss": 1.7347,
+      "step": 28450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.154e-05,
+      "loss": 1.7033,
+      "step": 28460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.153e-05,
+      "loss": 1.7937,
+      "step": 28470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.152e-05,
+      "loss": 1.8599,
+      "step": 28480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1510000000000002e-05,
+      "loss": 1.8193,
+      "step": 28490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.15e-05,
+      "loss": 1.7272,
+      "step": 28500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1490000000000003e-05,
+      "loss": 1.7886,
+      "step": 28510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.148e-05,
+      "loss": 1.6067,
+      "step": 28520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1470000000000003e-05,
+      "loss": 1.7905,
+      "step": 28530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.146e-05,
+      "loss": 1.8225,
+      "step": 28540
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.145e-05,
+      "loss": 1.697,
+      "step": 28550
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.144e-05,
+      "loss": 1.8633,
+      "step": 28560
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.143e-05,
+      "loss": 1.7384,
+      "step": 28570
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.142e-05,
+      "loss": 1.7507,
+      "step": 28580
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1410000000000003e-05,
+      "loss": 1.8902,
+      "step": 28590
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1400000000000002e-05,
+      "loss": 1.6339,
+      "step": 28600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.139e-05,
+      "loss": 1.711,
+      "step": 28610
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.138e-05,
+      "loss": 1.6593,
+      "step": 28620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.137e-05,
+      "loss": 1.6229,
+      "step": 28630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1360000000000002e-05,
+      "loss": 1.642,
+      "step": 28640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.135e-05,
+      "loss": 1.6759,
+      "step": 28650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.1340000000000002e-05,
+      "loss": 1.5142,
+      "step": 28660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.133e-05,
+      "loss": 1.7527,
+      "step": 28670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.1320000000000003e-05,
+      "loss": 1.5974,
+      "step": 28680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.131e-05,
+      "loss": 1.6775,
+      "step": 28690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.13e-05,
+      "loss": 1.7238,
+      "step": 28700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.129e-05,
+      "loss": 1.633,
+      "step": 28710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.128e-05,
+      "loss": 1.6799,
+      "step": 28720
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.127e-05,
+      "loss": 1.7203,
+      "step": 28730
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.1260000000000003e-05,
+      "loss": 1.6795,
+      "step": 28740
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.125e-05,
+      "loss": 1.5905,
+      "step": 28750
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.124e-05,
+      "loss": 1.6862,
+      "step": 28760
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.123e-05,
+      "loss": 1.9181,
+      "step": 28770
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.122e-05,
+      "loss": 1.6523,
+      "step": 28780
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.1210000000000002e-05,
+      "loss": 1.6281,
+      "step": 28790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.12e-05,
+      "loss": 1.7445,
+      "step": 28800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.1190000000000002e-05,
+      "loss": 1.6548,
+      "step": 28810
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.118e-05,
+      "loss": 1.8021,
+      "step": 28820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.1170000000000002e-05,
+      "loss": 1.7611,
+      "step": 28830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.116e-05,
+      "loss": 1.7381,
+      "step": 28840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.115e-05,
+      "loss": 1.7194,
+      "step": 28850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.114e-05,
+      "loss": 1.6736,
+      "step": 28860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.113e-05,
+      "loss": 1.5164,
+      "step": 28870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.112e-05,
+      "loss": 1.8062,
+      "step": 28880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.1110000000000003e-05,
+      "loss": 1.6534,
+      "step": 28890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.11e-05,
+      "loss": 1.7804,
+      "step": 28900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.1090000000000003e-05,
+      "loss": 1.4374,
+      "step": 28910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.1079999999999998e-05,
+      "loss": 1.6392,
+      "step": 28920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.107e-05,
+      "loss": 1.7398,
+      "step": 28930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.106e-05,
+      "loss": 1.7145,
+      "step": 28940
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.105e-05,
+      "loss": 1.6976,
+      "step": 28950
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.1040000000000002e-05,
+      "loss": 1.7686,
+      "step": 28960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.103e-05,
+      "loss": 1.6854,
+      "step": 28970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.1020000000000002e-05,
+      "loss": 1.7198,
+      "step": 28980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.101e-05,
+      "loss": 1.6709,
+      "step": 28990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.1e-05,
+      "loss": 1.686,
+      "step": 29000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.099e-05,
+      "loss": 1.7976,
+      "step": 29010
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.098e-05,
+      "loss": 1.7279,
+      "step": 29020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.097e-05,
+      "loss": 1.7858,
+      "step": 29030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.0960000000000003e-05,
+      "loss": 1.6563,
+      "step": 29040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.095e-05,
+      "loss": 1.7685,
+      "step": 29050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.0940000000000003e-05,
+      "loss": 1.6627,
+      "step": 29060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.093e-05,
+      "loss": 1.6384,
+      "step": 29070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.092e-05,
+      "loss": 1.8181,
+      "step": 29080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.091e-05,
+      "loss": 1.6586,
+      "step": 29090
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.09e-05,
+      "loss": 1.5677,
+      "step": 29100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.089e-05,
+      "loss": 1.8111,
+      "step": 29110
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.0880000000000003e-05,
+      "loss": 1.657,
+      "step": 29120
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.0870000000000002e-05,
+      "loss": 1.8505,
+      "step": 29130
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.086e-05,
+      "loss": 1.8061,
+      "step": 29140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.085e-05,
+      "loss": 1.6564,
+      "step": 29150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.084e-05,
+      "loss": 1.9089,
+      "step": 29160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.0830000000000002e-05,
+      "loss": 1.7276,
+      "step": 29170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.082e-05,
+      "loss": 1.7835,
+      "step": 29180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.0810000000000002e-05,
+      "loss": 1.7184,
+      "step": 29190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.08e-05,
+      "loss": 1.6308,
+      "step": 29200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.0790000000000003e-05,
+      "loss": 1.6982,
+      "step": 29210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.078e-05,
+      "loss": 1.7386,
+      "step": 29220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.077e-05,
+      "loss": 1.6054,
+      "step": 29230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.076e-05,
+      "loss": 1.8456,
+      "step": 29240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.075e-05,
+      "loss": 1.684,
+      "step": 29250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.074e-05,
+      "loss": 1.6975,
+      "step": 29260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.0730000000000003e-05,
+      "loss": 1.5833,
+      "step": 29270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.072e-05,
+      "loss": 1.783,
+      "step": 29280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.0710000000000003e-05,
+      "loss": 1.7516,
+      "step": 29290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.07e-05,
+      "loss": 1.7071,
+      "step": 29300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.069e-05,
+      "loss": 1.6139,
+      "step": 29310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.0680000000000002e-05,
+      "loss": 1.6468,
+      "step": 29320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.067e-05,
+      "loss": 1.786,
+      "step": 29330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.0660000000000002e-05,
+      "loss": 1.873,
+      "step": 29340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.065e-05,
+      "loss": 1.5721,
+      "step": 29350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.0640000000000002e-05,
+      "loss": 1.5561,
+      "step": 29360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.063e-05,
+      "loss": 1.67,
+      "step": 29370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.062e-05,
+      "loss": 1.5956,
+      "step": 29380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.061e-05,
+      "loss": 1.8119,
+      "step": 29390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.06e-05,
+      "loss": 1.6707,
+      "step": 29400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.059e-05,
+      "loss": 1.6378,
+      "step": 29410
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.0580000000000003e-05,
+      "loss": 1.6778,
+      "step": 29420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.057e-05,
+      "loss": 1.7511,
+      "step": 29430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.0560000000000003e-05,
+      "loss": 1.7418,
+      "step": 29440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.055e-05,
+      "loss": 1.7055,
+      "step": 29450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.054e-05,
+      "loss": 1.4812,
+      "step": 29460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.053e-05,
+      "loss": 1.6969,
+      "step": 29470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.052e-05,
+      "loss": 1.523,
+      "step": 29480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.0510000000000002e-05,
+      "loss": 1.5878,
+      "step": 29490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.05e-05,
+      "loss": 1.802,
+      "step": 29500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.0490000000000002e-05,
+      "loss": 1.8271,
+      "step": 29510
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.048e-05,
+      "loss": 1.7373,
+      "step": 29520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.047e-05,
+      "loss": 1.713,
+      "step": 29530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.046e-05,
+      "loss": 1.7754,
+      "step": 29540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.045e-05,
+      "loss": 1.6347,
+      "step": 29550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.044e-05,
+      "loss": 1.714,
+      "step": 29560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.0430000000000003e-05,
+      "loss": 1.7778,
+      "step": 29570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.042e-05,
+      "loss": 1.6748,
+      "step": 29580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.0410000000000003e-05,
+      "loss": 1.7557,
+      "step": 29590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.04e-05,
+      "loss": 1.7256,
+      "step": 29600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.039e-05,
+      "loss": 1.5577,
+      "step": 29610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.038e-05,
+      "loss": 1.6099,
+      "step": 29620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.037e-05,
+      "loss": 1.7835,
+      "step": 29630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.036e-05,
+      "loss": 1.5758,
+      "step": 29640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.035e-05,
+      "loss": 1.5331,
+      "step": 29650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.0340000000000002e-05,
+      "loss": 1.5954,
+      "step": 29660
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.033e-05,
+      "loss": 1.7826,
+      "step": 29670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.032e-05,
+      "loss": 1.7583,
+      "step": 29680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.031e-05,
+      "loss": 1.6814,
+      "step": 29690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.0300000000000002e-05,
+      "loss": 1.9004,
+      "step": 29700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.029e-05,
+      "loss": 1.6956,
+      "step": 29710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.0280000000000002e-05,
+      "loss": 1.5276,
+      "step": 29720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.027e-05,
+      "loss": 1.7675,
+      "step": 29730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.0260000000000003e-05,
+      "loss": 1.5621,
+      "step": 29740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.025e-05,
+      "loss": 1.7218,
+      "step": 29750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.024e-05,
+      "loss": 1.5254,
+      "step": 29760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.023e-05,
+      "loss": 1.5399,
+      "step": 29770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.022e-05,
+      "loss": 1.8018,
+      "step": 29780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.021e-05,
+      "loss": 1.5771,
+      "step": 29790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.0200000000000003e-05,
+      "loss": 1.7906,
+      "step": 29800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.019e-05,
+      "loss": 1.7469,
+      "step": 29810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.0180000000000003e-05,
+      "loss": 1.5161,
+      "step": 29820
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.017e-05,
+      "loss": 1.7,
+      "step": 29830
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.016e-05,
+      "loss": 1.6838,
+      "step": 29840
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.0150000000000002e-05,
+      "loss": 1.745,
+      "step": 29850
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.014e-05,
+      "loss": 1.6218,
+      "step": 29860
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.0130000000000002e-05,
+      "loss": 1.8696,
+      "step": 29870
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.012e-05,
+      "loss": 1.7006,
+      "step": 29880
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.0110000000000002e-05,
+      "loss": 1.547,
+      "step": 29890
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.01e-05,
+      "loss": 1.7049,
+      "step": 29900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.009e-05,
+      "loss": 1.4708,
+      "step": 29910
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.008e-05,
+      "loss": 1.6427,
+      "step": 29920
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.007e-05,
+      "loss": 1.649,
+      "step": 29930
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.006e-05,
+      "loss": 1.736,
+      "step": 29940
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.0050000000000003e-05,
+      "loss": 1.6757,
+      "step": 29950
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.004e-05,
+      "loss": 1.7119,
+      "step": 29960
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.0030000000000003e-05,
+      "loss": 1.6503,
+      "step": 29970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.002e-05,
+      "loss": 1.6309,
+      "step": 29980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.001e-05,
+      "loss": 1.5919,
+      "step": 29990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 1.7332,
+      "step": 30000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999e-05,
+      "loss": 1.6572,
+      "step": 30010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 1.833,
+      "step": 30020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.997e-05,
+      "loss": 1.7864,
+      "step": 30030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9960000000000002e-05,
+      "loss": 1.7469,
+      "step": 30040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.995e-05,
+      "loss": 1.667,
+      "step": 30050
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.994e-05,
+      "loss": 1.7012,
+      "step": 30060
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.993e-05,
+      "loss": 1.6135,
+      "step": 30070
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.992e-05,
+      "loss": 1.7816,
+      "step": 30080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.991e-05,
+      "loss": 1.6835,
+      "step": 30090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9900000000000003e-05,
+      "loss": 1.5736,
+      "step": 30100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.989e-05,
+      "loss": 1.6714,
+      "step": 30110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9880000000000003e-05,
+      "loss": 1.6035,
+      "step": 30120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.987e-05,
+      "loss": 1.5046,
+      "step": 30130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.986e-05,
+      "loss": 1.6579,
+      "step": 30140
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.985e-05,
+      "loss": 1.5405,
+      "step": 30150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.984e-05,
+      "loss": 1.5525,
+      "step": 30160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.983e-05,
+      "loss": 1.6951,
+      "step": 30170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.982e-05,
+      "loss": 1.579,
+      "step": 30180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9810000000000002e-05,
+      "loss": 1.7095,
+      "step": 30190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 1.5394,
+      "step": 30200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.979e-05,
+      "loss": 1.8664,
+      "step": 30210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.978e-05,
+      "loss": 1.8439,
+      "step": 30220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.977e-05,
+      "loss": 1.5523,
+      "step": 30230
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.976e-05,
+      "loss": 1.7964,
+      "step": 30240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9750000000000002e-05,
+      "loss": 1.5611,
+      "step": 30250
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.974e-05,
+      "loss": 1.4753,
+      "step": 30260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9730000000000003e-05,
+      "loss": 1.5579,
+      "step": 30270
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.972e-05,
+      "loss": 1.6952,
+      "step": 30280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.971e-05,
+      "loss": 1.7253,
+      "step": 30290
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.97e-05,
+      "loss": 1.6226,
+      "step": 30300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.969e-05,
+      "loss": 1.5602,
+      "step": 30310
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.968e-05,
+      "loss": 1.54,
+      "step": 30320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9670000000000003e-05,
+      "loss": 1.7408,
+      "step": 30330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.966e-05,
+      "loss": 1.6387,
+      "step": 30340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9650000000000003e-05,
+      "loss": 1.6204,
+      "step": 30350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9640000000000002e-05,
+      "loss": 1.7485,
+      "step": 30360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.963e-05,
+      "loss": 1.4868,
+      "step": 30370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9620000000000002e-05,
+      "loss": 1.7123,
+      "step": 30380
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.961e-05,
+      "loss": 1.5183,
+      "step": 30390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 1.7144,
+      "step": 30400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.959e-05,
+      "loss": 1.6784,
+      "step": 30410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 1.578,
+      "step": 30420
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.957e-05,
+      "loss": 1.7343,
+      "step": 30430
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.956e-05,
+      "loss": 1.5555,
+      "step": 30440
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.955e-05,
+      "loss": 1.6819,
+      "step": 30450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.954e-05,
+      "loss": 1.8272,
+      "step": 30460
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.953e-05,
+      "loss": 1.5441,
+      "step": 30470
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9520000000000003e-05,
+      "loss": 1.471,
+      "step": 30480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.951e-05,
+      "loss": 1.5941,
+      "step": 30490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9500000000000003e-05,
+      "loss": 1.5063,
+      "step": 30500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.949e-05,
+      "loss": 1.5973,
+      "step": 30510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.948e-05,
+      "loss": 1.6185,
+      "step": 30520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.947e-05,
+      "loss": 1.7955,
+      "step": 30530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.946e-05,
+      "loss": 1.6562,
+      "step": 30540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9450000000000002e-05,
+      "loss": 1.5953,
+      "step": 30550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.944e-05,
+      "loss": 1.6702,
+      "step": 30560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9430000000000002e-05,
+      "loss": 1.7126,
+      "step": 30570
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.942e-05,
+      "loss": 1.6902,
+      "step": 30580
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.941e-05,
+      "loss": 1.4949,
+      "step": 30590
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.94e-05,
+      "loss": 1.6385,
+      "step": 30600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.939e-05,
+      "loss": 1.5749,
+      "step": 30610
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.938e-05,
+      "loss": 1.5695,
+      "step": 30620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9370000000000003e-05,
+      "loss": 1.4914,
+      "step": 30630
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.936e-05,
+      "loss": 1.6691,
+      "step": 30640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9350000000000003e-05,
+      "loss": 1.6542,
+      "step": 30650
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.934e-05,
+      "loss": 1.7178,
+      "step": 30660
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.933e-05,
+      "loss": 1.6096,
+      "step": 30670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.932e-05,
+      "loss": 1.5688,
+      "step": 30680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.931e-05,
+      "loss": 1.5784,
+      "step": 30690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.93e-05,
+      "loss": 1.716,
+      "step": 30700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.929e-05,
+      "loss": 1.6669,
+      "step": 30710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9280000000000002e-05,
+      "loss": 1.6759,
+      "step": 30720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9270000000000004e-05,
+      "loss": 1.5769,
+      "step": 30730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9260000000000002e-05,
+      "loss": 1.6725,
+      "step": 30740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.925e-05,
+      "loss": 1.6252,
+      "step": 30750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.924e-05,
+      "loss": 1.619,
+      "step": 30760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.923e-05,
+      "loss": 1.6372,
+      "step": 30770
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9220000000000002e-05,
+      "loss": 1.6887,
+      "step": 30780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.921e-05,
+      "loss": 1.7176,
+      "step": 30790
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 1.5836,
+      "step": 30800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.919e-05,
+      "loss": 1.7729,
+      "step": 30810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.918e-05,
+      "loss": 1.7222,
+      "step": 30820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.917e-05,
+      "loss": 1.5124,
+      "step": 30830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.916e-05,
+      "loss": 1.6695,
+      "step": 30840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.915e-05,
+      "loss": 1.6258,
+      "step": 30850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.914e-05,
+      "loss": 1.6581,
+      "step": 30860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.913e-05,
+      "loss": 1.6797,
+      "step": 30870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9120000000000003e-05,
+      "loss": 1.6178,
+      "step": 30880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.911e-05,
+      "loss": 1.6714,
+      "step": 30890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.91e-05,
+      "loss": 1.4649,
+      "step": 30900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.909e-05,
+      "loss": 1.5849,
+      "step": 30910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.908e-05,
+      "loss": 1.6169,
+      "step": 30920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9070000000000002e-05,
+      "loss": 1.5488,
+      "step": 30930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.906e-05,
+      "loss": 1.6514,
+      "step": 30940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9050000000000002e-05,
+      "loss": 1.4876,
+      "step": 30950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.904e-05,
+      "loss": 1.6373,
+      "step": 30960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.903e-05,
+      "loss": 1.6916,
+      "step": 30970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.902e-05,
+      "loss": 1.546,
+      "step": 30980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.901e-05,
+      "loss": 1.6881,
+      "step": 30990
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9e-05,
+      "loss": 1.678,
+      "step": 31000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8990000000000003e-05,
+      "loss": 1.6678,
+      "step": 31010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.898e-05,
+      "loss": 1.5599,
+      "step": 31020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8970000000000003e-05,
+      "loss": 1.6641,
+      "step": 31030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.896e-05,
+      "loss": 1.6307,
+      "step": 31040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.895e-05,
+      "loss": 1.5987,
+      "step": 31050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.894e-05,
+      "loss": 1.6824,
+      "step": 31060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.893e-05,
+      "loss": 1.5804,
+      "step": 31070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8920000000000002e-05,
+      "loss": 1.6239,
+      "step": 31080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.891e-05,
+      "loss": 1.538,
+      "step": 31090
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8900000000000002e-05,
+      "loss": 1.6155,
+      "step": 31100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8890000000000004e-05,
+      "loss": 1.6709,
+      "step": 31110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.888e-05,
+      "loss": 1.6782,
+      "step": 31120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887e-05,
+      "loss": 1.7043,
+      "step": 31130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886e-05,
+      "loss": 1.7087,
+      "step": 31140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.885e-05,
+      "loss": 1.5487,
+      "step": 31150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8840000000000003e-05,
+      "loss": 1.6752,
+      "step": 31160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.883e-05,
+      "loss": 1.6471,
+      "step": 31170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8820000000000003e-05,
+      "loss": 1.5962,
+      "step": 31180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.881e-05,
+      "loss": 1.5945,
+      "step": 31190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.88e-05,
+      "loss": 1.5677,
+      "step": 31200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.879e-05,
+      "loss": 1.5313,
+      "step": 31210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.878e-05,
+      "loss": 1.5828,
+      "step": 31220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.877e-05,
+      "loss": 1.5799,
+      "step": 31230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.876e-05,
+      "loss": 1.6871,
+      "step": 31240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 1.6628,
+      "step": 31250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8740000000000004e-05,
+      "loss": 1.6038,
+      "step": 31260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8730000000000002e-05,
+      "loss": 1.6646,
+      "step": 31270
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.872e-05,
+      "loss": 1.6872,
+      "step": 31280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.871e-05,
+      "loss": 1.5406,
+      "step": 31290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.87e-05,
+      "loss": 1.6441,
+      "step": 31300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8690000000000002e-05,
+      "loss": 1.6494,
+      "step": 31310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.868e-05,
+      "loss": 1.5994,
+      "step": 31320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8670000000000003e-05,
+      "loss": 1.4365,
+      "step": 31330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.866e-05,
+      "loss": 1.5138,
+      "step": 31340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.865e-05,
+      "loss": 1.5579,
+      "step": 31350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.864e-05,
+      "loss": 1.6254,
+      "step": 31360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.863e-05,
+      "loss": 1.678,
+      "step": 31370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.862e-05,
+      "loss": 1.5616,
+      "step": 31380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.861e-05,
+      "loss": 1.6192,
+      "step": 31390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.86e-05,
+      "loss": 1.5705,
+      "step": 31400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8590000000000003e-05,
+      "loss": 1.5705,
+      "step": 31410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.858e-05,
+      "loss": 1.5358,
+      "step": 31420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.857e-05,
+      "loss": 1.7382,
+      "step": 31430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.856e-05,
+      "loss": 1.6473,
+      "step": 31440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.855e-05,
+      "loss": 1.4963,
+      "step": 31450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8540000000000002e-05,
+      "loss": 1.5669,
+      "step": 31460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.853e-05,
+      "loss": 1.6247,
+      "step": 31470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8520000000000002e-05,
+      "loss": 1.5737,
+      "step": 31480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.851e-05,
+      "loss": 1.5284,
+      "step": 31490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.85e-05,
+      "loss": 1.6028,
+      "step": 31500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.849e-05,
+      "loss": 1.6826,
+      "step": 31510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.848e-05,
+      "loss": 1.7052,
+      "step": 31520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.847e-05,
+      "loss": 1.6226,
+      "step": 31530
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.846e-05,
+      "loss": 1.4318,
+      "step": 31540
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.845e-05,
+      "loss": 1.6176,
+      "step": 31550
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8440000000000003e-05,
+      "loss": 1.6029,
+      "step": 31560
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.843e-05,
+      "loss": 1.5352,
+      "step": 31570
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.842e-05,
+      "loss": 1.5698,
+      "step": 31580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.841e-05,
+      "loss": 1.5083,
+      "step": 31590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.84e-05,
+      "loss": 1.6225,
+      "step": 31600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8390000000000002e-05,
+      "loss": 1.6022,
+      "step": 31610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.838e-05,
+      "loss": 1.5386,
+      "step": 31620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8370000000000002e-05,
+      "loss": 1.5376,
+      "step": 31630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8360000000000004e-05,
+      "loss": 1.591,
+      "step": 31640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8350000000000002e-05,
+      "loss": 1.5229,
+      "step": 31650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.834e-05,
+      "loss": 1.607,
+      "step": 31660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.833e-05,
+      "loss": 1.7725,
+      "step": 31670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.832e-05,
+      "loss": 1.5698,
+      "step": 31680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8310000000000003e-05,
+      "loss": 1.6618,
+      "step": 31690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.83e-05,
+      "loss": 1.6779,
+      "step": 31700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8290000000000003e-05,
+      "loss": 1.503,
+      "step": 31710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.828e-05,
+      "loss": 1.6703,
+      "step": 31720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.827e-05,
+      "loss": 1.5668,
+      "step": 31730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.826e-05,
+      "loss": 1.654,
+      "step": 31740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.825e-05,
+      "loss": 1.6783,
+      "step": 31750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.824e-05,
+      "loss": 1.6835,
+      "step": 31760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.823e-05,
+      "loss": 1.6078,
+      "step": 31770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8220000000000002e-05,
+      "loss": 1.5454,
+      "step": 31780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8210000000000004e-05,
+      "loss": 1.5621,
+      "step": 31790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 1.601,
+      "step": 31800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.819e-05,
+      "loss": 1.6797,
+      "step": 31810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.818e-05,
+      "loss": 1.6359,
+      "step": 31820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.817e-05,
+      "loss": 1.5574,
+      "step": 31830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8160000000000002e-05,
+      "loss": 1.5116,
+      "step": 31840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.815e-05,
+      "loss": 1.7029,
+      "step": 31850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8140000000000003e-05,
+      "loss": 1.4669,
+      "step": 31860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.813e-05,
+      "loss": 1.5774,
+      "step": 31870
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.812e-05,
+      "loss": 1.7258,
+      "step": 31880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.811e-05,
+      "loss": 1.5738,
+      "step": 31890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.81e-05,
+      "loss": 1.6328,
+      "step": 31900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.809e-05,
+      "loss": 1.6144,
+      "step": 31910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.808e-05,
+      "loss": 1.5906,
+      "step": 31920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.807e-05,
+      "loss": 1.4574,
+      "step": 31930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8060000000000003e-05,
+      "loss": 1.5081,
+      "step": 31940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.805e-05,
+      "loss": 1.619,
+      "step": 31950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.804e-05,
+      "loss": 1.6831,
+      "step": 31960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.803e-05,
+      "loss": 1.6781,
+      "step": 31970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.802e-05,
+      "loss": 1.6808,
+      "step": 31980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8010000000000002e-05,
+      "loss": 1.5649,
+      "step": 31990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8e-05,
+      "loss": 1.6694,
+      "step": 32000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7990000000000002e-05,
+      "loss": 1.6122,
+      "step": 32010
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.798e-05,
+      "loss": 1.5585,
+      "step": 32020
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.797e-05,
+      "loss": 1.5524,
+      "step": 32030
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.796e-05,
+      "loss": 1.6081,
+      "step": 32040
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.795e-05,
+      "loss": 1.687,
+      "step": 32050
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.794e-05,
+      "loss": 1.5598,
+      "step": 32060
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.793e-05,
+      "loss": 1.5645,
+      "step": 32070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.792e-05,
+      "loss": 1.6375,
+      "step": 32080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7910000000000003e-05,
+      "loss": 1.7199,
+      "step": 32090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.79e-05,
+      "loss": 1.4672,
+      "step": 32100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.789e-05,
+      "loss": 1.5462,
+      "step": 32110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7879999999999998e-05,
+      "loss": 1.3889,
+      "step": 32120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.787e-05,
+      "loss": 1.6909,
+      "step": 32130
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7860000000000002e-05,
+      "loss": 1.5881,
+      "step": 32140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.785e-05,
+      "loss": 1.7833,
+      "step": 32150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7840000000000002e-05,
+      "loss": 1.5657,
+      "step": 32160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.783e-05,
+      "loss": 1.5098,
+      "step": 32170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7820000000000002e-05,
+      "loss": 1.6481,
+      "step": 32180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.781e-05,
+      "loss": 1.6807,
+      "step": 32190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.78e-05,
+      "loss": 1.6463,
+      "step": 32200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.779e-05,
+      "loss": 1.6883,
+      "step": 32210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 1.4874,
+      "step": 32220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.777e-05,
+      "loss": 1.667,
+      "step": 32230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7760000000000003e-05,
+      "loss": 1.6637,
+      "step": 32240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.775e-05,
+      "loss": 1.5136,
+      "step": 32250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.774e-05,
+      "loss": 1.4885,
+      "step": 32260
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.773e-05,
+      "loss": 1.5778,
+      "step": 32270
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.772e-05,
+      "loss": 1.6508,
+      "step": 32280
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.771e-05,
+      "loss": 1.637,
+      "step": 32290
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.77e-05,
+      "loss": 1.5288,
+      "step": 32300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7690000000000002e-05,
+      "loss": 1.5295,
+      "step": 32310
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7680000000000004e-05,
+      "loss": 1.5726,
+      "step": 32320
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7670000000000002e-05,
+      "loss": 1.5514,
+      "step": 32330
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.766e-05,
+      "loss": 1.5181,
+      "step": 32340
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.765e-05,
+      "loss": 1.5839,
+      "step": 32350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.764e-05,
+      "loss": 1.5892,
+      "step": 32360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7630000000000002e-05,
+      "loss": 1.6358,
+      "step": 32370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.762e-05,
+      "loss": 1.5642,
+      "step": 32380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7610000000000002e-05,
+      "loss": 1.5474,
+      "step": 32390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.76e-05,
+      "loss": 1.6437,
+      "step": 32400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.759e-05,
+      "loss": 1.6494,
+      "step": 32410
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.758e-05,
+      "loss": 1.5676,
+      "step": 32420
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.757e-05,
+      "loss": 1.4555,
+      "step": 32430
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.756e-05,
+      "loss": 1.7098,
+      "step": 32440
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.755e-05,
+      "loss": 1.6517,
+      "step": 32450
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.754e-05,
+      "loss": 1.4782,
+      "step": 32460
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7530000000000003e-05,
+      "loss": 1.6666,
+      "step": 32470
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.752e-05,
+      "loss": 1.8153,
+      "step": 32480
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.751e-05,
+      "loss": 1.6375,
+      "step": 32490
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.75e-05,
+      "loss": 1.5815,
+      "step": 32500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.749e-05,
+      "loss": 1.58,
+      "step": 32510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7480000000000002e-05,
+      "loss": 1.5614,
+      "step": 32520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.747e-05,
+      "loss": 1.6765,
+      "step": 32530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7460000000000002e-05,
+      "loss": 1.5322,
+      "step": 32540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.745e-05,
+      "loss": 1.6176,
+      "step": 32550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7440000000000002e-05,
+      "loss": 1.4826,
+      "step": 32560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.743e-05,
+      "loss": 1.6149,
+      "step": 32570
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.742e-05,
+      "loss": 1.394,
+      "step": 32580
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.741e-05,
+      "loss": 1.6397,
+      "step": 32590
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.74e-05,
+      "loss": 1.661,
+      "step": 32600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.739e-05,
+      "loss": 1.5637,
+      "step": 32610
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 1.5044,
+      "step": 32620
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.737e-05,
+      "loss": 1.4704,
+      "step": 32630
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.736e-05,
+      "loss": 1.7192,
+      "step": 32640
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7349999999999998e-05,
+      "loss": 1.4885,
+      "step": 32650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.734e-05,
+      "loss": 1.5379,
+      "step": 32660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7330000000000002e-05,
+      "loss": 1.521,
+      "step": 32670
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.732e-05,
+      "loss": 1.7711,
+      "step": 32680
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7310000000000002e-05,
+      "loss": 1.6033,
+      "step": 32690
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.73e-05,
+      "loss": 1.6607,
+      "step": 32700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7290000000000002e-05,
+      "loss": 1.6196,
+      "step": 32710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.728e-05,
+      "loss": 1.399,
+      "step": 32720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.727e-05,
+      "loss": 1.5823,
+      "step": 32730
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.726e-05,
+      "loss": 1.6349,
+      "step": 32740
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.725e-05,
+      "loss": 1.6323,
+      "step": 32750
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.724e-05,
+      "loss": 1.5706,
+      "step": 32760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7230000000000003e-05,
+      "loss": 1.4294,
+      "step": 32770
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.722e-05,
+      "loss": 1.6323,
+      "step": 32780
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.721e-05,
+      "loss": 1.6126,
+      "step": 32790
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7199999999999998e-05,
+      "loss": 1.5504,
+      "step": 32800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.719e-05,
+      "loss": 1.5418,
+      "step": 32810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.718e-05,
+      "loss": 1.5218,
+      "step": 32820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.717e-05,
+      "loss": 1.5771,
+      "step": 32830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7160000000000002e-05,
+      "loss": 1.7588,
+      "step": 32840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7150000000000004e-05,
+      "loss": 1.5131,
+      "step": 32850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7140000000000002e-05,
+      "loss": 1.342,
+      "step": 32860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.713e-05,
+      "loss": 1.5584,
+      "step": 32870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.712e-05,
+      "loss": 1.424,
+      "step": 32880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.711e-05,
+      "loss": 1.5464,
+      "step": 32890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7100000000000002e-05,
+      "loss": 1.5894,
+      "step": 32900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.709e-05,
+      "loss": 1.5866,
+      "step": 32910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7080000000000002e-05,
+      "loss": 1.5343,
+      "step": 32920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.707e-05,
+      "loss": 1.5855,
+      "step": 32930
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.706e-05,
+      "loss": 1.5946,
+      "step": 32940
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.705e-05,
+      "loss": 1.4115,
+      "step": 32950
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.704e-05,
+      "loss": 1.576,
+      "step": 32960
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.703e-05,
+      "loss": 1.5018,
+      "step": 32970
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.702e-05,
+      "loss": 1.5313,
+      "step": 32980
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.701e-05,
+      "loss": 1.5536,
+      "step": 32990
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 1.6653,
+      "step": 33000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.699e-05,
+      "loss": 1.4715,
+      "step": 33010
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.698e-05,
+      "loss": 1.6664,
+      "step": 33020
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.697e-05,
+      "loss": 1.5443,
+      "step": 33030
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.696e-05,
+      "loss": 1.6094,
+      "step": 33040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6950000000000002e-05,
+      "loss": 1.5602,
+      "step": 33050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.694e-05,
+      "loss": 1.5503,
+      "step": 33060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6930000000000002e-05,
+      "loss": 1.4777,
+      "step": 33070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.692e-05,
+      "loss": 1.5604,
+      "step": 33080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6910000000000002e-05,
+      "loss": 1.5951,
+      "step": 33090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.69e-05,
+      "loss": 1.5496,
+      "step": 33100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.689e-05,
+      "loss": 1.4574,
+      "step": 33110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.688e-05,
+      "loss": 1.6681,
+      "step": 33120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.687e-05,
+      "loss": 1.4438,
+      "step": 33130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.686e-05,
+      "loss": 1.6128,
+      "step": 33140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6850000000000003e-05,
+      "loss": 1.4319,
+      "step": 33150
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.684e-05,
+      "loss": 1.6014,
+      "step": 33160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.683e-05,
+      "loss": 1.5761,
+      "step": 33170
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6819999999999998e-05,
+      "loss": 1.7849,
+      "step": 33180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.681e-05,
+      "loss": 1.7283,
+      "step": 33190
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 1.4846,
+      "step": 33200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.679e-05,
+      "loss": 1.5232,
+      "step": 33210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 1.6088,
+      "step": 33220
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.677e-05,
+      "loss": 1.4777,
+      "step": 33230
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6760000000000002e-05,
+      "loss": 1.4768,
+      "step": 33240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.675e-05,
+      "loss": 1.4708,
+      "step": 33250
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.674e-05,
+      "loss": 1.5047,
+      "step": 33260
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.673e-05,
+      "loss": 1.3784,
+      "step": 33270
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.672e-05,
+      "loss": 1.6715,
+      "step": 33280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.671e-05,
+      "loss": 1.5498,
+      "step": 33290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6700000000000003e-05,
+      "loss": 1.6435,
+      "step": 33300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.669e-05,
+      "loss": 1.6373,
+      "step": 33310
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.668e-05,
+      "loss": 1.6681,
+      "step": 33320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6669999999999998e-05,
+      "loss": 1.7173,
+      "step": 33330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.666e-05,
+      "loss": 1.6158,
+      "step": 33340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.665e-05,
+      "loss": 1.5765,
+      "step": 33350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.664e-05,
+      "loss": 1.6132,
+      "step": 33360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6630000000000002e-05,
+      "loss": 1.5651,
+      "step": 33370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.662e-05,
+      "loss": 1.5856,
+      "step": 33380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6610000000000002e-05,
+      "loss": 1.5339,
+      "step": 33390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.66e-05,
+      "loss": 1.4839,
+      "step": 33400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.659e-05,
+      "loss": 1.4856,
+      "step": 33410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.658e-05,
+      "loss": 1.4617,
+      "step": 33420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.657e-05,
+      "loss": 1.4803,
+      "step": 33430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.656e-05,
+      "loss": 1.6247,
+      "step": 33440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6550000000000002e-05,
+      "loss": 1.5684,
+      "step": 33450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.654e-05,
+      "loss": 1.38,
+      "step": 33460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6530000000000003e-05,
+      "loss": 1.6013,
+      "step": 33470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.652e-05,
+      "loss": 1.6755,
+      "step": 33480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.651e-05,
+      "loss": 1.5009,
+      "step": 33490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.65e-05,
+      "loss": 1.4229,
+      "step": 33500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.649e-05,
+      "loss": 1.6325,
+      "step": 33510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.648e-05,
+      "loss": 1.4442,
+      "step": 33520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6470000000000003e-05,
+      "loss": 1.6142,
+      "step": 33530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.646e-05,
+      "loss": 1.6782,
+      "step": 33540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.645e-05,
+      "loss": 1.547,
+      "step": 33550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.644e-05,
+      "loss": 1.5355,
+      "step": 33560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.643e-05,
+      "loss": 1.5504,
+      "step": 33570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6420000000000002e-05,
+      "loss": 1.5749,
+      "step": 33580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.641e-05,
+      "loss": 1.6913,
+      "step": 33590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6400000000000002e-05,
+      "loss": 1.4766,
+      "step": 33600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.639e-05,
+      "loss": 1.6757,
+      "step": 33610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6380000000000002e-05,
+      "loss": 1.4728,
+      "step": 33620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.637e-05,
+      "loss": 1.5342,
+      "step": 33630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.636e-05,
+      "loss": 1.4933,
+      "step": 33640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.635e-05,
+      "loss": 1.5014,
+      "step": 33650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.634e-05,
+      "loss": 1.4951,
+      "step": 33660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.633e-05,
+      "loss": 1.5135,
+      "step": 33670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6320000000000003e-05,
+      "loss": 1.6,
+      "step": 33680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.631e-05,
+      "loss": 1.5986,
+      "step": 33690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.63e-05,
+      "loss": 1.5429,
+      "step": 33700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6289999999999998e-05,
+      "loss": 1.6195,
+      "step": 33710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.628e-05,
+      "loss": 1.5094,
+      "step": 33720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6270000000000002e-05,
+      "loss": 1.6093,
+      "step": 33730
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.626e-05,
+      "loss": 1.5038,
+      "step": 33740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6250000000000002e-05,
+      "loss": 1.4515,
+      "step": 33750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.624e-05,
+      "loss": 1.6037,
+      "step": 33760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6230000000000002e-05,
+      "loss": 1.4327,
+      "step": 33770
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.622e-05,
+      "loss": 1.65,
+      "step": 33780
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.621e-05,
+      "loss": 1.5791,
+      "step": 33790
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.62e-05,
+      "loss": 1.5776,
+      "step": 33800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.619e-05,
+      "loss": 1.5886,
+      "step": 33810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.618e-05,
+      "loss": 1.6591,
+      "step": 33820
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6170000000000003e-05,
+      "loss": 1.5032,
+      "step": 33830
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.616e-05,
+      "loss": 1.5456,
+      "step": 33840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6150000000000003e-05,
+      "loss": 1.5276,
+      "step": 33850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6139999999999998e-05,
+      "loss": 1.561,
+      "step": 33860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.613e-05,
+      "loss": 1.4646,
+      "step": 33870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.612e-05,
+      "loss": 1.5524,
+      "step": 33880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.611e-05,
+      "loss": 1.5296,
+      "step": 33890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6100000000000002e-05,
+      "loss": 1.5574,
+      "step": 33900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.609e-05,
+      "loss": 1.6273,
+      "step": 33910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6080000000000002e-05,
+      "loss": 1.5632,
+      "step": 33920
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.607e-05,
+      "loss": 1.6057,
+      "step": 33930
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.606e-05,
+      "loss": 1.5769,
+      "step": 33940
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.605e-05,
+      "loss": 1.666,
+      "step": 33950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.604e-05,
+      "loss": 1.4148,
+      "step": 33960
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.603e-05,
+      "loss": 1.4835,
+      "step": 33970
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6020000000000002e-05,
+      "loss": 1.624,
+      "step": 33980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.601e-05,
+      "loss": 1.6224,
+      "step": 33990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.5654,
+      "step": 34000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.599e-05,
+      "loss": 1.4991,
+      "step": 34010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.598e-05,
+      "loss": 1.5585,
+      "step": 34020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.597e-05,
+      "loss": 1.4943,
+      "step": 34030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.596e-05,
+      "loss": 1.6055,
+      "step": 34040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.595e-05,
+      "loss": 1.4917,
+      "step": 34050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.594e-05,
+      "loss": 1.3825,
+      "step": 34060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.593e-05,
+      "loss": 1.6018,
+      "step": 34070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.592e-05,
+      "loss": 1.5904,
+      "step": 34080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.591e-05,
+      "loss": 1.7278,
+      "step": 34090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.59e-05,
+      "loss": 1.646,
+      "step": 34100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5890000000000002e-05,
+      "loss": 1.5737,
+      "step": 34110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.588e-05,
+      "loss": 1.5249,
+      "step": 34120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5870000000000002e-05,
+      "loss": 1.6186,
+      "step": 34130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.586e-05,
+      "loss": 1.4194,
+      "step": 34140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5850000000000002e-05,
+      "loss": 1.4427,
+      "step": 34150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584e-05,
+      "loss": 1.5352,
+      "step": 34160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583e-05,
+      "loss": 1.2469,
+      "step": 34170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.582e-05,
+      "loss": 1.6457,
+      "step": 34180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581e-05,
+      "loss": 1.5585,
+      "step": 34190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.58e-05,
+      "loss": 1.5788,
+      "step": 34200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5790000000000003e-05,
+      "loss": 1.6226,
+      "step": 34210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578e-05,
+      "loss": 1.3848,
+      "step": 34220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.577e-05,
+      "loss": 1.5528,
+      "step": 34230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5759999999999998e-05,
+      "loss": 1.5069,
+      "step": 34240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.575e-05,
+      "loss": 1.3937,
+      "step": 34250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5740000000000002e-05,
+      "loss": 1.568,
+      "step": 34260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.573e-05,
+      "loss": 1.5333,
+      "step": 34270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5720000000000002e-05,
+      "loss": 1.4354,
+      "step": 34280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571e-05,
+      "loss": 1.7531,
+      "step": 34290
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5700000000000002e-05,
+      "loss": 1.5026,
+      "step": 34300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.569e-05,
+      "loss": 1.6265,
+      "step": 34310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.568e-05,
+      "loss": 1.453,
+      "step": 34320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.567e-05,
+      "loss": 1.5748,
+      "step": 34330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.566e-05,
+      "loss": 1.6002,
+      "step": 34340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.565e-05,
+      "loss": 1.539,
+      "step": 34350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5640000000000003e-05,
+      "loss": 1.4949,
+      "step": 34360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.563e-05,
+      "loss": 1.4835,
+      "step": 34370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5620000000000003e-05,
+      "loss": 1.6206,
+      "step": 34380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.561e-05,
+      "loss": 1.6137,
+      "step": 34390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.56e-05,
+      "loss": 1.4324,
+      "step": 34400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.559e-05,
+      "loss": 1.491,
+      "step": 34410
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.558e-05,
+      "loss": 1.435,
+      "step": 34420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5570000000000002e-05,
+      "loss": 1.4853,
+      "step": 34430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.556e-05,
+      "loss": 1.7116,
+      "step": 34440
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5550000000000002e-05,
+      "loss": 1.4253,
+      "step": 34450
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.554e-05,
+      "loss": 1.3853,
+      "step": 34460
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.553e-05,
+      "loss": 1.464,
+      "step": 34470
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.552e-05,
+      "loss": 1.4171,
+      "step": 34480
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.551e-05,
+      "loss": 1.5743,
+      "step": 34490
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.55e-05,
+      "loss": 1.5398,
+      "step": 34500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5490000000000002e-05,
+      "loss": 1.4146,
+      "step": 34510
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.548e-05,
+      "loss": 1.5169,
+      "step": 34520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5470000000000003e-05,
+      "loss": 1.4058,
+      "step": 34530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.546e-05,
+      "loss": 1.5804,
+      "step": 34540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.545e-05,
+      "loss": 1.4429,
+      "step": 34550
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.544e-05,
+      "loss": 1.6216,
+      "step": 34560
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.543e-05,
+      "loss": 1.4709,
+      "step": 34570
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.542e-05,
+      "loss": 1.5672,
+      "step": 34580
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.541e-05,
+      "loss": 1.5383,
+      "step": 34590
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.54e-05,
+      "loss": 1.447,
+      "step": 34600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.539e-05,
+      "loss": 1.5471,
+      "step": 34610
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.538e-05,
+      "loss": 1.4853,
+      "step": 34620
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.537e-05,
+      "loss": 1.4883,
+      "step": 34630
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.536e-05,
+      "loss": 1.5951,
+      "step": 34640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.535e-05,
+      "loss": 1.6201,
+      "step": 34650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5340000000000002e-05,
+      "loss": 1.3994,
+      "step": 34660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.533e-05,
+      "loss": 1.4461,
+      "step": 34670
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5320000000000002e-05,
+      "loss": 1.5525,
+      "step": 34680
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.531e-05,
+      "loss": 1.4115,
+      "step": 34690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.53e-05,
+      "loss": 1.4108,
+      "step": 34700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.529e-05,
+      "loss": 1.4192,
+      "step": 34710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.528e-05,
+      "loss": 1.5091,
+      "step": 34720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.527e-05,
+      "loss": 1.5601,
+      "step": 34730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5260000000000003e-05,
+      "loss": 1.5152,
+      "step": 34740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.525e-05,
+      "loss": 1.5596,
+      "step": 34750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5240000000000001e-05,
+      "loss": 1.5418,
+      "step": 34760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.523e-05,
+      "loss": 1.4742,
+      "step": 34770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5220000000000002e-05,
+      "loss": 1.5554,
+      "step": 34780
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5210000000000002e-05,
+      "loss": 1.3983,
+      "step": 34790
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.52e-05,
+      "loss": 1.5212,
+      "step": 34800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5190000000000002e-05,
+      "loss": 1.5942,
+      "step": 34810
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.518e-05,
+      "loss": 1.5306,
+      "step": 34820
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.517e-05,
+      "loss": 1.6812,
+      "step": 34830
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5160000000000002e-05,
+      "loss": 1.5368,
+      "step": 34840
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.515e-05,
+      "loss": 1.7089,
+      "step": 34850
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.514e-05,
+      "loss": 1.4018,
+      "step": 34860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5129999999999999e-05,
+      "loss": 1.3671,
+      "step": 34870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5120000000000001e-05,
+      "loss": 1.6032,
+      "step": 34880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5110000000000003e-05,
+      "loss": 1.582,
+      "step": 34890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.51e-05,
+      "loss": 1.3411,
+      "step": 34900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5090000000000001e-05,
+      "loss": 1.3608,
+      "step": 34910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.508e-05,
+      "loss": 1.5319,
+      "step": 34920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5070000000000001e-05,
+      "loss": 1.5031,
+      "step": 34930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5060000000000001e-05,
+      "loss": 1.5158,
+      "step": 34940
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.505e-05,
+      "loss": 1.4664,
+      "step": 34950
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5040000000000002e-05,
+      "loss": 1.5385,
+      "step": 34960
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.503e-05,
+      "loss": 1.4345,
+      "step": 34970
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.502e-05,
+      "loss": 1.4862,
+      "step": 34980
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5010000000000002e-05,
+      "loss": 1.5043,
+      "step": 34990
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5e-05,
+      "loss": 1.5508,
+      "step": 35000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.499e-05,
+      "loss": 1.4539,
+      "step": 35010
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4979999999999999e-05,
+      "loss": 1.5337,
+      "step": 35020
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.497e-05,
+      "loss": 1.6445,
+      "step": 35030
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4960000000000002e-05,
+      "loss": 1.5792,
+      "step": 35040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4950000000000001e-05,
+      "loss": 1.5738,
+      "step": 35050
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4940000000000001e-05,
+      "loss": 1.4393,
+      "step": 35060
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.493e-05,
+      "loss": 1.363,
+      "step": 35070
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4920000000000001e-05,
+      "loss": 1.5937,
+      "step": 35080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4910000000000001e-05,
+      "loss": 1.4869,
+      "step": 35090
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.49e-05,
+      "loss": 1.4562,
+      "step": 35100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4890000000000001e-05,
+      "loss": 1.5453,
+      "step": 35110
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.488e-05,
+      "loss": 1.6667,
+      "step": 35120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.487e-05,
+      "loss": 1.5206,
+      "step": 35130
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4860000000000002e-05,
+      "loss": 1.3808,
+      "step": 35140
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.485e-05,
+      "loss": 1.43,
+      "step": 35150
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4840000000000002e-05,
+      "loss": 1.6952,
+      "step": 35160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4829999999999999e-05,
+      "loss": 1.4203,
+      "step": 35170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.482e-05,
+      "loss": 1.4637,
+      "step": 35180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4810000000000002e-05,
+      "loss": 1.3679,
+      "step": 35190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.48e-05,
+      "loss": 1.5311,
+      "step": 35200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.479e-05,
+      "loss": 1.5097,
+      "step": 35210
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4779999999999999e-05,
+      "loss": 1.4232,
+      "step": 35220
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4770000000000001e-05,
+      "loss": 1.418,
+      "step": 35230
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4760000000000001e-05,
+      "loss": 1.4844,
+      "step": 35240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.475e-05,
+      "loss": 1.2521,
+      "step": 35250
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4740000000000001e-05,
+      "loss": 1.4203,
+      "step": 35260
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.473e-05,
+      "loss": 1.4818,
+      "step": 35270
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.472e-05,
+      "loss": 1.4072,
+      "step": 35280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4710000000000001e-05,
+      "loss": 1.3409,
+      "step": 35290
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.47e-05,
+      "loss": 1.3022,
+      "step": 35300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4690000000000002e-05,
+      "loss": 1.3601,
+      "step": 35310
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4680000000000002e-05,
+      "loss": 1.2842,
+      "step": 35320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.467e-05,
+      "loss": 1.4555,
+      "step": 35330
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4660000000000002e-05,
+      "loss": 1.3873,
+      "step": 35340
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.465e-05,
+      "loss": 1.498,
+      "step": 35350
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.464e-05,
+      "loss": 1.4959,
+      "step": 35360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4630000000000002e-05,
+      "loss": 1.393,
+      "step": 35370
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.462e-05,
+      "loss": 1.3951,
+      "step": 35380
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.461e-05,
+      "loss": 1.3346,
+      "step": 35390
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4599999999999999e-05,
+      "loss": 1.369,
+      "step": 35400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4590000000000001e-05,
+      "loss": 1.5327,
+      "step": 35410
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4580000000000003e-05,
+      "loss": 1.2657,
+      "step": 35420
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4570000000000001e-05,
+      "loss": 1.5327,
+      "step": 35430
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4560000000000001e-05,
+      "loss": 1.3794,
+      "step": 35440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.455e-05,
+      "loss": 1.6058,
+      "step": 35450
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4540000000000001e-05,
+      "loss": 1.3079,
+      "step": 35460
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4530000000000001e-05,
+      "loss": 1.4226,
+      "step": 35470
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.452e-05,
+      "loss": 1.384,
+      "step": 35480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4510000000000002e-05,
+      "loss": 1.4655,
+      "step": 35490
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.45e-05,
+      "loss": 1.3957,
+      "step": 35500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.449e-05,
+      "loss": 1.3034,
+      "step": 35510
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4480000000000002e-05,
+      "loss": 1.3896,
+      "step": 35520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.447e-05,
+      "loss": 1.4504,
+      "step": 35530
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4460000000000002e-05,
+      "loss": 1.3719,
+      "step": 35540
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4449999999999999e-05,
+      "loss": 1.4992,
+      "step": 35550
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.444e-05,
+      "loss": 1.5756,
+      "step": 35560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4430000000000002e-05,
+      "loss": 1.2369,
+      "step": 35570
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4420000000000001e-05,
+      "loss": 1.2538,
+      "step": 35580
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4410000000000001e-05,
+      "loss": 1.4048,
+      "step": 35590
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.44e-05,
+      "loss": 1.4374,
+      "step": 35600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4390000000000001e-05,
+      "loss": 1.3871,
+      "step": 35610
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 1.4352,
+      "step": 35620
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.437e-05,
+      "loss": 1.3957,
+      "step": 35630
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4360000000000001e-05,
+      "loss": 1.3317,
+      "step": 35640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.435e-05,
+      "loss": 1.3447,
+      "step": 35650
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.434e-05,
+      "loss": 1.3128,
+      "step": 35660
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4330000000000002e-05,
+      "loss": 1.4563,
+      "step": 35670
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.432e-05,
+      "loss": 1.2585,
+      "step": 35680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4310000000000002e-05,
+      "loss": 1.4495,
+      "step": 35690
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.43e-05,
+      "loss": 1.502,
+      "step": 35700
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.429e-05,
+      "loss": 1.3879,
+      "step": 35710
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4280000000000002e-05,
+      "loss": 1.242,
+      "step": 35720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.427e-05,
+      "loss": 1.2986,
+      "step": 35730
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.426e-05,
+      "loss": 1.4736,
+      "step": 35740
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4249999999999999e-05,
+      "loss": 1.4086,
+      "step": 35750
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4240000000000001e-05,
+      "loss": 1.2806,
+      "step": 35760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4230000000000001e-05,
+      "loss": 1.3898,
+      "step": 35770
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.422e-05,
+      "loss": 1.3205,
+      "step": 35780
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4210000000000001e-05,
+      "loss": 1.4046,
+      "step": 35790
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.42e-05,
+      "loss": 1.498,
+      "step": 35800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4190000000000001e-05,
+      "loss": 1.3817,
+      "step": 35810
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4180000000000001e-05,
+      "loss": 1.4121,
+      "step": 35820
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.417e-05,
+      "loss": 1.3639,
+      "step": 35830
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4160000000000002e-05,
+      "loss": 1.3307,
+      "step": 35840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.415e-05,
+      "loss": 1.3568,
+      "step": 35850
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.414e-05,
+      "loss": 1.4143,
+      "step": 35860
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4130000000000002e-05,
+      "loss": 1.2825,
+      "step": 35870
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.412e-05,
+      "loss": 1.4246,
+      "step": 35880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.411e-05,
+      "loss": 1.3914,
+      "step": 35890
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4099999999999999e-05,
+      "loss": 1.4348,
+      "step": 35900
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.409e-05,
+      "loss": 1.4118,
+      "step": 35910
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.408e-05,
+      "loss": 1.3044,
+      "step": 35920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4069999999999999e-05,
+      "loss": 1.2948,
+      "step": 35930
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4060000000000001e-05,
+      "loss": 1.3778,
+      "step": 35940
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4050000000000003e-05,
+      "loss": 1.3776,
+      "step": 35950
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4040000000000001e-05,
+      "loss": 1.3391,
+      "step": 35960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4030000000000001e-05,
+      "loss": 1.446,
+      "step": 35970
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.402e-05,
+      "loss": 1.3719,
+      "step": 35980
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4010000000000001e-05,
+      "loss": 1.4236,
+      "step": 35990
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 1.3628,
+      "step": 36000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.399e-05,
+      "loss": 1.4474,
+      "step": 36010
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3980000000000002e-05,
+      "loss": 1.355,
+      "step": 36020
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.397e-05,
+      "loss": 1.2502,
+      "step": 36030
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.396e-05,
+      "loss": 1.4182,
+      "step": 36040
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3950000000000002e-05,
+      "loss": 1.4184,
+      "step": 36050
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.394e-05,
+      "loss": 1.3637,
+      "step": 36060
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3930000000000002e-05,
+      "loss": 1.3679,
+      "step": 36070
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3919999999999999e-05,
+      "loss": 1.3508,
+      "step": 36080
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.391e-05,
+      "loss": 1.3874,
+      "step": 36090
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3900000000000002e-05,
+      "loss": 1.383,
+      "step": 36100
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.389e-05,
+      "loss": 1.3336,
+      "step": 36110
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3880000000000001e-05,
+      "loss": 1.3832,
+      "step": 36120
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.387e-05,
+      "loss": 1.4573,
+      "step": 36130
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3860000000000001e-05,
+      "loss": 1.4079,
+      "step": 36140
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3850000000000001e-05,
+      "loss": 1.323,
+      "step": 36150
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.384e-05,
+      "loss": 1.4184,
+      "step": 36160
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3830000000000001e-05,
+      "loss": 1.4465,
+      "step": 36170
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.382e-05,
+      "loss": 1.4724,
+      "step": 36180
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.381e-05,
+      "loss": 1.4868,
+      "step": 36190
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 1.3458,
+      "step": 36200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.379e-05,
+      "loss": 1.2263,
+      "step": 36210
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3780000000000002e-05,
+      "loss": 1.364,
+      "step": 36220
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.377e-05,
+      "loss": 1.4216,
+      "step": 36230
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.376e-05,
+      "loss": 1.4149,
+      "step": 36240
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3750000000000002e-05,
+      "loss": 1.279,
+      "step": 36250
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.374e-05,
+      "loss": 1.4204,
+      "step": 36260
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.373e-05,
+      "loss": 1.3856,
+      "step": 36270
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3719999999999999e-05,
+      "loss": 1.3505,
+      "step": 36280
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3710000000000001e-05,
+      "loss": 1.2993,
+      "step": 36290
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3700000000000001e-05,
+      "loss": 1.3599,
+      "step": 36300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.369e-05,
+      "loss": 1.3869,
+      "step": 36310
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3680000000000001e-05,
+      "loss": 1.4485,
+      "step": 36320
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.367e-05,
+      "loss": 1.353,
+      "step": 36330
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3660000000000001e-05,
+      "loss": 1.4759,
+      "step": 36340
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3650000000000001e-05,
+      "loss": 1.4572,
+      "step": 36350
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.364e-05,
+      "loss": 1.4902,
+      "step": 36360
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3630000000000002e-05,
+      "loss": 1.3714,
+      "step": 36370
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.362e-05,
+      "loss": 1.4098,
+      "step": 36380
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.361e-05,
+      "loss": 1.2903,
+      "step": 36390
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 1.3919,
+      "step": 36400
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.359e-05,
+      "loss": 1.4261,
+      "step": 36410
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.358e-05,
+      "loss": 1.5104,
+      "step": 36420
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3569999999999999e-05,
+      "loss": 1.3817,
+      "step": 36430
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.356e-05,
+      "loss": 1.3509,
+      "step": 36440
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3550000000000002e-05,
+      "loss": 1.2676,
+      "step": 36450
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3539999999999999e-05,
+      "loss": 1.3815,
+      "step": 36460
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3530000000000001e-05,
+      "loss": 1.3666,
+      "step": 36470
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.352e-05,
+      "loss": 1.2362,
+      "step": 36480
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3510000000000001e-05,
+      "loss": 1.3857,
+      "step": 36490
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 1.5031,
+      "step": 36500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.349e-05,
+      "loss": 1.337,
+      "step": 36510
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3480000000000001e-05,
+      "loss": 1.3405,
+      "step": 36520
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.347e-05,
+      "loss": 1.3528,
+      "step": 36530
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.346e-05,
+      "loss": 1.3999,
+      "step": 36540
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3450000000000002e-05,
+      "loss": 1.4297,
+      "step": 36550
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.344e-05,
+      "loss": 1.4576,
+      "step": 36560
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.343e-05,
+      "loss": 1.5117,
+      "step": 36570
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3420000000000002e-05,
+      "loss": 1.6488,
+      "step": 36580
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.341e-05,
+      "loss": 1.3059,
+      "step": 36590
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 1.3258,
+      "step": 36600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.339e-05,
+      "loss": 1.353,
+      "step": 36610
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.338e-05,
+      "loss": 1.4501,
+      "step": 36620
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3370000000000002e-05,
+      "loss": 1.3433,
+      "step": 36630
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.336e-05,
+      "loss": 1.3696,
+      "step": 36640
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3350000000000001e-05,
+      "loss": 1.4391,
+      "step": 36650
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.334e-05,
+      "loss": 1.2244,
+      "step": 36660
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3330000000000001e-05,
+      "loss": 1.4219,
+      "step": 36670
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3320000000000001e-05,
+      "loss": 1.4823,
+      "step": 36680
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.331e-05,
+      "loss": 1.4561,
+      "step": 36690
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3300000000000001e-05,
+      "loss": 1.4247,
+      "step": 36700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.329e-05,
+      "loss": 1.2447,
+      "step": 36710
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3280000000000002e-05,
+      "loss": 1.4891,
+      "step": 36720
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3270000000000002e-05,
+      "loss": 1.4782,
+      "step": 36730
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.326e-05,
+      "loss": 1.3429,
+      "step": 36740
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3250000000000002e-05,
+      "loss": 1.5429,
+      "step": 36750
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.324e-05,
+      "loss": 1.3594,
+      "step": 36760
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.323e-05,
+      "loss": 1.2555,
+      "step": 36770
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3220000000000002e-05,
+      "loss": 1.2696,
+      "step": 36780
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.321e-05,
+      "loss": 1.5175,
+      "step": 36790
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.32e-05,
+      "loss": 1.4523,
+      "step": 36800
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3189999999999999e-05,
+      "loss": 1.2392,
+      "step": 36810
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 1.2724,
+      "step": 36820
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3170000000000001e-05,
+      "loss": 1.3662,
+      "step": 36830
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.316e-05,
+      "loss": 1.2657,
+      "step": 36840
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3150000000000001e-05,
+      "loss": 1.3853,
+      "step": 36850
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.314e-05,
+      "loss": 1.3473,
+      "step": 36860
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3130000000000001e-05,
+      "loss": 1.31,
+      "step": 36870
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3120000000000001e-05,
+      "loss": 1.4584,
+      "step": 36880
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.311e-05,
+      "loss": 1.386,
+      "step": 36890
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3100000000000002e-05,
+      "loss": 1.2545,
+      "step": 36900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.309e-05,
+      "loss": 1.4005,
+      "step": 36910
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.308e-05,
+      "loss": 1.4215,
+      "step": 36920
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3070000000000002e-05,
+      "loss": 1.3511,
+      "step": 36930
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.306e-05,
+      "loss": 1.4043,
+      "step": 36940
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.305e-05,
+      "loss": 1.2313,
+      "step": 36950
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3039999999999999e-05,
+      "loss": 1.2897,
+      "step": 36960
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.303e-05,
+      "loss": 1.446,
+      "step": 36970
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3020000000000002e-05,
+      "loss": 1.3273,
+      "step": 36980
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.301e-05,
+      "loss": 1.3421,
+      "step": 36990
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 1.3716,
+      "step": 37000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.299e-05,
+      "loss": 1.2968,
+      "step": 37010
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 1.215,
+      "step": 37020
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2970000000000001e-05,
+      "loss": 1.2949,
+      "step": 37030
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.296e-05,
+      "loss": 1.4877,
+      "step": 37040
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2950000000000001e-05,
+      "loss": 1.3503,
+      "step": 37050
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.294e-05,
+      "loss": 1.4108,
+      "step": 37060
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.293e-05,
+      "loss": 1.4562,
+      "step": 37070
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2920000000000002e-05,
+      "loss": 1.4018,
+      "step": 37080
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.291e-05,
+      "loss": 1.392,
+      "step": 37090
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.29e-05,
+      "loss": 1.4771,
+      "step": 37100
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2889999999999999e-05,
+      "loss": 1.3341,
+      "step": 37110
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.288e-05,
+      "loss": 1.3807,
+      "step": 37120
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2870000000000002e-05,
+      "loss": 1.3116,
+      "step": 37130
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.286e-05,
+      "loss": 1.394,
+      "step": 37140
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.285e-05,
+      "loss": 1.4696,
+      "step": 37150
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2839999999999999e-05,
+      "loss": 1.3209,
+      "step": 37160
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.283e-05,
+      "loss": 1.2461,
+      "step": 37170
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2820000000000001e-05,
+      "loss": 1.4023,
+      "step": 37180
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.281e-05,
+      "loss": 1.312,
+      "step": 37190
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 1.3554,
+      "step": 37200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2790000000000001e-05,
+      "loss": 1.3303,
+      "step": 37210
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.278e-05,
+      "loss": 1.4068,
+      "step": 37220
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2770000000000001e-05,
+      "loss": 1.3265,
+      "step": 37230
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.276e-05,
+      "loss": 1.2341,
+      "step": 37240
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2750000000000002e-05,
+      "loss": 1.4715,
+      "step": 37250
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2740000000000002e-05,
+      "loss": 1.3706,
+      "step": 37260
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.273e-05,
+      "loss": 1.4313,
+      "step": 37270
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2720000000000002e-05,
+      "loss": 1.4254,
+      "step": 37280
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.271e-05,
+      "loss": 1.446,
+      "step": 37290
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.27e-05,
+      "loss": 1.4801,
+      "step": 37300
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2690000000000002e-05,
+      "loss": 1.2048,
+      "step": 37310
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.268e-05,
+      "loss": 1.5393,
+      "step": 37320
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.267e-05,
+      "loss": 1.2967,
+      "step": 37330
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2659999999999999e-05,
+      "loss": 1.2133,
+      "step": 37340
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2650000000000001e-05,
+      "loss": 1.3761,
+      "step": 37350
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2640000000000003e-05,
+      "loss": 1.408,
+      "step": 37360
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.263e-05,
+      "loss": 1.4762,
+      "step": 37370
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2620000000000001e-05,
+      "loss": 1.3033,
+      "step": 37380
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.261e-05,
+      "loss": 1.4278,
+      "step": 37390
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 1.4756,
+      "step": 37400
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2590000000000001e-05,
+      "loss": 1.3198,
+      "step": 37410
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.258e-05,
+      "loss": 1.4318,
+      "step": 37420
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2570000000000002e-05,
+      "loss": 1.2304,
+      "step": 37430
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.256e-05,
+      "loss": 1.3657,
+      "step": 37440
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.255e-05,
+      "loss": 1.4345,
+      "step": 37450
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2540000000000002e-05,
+      "loss": 1.36,
+      "step": 37460
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.253e-05,
+      "loss": 1.4262,
+      "step": 37470
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.252e-05,
+      "loss": 1.3757,
+      "step": 37480
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2509999999999999e-05,
+      "loss": 1.4093,
+      "step": 37490
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.25e-05,
+      "loss": 1.353,
+      "step": 37500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.249e-05,
+      "loss": 1.397,
+      "step": 37510
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.248e-05,
+      "loss": 1.2824,
+      "step": 37520
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2470000000000001e-05,
+      "loss": 1.3538,
+      "step": 37530
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2460000000000001e-05,
+      "loss": 1.4456,
+      "step": 37540
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2450000000000001e-05,
+      "loss": 1.444,
+      "step": 37550
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.244e-05,
+      "loss": 1.4812,
+      "step": 37560
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.243e-05,
+      "loss": 1.3241,
+      "step": 37570
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2420000000000001e-05,
+      "loss": 1.4402,
+      "step": 37580
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2410000000000001e-05,
+      "loss": 1.3404,
+      "step": 37590
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.24e-05,
+      "loss": 1.3793,
+      "step": 37600
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.239e-05,
+      "loss": 1.3223,
+      "step": 37610
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.238e-05,
+      "loss": 1.4116,
+      "step": 37620
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2370000000000002e-05,
+      "loss": 1.3277,
+      "step": 37630
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.236e-05,
+      "loss": 1.3178,
+      "step": 37640
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.235e-05,
+      "loss": 1.4073,
+      "step": 37650
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.234e-05,
+      "loss": 1.4598,
+      "step": 37660
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.233e-05,
+      "loss": 1.3383,
+      "step": 37670
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.232e-05,
+      "loss": 1.3967,
+      "step": 37680
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.231e-05,
+      "loss": 1.291,
+      "step": 37690
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.23e-05,
+      "loss": 1.2943,
+      "step": 37700
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2290000000000001e-05,
+      "loss": 1.3859,
+      "step": 37710
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2280000000000001e-05,
+      "loss": 1.23,
+      "step": 37720
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2270000000000001e-05,
+      "loss": 1.3502,
+      "step": 37730
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2260000000000001e-05,
+      "loss": 1.3822,
+      "step": 37740
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.225e-05,
+      "loss": 1.4754,
+      "step": 37750
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.224e-05,
+      "loss": 1.4368,
+      "step": 37760
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2230000000000001e-05,
+      "loss": 1.4885,
+      "step": 37770
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2220000000000002e-05,
+      "loss": 1.3053,
+      "step": 37780
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.221e-05,
+      "loss": 1.3452,
+      "step": 37790
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.22e-05,
+      "loss": 1.2644,
+      "step": 37800
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.219e-05,
+      "loss": 1.2428,
+      "step": 37810
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2180000000000002e-05,
+      "loss": 1.3195,
+      "step": 37820
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.217e-05,
+      "loss": 1.3353,
+      "step": 37830
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.216e-05,
+      "loss": 1.328,
+      "step": 37840
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.215e-05,
+      "loss": 1.4342,
+      "step": 37850
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.214e-05,
+      "loss": 1.3312,
+      "step": 37860
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.213e-05,
+      "loss": 1.4222,
+      "step": 37870
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2120000000000001e-05,
+      "loss": 1.3013,
+      "step": 37880
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2110000000000001e-05,
+      "loss": 1.2939,
+      "step": 37890
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2100000000000001e-05,
+      "loss": 1.2294,
+      "step": 37900
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.209e-05,
+      "loss": 1.2581,
+      "step": 37910
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2080000000000001e-05,
+      "loss": 1.4029,
+      "step": 37920
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2070000000000001e-05,
+      "loss": 1.3436,
+      "step": 37930
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.206e-05,
+      "loss": 1.3003,
+      "step": 37940
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.205e-05,
+      "loss": 1.4488,
+      "step": 37950
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.204e-05,
+      "loss": 1.2492,
+      "step": 37960
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2030000000000002e-05,
+      "loss": 1.2608,
+      "step": 37970
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.202e-05,
+      "loss": 1.2657,
+      "step": 37980
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.201e-05,
+      "loss": 1.308,
+      "step": 37990
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2e-05,
+      "loss": 1.3667,
+      "step": 38000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.199e-05,
+      "loss": 1.3203,
+      "step": 38010
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.198e-05,
+      "loss": 1.3152,
+      "step": 38020
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.197e-05,
+      "loss": 1.2843,
+      "step": 38030
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.196e-05,
+      "loss": 1.4222,
+      "step": 38040
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.195e-05,
+      "loss": 1.4076,
+      "step": 38050
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.1940000000000001e-05,
+      "loss": 1.2725,
+      "step": 38060
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.1930000000000001e-05,
+      "loss": 1.3472,
+      "step": 38070
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.1920000000000001e-05,
+      "loss": 1.5016,
+      "step": 38080
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.1910000000000001e-05,
+      "loss": 1.4567,
+      "step": 38090
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.19e-05,
+      "loss": 1.45,
+      "step": 38100
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.1890000000000001e-05,
+      "loss": 1.3328,
+      "step": 38110
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.1880000000000001e-05,
+      "loss": 1.2008,
+      "step": 38120
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.187e-05,
+      "loss": 1.3816,
+      "step": 38130
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.186e-05,
+      "loss": 1.3541,
+      "step": 38140
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.185e-05,
+      "loss": 1.2862,
+      "step": 38150
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1840000000000002e-05,
+      "loss": 1.2611,
+      "step": 38160
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.183e-05,
+      "loss": 1.4465,
+      "step": 38170
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.182e-05,
+      "loss": 1.2863,
+      "step": 38180
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.181e-05,
+      "loss": 1.4014,
+      "step": 38190
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.18e-05,
+      "loss": 1.3359,
+      "step": 38200
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.179e-05,
+      "loss": 1.4163,
+      "step": 38210
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.178e-05,
+      "loss": 1.3012,
+      "step": 38220
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.177e-05,
+      "loss": 1.3246,
+      "step": 38230
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1760000000000001e-05,
+      "loss": 1.2409,
+      "step": 38240
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.175e-05,
+      "loss": 1.3827,
+      "step": 38250
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1740000000000001e-05,
+      "loss": 1.24,
+      "step": 38260
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1730000000000001e-05,
+      "loss": 1.2846,
+      "step": 38270
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.172e-05,
+      "loss": 1.257,
+      "step": 38280
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.171e-05,
+      "loss": 1.2937,
+      "step": 38290
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1700000000000001e-05,
+      "loss": 1.5159,
+      "step": 38300
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1690000000000002e-05,
+      "loss": 1.3241,
+      "step": 38310
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.168e-05,
+      "loss": 1.1941,
+      "step": 38320
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.167e-05,
+      "loss": 1.3509,
+      "step": 38330
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.166e-05,
+      "loss": 1.3004,
+      "step": 38340
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1650000000000002e-05,
+      "loss": 1.3355,
+      "step": 38350
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.164e-05,
+      "loss": 1.3675,
+      "step": 38360
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.163e-05,
+      "loss": 1.2031,
+      "step": 38370
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.162e-05,
+      "loss": 1.3903,
+      "step": 38380
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.161e-05,
+      "loss": 1.3477,
+      "step": 38390
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.16e-05,
+      "loss": 1.3441,
+      "step": 38400
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.159e-05,
+      "loss": 1.4114,
+      "step": 38410
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 1.3636,
+      "step": 38420
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1570000000000001e-05,
+      "loss": 1.2267,
+      "step": 38430
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.156e-05,
+      "loss": 1.3999,
+      "step": 38440
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1550000000000001e-05,
+      "loss": 1.295,
+      "step": 38450
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1540000000000001e-05,
+      "loss": 1.4234,
+      "step": 38460
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.153e-05,
+      "loss": 1.4471,
+      "step": 38470
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.152e-05,
+      "loss": 1.338,
+      "step": 38480
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.151e-05,
+      "loss": 1.323,
+      "step": 38490
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1500000000000002e-05,
+      "loss": 1.5094,
+      "step": 38500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.149e-05,
+      "loss": 1.3467,
+      "step": 38510
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.148e-05,
+      "loss": 1.3065,
+      "step": 38520
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.147e-05,
+      "loss": 1.218,
+      "step": 38530
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.146e-05,
+      "loss": 1.2523,
+      "step": 38540
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.145e-05,
+      "loss": 1.4649,
+      "step": 38550
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.144e-05,
+      "loss": 1.4083,
+      "step": 38560
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.143e-05,
+      "loss": 1.3747,
+      "step": 38570
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.142e-05,
+      "loss": 1.2862,
+      "step": 38580
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.141e-05,
+      "loss": 1.3225,
+      "step": 38590
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 1.3258,
+      "step": 38600
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1390000000000001e-05,
+      "loss": 1.4692,
+      "step": 38610
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1380000000000001e-05,
+      "loss": 1.2478,
+      "step": 38620
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.137e-05,
+      "loss": 1.2994,
+      "step": 38630
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1360000000000001e-05,
+      "loss": 1.4722,
+      "step": 38640
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1350000000000001e-05,
+      "loss": 1.4871,
+      "step": 38650
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.134e-05,
+      "loss": 1.4225,
+      "step": 38660
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.133e-05,
+      "loss": 1.3213,
+      "step": 38670
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.132e-05,
+      "loss": 1.3607,
+      "step": 38680
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1310000000000002e-05,
+      "loss": 1.3798,
+      "step": 38690
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.13e-05,
+      "loss": 1.3316,
+      "step": 38700
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.129e-05,
+      "loss": 1.3665,
+      "step": 38710
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.128e-05,
+      "loss": 1.3858,
+      "step": 38720
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.127e-05,
+      "loss": 1.3606,
+      "step": 38730
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.126e-05,
+      "loss": 1.3015,
+      "step": 38740
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.125e-05,
+      "loss": 1.4105,
+      "step": 38750
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.124e-05,
+      "loss": 1.2787,
+      "step": 38760
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1230000000000001e-05,
+      "loss": 1.3526,
+      "step": 38770
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.122e-05,
+      "loss": 1.2814,
+      "step": 38780
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1210000000000001e-05,
+      "loss": 1.3639,
+      "step": 38790
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 1.3234,
+      "step": 38800
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1190000000000001e-05,
+      "loss": 1.4002,
+      "step": 38810
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.118e-05,
+      "loss": 1.4197,
+      "step": 38820
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.117e-05,
+      "loss": 1.3118,
+      "step": 38830
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1160000000000002e-05,
+      "loss": 1.2426,
+      "step": 38840
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.115e-05,
+      "loss": 1.4406,
+      "step": 38850
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.114e-05,
+      "loss": 1.2891,
+      "step": 38860
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.113e-05,
+      "loss": 1.1876,
+      "step": 38870
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.112e-05,
+      "loss": 1.3883,
+      "step": 38880
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.111e-05,
+      "loss": 1.3635,
+      "step": 38890
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.11e-05,
+      "loss": 1.3895,
+      "step": 38900
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.109e-05,
+      "loss": 1.3333,
+      "step": 38910
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.108e-05,
+      "loss": 1.3223,
+      "step": 38920
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.107e-05,
+      "loss": 1.1986,
+      "step": 38930
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.106e-05,
+      "loss": 1.2862,
+      "step": 38940
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1050000000000001e-05,
+      "loss": 1.3199,
+      "step": 38950
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1040000000000001e-05,
+      "loss": 1.2965,
+      "step": 38960
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.103e-05,
+      "loss": 1.388,
+      "step": 38970
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1020000000000001e-05,
+      "loss": 1.3333,
+      "step": 38980
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1010000000000001e-05,
+      "loss": 1.3257,
+      "step": 38990
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 1.359,
+      "step": 39000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.099e-05,
+      "loss": 1.2966,
+      "step": 39010
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.098e-05,
+      "loss": 1.238,
+      "step": 39020
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0970000000000002e-05,
+      "loss": 1.3741,
+      "step": 39030
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.096e-05,
+      "loss": 1.292,
+      "step": 39040
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.095e-05,
+      "loss": 1.2355,
+      "step": 39050
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.094e-05,
+      "loss": 1.2614,
+      "step": 39060
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.093e-05,
+      "loss": 1.4091,
+      "step": 39070
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.092e-05,
+      "loss": 1.357,
+      "step": 39080
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.091e-05,
+      "loss": 1.4129,
+      "step": 39090
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.09e-05,
+      "loss": 1.3807,
+      "step": 39100
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.089e-05,
+      "loss": 1.3223,
+      "step": 39110
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.088e-05,
+      "loss": 1.391,
+      "step": 39120
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0870000000000001e-05,
+      "loss": 1.305,
+      "step": 39130
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0860000000000001e-05,
+      "loss": 1.2176,
+      "step": 39140
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0850000000000001e-05,
+      "loss": 1.2086,
+      "step": 39150
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.084e-05,
+      "loss": 1.3136,
+      "step": 39160
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.083e-05,
+      "loss": 1.2813,
+      "step": 39170
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0820000000000001e-05,
+      "loss": 1.4621,
+      "step": 39180
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.081e-05,
+      "loss": 1.3954,
+      "step": 39190
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.08e-05,
+      "loss": 1.2242,
+      "step": 39200
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.079e-05,
+      "loss": 1.3113,
+      "step": 39210
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 1.2257,
+      "step": 39220
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.077e-05,
+      "loss": 1.3742,
+      "step": 39230
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.076e-05,
+      "loss": 1.2616,
+      "step": 39240
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.075e-05,
+      "loss": 1.2266,
+      "step": 39250
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.074e-05,
+      "loss": 1.2814,
+      "step": 39260
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.073e-05,
+      "loss": 1.3733,
+      "step": 39270
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.072e-05,
+      "loss": 1.3419,
+      "step": 39280
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.071e-05,
+      "loss": 1.3823,
+      "step": 39290
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.0700000000000001e-05,
+      "loss": 1.481,
+      "step": 39300
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.069e-05,
+      "loss": 1.3059,
+      "step": 39310
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.0680000000000001e-05,
+      "loss": 1.4255,
+      "step": 39320
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.0670000000000001e-05,
+      "loss": 1.3426,
+      "step": 39330
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.0660000000000001e-05,
+      "loss": 1.25,
+      "step": 39340
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.065e-05,
+      "loss": 1.3455,
+      "step": 39350
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.064e-05,
+      "loss": 1.3877,
+      "step": 39360
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.0630000000000002e-05,
+      "loss": 1.3466,
+      "step": 39370
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.062e-05,
+      "loss": 1.2849,
+      "step": 39380
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.061e-05,
+      "loss": 1.4092,
+      "step": 39390
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.06e-05,
+      "loss": 1.2894,
+      "step": 39400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.059e-05,
+      "loss": 1.2772,
+      "step": 39410
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.058e-05,
+      "loss": 1.4024,
+      "step": 39420
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.057e-05,
+      "loss": 1.2522,
+      "step": 39430
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.056e-05,
+      "loss": 1.4216,
+      "step": 39440
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.055e-05,
+      "loss": 1.2475,
+      "step": 39450
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.0539999999999999e-05,
+      "loss": 1.2786,
+      "step": 39460
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.053e-05,
+      "loss": 1.2487,
+      "step": 39470
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.0520000000000001e-05,
+      "loss": 1.3656,
+      "step": 39480
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.0510000000000001e-05,
+      "loss": 1.2963,
+      "step": 39490
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.05e-05,
+      "loss": 1.5123,
+      "step": 39500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.049e-05,
+      "loss": 1.3736,
+      "step": 39510
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.0480000000000001e-05,
+      "loss": 1.3812,
+      "step": 39520
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.0470000000000001e-05,
+      "loss": 1.4691,
+      "step": 39530
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.046e-05,
+      "loss": 1.4527,
+      "step": 39540
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.045e-05,
+      "loss": 1.2337,
+      "step": 39550
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.0440000000000002e-05,
+      "loss": 1.405,
+      "step": 39560
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.043e-05,
+      "loss": 1.3959,
+      "step": 39570
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.042e-05,
+      "loss": 1.4396,
+      "step": 39580
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.041e-05,
+      "loss": 1.4474,
+      "step": 39590
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.04e-05,
+      "loss": 1.3464,
+      "step": 39600
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.039e-05,
+      "loss": 1.4614,
+      "step": 39610
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.038e-05,
+      "loss": 1.2985,
+      "step": 39620
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.037e-05,
+      "loss": 1.3226,
+      "step": 39630
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.036e-05,
+      "loss": 1.2193,
+      "step": 39640
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.035e-05,
+      "loss": 1.334,
+      "step": 39650
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.0340000000000001e-05,
+      "loss": 1.2631,
+      "step": 39660
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.0330000000000001e-05,
+      "loss": 1.3029,
+      "step": 39670
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.0320000000000001e-05,
+      "loss": 1.3253,
+      "step": 39680
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.031e-05,
+      "loss": 1.3966,
+      "step": 39690
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.03e-05,
+      "loss": 1.3257,
+      "step": 39700
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.0290000000000001e-05,
+      "loss": 1.3082,
+      "step": 39710
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.0280000000000002e-05,
+      "loss": 1.4655,
+      "step": 39720
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.027e-05,
+      "loss": 1.1884,
+      "step": 39730
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.026e-05,
+      "loss": 1.2131,
+      "step": 39740
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.025e-05,
+      "loss": 1.3655,
+      "step": 39750
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.024e-05,
+      "loss": 1.335,
+      "step": 39760
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.023e-05,
+      "loss": 1.3611,
+      "step": 39770
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.022e-05,
+      "loss": 1.3897,
+      "step": 39780
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.021e-05,
+      "loss": 1.2153,
+      "step": 39790
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.02e-05,
+      "loss": 1.3384,
+      "step": 39800
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.019e-05,
+      "loss": 1.3191,
+      "step": 39810
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.018e-05,
+      "loss": 1.1967,
+      "step": 39820
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.0170000000000001e-05,
+      "loss": 1.4688,
+      "step": 39830
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.016e-05,
+      "loss": 1.3611,
+      "step": 39840
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.0150000000000001e-05,
+      "loss": 1.3648,
+      "step": 39850
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.0140000000000001e-05,
+      "loss": 1.3712,
+      "step": 39860
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.0130000000000001e-05,
+      "loss": 1.2303,
+      "step": 39870
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.012e-05,
+      "loss": 1.2751,
+      "step": 39880
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.011e-05,
+      "loss": 1.4242,
+      "step": 39890
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.0100000000000002e-05,
+      "loss": 1.3618,
+      "step": 39900
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.0090000000000002e-05,
+      "loss": 1.2793,
+      "step": 39910
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.008e-05,
+      "loss": 1.2977,
+      "step": 39920
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.007e-05,
+      "loss": 1.3266,
+      "step": 39930
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.006e-05,
+      "loss": 1.2405,
+      "step": 39940
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.005e-05,
+      "loss": 1.2591,
+      "step": 39950
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.004e-05,
+      "loss": 1.4465,
+      "step": 39960
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.003e-05,
+      "loss": 1.1479,
+      "step": 39970
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.002e-05,
+      "loss": 1.2786,
+      "step": 39980
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.001e-05,
+      "loss": 1.3093,
+      "step": 39990
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1e-05,
+      "loss": 1.22,
+      "step": 40000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.990000000000001e-06,
+      "loss": 1.2732,
+      "step": 40010
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 1.3582,
+      "step": 40020
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.97e-06,
+      "loss": 1.3128,
+      "step": 40030
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.96e-06,
+      "loss": 1.2566,
+      "step": 40040
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.950000000000001e-06,
+      "loss": 1.3225,
+      "step": 40050
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.940000000000001e-06,
+      "loss": 1.3654,
+      "step": 40060
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.93e-06,
+      "loss": 1.3105,
+      "step": 40070
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.92e-06,
+      "loss": 1.407,
+      "step": 40080
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.91e-06,
+      "loss": 1.4113,
+      "step": 40090
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.900000000000002e-06,
+      "loss": 1.4563,
+      "step": 40100
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.89e-06,
+      "loss": 1.3457,
+      "step": 40110
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.88e-06,
+      "loss": 1.5044,
+      "step": 40120
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.87e-06,
+      "loss": 1.3321,
+      "step": 40130
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.86e-06,
+      "loss": 1.4397,
+      "step": 40140
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.85e-06,
+      "loss": 1.3765,
+      "step": 40150
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.84e-06,
+      "loss": 1.2651,
+      "step": 40160
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.83e-06,
+      "loss": 1.3176,
+      "step": 40170
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.820000000000001e-06,
+      "loss": 1.4558,
+      "step": 40180
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.810000000000001e-06,
+      "loss": 1.269,
+      "step": 40190
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 1.2556,
+      "step": 40200
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.790000000000001e-06,
+      "loss": 1.2471,
+      "step": 40210
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.78e-06,
+      "loss": 1.3681,
+      "step": 40220
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.77e-06,
+      "loss": 1.4189,
+      "step": 40230
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 1.3062,
+      "step": 40240
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.750000000000002e-06,
+      "loss": 1.2628,
+      "step": 40250
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.74e-06,
+      "loss": 1.2659,
+      "step": 40260
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.73e-06,
+      "loss": 1.3026,
+      "step": 40270
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.72e-06,
+      "loss": 1.4091,
+      "step": 40280
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.71e-06,
+      "loss": 1.3569,
+      "step": 40290
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.7e-06,
+      "loss": 1.2862,
+      "step": 40300
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.69e-06,
+      "loss": 1.3514,
+      "step": 40310
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.68e-06,
+      "loss": 1.2899,
+      "step": 40320
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.67e-06,
+      "loss": 1.3161,
+      "step": 40330
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.66e-06,
+      "loss": 1.2684,
+      "step": 40340
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.65e-06,
+      "loss": 1.3948,
+      "step": 40350
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.640000000000001e-06,
+      "loss": 1.1765,
+      "step": 40360
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.630000000000001e-06,
+      "loss": 1.2961,
+      "step": 40370
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.62e-06,
+      "loss": 1.278,
+      "step": 40380
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.610000000000001e-06,
+      "loss": 1.312,
+      "step": 40390
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 1.3472,
+      "step": 40400
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.59e-06,
+      "loss": 1.3545,
+      "step": 40410
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.58e-06,
+      "loss": 1.4103,
+      "step": 40420
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.57e-06,
+      "loss": 1.2508,
+      "step": 40430
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.560000000000002e-06,
+      "loss": 1.3397,
+      "step": 40440
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.55e-06,
+      "loss": 1.2906,
+      "step": 40450
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.54e-06,
+      "loss": 1.3536,
+      "step": 40460
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.53e-06,
+      "loss": 1.3033,
+      "step": 40470
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.52e-06,
+      "loss": 1.4199,
+      "step": 40480
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.51e-06,
+      "loss": 1.3462,
+      "step": 40490
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.5e-06,
+      "loss": 1.2398,
+      "step": 40500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.49e-06,
+      "loss": 1.3633,
+      "step": 40510
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.48e-06,
+      "loss": 1.3117,
+      "step": 40520
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.47e-06,
+      "loss": 1.4947,
+      "step": 40530
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.460000000000001e-06,
+      "loss": 1.1845,
+      "step": 40540
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.450000000000001e-06,
+      "loss": 1.2453,
+      "step": 40550
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.44e-06,
+      "loss": 1.4108,
+      "step": 40560
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.43e-06,
+      "loss": 1.3919,
+      "step": 40570
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.420000000000001e-06,
+      "loss": 1.348,
+      "step": 40580
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.410000000000001e-06,
+      "loss": 1.3472,
+      "step": 40590
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.4e-06,
+      "loss": 1.205,
+      "step": 40600
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.39e-06,
+      "loss": 1.3321,
+      "step": 40610
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.38e-06,
+      "loss": 1.2628,
+      "step": 40620
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.370000000000002e-06,
+      "loss": 1.2522,
+      "step": 40630
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.36e-06,
+      "loss": 1.2814,
+      "step": 40640
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.35e-06,
+      "loss": 1.2579,
+      "step": 40650
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.34e-06,
+      "loss": 1.3425,
+      "step": 40660
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.33e-06,
+      "loss": 1.2438,
+      "step": 40670
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.32e-06,
+      "loss": 1.3084,
+      "step": 40680
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.31e-06,
+      "loss": 1.4449,
+      "step": 40690
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.3e-06,
+      "loss": 1.3268,
+      "step": 40700
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.29e-06,
+      "loss": 1.3039,
+      "step": 40710
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.28e-06,
+      "loss": 1.2879,
+      "step": 40720
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.270000000000001e-06,
+      "loss": 1.2918,
+      "step": 40730
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.260000000000001e-06,
+      "loss": 1.2754,
+      "step": 40740
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.25e-06,
+      "loss": 1.2296,
+      "step": 40750
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.24e-06,
+      "loss": 1.4216,
+      "step": 40760
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.23e-06,
+      "loss": 1.1641,
+      "step": 40770
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.220000000000002e-06,
+      "loss": 1.4424,
+      "step": 40780
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.21e-06,
+      "loss": 1.31,
+      "step": 40790
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.2e-06,
+      "loss": 1.2882,
+      "step": 40800
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.19e-06,
+      "loss": 1.3633,
+      "step": 40810
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 1.3123,
+      "step": 40820
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.17e-06,
+      "loss": 1.3247,
+      "step": 40830
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.16e-06,
+      "loss": 1.271,
+      "step": 40840
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.15e-06,
+      "loss": 1.3142,
+      "step": 40850
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.14e-06,
+      "loss": 1.2793,
+      "step": 40860
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.13e-06,
+      "loss": 1.2286,
+      "step": 40870
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.12e-06,
+      "loss": 1.3923,
+      "step": 40880
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.110000000000001e-06,
+      "loss": 1.4105,
+      "step": 40890
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 1.2724,
+      "step": 40900
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.09e-06,
+      "loss": 1.286,
+      "step": 40910
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.080000000000001e-06,
+      "loss": 1.2681,
+      "step": 40920
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.070000000000001e-06,
+      "loss": 1.3611,
+      "step": 40930
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.06e-06,
+      "loss": 1.336,
+      "step": 40940
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.05e-06,
+      "loss": 1.286,
+      "step": 40950
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.04e-06,
+      "loss": 1.2387,
+      "step": 40960
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.030000000000002e-06,
+      "loss": 1.332,
+      "step": 40970
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.02e-06,
+      "loss": 1.2897,
+      "step": 40980
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.01e-06,
+      "loss": 1.1832,
+      "step": 40990
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9e-06,
+      "loss": 1.3335,
+      "step": 41000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.99e-06,
+      "loss": 1.3157,
+      "step": 41010
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.98e-06,
+      "loss": 1.3276,
+      "step": 41020
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.97e-06,
+      "loss": 1.4031,
+      "step": 41030
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.96e-06,
+      "loss": 1.3743,
+      "step": 41040
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.95e-06,
+      "loss": 1.3984,
+      "step": 41050
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.939999999999999e-06,
+      "loss": 1.3916,
+      "step": 41060
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.930000000000001e-06,
+      "loss": 1.3207,
+      "step": 41070
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.920000000000001e-06,
+      "loss": 1.443,
+      "step": 41080
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.910000000000001e-06,
+      "loss": 1.4559,
+      "step": 41090
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.9e-06,
+      "loss": 1.2645,
+      "step": 41100
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.890000000000001e-06,
+      "loss": 1.3783,
+      "step": 41110
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.880000000000001e-06,
+      "loss": 1.2464,
+      "step": 41120
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.87e-06,
+      "loss": 1.3725,
+      "step": 41130
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.86e-06,
+      "loss": 1.2492,
+      "step": 41140
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.85e-06,
+      "loss": 1.2853,
+      "step": 41150
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.840000000000002e-06,
+      "loss": 1.2381,
+      "step": 41160
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.83e-06,
+      "loss": 1.2788,
+      "step": 41170
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.82e-06,
+      "loss": 1.338,
+      "step": 41180
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.81e-06,
+      "loss": 1.2646,
+      "step": 41190
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.8e-06,
+      "loss": 1.3866,
+      "step": 41200
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.79e-06,
+      "loss": 1.2984,
+      "step": 41210
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.78e-06,
+      "loss": 1.3515,
+      "step": 41220
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.77e-06,
+      "loss": 1.3124,
+      "step": 41230
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.76e-06,
+      "loss": 1.4449,
+      "step": 41240
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.75e-06,
+      "loss": 1.4423,
+      "step": 41250
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.740000000000001e-06,
+      "loss": 1.2997,
+      "step": 41260
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.730000000000001e-06,
+      "loss": 1.4163,
+      "step": 41270
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.720000000000001e-06,
+      "loss": 1.1633,
+      "step": 41280
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.71e-06,
+      "loss": 1.3973,
+      "step": 41290
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.7e-06,
+      "loss": 1.37,
+      "step": 41300
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.690000000000002e-06,
+      "loss": 1.3014,
+      "step": 41310
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.68e-06,
+      "loss": 1.4964,
+      "step": 41320
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.67e-06,
+      "loss": 1.1413,
+      "step": 41330
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.66e-06,
+      "loss": 1.2944,
+      "step": 41340
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.65e-06,
+      "loss": 1.3713,
+      "step": 41350
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.64e-06,
+      "loss": 1.2067,
+      "step": 41360
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.63e-06,
+      "loss": 1.3532,
+      "step": 41370
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.62e-06,
+      "loss": 1.267,
+      "step": 41380
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.61e-06,
+      "loss": 1.3744,
+      "step": 41390
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.599999999999999e-06,
+      "loss": 1.2529,
+      "step": 41400
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.59e-06,
+      "loss": 1.4045,
+      "step": 41410
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 1.2355,
+      "step": 41420
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.570000000000001e-06,
+      "loss": 1.2456,
+      "step": 41430
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.56e-06,
+      "loss": 1.3104,
+      "step": 41440
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.550000000000001e-06,
+      "loss": 1.3291,
+      "step": 41450
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.540000000000001e-06,
+      "loss": 1.2944,
+      "step": 41460
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.53e-06,
+      "loss": 1.2788,
+      "step": 41470
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.52e-06,
+      "loss": 1.4546,
+      "step": 41480
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.51e-06,
+      "loss": 1.1498,
+      "step": 41490
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.500000000000002e-06,
+      "loss": 1.2491,
+      "step": 41500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.49e-06,
+      "loss": 1.3705,
+      "step": 41510
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.48e-06,
+      "loss": 1.378,
+      "step": 41520
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.47e-06,
+      "loss": 1.2467,
+      "step": 41530
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.46e-06,
+      "loss": 1.3562,
+      "step": 41540
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.45e-06,
+      "loss": 1.3445,
+      "step": 41550
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.44e-06,
+      "loss": 1.2485,
+      "step": 41560
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.43e-06,
+      "loss": 1.3913,
+      "step": 41570
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.42e-06,
+      "loss": 1.1722,
+      "step": 41580
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.409999999999999e-06,
+      "loss": 1.2928,
+      "step": 41590
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 1.2624,
+      "step": 41600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.390000000000001e-06,
+      "loss": 1.286,
+      "step": 41610
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 1.3641,
+      "step": 41620
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.37e-06,
+      "loss": 1.2181,
+      "step": 41630
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.36e-06,
+      "loss": 1.3921,
+      "step": 41640
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.350000000000001e-06,
+      "loss": 1.2583,
+      "step": 41650
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.34e-06,
+      "loss": 1.3563,
+      "step": 41660
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.33e-06,
+      "loss": 1.2684,
+      "step": 41670
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.32e-06,
+      "loss": 1.2777,
+      "step": 41680
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.31e-06,
+      "loss": 1.2504,
+      "step": 41690
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.3e-06,
+      "loss": 1.1974,
+      "step": 41700
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.29e-06,
+      "loss": 1.3091,
+      "step": 41710
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.28e-06,
+      "loss": 1.2756,
+      "step": 41720
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.27e-06,
+      "loss": 1.3452,
+      "step": 41730
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.26e-06,
+      "loss": 1.2912,
+      "step": 41740
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.25e-06,
+      "loss": 1.2092,
+      "step": 41750
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.24e-06,
+      "loss": 1.2742,
+      "step": 41760
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.23e-06,
+      "loss": 1.4791,
+      "step": 41770
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.22e-06,
+      "loss": 1.2518,
+      "step": 41780
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.210000000000001e-06,
+      "loss": 1.2888,
+      "step": 41790
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.200000000000001e-06,
+      "loss": 1.4257,
+      "step": 41800
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.190000000000001e-06,
+      "loss": 1.2414,
+      "step": 41810
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.18e-06,
+      "loss": 1.2148,
+      "step": 41820
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.17e-06,
+      "loss": 1.3737,
+      "step": 41830
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.160000000000001e-06,
+      "loss": 1.292,
+      "step": 41840
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.15e-06,
+      "loss": 1.43,
+      "step": 41850
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.14e-06,
+      "loss": 1.339,
+      "step": 41860
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.13e-06,
+      "loss": 1.2444,
+      "step": 41870
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.12e-06,
+      "loss": 1.2952,
+      "step": 41880
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.11e-06,
+      "loss": 1.2866,
+      "step": 41890
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.1e-06,
+      "loss": 1.3008,
+      "step": 41900
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.09e-06,
+      "loss": 1.3721,
+      "step": 41910
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.08e-06,
+      "loss": 1.3857,
+      "step": 41920
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.069999999999999e-06,
+      "loss": 1.1088,
+      "step": 41930
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.06e-06,
+      "loss": 1.2905,
+      "step": 41940
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.050000000000001e-06,
+      "loss": 1.1947,
+      "step": 41950
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 1.3368,
+      "step": 41960
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.03e-06,
+      "loss": 1.2909,
+      "step": 41970
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.02e-06,
+      "loss": 1.2874,
+      "step": 41980
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.010000000000001e-06,
+      "loss": 1.4011,
+      "step": 41990
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.3852,
+      "step": 42000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.99e-06,
+      "loss": 1.5664,
+      "step": 42010
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.98e-06,
+      "loss": 1.3919,
+      "step": 42020
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.97e-06,
+      "loss": 1.2651,
+      "step": 42030
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.96e-06,
+      "loss": 1.2058,
+      "step": 42040
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.95e-06,
+      "loss": 1.3173,
+      "step": 42050
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.94e-06,
+      "loss": 1.1665,
+      "step": 42060
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.93e-06,
+      "loss": 1.3173,
+      "step": 42070
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.92e-06,
+      "loss": 1.3311,
+      "step": 42080
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.91e-06,
+      "loss": 1.3068,
+      "step": 42090
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.9e-06,
+      "loss": 1.2308,
+      "step": 42100
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.89e-06,
+      "loss": 1.2509,
+      "step": 42110
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.879999999999999e-06,
+      "loss": 1.3574,
+      "step": 42120
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.870000000000001e-06,
+      "loss": 1.3122,
+      "step": 42130
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 1.26,
+      "step": 42140
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.850000000000001e-06,
+      "loss": 1.3586,
+      "step": 42150
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.84e-06,
+      "loss": 1.3053,
+      "step": 42160
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.83e-06,
+      "loss": 1.3276,
+      "step": 42170
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.820000000000001e-06,
+      "loss": 1.2551,
+      "step": 42180
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.810000000000001e-06,
+      "loss": 1.251,
+      "step": 42190
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.8e-06,
+      "loss": 1.3546,
+      "step": 42200
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.79e-06,
+      "loss": 1.1939,
+      "step": 42210
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.78e-06,
+      "loss": 1.3409,
+      "step": 42220
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.77e-06,
+      "loss": 1.4121,
+      "step": 42230
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.76e-06,
+      "loss": 1.2573,
+      "step": 42240
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.75e-06,
+      "loss": 1.3738,
+      "step": 42250
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.74e-06,
+      "loss": 1.3257,
+      "step": 42260
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.73e-06,
+      "loss": 1.3133,
+      "step": 42270
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.72e-06,
+      "loss": 1.2511,
+      "step": 42280
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.71e-06,
+      "loss": 1.198,
+      "step": 42290
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.7e-06,
+      "loss": 1.3476,
+      "step": 42300
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.69e-06,
+      "loss": 1.342,
+      "step": 42310
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.68e-06,
+      "loss": 1.2334,
+      "step": 42320
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.670000000000001e-06,
+      "loss": 1.1956,
+      "step": 42330
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.660000000000001e-06,
+      "loss": 1.3127,
+      "step": 42340
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.65e-06,
+      "loss": 1.2715,
+      "step": 42350
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.64e-06,
+      "loss": 1.3362,
+      "step": 42360
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.630000000000001e-06,
+      "loss": 1.3841,
+      "step": 42370
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.620000000000001e-06,
+      "loss": 1.336,
+      "step": 42380
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.610000000000001e-06,
+      "loss": 1.1743,
+      "step": 42390
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.6e-06,
+      "loss": 1.205,
+      "step": 42400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.59e-06,
+      "loss": 1.3501,
+      "step": 42410
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.580000000000001e-06,
+      "loss": 1.28,
+      "step": 42420
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.57e-06,
+      "loss": 1.3001,
+      "step": 42430
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 1.1826,
+      "step": 42440
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.55e-06,
+      "loss": 1.2382,
+      "step": 42450
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.54e-06,
+      "loss": 1.1513,
+      "step": 42460
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.530000000000001e-06,
+      "loss": 1.4111,
+      "step": 42470
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 1.1969,
+      "step": 42480
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.51e-06,
+      "loss": 1.3472,
+      "step": 42490
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.5e-06,
+      "loss": 1.226,
+      "step": 42500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.4899999999999994e-06,
+      "loss": 1.259,
+      "step": 42510
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.480000000000001e-06,
+      "loss": 1.3286,
+      "step": 42520
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.4700000000000005e-06,
+      "loss": 1.4035,
+      "step": 42530
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.4600000000000006e-06,
+      "loss": 1.3228,
+      "step": 42540
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.45e-06,
+      "loss": 1.3083,
+      "step": 42550
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.44e-06,
+      "loss": 1.2374,
+      "step": 42560
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.430000000000001e-06,
+      "loss": 1.2319,
+      "step": 42570
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.420000000000001e-06,
+      "loss": 1.2132,
+      "step": 42580
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.41e-06,
+      "loss": 1.2928,
+      "step": 42590
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.4e-06,
+      "loss": 1.2919,
+      "step": 42600
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.3899999999999995e-06,
+      "loss": 1.2591,
+      "step": 42610
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 1.3073,
+      "step": 42620
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.370000000000001e-06,
+      "loss": 1.2382,
+      "step": 42630
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.36e-06,
+      "loss": 1.3987,
+      "step": 42640
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.35e-06,
+      "loss": 1.4695,
+      "step": 42650
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.340000000000001e-06,
+      "loss": 1.3741,
+      "step": 42660
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.330000000000001e-06,
+      "loss": 1.2731,
+      "step": 42670
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.32e-06,
+      "loss": 1.3629,
+      "step": 42680
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.31e-06,
+      "loss": 1.3708,
+      "step": 42690
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.2999999999999996e-06,
+      "loss": 1.2818,
+      "step": 42700
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.290000000000001e-06,
+      "loss": 1.2068,
+      "step": 42710
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.280000000000001e-06,
+      "loss": 1.2188,
+      "step": 42720
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.270000000000001e-06,
+      "loss": 1.3633,
+      "step": 42730
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.26e-06,
+      "loss": 1.3143,
+      "step": 42740
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.25e-06,
+      "loss": 1.1786,
+      "step": 42750
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.240000000000001e-06,
+      "loss": 1.3086,
+      "step": 42760
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.230000000000001e-06,
+      "loss": 1.3826,
+      "step": 42770
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.22e-06,
+      "loss": 1.2297,
+      "step": 42780
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.2100000000000004e-06,
+      "loss": 1.4001,
+      "step": 42790
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.2e-06,
+      "loss": 1.2914,
+      "step": 42800
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.190000000000001e-06,
+      "loss": 1.4607,
+      "step": 42810
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 1.3925,
+      "step": 42820
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.17e-06,
+      "loss": 1.3668,
+      "step": 42830
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.16e-06,
+      "loss": 1.2983,
+      "step": 42840
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.15e-06,
+      "loss": 1.1582,
+      "step": 42850
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.140000000000001e-06,
+      "loss": 1.1996,
+      "step": 42860
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.13e-06,
+      "loss": 1.1986,
+      "step": 42870
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.1200000000000004e-06,
+      "loss": 1.3909,
+      "step": 42880
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.11e-06,
+      "loss": 1.5094,
+      "step": 42890
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.1e-06,
+      "loss": 1.298,
+      "step": 42900
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.090000000000001e-06,
+      "loss": 1.1636,
+      "step": 42910
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.080000000000001e-06,
+      "loss": 1.2413,
+      "step": 42920
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.07e-06,
+      "loss": 1.3203,
+      "step": 42930
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.06e-06,
+      "loss": 1.1568,
+      "step": 42940
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.049999999999999e-06,
+      "loss": 1.2527,
+      "step": 42950
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.04e-06,
+      "loss": 1.3553,
+      "step": 42960
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.0300000000000005e-06,
+      "loss": 1.3023,
+      "step": 42970
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 1.3034,
+      "step": 42980
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.01e-06,
+      "loss": 1.2169,
+      "step": 42990
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 1.1894,
+      "step": 43000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.990000000000001e-06,
+      "loss": 1.2881,
+      "step": 43010
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.98e-06,
+      "loss": 1.279,
+      "step": 43020
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.97e-06,
+      "loss": 1.2646,
+      "step": 43030
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9599999999999994e-06,
+      "loss": 1.2385,
+      "step": 43040
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.950000000000001e-06,
+      "loss": 1.3341,
+      "step": 43050
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9400000000000005e-06,
+      "loss": 1.3062,
+      "step": 43060
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.9300000000000006e-06,
+      "loss": 1.196,
+      "step": 43070
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.92e-06,
+      "loss": 1.3633,
+      "step": 43080
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.91e-06,
+      "loss": 1.2536,
+      "step": 43090
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.900000000000001e-06,
+      "loss": 1.2432,
+      "step": 43100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.890000000000001e-06,
+      "loss": 1.1644,
+      "step": 43110
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.88e-06,
+      "loss": 1.2294,
+      "step": 43120
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.87e-06,
+      "loss": 1.2571,
+      "step": 43130
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8599999999999995e-06,
+      "loss": 1.3285,
+      "step": 43140
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8500000000000005e-06,
+      "loss": 1.3266,
+      "step": 43150
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.840000000000001e-06,
+      "loss": 1.2676,
+      "step": 43160
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.830000000000001e-06,
+      "loss": 1.2513,
+      "step": 43170
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.82e-06,
+      "loss": 1.3143,
+      "step": 43180
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.81e-06,
+      "loss": 1.4012,
+      "step": 43190
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 1.3152,
+      "step": 43200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.79e-06,
+      "loss": 1.3458,
+      "step": 43210
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.78e-06,
+      "loss": 1.2049,
+      "step": 43220
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.7699999999999996e-06,
+      "loss": 1.2596,
+      "step": 43230
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.76e-06,
+      "loss": 1.2926,
+      "step": 43240
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.750000000000001e-06,
+      "loss": 1.2604,
+      "step": 43250
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.740000000000001e-06,
+      "loss": 1.2228,
+      "step": 43260
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.73e-06,
+      "loss": 1.24,
+      "step": 43270
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.72e-06,
+      "loss": 1.3388,
+      "step": 43280
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.710000000000001e-06,
+      "loss": 1.365,
+      "step": 43290
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 1.1411,
+      "step": 43300
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.69e-06,
+      "loss": 1.1983,
+      "step": 43310
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.68e-06,
+      "loss": 1.3011,
+      "step": 43320
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.67e-06,
+      "loss": 1.3404,
+      "step": 43330
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 1.1713,
+      "step": 43340
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.650000000000001e-06,
+      "loss": 1.3775,
+      "step": 43350
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.640000000000001e-06,
+      "loss": 1.3299,
+      "step": 43360
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.63e-06,
+      "loss": 1.345,
+      "step": 43370
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.62e-06,
+      "loss": 1.2792,
+      "step": 43380
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.610000000000001e-06,
+      "loss": 1.3265,
+      "step": 43390
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.6e-06,
+      "loss": 1.338,
+      "step": 43400
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.5900000000000004e-06,
+      "loss": 1.2858,
+      "step": 43410
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.58e-06,
+      "loss": 1.1493,
+      "step": 43420
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.57e-06,
+      "loss": 1.1657,
+      "step": 43430
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 1.2674,
+      "step": 43440
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.550000000000001e-06,
+      "loss": 1.2431,
+      "step": 43450
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.54e-06,
+      "loss": 1.4673,
+      "step": 43460
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.53e-06,
+      "loss": 1.1438,
+      "step": 43470
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.519999999999999e-06,
+      "loss": 1.3224,
+      "step": 43480
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.510000000000001e-06,
+      "loss": 1.3733,
+      "step": 43490
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 1.2527,
+      "step": 43500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.4900000000000005e-06,
+      "loss": 1.1841,
+      "step": 43510
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.48e-06,
+      "loss": 1.3026,
+      "step": 43520
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.47e-06,
+      "loss": 1.3302,
+      "step": 43530
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.460000000000001e-06,
+      "loss": 1.2586,
+      "step": 43540
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.45e-06,
+      "loss": 1.2909,
+      "step": 43550
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.44e-06,
+      "loss": 1.1415,
+      "step": 43560
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.43e-06,
+      "loss": 1.2407,
+      "step": 43570
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.4199999999999995e-06,
+      "loss": 1.27,
+      "step": 43580
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.4100000000000005e-06,
+      "loss": 1.3152,
+      "step": 43590
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 1.2101,
+      "step": 43600
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.39e-06,
+      "loss": 1.3065,
+      "step": 43610
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.38e-06,
+      "loss": 1.3529,
+      "step": 43620
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.370000000000001e-06,
+      "loss": 1.2567,
+      "step": 43630
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.360000000000001e-06,
+      "loss": 1.2236,
+      "step": 43640
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.35e-06,
+      "loss": 1.2776,
+      "step": 43650
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.34e-06,
+      "loss": 1.3248,
+      "step": 43660
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.3299999999999995e-06,
+      "loss": 1.1494,
+      "step": 43670
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.320000000000001e-06,
+      "loss": 1.2815,
+      "step": 43680
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.3100000000000006e-06,
+      "loss": 1.1915,
+      "step": 43690
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 1.2651,
+      "step": 43700
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.29e-06,
+      "loss": 1.3039,
+      "step": 43710
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.28e-06,
+      "loss": 1.2148,
+      "step": 43720
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.270000000000001e-06,
+      "loss": 1.2404,
+      "step": 43730
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.26e-06,
+      "loss": 1.2506,
+      "step": 43740
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.25e-06,
+      "loss": 1.251,
+      "step": 43750
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.24e-06,
+      "loss": 1.2743,
+      "step": 43760
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.2300000000000005e-06,
+      "loss": 1.3676,
+      "step": 43770
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.22e-06,
+      "loss": 1.3241,
+      "step": 43780
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.210000000000001e-06,
+      "loss": 1.269,
+      "step": 43790
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.2e-06,
+      "loss": 1.2237,
+      "step": 43800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.19e-06,
+      "loss": 1.3183,
+      "step": 43810
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.18e-06,
+      "loss": 1.1611,
+      "step": 43820
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.17e-06,
+      "loss": 1.3027,
+      "step": 43830
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.16e-06,
+      "loss": 1.2482,
+      "step": 43840
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.15e-06,
+      "loss": 1.2882,
+      "step": 43850
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.1400000000000005e-06,
+      "loss": 1.3727,
+      "step": 43860
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.130000000000001e-06,
+      "loss": 1.3317,
+      "step": 43870
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.12e-06,
+      "loss": 1.2708,
+      "step": 43880
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.110000000000001e-06,
+      "loss": 1.3278,
+      "step": 43890
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.1e-06,
+      "loss": 1.2421,
+      "step": 43900
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.090000000000001e-06,
+      "loss": 1.2746,
+      "step": 43910
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.08e-06,
+      "loss": 1.3352,
+      "step": 43920
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.07e-06,
+      "loss": 1.2934,
+      "step": 43930
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 1.2514,
+      "step": 43940
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.0500000000000005e-06,
+      "loss": 1.3058,
+      "step": 43950
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.040000000000001e-06,
+      "loss": 1.2591,
+      "step": 43960
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.03e-06,
+      "loss": 1.2307,
+      "step": 43970
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.02e-06,
+      "loss": 1.3413,
+      "step": 43980
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.01e-06,
+      "loss": 1.2836,
+      "step": 43990
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6e-06,
+      "loss": 1.3022,
+      "step": 44000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.99e-06,
+      "loss": 1.257,
+      "step": 44010
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.98e-06,
+      "loss": 1.2853,
+      "step": 44020
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.9700000000000004e-06,
+      "loss": 1.2551,
+      "step": 44030
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.9600000000000005e-06,
+      "loss": 1.2821,
+      "step": 44040
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.95e-06,
+      "loss": 1.2429,
+      "step": 44050
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.940000000000001e-06,
+      "loss": 1.2501,
+      "step": 44060
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.93e-06,
+      "loss": 1.398,
+      "step": 44070
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.920000000000001e-06,
+      "loss": 1.2358,
+      "step": 44080
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.91e-06,
+      "loss": 1.323,
+      "step": 44090
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.9e-06,
+      "loss": 1.1571,
+      "step": 44100
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.89e-06,
+      "loss": 1.2472,
+      "step": 44110
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 1.138,
+      "step": 44120
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.8700000000000005e-06,
+      "loss": 1.1883,
+      "step": 44130
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.86e-06,
+      "loss": 1.2692,
+      "step": 44140
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.850000000000001e-06,
+      "loss": 1.291,
+      "step": 44150
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.84e-06,
+      "loss": 1.222,
+      "step": 44160
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.83e-06,
+      "loss": 1.074,
+      "step": 44170
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.82e-06,
+      "loss": 1.2711,
+      "step": 44180
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.81e-06,
+      "loss": 1.3709,
+      "step": 44190
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.8e-06,
+      "loss": 1.2457,
+      "step": 44200
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.7900000000000005e-06,
+      "loss": 1.2147,
+      "step": 44210
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.78e-06,
+      "loss": 1.2829,
+      "step": 44220
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.770000000000001e-06,
+      "loss": 1.1952,
+      "step": 44230
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.76e-06,
+      "loss": 1.1646,
+      "step": 44240
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.750000000000001e-06,
+      "loss": 1.2839,
+      "step": 44250
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.74e-06,
+      "loss": 1.3046,
+      "step": 44260
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.73e-06,
+      "loss": 1.2978,
+      "step": 44270
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.72e-06,
+      "loss": 1.2057,
+      "step": 44280
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.71e-06,
+      "loss": 1.3419,
+      "step": 44290
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.7000000000000005e-06,
+      "loss": 1.211,
+      "step": 44300
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.690000000000001e-06,
+      "loss": 1.2394,
+      "step": 44310
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.680000000000001e-06,
+      "loss": 1.2638,
+      "step": 44320
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.67e-06,
+      "loss": 1.3295,
+      "step": 44330
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.66e-06,
+      "loss": 1.2362,
+      "step": 44340
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.65e-06,
+      "loss": 1.2827,
+      "step": 44350
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.64e-06,
+      "loss": 1.3807,
+      "step": 44360
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.63e-06,
+      "loss": 1.3245,
+      "step": 44370
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.62e-06,
+      "loss": 1.1902,
+      "step": 44380
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.61e-06,
+      "loss": 1.2908,
+      "step": 44390
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 1.4219,
+      "step": 44400
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.59e-06,
+      "loss": 1.2473,
+      "step": 44410
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 1.2334,
+      "step": 44420
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.57e-06,
+      "loss": 1.2362,
+      "step": 44430
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.56e-06,
+      "loss": 1.3199,
+      "step": 44440
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.55e-06,
+      "loss": 1.2572,
+      "step": 44450
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.54e-06,
+      "loss": 1.2353,
+      "step": 44460
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.53e-06,
+      "loss": 1.132,
+      "step": 44470
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.5200000000000005e-06,
+      "loss": 1.1656,
+      "step": 44480
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.510000000000001e-06,
+      "loss": 1.304,
+      "step": 44490
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 1.1656,
+      "step": 44500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.49e-06,
+      "loss": 1.2218,
+      "step": 44510
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.48e-06,
+      "loss": 1.3633,
+      "step": 44520
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.47e-06,
+      "loss": 1.3644,
+      "step": 44530
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.46e-06,
+      "loss": 1.2479,
+      "step": 44540
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.45e-06,
+      "loss": 1.2368,
+      "step": 44550
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.44e-06,
+      "loss": 1.2527,
+      "step": 44560
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.4300000000000005e-06,
+      "loss": 1.3678,
+      "step": 44570
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.42e-06,
+      "loss": 1.1672,
+      "step": 44580
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.410000000000001e-06,
+      "loss": 1.2568,
+      "step": 44590
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.4e-06,
+      "loss": 1.2889,
+      "step": 44600
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.390000000000001e-06,
+      "loss": 1.2438,
+      "step": 44610
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.38e-06,
+      "loss": 1.2663,
+      "step": 44620
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.37e-06,
+      "loss": 1.4178,
+      "step": 44630
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.36e-06,
+      "loss": 1.3832,
+      "step": 44640
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.3500000000000004e-06,
+      "loss": 1.3323,
+      "step": 44650
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.3400000000000005e-06,
+      "loss": 1.2418,
+      "step": 44660
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.330000000000001e-06,
+      "loss": 1.3192,
+      "step": 44670
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.32e-06,
+      "loss": 1.2265,
+      "step": 44680
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.31e-06,
+      "loss": 1.4028,
+      "step": 44690
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.3e-06,
+      "loss": 1.145,
+      "step": 44700
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.29e-06,
+      "loss": 1.2729,
+      "step": 44710
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.28e-06,
+      "loss": 1.319,
+      "step": 44720
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.2699999999999995e-06,
+      "loss": 1.3849,
+      "step": 44730
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.2600000000000005e-06,
+      "loss": 1.1419,
+      "step": 44740
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.25e-06,
+      "loss": 1.2097,
+      "step": 44750
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.240000000000001e-06,
+      "loss": 1.2692,
+      "step": 44760
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.23e-06,
+      "loss": 1.3678,
+      "step": 44770
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.220000000000001e-06,
+      "loss": 1.1884,
+      "step": 44780
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.21e-06,
+      "loss": 1.2722,
+      "step": 44790
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.2e-06,
+      "loss": 1.1888,
+      "step": 44800
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.19e-06,
+      "loss": 1.3669,
+      "step": 44810
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.18e-06,
+      "loss": 1.3735,
+      "step": 44820
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.1700000000000005e-06,
+      "loss": 1.1961,
+      "step": 44830
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.1600000000000006e-06,
+      "loss": 1.2724,
+      "step": 44840
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.15e-06,
+      "loss": 1.172,
+      "step": 44850
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.140000000000001e-06,
+      "loss": 1.4575,
+      "step": 44860
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.13e-06,
+      "loss": 1.448,
+      "step": 44870
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.12e-06,
+      "loss": 1.2183,
+      "step": 44880
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.11e-06,
+      "loss": 1.4545,
+      "step": 44890
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.1e-06,
+      "loss": 1.1989,
+      "step": 44900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.09e-06,
+      "loss": 1.3454,
+      "step": 44910
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.08e-06,
+      "loss": 1.2396,
+      "step": 44920
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.070000000000001e-06,
+      "loss": 1.305,
+      "step": 44930
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.06e-06,
+      "loss": 1.4568,
+      "step": 44940
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.050000000000001e-06,
+      "loss": 1.254,
+      "step": 44950
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.04e-06,
+      "loss": 1.3864,
+      "step": 44960
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.03e-06,
+      "loss": 1.3427,
+      "step": 44970
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.02e-06,
+      "loss": 1.3463,
+      "step": 44980
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.01e-06,
+      "loss": 1.3483,
+      "step": 44990
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5e-06,
+      "loss": 1.3092,
+      "step": 45000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.9900000000000005e-06,
+      "loss": 1.2444,
+      "step": 45010
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.98e-06,
+      "loss": 1.062,
+      "step": 45020
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.970000000000001e-06,
+      "loss": 1.2879,
+      "step": 45030
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.96e-06,
+      "loss": 1.3013,
+      "step": 45040
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.950000000000001e-06,
+      "loss": 1.1945,
+      "step": 45050
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.94e-06,
+      "loss": 1.0963,
+      "step": 45060
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.93e-06,
+      "loss": 1.2513,
+      "step": 45070
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.92e-06,
+      "loss": 1.1753,
+      "step": 45080
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.9100000000000004e-06,
+      "loss": 1.2199,
+      "step": 45090
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 1.2156,
+      "step": 45100
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.89e-06,
+      "loss": 1.1678,
+      "step": 45110
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 1.3041,
+      "step": 45120
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.87e-06,
+      "loss": 1.3274,
+      "step": 45130
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.86e-06,
+      "loss": 1.2491,
+      "step": 45140
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.85e-06,
+      "loss": 1.4127,
+      "step": 45150
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.84e-06,
+      "loss": 1.3052,
+      "step": 45160
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.83e-06,
+      "loss": 1.153,
+      "step": 45170
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.8200000000000004e-06,
+      "loss": 1.3809,
+      "step": 45180
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.81e-06,
+      "loss": 1.2864,
+      "step": 45190
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.2444,
+      "step": 45200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.79e-06,
+      "loss": 1.3771,
+      "step": 45210
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.780000000000001e-06,
+      "loss": 1.3487,
+      "step": 45220
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.77e-06,
+      "loss": 1.3676,
+      "step": 45230
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.76e-06,
+      "loss": 1.2946,
+      "step": 45240
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.75e-06,
+      "loss": 1.2728,
+      "step": 45250
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.74e-06,
+      "loss": 1.2025,
+      "step": 45260
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7300000000000005e-06,
+      "loss": 1.3123,
+      "step": 45270
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.72e-06,
+      "loss": 1.1974,
+      "step": 45280
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.710000000000001e-06,
+      "loss": 1.335,
+      "step": 45290
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.7e-06,
+      "loss": 1.2504,
+      "step": 45300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.69e-06,
+      "loss": 1.34,
+      "step": 45310
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.68e-06,
+      "loss": 1.2481,
+      "step": 45320
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.67e-06,
+      "loss": 1.3821,
+      "step": 45330
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.66e-06,
+      "loss": 1.2713,
+      "step": 45340
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.65e-06,
+      "loss": 1.1722,
+      "step": 45350
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.64e-06,
+      "loss": 1.2658,
+      "step": 45360
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.6300000000000006e-06,
+      "loss": 1.2674,
+      "step": 45370
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.62e-06,
+      "loss": 1.3148,
+      "step": 45380
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.610000000000001e-06,
+      "loss": 1.1645,
+      "step": 45390
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.6e-06,
+      "loss": 1.2859,
+      "step": 45400
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.590000000000001e-06,
+      "loss": 1.321,
+      "step": 45410
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.58e-06,
+      "loss": 1.3354,
+      "step": 45420
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.57e-06,
+      "loss": 1.2083,
+      "step": 45430
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.56e-06,
+      "loss": 1.3196,
+      "step": 45440
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.5500000000000005e-06,
+      "loss": 1.2785,
+      "step": 45450
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.540000000000001e-06,
+      "loss": 1.2376,
+      "step": 45460
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.53e-06,
+      "loss": 1.4352,
+      "step": 45470
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.52e-06,
+      "loss": 1.1992,
+      "step": 45480
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.51e-06,
+      "loss": 1.3473,
+      "step": 45490
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.5e-06,
+      "loss": 1.1109,
+      "step": 45500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.49e-06,
+      "loss": 1.209,
+      "step": 45510
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.48e-06,
+      "loss": 1.3362,
+      "step": 45520
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.4699999999999996e-06,
+      "loss": 1.3299,
+      "step": 45530
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.4600000000000005e-06,
+      "loss": 1.3828,
+      "step": 45540
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.45e-06,
+      "loss": 1.3032,
+      "step": 45550
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.440000000000001e-06,
+      "loss": 1.2685,
+      "step": 45560
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.43e-06,
+      "loss": 1.2414,
+      "step": 45570
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.420000000000001e-06,
+      "loss": 1.2131,
+      "step": 45580
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.41e-06,
+      "loss": 1.1776,
+      "step": 45590
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.4e-06,
+      "loss": 1.3444,
+      "step": 45600
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.39e-06,
+      "loss": 1.165,
+      "step": 45610
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.38e-06,
+      "loss": 1.1455,
+      "step": 45620
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3700000000000005e-06,
+      "loss": 1.0943,
+      "step": 45630
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.360000000000001e-06,
+      "loss": 1.205,
+      "step": 45640
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.35e-06,
+      "loss": 1.1425,
+      "step": 45650
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.34e-06,
+      "loss": 1.3058,
+      "step": 45660
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.33e-06,
+      "loss": 1.2079,
+      "step": 45670
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.32e-06,
+      "loss": 1.2874,
+      "step": 45680
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.31e-06,
+      "loss": 1.1352,
+      "step": 45690
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.2999999999999995e-06,
+      "loss": 1.2044,
+      "step": 45700
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.2900000000000004e-06,
+      "loss": 1.2471,
+      "step": 45710
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.28e-06,
+      "loss": 1.1842,
+      "step": 45720
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.270000000000001e-06,
+      "loss": 1.2876,
+      "step": 45730
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.26e-06,
+      "loss": 1.2764,
+      "step": 45740
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.250000000000001e-06,
+      "loss": 1.3487,
+      "step": 45750
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.24e-06,
+      "loss": 1.1958,
+      "step": 45760
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.23e-06,
+      "loss": 1.2823,
+      "step": 45770
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.22e-06,
+      "loss": 1.2562,
+      "step": 45780
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.21e-06,
+      "loss": 1.3411,
+      "step": 45790
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 1.3165,
+      "step": 45800
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.1900000000000005e-06,
+      "loss": 1.3583,
+      "step": 45810
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.18e-06,
+      "loss": 1.1855,
+      "step": 45820
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.17e-06,
+      "loss": 1.2212,
+      "step": 45830
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.16e-06,
+      "loss": 1.2906,
+      "step": 45840
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.15e-06,
+      "loss": 1.2223,
+      "step": 45850
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.14e-06,
+      "loss": 1.3016,
+      "step": 45860
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.13e-06,
+      "loss": 1.2698,
+      "step": 45870
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.12e-06,
+      "loss": 1.3288,
+      "step": 45880
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.11e-06,
+      "loss": 1.2217,
+      "step": 45890
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.1000000000000006e-06,
+      "loss": 1.268,
+      "step": 45900
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.09e-06,
+      "loss": 1.3645,
+      "step": 45910
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.080000000000001e-06,
+      "loss": 1.3289,
+      "step": 45920
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.07e-06,
+      "loss": 1.2371,
+      "step": 45930
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.06e-06,
+      "loss": 1.316,
+      "step": 45940
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.05e-06,
+      "loss": 1.2095,
+      "step": 45950
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.04e-06,
+      "loss": 1.329,
+      "step": 45960
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.03e-06,
+      "loss": 1.2408,
+      "step": 45970
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 1.2545,
+      "step": 45980
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.01e-06,
+      "loss": 1.2817,
+      "step": 45990
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.0819,
+      "step": 46000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.99e-06,
+      "loss": 1.2535,
+      "step": 46010
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.98e-06,
+      "loss": 1.1581,
+      "step": 46020
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.97e-06,
+      "loss": 1.3384,
+      "step": 46030
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.96e-06,
+      "loss": 1.3365,
+      "step": 46040
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.95e-06,
+      "loss": 1.2391,
+      "step": 46050
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.9399999999999995e-06,
+      "loss": 1.2459,
+      "step": 46060
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.9300000000000005e-06,
+      "loss": 1.2626,
+      "step": 46070
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.92e-06,
+      "loss": 1.1914,
+      "step": 46080
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.910000000000001e-06,
+      "loss": 1.3306,
+      "step": 46090
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.9e-06,
+      "loss": 1.387,
+      "step": 46100
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.89e-06,
+      "loss": 1.3644,
+      "step": 46110
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.88e-06,
+      "loss": 1.3131,
+      "step": 46120
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.87e-06,
+      "loss": 1.2043,
+      "step": 46130
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.86e-06,
+      "loss": 1.2163,
+      "step": 46140
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.85e-06,
+      "loss": 1.3426,
+      "step": 46150
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.84e-06,
+      "loss": 1.1194,
+      "step": 46160
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.830000000000001e-06,
+      "loss": 1.2989,
+      "step": 46170
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.82e-06,
+      "loss": 1.1763,
+      "step": 46180
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.8100000000000004e-06,
+      "loss": 1.234,
+      "step": 46190
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.8e-06,
+      "loss": 1.3254,
+      "step": 46200
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7900000000000006e-06,
+      "loss": 1.1991,
+      "step": 46210
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 1.449,
+      "step": 46220
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.77e-06,
+      "loss": 1.3267,
+      "step": 46230
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 1.2028,
+      "step": 46240
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.75e-06,
+      "loss": 1.3977,
+      "step": 46250
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7400000000000006e-06,
+      "loss": 1.236,
+      "step": 46260
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7300000000000003e-06,
+      "loss": 1.1958,
+      "step": 46270
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.72e-06,
+      "loss": 1.2296,
+      "step": 46280
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7100000000000005e-06,
+      "loss": 1.3118,
+      "step": 46290
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7e-06,
+      "loss": 1.3586,
+      "step": 46300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6900000000000002e-06,
+      "loss": 1.2043,
+      "step": 46310
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.68e-06,
+      "loss": 1.1487,
+      "step": 46320
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6700000000000004e-06,
+      "loss": 1.2722,
+      "step": 46330
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.66e-06,
+      "loss": 1.2101,
+      "step": 46340
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6499999999999998e-06,
+      "loss": 1.2046,
+      "step": 46350
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6400000000000003e-06,
+      "loss": 1.1743,
+      "step": 46360
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.63e-06,
+      "loss": 1.1432,
+      "step": 46370
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6200000000000005e-06,
+      "loss": 1.3718,
+      "step": 46380
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.61e-06,
+      "loss": 1.2621,
+      "step": 46390
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6e-06,
+      "loss": 1.2279,
+      "step": 46400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.5900000000000004e-06,
+      "loss": 1.2855,
+      "step": 46410
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.58e-06,
+      "loss": 1.2024,
+      "step": 46420
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.5700000000000005e-06,
+      "loss": 1.1617,
+      "step": 46430
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.5600000000000002e-06,
+      "loss": 1.2213,
+      "step": 46440
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.55e-06,
+      "loss": 1.29,
+      "step": 46450
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.5400000000000004e-06,
+      "loss": 1.2453,
+      "step": 46460
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.53e-06,
+      "loss": 1.2931,
+      "step": 46470
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.52e-06,
+      "loss": 1.4254,
+      "step": 46480
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.5100000000000003e-06,
+      "loss": 1.3507,
+      "step": 46490
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.5000000000000004e-06,
+      "loss": 1.3312,
+      "step": 46500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.49e-06,
+      "loss": 1.2059,
+      "step": 46510
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.4799999999999997e-06,
+      "loss": 1.5145,
+      "step": 46520
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.4700000000000002e-06,
+      "loss": 1.2504,
+      "step": 46530
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.46e-06,
+      "loss": 1.274,
+      "step": 46540
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.4500000000000004e-06,
+      "loss": 1.1626,
+      "step": 46550
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.44e-06,
+      "loss": 1.2728,
+      "step": 46560
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.4299999999999998e-06,
+      "loss": 1.3131,
+      "step": 46570
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.4200000000000003e-06,
+      "loss": 1.2379,
+      "step": 46580
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.41e-06,
+      "loss": 1.1546,
+      "step": 46590
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 1.3334,
+      "step": 46600
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.39e-06,
+      "loss": 1.1895,
+      "step": 46610
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.38e-06,
+      "loss": 1.2382,
+      "step": 46620
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.3700000000000003e-06,
+      "loss": 1.2427,
+      "step": 46630
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.36e-06,
+      "loss": 1.2171,
+      "step": 46640
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.3500000000000005e-06,
+      "loss": 1.2543,
+      "step": 46650
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.34e-06,
+      "loss": 1.4486,
+      "step": 46660
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.3300000000000003e-06,
+      "loss": 1.3126,
+      "step": 46670
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 1.2587,
+      "step": 46680
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.31e-06,
+      "loss": 1.3859,
+      "step": 46690
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.3e-06,
+      "loss": 1.307,
+      "step": 46700
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.29e-06,
+      "loss": 1.188,
+      "step": 46710
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 1.3121,
+      "step": 46720
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.27e-06,
+      "loss": 1.2738,
+      "step": 46730
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.2599999999999997e-06,
+      "loss": 1.2506,
+      "step": 46740
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.2500000000000002e-06,
+      "loss": 1.1333,
+      "step": 46750
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.24e-06,
+      "loss": 1.2236,
+      "step": 46760
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.2300000000000004e-06,
+      "loss": 1.4532,
+      "step": 46770
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.22e-06,
+      "loss": 1.2699,
+      "step": 46780
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.2099999999999998e-06,
+      "loss": 1.199,
+      "step": 46790
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 1.2247,
+      "step": 46800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.19e-06,
+      "loss": 1.1958,
+      "step": 46810
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 1.2497,
+      "step": 46820
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.17e-06,
+      "loss": 1.2674,
+      "step": 46830
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.1600000000000007e-06,
+      "loss": 1.2301,
+      "step": 46840
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.1500000000000003e-06,
+      "loss": 1.2675,
+      "step": 46850
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.14e-06,
+      "loss": 1.3908,
+      "step": 46860
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.13e-06,
+      "loss": 1.2608,
+      "step": 46870
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.12e-06,
+      "loss": 1.2438,
+      "step": 46880
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.11e-06,
+      "loss": 1.1902,
+      "step": 46890
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.1e-06,
+      "loss": 1.2846,
+      "step": 46900
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.09e-06,
+      "loss": 1.2815,
+      "step": 46910
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.08e-06,
+      "loss": 1.207,
+      "step": 46920
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.0700000000000003e-06,
+      "loss": 1.1859,
+      "step": 46930
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.06e-06,
+      "loss": 1.2278,
+      "step": 46940
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.05e-06,
+      "loss": 1.3061,
+      "step": 46950
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.04e-06,
+      "loss": 1.2305,
+      "step": 46960
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.0300000000000002e-06,
+      "loss": 1.3097,
+      "step": 46970
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.0200000000000003e-06,
+      "loss": 1.2655,
+      "step": 46980
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.01e-06,
+      "loss": 1.321,
+      "step": 46990
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3e-06,
+      "loss": 1.1803,
+      "step": 47000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.99e-06,
+      "loss": 1.2844,
+      "step": 47010
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 1.2091,
+      "step": 47020
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.9700000000000004e-06,
+      "loss": 1.2416,
+      "step": 47030
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.9600000000000005e-06,
+      "loss": 1.331,
+      "step": 47040
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.95e-06,
+      "loss": 1.3034,
+      "step": 47050
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 1.2473,
+      "step": 47060
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.93e-06,
+      "loss": 1.216,
+      "step": 47070
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.92e-06,
+      "loss": 1.2535,
+      "step": 47080
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.91e-06,
+      "loss": 1.2177,
+      "step": 47090
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.9e-06,
+      "loss": 1.1994,
+      "step": 47100
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.89e-06,
+      "loss": 1.2631,
+      "step": 47110
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.88e-06,
+      "loss": 1.2383,
+      "step": 47120
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.87e-06,
+      "loss": 1.2278,
+      "step": 47130
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.86e-06,
+      "loss": 1.2227,
+      "step": 47140
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.8500000000000002e-06,
+      "loss": 1.2084,
+      "step": 47150
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.8400000000000003e-06,
+      "loss": 1.3369,
+      "step": 47160
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.83e-06,
+      "loss": 1.346,
+      "step": 47170
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.82e-06,
+      "loss": 1.2313,
+      "step": 47180
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.81e-06,
+      "loss": 1.2203,
+      "step": 47190
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 1.3039,
+      "step": 47200
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7900000000000004e-06,
+      "loss": 1.1866,
+      "step": 47210
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.78e-06,
+      "loss": 1.3063,
+      "step": 47220
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.77e-06,
+      "loss": 1.2415,
+      "step": 47230
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 1.1617,
+      "step": 47240
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7500000000000004e-06,
+      "loss": 1.2577,
+      "step": 47250
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.74e-06,
+      "loss": 1.2152,
+      "step": 47260
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.73e-06,
+      "loss": 1.2824,
+      "step": 47270
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.72e-06,
+      "loss": 1.1884,
+      "step": 47280
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.71e-06,
+      "loss": 1.2503,
+      "step": 47290
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7e-06,
+      "loss": 1.2633,
+      "step": 47300
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.69e-06,
+      "loss": 1.1045,
+      "step": 47310
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.68e-06,
+      "loss": 1.1565,
+      "step": 47320
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6700000000000003e-06,
+      "loss": 1.3789,
+      "step": 47330
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.66e-06,
+      "loss": 1.135,
+      "step": 47340
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.65e-06,
+      "loss": 1.2283,
+      "step": 47350
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.64e-06,
+      "loss": 1.1148,
+      "step": 47360
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6300000000000002e-06,
+      "loss": 1.2157,
+      "step": 47370
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6200000000000003e-06,
+      "loss": 1.2751,
+      "step": 47380
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6100000000000004e-06,
+      "loss": 1.2528,
+      "step": 47390
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6e-06,
+      "loss": 1.1917,
+      "step": 47400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.59e-06,
+      "loss": 1.2776,
+      "step": 47410
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 1.2521,
+      "step": 47420
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.5700000000000004e-06,
+      "loss": 1.1229,
+      "step": 47430
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.56e-06,
+      "loss": 1.1912,
+      "step": 47440
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.55e-06,
+      "loss": 1.2905,
+      "step": 47450
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.54e-06,
+      "loss": 1.2067,
+      "step": 47460
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.53e-06,
+      "loss": 1.2438,
+      "step": 47470
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.52e-06,
+      "loss": 1.3353,
+      "step": 47480
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.51e-06,
+      "loss": 1.1481,
+      "step": 47490
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.5e-06,
+      "loss": 1.2308,
+      "step": 47500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.49e-06,
+      "loss": 1.231,
+      "step": 47510
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.48e-06,
+      "loss": 1.2317,
+      "step": 47520
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.47e-06,
+      "loss": 1.3854,
+      "step": 47530
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.46e-06,
+      "loss": 1.259,
+      "step": 47540
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.4500000000000003e-06,
+      "loss": 1.1574,
+      "step": 47550
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.4400000000000004e-06,
+      "loss": 1.2615,
+      "step": 47560
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.43e-06,
+      "loss": 1.1619,
+      "step": 47570
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.42e-06,
+      "loss": 1.2308,
+      "step": 47580
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.4100000000000002e-06,
+      "loss": 1.1556,
+      "step": 47590
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.2297,
+      "step": 47600
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.3900000000000004e-06,
+      "loss": 1.3418,
+      "step": 47610
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.38e-06,
+      "loss": 1.1463,
+      "step": 47620
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.37e-06,
+      "loss": 1.2665,
+      "step": 47630
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.36e-06,
+      "loss": 1.346,
+      "step": 47640
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.35e-06,
+      "loss": 1.3292,
+      "step": 47650
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.34e-06,
+      "loss": 1.2078,
+      "step": 47660
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.33e-06,
+      "loss": 1.2569,
+      "step": 47670
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.32e-06,
+      "loss": 1.3267,
+      "step": 47680
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.31e-06,
+      "loss": 1.2662,
+      "step": 47690
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.3e-06,
+      "loss": 1.166,
+      "step": 47700
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.29e-06,
+      "loss": 1.3351,
+      "step": 47710
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.28e-06,
+      "loss": 1.1999,
+      "step": 47720
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.2700000000000003e-06,
+      "loss": 1.2256,
+      "step": 47730
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.26e-06,
+      "loss": 1.2048,
+      "step": 47740
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.25e-06,
+      "loss": 1.2535,
+      "step": 47750
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.24e-06,
+      "loss": 1.1992,
+      "step": 47760
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.2300000000000002e-06,
+      "loss": 1.2326,
+      "step": 47770
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.2200000000000003e-06,
+      "loss": 1.1266,
+      "step": 47780
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.2100000000000004e-06,
+      "loss": 1.1183,
+      "step": 47790
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.2e-06,
+      "loss": 1.1773,
+      "step": 47800
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.19e-06,
+      "loss": 1.1674,
+      "step": 47810
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 1.3528,
+      "step": 47820
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.17e-06,
+      "loss": 1.2002,
+      "step": 47830
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.16e-06,
+      "loss": 1.0927,
+      "step": 47840
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.1499999999999997e-06,
+      "loss": 1.3421,
+      "step": 47850
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.14e-06,
+      "loss": 1.2861,
+      "step": 47860
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.13e-06,
+      "loss": 1.3018,
+      "step": 47870
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.12e-06,
+      "loss": 1.2516,
+      "step": 47880
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.11e-06,
+      "loss": 1.3106,
+      "step": 47890
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 1.3275,
+      "step": 47900
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.09e-06,
+      "loss": 1.3153,
+      "step": 47910
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.08e-06,
+      "loss": 1.2137,
+      "step": 47920
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.07e-06,
+      "loss": 1.1559,
+      "step": 47930
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.06e-06,
+      "loss": 1.2132,
+      "step": 47940
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.0500000000000003e-06,
+      "loss": 1.2096,
+      "step": 47950
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.0400000000000004e-06,
+      "loss": 1.1562,
+      "step": 47960
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.03e-06,
+      "loss": 1.1882,
+      "step": 47970
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.02e-06,
+      "loss": 1.193,
+      "step": 47980
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.0100000000000002e-06,
+      "loss": 1.172,
+      "step": 47990
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.2852,
+      "step": 48000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.99e-06,
+      "loss": 1.2759,
+      "step": 48010
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.98e-06,
+      "loss": 1.3055,
+      "step": 48020
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9699999999999998e-06,
+      "loss": 1.2903,
+      "step": 48030
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.96e-06,
+      "loss": 1.2695,
+      "step": 48040
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.95e-06,
+      "loss": 1.2606,
+      "step": 48050
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.94e-06,
+      "loss": 1.2405,
+      "step": 48060
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.93e-06,
+      "loss": 1.1725,
+      "step": 48070
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.92e-06,
+      "loss": 1.1413,
+      "step": 48080
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.91e-06,
+      "loss": 1.3252,
+      "step": 48090
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9e-06,
+      "loss": 1.3112,
+      "step": 48100
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8900000000000001e-06,
+      "loss": 1.2518,
+      "step": 48110
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 1.1808,
+      "step": 48120
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8700000000000003e-06,
+      "loss": 1.1726,
+      "step": 48130
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.86e-06,
+      "loss": 1.2757,
+      "step": 48140
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.85e-06,
+      "loss": 1.277,
+      "step": 48150
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.84e-06,
+      "loss": 1.2175,
+      "step": 48160
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.83e-06,
+      "loss": 1.2638,
+      "step": 48170
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8200000000000002e-06,
+      "loss": 1.304,
+      "step": 48180
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8100000000000002e-06,
+      "loss": 1.3357,
+      "step": 48190
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8e-06,
+      "loss": 1.1893,
+      "step": 48200
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.79e-06,
+      "loss": 1.339,
+      "step": 48210
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 1.3387,
+      "step": 48220
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7700000000000002e-06,
+      "loss": 1.2159,
+      "step": 48230
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.76e-06,
+      "loss": 1.1293,
+      "step": 48240
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7500000000000002e-06,
+      "loss": 1.1566,
+      "step": 48250
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7399999999999999e-06,
+      "loss": 1.3019,
+      "step": 48260
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.73e-06,
+      "loss": 1.3018,
+      "step": 48270
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.72e-06,
+      "loss": 1.2026,
+      "step": 48280
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7100000000000001e-06,
+      "loss": 1.232,
+      "step": 48290
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 1.3659,
+      "step": 48300
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.69e-06,
+      "loss": 1.2638,
+      "step": 48310
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.68e-06,
+      "loss": 1.2624,
+      "step": 48320
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.67e-06,
+      "loss": 1.1901,
+      "step": 48330
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6600000000000002e-06,
+      "loss": 1.2311,
+      "step": 48340
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.65e-06,
+      "loss": 1.2476,
+      "step": 48350
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6400000000000002e-06,
+      "loss": 1.1607,
+      "step": 48360
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6299999999999999e-06,
+      "loss": 1.2204,
+      "step": 48370
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.62e-06,
+      "loss": 1.1779,
+      "step": 48380
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.61e-06,
+      "loss": 1.2311,
+      "step": 48390
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.3532,
+      "step": 48400
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5900000000000002e-06,
+      "loss": 1.1511,
+      "step": 48410
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5800000000000003e-06,
+      "loss": 1.3186,
+      "step": 48420
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.57e-06,
+      "loss": 1.2204,
+      "step": 48430
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.56e-06,
+      "loss": 1.0756,
+      "step": 48440
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.55e-06,
+      "loss": 1.231,
+      "step": 48450
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.54e-06,
+      "loss": 1.2607,
+      "step": 48460
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.53e-06,
+      "loss": 1.2036,
+      "step": 48470
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.52e-06,
+      "loss": 1.3247,
+      "step": 48480
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5100000000000002e-06,
+      "loss": 1.2768,
+      "step": 48490
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5e-06,
+      "loss": 1.2648,
+      "step": 48500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4900000000000001e-06,
+      "loss": 1.1921,
+      "step": 48510
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4800000000000002e-06,
+      "loss": 1.182,
+      "step": 48520
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4700000000000001e-06,
+      "loss": 1.2016,
+      "step": 48530
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.46e-06,
+      "loss": 1.3922,
+      "step": 48540
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.45e-06,
+      "loss": 1.255,
+      "step": 48550
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.44e-06,
+      "loss": 1.2747,
+      "step": 48560
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.43e-06,
+      "loss": 1.1842,
+      "step": 48570
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4200000000000002e-06,
+      "loss": 1.321,
+      "step": 48580
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.41e-06,
+      "loss": 1.1276,
+      "step": 48590
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 1.2196,
+      "step": 48600
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.39e-06,
+      "loss": 1.2966,
+      "step": 48610
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 1.1923,
+      "step": 48620
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.37e-06,
+      "loss": 1.1791,
+      "step": 48630
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.36e-06,
+      "loss": 1.2007,
+      "step": 48640
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.35e-06,
+      "loss": 1.2946,
+      "step": 48650
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.34e-06,
+      "loss": 1.2567,
+      "step": 48660
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.33e-06,
+      "loss": 1.1534,
+      "step": 48670
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.32e-06,
+      "loss": 1.2397,
+      "step": 48680
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3100000000000002e-06,
+      "loss": 1.2032,
+      "step": 48690
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3e-06,
+      "loss": 1.1944,
+      "step": 48700
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2900000000000001e-06,
+      "loss": 1.2335,
+      "step": 48710
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.28e-06,
+      "loss": 1.2502,
+      "step": 48720
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.27e-06,
+      "loss": 1.2653,
+      "step": 48730
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.26e-06,
+      "loss": 1.3404,
+      "step": 48740
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.25e-06,
+      "loss": 1.2053,
+      "step": 48750
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.24e-06,
+      "loss": 1.3022,
+      "step": 48760
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.23e-06,
+      "loss": 1.2656,
+      "step": 48770
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2200000000000002e-06,
+      "loss": 1.2367,
+      "step": 48780
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.21e-06,
+      "loss": 1.2221,
+      "step": 48790
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 1.4039,
+      "step": 48800
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.19e-06,
+      "loss": 1.1495,
+      "step": 48810
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.18e-06,
+      "loss": 1.3488,
+      "step": 48820
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.17e-06,
+      "loss": 1.3478,
+      "step": 48830
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.16e-06,
+      "loss": 1.1638,
+      "step": 48840
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.15e-06,
+      "loss": 1.3423,
+      "step": 48850
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.14e-06,
+      "loss": 1.2979,
+      "step": 48860
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.13e-06,
+      "loss": 1.1761,
+      "step": 48870
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.12e-06,
+      "loss": 1.2002,
+      "step": 48880
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1100000000000002e-06,
+      "loss": 1.2773,
+      "step": 48890
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1e-06,
+      "loss": 1.2885,
+      "step": 48900
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0900000000000002e-06,
+      "loss": 1.2756,
+      "step": 48910
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.08e-06,
+      "loss": 1.3772,
+      "step": 48920
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.07e-06,
+      "loss": 1.2726,
+      "step": 48930
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.06e-06,
+      "loss": 1.3269,
+      "step": 48940
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0500000000000001e-06,
+      "loss": 1.1859,
+      "step": 48950
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.04e-06,
+      "loss": 1.2663,
+      "step": 48960
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.03e-06,
+      "loss": 1.2277,
+      "step": 48970
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0200000000000002e-06,
+      "loss": 1.2873,
+      "step": 48980
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.01e-06,
+      "loss": 1.3076,
+      "step": 48990
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 1.1484,
+      "step": 49000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.9e-07,
+      "loss": 1.1667,
+      "step": 49010
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.8e-07,
+      "loss": 1.2986,
+      "step": 49020
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.7e-07,
+      "loss": 1.2242,
+      "step": 49030
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.6e-07,
+      "loss": 1.2124,
+      "step": 49040
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.5e-07,
+      "loss": 1.1419,
+      "step": 49050
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 1.4705,
+      "step": 49060
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.3e-07,
+      "loss": 1.3673,
+      "step": 49070
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.2e-07,
+      "loss": 1.2825,
+      "step": 49080
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.100000000000001e-07,
+      "loss": 1.2816,
+      "step": 49090
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9e-07,
+      "loss": 1.3387,
+      "step": 49100
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.900000000000001e-07,
+      "loss": 1.2206,
+      "step": 49110
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.8e-07,
+      "loss": 1.3244,
+      "step": 49120
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.699999999999999e-07,
+      "loss": 1.2446,
+      "step": 49130
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.6e-07,
+      "loss": 1.2923,
+      "step": 49140
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.500000000000001e-07,
+      "loss": 1.3434,
+      "step": 49150
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.4e-07,
+      "loss": 1.2867,
+      "step": 49160
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.300000000000001e-07,
+      "loss": 1.3705,
+      "step": 49170
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.200000000000001e-07,
+      "loss": 1.282,
+      "step": 49180
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.1e-07,
+      "loss": 1.178,
+      "step": 49190
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 1.3113,
+      "step": 49200
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.900000000000002e-07,
+      "loss": 1.1408,
+      "step": 49210
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.8e-07,
+      "loss": 1.2628,
+      "step": 49220
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.7e-07,
+      "loss": 1.2198,
+      "step": 49230
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.6e-07,
+      "loss": 1.2663,
+      "step": 49240
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.5e-07,
+      "loss": 1.2159,
+      "step": 49250
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.400000000000001e-07,
+      "loss": 1.3291,
+      "step": 49260
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.3e-07,
+      "loss": 1.1602,
+      "step": 49270
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.2e-07,
+      "loss": 1.2033,
+      "step": 49280
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.100000000000001e-07,
+      "loss": 1.2634,
+      "step": 49290
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 1.232,
+      "step": 49300
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.900000000000001e-07,
+      "loss": 1.32,
+      "step": 49310
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.8e-07,
+      "loss": 1.2847,
+      "step": 49320
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.7e-07,
+      "loss": 1.2841,
+      "step": 49330
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.6e-07,
+      "loss": 1.1504,
+      "step": 49340
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.5e-07,
+      "loss": 1.209,
+      "step": 49350
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.4e-07,
+      "loss": 1.1632,
+      "step": 49360
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.3e-07,
+      "loss": 1.3084,
+      "step": 49370
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.2e-07,
+      "loss": 1.2311,
+      "step": 49380
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.100000000000001e-07,
+      "loss": 1.3207,
+      "step": 49390
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 1.212,
+      "step": 49400
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.9e-07,
+      "loss": 1.1591,
+      "step": 49410
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.8e-07,
+      "loss": 1.1745,
+      "step": 49420
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.7e-07,
+      "loss": 1.14,
+      "step": 49430
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.6e-07,
+      "loss": 1.1459,
+      "step": 49440
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.5e-07,
+      "loss": 1.2258,
+      "step": 49450
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.4e-07,
+      "loss": 1.2463,
+      "step": 49460
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.3e-07,
+      "loss": 1.2719,
+      "step": 49470
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.2e-07,
+      "loss": 1.2549,
+      "step": 49480
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.100000000000001e-07,
+      "loss": 1.232,
+      "step": 49490
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 1.2823,
+      "step": 49500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.9e-07,
+      "loss": 1.2643,
+      "step": 49510
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.8e-07,
+      "loss": 1.2605,
+      "step": 49520
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.7000000000000005e-07,
+      "loss": 1.2048,
+      "step": 49530
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.6e-07,
+      "loss": 1.2467,
+      "step": 49540
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.5e-07,
+      "loss": 1.2734,
+      "step": 49550
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.4e-07,
+      "loss": 1.1696,
+      "step": 49560
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.3e-07,
+      "loss": 1.2643,
+      "step": 49570
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2e-07,
+      "loss": 1.271,
+      "step": 49580
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1000000000000004e-07,
+      "loss": 1.3499,
+      "step": 49590
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 1.2445,
+      "step": 49600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.9e-07,
+      "loss": 1.1398,
+      "step": 49610
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.8e-07,
+      "loss": 1.116,
+      "step": 49620
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.7000000000000006e-07,
+      "loss": 1.265,
+      "step": 49630
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.6e-07,
+      "loss": 1.3331,
+      "step": 49640
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5000000000000004e-07,
+      "loss": 1.1395,
+      "step": 49650
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.4e-07,
+      "loss": 1.1937,
+      "step": 49660
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.3e-07,
+      "loss": 1.2158,
+      "step": 49670
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.2e-07,
+      "loss": 1.1865,
+      "step": 49680
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.1e-07,
+      "loss": 1.2871,
+      "step": 49690
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 1.3864,
+      "step": 49700
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.9e-07,
+      "loss": 1.0766,
+      "step": 49710
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.8e-07,
+      "loss": 1.334,
+      "step": 49720
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.7e-07,
+      "loss": 1.2829,
+      "step": 49730
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6e-07,
+      "loss": 1.2309,
+      "step": 49740
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5000000000000004e-07,
+      "loss": 1.2767,
+      "step": 49750
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.4e-07,
+      "loss": 1.2445,
+      "step": 49760
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.3e-07,
+      "loss": 1.2141,
+      "step": 49770
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.2e-07,
+      "loss": 1.3221,
+      "step": 49780
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.1e-07,
+      "loss": 1.3357,
+      "step": 49790
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 1.1666,
+      "step": 49800
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9e-07,
+      "loss": 1.2364,
+      "step": 49810
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8e-07,
+      "loss": 1.0952,
+      "step": 49820
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7e-07,
+      "loss": 1.2958,
+      "step": 49830
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6e-07,
+      "loss": 1.261,
+      "step": 49840
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.5000000000000002e-07,
+      "loss": 1.2657,
+      "step": 49850
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4e-07,
+      "loss": 1.1989,
+      "step": 49860
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.3e-07,
+      "loss": 1.2341,
+      "step": 49870
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.2e-07,
+      "loss": 1.2332,
+      "step": 49880
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1e-07,
+      "loss": 1.3623,
+      "step": 49890
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 1.2791,
+      "step": 49900
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9e-08,
+      "loss": 1.1961,
+      "step": 49910
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8e-08,
+      "loss": 1.3181,
+      "step": 49920
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7e-08,
+      "loss": 1.1396,
+      "step": 49930
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6e-08,
+      "loss": 1.2849,
+      "step": 49940
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.0000000000000004e-08,
+      "loss": 1.3224,
+      "step": 49950
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4e-08,
+      "loss": 1.2267,
+      "step": 49960
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3e-08,
+      "loss": 1.2261,
+      "step": 49970
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-08,
+      "loss": 1.2057,
+      "step": 49980
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1e-08,
+      "loss": 1.0756,
+      "step": 49990
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0,
+      "loss": 1.1995,
+      "step": 50000
+    },
+    {
+      "epoch": 1.42,
+      "step": 50000,
+      "total_flos": 8.941575326722023e+19,
+      "train_loss": 2.0794223800468443,
+      "train_runtime": 42713.4006,
+      "train_samples_per_second": 9.365,
+      "train_steps_per_second": 1.171
+    }
+  ],
+  "max_steps": 50000,
+  "num_train_epochs": 2,
+  "total_flos": 8.941575326722023e+19,
+  "trial_name": null,
+  "trial_params": null
+}