diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,12024 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 6.622516556291391,
+  "global_step": 20000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 3.1336,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1e-05,
+      "loss": 3.0398,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1e-05,
+      "loss": 2.8756,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1e-05,
+      "loss": 2.7637,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 2.8073,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 2.6429,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 2.4954,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 2.4862,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 2.4175,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 2.3694,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1e-05,
+      "loss": 2.332,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1e-05,
+      "loss": 2.391,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1e-05,
+      "loss": 2.3132,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 2.2373,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 2.1669,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 2.1628,
+      "step": 160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1e-05,
+      "loss": 2.1119,
+      "step": 170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1e-05,
+      "loss": 2.0413,
+      "step": 180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1e-05,
+      "loss": 2.0823,
+      "step": 190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1e-05,
+      "loss": 2.0819,
+      "step": 200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1e-05,
+      "loss": 2.0258,
+      "step": 210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1e-05,
+      "loss": 2.0247,
+      "step": 220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1e-05,
+      "loss": 2.0177,
+      "step": 230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1e-05,
+      "loss": 2.0268,
+      "step": 240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1e-05,
+      "loss": 1.9926,
+      "step": 250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1e-05,
+      "loss": 1.9454,
+      "step": 260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1e-05,
+      "loss": 1.973,
+      "step": 270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1e-05,
+      "loss": 1.9811,
+      "step": 280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1e-05,
+      "loss": 1.8972,
+      "step": 290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1e-05,
+      "loss": 1.8972,
+      "step": 300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1e-05,
+      "loss": 1.8934,
+      "step": 310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1e-05,
+      "loss": 1.9546,
+      "step": 320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1e-05,
+      "loss": 1.8864,
+      "step": 330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1e-05,
+      "loss": 1.9125,
+      "step": 340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1e-05,
+      "loss": 1.942,
+      "step": 350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1e-05,
+      "loss": 1.8591,
+      "step": 360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1e-05,
+      "loss": 1.8922,
+      "step": 370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1e-05,
+      "loss": 1.862,
+      "step": 380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1e-05,
+      "loss": 1.8979,
+      "step": 390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1e-05,
+      "loss": 1.821,
+      "step": 400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1e-05,
+      "loss": 1.8045,
+      "step": 410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1e-05,
+      "loss": 1.8085,
+      "step": 420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1e-05,
+      "loss": 1.7711,
+      "step": 430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1e-05,
+      "loss": 1.8138,
+      "step": 440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1e-05,
+      "loss": 1.8338,
+      "step": 450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1e-05,
+      "loss": 1.7859,
+      "step": 460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1e-05,
+      "loss": 1.7567,
+      "step": 470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1e-05,
+      "loss": 1.8074,
+      "step": 480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1e-05,
+      "loss": 1.7722,
+      "step": 490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1e-05,
+      "loss": 1.785,
+      "step": 500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1e-05,
+      "loss": 1.7683,
+      "step": 510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1e-05,
+      "loss": 1.787,
+      "step": 520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1e-05,
+      "loss": 1.783,
+      "step": 530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1e-05,
+      "loss": 1.7691,
+      "step": 540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1e-05,
+      "loss": 1.779,
+      "step": 550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1e-05,
+      "loss": 1.7704,
+      "step": 560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1e-05,
+      "loss": 1.7574,
+      "step": 570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1e-05,
+      "loss": 1.8168,
+      "step": 580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1e-05,
+      "loss": 1.777,
+      "step": 590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1e-05,
+      "loss": 1.7237,
+      "step": 600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1e-05,
+      "loss": 1.7167,
+      "step": 610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1e-05,
+      "loss": 1.7372,
+      "step": 620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1e-05,
+      "loss": 1.7565,
+      "step": 630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1e-05,
+      "loss": 1.7251,
+      "step": 640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1e-05,
+      "loss": 1.7993,
+      "step": 650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1e-05,
+      "loss": 1.6941,
+      "step": 660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1e-05,
+      "loss": 1.6991,
+      "step": 670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1e-05,
+      "loss": 1.7539,
+      "step": 680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1e-05,
+      "loss": 1.6891,
+      "step": 690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1e-05,
+      "loss": 1.6789,
+      "step": 700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1e-05,
+      "loss": 1.7576,
+      "step": 710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1e-05,
+      "loss": 1.7513,
+      "step": 720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1e-05,
+      "loss": 1.7029,
+      "step": 730
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1e-05,
+      "loss": 1.6692,
+      "step": 740
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1e-05,
+      "loss": 1.6625,
+      "step": 750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1e-05,
+      "loss": 1.7496,
+      "step": 760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1e-05,
+      "loss": 1.7231,
+      "step": 770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1e-05,
+      "loss": 1.7262,
+      "step": 780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1e-05,
+      "loss": 1.6984,
+      "step": 790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1e-05,
+      "loss": 1.6897,
+      "step": 800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1e-05,
+      "loss": 1.6853,
+      "step": 810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1e-05,
+      "loss": 1.6822,
+      "step": 820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1e-05,
+      "loss": 1.7174,
+      "step": 830
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1e-05,
+      "loss": 1.687,
+      "step": 840
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1e-05,
+      "loss": 1.6791,
+      "step": 850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1e-05,
+      "loss": 1.7481,
+      "step": 860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1e-05,
+      "loss": 1.6298,
+      "step": 870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1e-05,
+      "loss": 1.7056,
+      "step": 880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1e-05,
+      "loss": 1.6225,
+      "step": 890
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1e-05,
+      "loss": 1.6684,
+      "step": 900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1e-05,
+      "loss": 1.6496,
+      "step": 910
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1e-05,
+      "loss": 1.7033,
+      "step": 920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1e-05,
+      "loss": 1.6358,
+      "step": 930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1e-05,
+      "loss": 1.6297,
+      "step": 940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1e-05,
+      "loss": 1.6304,
+      "step": 950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1e-05,
+      "loss": 1.6235,
+      "step": 960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1e-05,
+      "loss": 1.6883,
+      "step": 970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1e-05,
+      "loss": 1.6659,
+      "step": 980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1e-05,
+      "loss": 1.6617,
+      "step": 990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1e-05,
+      "loss": 1.662,
+      "step": 1000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1e-05,
+      "loss": 1.6717,
+      "step": 1010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1e-05,
+      "loss": 1.6247,
+      "step": 1020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1e-05,
+      "loss": 1.6542,
+      "step": 1030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1e-05,
+      "loss": 1.6591,
+      "step": 1040
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1e-05,
+      "loss": 1.6574,
+      "step": 1050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1e-05,
+      "loss": 1.659,
+      "step": 1060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1e-05,
+      "loss": 1.6471,
+      "step": 1070
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1e-05,
+      "loss": 1.6781,
+      "step": 1080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1e-05,
+      "loss": 1.6282,
+      "step": 1090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1e-05,
+      "loss": 1.622,
+      "step": 1100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1e-05,
+      "loss": 1.6326,
+      "step": 1110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1e-05,
+      "loss": 1.6366,
+      "step": 1120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1e-05,
+      "loss": 1.655,
+      "step": 1130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1e-05,
+      "loss": 1.6051,
+      "step": 1140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1e-05,
+      "loss": 1.6375,
+      "step": 1150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1e-05,
+      "loss": 1.6372,
+      "step": 1160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1e-05,
+      "loss": 1.6227,
+      "step": 1170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1e-05,
+      "loss": 1.6234,
+      "step": 1180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1e-05,
+      "loss": 1.6219,
+      "step": 1190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1e-05,
+      "loss": 1.6338,
+      "step": 1200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1e-05,
+      "loss": 1.6333,
+      "step": 1210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1e-05,
+      "loss": 1.6488,
+      "step": 1220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1e-05,
+      "loss": 1.6278,
+      "step": 1230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1e-05,
+      "loss": 1.6314,
+      "step": 1240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1e-05,
+      "loss": 1.5687,
+      "step": 1250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1e-05,
+      "loss": 1.6316,
+      "step": 1260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1e-05,
+      "loss": 1.6234,
+      "step": 1270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1e-05,
+      "loss": 1.6042,
+      "step": 1280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1e-05,
+      "loss": 1.6176,
+      "step": 1290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1e-05,
+      "loss": 1.6126,
+      "step": 1300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1e-05,
+      "loss": 1.6224,
+      "step": 1310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1e-05,
+      "loss": 1.6026,
+      "step": 1320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1e-05,
+      "loss": 1.5838,
+      "step": 1330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1e-05,
+      "loss": 1.6111,
+      "step": 1340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1e-05,
+      "loss": 1.5864,
+      "step": 1350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1e-05,
+      "loss": 1.5923,
+      "step": 1360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1e-05,
+      "loss": 1.627,
+      "step": 1370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1e-05,
+      "loss": 1.6162,
+      "step": 1380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1e-05,
+      "loss": 1.5858,
+      "step": 1390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1e-05,
+      "loss": 1.6053,
+      "step": 1400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1e-05,
+      "loss": 1.593,
+      "step": 1410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1e-05,
+      "loss": 1.6104,
+      "step": 1420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1e-05,
+      "loss": 1.5995,
+      "step": 1430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1e-05,
+      "loss": 1.6276,
+      "step": 1440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1e-05,
+      "loss": 1.6072,
+      "step": 1450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1e-05,
+      "loss": 1.605,
+      "step": 1460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1e-05,
+      "loss": 1.5998,
+      "step": 1470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1e-05,
+      "loss": 1.5901,
+      "step": 1480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1e-05,
+      "loss": 1.5852,
+      "step": 1490
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1e-05,
+      "loss": 1.5957,
+      "step": 1500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1e-05,
+      "loss": 1.6031,
+      "step": 1510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1e-05,
+      "loss": 1.5938,
+      "step": 1520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1e-05,
+      "loss": 1.5941,
+      "step": 1530
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1e-05,
+      "loss": 1.5686,
+      "step": 1540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1e-05,
+      "loss": 1.5965,
+      "step": 1550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 1.5728,
+      "step": 1560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 1.5887,
+      "step": 1570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 1.5972,
+      "step": 1580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1e-05,
+      "loss": 1.5899,
+      "step": 1590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1e-05,
+      "loss": 1.573,
+      "step": 1600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1e-05,
+      "loss": 1.5697,
+      "step": 1610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1e-05,
+      "loss": 1.5732,
+      "step": 1620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1e-05,
+      "loss": 1.567,
+      "step": 1630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1e-05,
+      "loss": 1.5478,
+      "step": 1640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1e-05,
+      "loss": 1.5869,
+      "step": 1650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1e-05,
+      "loss": 1.5467,
+      "step": 1660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1e-05,
+      "loss": 1.5804,
+      "step": 1670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1e-05,
+      "loss": 1.5817,
+      "step": 1680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1e-05,
+      "loss": 1.5662,
+      "step": 1690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1e-05,
+      "loss": 1.5815,
+      "step": 1700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1e-05,
+      "loss": 1.574,
+      "step": 1710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1e-05,
+      "loss": 1.5662,
+      "step": 1720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1e-05,
+      "loss": 1.5477,
+      "step": 1730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1e-05,
+      "loss": 1.5759,
+      "step": 1740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1e-05,
+      "loss": 1.5687,
+      "step": 1750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1e-05,
+      "loss": 1.5663,
+      "step": 1760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1e-05,
+      "loss": 1.6025,
+      "step": 1770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1e-05,
+      "loss": 1.5835,
+      "step": 1780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1e-05,
+      "loss": 1.5775,
+      "step": 1790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1e-05,
+      "loss": 1.58,
+      "step": 1800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1e-05,
+      "loss": 1.5571,
+      "step": 1810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1e-05,
+      "loss": 1.5652,
+      "step": 1820
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1e-05,
+      "loss": 1.5618,
+      "step": 1830
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1e-05,
+      "loss": 1.5754,
+      "step": 1840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1e-05,
+      "loss": 1.5711,
+      "step": 1850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1e-05,
+      "loss": 1.5758,
+      "step": 1860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1e-05,
+      "loss": 1.5631,
+      "step": 1870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1e-05,
+      "loss": 1.5735,
+      "step": 1880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1e-05,
+      "loss": 1.5671,
+      "step": 1890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1e-05,
+      "loss": 1.5622,
+      "step": 1900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1e-05,
+      "loss": 1.5552,
+      "step": 1910
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1e-05,
+      "loss": 1.5743,
+      "step": 1920
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1e-05,
+      "loss": 1.5508,
+      "step": 1930
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1e-05,
+      "loss": 1.5454,
+      "step": 1940
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1e-05,
+      "loss": 1.5481,
+      "step": 1950
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1e-05,
+      "loss": 1.5428,
+      "step": 1960
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1e-05,
+      "loss": 1.5566,
+      "step": 1970
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1e-05,
+      "loss": 1.5379,
+      "step": 1980
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1e-05,
+      "loss": 1.5503,
+      "step": 1990
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1e-05,
+      "loss": 1.5545,
+      "step": 2000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1e-05,
+      "loss": 1.5488,
+      "step": 2010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1e-05,
+      "loss": 1.5547,
+      "step": 2020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1e-05,
+      "loss": 1.5533,
+      "step": 2030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1e-05,
+      "loss": 1.5438,
+      "step": 2040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1e-05,
+      "loss": 1.5376,
+      "step": 2050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1e-05,
+      "loss": 1.5208,
+      "step": 2060
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1e-05,
+      "loss": 1.5414,
+      "step": 2070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1e-05,
+      "loss": 1.5406,
+      "step": 2080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1e-05,
+      "loss": 1.5336,
+      "step": 2090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1e-05,
+      "loss": 1.5498,
+      "step": 2100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1e-05,
+      "loss": 1.5409,
+      "step": 2110
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1e-05,
+      "loss": 1.5468,
+      "step": 2120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1e-05,
+      "loss": 1.5399,
+      "step": 2130
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1e-05,
+      "loss": 1.5362,
+      "step": 2140
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1e-05,
+      "loss": 1.5442,
+      "step": 2150
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1e-05,
+      "loss": 1.5447,
+      "step": 2160
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1e-05,
+      "loss": 1.5251,
+      "step": 2170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1e-05,
+      "loss": 1.5543,
+      "step": 2180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1e-05,
+      "loss": 1.5498,
+      "step": 2190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1e-05,
+      "loss": 1.5488,
+      "step": 2200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1e-05,
+      "loss": 1.5638,
+      "step": 2210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1e-05,
+      "loss": 1.5441,
+      "step": 2220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1e-05,
+      "loss": 1.5702,
+      "step": 2230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1e-05,
+      "loss": 1.5235,
+      "step": 2240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1e-05,
+      "loss": 1.5513,
+      "step": 2250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1e-05,
+      "loss": 1.5773,
+      "step": 2260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1e-05,
+      "loss": 1.5334,
+      "step": 2270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1e-05,
+      "loss": 1.5438,
+      "step": 2280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1e-05,
+      "loss": 1.5314,
+      "step": 2290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1e-05,
+      "loss": 1.5513,
+      "step": 2300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1e-05,
+      "loss": 1.5455,
+      "step": 2310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1e-05,
+      "loss": 1.5376,
+      "step": 2320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1e-05,
+      "loss": 1.5342,
+      "step": 2330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1e-05,
+      "loss": 1.5529,
+      "step": 2340
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1e-05,
+      "loss": 1.5275,
+      "step": 2350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1e-05,
+      "loss": 1.5225,
+      "step": 2360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1e-05,
+      "loss": 1.5365,
+      "step": 2370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1e-05,
+      "loss": 1.5259,
+      "step": 2380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1e-05,
+      "loss": 1.5132,
+      "step": 2390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1e-05,
+      "loss": 1.5348,
+      "step": 2400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1e-05,
+      "loss": 1.526,
+      "step": 2410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1e-05,
+      "loss": 1.504,
+      "step": 2420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1e-05,
+      "loss": 1.5165,
+      "step": 2430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1e-05,
+      "loss": 1.5198,
+      "step": 2440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1e-05,
+      "loss": 1.5217,
+      "step": 2450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1e-05,
+      "loss": 1.5337,
+      "step": 2460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1e-05,
+      "loss": 1.5346,
+      "step": 2470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1e-05,
+      "loss": 1.5398,
+      "step": 2480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1e-05,
+      "loss": 1.5354,
+      "step": 2490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1e-05,
+      "loss": 1.5319,
+      "step": 2500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1e-05,
+      "loss": 1.5213,
+      "step": 2510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1e-05,
+      "loss": 1.5372,
+      "step": 2520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1e-05,
+      "loss": 1.5342,
+      "step": 2530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1e-05,
+      "loss": 1.5207,
+      "step": 2540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1e-05,
+      "loss": 1.5374,
+      "step": 2550
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1e-05,
+      "loss": 1.524,
+      "step": 2560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1e-05,
+      "loss": 1.5227,
+      "step": 2570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1e-05,
+      "loss": 1.5256,
+      "step": 2580
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1e-05,
+      "loss": 1.5234,
+      "step": 2590
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1e-05,
+      "loss": 1.5255,
+      "step": 2600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1e-05,
+      "loss": 1.5438,
+      "step": 2610
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1e-05,
+      "loss": 1.5142,
+      "step": 2620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1e-05,
+      "loss": 1.5102,
+      "step": 2630
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1e-05,
+      "loss": 1.523,
+      "step": 2640
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1e-05,
+      "loss": 1.5093,
+      "step": 2650
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1e-05,
+      "loss": 1.511,
+      "step": 2660
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1e-05,
+      "loss": 1.5135,
+      "step": 2670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1e-05,
+      "loss": 1.521,
+      "step": 2680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1e-05,
+      "loss": 1.5128,
+      "step": 2690
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1e-05,
+      "loss": 1.5314,
+      "step": 2700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1e-05,
+      "loss": 1.528,
+      "step": 2710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1e-05,
+      "loss": 1.5362,
+      "step": 2720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1e-05,
+      "loss": 1.5091,
+      "step": 2730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1e-05,
+      "loss": 1.503,
+      "step": 2740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1e-05,
+      "loss": 1.5113,
+      "step": 2750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1e-05,
+      "loss": 1.498,
+      "step": 2760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1e-05,
+      "loss": 1.5127,
+      "step": 2770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1e-05,
+      "loss": 1.5028,
+      "step": 2780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1e-05,
+      "loss": 1.5245,
+      "step": 2790
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1e-05,
+      "loss": 1.5197,
+      "step": 2800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1e-05,
+      "loss": 1.5238,
+      "step": 2810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1e-05,
+      "loss": 1.5069,
+      "step": 2820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1e-05,
+      "loss": 1.5202,
+      "step": 2830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1e-05,
+      "loss": 1.5388,
+      "step": 2840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1e-05,
+      "loss": 1.5061,
+      "step": 2850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1e-05,
+      "loss": 1.5297,
+      "step": 2860
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1e-05,
+      "loss": 1.5185,
+      "step": 2870
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1e-05,
+      "loss": 1.5213,
+      "step": 2880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1e-05,
+      "loss": 1.5189,
+      "step": 2890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1e-05,
+      "loss": 1.4997,
+      "step": 2900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1e-05,
+      "loss": 1.5122,
+      "step": 2910
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1e-05,
+      "loss": 1.5351,
+      "step": 2920
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1e-05,
+      "loss": 1.5031,
+      "step": 2930
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1e-05,
+      "loss": 1.5108,
+      "step": 2940
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1e-05,
+      "loss": 1.514,
+      "step": 2950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1e-05,
+      "loss": 1.5275,
+      "step": 2960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1e-05,
+      "loss": 1.5156,
+      "step": 2970
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1e-05,
+      "loss": 1.5315,
+      "step": 2980
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1e-05,
+      "loss": 1.5007,
+      "step": 2990
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1e-05,
+      "loss": 1.5104,
+      "step": 3000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1e-05,
+      "loss": 1.5182,
+      "step": 3010
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1e-05,
+      "loss": 1.5091,
+      "step": 3020
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1e-05,
+      "loss": 1.5054,
+      "step": 3030
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1e-05,
+      "loss": 1.4926,
+      "step": 3040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1e-05,
+      "loss": 1.5106,
+      "step": 3050
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1e-05,
+      "loss": 1.5055,
+      "step": 3060
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1e-05,
+      "loss": 1.5039,
+      "step": 3070
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1e-05,
+      "loss": 1.4901,
+      "step": 3080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1e-05,
+      "loss": 1.5109,
+      "step": 3090
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1e-05,
+      "loss": 1.5086,
+      "step": 3100
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1e-05,
+      "loss": 1.4979,
+      "step": 3110
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1e-05,
+      "loss": 1.4867,
+      "step": 3120
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1e-05,
+      "loss": 1.5192,
+      "step": 3130
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1e-05,
+      "loss": 1.5144,
+      "step": 3140
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1e-05,
+      "loss": 1.4885,
+      "step": 3150
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1e-05,
+      "loss": 1.4952,
+      "step": 3160
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1e-05,
+      "loss": 1.525,
+      "step": 3170
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1e-05,
+      "loss": 1.4877,
+      "step": 3180
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1e-05,
+      "loss": 1.5056,
+      "step": 3190
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1e-05,
+      "loss": 1.4994,
+      "step": 3200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1e-05,
+      "loss": 1.4996,
+      "step": 3210
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1e-05,
+      "loss": 1.4979,
+      "step": 3220
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1e-05,
+      "loss": 1.4815,
+      "step": 3230
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1e-05,
+      "loss": 1.4926,
+      "step": 3240
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1e-05,
+      "loss": 1.5056,
+      "step": 3250
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1e-05,
+      "loss": 1.4986,
+      "step": 3260
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1e-05,
+      "loss": 1.5161,
+      "step": 3270
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1e-05,
+      "loss": 1.5024,
+      "step": 3280
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1e-05,
+      "loss": 1.4915,
+      "step": 3290
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1e-05,
+      "loss": 1.5146,
+      "step": 3300
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1e-05,
+      "loss": 1.4937,
+      "step": 3310
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1e-05,
+      "loss": 1.5194,
+      "step": 3320
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1e-05,
+      "loss": 1.4959,
+      "step": 3330
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1e-05,
+      "loss": 1.4841,
+      "step": 3340
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1e-05,
+      "loss": 1.5012,
+      "step": 3350
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1e-05,
+      "loss": 1.5101,
+      "step": 3360
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1e-05,
+      "loss": 1.4965,
+      "step": 3370
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1e-05,
+      "loss": 1.4951,
+      "step": 3380
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1e-05,
+      "loss": 1.5091,
+      "step": 3390
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1e-05,
+      "loss": 1.4971,
+      "step": 3400
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1e-05,
+      "loss": 1.5015,
+      "step": 3410
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1e-05,
+      "loss": 1.4962,
+      "step": 3420
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1e-05,
+      "loss": 1.5075,
+      "step": 3430
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1e-05,
+      "loss": 1.4893,
+      "step": 3440
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1e-05,
+      "loss": 1.4908,
+      "step": 3450
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1e-05,
+      "loss": 1.4952,
+      "step": 3460
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1e-05,
+      "loss": 1.5181,
+      "step": 3470
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1e-05,
+      "loss": 1.4825,
+      "step": 3480
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1e-05,
+      "loss": 1.5025,
+      "step": 3490
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1e-05,
+      "loss": 1.4843,
+      "step": 3500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1e-05,
+      "loss": 1.4863,
+      "step": 3510
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1e-05,
+      "loss": 1.4904,
+      "step": 3520
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1e-05,
+      "loss": 1.4903,
+      "step": 3530
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1e-05,
+      "loss": 1.4866,
+      "step": 3540
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1e-05,
+      "loss": 1.4906,
+      "step": 3550
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1e-05,
+      "loss": 1.4824,
+      "step": 3560
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1e-05,
+      "loss": 1.4775,
+      "step": 3570
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1e-05,
+      "loss": 1.5007,
+      "step": 3580
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1e-05,
+      "loss": 1.4702,
+      "step": 3590
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1e-05,
+      "loss": 1.4776,
+      "step": 3600
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1e-05,
+      "loss": 1.4917,
+      "step": 3610
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1e-05,
+      "loss": 1.4886,
+      "step": 3620
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1e-05,
+      "loss": 1.5013,
+      "step": 3630
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1e-05,
+      "loss": 1.4929,
+      "step": 3640
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1e-05,
+      "loss": 1.4898,
+      "step": 3650
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1e-05,
+      "loss": 1.496,
+      "step": 3660
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1e-05,
+      "loss": 1.488,
+      "step": 3670
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1e-05,
+      "loss": 1.4848,
+      "step": 3680
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1e-05,
+      "loss": 1.4874,
+      "step": 3690
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1e-05,
+      "loss": 1.488,
+      "step": 3700
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1e-05,
+      "loss": 1.4925,
+      "step": 3710
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1e-05,
+      "loss": 1.4783,
+      "step": 3720
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1e-05,
+      "loss": 1.484,
+      "step": 3730
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1e-05,
+      "loss": 1.4915,
+      "step": 3740
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1e-05,
+      "loss": 1.4918,
+      "step": 3750
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1e-05,
+      "loss": 1.4804,
+      "step": 3760
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1e-05,
+      "loss": 1.4816,
+      "step": 3770
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1e-05,
+      "loss": 1.4933,
+      "step": 3780
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1e-05,
+      "loss": 1.4813,
+      "step": 3790
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1e-05,
+      "loss": 1.4996,
+      "step": 3800
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1e-05,
+      "loss": 1.4839,
+      "step": 3810
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1e-05,
+      "loss": 1.5016,
+      "step": 3820
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1e-05,
+      "loss": 1.4764,
+      "step": 3830
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1e-05,
+      "loss": 1.4844,
+      "step": 3840
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1e-05,
+      "loss": 1.4948,
+      "step": 3850
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1e-05,
+      "loss": 1.4711,
+      "step": 3860
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1e-05,
+      "loss": 1.5079,
+      "step": 3870
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1e-05,
+      "loss": 1.4873,
+      "step": 3880
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1e-05,
+      "loss": 1.4951,
+      "step": 3890
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1e-05,
+      "loss": 1.4874,
+      "step": 3900
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1e-05,
+      "loss": 1.49,
+      "step": 3910
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1e-05,
+      "loss": 1.4918,
+      "step": 3920
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1e-05,
+      "loss": 1.4807,
+      "step": 3930
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1e-05,
+      "loss": 1.4754,
+      "step": 3940
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1e-05,
+      "loss": 1.489,
+      "step": 3950
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1e-05,
+      "loss": 1.4829,
+      "step": 3960
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1e-05,
+      "loss": 1.4921,
+      "step": 3970
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1e-05,
+      "loss": 1.465,
+      "step": 3980
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1e-05,
+      "loss": 1.4717,
+      "step": 3990
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1e-05,
+      "loss": 1.4777,
+      "step": 4000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1e-05,
+      "loss": 1.4897,
+      "step": 4010
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1e-05,
+      "loss": 1.4942,
+      "step": 4020
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1e-05,
+      "loss": 1.4903,
+      "step": 4030
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1e-05,
+      "loss": 1.4869,
+      "step": 4040
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1e-05,
+      "loss": 1.4905,
+      "step": 4050
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1e-05,
+      "loss": 1.5023,
+      "step": 4060
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1e-05,
+      "loss": 1.4783,
+      "step": 4070
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1e-05,
+      "loss": 1.498,
+      "step": 4080
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1e-05,
+      "loss": 1.4699,
+      "step": 4090
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1e-05,
+      "loss": 1.4863,
+      "step": 4100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1e-05,
+      "loss": 1.4769,
+      "step": 4110
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1e-05,
+      "loss": 1.4839,
+      "step": 4120
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1e-05,
+      "loss": 1.4855,
+      "step": 4130
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1e-05,
+      "loss": 1.4696,
+      "step": 4140
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1e-05,
+      "loss": 1.4816,
+      "step": 4150
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1e-05,
+      "loss": 1.4841,
+      "step": 4160
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1e-05,
+      "loss": 1.468,
+      "step": 4170
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1e-05,
+      "loss": 1.4788,
+      "step": 4180
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1e-05,
+      "loss": 1.4826,
+      "step": 4190
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1e-05,
+      "loss": 1.4893,
+      "step": 4200
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1e-05,
+      "loss": 1.4789,
+      "step": 4210
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1e-05,
+      "loss": 1.4815,
+      "step": 4220
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1e-05,
+      "loss": 1.4819,
+      "step": 4230
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1e-05,
+      "loss": 1.4779,
+      "step": 4240
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1e-05,
+      "loss": 1.4781,
+      "step": 4250
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1e-05,
+      "loss": 1.4796,
+      "step": 4260
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1e-05,
+      "loss": 1.4778,
+      "step": 4270
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1e-05,
+      "loss": 1.4763,
+      "step": 4280
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1e-05,
+      "loss": 1.5026,
+      "step": 4290
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1e-05,
+      "loss": 1.4951,
+      "step": 4300
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1e-05,
+      "loss": 1.471,
+      "step": 4310
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1e-05,
+      "loss": 1.4893,
+      "step": 4320
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1e-05,
+      "loss": 1.4742,
+      "step": 4330
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1e-05,
+      "loss": 1.4766,
+      "step": 4340
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1e-05,
+      "loss": 1.4802,
+      "step": 4350
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1e-05,
+      "loss": 1.4879,
+      "step": 4360
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1e-05,
+      "loss": 1.4758,
+      "step": 4370
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1e-05,
+      "loss": 1.4751,
+      "step": 4380
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1e-05,
+      "loss": 1.4991,
+      "step": 4390
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1e-05,
+      "loss": 1.4764,
+      "step": 4400
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1e-05,
+      "loss": 1.4698,
+      "step": 4410
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1e-05,
+      "loss": 1.4743,
+      "step": 4420
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1e-05,
+      "loss": 1.4766,
+      "step": 4430
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1e-05,
+      "loss": 1.4715,
+      "step": 4440
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1e-05,
+      "loss": 1.4749,
+      "step": 4450
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1e-05,
+      "loss": 1.4775,
+      "step": 4460
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1e-05,
+      "loss": 1.4698,
+      "step": 4470
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1e-05,
+      "loss": 1.4672,
+      "step": 4480
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1e-05,
+      "loss": 1.4768,
+      "step": 4490
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1e-05,
+      "loss": 1.4713,
+      "step": 4500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1e-05,
+      "loss": 1.4765,
+      "step": 4510
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1e-05,
+      "loss": 1.471,
+      "step": 4520
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1e-05,
+      "loss": 1.4752,
+      "step": 4530
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1e-05,
+      "loss": 1.475,
+      "step": 4540
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1e-05,
+      "loss": 1.4761,
+      "step": 4550
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1e-05,
+      "loss": 1.4733,
+      "step": 4560
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1e-05,
+      "loss": 1.4685,
+      "step": 4570
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1e-05,
+      "loss": 1.4782,
+      "step": 4580
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1e-05,
+      "loss": 1.4714,
+      "step": 4590
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1e-05,
+      "loss": 1.4844,
+      "step": 4600
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1e-05,
+      "loss": 1.4662,
+      "step": 4610
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1e-05,
+      "loss": 1.4724,
+      "step": 4620
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1e-05,
+      "loss": 1.4831,
+      "step": 4630
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1e-05,
+      "loss": 1.4775,
+      "step": 4640
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1e-05,
+      "loss": 1.4691,
+      "step": 4650
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1e-05,
+      "loss": 1.4794,
+      "step": 4660
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1e-05,
+      "loss": 1.4742,
+      "step": 4670
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1e-05,
+      "loss": 1.4686,
+      "step": 4680
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1e-05,
+      "loss": 1.4724,
+      "step": 4690
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1e-05,
+      "loss": 1.4771,
+      "step": 4700
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1e-05,
+      "loss": 1.4761,
+      "step": 4710
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1e-05,
+      "loss": 1.4675,
+      "step": 4720
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1e-05,
+      "loss": 1.4871,
+      "step": 4730
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1e-05,
+      "loss": 1.4745,
+      "step": 4740
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1e-05,
+      "loss": 1.4722,
+      "step": 4750
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1e-05,
+      "loss": 1.4756,
+      "step": 4760
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1e-05,
+      "loss": 1.4697,
+      "step": 4770
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1e-05,
+      "loss": 1.4713,
+      "step": 4780
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1e-05,
+      "loss": 1.4576,
+      "step": 4790
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1e-05,
+      "loss": 1.4642,
+      "step": 4800
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1e-05,
+      "loss": 1.478,
+      "step": 4810
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1e-05,
+      "loss": 1.4718,
+      "step": 4820
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1e-05,
+      "loss": 1.4681,
+      "step": 4830
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1e-05,
+      "loss": 1.4749,
+      "step": 4840
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1e-05,
+      "loss": 1.4612,
+      "step": 4850
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1e-05,
+      "loss": 1.4878,
+      "step": 4860
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1e-05,
+      "loss": 1.4638,
+      "step": 4870
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1e-05,
+      "loss": 1.464,
+      "step": 4880
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1e-05,
+      "loss": 1.461,
+      "step": 4890
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1e-05,
+      "loss": 1.4787,
+      "step": 4900
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1e-05,
+      "loss": 1.473,
+      "step": 4910
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1e-05,
+      "loss": 1.4645,
+      "step": 4920
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1e-05,
+      "loss": 1.4648,
+      "step": 4930
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1e-05,
+      "loss": 1.4685,
+      "step": 4940
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1e-05,
+      "loss": 1.4717,
+      "step": 4950
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1e-05,
+      "loss": 1.4697,
+      "step": 4960
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1e-05,
+      "loss": 1.4648,
+      "step": 4970
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1e-05,
+      "loss": 1.4641,
+      "step": 4980
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1e-05,
+      "loss": 1.4672,
+      "step": 4990
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1e-05,
+      "loss": 1.4563,
+      "step": 5000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1e-05,
+      "loss": 1.4808,
+      "step": 5010
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1e-05,
+      "loss": 1.4689,
+      "step": 5020
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1e-05,
+      "loss": 1.4641,
+      "step": 5030
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1e-05,
+      "loss": 1.4687,
+      "step": 5040
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1e-05,
+      "loss": 1.4688,
+      "step": 5050
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1e-05,
+      "loss": 1.4814,
+      "step": 5060
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1e-05,
+      "loss": 1.4705,
+      "step": 5070
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1e-05,
+      "loss": 1.471,
+      "step": 5080
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1e-05,
+      "loss": 1.4545,
+      "step": 5090
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1e-05,
+      "loss": 1.4756,
+      "step": 5100
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1e-05,
+      "loss": 1.4614,
+      "step": 5110
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1e-05,
+      "loss": 1.456,
+      "step": 5120
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1e-05,
+      "loss": 1.4703,
+      "step": 5130
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1e-05,
+      "loss": 1.4682,
+      "step": 5140
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1e-05,
+      "loss": 1.468,
+      "step": 5150
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1e-05,
+      "loss": 1.4604,
+      "step": 5160
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1e-05,
+      "loss": 1.4511,
+      "step": 5170
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1e-05,
+      "loss": 1.4853,
+      "step": 5180
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1e-05,
+      "loss": 1.4745,
+      "step": 5190
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1e-05,
+      "loss": 1.4743,
+      "step": 5200
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1e-05,
+      "loss": 1.4704,
+      "step": 5210
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1e-05,
+      "loss": 1.4575,
+      "step": 5220
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1e-05,
+      "loss": 1.4601,
+      "step": 5230
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1e-05,
+      "loss": 1.4792,
+      "step": 5240
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1e-05,
+      "loss": 1.4582,
+      "step": 5250
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1e-05,
+      "loss": 1.4567,
+      "step": 5260
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1e-05,
+      "loss": 1.4637,
+      "step": 5270
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1e-05,
+      "loss": 1.4711,
+      "step": 5280
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1e-05,
+      "loss": 1.4801,
+      "step": 5290
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1e-05,
+      "loss": 1.473,
+      "step": 5300
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1e-05,
+      "loss": 1.4644,
+      "step": 5310
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1e-05,
+      "loss": 1.4599,
+      "step": 5320
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1e-05,
+      "loss": 1.4652,
+      "step": 5330
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1e-05,
+      "loss": 1.4602,
+      "step": 5340
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1e-05,
+      "loss": 1.4643,
+      "step": 5350
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1e-05,
+      "loss": 1.466,
+      "step": 5360
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1e-05,
+      "loss": 1.4643,
+      "step": 5370
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1e-05,
+      "loss": 1.4512,
+      "step": 5380
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1e-05,
+      "loss": 1.468,
+      "step": 5390
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1e-05,
+      "loss": 1.4616,
+      "step": 5400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1e-05,
+      "loss": 1.4822,
+      "step": 5410
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1e-05,
+      "loss": 1.4737,
+      "step": 5420
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1e-05,
+      "loss": 1.454,
+      "step": 5430
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1e-05,
+      "loss": 1.4657,
+      "step": 5440
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1e-05,
+      "loss": 1.4648,
+      "step": 5450
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1e-05,
+      "loss": 1.4629,
+      "step": 5460
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1e-05,
+      "loss": 1.4727,
+      "step": 5470
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1e-05,
+      "loss": 1.4669,
+      "step": 5480
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1e-05,
+      "loss": 1.4656,
+      "step": 5490
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1e-05,
+      "loss": 1.4642,
+      "step": 5500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1e-05,
+      "loss": 1.4582,
+      "step": 5510
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1e-05,
+      "loss": 1.4677,
+      "step": 5520
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1e-05,
+      "loss": 1.4576,
+      "step": 5530
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1e-05,
+      "loss": 1.4528,
+      "step": 5540
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1e-05,
+      "loss": 1.461,
+      "step": 5550
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1e-05,
+      "loss": 1.4747,
+      "step": 5560
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1e-05,
+      "loss": 1.4508,
+      "step": 5570
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1e-05,
+      "loss": 1.4614,
+      "step": 5580
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1e-05,
+      "loss": 1.4644,
+      "step": 5590
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1e-05,
+      "loss": 1.4627,
+      "step": 5600
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1e-05,
+      "loss": 1.4556,
+      "step": 5610
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1e-05,
+      "loss": 1.4626,
+      "step": 5620
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1e-05,
+      "loss": 1.4611,
+      "step": 5630
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1e-05,
+      "loss": 1.4739,
+      "step": 5640
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1e-05,
+      "loss": 1.4599,
+      "step": 5650
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1e-05,
+      "loss": 1.4639,
+      "step": 5660
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1e-05,
+      "loss": 1.4537,
+      "step": 5670
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1e-05,
+      "loss": 1.4563,
+      "step": 5680
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1e-05,
+      "loss": 1.4612,
+      "step": 5690
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1e-05,
+      "loss": 1.4663,
+      "step": 5700
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1e-05,
+      "loss": 1.4564,
+      "step": 5710
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1e-05,
+      "loss": 1.4523,
+      "step": 5720
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1e-05,
+      "loss": 1.4774,
+      "step": 5730
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1e-05,
+      "loss": 1.4514,
+      "step": 5740
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1e-05,
+      "loss": 1.4627,
+      "step": 5750
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1e-05,
+      "loss": 1.4512,
+      "step": 5760
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1e-05,
+      "loss": 1.4629,
+      "step": 5770
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1e-05,
+      "loss": 1.4642,
+      "step": 5780
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1e-05,
+      "loss": 1.4602,
+      "step": 5790
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1e-05,
+      "loss": 1.4664,
+      "step": 5800
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1e-05,
+      "loss": 1.4642,
+      "step": 5810
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1e-05,
+      "loss": 1.4638,
+      "step": 5820
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1e-05,
+      "loss": 1.4537,
+      "step": 5830
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1e-05,
+      "loss": 1.4565,
+      "step": 5840
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1e-05,
+      "loss": 1.4523,
+      "step": 5850
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1e-05,
+      "loss": 1.4508,
+      "step": 5860
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1e-05,
+      "loss": 1.4555,
+      "step": 5870
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1e-05,
+      "loss": 1.4592,
+      "step": 5880
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1e-05,
+      "loss": 1.4515,
+      "step": 5890
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1e-05,
+      "loss": 1.4689,
+      "step": 5900
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1e-05,
+      "loss": 1.4648,
+      "step": 5910
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1e-05,
+      "loss": 1.458,
+      "step": 5920
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1e-05,
+      "loss": 1.4539,
+      "step": 5930
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1e-05,
+      "loss": 1.4495,
+      "step": 5940
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1e-05,
+      "loss": 1.465,
+      "step": 5950
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1e-05,
+      "loss": 1.4659,
+      "step": 5960
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1e-05,
+      "loss": 1.4494,
+      "step": 5970
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1e-05,
+      "loss": 1.453,
+      "step": 5980
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1e-05,
+      "loss": 1.45,
+      "step": 5990
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1e-05,
+      "loss": 1.4568,
+      "step": 6000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1e-05,
+      "loss": 1.4639,
+      "step": 6010
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1e-05,
+      "loss": 1.449,
+      "step": 6020
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1e-05,
+      "loss": 1.4586,
+      "step": 6030
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1e-05,
+      "loss": 1.449,
+      "step": 6040
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1e-05,
+      "loss": 1.457,
+      "step": 6050
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1e-05,
+      "loss": 1.4501,
+      "step": 6060
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1e-05,
+      "loss": 1.4494,
+      "step": 6070
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1e-05,
+      "loss": 1.4821,
+      "step": 6080
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1e-05,
+      "loss": 1.4693,
+      "step": 6090
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1e-05,
+      "loss": 1.4477,
+      "step": 6100
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1e-05,
+      "loss": 1.4633,
+      "step": 6110
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1e-05,
+      "loss": 1.4571,
+      "step": 6120
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1e-05,
+      "loss": 1.4721,
+      "step": 6130
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1e-05,
+      "loss": 1.4505,
+      "step": 6140
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1e-05,
+      "loss": 1.4462,
+      "step": 6150
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1e-05,
+      "loss": 1.4586,
+      "step": 6160
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1e-05,
+      "loss": 1.4491,
+      "step": 6170
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1e-05,
+      "loss": 1.4635,
+      "step": 6180
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1e-05,
+      "loss": 1.4448,
+      "step": 6190
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1e-05,
+      "loss": 1.4472,
+      "step": 6200
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1e-05,
+      "loss": 1.4522,
+      "step": 6210
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1e-05,
+      "loss": 1.4733,
+      "step": 6220
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1e-05,
+      "loss": 1.4585,
+      "step": 6230
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1e-05,
+      "loss": 1.4465,
+      "step": 6240
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1e-05,
+      "loss": 1.4638,
+      "step": 6250
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1e-05,
+      "loss": 1.4488,
+      "step": 6260
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1e-05,
+      "loss": 1.4497,
+      "step": 6270
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1e-05,
+      "loss": 1.4574,
+      "step": 6280
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1e-05,
+      "loss": 1.4512,
+      "step": 6290
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1e-05,
+      "loss": 1.4544,
+      "step": 6300
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1e-05,
+      "loss": 1.4583,
+      "step": 6310
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1e-05,
+      "loss": 1.4556,
+      "step": 6320
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1e-05,
+      "loss": 1.4475,
+      "step": 6330
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1e-05,
+      "loss": 1.4462,
+      "step": 6340
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1e-05,
+      "loss": 1.4468,
+      "step": 6350
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1e-05,
+      "loss": 1.464,
+      "step": 6360
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1e-05,
+      "loss": 1.4508,
+      "step": 6370
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1e-05,
+      "loss": 1.4548,
+      "step": 6380
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1e-05,
+      "loss": 1.4553,
+      "step": 6390
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1e-05,
+      "loss": 1.4522,
+      "step": 6400
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1e-05,
+      "loss": 1.4527,
+      "step": 6410
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1e-05,
+      "loss": 1.4602,
+      "step": 6420
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1e-05,
+      "loss": 1.4537,
+      "step": 6430
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1e-05,
+      "loss": 1.4505,
+      "step": 6440
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1e-05,
+      "loss": 1.4549,
+      "step": 6450
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1e-05,
+      "loss": 1.4499,
+      "step": 6460
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1e-05,
+      "loss": 1.4453,
+      "step": 6470
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1e-05,
+      "loss": 1.4602,
+      "step": 6480
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1e-05,
+      "loss": 1.4512,
+      "step": 6490
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1e-05,
+      "loss": 1.4575,
+      "step": 6500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1e-05,
+      "loss": 1.4443,
+      "step": 6510
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1e-05,
+      "loss": 1.4569,
+      "step": 6520
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1e-05,
+      "loss": 1.448,
+      "step": 6530
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1e-05,
+      "loss": 1.4603,
+      "step": 6540
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1e-05,
+      "loss": 1.4558,
+      "step": 6550
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1e-05,
+      "loss": 1.463,
+      "step": 6560
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1e-05,
+      "loss": 1.4477,
+      "step": 6570
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1e-05,
+      "loss": 1.4502,
+      "step": 6580
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1e-05,
+      "loss": 1.4489,
+      "step": 6590
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1e-05,
+      "loss": 1.448,
+      "step": 6600
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1e-05,
+      "loss": 1.4444,
+      "step": 6610
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1e-05,
+      "loss": 1.4472,
+      "step": 6620
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1e-05,
+      "loss": 1.4482,
+      "step": 6630
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1e-05,
+      "loss": 1.4458,
+      "step": 6640
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1e-05,
+      "loss": 1.4584,
+      "step": 6650
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1e-05,
+      "loss": 1.4628,
+      "step": 6660
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1e-05,
+      "loss": 1.4557,
+      "step": 6670
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1e-05,
+      "loss": 1.4504,
+      "step": 6680
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1e-05,
+      "loss": 1.4644,
+      "step": 6690
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1e-05,
+      "loss": 1.4547,
+      "step": 6700
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1e-05,
+      "loss": 1.4521,
+      "step": 6710
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1e-05,
+      "loss": 1.4499,
+      "step": 6720
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1e-05,
+      "loss": 1.4533,
+      "step": 6730
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1e-05,
+      "loss": 1.4572,
+      "step": 6740
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1e-05,
+      "loss": 1.4442,
+      "step": 6750
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1e-05,
+      "loss": 1.4526,
+      "step": 6760
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1e-05,
+      "loss": 1.4507,
+      "step": 6770
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1e-05,
+      "loss": 1.4567,
+      "step": 6780
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1e-05,
+      "loss": 1.4408,
+      "step": 6790
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1e-05,
+      "loss": 1.4541,
+      "step": 6800
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1e-05,
+      "loss": 1.4469,
+      "step": 6810
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1e-05,
+      "loss": 1.4482,
+      "step": 6820
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1e-05,
+      "loss": 1.4528,
+      "step": 6830
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1e-05,
+      "loss": 1.4591,
+      "step": 6840
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1e-05,
+      "loss": 1.4537,
+      "step": 6850
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1e-05,
+      "loss": 1.4497,
+      "step": 6860
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1e-05,
+      "loss": 1.4475,
+      "step": 6870
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1e-05,
+      "loss": 1.451,
+      "step": 6880
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1e-05,
+      "loss": 1.4578,
+      "step": 6890
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1e-05,
+      "loss": 1.4415,
+      "step": 6900
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1e-05,
+      "loss": 1.4551,
+      "step": 6910
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1e-05,
+      "loss": 1.4501,
+      "step": 6920
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1e-05,
+      "loss": 1.4475,
+      "step": 6930
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1e-05,
+      "loss": 1.451,
+      "step": 6940
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1e-05,
+      "loss": 1.4684,
+      "step": 6950
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1e-05,
+      "loss": 1.4556,
+      "step": 6960
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1e-05,
+      "loss": 1.4688,
+      "step": 6970
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1e-05,
+      "loss": 1.4557,
+      "step": 6980
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1e-05,
+      "loss": 1.4537,
+      "step": 6990
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1e-05,
+      "loss": 1.4593,
+      "step": 7000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1e-05,
+      "loss": 1.4705,
+      "step": 7010
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1e-05,
+      "loss": 1.4614,
+      "step": 7020
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1e-05,
+      "loss": 1.4405,
+      "step": 7030
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1e-05,
+      "loss": 1.4531,
+      "step": 7040
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1e-05,
+      "loss": 1.4501,
+      "step": 7050
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1e-05,
+      "loss": 1.4558,
+      "step": 7060
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1e-05,
+      "loss": 1.4569,
+      "step": 7070
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1e-05,
+      "loss": 1.4567,
+      "step": 7080
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1e-05,
+      "loss": 1.4478,
+      "step": 7090
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1e-05,
+      "loss": 1.4526,
+      "step": 7100
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1e-05,
+      "loss": 1.4474,
+      "step": 7110
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1e-05,
+      "loss": 1.4493,
+      "step": 7120
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1e-05,
+      "loss": 1.4547,
+      "step": 7130
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1e-05,
+      "loss": 1.4543,
+      "step": 7140
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1e-05,
+      "loss": 1.4505,
+      "step": 7150
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1e-05,
+      "loss": 1.4525,
+      "step": 7160
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1e-05,
+      "loss": 1.4532,
+      "step": 7170
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1e-05,
+      "loss": 1.4479,
+      "step": 7180
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1e-05,
+      "loss": 1.4426,
+      "step": 7190
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1e-05,
+      "loss": 1.4539,
+      "step": 7200
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1e-05,
+      "loss": 1.4566,
+      "step": 7210
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1e-05,
+      "loss": 1.4499,
+      "step": 7220
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1e-05,
+      "loss": 1.445,
+      "step": 7230
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1e-05,
+      "loss": 1.4395,
+      "step": 7240
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1e-05,
+      "loss": 1.4486,
+      "step": 7250
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1e-05,
+      "loss": 1.4454,
+      "step": 7260
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1e-05,
+      "loss": 1.4505,
+      "step": 7270
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1e-05,
+      "loss": 1.4463,
+      "step": 7280
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1e-05,
+      "loss": 1.4406,
+      "step": 7290
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1e-05,
+      "loss": 1.4474,
+      "step": 7300
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1e-05,
+      "loss": 1.4562,
+      "step": 7310
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1e-05,
+      "loss": 1.4549,
+      "step": 7320
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1e-05,
+      "loss": 1.4443,
+      "step": 7330
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1e-05,
+      "loss": 1.4483,
+      "step": 7340
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1e-05,
+      "loss": 1.4502,
+      "step": 7350
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1e-05,
+      "loss": 1.4539,
+      "step": 7360
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1e-05,
+      "loss": 1.4421,
+      "step": 7370
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1e-05,
+      "loss": 1.4532,
+      "step": 7380
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1e-05,
+      "loss": 1.4516,
+      "step": 7390
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1e-05,
+      "loss": 1.4567,
+      "step": 7400
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1e-05,
+      "loss": 1.4469,
+      "step": 7410
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1e-05,
+      "loss": 1.4443,
+      "step": 7420
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1e-05,
+      "loss": 1.449,
+      "step": 7430
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1e-05,
+      "loss": 1.4554,
+      "step": 7440
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1e-05,
+      "loss": 1.4566,
+      "step": 7450
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1e-05,
+      "loss": 1.4508,
+      "step": 7460
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1e-05,
+      "loss": 1.4478,
+      "step": 7470
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1e-05,
+      "loss": 1.4531,
+      "step": 7480
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1e-05,
+      "loss": 1.4468,
+      "step": 7490
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1e-05,
+      "loss": 1.4426,
+      "step": 7500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1e-05,
+      "loss": 1.4586,
+      "step": 7510
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1e-05,
+      "loss": 1.4406,
+      "step": 7520
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1e-05,
+      "loss": 1.4466,
+      "step": 7530
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1e-05,
+      "loss": 1.451,
+      "step": 7540
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1e-05,
+      "loss": 1.444,
+      "step": 7550
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1e-05,
+      "loss": 1.4467,
+      "step": 7560
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1e-05,
+      "loss": 1.4501,
+      "step": 7570
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1e-05,
+      "loss": 1.449,
+      "step": 7580
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1e-05,
+      "loss": 1.4501,
+      "step": 7590
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1e-05,
+      "loss": 1.4383,
+      "step": 7600
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1e-05,
+      "loss": 1.4545,
+      "step": 7610
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1e-05,
+      "loss": 1.4452,
+      "step": 7620
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1e-05,
+      "loss": 1.4431,
+      "step": 7630
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1e-05,
+      "loss": 1.4506,
+      "step": 7640
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1e-05,
+      "loss": 1.4454,
+      "step": 7650
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1e-05,
+      "loss": 1.4491,
+      "step": 7660
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1e-05,
+      "loss": 1.4532,
+      "step": 7670
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1e-05,
+      "loss": 1.4514,
+      "step": 7680
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1e-05,
+      "loss": 1.4492,
+      "step": 7690
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1e-05,
+      "loss": 1.4531,
+      "step": 7700
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1e-05,
+      "loss": 1.4376,
+      "step": 7710
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1e-05,
+      "loss": 1.4526,
+      "step": 7720
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1e-05,
+      "loss": 1.4385,
+      "step": 7730
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1e-05,
+      "loss": 1.4469,
+      "step": 7740
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1e-05,
+      "loss": 1.4385,
+      "step": 7750
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1e-05,
+      "loss": 1.4452,
+      "step": 7760
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1e-05,
+      "loss": 1.4586,
+      "step": 7770
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1e-05,
+      "loss": 1.4489,
+      "step": 7780
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1e-05,
+      "loss": 1.4446,
+      "step": 7790
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1e-05,
+      "loss": 1.439,
+      "step": 7800
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1e-05,
+      "loss": 1.4538,
+      "step": 7810
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1e-05,
+      "loss": 1.4466,
+      "step": 7820
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1e-05,
+      "loss": 1.4464,
+      "step": 7830
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1e-05,
+      "loss": 1.4427,
+      "step": 7840
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1e-05,
+      "loss": 1.4535,
+      "step": 7850
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1e-05,
+      "loss": 1.4437,
+      "step": 7860
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1e-05,
+      "loss": 1.4378,
+      "step": 7870
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1e-05,
+      "loss": 1.4448,
+      "step": 7880
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1e-05,
+      "loss": 1.4419,
+      "step": 7890
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1e-05,
+      "loss": 1.4438,
+      "step": 7900
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1e-05,
+      "loss": 1.4387,
+      "step": 7910
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1e-05,
+      "loss": 1.4395,
+      "step": 7920
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1e-05,
+      "loss": 1.452,
+      "step": 7930
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1e-05,
+      "loss": 1.4531,
+      "step": 7940
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1e-05,
+      "loss": 1.445,
+      "step": 7950
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1e-05,
+      "loss": 1.4437,
+      "step": 7960
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1e-05,
+      "loss": 1.4479,
+      "step": 7970
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1e-05,
+      "loss": 1.4526,
+      "step": 7980
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1e-05,
+      "loss": 1.4526,
+      "step": 7990
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1e-05,
+      "loss": 1.4429,
+      "step": 8000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1e-05,
+      "loss": 1.44,
+      "step": 8010
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1e-05,
+      "loss": 1.4446,
+      "step": 8020
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1e-05,
+      "loss": 1.4461,
+      "step": 8030
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1e-05,
+      "loss": 1.4404,
+      "step": 8040
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1e-05,
+      "loss": 1.4485,
+      "step": 8050
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1e-05,
+      "loss": 1.4486,
+      "step": 8060
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1e-05,
+      "loss": 1.448,
+      "step": 8070
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1e-05,
+      "loss": 1.4399,
+      "step": 8080
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1e-05,
+      "loss": 1.44,
+      "step": 8090
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1e-05,
+      "loss": 1.4449,
+      "step": 8100
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1e-05,
+      "loss": 1.4369,
+      "step": 8110
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1e-05,
+      "loss": 1.4408,
+      "step": 8120
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1e-05,
+      "loss": 1.4461,
+      "step": 8130
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1e-05,
+      "loss": 1.4554,
+      "step": 8140
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1e-05,
+      "loss": 1.4418,
+      "step": 8150
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1e-05,
+      "loss": 1.4391,
+      "step": 8160
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1e-05,
+      "loss": 1.4367,
+      "step": 8170
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1e-05,
+      "loss": 1.4381,
+      "step": 8180
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1e-05,
+      "loss": 1.4391,
+      "step": 8190
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1e-05,
+      "loss": 1.4434,
+      "step": 8200
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1e-05,
+      "loss": 1.4485,
+      "step": 8210
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1e-05,
+      "loss": 1.4412,
+      "step": 8220
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1e-05,
+      "loss": 1.4462,
+      "step": 8230
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1e-05,
+      "loss": 1.4419,
+      "step": 8240
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1e-05,
+      "loss": 1.4479,
+      "step": 8250
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1e-05,
+      "loss": 1.4384,
+      "step": 8260
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1e-05,
+      "loss": 1.4423,
+      "step": 8270
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1e-05,
+      "loss": 1.4444,
+      "step": 8280
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1e-05,
+      "loss": 1.4374,
+      "step": 8290
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1e-05,
+      "loss": 1.4462,
+      "step": 8300
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1e-05,
+      "loss": 1.4347,
+      "step": 8310
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1e-05,
+      "loss": 1.4482,
+      "step": 8320
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1e-05,
+      "loss": 1.4457,
+      "step": 8330
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1e-05,
+      "loss": 1.4493,
+      "step": 8340
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1e-05,
+      "loss": 1.4358,
+      "step": 8350
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1e-05,
+      "loss": 1.4402,
+      "step": 8360
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1e-05,
+      "loss": 1.443,
+      "step": 8370
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1e-05,
+      "loss": 1.4362,
+      "step": 8380
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1e-05,
+      "loss": 1.4414,
+      "step": 8390
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1e-05,
+      "loss": 1.4415,
+      "step": 8400
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1e-05,
+      "loss": 1.4382,
+      "step": 8410
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1e-05,
+      "loss": 1.4446,
+      "step": 8420
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1e-05,
+      "loss": 1.4326,
+      "step": 8430
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1e-05,
+      "loss": 1.4455,
+      "step": 8440
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1e-05,
+      "loss": 1.4435,
+      "step": 8450
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1e-05,
+      "loss": 1.4386,
+      "step": 8460
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1e-05,
+      "loss": 1.4387,
+      "step": 8470
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1e-05,
+      "loss": 1.4395,
+      "step": 8480
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1e-05,
+      "loss": 1.4412,
+      "step": 8490
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1e-05,
+      "loss": 1.44,
+      "step": 8500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1e-05,
+      "loss": 1.4366,
+      "step": 8510
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1e-05,
+      "loss": 1.4376,
+      "step": 8520
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1e-05,
+      "loss": 1.4385,
+      "step": 8530
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1e-05,
+      "loss": 1.4432,
+      "step": 8540
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1e-05,
+      "loss": 1.4344,
+      "step": 8550
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1e-05,
+      "loss": 1.4466,
+      "step": 8560
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1e-05,
+      "loss": 1.4393,
+      "step": 8570
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1e-05,
+      "loss": 1.4348,
+      "step": 8580
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1e-05,
+      "loss": 1.4447,
+      "step": 8590
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1e-05,
+      "loss": 1.4392,
+      "step": 8600
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1e-05,
+      "loss": 1.4412,
+      "step": 8610
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1e-05,
+      "loss": 1.4345,
+      "step": 8620
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1e-05,
+      "loss": 1.4448,
+      "step": 8630
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1e-05,
+      "loss": 1.4396,
+      "step": 8640
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1e-05,
+      "loss": 1.4334,
+      "step": 8650
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1e-05,
+      "loss": 1.4356,
+      "step": 8660
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1e-05,
+      "loss": 1.4403,
+      "step": 8670
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1e-05,
+      "loss": 1.4378,
+      "step": 8680
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1e-05,
+      "loss": 1.4445,
+      "step": 8690
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1e-05,
+      "loss": 1.4332,
+      "step": 8700
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1e-05,
+      "loss": 1.442,
+      "step": 8710
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1e-05,
+      "loss": 1.4352,
+      "step": 8720
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1e-05,
+      "loss": 1.4373,
+      "step": 8730
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1e-05,
+      "loss": 1.4422,
+      "step": 8740
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1e-05,
+      "loss": 1.4413,
+      "step": 8750
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1e-05,
+      "loss": 1.4457,
+      "step": 8760
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1e-05,
+      "loss": 1.4528,
+      "step": 8770
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1e-05,
+      "loss": 1.4352,
+      "step": 8780
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1e-05,
+      "loss": 1.4475,
+      "step": 8790
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1e-05,
+      "loss": 1.4497,
+      "step": 8800
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1e-05,
+      "loss": 1.4368,
+      "step": 8810
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1e-05,
+      "loss": 1.4376,
+      "step": 8820
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1e-05,
+      "loss": 1.4431,
+      "step": 8830
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1e-05,
+      "loss": 1.4338,
+      "step": 8840
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1e-05,
+      "loss": 1.4449,
+      "step": 8850
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1e-05,
+      "loss": 1.4395,
+      "step": 8860
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1e-05,
+      "loss": 1.4432,
+      "step": 8870
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1e-05,
+      "loss": 1.4433,
+      "step": 8880
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1e-05,
+      "loss": 1.438,
+      "step": 8890
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1e-05,
+      "loss": 1.439,
+      "step": 8900
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1e-05,
+      "loss": 1.4357,
+      "step": 8910
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1e-05,
+      "loss": 1.4422,
+      "step": 8920
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1e-05,
+      "loss": 1.4514,
+      "step": 8930
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1e-05,
+      "loss": 1.441,
+      "step": 8940
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1e-05,
+      "loss": 1.4407,
+      "step": 8950
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1e-05,
+      "loss": 1.4381,
+      "step": 8960
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1e-05,
+      "loss": 1.4456,
+      "step": 8970
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1e-05,
+      "loss": 1.4426,
+      "step": 8980
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1e-05,
+      "loss": 1.4454,
+      "step": 8990
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1e-05,
+      "loss": 1.4401,
+      "step": 9000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1e-05,
+      "loss": 1.4458,
+      "step": 9010
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1e-05,
+      "loss": 1.4446,
+      "step": 9020
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1e-05,
+      "loss": 1.4391,
+      "step": 9030
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1e-05,
+      "loss": 1.4433,
+      "step": 9040
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1e-05,
+      "loss": 1.4411,
+      "step": 9050
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1e-05,
+      "loss": 1.4397,
+      "step": 9060
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1e-05,
+      "loss": 1.4378,
+      "step": 9070
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1e-05,
+      "loss": 1.4397,
+      "step": 9080
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1e-05,
+      "loss": 1.4421,
+      "step": 9090
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1e-05,
+      "loss": 1.4418,
+      "step": 9100
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1e-05,
+      "loss": 1.4341,
+      "step": 9110
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1e-05,
+      "loss": 1.447,
+      "step": 9120
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1e-05,
+      "loss": 1.437,
+      "step": 9130
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1e-05,
+      "loss": 1.4387,
+      "step": 9140
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1e-05,
+      "loss": 1.4347,
+      "step": 9150
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1e-05,
+      "loss": 1.4437,
+      "step": 9160
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1e-05,
+      "loss": 1.4309,
+      "step": 9170
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1e-05,
+      "loss": 1.439,
+      "step": 9180
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1e-05,
+      "loss": 1.4349,
+      "step": 9190
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1e-05,
+      "loss": 1.4361,
+      "step": 9200
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1e-05,
+      "loss": 1.4505,
+      "step": 9210
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1e-05,
+      "loss": 1.4386,
+      "step": 9220
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1e-05,
+      "loss": 1.4343,
+      "step": 9230
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1e-05,
+      "loss": 1.4368,
+      "step": 9240
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1e-05,
+      "loss": 1.4329,
+      "step": 9250
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1e-05,
+      "loss": 1.4368,
+      "step": 9260
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1e-05,
+      "loss": 1.4418,
+      "step": 9270
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1e-05,
+      "loss": 1.4349,
+      "step": 9280
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1e-05,
+      "loss": 1.4426,
+      "step": 9290
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1e-05,
+      "loss": 1.445,
+      "step": 9300
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1e-05,
+      "loss": 1.4339,
+      "step": 9310
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1e-05,
+      "loss": 1.4394,
+      "step": 9320
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1e-05,
+      "loss": 1.4421,
+      "step": 9330
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1e-05,
+      "loss": 1.4448,
+      "step": 9340
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1e-05,
+      "loss": 1.4355,
+      "step": 9350
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1e-05,
+      "loss": 1.4426,
+      "step": 9360
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1e-05,
+      "loss": 1.4441,
+      "step": 9370
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1e-05,
+      "loss": 1.4389,
+      "step": 9380
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1e-05,
+      "loss": 1.4393,
+      "step": 9390
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1e-05,
+      "loss": 1.4367,
+      "step": 9400
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1e-05,
+      "loss": 1.4373,
+      "step": 9410
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1e-05,
+      "loss": 1.4367,
+      "step": 9420
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1e-05,
+      "loss": 1.452,
+      "step": 9430
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1e-05,
+      "loss": 1.438,
+      "step": 9440
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1e-05,
+      "loss": 1.4311,
+      "step": 9450
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1e-05,
+      "loss": 1.4384,
+      "step": 9460
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1e-05,
+      "loss": 1.4374,
+      "step": 9470
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1e-05,
+      "loss": 1.4331,
+      "step": 9480
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1e-05,
+      "loss": 1.4535,
+      "step": 9490
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1e-05,
+      "loss": 1.439,
+      "step": 9500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1e-05,
+      "loss": 1.4377,
+      "step": 9510
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1e-05,
+      "loss": 1.4345,
+      "step": 9520
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1e-05,
+      "loss": 1.4498,
+      "step": 9530
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1e-05,
+      "loss": 1.439,
+      "step": 9540
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1e-05,
+      "loss": 1.4389,
+      "step": 9550
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1e-05,
+      "loss": 1.442,
+      "step": 9560
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1e-05,
+      "loss": 1.4384,
+      "step": 9570
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1e-05,
+      "loss": 1.4315,
+      "step": 9580
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1e-05,
+      "loss": 1.434,
+      "step": 9590
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1e-05,
+      "loss": 1.4388,
+      "step": 9600
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1e-05,
+      "loss": 1.4389,
+      "step": 9610
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1e-05,
+      "loss": 1.4383,
+      "step": 9620
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1e-05,
+      "loss": 1.4369,
+      "step": 9630
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1e-05,
+      "loss": 1.4337,
+      "step": 9640
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1e-05,
+      "loss": 1.4363,
+      "step": 9650
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1e-05,
+      "loss": 1.4364,
+      "step": 9660
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1e-05,
+      "loss": 1.4385,
+      "step": 9670
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1e-05,
+      "loss": 1.4367,
+      "step": 9680
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1e-05,
+      "loss": 1.4452,
+      "step": 9690
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1e-05,
+      "loss": 1.4316,
+      "step": 9700
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1e-05,
+      "loss": 1.4325,
+      "step": 9710
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1e-05,
+      "loss": 1.433,
+      "step": 9720
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1e-05,
+      "loss": 1.438,
+      "step": 9730
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1e-05,
+      "loss": 1.4377,
+      "step": 9740
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1e-05,
+      "loss": 1.4326,
+      "step": 9750
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1e-05,
+      "loss": 1.4325,
+      "step": 9760
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1e-05,
+      "loss": 1.4451,
+      "step": 9770
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1e-05,
+      "loss": 1.4381,
+      "step": 9780
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1e-05,
+      "loss": 1.434,
+      "step": 9790
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1e-05,
+      "loss": 1.435,
+      "step": 9800
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1e-05,
+      "loss": 1.4371,
+      "step": 9810
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1e-05,
+      "loss": 1.4453,
+      "step": 9820
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1e-05,
+      "loss": 1.4297,
+      "step": 9830
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1e-05,
+      "loss": 1.4314,
+      "step": 9840
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1e-05,
+      "loss": 1.4378,
+      "step": 9850
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1e-05,
+      "loss": 1.4244,
+      "step": 9860
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1e-05,
+      "loss": 1.4379,
+      "step": 9870
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1e-05,
+      "loss": 1.4311,
+      "step": 9880
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1e-05,
+      "loss": 1.4369,
+      "step": 9890
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1e-05,
+      "loss": 1.4376,
+      "step": 9900
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1e-05,
+      "loss": 1.4423,
+      "step": 9910
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1e-05,
+      "loss": 1.4378,
+      "step": 9920
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1e-05,
+      "loss": 1.4426,
+      "step": 9930
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1e-05,
+      "loss": 1.434,
+      "step": 9940
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1e-05,
+      "loss": 1.4293,
+      "step": 9950
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1e-05,
+      "loss": 1.4356,
+      "step": 9960
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1e-05,
+      "loss": 1.4339,
+      "step": 9970
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1e-05,
+      "loss": 1.4335,
+      "step": 9980
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1e-05,
+      "loss": 1.4308,
+      "step": 9990
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1e-05,
+      "loss": 1.4274,
+      "step": 10000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1e-05,
+      "loss": 1.4452,
+      "step": 10010
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1e-05,
+      "loss": 1.4445,
+      "step": 10020
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1e-05,
+      "loss": 1.4313,
+      "step": 10030
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1e-05,
+      "loss": 1.4376,
+      "step": 10040
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1e-05,
+      "loss": 1.4387,
+      "step": 10050
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1e-05,
+      "loss": 1.4311,
+      "step": 10060
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1e-05,
+      "loss": 1.4368,
+      "step": 10070
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1e-05,
+      "loss": 1.4311,
+      "step": 10080
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1e-05,
+      "loss": 1.4302,
+      "step": 10090
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1e-05,
+      "loss": 1.4359,
+      "step": 10100
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1e-05,
+      "loss": 1.4354,
+      "step": 10110
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1e-05,
+      "loss": 1.4366,
+      "step": 10120
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1e-05,
+      "loss": 1.4367,
+      "step": 10130
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1e-05,
+      "loss": 1.4274,
+      "step": 10140
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1e-05,
+      "loss": 1.4336,
+      "step": 10150
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1e-05,
+      "loss": 1.4294,
+      "step": 10160
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1e-05,
+      "loss": 1.4327,
+      "step": 10170
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1e-05,
+      "loss": 1.4315,
+      "step": 10180
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1e-05,
+      "loss": 1.4318,
+      "step": 10190
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1e-05,
+      "loss": 1.431,
+      "step": 10200
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1e-05,
+      "loss": 1.4331,
+      "step": 10210
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1e-05,
+      "loss": 1.4292,
+      "step": 10220
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1e-05,
+      "loss": 1.4395,
+      "step": 10230
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1e-05,
+      "loss": 1.4317,
+      "step": 10240
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1e-05,
+      "loss": 1.4303,
+      "step": 10250
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1e-05,
+      "loss": 1.436,
+      "step": 10260
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1e-05,
+      "loss": 1.4339,
+      "step": 10270
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1e-05,
+      "loss": 1.4356,
+      "step": 10280
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1e-05,
+      "loss": 1.4322,
+      "step": 10290
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1e-05,
+      "loss": 1.4313,
+      "step": 10300
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1e-05,
+      "loss": 1.4354,
+      "step": 10310
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1e-05,
+      "loss": 1.4358,
+      "step": 10320
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1e-05,
+      "loss": 1.4293,
+      "step": 10330
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1e-05,
+      "loss": 1.4317,
+      "step": 10340
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1e-05,
+      "loss": 1.4463,
+      "step": 10350
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1e-05,
+      "loss": 1.4283,
+      "step": 10360
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1e-05,
+      "loss": 1.432,
+      "step": 10370
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1e-05,
+      "loss": 1.4332,
+      "step": 10380
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1e-05,
+      "loss": 1.4359,
+      "step": 10390
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1e-05,
+      "loss": 1.4293,
+      "step": 10400
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1e-05,
+      "loss": 1.4376,
+      "step": 10410
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1e-05,
+      "loss": 1.4451,
+      "step": 10420
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1e-05,
+      "loss": 1.4359,
+      "step": 10430
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1e-05,
+      "loss": 1.4322,
+      "step": 10440
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1e-05,
+      "loss": 1.4331,
+      "step": 10450
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1e-05,
+      "loss": 1.4306,
+      "step": 10460
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1e-05,
+      "loss": 1.4294,
+      "step": 10470
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1e-05,
+      "loss": 1.4343,
+      "step": 10480
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1e-05,
+      "loss": 1.4446,
+      "step": 10490
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1e-05,
+      "loss": 1.4369,
+      "step": 10500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1e-05,
+      "loss": 1.4402,
+      "step": 10510
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1e-05,
+      "loss": 1.4327,
+      "step": 10520
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1e-05,
+      "loss": 1.4339,
+      "step": 10530
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1e-05,
+      "loss": 1.4397,
+      "step": 10540
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1e-05,
+      "loss": 1.4308,
+      "step": 10550
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1e-05,
+      "loss": 1.433,
+      "step": 10560
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1e-05,
+      "loss": 1.4362,
+      "step": 10570
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1e-05,
+      "loss": 1.4413,
+      "step": 10580
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1e-05,
+      "loss": 1.4408,
+      "step": 10590
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1e-05,
+      "loss": 1.4318,
+      "step": 10600
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1e-05,
+      "loss": 1.4346,
+      "step": 10610
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1e-05,
+      "loss": 1.4308,
+      "step": 10620
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1e-05,
+      "loss": 1.427,
+      "step": 10630
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1e-05,
+      "loss": 1.4272,
+      "step": 10640
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1e-05,
+      "loss": 1.4334,
+      "step": 10650
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1e-05,
+      "loss": 1.4347,
+      "step": 10660
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1e-05,
+      "loss": 1.4312,
+      "step": 10670
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1e-05,
+      "loss": 1.4448,
+      "step": 10680
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1e-05,
+      "loss": 1.4313,
+      "step": 10690
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1e-05,
+      "loss": 1.4305,
+      "step": 10700
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1e-05,
+      "loss": 1.4329,
+      "step": 10710
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1e-05,
+      "loss": 1.4363,
+      "step": 10720
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1e-05,
+      "loss": 1.4303,
+      "step": 10730
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1e-05,
+      "loss": 1.4329,
+      "step": 10740
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1e-05,
+      "loss": 1.4284,
+      "step": 10750
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1e-05,
+      "loss": 1.4318,
+      "step": 10760
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1e-05,
+      "loss": 1.4344,
+      "step": 10770
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1e-05,
+      "loss": 1.4343,
+      "step": 10780
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1e-05,
+      "loss": 1.4294,
+      "step": 10790
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1e-05,
+      "loss": 1.4366,
+      "step": 10800
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1e-05,
+      "loss": 1.4425,
+      "step": 10810
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1e-05,
+      "loss": 1.4352,
+      "step": 10820
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1e-05,
+      "loss": 1.432,
+      "step": 10830
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1e-05,
+      "loss": 1.431,
+      "step": 10840
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1e-05,
+      "loss": 1.4335,
+      "step": 10850
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1e-05,
+      "loss": 1.4403,
+      "step": 10860
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1e-05,
+      "loss": 1.4397,
+      "step": 10870
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1e-05,
+      "loss": 1.4278,
+      "step": 10880
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1e-05,
+      "loss": 1.4312,
+      "step": 10890
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1e-05,
+      "loss": 1.4338,
+      "step": 10900
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1e-05,
+      "loss": 1.4313,
+      "step": 10910
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1e-05,
+      "loss": 1.4309,
+      "step": 10920
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1e-05,
+      "loss": 1.43,
+      "step": 10930
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1e-05,
+      "loss": 1.4376,
+      "step": 10940
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1e-05,
+      "loss": 1.4339,
+      "step": 10950
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1e-05,
+      "loss": 1.4311,
+      "step": 10960
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1e-05,
+      "loss": 1.4319,
+      "step": 10970
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1e-05,
+      "loss": 1.4313,
+      "step": 10980
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1e-05,
+      "loss": 1.4311,
+      "step": 10990
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1e-05,
+      "loss": 1.4353,
+      "step": 11000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1e-05,
+      "loss": 1.4301,
+      "step": 11010
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1e-05,
+      "loss": 1.4416,
+      "step": 11020
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1e-05,
+      "loss": 1.426,
+      "step": 11030
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1e-05,
+      "loss": 1.4309,
+      "step": 11040
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1e-05,
+      "loss": 1.4327,
+      "step": 11050
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1e-05,
+      "loss": 1.4334,
+      "step": 11060
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1e-05,
+      "loss": 1.4336,
+      "step": 11070
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1e-05,
+      "loss": 1.4266,
+      "step": 11080
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1e-05,
+      "loss": 1.4303,
+      "step": 11090
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1e-05,
+      "loss": 1.4333,
+      "step": 11100
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1e-05,
+      "loss": 1.431,
+      "step": 11110
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1e-05,
+      "loss": 1.4279,
+      "step": 11120
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1e-05,
+      "loss": 1.4295,
+      "step": 11130
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1e-05,
+      "loss": 1.4357,
+      "step": 11140
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1e-05,
+      "loss": 1.4292,
+      "step": 11150
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1e-05,
+      "loss": 1.4299,
+      "step": 11160
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1e-05,
+      "loss": 1.43,
+      "step": 11170
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1e-05,
+      "loss": 1.4293,
+      "step": 11180
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1e-05,
+      "loss": 1.435,
+      "step": 11190
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1e-05,
+      "loss": 1.4285,
+      "step": 11200
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1e-05,
+      "loss": 1.431,
+      "step": 11210
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1e-05,
+      "loss": 1.4331,
+      "step": 11220
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1e-05,
+      "loss": 1.4315,
+      "step": 11230
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1e-05,
+      "loss": 1.4274,
+      "step": 11240
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1e-05,
+      "loss": 1.4429,
+      "step": 11250
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1e-05,
+      "loss": 1.4252,
+      "step": 11260
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1e-05,
+      "loss": 1.4278,
+      "step": 11270
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1e-05,
+      "loss": 1.4296,
+      "step": 11280
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1e-05,
+      "loss": 1.431,
+      "step": 11290
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1e-05,
+      "loss": 1.4282,
+      "step": 11300
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1e-05,
+      "loss": 1.4262,
+      "step": 11310
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1e-05,
+      "loss": 1.4251,
+      "step": 11320
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1e-05,
+      "loss": 1.4276,
+      "step": 11330
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1e-05,
+      "loss": 1.4282,
+      "step": 11340
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1e-05,
+      "loss": 1.4293,
+      "step": 11350
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1e-05,
+      "loss": 1.4261,
+      "step": 11360
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1e-05,
+      "loss": 1.4314,
+      "step": 11370
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1e-05,
+      "loss": 1.4295,
+      "step": 11380
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1e-05,
+      "loss": 1.4362,
+      "step": 11390
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1e-05,
+      "loss": 1.4271,
+      "step": 11400
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1e-05,
+      "loss": 1.4348,
+      "step": 11410
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1e-05,
+      "loss": 1.4384,
+      "step": 11420
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1e-05,
+      "loss": 1.4272,
+      "step": 11430
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1e-05,
+      "loss": 1.4317,
+      "step": 11440
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1e-05,
+      "loss": 1.4296,
+      "step": 11450
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1e-05,
+      "loss": 1.4374,
+      "step": 11460
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1e-05,
+      "loss": 1.4295,
+      "step": 11470
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1e-05,
+      "loss": 1.4298,
+      "step": 11480
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1e-05,
+      "loss": 1.4318,
+      "step": 11490
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1e-05,
+      "loss": 1.4309,
+      "step": 11500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1e-05,
+      "loss": 1.4325,
+      "step": 11510
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1e-05,
+      "loss": 1.4288,
+      "step": 11520
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1e-05,
+      "loss": 1.4259,
+      "step": 11530
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1e-05,
+      "loss": 1.4337,
+      "step": 11540
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1e-05,
+      "loss": 1.4328,
+      "step": 11550
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1e-05,
+      "loss": 1.437,
+      "step": 11560
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1e-05,
+      "loss": 1.4298,
+      "step": 11570
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1e-05,
+      "loss": 1.4413,
+      "step": 11580
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1e-05,
+      "loss": 1.4233,
+      "step": 11590
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1e-05,
+      "loss": 1.4288,
+      "step": 11600
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1e-05,
+      "loss": 1.4309,
+      "step": 11610
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1e-05,
+      "loss": 1.4303,
+      "step": 11620
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1e-05,
+      "loss": 1.4307,
+      "step": 11630
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1e-05,
+      "loss": 1.4291,
+      "step": 11640
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1e-05,
+      "loss": 1.4378,
+      "step": 11650
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1e-05,
+      "loss": 1.4219,
+      "step": 11660
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1e-05,
+      "loss": 1.4276,
+      "step": 11670
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1e-05,
+      "loss": 1.4304,
+      "step": 11680
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1e-05,
+      "loss": 1.4278,
+      "step": 11690
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1e-05,
+      "loss": 1.4331,
+      "step": 11700
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1e-05,
+      "loss": 1.4271,
+      "step": 11710
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1e-05,
+      "loss": 1.4352,
+      "step": 11720
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1e-05,
+      "loss": 1.4361,
+      "step": 11730
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1e-05,
+      "loss": 1.4316,
+      "step": 11740
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1e-05,
+      "loss": 1.4337,
+      "step": 11750
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1e-05,
+      "loss": 1.4302,
+      "step": 11760
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1e-05,
+      "loss": 1.4269,
+      "step": 11770
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1e-05,
+      "loss": 1.4283,
+      "step": 11780
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1e-05,
+      "loss": 1.4329,
+      "step": 11790
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1e-05,
+      "loss": 1.4327,
+      "step": 11800
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1e-05,
+      "loss": 1.431,
+      "step": 11810
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1e-05,
+      "loss": 1.4343,
+      "step": 11820
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1e-05,
+      "loss": 1.4262,
+      "step": 11830
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1e-05,
+      "loss": 1.4343,
+      "step": 11840
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1e-05,
+      "loss": 1.4304,
+      "step": 11850
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1e-05,
+      "loss": 1.4267,
+      "step": 11860
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1e-05,
+      "loss": 1.4255,
+      "step": 11870
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1e-05,
+      "loss": 1.4281,
+      "step": 11880
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1e-05,
+      "loss": 1.4324,
+      "step": 11890
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1e-05,
+      "loss": 1.4354,
+      "step": 11900
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1e-05,
+      "loss": 1.4259,
+      "step": 11910
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1e-05,
+      "loss": 1.4392,
+      "step": 11920
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1e-05,
+      "loss": 1.4295,
+      "step": 11930
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1e-05,
+      "loss": 1.4262,
+      "step": 11940
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1e-05,
+      "loss": 1.4269,
+      "step": 11950
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1e-05,
+      "loss": 1.4312,
+      "step": 11960
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1e-05,
+      "loss": 1.4365,
+      "step": 11970
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1e-05,
+      "loss": 1.4301,
+      "step": 11980
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1e-05,
+      "loss": 1.4248,
+      "step": 11990
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1e-05,
+      "loss": 1.4269,
+      "step": 12000
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1e-05,
+      "loss": 1.4264,
+      "step": 12010
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1e-05,
+      "loss": 1.4287,
+      "step": 12020
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1e-05,
+      "loss": 1.4275,
+      "step": 12030
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1e-05,
+      "loss": 1.4241,
+      "step": 12040
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1e-05,
+      "loss": 1.4247,
+      "step": 12050
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1e-05,
+      "loss": 1.4228,
+      "step": 12060
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1e-05,
+      "loss": 1.4271,
+      "step": 12070
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1e-05,
+      "loss": 1.4316,
+      "step": 12080
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1e-05,
+      "loss": 1.4306,
+      "step": 12090
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1e-05,
+      "loss": 1.4245,
+      "step": 12100
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1e-05,
+      "loss": 1.4253,
+      "step": 12110
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1e-05,
+      "loss": 1.4246,
+      "step": 12120
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1e-05,
+      "loss": 1.4255,
+      "step": 12130
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1e-05,
+      "loss": 1.4302,
+      "step": 12140
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1e-05,
+      "loss": 1.4327,
+      "step": 12150
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1e-05,
+      "loss": 1.4386,
+      "step": 12160
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1e-05,
+      "loss": 1.4247,
+      "step": 12170
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1e-05,
+      "loss": 1.43,
+      "step": 12180
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1e-05,
+      "loss": 1.428,
+      "step": 12190
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1e-05,
+      "loss": 1.4226,
+      "step": 12200
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1e-05,
+      "loss": 1.4269,
+      "step": 12210
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1e-05,
+      "loss": 1.4321,
+      "step": 12220
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1e-05,
+      "loss": 1.4288,
+      "step": 12230
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1e-05,
+      "loss": 1.4256,
+      "step": 12240
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1e-05,
+      "loss": 1.4315,
+      "step": 12250
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1e-05,
+      "loss": 1.4318,
+      "step": 12260
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1e-05,
+      "loss": 1.4292,
+      "step": 12270
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1e-05,
+      "loss": 1.4277,
+      "step": 12280
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1e-05,
+      "loss": 1.4288,
+      "step": 12290
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1e-05,
+      "loss": 1.43,
+      "step": 12300
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1e-05,
+      "loss": 1.4267,
+      "step": 12310
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1e-05,
+      "loss": 1.4273,
+      "step": 12320
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1e-05,
+      "loss": 1.4282,
+      "step": 12330
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1e-05,
+      "loss": 1.4281,
+      "step": 12340
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1e-05,
+      "loss": 1.4344,
+      "step": 12350
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1e-05,
+      "loss": 1.4295,
+      "step": 12360
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1e-05,
+      "loss": 1.4311,
+      "step": 12370
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1e-05,
+      "loss": 1.4312,
+      "step": 12380
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1e-05,
+      "loss": 1.4354,
+      "step": 12390
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1e-05,
+      "loss": 1.4364,
+      "step": 12400
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1e-05,
+      "loss": 1.428,
+      "step": 12410
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1e-05,
+      "loss": 1.4333,
+      "step": 12420
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1e-05,
+      "loss": 1.4263,
+      "step": 12430
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1e-05,
+      "loss": 1.4262,
+      "step": 12440
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1e-05,
+      "loss": 1.4268,
+      "step": 12450
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1e-05,
+      "loss": 1.4234,
+      "step": 12460
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1e-05,
+      "loss": 1.4282,
+      "step": 12470
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1e-05,
+      "loss": 1.4261,
+      "step": 12480
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1e-05,
+      "loss": 1.4258,
+      "step": 12490
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1e-05,
+      "loss": 1.4222,
+      "step": 12500
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1e-05,
+      "loss": 1.4327,
+      "step": 12510
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1e-05,
+      "loss": 1.4336,
+      "step": 12520
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1e-05,
+      "loss": 1.4262,
+      "step": 12530
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1e-05,
+      "loss": 1.4262,
+      "step": 12540
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1e-05,
+      "loss": 1.4245,
+      "step": 12550
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1e-05,
+      "loss": 1.431,
+      "step": 12560
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1e-05,
+      "loss": 1.4399,
+      "step": 12570
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1e-05,
+      "loss": 1.4295,
+      "step": 12580
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1e-05,
+      "loss": 1.4294,
+      "step": 12590
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1e-05,
+      "loss": 1.4273,
+      "step": 12600
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1e-05,
+      "loss": 1.4221,
+      "step": 12610
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1e-05,
+      "loss": 1.4301,
+      "step": 12620
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1e-05,
+      "loss": 1.4294,
+      "step": 12630
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1e-05,
+      "loss": 1.4283,
+      "step": 12640
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1e-05,
+      "loss": 1.429,
+      "step": 12650
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1e-05,
+      "loss": 1.4247,
+      "step": 12660
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1e-05,
+      "loss": 1.4253,
+      "step": 12670
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1e-05,
+      "loss": 1.4275,
+      "step": 12680
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1e-05,
+      "loss": 1.4302,
+      "step": 12690
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1e-05,
+      "loss": 1.4255,
+      "step": 12700
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1e-05,
+      "loss": 1.4348,
+      "step": 12710
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1e-05,
+      "loss": 1.4323,
+      "step": 12720
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1e-05,
+      "loss": 1.443,
+      "step": 12730
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1e-05,
+      "loss": 1.4214,
+      "step": 12740
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1e-05,
+      "loss": 1.4299,
+      "step": 12750
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1e-05,
+      "loss": 1.4309,
+      "step": 12760
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1e-05,
+      "loss": 1.4235,
+      "step": 12770
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1e-05,
+      "loss": 1.4322,
+      "step": 12780
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1e-05,
+      "loss": 1.4296,
+      "step": 12790
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1e-05,
+      "loss": 1.429,
+      "step": 12800
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1e-05,
+      "loss": 1.4252,
+      "step": 12810
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1e-05,
+      "loss": 1.4251,
+      "step": 12820
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1e-05,
+      "loss": 1.4299,
+      "step": 12830
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1e-05,
+      "loss": 1.4257,
+      "step": 12840
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1e-05,
+      "loss": 1.4286,
+      "step": 12850
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1e-05,
+      "loss": 1.4264,
+      "step": 12860
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1e-05,
+      "loss": 1.4257,
+      "step": 12870
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1e-05,
+      "loss": 1.4238,
+      "step": 12880
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1e-05,
+      "loss": 1.4296,
+      "step": 12890
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1e-05,
+      "loss": 1.4271,
+      "step": 12900
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1e-05,
+      "loss": 1.4261,
+      "step": 12910
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1e-05,
+      "loss": 1.4344,
+      "step": 12920
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1e-05,
+      "loss": 1.4257,
+      "step": 12930
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1e-05,
+      "loss": 1.4278,
+      "step": 12940
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1e-05,
+      "loss": 1.4268,
+      "step": 12950
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1e-05,
+      "loss": 1.427,
+      "step": 12960
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1e-05,
+      "loss": 1.4298,
+      "step": 12970
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1e-05,
+      "loss": 1.4248,
+      "step": 12980
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1e-05,
+      "loss": 1.4255,
+      "step": 12990
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1e-05,
+      "loss": 1.4277,
+      "step": 13000
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1e-05,
+      "loss": 1.4218,
+      "step": 13010
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1e-05,
+      "loss": 1.4233,
+      "step": 13020
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1e-05,
+      "loss": 1.4258,
+      "step": 13030
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1e-05,
+      "loss": 1.4265,
+      "step": 13040
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1e-05,
+      "loss": 1.4266,
+      "step": 13050
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1e-05,
+      "loss": 1.4228,
+      "step": 13060
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1e-05,
+      "loss": 1.424,
+      "step": 13070
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1e-05,
+      "loss": 1.4281,
+      "step": 13080
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1e-05,
+      "loss": 1.4225,
+      "step": 13090
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1e-05,
+      "loss": 1.4295,
+      "step": 13100
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1e-05,
+      "loss": 1.4249,
+      "step": 13110
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1e-05,
+      "loss": 1.4317,
+      "step": 13120
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1e-05,
+      "loss": 1.4226,
+      "step": 13130
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1e-05,
+      "loss": 1.4249,
+      "step": 13140
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1e-05,
+      "loss": 1.4237,
+      "step": 13150
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1e-05,
+      "loss": 1.4251,
+      "step": 13160
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1e-05,
+      "loss": 1.4262,
+      "step": 13170
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1e-05,
+      "loss": 1.4262,
+      "step": 13180
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1e-05,
+      "loss": 1.4204,
+      "step": 13190
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1e-05,
+      "loss": 1.4235,
+      "step": 13200
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1e-05,
+      "loss": 1.4304,
+      "step": 13210
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1e-05,
+      "loss": 1.4279,
+      "step": 13220
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1e-05,
+      "loss": 1.4257,
+      "step": 13230
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1e-05,
+      "loss": 1.4233,
+      "step": 13240
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1e-05,
+      "loss": 1.4268,
+      "step": 13250
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1e-05,
+      "loss": 1.429,
+      "step": 13260
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1e-05,
+      "loss": 1.4238,
+      "step": 13270
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1e-05,
+      "loss": 1.4289,
+      "step": 13280
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1e-05,
+      "loss": 1.4251,
+      "step": 13290
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1e-05,
+      "loss": 1.43,
+      "step": 13300
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1e-05,
+      "loss": 1.4238,
+      "step": 13310
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1e-05,
+      "loss": 1.4245,
+      "step": 13320
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1e-05,
+      "loss": 1.4245,
+      "step": 13330
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1e-05,
+      "loss": 1.4315,
+      "step": 13340
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1e-05,
+      "loss": 1.4275,
+      "step": 13350
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1e-05,
+      "loss": 1.4288,
+      "step": 13360
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1e-05,
+      "loss": 1.4332,
+      "step": 13370
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1e-05,
+      "loss": 1.4245,
+      "step": 13380
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1e-05,
+      "loss": 1.4237,
+      "step": 13390
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1e-05,
+      "loss": 1.4223,
+      "step": 13400
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1e-05,
+      "loss": 1.4215,
+      "step": 13410
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1e-05,
+      "loss": 1.4242,
+      "step": 13420
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1e-05,
+      "loss": 1.4279,
+      "step": 13430
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1e-05,
+      "loss": 1.4243,
+      "step": 13440
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1e-05,
+      "loss": 1.4265,
+      "step": 13450
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1e-05,
+      "loss": 1.4268,
+      "step": 13460
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1e-05,
+      "loss": 1.4233,
+      "step": 13470
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1e-05,
+      "loss": 1.4218,
+      "step": 13480
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1e-05,
+      "loss": 1.422,
+      "step": 13490
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1e-05,
+      "loss": 1.4263,
+      "step": 13500
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1e-05,
+      "loss": 1.4295,
+      "step": 13510
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1e-05,
+      "loss": 1.427,
+      "step": 13520
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1e-05,
+      "loss": 1.4257,
+      "step": 13530
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1e-05,
+      "loss": 1.4273,
+      "step": 13540
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1e-05,
+      "loss": 1.424,
+      "step": 13550
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1e-05,
+      "loss": 1.4226,
+      "step": 13560
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1e-05,
+      "loss": 1.4198,
+      "step": 13570
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1e-05,
+      "loss": 1.4297,
+      "step": 13580
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1e-05,
+      "loss": 1.4241,
+      "step": 13590
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1e-05,
+      "loss": 1.4302,
+      "step": 13600
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1e-05,
+      "loss": 1.4249,
+      "step": 13610
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1e-05,
+      "loss": 1.426,
+      "step": 13620
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1e-05,
+      "loss": 1.4226,
+      "step": 13630
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1e-05,
+      "loss": 1.4221,
+      "step": 13640
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1e-05,
+      "loss": 1.4255,
+      "step": 13650
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1e-05,
+      "loss": 1.4293,
+      "step": 13660
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1e-05,
+      "loss": 1.4267,
+      "step": 13670
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1e-05,
+      "loss": 1.421,
+      "step": 13680
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1e-05,
+      "loss": 1.4264,
+      "step": 13690
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1e-05,
+      "loss": 1.4253,
+      "step": 13700
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1e-05,
+      "loss": 1.4249,
+      "step": 13710
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1e-05,
+      "loss": 1.4307,
+      "step": 13720
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1e-05,
+      "loss": 1.4235,
+      "step": 13730
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1e-05,
+      "loss": 1.4233,
+      "step": 13740
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1e-05,
+      "loss": 1.4322,
+      "step": 13750
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1e-05,
+      "loss": 1.4276,
+      "step": 13760
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1e-05,
+      "loss": 1.4241,
+      "step": 13770
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1e-05,
+      "loss": 1.4221,
+      "step": 13780
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1e-05,
+      "loss": 1.4283,
+      "step": 13790
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1e-05,
+      "loss": 1.4255,
+      "step": 13800
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1e-05,
+      "loss": 1.423,
+      "step": 13810
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1e-05,
+      "loss": 1.4275,
+      "step": 13820
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1e-05,
+      "loss": 1.4213,
+      "step": 13830
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1e-05,
+      "loss": 1.4229,
+      "step": 13840
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1e-05,
+      "loss": 1.4228,
+      "step": 13850
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1e-05,
+      "loss": 1.4256,
+      "step": 13860
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1e-05,
+      "loss": 1.422,
+      "step": 13870
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1e-05,
+      "loss": 1.4221,
+      "step": 13880
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1e-05,
+      "loss": 1.4239,
+      "step": 13890
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1e-05,
+      "loss": 1.4237,
+      "step": 13900
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1e-05,
+      "loss": 1.4212,
+      "step": 13910
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1e-05,
+      "loss": 1.4225,
+      "step": 13920
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1e-05,
+      "loss": 1.4323,
+      "step": 13930
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1e-05,
+      "loss": 1.4249,
+      "step": 13940
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1e-05,
+      "loss": 1.431,
+      "step": 13950
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1e-05,
+      "loss": 1.4227,
+      "step": 13960
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1e-05,
+      "loss": 1.4232,
+      "step": 13970
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1e-05,
+      "loss": 1.4327,
+      "step": 13980
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1e-05,
+      "loss": 1.4241,
+      "step": 13990
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1e-05,
+      "loss": 1.4278,
+      "step": 14000
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1e-05,
+      "loss": 1.4246,
+      "step": 14010
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1e-05,
+      "loss": 1.431,
+      "step": 14020
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1e-05,
+      "loss": 1.4254,
+      "step": 14030
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1e-05,
+      "loss": 1.4266,
+      "step": 14040
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1e-05,
+      "loss": 1.4259,
+      "step": 14050
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1e-05,
+      "loss": 1.427,
+      "step": 14060
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1e-05,
+      "loss": 1.4234,
+      "step": 14070
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1e-05,
+      "loss": 1.4247,
+      "step": 14080
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1e-05,
+      "loss": 1.4202,
+      "step": 14090
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1e-05,
+      "loss": 1.4308,
+      "step": 14100
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1e-05,
+      "loss": 1.4202,
+      "step": 14110
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1e-05,
+      "loss": 1.4225,
+      "step": 14120
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1e-05,
+      "loss": 1.4238,
+      "step": 14130
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1e-05,
+      "loss": 1.4225,
+      "step": 14140
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1e-05,
+      "loss": 1.4244,
+      "step": 14150
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1e-05,
+      "loss": 1.4204,
+      "step": 14160
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1e-05,
+      "loss": 1.4262,
+      "step": 14170
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1e-05,
+      "loss": 1.4263,
+      "step": 14180
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1e-05,
+      "loss": 1.4245,
+      "step": 14190
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1e-05,
+      "loss": 1.4208,
+      "step": 14200
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1e-05,
+      "loss": 1.4233,
+      "step": 14210
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1e-05,
+      "loss": 1.4239,
+      "step": 14220
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1e-05,
+      "loss": 1.4273,
+      "step": 14230
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1e-05,
+      "loss": 1.4252,
+      "step": 14240
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1e-05,
+      "loss": 1.424,
+      "step": 14250
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1e-05,
+      "loss": 1.4246,
+      "step": 14260
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1e-05,
+      "loss": 1.423,
+      "step": 14270
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1e-05,
+      "loss": 1.4232,
+      "step": 14280
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1e-05,
+      "loss": 1.4238,
+      "step": 14290
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1e-05,
+      "loss": 1.4201,
+      "step": 14300
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1e-05,
+      "loss": 1.422,
+      "step": 14310
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1e-05,
+      "loss": 1.4215,
+      "step": 14320
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1e-05,
+      "loss": 1.4261,
+      "step": 14330
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1e-05,
+      "loss": 1.4254,
+      "step": 14340
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1e-05,
+      "loss": 1.4214,
+      "step": 14350
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1e-05,
+      "loss": 1.4214,
+      "step": 14360
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1e-05,
+      "loss": 1.424,
+      "step": 14370
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1e-05,
+      "loss": 1.4232,
+      "step": 14380
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1e-05,
+      "loss": 1.4218,
+      "step": 14390
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1e-05,
+      "loss": 1.4264,
+      "step": 14400
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1e-05,
+      "loss": 1.4229,
+      "step": 14410
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1e-05,
+      "loss": 1.4214,
+      "step": 14420
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1e-05,
+      "loss": 1.4218,
+      "step": 14430
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1e-05,
+      "loss": 1.4266,
+      "step": 14440
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1e-05,
+      "loss": 1.424,
+      "step": 14450
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1e-05,
+      "loss": 1.4223,
+      "step": 14460
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1e-05,
+      "loss": 1.4246,
+      "step": 14470
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1e-05,
+      "loss": 1.4246,
+      "step": 14480
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1e-05,
+      "loss": 1.4243,
+      "step": 14490
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1e-05,
+      "loss": 1.4238,
+      "step": 14500
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1e-05,
+      "loss": 1.4217,
+      "step": 14510
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1e-05,
+      "loss": 1.423,
+      "step": 14520
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1e-05,
+      "loss": 1.4252,
+      "step": 14530
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1e-05,
+      "loss": 1.4271,
+      "step": 14540
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1e-05,
+      "loss": 1.4226,
+      "step": 14550
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1e-05,
+      "loss": 1.4236,
+      "step": 14560
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1e-05,
+      "loss": 1.4242,
+      "step": 14570
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1e-05,
+      "loss": 1.4223,
+      "step": 14580
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1e-05,
+      "loss": 1.4232,
+      "step": 14590
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1e-05,
+      "loss": 1.4217,
+      "step": 14600
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1e-05,
+      "loss": 1.4219,
+      "step": 14610
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1e-05,
+      "loss": 1.426,
+      "step": 14620
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1e-05,
+      "loss": 1.4203,
+      "step": 14630
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1e-05,
+      "loss": 1.4236,
+      "step": 14640
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1e-05,
+      "loss": 1.4302,
+      "step": 14650
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1e-05,
+      "loss": 1.4205,
+      "step": 14660
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1e-05,
+      "loss": 1.4259,
+      "step": 14670
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1e-05,
+      "loss": 1.426,
+      "step": 14680
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1e-05,
+      "loss": 1.4215,
+      "step": 14690
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1e-05,
+      "loss": 1.4218,
+      "step": 14700
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1e-05,
+      "loss": 1.4226,
+      "step": 14710
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1e-05,
+      "loss": 1.4221,
+      "step": 14720
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1e-05,
+      "loss": 1.4198,
+      "step": 14730
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1e-05,
+      "loss": 1.4271,
+      "step": 14740
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1e-05,
+      "loss": 1.4209,
+      "step": 14750
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1e-05,
+      "loss": 1.4197,
+      "step": 14760
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1e-05,
+      "loss": 1.4249,
+      "step": 14770
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1e-05,
+      "loss": 1.4249,
+      "step": 14780
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1e-05,
+      "loss": 1.4284,
+      "step": 14790
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1e-05,
+      "loss": 1.4256,
+      "step": 14800
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1e-05,
+      "loss": 1.4221,
+      "step": 14810
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1e-05,
+      "loss": 1.4266,
+      "step": 14820
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1e-05,
+      "loss": 1.4232,
+      "step": 14830
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1e-05,
+      "loss": 1.4229,
+      "step": 14840
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1e-05,
+      "loss": 1.425,
+      "step": 14850
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1e-05,
+      "loss": 1.4249,
+      "step": 14860
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1e-05,
+      "loss": 1.4265,
+      "step": 14870
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1e-05,
+      "loss": 1.4204,
+      "step": 14880
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1e-05,
+      "loss": 1.4226,
+      "step": 14890
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1e-05,
+      "loss": 1.4226,
+      "step": 14900
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1e-05,
+      "loss": 1.4213,
+      "step": 14910
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1e-05,
+      "loss": 1.4244,
+      "step": 14920
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1e-05,
+      "loss": 1.4225,
+      "step": 14930
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1e-05,
+      "loss": 1.4206,
+      "step": 14940
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1e-05,
+      "loss": 1.4277,
+      "step": 14950
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1e-05,
+      "loss": 1.4232,
+      "step": 14960
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1e-05,
+      "loss": 1.425,
+      "step": 14970
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1e-05,
+      "loss": 1.4265,
+      "step": 14980
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1e-05,
+      "loss": 1.4232,
+      "step": 14990
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1e-05,
+      "loss": 1.4189,
+      "step": 15000
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1e-05,
+      "loss": 1.4303,
+      "step": 15010
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1e-05,
+      "loss": 1.4211,
+      "step": 15020
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1e-05,
+      "loss": 1.4233,
+      "step": 15030
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1e-05,
+      "loss": 1.4224,
+      "step": 15040
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1e-05,
+      "loss": 1.4197,
+      "step": 15050
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1e-05,
+      "loss": 1.4228,
+      "step": 15060
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1e-05,
+      "loss": 1.417,
+      "step": 15070
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1e-05,
+      "loss": 1.4241,
+      "step": 15080
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1e-05,
+      "loss": 1.426,
+      "step": 15090
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1e-05,
+      "loss": 1.4273,
+      "step": 15100
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1e-05,
+      "loss": 1.4224,
+      "step": 15110
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1e-05,
+      "loss": 1.4233,
+      "step": 15120
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1e-05,
+      "loss": 1.4233,
+      "step": 15130
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1e-05,
+      "loss": 1.419,
+      "step": 15140
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1e-05,
+      "loss": 1.4191,
+      "step": 15150
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1e-05,
+      "loss": 1.4181,
+      "step": 15160
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1e-05,
+      "loss": 1.4214,
+      "step": 15170
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1e-05,
+      "loss": 1.4277,
+      "step": 15180
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1e-05,
+      "loss": 1.4215,
+      "step": 15190
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1e-05,
+      "loss": 1.4235,
+      "step": 15200
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1e-05,
+      "loss": 1.4225,
+      "step": 15210
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1e-05,
+      "loss": 1.4202,
+      "step": 15220
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1e-05,
+      "loss": 1.4196,
+      "step": 15230
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1e-05,
+      "loss": 1.4238,
+      "step": 15240
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1e-05,
+      "loss": 1.4218,
+      "step": 15250
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1e-05,
+      "loss": 1.4241,
+      "step": 15260
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1e-05,
+      "loss": 1.4202,
+      "step": 15270
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1e-05,
+      "loss": 1.4268,
+      "step": 15280
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1e-05,
+      "loss": 1.4217,
+      "step": 15290
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1e-05,
+      "loss": 1.4317,
+      "step": 15300
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1e-05,
+      "loss": 1.4217,
+      "step": 15310
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1e-05,
+      "loss": 1.4181,
+      "step": 15320
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1e-05,
+      "loss": 1.4265,
+      "step": 15330
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1e-05,
+      "loss": 1.4186,
+      "step": 15340
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1e-05,
+      "loss": 1.4198,
+      "step": 15350
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1e-05,
+      "loss": 1.4211,
+      "step": 15360
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1e-05,
+      "loss": 1.4201,
+      "step": 15370
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1e-05,
+      "loss": 1.4262,
+      "step": 15380
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1e-05,
+      "loss": 1.4243,
+      "step": 15390
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1e-05,
+      "loss": 1.4221,
+      "step": 15400
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1e-05,
+      "loss": 1.4205,
+      "step": 15410
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1e-05,
+      "loss": 1.4241,
+      "step": 15420
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1e-05,
+      "loss": 1.4243,
+      "step": 15430
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1e-05,
+      "loss": 1.4198,
+      "step": 15440
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1e-05,
+      "loss": 1.42,
+      "step": 15450
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1e-05,
+      "loss": 1.4207,
+      "step": 15460
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1e-05,
+      "loss": 1.4239,
+      "step": 15470
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1e-05,
+      "loss": 1.4238,
+      "step": 15480
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1e-05,
+      "loss": 1.4221,
+      "step": 15490
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1e-05,
+      "loss": 1.4203,
+      "step": 15500
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1e-05,
+      "loss": 1.4266,
+      "step": 15510
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1e-05,
+      "loss": 1.4166,
+      "step": 15520
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1e-05,
+      "loss": 1.4272,
+      "step": 15530
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1e-05,
+      "loss": 1.4203,
+      "step": 15540
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1e-05,
+      "loss": 1.4175,
+      "step": 15550
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1e-05,
+      "loss": 1.4191,
+      "step": 15560
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1e-05,
+      "loss": 1.4212,
+      "step": 15570
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1e-05,
+      "loss": 1.4195,
+      "step": 15580
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1e-05,
+      "loss": 1.4174,
+      "step": 15590
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1e-05,
+      "loss": 1.4209,
+      "step": 15600
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1e-05,
+      "loss": 1.4196,
+      "step": 15610
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1e-05,
+      "loss": 1.4236,
+      "step": 15620
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1e-05,
+      "loss": 1.4244,
+      "step": 15630
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1e-05,
+      "loss": 1.4242,
+      "step": 15640
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1e-05,
+      "loss": 1.4271,
+      "step": 15650
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1e-05,
+      "loss": 1.4265,
+      "step": 15660
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1e-05,
+      "loss": 1.4179,
+      "step": 15670
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1e-05,
+      "loss": 1.4204,
+      "step": 15680
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1e-05,
+      "loss": 1.4337,
+      "step": 15690
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1e-05,
+      "loss": 1.4216,
+      "step": 15700
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1e-05,
+      "loss": 1.4189,
+      "step": 15710
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1e-05,
+      "loss": 1.4298,
+      "step": 15720
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1e-05,
+      "loss": 1.4179,
+      "step": 15730
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1e-05,
+      "loss": 1.4245,
+      "step": 15740
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1e-05,
+      "loss": 1.4201,
+      "step": 15750
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1e-05,
+      "loss": 1.4208,
+      "step": 15760
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1e-05,
+      "loss": 1.4245,
+      "step": 15770
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1e-05,
+      "loss": 1.4171,
+      "step": 15780
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1e-05,
+      "loss": 1.4157,
+      "step": 15790
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1e-05,
+      "loss": 1.4178,
+      "step": 15800
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1e-05,
+      "loss": 1.4233,
+      "step": 15810
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1e-05,
+      "loss": 1.4192,
+      "step": 15820
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1e-05,
+      "loss": 1.4214,
+      "step": 15830
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1e-05,
+      "loss": 1.4256,
+      "step": 15840
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1e-05,
+      "loss": 1.4224,
+      "step": 15850
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1e-05,
+      "loss": 1.4229,
+      "step": 15860
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1e-05,
+      "loss": 1.4241,
+      "step": 15870
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1e-05,
+      "loss": 1.4189,
+      "step": 15880
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1e-05,
+      "loss": 1.4199,
+      "step": 15890
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1e-05,
+      "loss": 1.42,
+      "step": 15900
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1e-05,
+      "loss": 1.4232,
+      "step": 15910
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1e-05,
+      "loss": 1.4228,
+      "step": 15920
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1e-05,
+      "loss": 1.4209,
+      "step": 15930
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1e-05,
+      "loss": 1.4178,
+      "step": 15940
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1e-05,
+      "loss": 1.4186,
+      "step": 15950
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1e-05,
+      "loss": 1.4207,
+      "step": 15960
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1e-05,
+      "loss": 1.4204,
+      "step": 15970
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1e-05,
+      "loss": 1.4215,
+      "step": 15980
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1e-05,
+      "loss": 1.4257,
+      "step": 15990
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1e-05,
+      "loss": 1.4222,
+      "step": 16000
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1e-05,
+      "loss": 1.4191,
+      "step": 16010
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1e-05,
+      "loss": 1.4225,
+      "step": 16020
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1e-05,
+      "loss": 1.4186,
+      "step": 16030
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1e-05,
+      "loss": 1.4201,
+      "step": 16040
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1e-05,
+      "loss": 1.4266,
+      "step": 16050
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1e-05,
+      "loss": 1.4254,
+      "step": 16060
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1e-05,
+      "loss": 1.4231,
+      "step": 16070
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1e-05,
+      "loss": 1.4173,
+      "step": 16080
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1e-05,
+      "loss": 1.416,
+      "step": 16090
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1e-05,
+      "loss": 1.418,
+      "step": 16100
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1e-05,
+      "loss": 1.4195,
+      "step": 16110
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1e-05,
+      "loss": 1.4244,
+      "step": 16120
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1e-05,
+      "loss": 1.4213,
+      "step": 16130
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1e-05,
+      "loss": 1.4167,
+      "step": 16140
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1e-05,
+      "loss": 1.4213,
+      "step": 16150
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1e-05,
+      "loss": 1.4251,
+      "step": 16160
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1e-05,
+      "loss": 1.4334,
+      "step": 16170
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1e-05,
+      "loss": 1.4175,
+      "step": 16180
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1e-05,
+      "loss": 1.4185,
+      "step": 16190
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1e-05,
+      "loss": 1.4197,
+      "step": 16200
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1e-05,
+      "loss": 1.4141,
+      "step": 16210
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1e-05,
+      "loss": 1.418,
+      "step": 16220
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1e-05,
+      "loss": 1.4313,
+      "step": 16230
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1e-05,
+      "loss": 1.4182,
+      "step": 16240
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1e-05,
+      "loss": 1.4176,
+      "step": 16250
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1e-05,
+      "loss": 1.4198,
+      "step": 16260
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1e-05,
+      "loss": 1.4216,
+      "step": 16270
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1e-05,
+      "loss": 1.4192,
+      "step": 16280
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1e-05,
+      "loss": 1.4247,
+      "step": 16290
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1e-05,
+      "loss": 1.4199,
+      "step": 16300
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1e-05,
+      "loss": 1.4215,
+      "step": 16310
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1e-05,
+      "loss": 1.4178,
+      "step": 16320
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1e-05,
+      "loss": 1.4223,
+      "step": 16330
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1e-05,
+      "loss": 1.4191,
+      "step": 16340
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1e-05,
+      "loss": 1.4192,
+      "step": 16350
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1e-05,
+      "loss": 1.4204,
+      "step": 16360
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1e-05,
+      "loss": 1.42,
+      "step": 16370
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1e-05,
+      "loss": 1.4225,
+      "step": 16380
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1e-05,
+      "loss": 1.4248,
+      "step": 16390
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1e-05,
+      "loss": 1.4183,
+      "step": 16400
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1e-05,
+      "loss": 1.419,
+      "step": 16410
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1e-05,
+      "loss": 1.4209,
+      "step": 16420
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1e-05,
+      "loss": 1.423,
+      "step": 16430
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1e-05,
+      "loss": 1.4199,
+      "step": 16440
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1e-05,
+      "loss": 1.4176,
+      "step": 16450
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1e-05,
+      "loss": 1.4161,
+      "step": 16460
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1e-05,
+      "loss": 1.4224,
+      "step": 16470
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1e-05,
+      "loss": 1.4173,
+      "step": 16480
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1e-05,
+      "loss": 1.4305,
+      "step": 16490
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1e-05,
+      "loss": 1.4182,
+      "step": 16500
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1e-05,
+      "loss": 1.42,
+      "step": 16510
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1e-05,
+      "loss": 1.4174,
+      "step": 16520
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1e-05,
+      "loss": 1.4212,
+      "step": 16530
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1e-05,
+      "loss": 1.4213,
+      "step": 16540
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1e-05,
+      "loss": 1.421,
+      "step": 16550
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1e-05,
+      "loss": 1.4251,
+      "step": 16560
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1e-05,
+      "loss": 1.4217,
+      "step": 16570
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1e-05,
+      "loss": 1.4208,
+      "step": 16580
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1e-05,
+      "loss": 1.4174,
+      "step": 16590
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1e-05,
+      "loss": 1.4193,
+      "step": 16600
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1e-05,
+      "loss": 1.4188,
+      "step": 16610
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1e-05,
+      "loss": 1.4214,
+      "step": 16620
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1e-05,
+      "loss": 1.4248,
+      "step": 16630
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1e-05,
+      "loss": 1.4173,
+      "step": 16640
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1e-05,
+      "loss": 1.4176,
+      "step": 16650
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1e-05,
+      "loss": 1.4257,
+      "step": 16660
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1e-05,
+      "loss": 1.4215,
+      "step": 16670
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1e-05,
+      "loss": 1.4225,
+      "step": 16680
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1e-05,
+      "loss": 1.4303,
+      "step": 16690
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1e-05,
+      "loss": 1.4234,
+      "step": 16700
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1e-05,
+      "loss": 1.4194,
+      "step": 16710
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1e-05,
+      "loss": 1.4213,
+      "step": 16720
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1e-05,
+      "loss": 1.4218,
+      "step": 16730
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1e-05,
+      "loss": 1.4196,
+      "step": 16740
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1e-05,
+      "loss": 1.4185,
+      "step": 16750
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1e-05,
+      "loss": 1.4263,
+      "step": 16760
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1e-05,
+      "loss": 1.4217,
+      "step": 16770
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1e-05,
+      "loss": 1.4151,
+      "step": 16780
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1e-05,
+      "loss": 1.424,
+      "step": 16790
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1e-05,
+      "loss": 1.4191,
+      "step": 16800
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1e-05,
+      "loss": 1.4229,
+      "step": 16810
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1e-05,
+      "loss": 1.4212,
+      "step": 16820
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1e-05,
+      "loss": 1.4194,
+      "step": 16830
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1e-05,
+      "loss": 1.4177,
+      "step": 16840
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1e-05,
+      "loss": 1.4235,
+      "step": 16850
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1e-05,
+      "loss": 1.4176,
+      "step": 16860
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1e-05,
+      "loss": 1.4234,
+      "step": 16870
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1e-05,
+      "loss": 1.422,
+      "step": 16880
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1e-05,
+      "loss": 1.4217,
+      "step": 16890
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1e-05,
+      "loss": 1.4178,
+      "step": 16900
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1e-05,
+      "loss": 1.4193,
+      "step": 16910
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1e-05,
+      "loss": 1.42,
+      "step": 16920
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1e-05,
+      "loss": 1.42,
+      "step": 16930
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1e-05,
+      "loss": 1.4231,
+      "step": 16940
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1e-05,
+      "loss": 1.4258,
+      "step": 16950
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1e-05,
+      "loss": 1.4161,
+      "step": 16960
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1e-05,
+      "loss": 1.4252,
+      "step": 16970
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1e-05,
+      "loss": 1.4201,
+      "step": 16980
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1e-05,
+      "loss": 1.4189,
+      "step": 16990
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1e-05,
+      "loss": 1.4206,
+      "step": 17000
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1e-05,
+      "loss": 1.4157,
+      "step": 17010
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1e-05,
+      "loss": 1.4276,
+      "step": 17020
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1e-05,
+      "loss": 1.4187,
+      "step": 17030
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1e-05,
+      "loss": 1.4192,
+      "step": 17040
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1e-05,
+      "loss": 1.4161,
+      "step": 17050
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1e-05,
+      "loss": 1.4184,
+      "step": 17060
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1e-05,
+      "loss": 1.4234,
+      "step": 17070
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1e-05,
+      "loss": 1.4221,
+      "step": 17080
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1e-05,
+      "loss": 1.4173,
+      "step": 17090
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1e-05,
+      "loss": 1.4207,
+      "step": 17100
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1e-05,
+      "loss": 1.4255,
+      "step": 17110
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1e-05,
+      "loss": 1.4236,
+      "step": 17120
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1e-05,
+      "loss": 1.4227,
+      "step": 17130
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1e-05,
+      "loss": 1.4168,
+      "step": 17140
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1e-05,
+      "loss": 1.4262,
+      "step": 17150
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1e-05,
+      "loss": 1.4272,
+      "step": 17160
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1e-05,
+      "loss": 1.4209,
+      "step": 17170
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1e-05,
+      "loss": 1.4205,
+      "step": 17180
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1e-05,
+      "loss": 1.4199,
+      "step": 17190
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1e-05,
+      "loss": 1.4215,
+      "step": 17200
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1e-05,
+      "loss": 1.4159,
+      "step": 17210
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1e-05,
+      "loss": 1.421,
+      "step": 17220
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1e-05,
+      "loss": 1.4172,
+      "step": 17230
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1e-05,
+      "loss": 1.4196,
+      "step": 17240
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1e-05,
+      "loss": 1.419,
+      "step": 17250
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1e-05,
+      "loss": 1.4172,
+      "step": 17260
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1e-05,
+      "loss": 1.4276,
+      "step": 17270
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1e-05,
+      "loss": 1.4193,
+      "step": 17280
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1e-05,
+      "loss": 1.4187,
+      "step": 17290
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1e-05,
+      "loss": 1.4175,
+      "step": 17300
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1e-05,
+      "loss": 1.4212,
+      "step": 17310
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1e-05,
+      "loss": 1.42,
+      "step": 17320
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1e-05,
+      "loss": 1.4153,
+      "step": 17330
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1e-05,
+      "loss": 1.4216,
+      "step": 17340
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1e-05,
+      "loss": 1.4156,
+      "step": 17350
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1e-05,
+      "loss": 1.4197,
+      "step": 17360
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1e-05,
+      "loss": 1.4174,
+      "step": 17370
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1e-05,
+      "loss": 1.4168,
+      "step": 17380
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1e-05,
+      "loss": 1.4159,
+      "step": 17390
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1e-05,
+      "loss": 1.422,
+      "step": 17400
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1e-05,
+      "loss": 1.4207,
+      "step": 17410
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1e-05,
+      "loss": 1.4214,
+      "step": 17420
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1e-05,
+      "loss": 1.4224,
+      "step": 17430
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1e-05,
+      "loss": 1.4202,
+      "step": 17440
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1e-05,
+      "loss": 1.4179,
+      "step": 17450
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1e-05,
+      "loss": 1.415,
+      "step": 17460
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1e-05,
+      "loss": 1.4199,
+      "step": 17470
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1e-05,
+      "loss": 1.4198,
+      "step": 17480
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1e-05,
+      "loss": 1.4162,
+      "step": 17490
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1e-05,
+      "loss": 1.4217,
+      "step": 17500
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1e-05,
+      "loss": 1.4235,
+      "step": 17510
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1e-05,
+      "loss": 1.4204,
+      "step": 17520
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1e-05,
+      "loss": 1.4213,
+      "step": 17530
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1e-05,
+      "loss": 1.4236,
+      "step": 17540
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1e-05,
+      "loss": 1.4195,
+      "step": 17550
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1e-05,
+      "loss": 1.4175,
+      "step": 17560
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1e-05,
+      "loss": 1.4244,
+      "step": 17570
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1e-05,
+      "loss": 1.4165,
+      "step": 17580
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1e-05,
+      "loss": 1.4204,
+      "step": 17590
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1e-05,
+      "loss": 1.4246,
+      "step": 17600
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1e-05,
+      "loss": 1.4168,
+      "step": 17610
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1e-05,
+      "loss": 1.4186,
+      "step": 17620
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1e-05,
+      "loss": 1.4198,
+      "step": 17630
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1e-05,
+      "loss": 1.4182,
+      "step": 17640
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1e-05,
+      "loss": 1.4214,
+      "step": 17650
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1e-05,
+      "loss": 1.4192,
+      "step": 17660
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1e-05,
+      "loss": 1.4214,
+      "step": 17670
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1e-05,
+      "loss": 1.4202,
+      "step": 17680
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1e-05,
+      "loss": 1.4232,
+      "step": 17690
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1e-05,
+      "loss": 1.4143,
+      "step": 17700
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1e-05,
+      "loss": 1.4205,
+      "step": 17710
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1e-05,
+      "loss": 1.4249,
+      "step": 17720
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1e-05,
+      "loss": 1.4207,
+      "step": 17730
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1e-05,
+      "loss": 1.4185,
+      "step": 17740
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1e-05,
+      "loss": 1.4142,
+      "step": 17750
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1e-05,
+      "loss": 1.4226,
+      "step": 17760
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1e-05,
+      "loss": 1.4153,
+      "step": 17770
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1e-05,
+      "loss": 1.4136,
+      "step": 17780
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1e-05,
+      "loss": 1.4156,
+      "step": 17790
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1e-05,
+      "loss": 1.4165,
+      "step": 17800
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1e-05,
+      "loss": 1.4172,
+      "step": 17810
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1e-05,
+      "loss": 1.4152,
+      "step": 17820
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1e-05,
+      "loss": 1.4148,
+      "step": 17830
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1e-05,
+      "loss": 1.4207,
+      "step": 17840
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1e-05,
+      "loss": 1.4197,
+      "step": 17850
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1e-05,
+      "loss": 1.4142,
+      "step": 17860
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1e-05,
+      "loss": 1.4184,
+      "step": 17870
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1e-05,
+      "loss": 1.4176,
+      "step": 17880
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1e-05,
+      "loss": 1.4165,
+      "step": 17890
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1e-05,
+      "loss": 1.4181,
+      "step": 17900
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1e-05,
+      "loss": 1.4181,
+      "step": 17910
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1e-05,
+      "loss": 1.4181,
+      "step": 17920
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1e-05,
+      "loss": 1.4159,
+      "step": 17930
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1e-05,
+      "loss": 1.4188,
+      "step": 17940
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1e-05,
+      "loss": 1.4147,
+      "step": 17950
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1e-05,
+      "loss": 1.4176,
+      "step": 17960
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1e-05,
+      "loss": 1.417,
+      "step": 17970
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1e-05,
+      "loss": 1.4146,
+      "step": 17980
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1e-05,
+      "loss": 1.4174,
+      "step": 17990
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1e-05,
+      "loss": 1.4165,
+      "step": 18000
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1e-05,
+      "loss": 1.4178,
+      "step": 18010
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1e-05,
+      "loss": 1.4203,
+      "step": 18020
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1e-05,
+      "loss": 1.4177,
+      "step": 18030
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1e-05,
+      "loss": 1.4192,
+      "step": 18040
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1e-05,
+      "loss": 1.4203,
+      "step": 18050
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1e-05,
+      "loss": 1.4139,
+      "step": 18060
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1e-05,
+      "loss": 1.423,
+      "step": 18070
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1e-05,
+      "loss": 1.4169,
+      "step": 18080
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1e-05,
+      "loss": 1.4161,
+      "step": 18090
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1e-05,
+      "loss": 1.4135,
+      "step": 18100
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1e-05,
+      "loss": 1.4201,
+      "step": 18110
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1e-05,
+      "loss": 1.4178,
+      "step": 18120
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1e-05,
+      "loss": 1.4166,
+      "step": 18130
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1e-05,
+      "loss": 1.4188,
+      "step": 18140
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1e-05,
+      "loss": 1.4167,
+      "step": 18150
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1e-05,
+      "loss": 1.417,
+      "step": 18160
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1e-05,
+      "loss": 1.416,
+      "step": 18170
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1e-05,
+      "loss": 1.4214,
+      "step": 18180
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1e-05,
+      "loss": 1.4185,
+      "step": 18190
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1e-05,
+      "loss": 1.4188,
+      "step": 18200
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1e-05,
+      "loss": 1.4244,
+      "step": 18210
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1e-05,
+      "loss": 1.4179,
+      "step": 18220
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1e-05,
+      "loss": 1.4155,
+      "step": 18230
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1e-05,
+      "loss": 1.4177,
+      "step": 18240
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1e-05,
+      "loss": 1.4183,
+      "step": 18250
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1e-05,
+      "loss": 1.4206,
+      "step": 18260
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1e-05,
+      "loss": 1.4158,
+      "step": 18270
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1e-05,
+      "loss": 1.4196,
+      "step": 18280
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1e-05,
+      "loss": 1.4196,
+      "step": 18290
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1e-05,
+      "loss": 1.4163,
+      "step": 18300
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1e-05,
+      "loss": 1.4156,
+      "step": 18310
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1e-05,
+      "loss": 1.4166,
+      "step": 18320
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1e-05,
+      "loss": 1.4173,
+      "step": 18330
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1e-05,
+      "loss": 1.418,
+      "step": 18340
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1e-05,
+      "loss": 1.4173,
+      "step": 18350
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1e-05,
+      "loss": 1.4198,
+      "step": 18360
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1e-05,
+      "loss": 1.4174,
+      "step": 18370
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1e-05,
+      "loss": 1.4158,
+      "step": 18380
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1e-05,
+      "loss": 1.4191,
+      "step": 18390
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1e-05,
+      "loss": 1.4157,
+      "step": 18400
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1e-05,
+      "loss": 1.4163,
+      "step": 18410
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1e-05,
+      "loss": 1.4169,
+      "step": 18420
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1e-05,
+      "loss": 1.4181,
+      "step": 18430
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1e-05,
+      "loss": 1.4145,
+      "step": 18440
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1e-05,
+      "loss": 1.4164,
+      "step": 18450
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1e-05,
+      "loss": 1.4183,
+      "step": 18460
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1e-05,
+      "loss": 1.4155,
+      "step": 18470
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1e-05,
+      "loss": 1.4243,
+      "step": 18480
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1e-05,
+      "loss": 1.4188,
+      "step": 18490
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1e-05,
+      "loss": 1.4154,
+      "step": 18500
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1e-05,
+      "loss": 1.4178,
+      "step": 18510
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1e-05,
+      "loss": 1.4183,
+      "step": 18520
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1e-05,
+      "loss": 1.4165,
+      "step": 18530
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1e-05,
+      "loss": 1.4155,
+      "step": 18540
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1e-05,
+      "loss": 1.4143,
+      "step": 18550
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1e-05,
+      "loss": 1.4232,
+      "step": 18560
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1e-05,
+      "loss": 1.4158,
+      "step": 18570
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1e-05,
+      "loss": 1.4168,
+      "step": 18580
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1e-05,
+      "loss": 1.4133,
+      "step": 18590
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1e-05,
+      "loss": 1.4156,
+      "step": 18600
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1e-05,
+      "loss": 1.4178,
+      "step": 18610
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1e-05,
+      "loss": 1.415,
+      "step": 18620
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1e-05,
+      "loss": 1.416,
+      "step": 18630
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1e-05,
+      "loss": 1.4194,
+      "step": 18640
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1e-05,
+      "loss": 1.4204,
+      "step": 18650
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1e-05,
+      "loss": 1.4229,
+      "step": 18660
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1e-05,
+      "loss": 1.4175,
+      "step": 18670
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1e-05,
+      "loss": 1.4156,
+      "step": 18680
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1e-05,
+      "loss": 1.4164,
+      "step": 18690
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1e-05,
+      "loss": 1.4179,
+      "step": 18700
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1e-05,
+      "loss": 1.4197,
+      "step": 18710
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1e-05,
+      "loss": 1.4161,
+      "step": 18720
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1e-05,
+      "loss": 1.4149,
+      "step": 18730
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1e-05,
+      "loss": 1.4161,
+      "step": 18740
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1e-05,
+      "loss": 1.4173,
+      "step": 18750
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1e-05,
+      "loss": 1.4216,
+      "step": 18760
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1e-05,
+      "loss": 1.4162,
+      "step": 18770
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1e-05,
+      "loss": 1.4296,
+      "step": 18780
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1e-05,
+      "loss": 1.4146,
+      "step": 18790
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1e-05,
+      "loss": 1.418,
+      "step": 18800
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1e-05,
+      "loss": 1.4197,
+      "step": 18810
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1e-05,
+      "loss": 1.414,
+      "step": 18820
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1e-05,
+      "loss": 1.415,
+      "step": 18830
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1e-05,
+      "loss": 1.4126,
+      "step": 18840
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1e-05,
+      "loss": 1.4221,
+      "step": 18850
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1e-05,
+      "loss": 1.4161,
+      "step": 18860
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1e-05,
+      "loss": 1.4175,
+      "step": 18870
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1e-05,
+      "loss": 1.4174,
+      "step": 18880
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1e-05,
+      "loss": 1.4163,
+      "step": 18890
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1e-05,
+      "loss": 1.4169,
+      "step": 18900
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1e-05,
+      "loss": 1.4161,
+      "step": 18910
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1e-05,
+      "loss": 1.4158,
+      "step": 18920
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1e-05,
+      "loss": 1.4137,
+      "step": 18930
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1e-05,
+      "loss": 1.413,
+      "step": 18940
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1e-05,
+      "loss": 1.4153,
+      "step": 18950
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1e-05,
+      "loss": 1.4171,
+      "step": 18960
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1e-05,
+      "loss": 1.4147,
+      "step": 18970
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1e-05,
+      "loss": 1.4203,
+      "step": 18980
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1e-05,
+      "loss": 1.4155,
+      "step": 18990
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1e-05,
+      "loss": 1.4276,
+      "step": 19000
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1e-05,
+      "loss": 1.4197,
+      "step": 19010
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1e-05,
+      "loss": 1.4146,
+      "step": 19020
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1e-05,
+      "loss": 1.4145,
+      "step": 19030
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1e-05,
+      "loss": 1.4182,
+      "step": 19040
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1e-05,
+      "loss": 1.4176,
+      "step": 19050
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1e-05,
+      "loss": 1.4171,
+      "step": 19060
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1e-05,
+      "loss": 1.4129,
+      "step": 19070
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1e-05,
+      "loss": 1.415,
+      "step": 19080
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1e-05,
+      "loss": 1.416,
+      "step": 19090
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1e-05,
+      "loss": 1.4136,
+      "step": 19100
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1e-05,
+      "loss": 1.4139,
+      "step": 19110
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1e-05,
+      "loss": 1.4172,
+      "step": 19120
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1e-05,
+      "loss": 1.4165,
+      "step": 19130
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1e-05,
+      "loss": 1.4163,
+      "step": 19140
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1e-05,
+      "loss": 1.4145,
+      "step": 19150
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1e-05,
+      "loss": 1.4169,
+      "step": 19160
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1e-05,
+      "loss": 1.4151,
+      "step": 19170
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1e-05,
+      "loss": 1.4211,
+      "step": 19180
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1e-05,
+      "loss": 1.4168,
+      "step": 19190
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1e-05,
+      "loss": 1.4199,
+      "step": 19200
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1e-05,
+      "loss": 1.4185,
+      "step": 19210
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1e-05,
+      "loss": 1.418,
+      "step": 19220
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1e-05,
+      "loss": 1.4179,
+      "step": 19230
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1e-05,
+      "loss": 1.4199,
+      "step": 19240
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1e-05,
+      "loss": 1.4131,
+      "step": 19250
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1e-05,
+      "loss": 1.415,
+      "step": 19260
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1e-05,
+      "loss": 1.4173,
+      "step": 19270
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1e-05,
+      "loss": 1.4237,
+      "step": 19280
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1e-05,
+      "loss": 1.4172,
+      "step": 19290
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1e-05,
+      "loss": 1.4154,
+      "step": 19300
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1e-05,
+      "loss": 1.4146,
+      "step": 19310
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1e-05,
+      "loss": 1.4181,
+      "step": 19320
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1e-05,
+      "loss": 1.4141,
+      "step": 19330
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1e-05,
+      "loss": 1.4149,
+      "step": 19340
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1e-05,
+      "loss": 1.4145,
+      "step": 19350
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1e-05,
+      "loss": 1.4197,
+      "step": 19360
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1e-05,
+      "loss": 1.4162,
+      "step": 19370
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1e-05,
+      "loss": 1.4209,
+      "step": 19380
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1e-05,
+      "loss": 1.4152,
+      "step": 19390
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1e-05,
+      "loss": 1.4171,
+      "step": 19400
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1e-05,
+      "loss": 1.4172,
+      "step": 19410
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1e-05,
+      "loss": 1.414,
+      "step": 19420
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1e-05,
+      "loss": 1.4154,
+      "step": 19430
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1e-05,
+      "loss": 1.4183,
+      "step": 19440
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1e-05,
+      "loss": 1.4111,
+      "step": 19450
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1e-05,
+      "loss": 1.4211,
+      "step": 19460
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1e-05,
+      "loss": 1.4198,
+      "step": 19470
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1e-05,
+      "loss": 1.4165,
+      "step": 19480
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1e-05,
+      "loss": 1.4245,
+      "step": 19490
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1e-05,
+      "loss": 1.419,
+      "step": 19500
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1e-05,
+      "loss": 1.4166,
+      "step": 19510
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1e-05,
+      "loss": 1.4158,
+      "step": 19520
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1e-05,
+      "loss": 1.4139,
+      "step": 19530
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1e-05,
+      "loss": 1.4144,
+      "step": 19540
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1e-05,
+      "loss": 1.4152,
+      "step": 19550
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1e-05,
+      "loss": 1.4155,
+      "step": 19560
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1e-05,
+      "loss": 1.417,
+      "step": 19570
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1e-05,
+      "loss": 1.4129,
+      "step": 19580
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1e-05,
+      "loss": 1.4155,
+      "step": 19590
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1e-05,
+      "loss": 1.4144,
+      "step": 19600
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1e-05,
+      "loss": 1.4137,
+      "step": 19610
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1e-05,
+      "loss": 1.4149,
+      "step": 19620
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1e-05,
+      "loss": 1.4136,
+      "step": 19630
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1e-05,
+      "loss": 1.4173,
+      "step": 19640
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1e-05,
+      "loss": 1.4164,
+      "step": 19650
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1e-05,
+      "loss": 1.4142,
+      "step": 19660
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1e-05,
+      "loss": 1.4143,
+      "step": 19670
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1e-05,
+      "loss": 1.4124,
+      "step": 19680
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1e-05,
+      "loss": 1.4141,
+      "step": 19690
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1e-05,
+      "loss": 1.4149,
+      "step": 19700
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1e-05,
+      "loss": 1.4146,
+      "step": 19710
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1e-05,
+      "loss": 1.4142,
+      "step": 19720
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1e-05,
+      "loss": 1.4119,
+      "step": 19730
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1e-05,
+      "loss": 1.4214,
+      "step": 19740
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1e-05,
+      "loss": 1.4145,
+      "step": 19750
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1e-05,
+      "loss": 1.4162,
+      "step": 19760
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1e-05,
+      "loss": 1.417,
+      "step": 19770
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1e-05,
+      "loss": 1.4255,
+      "step": 19780
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1e-05,
+      "loss": 1.418,
+      "step": 19790
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1e-05,
+      "loss": 1.416,
+      "step": 19800
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1e-05,
+      "loss": 1.4229,
+      "step": 19810
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1e-05,
+      "loss": 1.4179,
+      "step": 19820
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1e-05,
+      "loss": 1.4178,
+      "step": 19830
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1e-05,
+      "loss": 1.4168,
+      "step": 19840
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1e-05,
+      "loss": 1.4179,
+      "step": 19850
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1e-05,
+      "loss": 1.4154,
+      "step": 19860
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1e-05,
+      "loss": 1.418,
+      "step": 19870
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1e-05,
+      "loss": 1.4179,
+      "step": 19880
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1e-05,
+      "loss": 1.4197,
+      "step": 19890
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1e-05,
+      "loss": 1.4226,
+      "step": 19900
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1e-05,
+      "loss": 1.4155,
+      "step": 19910
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1e-05,
+      "loss": 1.4148,
+      "step": 19920
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1e-05,
+      "loss": 1.4137,
+      "step": 19930
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1e-05,
+      "loss": 1.4215,
+      "step": 19940
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1e-05,
+      "loss": 1.4156,
+      "step": 19950
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1e-05,
+      "loss": 1.4176,
+      "step": 19960
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1e-05,
+      "loss": 1.4172,
+      "step": 19970
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1e-05,
+      "loss": 1.4161,
+      "step": 19980
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1e-05,
+      "loss": 1.4146,
+      "step": 19990
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1e-05,
+      "loss": 1.4129,
+      "step": 20000
+    },
+    {
+      "epoch": 6.62,
+      "eval_accuracy": 0.998,
+      "eval_loss": 1.4164189100265503,
+      "eval_runtime": 358.657,
+      "eval_samples_per_second": 8.365,
+      "step": 20000
+    }
+  ],
+  "max_steps": 20000,
+  "num_train_epochs": 7,
+  "total_flos": 1.3216276868497408e+16,
+  "trial_name": null,
+  "trial_params": null
+}