diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,13525 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 22500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999111111111111e-05,
+      "loss": 2.1847,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9982222222222224e-05,
+      "loss": 2.1475,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9973333333333334e-05,
+      "loss": 2.0361,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9964444444444447e-05,
+      "loss": 2.037,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9955555555555557e-05,
+      "loss": 2.0814,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9946666666666667e-05,
+      "loss": 2.0709,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.993777777777778e-05,
+      "loss": 2.0122,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.992888888888889e-05,
+      "loss": 2.0661,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9920000000000002e-05,
+      "loss": 1.8736,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9911111111111112e-05,
+      "loss": 2.0279,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9902222222222222e-05,
+      "loss": 1.9704,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9893333333333335e-05,
+      "loss": 2.0314,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9884444444444445e-05,
+      "loss": 2.0765,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9875555555555558e-05,
+      "loss": 1.9607,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9866666666666667e-05,
+      "loss": 2.0507,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.985777777777778e-05,
+      "loss": 2.1223,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.984888888888889e-05,
+      "loss": 1.9513,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9840000000000003e-05,
+      "loss": 1.9894,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9831111111111113e-05,
+      "loss": 1.94,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9822222222222226e-05,
+      "loss": 1.8444,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9813333333333336e-05,
+      "loss": 1.9063,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9804444444444445e-05,
+      "loss": 1.9352,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.979555555555556e-05,
+      "loss": 1.935,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9786666666666668e-05,
+      "loss": 1.8967,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.977777777777778e-05,
+      "loss": 2.0144,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.976888888888889e-05,
+      "loss": 1.9898,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.976e-05,
+      "loss": 2.012,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9751111111111114e-05,
+      "loss": 1.9699,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9742222222222223e-05,
+      "loss": 1.918,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9733333333333336e-05,
+      "loss": 1.9957,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9724444444444446e-05,
+      "loss": 2.0303,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9715555555555556e-05,
+      "loss": 1.9581,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.970666666666667e-05,
+      "loss": 1.8452,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.969777777777778e-05,
+      "loss": 1.9915,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.968888888888889e-05,
+      "loss": 1.8454,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.968e-05,
+      "loss": 2.0311,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.967111111111111e-05,
+      "loss": 1.9825,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9662222222222224e-05,
+      "loss": 1.7858,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9653333333333334e-05,
+      "loss": 1.9468,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9644444444444447e-05,
+      "loss": 1.9827,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9635555555555557e-05,
+      "loss": 1.9055,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9626666666666666e-05,
+      "loss": 1.9589,
+      "step": 420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.961777777777778e-05,
+      "loss": 1.9031,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.960888888888889e-05,
+      "loss": 1.9699,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 1.9873,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9591111111111112e-05,
+      "loss": 2.0331,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9582222222222225e-05,
+      "loss": 2.0458,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9573333333333335e-05,
+      "loss": 1.9029,
+      "step": 480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9564444444444444e-05,
+      "loss": 1.9421,
+      "step": 490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9555555555555557e-05,
+      "loss": 1.9881,
+      "step": 500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.954666666666667e-05,
+      "loss": 1.8729,
+      "step": 510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.953777777777778e-05,
+      "loss": 1.8436,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.952888888888889e-05,
+      "loss": 1.9262,
+      "step": 530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9520000000000003e-05,
+      "loss": 1.9438,
+      "step": 540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9511111111111113e-05,
+      "loss": 1.8841,
+      "step": 550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9502222222222226e-05,
+      "loss": 1.799,
+      "step": 560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9493333333333335e-05,
+      "loss": 1.9218,
+      "step": 570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9484444444444445e-05,
+      "loss": 1.8995,
+      "step": 580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9475555555555558e-05,
+      "loss": 1.9363,
+      "step": 590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9466666666666668e-05,
+      "loss": 1.8685,
+      "step": 600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.945777777777778e-05,
+      "loss": 1.8772,
+      "step": 610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.944888888888889e-05,
+      "loss": 1.8764,
+      "step": 620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.944e-05,
+      "loss": 1.9196,
+      "step": 630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9431111111111113e-05,
+      "loss": 1.9998,
+      "step": 640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9422222222222223e-05,
+      "loss": 1.9287,
+      "step": 650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9413333333333336e-05,
+      "loss": 1.971,
+      "step": 660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9404444444444446e-05,
+      "loss": 1.865,
+      "step": 670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9395555555555555e-05,
+      "loss": 1.9625,
+      "step": 680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.938666666666667e-05,
+      "loss": 2.102,
+      "step": 690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9377777777777778e-05,
+      "loss": 1.9093,
+      "step": 700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.936888888888889e-05,
+      "loss": 1.9673,
+      "step": 710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.936e-05,
+      "loss": 1.9121,
+      "step": 720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.935111111111111e-05,
+      "loss": 1.9261,
+      "step": 730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9342222222222224e-05,
+      "loss": 1.8706,
+      "step": 740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9333333333333333e-05,
+      "loss": 2.0241,
+      "step": 750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9324444444444447e-05,
+      "loss": 1.8352,
+      "step": 760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9315555555555556e-05,
+      "loss": 1.8939,
+      "step": 770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.930666666666667e-05,
+      "loss": 1.889,
+      "step": 780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.929777777777778e-05,
+      "loss": 1.9748,
+      "step": 790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.928888888888889e-05,
+      "loss": 1.8958,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9280000000000002e-05,
+      "loss": 1.9332,
+      "step": 810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9271111111111115e-05,
+      "loss": 1.9035,
+      "step": 820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9262222222222225e-05,
+      "loss": 1.9788,
+      "step": 830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9253333333333334e-05,
+      "loss": 1.8312,
+      "step": 840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9244444444444444e-05,
+      "loss": 1.9128,
+      "step": 850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9235555555555557e-05,
+      "loss": 1.9247,
+      "step": 860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.922666666666667e-05,
+      "loss": 1.9675,
+      "step": 870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.921777777777778e-05,
+      "loss": 1.9798,
+      "step": 880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.920888888888889e-05,
+      "loss": 1.8594,
+      "step": 890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 1.9372,
+      "step": 900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9191111111111112e-05,
+      "loss": 1.8272,
+      "step": 910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9182222222222225e-05,
+      "loss": 1.8978,
+      "step": 920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9173333333333335e-05,
+      "loss": 1.8588,
+      "step": 930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9164444444444445e-05,
+      "loss": 1.9056,
+      "step": 940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9155555555555558e-05,
+      "loss": 1.847,
+      "step": 950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9146666666666667e-05,
+      "loss": 1.8985,
+      "step": 960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.913777777777778e-05,
+      "loss": 1.8845,
+      "step": 970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.912888888888889e-05,
+      "loss": 1.9342,
+      "step": 980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.912e-05,
+      "loss": 1.9315,
+      "step": 990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9111111111111113e-05,
+      "loss": 1.9864,
+      "step": 1000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9102222222222223e-05,
+      "loss": 1.9387,
+      "step": 1010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9093333333333336e-05,
+      "loss": 2.0192,
+      "step": 1020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9084444444444445e-05,
+      "loss": 1.9337,
+      "step": 1030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9075555555555555e-05,
+      "loss": 1.8851,
+      "step": 1040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9066666666666668e-05,
+      "loss": 1.8107,
+      "step": 1050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9057777777777778e-05,
+      "loss": 1.8935,
+      "step": 1060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.904888888888889e-05,
+      "loss": 1.9739,
+      "step": 1070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.904e-05,
+      "loss": 1.8417,
+      "step": 1080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9031111111111114e-05,
+      "loss": 1.9473,
+      "step": 1090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9022222222222223e-05,
+      "loss": 1.9727,
+      "step": 1100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9013333333333333e-05,
+      "loss": 1.9843,
+      "step": 1110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9004444444444446e-05,
+      "loss": 1.9677,
+      "step": 1120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.899555555555556e-05,
+      "loss": 1.8819,
+      "step": 1130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.898666666666667e-05,
+      "loss": 1.922,
+      "step": 1140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.897777777777778e-05,
+      "loss": 1.8144,
+      "step": 1150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.896888888888889e-05,
+      "loss": 2.0499,
+      "step": 1160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.896e-05,
+      "loss": 1.9009,
+      "step": 1170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8951111111111115e-05,
+      "loss": 1.8605,
+      "step": 1180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8942222222222224e-05,
+      "loss": 1.9,
+      "step": 1190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8933333333333334e-05,
+      "loss": 1.9388,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8924444444444447e-05,
+      "loss": 1.8781,
+      "step": 1210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8915555555555557e-05,
+      "loss": 1.9763,
+      "step": 1220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.890666666666667e-05,
+      "loss": 1.8335,
+      "step": 1230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.889777777777778e-05,
+      "loss": 1.9232,
+      "step": 1240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.888888888888889e-05,
+      "loss": 1.8896,
+      "step": 1250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8880000000000002e-05,
+      "loss": 1.9744,
+      "step": 1260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8871111111111112e-05,
+      "loss": 1.8459,
+      "step": 1270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8862222222222225e-05,
+      "loss": 1.7448,
+      "step": 1280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8853333333333335e-05,
+      "loss": 1.9686,
+      "step": 1290
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8844444444444444e-05,
+      "loss": 1.8628,
+      "step": 1300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8835555555555557e-05,
+      "loss": 1.837,
+      "step": 1310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8826666666666667e-05,
+      "loss": 1.848,
+      "step": 1320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881777777777778e-05,
+      "loss": 1.859,
+      "step": 1330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880888888888889e-05,
+      "loss": 1.8516,
+      "step": 1340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88e-05,
+      "loss": 1.9106,
+      "step": 1350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8791111111111113e-05,
+      "loss": 1.8998,
+      "step": 1360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8782222222222222e-05,
+      "loss": 1.8505,
+      "step": 1370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8773333333333335e-05,
+      "loss": 1.869,
+      "step": 1380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8764444444444445e-05,
+      "loss": 1.9454,
+      "step": 1390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8755555555555558e-05,
+      "loss": 1.974,
+      "step": 1400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8746666666666668e-05,
+      "loss": 2.008,
+      "step": 1410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8737777777777778e-05,
+      "loss": 1.8839,
+      "step": 1420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872888888888889e-05,
+      "loss": 1.9231,
+      "step": 1430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8720000000000004e-05,
+      "loss": 1.9344,
+      "step": 1440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8711111111111113e-05,
+      "loss": 1.9142,
+      "step": 1450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8702222222222223e-05,
+      "loss": 1.9634,
+      "step": 1460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8693333333333333e-05,
+      "loss": 1.8861,
+      "step": 1470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8684444444444446e-05,
+      "loss": 1.7928,
+      "step": 1480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.867555555555556e-05,
+      "loss": 1.9583,
+      "step": 1490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 1.8707,
+      "step": 1500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.865777777777778e-05,
+      "loss": 1.8204,
+      "step": 1510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8648888888888888e-05,
+      "loss": 1.7685,
+      "step": 1520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.864e-05,
+      "loss": 1.7884,
+      "step": 1530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8631111111111114e-05,
+      "loss": 2.0554,
+      "step": 1540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8622222222222224e-05,
+      "loss": 1.9678,
+      "step": 1550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8613333333333334e-05,
+      "loss": 1.9219,
+      "step": 1560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8604444444444447e-05,
+      "loss": 1.7832,
+      "step": 1570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8595555555555556e-05,
+      "loss": 1.973,
+      "step": 1580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.858666666666667e-05,
+      "loss": 1.8929,
+      "step": 1590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.857777777777778e-05,
+      "loss": 1.8331,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.856888888888889e-05,
+      "loss": 1.9128,
+      "step": 1610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8560000000000002e-05,
+      "loss": 1.9168,
+      "step": 1620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.855111111111111e-05,
+      "loss": 1.9078,
+      "step": 1630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8542222222222225e-05,
+      "loss": 1.9619,
+      "step": 1640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8533333333333334e-05,
+      "loss": 2.0052,
+      "step": 1650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8524444444444444e-05,
+      "loss": 1.963,
+      "step": 1660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8515555555555557e-05,
+      "loss": 1.8751,
+      "step": 1670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8506666666666667e-05,
+      "loss": 1.9217,
+      "step": 1680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.849777777777778e-05,
+      "loss": 1.8591,
+      "step": 1690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.848888888888889e-05,
+      "loss": 1.7707,
+      "step": 1700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8480000000000003e-05,
+      "loss": 1.8575,
+      "step": 1710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8471111111111112e-05,
+      "loss": 1.7893,
+      "step": 1720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8462222222222222e-05,
+      "loss": 1.9336,
+      "step": 1730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8453333333333335e-05,
+      "loss": 1.814,
+      "step": 1740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8444444444444448e-05,
+      "loss": 1.825,
+      "step": 1750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8435555555555558e-05,
+      "loss": 1.9636,
+      "step": 1760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8426666666666668e-05,
+      "loss": 1.8628,
+      "step": 1770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8417777777777777e-05,
+      "loss": 1.8134,
+      "step": 1780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.840888888888889e-05,
+      "loss": 1.8187,
+      "step": 1790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 1.8987,
+      "step": 1800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8391111111111113e-05,
+      "loss": 1.8996,
+      "step": 1810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8382222222222223e-05,
+      "loss": 1.7418,
+      "step": 1820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8373333333333332e-05,
+      "loss": 1.8781,
+      "step": 1830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8364444444444446e-05,
+      "loss": 1.8355,
+      "step": 1840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.835555555555556e-05,
+      "loss": 1.8606,
+      "step": 1850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.834666666666667e-05,
+      "loss": 1.8834,
+      "step": 1860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8337777777777778e-05,
+      "loss": 1.9944,
+      "step": 1870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.832888888888889e-05,
+      "loss": 1.9366,
+      "step": 1880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.832e-05,
+      "loss": 1.852,
+      "step": 1890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8311111111111114e-05,
+      "loss": 1.9393,
+      "step": 1900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8302222222222224e-05,
+      "loss": 1.8783,
+      "step": 1910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8293333333333333e-05,
+      "loss": 1.9497,
+      "step": 1920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8284444444444446e-05,
+      "loss": 1.7848,
+      "step": 1930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8275555555555556e-05,
+      "loss": 1.9093,
+      "step": 1940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.826666666666667e-05,
+      "loss": 1.8496,
+      "step": 1950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.825777777777778e-05,
+      "loss": 1.8813,
+      "step": 1960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.824888888888889e-05,
+      "loss": 1.9528,
+      "step": 1970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.824e-05,
+      "loss": 1.7515,
+      "step": 1980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.823111111111111e-05,
+      "loss": 1.8742,
+      "step": 1990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8222222222222224e-05,
+      "loss": 1.7551,
+      "step": 2000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8213333333333334e-05,
+      "loss": 1.8996,
+      "step": 2010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8204444444444447e-05,
+      "loss": 1.9187,
+      "step": 2020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8195555555555557e-05,
+      "loss": 1.783,
+      "step": 2030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8186666666666666e-05,
+      "loss": 1.8344,
+      "step": 2040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.817777777777778e-05,
+      "loss": 1.9063,
+      "step": 2050
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8168888888888893e-05,
+      "loss": 1.9129,
+      "step": 2060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8160000000000002e-05,
+      "loss": 1.8462,
+      "step": 2070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8151111111111112e-05,
+      "loss": 1.8286,
+      "step": 2080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.814222222222222e-05,
+      "loss": 1.9012,
+      "step": 2090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8133333333333335e-05,
+      "loss": 1.8167,
+      "step": 2100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8124444444444448e-05,
+      "loss": 1.9675,
+      "step": 2110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8115555555555558e-05,
+      "loss": 1.8133,
+      "step": 2120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8106666666666667e-05,
+      "loss": 1.7176,
+      "step": 2130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8097777777777777e-05,
+      "loss": 1.9016,
+      "step": 2140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.808888888888889e-05,
+      "loss": 1.9175,
+      "step": 2150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8080000000000003e-05,
+      "loss": 1.9084,
+      "step": 2160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8071111111111113e-05,
+      "loss": 1.9775,
+      "step": 2170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8062222222222222e-05,
+      "loss": 1.9114,
+      "step": 2180
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8053333333333332e-05,
+      "loss": 1.9249,
+      "step": 2190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8044444444444445e-05,
+      "loss": 1.8737,
+      "step": 2200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.803555555555556e-05,
+      "loss": 1.9817,
+      "step": 2210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8026666666666668e-05,
+      "loss": 1.7711,
+      "step": 2220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8017777777777778e-05,
+      "loss": 1.9182,
+      "step": 2230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.800888888888889e-05,
+      "loss": 1.8014,
+      "step": 2240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8e-05,
+      "loss": 1.805,
+      "step": 2250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7991111111111114e-05,
+      "loss": 1.8304,
+      "step": 2260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7982222222222223e-05,
+      "loss": 1.8271,
+      "step": 2270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7973333333333333e-05,
+      "loss": 1.8199,
+      "step": 2280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7964444444444446e-05,
+      "loss": 1.8094,
+      "step": 2290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7955555555555556e-05,
+      "loss": 1.7927,
+      "step": 2300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.794666666666667e-05,
+      "loss": 1.8518,
+      "step": 2310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.793777777777778e-05,
+      "loss": 1.8703,
+      "step": 2320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.792888888888889e-05,
+      "loss": 1.8711,
+      "step": 2330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.792e-05,
+      "loss": 1.8136,
+      "step": 2340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.791111111111111e-05,
+      "loss": 1.7613,
+      "step": 2350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7902222222222224e-05,
+      "loss": 1.8932,
+      "step": 2360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7893333333333337e-05,
+      "loss": 1.8365,
+      "step": 2370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7884444444444447e-05,
+      "loss": 1.8905,
+      "step": 2380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7875555555555556e-05,
+      "loss": 1.8713,
+      "step": 2390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7866666666666666e-05,
+      "loss": 1.8353,
+      "step": 2400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.785777777777778e-05,
+      "loss": 1.7683,
+      "step": 2410
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7848888888888892e-05,
+      "loss": 2.0259,
+      "step": 2420
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7840000000000002e-05,
+      "loss": 1.7316,
+      "step": 2430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.783111111111111e-05,
+      "loss": 1.874,
+      "step": 2440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.782222222222222e-05,
+      "loss": 1.8563,
+      "step": 2450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7813333333333334e-05,
+      "loss": 1.8006,
+      "step": 2460
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7804444444444448e-05,
+      "loss": 1.8835,
+      "step": 2470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7795555555555557e-05,
+      "loss": 1.9641,
+      "step": 2480
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7786666666666667e-05,
+      "loss": 1.8004,
+      "step": 2490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 1.8723,
+      "step": 2500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.776888888888889e-05,
+      "loss": 1.9558,
+      "step": 2510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7760000000000003e-05,
+      "loss": 1.9045,
+      "step": 2520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7751111111111112e-05,
+      "loss": 1.9107,
+      "step": 2530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7742222222222222e-05,
+      "loss": 1.8421,
+      "step": 2540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7733333333333335e-05,
+      "loss": 1.8681,
+      "step": 2550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7724444444444445e-05,
+      "loss": 1.8218,
+      "step": 2560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7715555555555558e-05,
+      "loss": 1.8018,
+      "step": 2570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7706666666666668e-05,
+      "loss": 1.7739,
+      "step": 2580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7697777777777777e-05,
+      "loss": 1.8538,
+      "step": 2590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.768888888888889e-05,
+      "loss": 1.8228,
+      "step": 2600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.768e-05,
+      "loss": 1.8948,
+      "step": 2610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7671111111111113e-05,
+      "loss": 1.8888,
+      "step": 2620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7662222222222223e-05,
+      "loss": 1.747,
+      "step": 2630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7653333333333336e-05,
+      "loss": 1.783,
+      "step": 2640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7644444444444446e-05,
+      "loss": 1.825,
+      "step": 2650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7635555555555555e-05,
+      "loss": 1.9299,
+      "step": 2660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.762666666666667e-05,
+      "loss": 1.8301,
+      "step": 2670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.761777777777778e-05,
+      "loss": 1.7254,
+      "step": 2680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.760888888888889e-05,
+      "loss": 1.8504,
+      "step": 2690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.76e-05,
+      "loss": 1.8482,
+      "step": 2700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.759111111111111e-05,
+      "loss": 1.8365,
+      "step": 2710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.7582222222222224e-05,
+      "loss": 1.8017,
+      "step": 2720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.7573333333333337e-05,
+      "loss": 1.8361,
+      "step": 2730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.7564444444444446e-05,
+      "loss": 1.827,
+      "step": 2740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.7555555555555556e-05,
+      "loss": 1.7714,
+      "step": 2750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.7546666666666666e-05,
+      "loss": 1.8223,
+      "step": 2760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.753777777777778e-05,
+      "loss": 1.7868,
+      "step": 2770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.7528888888888892e-05,
+      "loss": 1.8806,
+      "step": 2780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.752e-05,
+      "loss": 1.8846,
+      "step": 2790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.751111111111111e-05,
+      "loss": 1.8832,
+      "step": 2800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.750222222222222e-05,
+      "loss": 1.8067,
+      "step": 2810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7493333333333334e-05,
+      "loss": 1.8158,
+      "step": 2820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7484444444444447e-05,
+      "loss": 1.8292,
+      "step": 2830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7475555555555557e-05,
+      "loss": 1.973,
+      "step": 2840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7466666666666667e-05,
+      "loss": 1.8248,
+      "step": 2850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.745777777777778e-05,
+      "loss": 1.8182,
+      "step": 2860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.744888888888889e-05,
+      "loss": 1.8932,
+      "step": 2870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7440000000000002e-05,
+      "loss": 1.8701,
+      "step": 2880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7431111111111112e-05,
+      "loss": 1.8347,
+      "step": 2890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7422222222222222e-05,
+      "loss": 1.839,
+      "step": 2900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7413333333333335e-05,
+      "loss": 1.8059,
+      "step": 2910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7404444444444445e-05,
+      "loss": 2.0315,
+      "step": 2920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7395555555555558e-05,
+      "loss": 1.8481,
+      "step": 2930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7386666666666667e-05,
+      "loss": 1.8156,
+      "step": 2940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.737777777777778e-05,
+      "loss": 1.8472,
+      "step": 2950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.736888888888889e-05,
+      "loss": 1.7449,
+      "step": 2960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.736e-05,
+      "loss": 1.787,
+      "step": 2970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7351111111111113e-05,
+      "loss": 1.779,
+      "step": 2980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7342222222222226e-05,
+      "loss": 1.8532,
+      "step": 2990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 1.901,
+      "step": 3000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7324444444444445e-05,
+      "loss": 1.7525,
+      "step": 3010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7315555555555555e-05,
+      "loss": 1.8174,
+      "step": 3020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7306666666666668e-05,
+      "loss": 1.9481,
+      "step": 3030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.729777777777778e-05,
+      "loss": 1.8545,
+      "step": 3040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.728888888888889e-05,
+      "loss": 1.8215,
+      "step": 3050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.728e-05,
+      "loss": 1.836,
+      "step": 3060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.727111111111111e-05,
+      "loss": 1.7745,
+      "step": 3070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7262222222222223e-05,
+      "loss": 1.8678,
+      "step": 3080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7253333333333336e-05,
+      "loss": 1.8757,
+      "step": 3090
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7244444444444446e-05,
+      "loss": 1.8523,
+      "step": 3100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7235555555555556e-05,
+      "loss": 1.8066,
+      "step": 3110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7226666666666665e-05,
+      "loss": 1.9035,
+      "step": 3120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.721777777777778e-05,
+      "loss": 1.7928,
+      "step": 3130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.720888888888889e-05,
+      "loss": 1.9414,
+      "step": 3140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.72e-05,
+      "loss": 1.8732,
+      "step": 3150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.719111111111111e-05,
+      "loss": 1.8059,
+      "step": 3160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7182222222222224e-05,
+      "loss": 1.8442,
+      "step": 3170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7173333333333334e-05,
+      "loss": 1.8084,
+      "step": 3180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7164444444444447e-05,
+      "loss": 1.8173,
+      "step": 3190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7155555555555557e-05,
+      "loss": 1.7362,
+      "step": 3200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7146666666666666e-05,
+      "loss": 1.7882,
+      "step": 3210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.713777777777778e-05,
+      "loss": 1.7937,
+      "step": 3220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.712888888888889e-05,
+      "loss": 1.8362,
+      "step": 3230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7120000000000002e-05,
+      "loss": 1.8387,
+      "step": 3240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7111111111111112e-05,
+      "loss": 1.7992,
+      "step": 3250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7102222222222225e-05,
+      "loss": 1.8306,
+      "step": 3260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7093333333333335e-05,
+      "loss": 1.8823,
+      "step": 3270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7084444444444444e-05,
+      "loss": 1.8358,
+      "step": 3280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7075555555555557e-05,
+      "loss": 1.8148,
+      "step": 3290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.706666666666667e-05,
+      "loss": 1.7982,
+      "step": 3300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.705777777777778e-05,
+      "loss": 1.6572,
+      "step": 3310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.704888888888889e-05,
+      "loss": 1.7404,
+      "step": 3320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.704e-05,
+      "loss": 1.7953,
+      "step": 3330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7031111111111113e-05,
+      "loss": 1.7513,
+      "step": 3340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7022222222222226e-05,
+      "loss": 1.7905,
+      "step": 3350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7013333333333335e-05,
+      "loss": 1.6971,
+      "step": 3360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7004444444444445e-05,
+      "loss": 1.8663,
+      "step": 3370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6995555555555555e-05,
+      "loss": 1.7496,
+      "step": 3380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6986666666666668e-05,
+      "loss": 1.8941,
+      "step": 3390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.697777777777778e-05,
+      "loss": 1.7819,
+      "step": 3400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.696888888888889e-05,
+      "loss": 1.8641,
+      "step": 3410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.696e-05,
+      "loss": 1.8365,
+      "step": 3420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.695111111111111e-05,
+      "loss": 1.8025,
+      "step": 3430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6942222222222223e-05,
+      "loss": 1.8163,
+      "step": 3440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6933333333333336e-05,
+      "loss": 1.8432,
+      "step": 3450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6924444444444446e-05,
+      "loss": 1.8422,
+      "step": 3460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6915555555555555e-05,
+      "loss": 1.7715,
+      "step": 3470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.690666666666667e-05,
+      "loss": 1.853,
+      "step": 3480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6897777777777778e-05,
+      "loss": 1.8708,
+      "step": 3490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.688888888888889e-05,
+      "loss": 1.9041,
+      "step": 3500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.688e-05,
+      "loss": 1.7341,
+      "step": 3510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.687111111111111e-05,
+      "loss": 1.8011,
+      "step": 3520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6862222222222224e-05,
+      "loss": 1.7929,
+      "step": 3530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6853333333333333e-05,
+      "loss": 1.7511,
+      "step": 3540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6844444444444447e-05,
+      "loss": 1.7666,
+      "step": 3550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6835555555555556e-05,
+      "loss": 1.8554,
+      "step": 3560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.682666666666667e-05,
+      "loss": 1.8977,
+      "step": 3570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.681777777777778e-05,
+      "loss": 1.7102,
+      "step": 3580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.680888888888889e-05,
+      "loss": 1.8676,
+      "step": 3590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 1.9091,
+      "step": 3600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6791111111111115e-05,
+      "loss": 1.8765,
+      "step": 3610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6782222222222225e-05,
+      "loss": 1.7471,
+      "step": 3620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6773333333333334e-05,
+      "loss": 1.8306,
+      "step": 3630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6764444444444444e-05,
+      "loss": 1.8785,
+      "step": 3640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6755555555555557e-05,
+      "loss": 1.7709,
+      "step": 3650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.674666666666667e-05,
+      "loss": 1.8463,
+      "step": 3660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.673777777777778e-05,
+      "loss": 1.8603,
+      "step": 3670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.672888888888889e-05,
+      "loss": 1.7914,
+      "step": 3680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.672e-05,
+      "loss": 1.7013,
+      "step": 3690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6711111111111112e-05,
+      "loss": 1.9074,
+      "step": 3700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6702222222222225e-05,
+      "loss": 1.773,
+      "step": 3710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6693333333333335e-05,
+      "loss": 1.8103,
+      "step": 3720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6684444444444445e-05,
+      "loss": 1.7782,
+      "step": 3730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6675555555555554e-05,
+      "loss": 1.7434,
+      "step": 3740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 1.8459,
+      "step": 3750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.665777777777778e-05,
+      "loss": 1.7029,
+      "step": 3760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.664888888888889e-05,
+      "loss": 1.7203,
+      "step": 3770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.664e-05,
+      "loss": 1.7825,
+      "step": 3780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6631111111111113e-05,
+      "loss": 1.8225,
+      "step": 3790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6622222222222223e-05,
+      "loss": 1.9212,
+      "step": 3800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6613333333333336e-05,
+      "loss": 1.7411,
+      "step": 3810
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6604444444444445e-05,
+      "loss": 1.782,
+      "step": 3820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6595555555555555e-05,
+      "loss": 1.8284,
+      "step": 3830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6586666666666668e-05,
+      "loss": 1.8202,
+      "step": 3840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6577777777777778e-05,
+      "loss": 1.7823,
+      "step": 3850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.656888888888889e-05,
+      "loss": 1.8564,
+      "step": 3860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.656e-05,
+      "loss": 1.8507,
+      "step": 3870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6551111111111114e-05,
+      "loss": 1.7557,
+      "step": 3880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6542222222222223e-05,
+      "loss": 1.8144,
+      "step": 3890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6533333333333333e-05,
+      "loss": 1.793,
+      "step": 3900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6524444444444446e-05,
+      "loss": 1.7302,
+      "step": 3910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.651555555555556e-05,
+      "loss": 1.9816,
+      "step": 3920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.650666666666667e-05,
+      "loss": 1.7437,
+      "step": 3930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.649777777777778e-05,
+      "loss": 1.7755,
+      "step": 3940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.648888888888889e-05,
+      "loss": 1.9136,
+      "step": 3950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.648e-05,
+      "loss": 1.7864,
+      "step": 3960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6471111111111115e-05,
+      "loss": 1.8314,
+      "step": 3970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6462222222222224e-05,
+      "loss": 1.8898,
+      "step": 3980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6453333333333334e-05,
+      "loss": 1.7912,
+      "step": 3990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6444444444444444e-05,
+      "loss": 1.8007,
+      "step": 4000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6435555555555557e-05,
+      "loss": 1.7512,
+      "step": 4010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.642666666666667e-05,
+      "loss": 1.7514,
+      "step": 4020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.641777777777778e-05,
+      "loss": 1.9658,
+      "step": 4030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.640888888888889e-05,
+      "loss": 1.7525,
+      "step": 4040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.64e-05,
+      "loss": 1.7224,
+      "step": 4050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6391111111111112e-05,
+      "loss": 1.888,
+      "step": 4060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6382222222222225e-05,
+      "loss": 1.851,
+      "step": 4070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6373333333333335e-05,
+      "loss": 1.815,
+      "step": 4080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6364444444444444e-05,
+      "loss": 1.8551,
+      "step": 4090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6355555555555557e-05,
+      "loss": 1.8181,
+      "step": 4100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6346666666666667e-05,
+      "loss": 1.889,
+      "step": 4110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.633777777777778e-05,
+      "loss": 1.7464,
+      "step": 4120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.632888888888889e-05,
+      "loss": 1.8543,
+      "step": 4130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.632e-05,
+      "loss": 1.8304,
+      "step": 4140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6311111111111113e-05,
+      "loss": 1.8297,
+      "step": 4150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6302222222222222e-05,
+      "loss": 1.8792,
+      "step": 4160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6293333333333335e-05,
+      "loss": 1.8107,
+      "step": 4170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6284444444444445e-05,
+      "loss": 1.744,
+      "step": 4180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6275555555555558e-05,
+      "loss": 1.8029,
+      "step": 4190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6266666666666668e-05,
+      "loss": 1.7973,
+      "step": 4200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6257777777777778e-05,
+      "loss": 1.854,
+      "step": 4210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.624888888888889e-05,
+      "loss": 1.8334,
+      "step": 4220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6240000000000004e-05,
+      "loss": 1.7622,
+      "step": 4230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6231111111111113e-05,
+      "loss": 1.826,
+      "step": 4240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6222222222222223e-05,
+      "loss": 1.7813,
+      "step": 4250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6213333333333333e-05,
+      "loss": 1.8041,
+      "step": 4260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6204444444444446e-05,
+      "loss": 1.7394,
+      "step": 4270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.619555555555556e-05,
+      "loss": 1.856,
+      "step": 4280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.618666666666667e-05,
+      "loss": 1.7458,
+      "step": 4290
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.617777777777778e-05,
+      "loss": 1.7687,
+      "step": 4300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6168888888888888e-05,
+      "loss": 1.8035,
+      "step": 4310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.616e-05,
+      "loss": 1.7987,
+      "step": 4320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6151111111111114e-05,
+      "loss": 1.7843,
+      "step": 4330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6142222222222224e-05,
+      "loss": 1.7192,
+      "step": 4340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6133333333333334e-05,
+      "loss": 1.7647,
+      "step": 4350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6124444444444443e-05,
+      "loss": 1.7249,
+      "step": 4360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6115555555555556e-05,
+      "loss": 1.7459,
+      "step": 4370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.610666666666667e-05,
+      "loss": 1.7646,
+      "step": 4380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.609777777777778e-05,
+      "loss": 1.9411,
+      "step": 4390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.608888888888889e-05,
+      "loss": 1.7212,
+      "step": 4400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6080000000000002e-05,
+      "loss": 1.8327,
+      "step": 4410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.607111111111111e-05,
+      "loss": 1.8452,
+      "step": 4420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6062222222222225e-05,
+      "loss": 1.7709,
+      "step": 4430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6053333333333334e-05,
+      "loss": 1.7413,
+      "step": 4440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6044444444444444e-05,
+      "loss": 1.7084,
+      "step": 4450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6035555555555557e-05,
+      "loss": 1.7451,
+      "step": 4460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6026666666666667e-05,
+      "loss": 1.8442,
+      "step": 4470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.601777777777778e-05,
+      "loss": 1.7965,
+      "step": 4480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.600888888888889e-05,
+      "loss": 1.7332,
+      "step": 4490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.8255,
+      "step": 4500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5991111111111112e-05,
+      "loss": 1.8302,
+      "step": 4510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5982222222222222e-05,
+      "loss": 1.7181,
+      "step": 4520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5973333333333335e-05,
+      "loss": 1.8474,
+      "step": 4530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5964444444444448e-05,
+      "loss": 1.7126,
+      "step": 4540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5955555555555558e-05,
+      "loss": 1.7683,
+      "step": 4550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5946666666666668e-05,
+      "loss": 1.8072,
+      "step": 4560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5937777777777777e-05,
+      "loss": 1.8631,
+      "step": 4570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.592888888888889e-05,
+      "loss": 1.8311,
+      "step": 4580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5920000000000003e-05,
+      "loss": 1.8399,
+      "step": 4590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5911111111111113e-05,
+      "loss": 1.8246,
+      "step": 4600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5902222222222223e-05,
+      "loss": 1.7888,
+      "step": 4610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5893333333333333e-05,
+      "loss": 1.8095,
+      "step": 4620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5884444444444446e-05,
+      "loss": 1.7484,
+      "step": 4630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.587555555555556e-05,
+      "loss": 1.8015,
+      "step": 4640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.586666666666667e-05,
+      "loss": 1.7268,
+      "step": 4650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5857777777777778e-05,
+      "loss": 1.7513,
+      "step": 4660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5848888888888888e-05,
+      "loss": 1.8132,
+      "step": 4670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.584e-05,
+      "loss": 1.8247,
+      "step": 4680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5831111111111114e-05,
+      "loss": 1.6942,
+      "step": 4690
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5822222222222224e-05,
+      "loss": 1.8149,
+      "step": 4700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5813333333333333e-05,
+      "loss": 1.7698,
+      "step": 4710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5804444444444446e-05,
+      "loss": 1.8132,
+      "step": 4720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5795555555555556e-05,
+      "loss": 1.7609,
+      "step": 4730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.578666666666667e-05,
+      "loss": 1.7464,
+      "step": 4740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.577777777777778e-05,
+      "loss": 1.797,
+      "step": 4750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.576888888888889e-05,
+      "loss": 1.8114,
+      "step": 4760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.576e-05,
+      "loss": 1.799,
+      "step": 4770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.575111111111111e-05,
+      "loss": 1.6768,
+      "step": 4780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5742222222222224e-05,
+      "loss": 1.8299,
+      "step": 4790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5733333333333334e-05,
+      "loss": 1.773,
+      "step": 4800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5724444444444447e-05,
+      "loss": 1.8671,
+      "step": 4810
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5715555555555557e-05,
+      "loss": 1.8164,
+      "step": 4820
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5706666666666666e-05,
+      "loss": 1.746,
+      "step": 4830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.569777777777778e-05,
+      "loss": 1.7676,
+      "step": 4840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5688888888888893e-05,
+      "loss": 1.7848,
+      "step": 4850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5680000000000002e-05,
+      "loss": 1.8719,
+      "step": 4860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5671111111111112e-05,
+      "loss": 1.7556,
+      "step": 4870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5662222222222222e-05,
+      "loss": 1.7636,
+      "step": 4880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5653333333333335e-05,
+      "loss": 1.7625,
+      "step": 4890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5644444444444448e-05,
+      "loss": 1.7795,
+      "step": 4900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5635555555555558e-05,
+      "loss": 1.7973,
+      "step": 4910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5626666666666667e-05,
+      "loss": 1.8464,
+      "step": 4920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5617777777777777e-05,
+      "loss": 1.7456,
+      "step": 4930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.560888888888889e-05,
+      "loss": 1.7502,
+      "step": 4940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5600000000000003e-05,
+      "loss": 1.7905,
+      "step": 4950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5591111111111113e-05,
+      "loss": 1.8346,
+      "step": 4960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5582222222222222e-05,
+      "loss": 1.8643,
+      "step": 4970
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5573333333333332e-05,
+      "loss": 1.8001,
+      "step": 4980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5564444444444445e-05,
+      "loss": 1.8832,
+      "step": 4990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.555555555555556e-05,
+      "loss": 1.7317,
+      "step": 5000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5546666666666668e-05,
+      "loss": 1.7269,
+      "step": 5010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5537777777777778e-05,
+      "loss": 1.8316,
+      "step": 5020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.552888888888889e-05,
+      "loss": 1.7392,
+      "step": 5030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.552e-05,
+      "loss": 1.7442,
+      "step": 5040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5511111111111114e-05,
+      "loss": 1.7217,
+      "step": 5050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5502222222222223e-05,
+      "loss": 1.7569,
+      "step": 5060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5493333333333333e-05,
+      "loss": 1.8681,
+      "step": 5070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5484444444444446e-05,
+      "loss": 1.663,
+      "step": 5080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5475555555555556e-05,
+      "loss": 1.7667,
+      "step": 5090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.546666666666667e-05,
+      "loss": 1.8662,
+      "step": 5100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.545777777777778e-05,
+      "loss": 1.8021,
+      "step": 5110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.544888888888889e-05,
+      "loss": 1.8165,
+      "step": 5120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.544e-05,
+      "loss": 1.6853,
+      "step": 5130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.543111111111111e-05,
+      "loss": 1.8108,
+      "step": 5140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5422222222222224e-05,
+      "loss": 1.7975,
+      "step": 5150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5413333333333337e-05,
+      "loss": 1.7447,
+      "step": 5160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5404444444444447e-05,
+      "loss": 1.8241,
+      "step": 5170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5395555555555556e-05,
+      "loss": 1.7634,
+      "step": 5180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5386666666666666e-05,
+      "loss": 1.8376,
+      "step": 5190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.537777777777778e-05,
+      "loss": 1.7088,
+      "step": 5200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5368888888888892e-05,
+      "loss": 1.8199,
+      "step": 5210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5360000000000002e-05,
+      "loss": 1.801,
+      "step": 5220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5351111111111112e-05,
+      "loss": 1.727,
+      "step": 5230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.534222222222222e-05,
+      "loss": 1.8676,
+      "step": 5240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5333333333333334e-05,
+      "loss": 1.8062,
+      "step": 5250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5324444444444448e-05,
+      "loss": 1.7656,
+      "step": 5260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5315555555555557e-05,
+      "loss": 1.7274,
+      "step": 5270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5306666666666667e-05,
+      "loss": 1.7541,
+      "step": 5280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5297777777777777e-05,
+      "loss": 1.8898,
+      "step": 5290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.528888888888889e-05,
+      "loss": 1.7473,
+      "step": 5300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5280000000000003e-05,
+      "loss": 1.7611,
+      "step": 5310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5271111111111112e-05,
+      "loss": 1.7796,
+      "step": 5320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5262222222222222e-05,
+      "loss": 1.8352,
+      "step": 5330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5253333333333335e-05,
+      "loss": 1.7571,
+      "step": 5340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5244444444444447e-05,
+      "loss": 1.7671,
+      "step": 5350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5235555555555556e-05,
+      "loss": 1.7342,
+      "step": 5360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5226666666666668e-05,
+      "loss": 1.8842,
+      "step": 5370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5217777777777777e-05,
+      "loss": 1.6502,
+      "step": 5380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.520888888888889e-05,
+      "loss": 1.6959,
+      "step": 5390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 1.8415,
+      "step": 5400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5191111111111112e-05,
+      "loss": 1.7555,
+      "step": 5410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5182222222222223e-05,
+      "loss": 1.7754,
+      "step": 5420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5173333333333336e-05,
+      "loss": 1.747,
+      "step": 5430
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5164444444444446e-05,
+      "loss": 1.8642,
+      "step": 5440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5155555555555557e-05,
+      "loss": 1.8813,
+      "step": 5450
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5146666666666667e-05,
+      "loss": 1.8394,
+      "step": 5460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.513777777777778e-05,
+      "loss": 1.7536,
+      "step": 5470
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5128888888888891e-05,
+      "loss": 1.6355,
+      "step": 5480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5120000000000001e-05,
+      "loss": 1.7599,
+      "step": 5490
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5111111111111112e-05,
+      "loss": 1.8124,
+      "step": 5500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5102222222222222e-05,
+      "loss": 1.7518,
+      "step": 5510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5093333333333335e-05,
+      "loss": 1.7243,
+      "step": 5520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5084444444444446e-05,
+      "loss": 1.7129,
+      "step": 5530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5075555555555556e-05,
+      "loss": 1.7691,
+      "step": 5540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5066666666666668e-05,
+      "loss": 1.7522,
+      "step": 5550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.505777777777778e-05,
+      "loss": 1.7405,
+      "step": 5560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.504888888888889e-05,
+      "loss": 1.873,
+      "step": 5570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5040000000000002e-05,
+      "loss": 1.654,
+      "step": 5580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5031111111111111e-05,
+      "loss": 1.7372,
+      "step": 5590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5022222222222223e-05,
+      "loss": 1.7489,
+      "step": 5600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5013333333333336e-05,
+      "loss": 1.72,
+      "step": 5610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5004444444444446e-05,
+      "loss": 1.775,
+      "step": 5620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4995555555555557e-05,
+      "loss": 1.7491,
+      "step": 5630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4986666666666667e-05,
+      "loss": 1.8408,
+      "step": 5640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.497777777777778e-05,
+      "loss": 1.7869,
+      "step": 5650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4968888888888891e-05,
+      "loss": 1.6827,
+      "step": 5660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.496e-05,
+      "loss": 1.8255,
+      "step": 5670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4951111111111112e-05,
+      "loss": 1.8083,
+      "step": 5680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4942222222222222e-05,
+      "loss": 1.8452,
+      "step": 5690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4933333333333335e-05,
+      "loss": 1.6767,
+      "step": 5700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4924444444444446e-05,
+      "loss": 1.7386,
+      "step": 5710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4915555555555556e-05,
+      "loss": 1.7684,
+      "step": 5720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4906666666666667e-05,
+      "loss": 1.9271,
+      "step": 5730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.489777777777778e-05,
+      "loss": 1.7283,
+      "step": 5740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.488888888888889e-05,
+      "loss": 1.7687,
+      "step": 5750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4880000000000002e-05,
+      "loss": 1.7687,
+      "step": 5760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4871111111111111e-05,
+      "loss": 1.6546,
+      "step": 5770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4862222222222223e-05,
+      "loss": 1.789,
+      "step": 5780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4853333333333336e-05,
+      "loss": 1.6584,
+      "step": 5790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4844444444444445e-05,
+      "loss": 1.7437,
+      "step": 5800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4835555555555557e-05,
+      "loss": 1.7673,
+      "step": 5810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4826666666666666e-05,
+      "loss": 1.7488,
+      "step": 5820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.481777777777778e-05,
+      "loss": 1.692,
+      "step": 5830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4808888888888891e-05,
+      "loss": 1.7309,
+      "step": 5840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.48e-05,
+      "loss": 1.8143,
+      "step": 5850
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4791111111111112e-05,
+      "loss": 1.7716,
+      "step": 5860
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4782222222222225e-05,
+      "loss": 1.7321,
+      "step": 5870
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4773333333333335e-05,
+      "loss": 1.7479,
+      "step": 5880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4764444444444446e-05,
+      "loss": 1.6669,
+      "step": 5890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4755555555555556e-05,
+      "loss": 1.6983,
+      "step": 5900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4746666666666667e-05,
+      "loss": 1.816,
+      "step": 5910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.473777777777778e-05,
+      "loss": 1.7526,
+      "step": 5920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.472888888888889e-05,
+      "loss": 1.8571,
+      "step": 5930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4720000000000001e-05,
+      "loss": 1.7855,
+      "step": 5940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4711111111111111e-05,
+      "loss": 1.7267,
+      "step": 5950
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4702222222222224e-05,
+      "loss": 1.86,
+      "step": 5960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4693333333333336e-05,
+      "loss": 1.7203,
+      "step": 5970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4684444444444445e-05,
+      "loss": 1.8028,
+      "step": 5980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4675555555555557e-05,
+      "loss": 1.7889,
+      "step": 5990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 1.7077,
+      "step": 6000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.465777777777778e-05,
+      "loss": 1.73,
+      "step": 6010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.464888888888889e-05,
+      "loss": 1.663,
+      "step": 6020
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.464e-05,
+      "loss": 1.825,
+      "step": 6030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4631111111111112e-05,
+      "loss": 1.7473,
+      "step": 6040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4622222222222225e-05,
+      "loss": 1.6385,
+      "step": 6050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4613333333333335e-05,
+      "loss": 1.7819,
+      "step": 6060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4604444444444446e-05,
+      "loss": 1.6604,
+      "step": 6070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4595555555555556e-05,
+      "loss": 1.6817,
+      "step": 6080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4586666666666667e-05,
+      "loss": 1.865,
+      "step": 6090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.457777777777778e-05,
+      "loss": 1.7235,
+      "step": 6100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.456888888888889e-05,
+      "loss": 1.7744,
+      "step": 6110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4560000000000001e-05,
+      "loss": 1.7588,
+      "step": 6120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4551111111111111e-05,
+      "loss": 1.8098,
+      "step": 6130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4542222222222224e-05,
+      "loss": 1.7252,
+      "step": 6140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4533333333333335e-05,
+      "loss": 1.6818,
+      "step": 6150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4524444444444445e-05,
+      "loss": 1.7926,
+      "step": 6160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4515555555555556e-05,
+      "loss": 1.6481,
+      "step": 6170
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.450666666666667e-05,
+      "loss": 1.8435,
+      "step": 6180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.449777777777778e-05,
+      "loss": 1.8226,
+      "step": 6190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.448888888888889e-05,
+      "loss": 1.7548,
+      "step": 6200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.448e-05,
+      "loss": 1.7843,
+      "step": 6210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4471111111111112e-05,
+      "loss": 1.7895,
+      "step": 6220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4462222222222225e-05,
+      "loss": 1.8373,
+      "step": 6230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4453333333333334e-05,
+      "loss": 1.6512,
+      "step": 6240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4444444444444446e-05,
+      "loss": 1.7296,
+      "step": 6250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4435555555555556e-05,
+      "loss": 1.6636,
+      "step": 6260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4426666666666669e-05,
+      "loss": 1.7419,
+      "step": 6270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.441777777777778e-05,
+      "loss": 1.8304,
+      "step": 6280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.440888888888889e-05,
+      "loss": 1.8009,
+      "step": 6290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 1.6717,
+      "step": 6300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.439111111111111e-05,
+      "loss": 1.9496,
+      "step": 6310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4382222222222224e-05,
+      "loss": 1.7164,
+      "step": 6320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4373333333333335e-05,
+      "loss": 1.658,
+      "step": 6330
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4364444444444445e-05,
+      "loss": 1.79,
+      "step": 6340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4355555555555556e-05,
+      "loss": 1.6781,
+      "step": 6350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.434666666666667e-05,
+      "loss": 1.6198,
+      "step": 6360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4337777777777779e-05,
+      "loss": 1.8389,
+      "step": 6370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.432888888888889e-05,
+      "loss": 1.8507,
+      "step": 6380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.432e-05,
+      "loss": 1.7003,
+      "step": 6390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4311111111111111e-05,
+      "loss": 1.7742,
+      "step": 6400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4302222222222225e-05,
+      "loss": 1.6982,
+      "step": 6410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4293333333333334e-05,
+      "loss": 1.7541,
+      "step": 6420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4284444444444446e-05,
+      "loss": 1.787,
+      "step": 6430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4275555555555555e-05,
+      "loss": 1.7205,
+      "step": 6440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4266666666666668e-05,
+      "loss": 1.7482,
+      "step": 6450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.425777777777778e-05,
+      "loss": 1.8702,
+      "step": 6460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.424888888888889e-05,
+      "loss": 1.7204,
+      "step": 6470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4240000000000001e-05,
+      "loss": 1.8014,
+      "step": 6480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4231111111111114e-05,
+      "loss": 1.7835,
+      "step": 6490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4222222222222224e-05,
+      "loss": 1.8101,
+      "step": 6500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4213333333333335e-05,
+      "loss": 1.8082,
+      "step": 6510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4204444444444445e-05,
+      "loss": 1.7988,
+      "step": 6520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4195555555555556e-05,
+      "loss": 1.7512,
+      "step": 6530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.418666666666667e-05,
+      "loss": 1.6861,
+      "step": 6540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4177777777777779e-05,
+      "loss": 1.7465,
+      "step": 6550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.416888888888889e-05,
+      "loss": 1.737,
+      "step": 6560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.416e-05,
+      "loss": 1.7074,
+      "step": 6570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4151111111111113e-05,
+      "loss": 1.7245,
+      "step": 6580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4142222222222224e-05,
+      "loss": 1.796,
+      "step": 6590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4133333333333334e-05,
+      "loss": 1.7493,
+      "step": 6600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4124444444444445e-05,
+      "loss": 1.7983,
+      "step": 6610
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4115555555555555e-05,
+      "loss": 1.7788,
+      "step": 6620
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4106666666666668e-05,
+      "loss": 1.7919,
+      "step": 6630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.409777777777778e-05,
+      "loss": 1.699,
+      "step": 6640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.408888888888889e-05,
+      "loss": 1.6976,
+      "step": 6650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.408e-05,
+      "loss": 1.8112,
+      "step": 6660
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4071111111111114e-05,
+      "loss": 1.7369,
+      "step": 6670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4062222222222223e-05,
+      "loss": 1.8199,
+      "step": 6680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4053333333333335e-05,
+      "loss": 1.7861,
+      "step": 6690
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4044444444444445e-05,
+      "loss": 1.6931,
+      "step": 6700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4035555555555556e-05,
+      "loss": 1.6584,
+      "step": 6710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4026666666666669e-05,
+      "loss": 1.6932,
+      "step": 6720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4017777777777779e-05,
+      "loss": 1.6603,
+      "step": 6730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.400888888888889e-05,
+      "loss": 1.8059,
+      "step": 6740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4e-05,
+      "loss": 1.8255,
+      "step": 6750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3991111111111113e-05,
+      "loss": 1.7258,
+      "step": 6760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3982222222222224e-05,
+      "loss": 1.6449,
+      "step": 6770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3973333333333334e-05,
+      "loss": 1.7384,
+      "step": 6780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3964444444444445e-05,
+      "loss": 1.735,
+      "step": 6790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3955555555555558e-05,
+      "loss": 1.7411,
+      "step": 6800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3946666666666668e-05,
+      "loss": 1.8766,
+      "step": 6810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.393777777777778e-05,
+      "loss": 1.7621,
+      "step": 6820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.392888888888889e-05,
+      "loss": 1.7103,
+      "step": 6830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.392e-05,
+      "loss": 1.6966,
+      "step": 6840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3911111111111114e-05,
+      "loss": 1.8009,
+      "step": 6850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3902222222222223e-05,
+      "loss": 1.7512,
+      "step": 6860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3893333333333335e-05,
+      "loss": 1.709,
+      "step": 6870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3884444444444444e-05,
+      "loss": 1.7806,
+      "step": 6880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3875555555555557e-05,
+      "loss": 1.7553,
+      "step": 6890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3866666666666669e-05,
+      "loss": 1.8478,
+      "step": 6900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3857777777777779e-05,
+      "loss": 1.662,
+      "step": 6910
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.384888888888889e-05,
+      "loss": 1.6461,
+      "step": 6920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.384e-05,
+      "loss": 1.6806,
+      "step": 6930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3831111111111113e-05,
+      "loss": 1.7531,
+      "step": 6940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3822222222222224e-05,
+      "loss": 1.6372,
+      "step": 6950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3813333333333334e-05,
+      "loss": 1.7367,
+      "step": 6960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3804444444444445e-05,
+      "loss": 1.7171,
+      "step": 6970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3795555555555558e-05,
+      "loss": 1.8186,
+      "step": 6980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3786666666666668e-05,
+      "loss": 1.6357,
+      "step": 6990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.377777777777778e-05,
+      "loss": 1.7448,
+      "step": 7000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3768888888888889e-05,
+      "loss": 1.793,
+      "step": 7010
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.376e-05,
+      "loss": 1.8142,
+      "step": 7020
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3751111111111113e-05,
+      "loss": 1.6924,
+      "step": 7030
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3742222222222223e-05,
+      "loss": 1.7925,
+      "step": 7040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3733333333333335e-05,
+      "loss": 1.7615,
+      "step": 7050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3724444444444444e-05,
+      "loss": 1.6352,
+      "step": 7060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3715555555555557e-05,
+      "loss": 1.6845,
+      "step": 7070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3706666666666669e-05,
+      "loss": 1.792,
+      "step": 7080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3697777777777778e-05,
+      "loss": 1.7692,
+      "step": 7090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.368888888888889e-05,
+      "loss": 1.7179,
+      "step": 7100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3680000000000003e-05,
+      "loss": 1.7061,
+      "step": 7110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3671111111111113e-05,
+      "loss": 1.8235,
+      "step": 7120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3662222222222224e-05,
+      "loss": 1.7675,
+      "step": 7130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3653333333333334e-05,
+      "loss": 1.754,
+      "step": 7140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3644444444444445e-05,
+      "loss": 1.7446,
+      "step": 7150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3635555555555558e-05,
+      "loss": 1.8608,
+      "step": 7160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3626666666666668e-05,
+      "loss": 1.7654,
+      "step": 7170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.361777777777778e-05,
+      "loss": 1.6608,
+      "step": 7180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3608888888888889e-05,
+      "loss": 1.6793,
+      "step": 7190
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 1.7995,
+      "step": 7200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3591111111111113e-05,
+      "loss": 1.6848,
+      "step": 7210
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3582222222222223e-05,
+      "loss": 1.7332,
+      "step": 7220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3573333333333334e-05,
+      "loss": 1.783,
+      "step": 7230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3564444444444444e-05,
+      "loss": 1.6675,
+      "step": 7240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3555555555555557e-05,
+      "loss": 1.6632,
+      "step": 7250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3546666666666669e-05,
+      "loss": 1.7886,
+      "step": 7260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3537777777777778e-05,
+      "loss": 1.8071,
+      "step": 7270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.352888888888889e-05,
+      "loss": 1.9045,
+      "step": 7280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3520000000000003e-05,
+      "loss": 1.7632,
+      "step": 7290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3511111111111112e-05,
+      "loss": 1.7776,
+      "step": 7300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3502222222222224e-05,
+      "loss": 1.8126,
+      "step": 7310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3493333333333333e-05,
+      "loss": 1.7974,
+      "step": 7320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3484444444444445e-05,
+      "loss": 1.6832,
+      "step": 7330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3475555555555558e-05,
+      "loss": 1.7427,
+      "step": 7340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3466666666666668e-05,
+      "loss": 1.6547,
+      "step": 7350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3457777777777779e-05,
+      "loss": 1.6917,
+      "step": 7360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3448888888888889e-05,
+      "loss": 1.6864,
+      "step": 7370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3440000000000002e-05,
+      "loss": 1.6384,
+      "step": 7380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3431111111111113e-05,
+      "loss": 1.7144,
+      "step": 7390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3422222222222223e-05,
+      "loss": 1.7891,
+      "step": 7400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3413333333333334e-05,
+      "loss": 1.734,
+      "step": 7410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3404444444444447e-05,
+      "loss": 1.6721,
+      "step": 7420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3395555555555557e-05,
+      "loss": 1.8492,
+      "step": 7430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3386666666666668e-05,
+      "loss": 1.7149,
+      "step": 7440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3377777777777778e-05,
+      "loss": 1.8493,
+      "step": 7450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.336888888888889e-05,
+      "loss": 1.8174,
+      "step": 7460
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3360000000000003e-05,
+      "loss": 1.8089,
+      "step": 7470
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3351111111111112e-05,
+      "loss": 1.7008,
+      "step": 7480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3342222222222224e-05,
+      "loss": 1.6404,
+      "step": 7490
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.7202,
+      "step": 7500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3324444444444446e-05,
+      "loss": 1.4175,
+      "step": 7510
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3315555555555558e-05,
+      "loss": 1.3014,
+      "step": 7520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3306666666666667e-05,
+      "loss": 1.3302,
+      "step": 7530
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3297777777777779e-05,
+      "loss": 1.3212,
+      "step": 7540
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3288888888888889e-05,
+      "loss": 1.2417,
+      "step": 7550
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3280000000000002e-05,
+      "loss": 1.318,
+      "step": 7560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3271111111111113e-05,
+      "loss": 1.3119,
+      "step": 7570
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3262222222222223e-05,
+      "loss": 1.3151,
+      "step": 7580
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3253333333333334e-05,
+      "loss": 1.2873,
+      "step": 7590
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3244444444444447e-05,
+      "loss": 1.3266,
+      "step": 7600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3235555555555557e-05,
+      "loss": 1.3181,
+      "step": 7610
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3226666666666668e-05,
+      "loss": 1.2659,
+      "step": 7620
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3217777777777778e-05,
+      "loss": 1.2871,
+      "step": 7630
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.320888888888889e-05,
+      "loss": 1.3292,
+      "step": 7640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3200000000000002e-05,
+      "loss": 1.2704,
+      "step": 7650
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3191111111111112e-05,
+      "loss": 1.2996,
+      "step": 7660
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3182222222222223e-05,
+      "loss": 1.3049,
+      "step": 7670
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3173333333333333e-05,
+      "loss": 1.2368,
+      "step": 7680
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3164444444444446e-05,
+      "loss": 1.2997,
+      "step": 7690
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3155555555555558e-05,
+      "loss": 1.2688,
+      "step": 7700
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3146666666666667e-05,
+      "loss": 1.2414,
+      "step": 7710
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3137777777777779e-05,
+      "loss": 1.2034,
+      "step": 7720
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3128888888888892e-05,
+      "loss": 1.3935,
+      "step": 7730
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3120000000000001e-05,
+      "loss": 1.3119,
+      "step": 7740
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3111111111111113e-05,
+      "loss": 1.2798,
+      "step": 7750
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3102222222222223e-05,
+      "loss": 1.2745,
+      "step": 7760
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3093333333333334e-05,
+      "loss": 1.2346,
+      "step": 7770
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3084444444444447e-05,
+      "loss": 1.2624,
+      "step": 7780
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3075555555555557e-05,
+      "loss": 1.314,
+      "step": 7790
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3066666666666668e-05,
+      "loss": 1.1897,
+      "step": 7800
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3057777777777778e-05,
+      "loss": 1.3161,
+      "step": 7810
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.304888888888889e-05,
+      "loss": 1.1974,
+      "step": 7820
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3040000000000002e-05,
+      "loss": 1.2967,
+      "step": 7830
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3031111111111112e-05,
+      "loss": 1.2904,
+      "step": 7840
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3022222222222223e-05,
+      "loss": 1.2711,
+      "step": 7850
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3013333333333333e-05,
+      "loss": 1.3287,
+      "step": 7860
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3004444444444446e-05,
+      "loss": 1.2833,
+      "step": 7870
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2995555555555557e-05,
+      "loss": 1.2644,
+      "step": 7880
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2986666666666667e-05,
+      "loss": 1.2906,
+      "step": 7890
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2977777777777779e-05,
+      "loss": 1.3713,
+      "step": 7900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2968888888888892e-05,
+      "loss": 1.2543,
+      "step": 7910
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2960000000000001e-05,
+      "loss": 1.3059,
+      "step": 7920
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2951111111111113e-05,
+      "loss": 1.2611,
+      "step": 7930
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2942222222222222e-05,
+      "loss": 1.3628,
+      "step": 7940
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2933333333333334e-05,
+      "loss": 1.3252,
+      "step": 7950
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2924444444444447e-05,
+      "loss": 1.3181,
+      "step": 7960
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2915555555555557e-05,
+      "loss": 1.3307,
+      "step": 7970
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2906666666666668e-05,
+      "loss": 1.3147,
+      "step": 7980
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2897777777777778e-05,
+      "loss": 1.4119,
+      "step": 7990
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.288888888888889e-05,
+      "loss": 1.3296,
+      "step": 8000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2880000000000002e-05,
+      "loss": 1.1675,
+      "step": 8010
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2871111111111112e-05,
+      "loss": 1.2783,
+      "step": 8020
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2862222222222223e-05,
+      "loss": 1.2509,
+      "step": 8030
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2853333333333336e-05,
+      "loss": 1.2814,
+      "step": 8040
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2844444444444446e-05,
+      "loss": 1.3343,
+      "step": 8050
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2835555555555557e-05,
+      "loss": 1.2097,
+      "step": 8060
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2826666666666667e-05,
+      "loss": 1.3686,
+      "step": 8070
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2817777777777778e-05,
+      "loss": 1.3122,
+      "step": 8080
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2808888888888891e-05,
+      "loss": 1.2564,
+      "step": 8090
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 1.1614,
+      "step": 8100
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2791111111111112e-05,
+      "loss": 1.391,
+      "step": 8110
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2782222222222222e-05,
+      "loss": 1.2443,
+      "step": 8120
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2773333333333335e-05,
+      "loss": 1.3692,
+      "step": 8130
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2764444444444447e-05,
+      "loss": 1.234,
+      "step": 8140
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2755555555555556e-05,
+      "loss": 1.3177,
+      "step": 8150
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2746666666666668e-05,
+      "loss": 1.3273,
+      "step": 8160
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2737777777777777e-05,
+      "loss": 1.4376,
+      "step": 8170
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.272888888888889e-05,
+      "loss": 1.3564,
+      "step": 8180
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2720000000000002e-05,
+      "loss": 1.409,
+      "step": 8190
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2711111111111112e-05,
+      "loss": 1.2854,
+      "step": 8200
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2702222222222223e-05,
+      "loss": 1.177,
+      "step": 8210
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2693333333333336e-05,
+      "loss": 1.2595,
+      "step": 8220
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2684444444444446e-05,
+      "loss": 1.3414,
+      "step": 8230
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2675555555555557e-05,
+      "loss": 1.4175,
+      "step": 8240
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2666666666666667e-05,
+      "loss": 1.2388,
+      "step": 8250
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2657777777777778e-05,
+      "loss": 1.262,
+      "step": 8260
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2648888888888891e-05,
+      "loss": 1.2163,
+      "step": 8270
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2640000000000001e-05,
+      "loss": 1.2121,
+      "step": 8280
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2631111111111112e-05,
+      "loss": 1.3183,
+      "step": 8290
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2622222222222222e-05,
+      "loss": 1.2941,
+      "step": 8300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2613333333333335e-05,
+      "loss": 1.3029,
+      "step": 8310
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2604444444444446e-05,
+      "loss": 1.2604,
+      "step": 8320
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2595555555555556e-05,
+      "loss": 1.2472,
+      "step": 8330
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2586666666666668e-05,
+      "loss": 1.3085,
+      "step": 8340
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.257777777777778e-05,
+      "loss": 1.2261,
+      "step": 8350
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.256888888888889e-05,
+      "loss": 1.2843,
+      "step": 8360
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2560000000000002e-05,
+      "loss": 1.2629,
+      "step": 8370
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2551111111111111e-05,
+      "loss": 1.3116,
+      "step": 8380
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2542222222222223e-05,
+      "loss": 1.3742,
+      "step": 8390
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2533333333333336e-05,
+      "loss": 1.2678,
+      "step": 8400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2524444444444446e-05,
+      "loss": 1.343,
+      "step": 8410
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2515555555555557e-05,
+      "loss": 1.2973,
+      "step": 8420
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2506666666666667e-05,
+      "loss": 1.3211,
+      "step": 8430
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.249777777777778e-05,
+      "loss": 1.3729,
+      "step": 8440
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2488888888888891e-05,
+      "loss": 1.2127,
+      "step": 8450
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.248e-05,
+      "loss": 1.3223,
+      "step": 8460
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2471111111111112e-05,
+      "loss": 1.2519,
+      "step": 8470
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2462222222222222e-05,
+      "loss": 1.2051,
+      "step": 8480
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2453333333333335e-05,
+      "loss": 1.274,
+      "step": 8490
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2444444444444446e-05,
+      "loss": 1.2387,
+      "step": 8500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2435555555555556e-05,
+      "loss": 1.2816,
+      "step": 8510
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2426666666666667e-05,
+      "loss": 1.2443,
+      "step": 8520
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.241777777777778e-05,
+      "loss": 1.2571,
+      "step": 8530
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.240888888888889e-05,
+      "loss": 1.3689,
+      "step": 8540
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2400000000000002e-05,
+      "loss": 1.2634,
+      "step": 8550
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2391111111111111e-05,
+      "loss": 1.3699,
+      "step": 8560
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2382222222222223e-05,
+      "loss": 1.3579,
+      "step": 8570
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2373333333333336e-05,
+      "loss": 1.3534,
+      "step": 8580
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2364444444444445e-05,
+      "loss": 1.32,
+      "step": 8590
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2355555555555557e-05,
+      "loss": 1.228,
+      "step": 8600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2346666666666666e-05,
+      "loss": 1.1653,
+      "step": 8610
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.233777777777778e-05,
+      "loss": 1.3089,
+      "step": 8620
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2328888888888891e-05,
+      "loss": 1.3228,
+      "step": 8630
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.232e-05,
+      "loss": 1.3981,
+      "step": 8640
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2311111111111112e-05,
+      "loss": 1.3072,
+      "step": 8650
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2302222222222225e-05,
+      "loss": 1.3044,
+      "step": 8660
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2293333333333335e-05,
+      "loss": 1.243,
+      "step": 8670
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2284444444444446e-05,
+      "loss": 1.2722,
+      "step": 8680
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2275555555555556e-05,
+      "loss": 1.3212,
+      "step": 8690
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2266666666666667e-05,
+      "loss": 1.3121,
+      "step": 8700
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.225777777777778e-05,
+      "loss": 1.3273,
+      "step": 8710
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.224888888888889e-05,
+      "loss": 1.3024,
+      "step": 8720
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2240000000000001e-05,
+      "loss": 1.3419,
+      "step": 8730
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2231111111111111e-05,
+      "loss": 1.3019,
+      "step": 8740
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2222222222222224e-05,
+      "loss": 1.2113,
+      "step": 8750
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2213333333333336e-05,
+      "loss": 1.3088,
+      "step": 8760
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2204444444444445e-05,
+      "loss": 1.3598,
+      "step": 8770
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2195555555555557e-05,
+      "loss": 1.3405,
+      "step": 8780
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2186666666666666e-05,
+      "loss": 1.2585,
+      "step": 8790
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.217777777777778e-05,
+      "loss": 1.3278,
+      "step": 8800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.216888888888889e-05,
+      "loss": 1.3222,
+      "step": 8810
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.216e-05,
+      "loss": 1.2708,
+      "step": 8820
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2151111111111112e-05,
+      "loss": 1.283,
+      "step": 8830
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2142222222222225e-05,
+      "loss": 1.3328,
+      "step": 8840
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2133333333333335e-05,
+      "loss": 1.3816,
+      "step": 8850
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2124444444444446e-05,
+      "loss": 1.3146,
+      "step": 8860
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2115555555555556e-05,
+      "loss": 1.3597,
+      "step": 8870
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2106666666666667e-05,
+      "loss": 1.3237,
+      "step": 8880
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.209777777777778e-05,
+      "loss": 1.3415,
+      "step": 8890
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.208888888888889e-05,
+      "loss": 1.2728,
+      "step": 8900
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2080000000000001e-05,
+      "loss": 1.3427,
+      "step": 8910
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2071111111111111e-05,
+      "loss": 1.3692,
+      "step": 8920
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2062222222222224e-05,
+      "loss": 1.2415,
+      "step": 8930
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2053333333333335e-05,
+      "loss": 1.2817,
+      "step": 8940
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2044444444444445e-05,
+      "loss": 1.3986,
+      "step": 8950
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2035555555555556e-05,
+      "loss": 1.2345,
+      "step": 8960
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.202666666666667e-05,
+      "loss": 1.3971,
+      "step": 8970
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.201777777777778e-05,
+      "loss": 1.3981,
+      "step": 8980
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.200888888888889e-05,
+      "loss": 1.2921,
+      "step": 8990
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2e-05,
+      "loss": 1.2349,
+      "step": 9000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1991111111111112e-05,
+      "loss": 1.3524,
+      "step": 9010
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1982222222222225e-05,
+      "loss": 1.347,
+      "step": 9020
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1973333333333334e-05,
+      "loss": 1.2996,
+      "step": 9030
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1964444444444446e-05,
+      "loss": 1.395,
+      "step": 9040
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1955555555555556e-05,
+      "loss": 1.3767,
+      "step": 9050
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1946666666666669e-05,
+      "loss": 1.2676,
+      "step": 9060
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.193777777777778e-05,
+      "loss": 1.2499,
+      "step": 9070
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.192888888888889e-05,
+      "loss": 1.2946,
+      "step": 9080
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1920000000000001e-05,
+      "loss": 1.2465,
+      "step": 9090
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.191111111111111e-05,
+      "loss": 1.3125,
+      "step": 9100
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1902222222222224e-05,
+      "loss": 1.2732,
+      "step": 9110
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1893333333333335e-05,
+      "loss": 1.3307,
+      "step": 9120
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1884444444444445e-05,
+      "loss": 1.2492,
+      "step": 9130
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1875555555555556e-05,
+      "loss": 1.241,
+      "step": 9140
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.186666666666667e-05,
+      "loss": 1.2814,
+      "step": 9150
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1857777777777779e-05,
+      "loss": 1.2771,
+      "step": 9160
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.184888888888889e-05,
+      "loss": 1.2252,
+      "step": 9170
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.184e-05,
+      "loss": 1.3893,
+      "step": 9180
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1831111111111112e-05,
+      "loss": 1.2163,
+      "step": 9190
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1822222222222225e-05,
+      "loss": 1.2887,
+      "step": 9200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1813333333333334e-05,
+      "loss": 1.2959,
+      "step": 9210
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1804444444444446e-05,
+      "loss": 1.2892,
+      "step": 9220
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1795555555555555e-05,
+      "loss": 1.2917,
+      "step": 9230
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1786666666666668e-05,
+      "loss": 1.4003,
+      "step": 9240
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.177777777777778e-05,
+      "loss": 1.3751,
+      "step": 9250
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.176888888888889e-05,
+      "loss": 1.4375,
+      "step": 9260
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1760000000000001e-05,
+      "loss": 1.3405,
+      "step": 9270
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1751111111111112e-05,
+      "loss": 1.2773,
+      "step": 9280
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1742222222222224e-05,
+      "loss": 1.288,
+      "step": 9290
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1733333333333335e-05,
+      "loss": 1.2501,
+      "step": 9300
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1724444444444445e-05,
+      "loss": 1.2706,
+      "step": 9310
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1715555555555556e-05,
+      "loss": 1.3799,
+      "step": 9320
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.170666666666667e-05,
+      "loss": 1.2151,
+      "step": 9330
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1697777777777779e-05,
+      "loss": 1.3166,
+      "step": 9340
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.168888888888889e-05,
+      "loss": 1.3336,
+      "step": 9350
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.168e-05,
+      "loss": 1.2973,
+      "step": 9360
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1671111111111113e-05,
+      "loss": 1.3714,
+      "step": 9370
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1662222222222224e-05,
+      "loss": 1.2007,
+      "step": 9380
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1653333333333334e-05,
+      "loss": 1.3166,
+      "step": 9390
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1644444444444446e-05,
+      "loss": 1.3583,
+      "step": 9400
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1635555555555555e-05,
+      "loss": 1.2303,
+      "step": 9410
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1626666666666668e-05,
+      "loss": 1.347,
+      "step": 9420
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.161777777777778e-05,
+      "loss": 1.289,
+      "step": 9430
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.160888888888889e-05,
+      "loss": 1.235,
+      "step": 9440
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.16e-05,
+      "loss": 1.345,
+      "step": 9450
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1591111111111114e-05,
+      "loss": 1.3017,
+      "step": 9460
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1582222222222224e-05,
+      "loss": 1.4081,
+      "step": 9470
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1573333333333335e-05,
+      "loss": 1.3683,
+      "step": 9480
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1564444444444445e-05,
+      "loss": 1.3107,
+      "step": 9490
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1555555555555556e-05,
+      "loss": 1.356,
+      "step": 9500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1546666666666669e-05,
+      "loss": 1.2774,
+      "step": 9510
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1537777777777779e-05,
+      "loss": 1.3332,
+      "step": 9520
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.152888888888889e-05,
+      "loss": 1.261,
+      "step": 9530
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.152e-05,
+      "loss": 1.3417,
+      "step": 9540
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1511111111111113e-05,
+      "loss": 1.3769,
+      "step": 9550
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1502222222222224e-05,
+      "loss": 1.3708,
+      "step": 9560
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1493333333333334e-05,
+      "loss": 1.2925,
+      "step": 9570
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1484444444444445e-05,
+      "loss": 1.2638,
+      "step": 9580
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1475555555555557e-05,
+      "loss": 1.2503,
+      "step": 9590
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1466666666666668e-05,
+      "loss": 1.2877,
+      "step": 9600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.145777777777778e-05,
+      "loss": 1.3441,
+      "step": 9610
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.144888888888889e-05,
+      "loss": 1.3011,
+      "step": 9620
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.144e-05,
+      "loss": 1.2564,
+      "step": 9630
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1431111111111114e-05,
+      "loss": 1.2985,
+      "step": 9640
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1422222222222223e-05,
+      "loss": 1.3709,
+      "step": 9650
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1413333333333335e-05,
+      "loss": 1.2996,
+      "step": 9660
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1404444444444444e-05,
+      "loss": 1.4201,
+      "step": 9670
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1395555555555558e-05,
+      "loss": 1.2211,
+      "step": 9680
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1386666666666669e-05,
+      "loss": 1.264,
+      "step": 9690
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1377777777777779e-05,
+      "loss": 1.3313,
+      "step": 9700
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.136888888888889e-05,
+      "loss": 1.3571,
+      "step": 9710
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.136e-05,
+      "loss": 1.3247,
+      "step": 9720
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1351111111111113e-05,
+      "loss": 1.3511,
+      "step": 9730
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1342222222222224e-05,
+      "loss": 1.3544,
+      "step": 9740
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1333333333333334e-05,
+      "loss": 1.3398,
+      "step": 9750
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1324444444444445e-05,
+      "loss": 1.3307,
+      "step": 9760
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1315555555555557e-05,
+      "loss": 1.2701,
+      "step": 9770
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1306666666666668e-05,
+      "loss": 1.314,
+      "step": 9780
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.129777777777778e-05,
+      "loss": 1.241,
+      "step": 9790
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1288888888888889e-05,
+      "loss": 1.2809,
+      "step": 9800
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.128e-05,
+      "loss": 1.3105,
+      "step": 9810
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1271111111111113e-05,
+      "loss": 1.3106,
+      "step": 9820
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1262222222222223e-05,
+      "loss": 1.2048,
+      "step": 9830
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1253333333333335e-05,
+      "loss": 1.2797,
+      "step": 9840
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1244444444444444e-05,
+      "loss": 1.2862,
+      "step": 9850
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1235555555555557e-05,
+      "loss": 1.3079,
+      "step": 9860
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1226666666666669e-05,
+      "loss": 1.1805,
+      "step": 9870
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1217777777777778e-05,
+      "loss": 1.3203,
+      "step": 9880
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.120888888888889e-05,
+      "loss": 1.2052,
+      "step": 9890
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 1.3102,
+      "step": 9900
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1191111111111113e-05,
+      "loss": 1.3467,
+      "step": 9910
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1182222222222224e-05,
+      "loss": 1.2594,
+      "step": 9920
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1173333333333334e-05,
+      "loss": 1.3147,
+      "step": 9930
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1164444444444445e-05,
+      "loss": 1.2157,
+      "step": 9940
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1155555555555556e-05,
+      "loss": 1.3225,
+      "step": 9950
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1146666666666668e-05,
+      "loss": 1.1717,
+      "step": 9960
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.113777777777778e-05,
+      "loss": 1.2983,
+      "step": 9970
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1128888888888889e-05,
+      "loss": 1.4037,
+      "step": 9980
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1120000000000002e-05,
+      "loss": 1.268,
+      "step": 9990
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1111111111111113e-05,
+      "loss": 1.2195,
+      "step": 10000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1102222222222223e-05,
+      "loss": 1.2844,
+      "step": 10010
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1093333333333334e-05,
+      "loss": 1.3308,
+      "step": 10020
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1084444444444444e-05,
+      "loss": 1.3302,
+      "step": 10030
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1075555555555557e-05,
+      "loss": 1.2522,
+      "step": 10040
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1066666666666669e-05,
+      "loss": 1.3182,
+      "step": 10050
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1057777777777778e-05,
+      "loss": 1.3563,
+      "step": 10060
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.104888888888889e-05,
+      "loss": 1.3619,
+      "step": 10070
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1040000000000001e-05,
+      "loss": 1.2642,
+      "step": 10080
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1031111111111112e-05,
+      "loss": 1.2412,
+      "step": 10090
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1022222222222224e-05,
+      "loss": 1.1924,
+      "step": 10100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1013333333333333e-05,
+      "loss": 1.3139,
+      "step": 10110
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1004444444444445e-05,
+      "loss": 1.3279,
+      "step": 10120
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0995555555555556e-05,
+      "loss": 1.3202,
+      "step": 10130
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0986666666666668e-05,
+      "loss": 1.2398,
+      "step": 10140
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0977777777777779e-05,
+      "loss": 1.2982,
+      "step": 10150
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0968888888888889e-05,
+      "loss": 1.2707,
+      "step": 10160
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.0960000000000002e-05,
+      "loss": 1.2492,
+      "step": 10170
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.0951111111111113e-05,
+      "loss": 1.2729,
+      "step": 10180
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.0942222222222223e-05,
+      "loss": 1.325,
+      "step": 10190
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.0933333333333334e-05,
+      "loss": 1.3168,
+      "step": 10200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.0924444444444446e-05,
+      "loss": 1.2955,
+      "step": 10210
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.0915555555555557e-05,
+      "loss": 1.3221,
+      "step": 10220
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.0906666666666668e-05,
+      "loss": 1.3121,
+      "step": 10230
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.0897777777777778e-05,
+      "loss": 1.2272,
+      "step": 10240
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.088888888888889e-05,
+      "loss": 1.2919,
+      "step": 10250
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.0880000000000001e-05,
+      "loss": 1.2944,
+      "step": 10260
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.0871111111111112e-05,
+      "loss": 1.2907,
+      "step": 10270
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.0862222222222224e-05,
+      "loss": 1.3381,
+      "step": 10280
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.0853333333333333e-05,
+      "loss": 1.3573,
+      "step": 10290
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.0844444444444446e-05,
+      "loss": 1.3328,
+      "step": 10300
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.0835555555555556e-05,
+      "loss": 1.3188,
+      "step": 10310
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0826666666666667e-05,
+      "loss": 1.2644,
+      "step": 10320
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0817777777777779e-05,
+      "loss": 1.3162,
+      "step": 10330
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0808888888888889e-05,
+      "loss": 1.4474,
+      "step": 10340
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0800000000000002e-05,
+      "loss": 1.3643,
+      "step": 10350
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0791111111111113e-05,
+      "loss": 1.3039,
+      "step": 10360
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0782222222222223e-05,
+      "loss": 1.2926,
+      "step": 10370
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0773333333333334e-05,
+      "loss": 1.4064,
+      "step": 10380
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0764444444444445e-05,
+      "loss": 1.1842,
+      "step": 10390
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0755555555555557e-05,
+      "loss": 1.3694,
+      "step": 10400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0746666666666668e-05,
+      "loss": 1.4161,
+      "step": 10410
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0737777777777778e-05,
+      "loss": 1.3584,
+      "step": 10420
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.072888888888889e-05,
+      "loss": 1.2626,
+      "step": 10430
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.072e-05,
+      "loss": 1.2891,
+      "step": 10440
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0711111111111112e-05,
+      "loss": 1.2715,
+      "step": 10450
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0702222222222223e-05,
+      "loss": 1.3825,
+      "step": 10460
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0693333333333333e-05,
+      "loss": 1.3211,
+      "step": 10470
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0684444444444446e-05,
+      "loss": 1.3667,
+      "step": 10480
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0675555555555558e-05,
+      "loss": 1.3165,
+      "step": 10490
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 1.2955,
+      "step": 10500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0657777777777779e-05,
+      "loss": 1.3286,
+      "step": 10510
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.064888888888889e-05,
+      "loss": 1.299,
+      "step": 10520
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0640000000000001e-05,
+      "loss": 1.3321,
+      "step": 10530
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0631111111111113e-05,
+      "loss": 1.2766,
+      "step": 10540
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0622222222222223e-05,
+      "loss": 1.3089,
+      "step": 10550
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0613333333333334e-05,
+      "loss": 1.3043,
+      "step": 10560
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0604444444444445e-05,
+      "loss": 1.2449,
+      "step": 10570
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0595555555555557e-05,
+      "loss": 1.2622,
+      "step": 10580
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0586666666666668e-05,
+      "loss": 1.3104,
+      "step": 10590
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0577777777777778e-05,
+      "loss": 1.2886,
+      "step": 10600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0568888888888891e-05,
+      "loss": 1.2681,
+      "step": 10610
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.056e-05,
+      "loss": 1.2302,
+      "step": 10620
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0551111111111112e-05,
+      "loss": 1.2633,
+      "step": 10630
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0542222222222223e-05,
+      "loss": 1.3342,
+      "step": 10640
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0533333333333333e-05,
+      "loss": 1.3311,
+      "step": 10650
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0524444444444446e-05,
+      "loss": 1.3098,
+      "step": 10660
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0515555555555557e-05,
+      "loss": 1.3537,
+      "step": 10670
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0506666666666667e-05,
+      "loss": 1.2273,
+      "step": 10680
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0497777777777779e-05,
+      "loss": 1.4083,
+      "step": 10690
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.048888888888889e-05,
+      "loss": 1.2744,
+      "step": 10700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0480000000000001e-05,
+      "loss": 1.2371,
+      "step": 10710
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0471111111111113e-05,
+      "loss": 1.2844,
+      "step": 10720
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0462222222222222e-05,
+      "loss": 1.3337,
+      "step": 10730
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0453333333333334e-05,
+      "loss": 1.2411,
+      "step": 10740
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0444444444444445e-05,
+      "loss": 1.1806,
+      "step": 10750
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0435555555555557e-05,
+      "loss": 1.3143,
+      "step": 10760
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0426666666666668e-05,
+      "loss": 1.3145,
+      "step": 10770
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0417777777777778e-05,
+      "loss": 1.3019,
+      "step": 10780
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.040888888888889e-05,
+      "loss": 1.2615,
+      "step": 10790
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.04e-05,
+      "loss": 1.3636,
+      "step": 10800
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0391111111111112e-05,
+      "loss": 1.2738,
+      "step": 10810
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0382222222222223e-05,
+      "loss": 1.3178,
+      "step": 10820
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0373333333333335e-05,
+      "loss": 1.2788,
+      "step": 10830
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0364444444444446e-05,
+      "loss": 1.2375,
+      "step": 10840
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0355555555555557e-05,
+      "loss": 1.2657,
+      "step": 10850
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0346666666666667e-05,
+      "loss": 1.3125,
+      "step": 10860
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0337777777777778e-05,
+      "loss": 1.3071,
+      "step": 10870
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.032888888888889e-05,
+      "loss": 1.3174,
+      "step": 10880
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0320000000000001e-05,
+      "loss": 1.2834,
+      "step": 10890
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0311111111111113e-05,
+      "loss": 1.3054,
+      "step": 10900
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0302222222222222e-05,
+      "loss": 1.179,
+      "step": 10910
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0293333333333335e-05,
+      "loss": 1.2345,
+      "step": 10920
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0284444444444445e-05,
+      "loss": 1.3735,
+      "step": 10930
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0275555555555556e-05,
+      "loss": 1.238,
+      "step": 10940
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0266666666666668e-05,
+      "loss": 1.3102,
+      "step": 10950
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0257777777777777e-05,
+      "loss": 1.172,
+      "step": 10960
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.024888888888889e-05,
+      "loss": 1.272,
+      "step": 10970
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.024e-05,
+      "loss": 1.2451,
+      "step": 10980
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0231111111111112e-05,
+      "loss": 1.2522,
+      "step": 10990
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0222222222222223e-05,
+      "loss": 1.2014,
+      "step": 11000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0213333333333334e-05,
+      "loss": 1.2308,
+      "step": 11010
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0204444444444446e-05,
+      "loss": 1.3126,
+      "step": 11020
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0195555555555557e-05,
+      "loss": 1.2188,
+      "step": 11030
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0186666666666667e-05,
+      "loss": 1.318,
+      "step": 11040
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0177777777777778e-05,
+      "loss": 1.3385,
+      "step": 11050
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.016888888888889e-05,
+      "loss": 1.3492,
+      "step": 11060
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0160000000000001e-05,
+      "loss": 1.2495,
+      "step": 11070
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0151111111111112e-05,
+      "loss": 1.268,
+      "step": 11080
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0142222222222222e-05,
+      "loss": 1.2586,
+      "step": 11090
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0133333333333335e-05,
+      "loss": 1.2718,
+      "step": 11100
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0124444444444445e-05,
+      "loss": 1.2452,
+      "step": 11110
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0115555555555556e-05,
+      "loss": 1.2685,
+      "step": 11120
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0106666666666668e-05,
+      "loss": 1.2924,
+      "step": 11130
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0097777777777779e-05,
+      "loss": 1.2851,
+      "step": 11140
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.008888888888889e-05,
+      "loss": 1.2606,
+      "step": 11150
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.008e-05,
+      "loss": 1.2501,
+      "step": 11160
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0071111111111111e-05,
+      "loss": 1.245,
+      "step": 11170
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0062222222222223e-05,
+      "loss": 1.2472,
+      "step": 11180
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0053333333333334e-05,
+      "loss": 1.2095,
+      "step": 11190
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0044444444444446e-05,
+      "loss": 1.3048,
+      "step": 11200
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0035555555555557e-05,
+      "loss": 1.2809,
+      "step": 11210
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0026666666666667e-05,
+      "loss": 1.3349,
+      "step": 11220
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.001777777777778e-05,
+      "loss": 1.2677,
+      "step": 11230
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.000888888888889e-05,
+      "loss": 1.3069,
+      "step": 11240
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1e-05,
+      "loss": 1.2846,
+      "step": 11250
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.991111111111112e-06,
+      "loss": 1.2455,
+      "step": 11260
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.982222222222224e-06,
+      "loss": 1.2334,
+      "step": 11270
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.973333333333333e-06,
+      "loss": 1.3109,
+      "step": 11280
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.964444444444445e-06,
+      "loss": 1.3119,
+      "step": 11290
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.955555555555556e-06,
+      "loss": 1.2835,
+      "step": 11300
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.946666666666667e-06,
+      "loss": 1.2573,
+      "step": 11310
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.937777777777779e-06,
+      "loss": 1.2544,
+      "step": 11320
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.92888888888889e-06,
+      "loss": 1.3619,
+      "step": 11330
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 1.2576,
+      "step": 11340
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.911111111111113e-06,
+      "loss": 1.285,
+      "step": 11350
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.902222222222223e-06,
+      "loss": 1.2528,
+      "step": 11360
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.893333333333334e-06,
+      "loss": 1.272,
+      "step": 11370
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.884444444444445e-06,
+      "loss": 1.297,
+      "step": 11380
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.875555555555557e-06,
+      "loss": 1.3149,
+      "step": 11390
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.866666666666668e-06,
+      "loss": 1.3019,
+      "step": 11400
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.857777777777778e-06,
+      "loss": 1.2598,
+      "step": 11410
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.84888888888889e-06,
+      "loss": 1.2781,
+      "step": 11420
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.84e-06,
+      "loss": 1.3739,
+      "step": 11430
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.831111111111112e-06,
+      "loss": 1.139,
+      "step": 11440
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.822222222222223e-06,
+      "loss": 1.2455,
+      "step": 11450
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.813333333333333e-06,
+      "loss": 1.2812,
+      "step": 11460
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.804444444444444e-06,
+      "loss": 1.2689,
+      "step": 11470
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.795555555555556e-06,
+      "loss": 1.3294,
+      "step": 11480
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.786666666666667e-06,
+      "loss": 1.311,
+      "step": 11490
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.777777777777779e-06,
+      "loss": 1.2169,
+      "step": 11500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.76888888888889e-06,
+      "loss": 1.3094,
+      "step": 11510
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 1.2648,
+      "step": 11520
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.751111111111113e-06,
+      "loss": 1.3215,
+      "step": 11530
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.742222222222222e-06,
+      "loss": 1.1701,
+      "step": 11540
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.733333333333334e-06,
+      "loss": 1.2496,
+      "step": 11550
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.724444444444445e-06,
+      "loss": 1.2892,
+      "step": 11560
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.715555555555557e-06,
+      "loss": 1.2079,
+      "step": 11570
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.706666666666668e-06,
+      "loss": 1.2575,
+      "step": 11580
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.697777777777778e-06,
+      "loss": 1.2917,
+      "step": 11590
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.688888888888889e-06,
+      "loss": 1.2598,
+      "step": 11600
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.68e-06,
+      "loss": 1.2973,
+      "step": 11610
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.671111111111112e-06,
+      "loss": 1.3112,
+      "step": 11620
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.662222222222223e-06,
+      "loss": 1.2299,
+      "step": 11630
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.653333333333335e-06,
+      "loss": 1.2611,
+      "step": 11640
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.644444444444444e-06,
+      "loss": 1.3104,
+      "step": 11650
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.635555555555557e-06,
+      "loss": 1.2473,
+      "step": 11660
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.626666666666667e-06,
+      "loss": 1.3142,
+      "step": 11670
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.617777777777778e-06,
+      "loss": 1.2361,
+      "step": 11680
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.60888888888889e-06,
+      "loss": 1.2597,
+      "step": 11690
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 1.321,
+      "step": 11700
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.591111111111113e-06,
+      "loss": 1.2949,
+      "step": 11710
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.582222222222222e-06,
+      "loss": 1.2847,
+      "step": 11720
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.573333333333334e-06,
+      "loss": 1.1879,
+      "step": 11730
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.564444444444445e-06,
+      "loss": 1.3084,
+      "step": 11740
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.555555555555556e-06,
+      "loss": 1.2484,
+      "step": 11750
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.546666666666668e-06,
+      "loss": 1.3331,
+      "step": 11760
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.537777777777778e-06,
+      "loss": 1.3249,
+      "step": 11770
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.528888888888889e-06,
+      "loss": 1.31,
+      "step": 11780
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.52e-06,
+      "loss": 1.3252,
+      "step": 11790
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.511111111111112e-06,
+      "loss": 1.3655,
+      "step": 11800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.502222222222223e-06,
+      "loss": 1.3759,
+      "step": 11810
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.493333333333334e-06,
+      "loss": 1.2118,
+      "step": 11820
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.484444444444444e-06,
+      "loss": 1.3105,
+      "step": 11830
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.475555555555557e-06,
+      "loss": 1.3224,
+      "step": 11840
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.466666666666667e-06,
+      "loss": 1.3524,
+      "step": 11850
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.457777777777778e-06,
+      "loss": 1.3058,
+      "step": 11860
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.44888888888889e-06,
+      "loss": 1.2499,
+      "step": 11870
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 1.2448,
+      "step": 11880
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.431111111111112e-06,
+      "loss": 1.3527,
+      "step": 11890
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.422222222222222e-06,
+      "loss": 1.2263,
+      "step": 11900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.413333333333334e-06,
+      "loss": 1.2851,
+      "step": 11910
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.404444444444445e-06,
+      "loss": 1.3441,
+      "step": 11920
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.395555555555556e-06,
+      "loss": 1.2497,
+      "step": 11930
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.386666666666668e-06,
+      "loss": 1.211,
+      "step": 11940
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.377777777777779e-06,
+      "loss": 1.2567,
+      "step": 11950
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.368888888888889e-06,
+      "loss": 1.2009,
+      "step": 11960
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.360000000000002e-06,
+      "loss": 1.3445,
+      "step": 11970
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.351111111111112e-06,
+      "loss": 1.2952,
+      "step": 11980
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.342222222222223e-06,
+      "loss": 1.3034,
+      "step": 11990
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 1.2728,
+      "step": 12000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.324444444444444e-06,
+      "loss": 1.2215,
+      "step": 12010
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.315555555555557e-06,
+      "loss": 1.2394,
+      "step": 12020
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.306666666666667e-06,
+      "loss": 1.2712,
+      "step": 12030
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.297777777777778e-06,
+      "loss": 1.2592,
+      "step": 12040
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.28888888888889e-06,
+      "loss": 1.3132,
+      "step": 12050
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 1.2691,
+      "step": 12060
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.271111111111112e-06,
+      "loss": 1.3253,
+      "step": 12070
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.262222222222222e-06,
+      "loss": 1.2733,
+      "step": 12080
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.253333333333333e-06,
+      "loss": 1.1862,
+      "step": 12090
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.244444444444445e-06,
+      "loss": 1.2831,
+      "step": 12100
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.235555555555556e-06,
+      "loss": 1.159,
+      "step": 12110
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.226666666666668e-06,
+      "loss": 1.2446,
+      "step": 12120
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.217777777777779e-06,
+      "loss": 1.2421,
+      "step": 12130
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.208888888888889e-06,
+      "loss": 1.2553,
+      "step": 12140
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 1.2619,
+      "step": 12150
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.191111111111111e-06,
+      "loss": 1.3435,
+      "step": 12160
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.182222222222223e-06,
+      "loss": 1.304,
+      "step": 12170
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.173333333333334e-06,
+      "loss": 1.2557,
+      "step": 12180
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.164444444444446e-06,
+      "loss": 1.3141,
+      "step": 12190
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.155555555555557e-06,
+      "loss": 1.2411,
+      "step": 12200
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.146666666666667e-06,
+      "loss": 1.2849,
+      "step": 12210
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.137777777777778e-06,
+      "loss": 1.3234,
+      "step": 12220
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.12888888888889e-06,
+      "loss": 1.3266,
+      "step": 12230
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.12e-06,
+      "loss": 1.1997,
+      "step": 12240
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.111111111111112e-06,
+      "loss": 1.3032,
+      "step": 12250
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.102222222222224e-06,
+      "loss": 1.2216,
+      "step": 12260
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.093333333333333e-06,
+      "loss": 1.2377,
+      "step": 12270
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.084444444444446e-06,
+      "loss": 1.3451,
+      "step": 12280
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.075555555555556e-06,
+      "loss": 1.2116,
+      "step": 12290
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.066666666666667e-06,
+      "loss": 1.2842,
+      "step": 12300
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.057777777777779e-06,
+      "loss": 1.2196,
+      "step": 12310
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.048888888888888e-06,
+      "loss": 1.3075,
+      "step": 12320
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.040000000000002e-06,
+      "loss": 1.2657,
+      "step": 12330
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.031111111111111e-06,
+      "loss": 1.2977,
+      "step": 12340
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.022222222222223e-06,
+      "loss": 1.3294,
+      "step": 12350
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.013333333333334e-06,
+      "loss": 1.3123,
+      "step": 12360
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.004444444444445e-06,
+      "loss": 1.2692,
+      "step": 12370
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.995555555555557e-06,
+      "loss": 1.169,
+      "step": 12380
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.986666666666666e-06,
+      "loss": 1.2132,
+      "step": 12390
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.977777777777778e-06,
+      "loss": 1.2555,
+      "step": 12400
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.96888888888889e-06,
+      "loss": 1.3716,
+      "step": 12410
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.96e-06,
+      "loss": 1.1621,
+      "step": 12420
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.951111111111112e-06,
+      "loss": 1.2929,
+      "step": 12430
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.942222222222223e-06,
+      "loss": 1.1761,
+      "step": 12440
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.933333333333333e-06,
+      "loss": 1.2787,
+      "step": 12450
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.924444444444446e-06,
+      "loss": 1.2931,
+      "step": 12460
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.915555555555556e-06,
+      "loss": 1.4209,
+      "step": 12470
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.906666666666667e-06,
+      "loss": 1.3084,
+      "step": 12480
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.897777777777779e-06,
+      "loss": 1.3079,
+      "step": 12490
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 1.2932,
+      "step": 12500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.880000000000001e-06,
+      "loss": 1.3199,
+      "step": 12510
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.871111111111111e-06,
+      "loss": 1.3068,
+      "step": 12520
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.862222222222222e-06,
+      "loss": 1.3364,
+      "step": 12530
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.853333333333334e-06,
+      "loss": 1.2202,
+      "step": 12540
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.844444444444445e-06,
+      "loss": 1.3451,
+      "step": 12550
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.835555555555557e-06,
+      "loss": 1.3206,
+      "step": 12560
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.826666666666668e-06,
+      "loss": 1.2827,
+      "step": 12570
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.817777777777778e-06,
+      "loss": 1.2676,
+      "step": 12580
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.80888888888889e-06,
+      "loss": 1.3322,
+      "step": 12590
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.8e-06,
+      "loss": 1.2761,
+      "step": 12600
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.791111111111112e-06,
+      "loss": 1.1287,
+      "step": 12610
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.782222222222223e-06,
+      "loss": 1.2975,
+      "step": 12620
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.773333333333333e-06,
+      "loss": 1.1639,
+      "step": 12630
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.764444444444446e-06,
+      "loss": 1.3206,
+      "step": 12640
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.755555555555556e-06,
+      "loss": 1.2569,
+      "step": 12650
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.746666666666667e-06,
+      "loss": 1.2338,
+      "step": 12660
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.737777777777778e-06,
+      "loss": 1.2053,
+      "step": 12670
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.72888888888889e-06,
+      "loss": 1.2774,
+      "step": 12680
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.720000000000001e-06,
+      "loss": 1.1958,
+      "step": 12690
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.711111111111111e-06,
+      "loss": 1.3249,
+      "step": 12700
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.702222222222222e-06,
+      "loss": 1.3396,
+      "step": 12710
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.693333333333334e-06,
+      "loss": 1.2735,
+      "step": 12720
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.684444444444445e-06,
+      "loss": 1.2483,
+      "step": 12730
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.675555555555556e-06,
+      "loss": 1.3765,
+      "step": 12740
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.666666666666668e-06,
+      "loss": 1.2317,
+      "step": 12750
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.657777777777778e-06,
+      "loss": 1.3384,
+      "step": 12760
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.64888888888889e-06,
+      "loss": 1.2925,
+      "step": 12770
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.64e-06,
+      "loss": 1.2024,
+      "step": 12780
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.631111111111112e-06,
+      "loss": 1.2548,
+      "step": 12790
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.622222222222223e-06,
+      "loss": 1.3613,
+      "step": 12800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.613333333333333e-06,
+      "loss": 1.2567,
+      "step": 12810
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.604444444444446e-06,
+      "loss": 1.2534,
+      "step": 12820
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.595555555555556e-06,
+      "loss": 1.2183,
+      "step": 12830
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.586666666666667e-06,
+      "loss": 1.3391,
+      "step": 12840
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.577777777777778e-06,
+      "loss": 1.2432,
+      "step": 12850
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.56888888888889e-06,
+      "loss": 1.3076,
+      "step": 12860
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.560000000000001e-06,
+      "loss": 1.3239,
+      "step": 12870
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.551111111111112e-06,
+      "loss": 1.2042,
+      "step": 12880
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.542222222222222e-06,
+      "loss": 1.2647,
+      "step": 12890
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.533333333333335e-06,
+      "loss": 1.2405,
+      "step": 12900
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.524444444444445e-06,
+      "loss": 1.3369,
+      "step": 12910
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.515555555555556e-06,
+      "loss": 1.2035,
+      "step": 12920
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.506666666666668e-06,
+      "loss": 1.3162,
+      "step": 12930
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.497777777777777e-06,
+      "loss": 1.2844,
+      "step": 12940
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.48888888888889e-06,
+      "loss": 1.2494,
+      "step": 12950
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.48e-06,
+      "loss": 1.2999,
+      "step": 12960
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.471111111111112e-06,
+      "loss": 1.2545,
+      "step": 12970
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.462222222222223e-06,
+      "loss": 1.2085,
+      "step": 12980
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.453333333333334e-06,
+      "loss": 1.2116,
+      "step": 12990
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.444444444444446e-06,
+      "loss": 1.2923,
+      "step": 13000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.435555555555555e-06,
+      "loss": 1.2093,
+      "step": 13010
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.426666666666667e-06,
+      "loss": 1.3004,
+      "step": 13020
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.417777777777778e-06,
+      "loss": 1.2888,
+      "step": 13030
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.40888888888889e-06,
+      "loss": 1.323,
+      "step": 13040
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 1.2195,
+      "step": 13050
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.391111111111112e-06,
+      "loss": 1.2865,
+      "step": 13060
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.382222222222222e-06,
+      "loss": 1.3062,
+      "step": 13070
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.373333333333335e-06,
+      "loss": 1.1727,
+      "step": 13080
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.364444444444445e-06,
+      "loss": 1.2901,
+      "step": 13090
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.355555555555556e-06,
+      "loss": 1.2663,
+      "step": 13100
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.346666666666668e-06,
+      "loss": 1.2479,
+      "step": 13110
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.337777777777777e-06,
+      "loss": 1.2259,
+      "step": 13120
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.32888888888889e-06,
+      "loss": 1.3694,
+      "step": 13130
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.32e-06,
+      "loss": 1.2462,
+      "step": 13140
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.311111111111111e-06,
+      "loss": 1.2609,
+      "step": 13150
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.302222222222223e-06,
+      "loss": 1.2285,
+      "step": 13160
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.293333333333334e-06,
+      "loss": 1.3475,
+      "step": 13170
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.284444444444446e-06,
+      "loss": 1.2668,
+      "step": 13180
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.275555555555557e-06,
+      "loss": 1.2987,
+      "step": 13190
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.266666666666667e-06,
+      "loss": 1.2537,
+      "step": 13200
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.25777777777778e-06,
+      "loss": 1.2572,
+      "step": 13210
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.24888888888889e-06,
+      "loss": 1.2471,
+      "step": 13220
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.24e-06,
+      "loss": 1.2263,
+      "step": 13230
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.231111111111112e-06,
+      "loss": 1.3185,
+      "step": 13240
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.222222222222222e-06,
+      "loss": 1.2645,
+      "step": 13250
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.213333333333335e-06,
+      "loss": 1.3063,
+      "step": 13260
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.204444444444445e-06,
+      "loss": 1.191,
+      "step": 13270
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.195555555555556e-06,
+      "loss": 1.2418,
+      "step": 13280
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.186666666666667e-06,
+      "loss": 1.3557,
+      "step": 13290
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.177777777777779e-06,
+      "loss": 1.3138,
+      "step": 13300
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.16888888888889e-06,
+      "loss": 1.2382,
+      "step": 13310
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.16e-06,
+      "loss": 1.268,
+      "step": 13320
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.151111111111111e-06,
+      "loss": 1.1662,
+      "step": 13330
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.142222222222223e-06,
+      "loss": 1.2772,
+      "step": 13340
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.133333333333334e-06,
+      "loss": 1.2846,
+      "step": 13350
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.124444444444445e-06,
+      "loss": 1.1947,
+      "step": 13360
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.115555555555557e-06,
+      "loss": 1.3121,
+      "step": 13370
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.106666666666666e-06,
+      "loss": 1.2876,
+      "step": 13380
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.09777777777778e-06,
+      "loss": 1.2993,
+      "step": 13390
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.08888888888889e-06,
+      "loss": 1.3019,
+      "step": 13400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.08e-06,
+      "loss": 1.3544,
+      "step": 13410
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.071111111111112e-06,
+      "loss": 1.2742,
+      "step": 13420
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.062222222222222e-06,
+      "loss": 1.2497,
+      "step": 13430
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.053333333333335e-06,
+      "loss": 1.2895,
+      "step": 13440
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.044444444444444e-06,
+      "loss": 1.3032,
+      "step": 13450
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.035555555555556e-06,
+      "loss": 1.2516,
+      "step": 13460
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.026666666666667e-06,
+      "loss": 1.3395,
+      "step": 13470
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.017777777777779e-06,
+      "loss": 1.2534,
+      "step": 13480
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.00888888888889e-06,
+      "loss": 1.2794,
+      "step": 13490
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.2171,
+      "step": 13500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.991111111111111e-06,
+      "loss": 1.1564,
+      "step": 13510
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.982222222222224e-06,
+      "loss": 1.18,
+      "step": 13520
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.973333333333334e-06,
+      "loss": 1.2665,
+      "step": 13530
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.964444444444445e-06,
+      "loss": 1.3516,
+      "step": 13540
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.955555555555557e-06,
+      "loss": 1.4101,
+      "step": 13550
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.946666666666666e-06,
+      "loss": 1.2157,
+      "step": 13560
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.93777777777778e-06,
+      "loss": 1.2586,
+      "step": 13570
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.928888888888889e-06,
+      "loss": 1.2547,
+      "step": 13580
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.92e-06,
+      "loss": 1.3429,
+      "step": 13590
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.911111111111112e-06,
+      "loss": 1.1899,
+      "step": 13600
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.902222222222223e-06,
+      "loss": 1.2836,
+      "step": 13610
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.893333333333335e-06,
+      "loss": 1.2881,
+      "step": 13620
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.884444444444444e-06,
+      "loss": 1.3739,
+      "step": 13630
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.875555555555556e-06,
+      "loss": 1.2875,
+      "step": 13640
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.866666666666667e-06,
+      "loss": 1.2278,
+      "step": 13650
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.857777777777778e-06,
+      "loss": 1.239,
+      "step": 13660
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.84888888888889e-06,
+      "loss": 1.2063,
+      "step": 13670
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.840000000000001e-06,
+      "loss": 1.2448,
+      "step": 13680
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.831111111111111e-06,
+      "loss": 1.2705,
+      "step": 13690
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.822222222222224e-06,
+      "loss": 1.3094,
+      "step": 13700
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.813333333333334e-06,
+      "loss": 1.1893,
+      "step": 13710
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.804444444444445e-06,
+      "loss": 1.3243,
+      "step": 13720
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.795555555555556e-06,
+      "loss": 1.2749,
+      "step": 13730
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.786666666666666e-06,
+      "loss": 1.2446,
+      "step": 13740
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.77777777777778e-06,
+      "loss": 1.3492,
+      "step": 13750
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.768888888888889e-06,
+      "loss": 1.2565,
+      "step": 13760
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.76e-06,
+      "loss": 1.2021,
+      "step": 13770
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.751111111111112e-06,
+      "loss": 1.2324,
+      "step": 13780
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.742222222222223e-06,
+      "loss": 1.2174,
+      "step": 13790
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.733333333333334e-06,
+      "loss": 1.1951,
+      "step": 13800
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.724444444444446e-06,
+      "loss": 1.1886,
+      "step": 13810
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.715555555555555e-06,
+      "loss": 1.2288,
+      "step": 13820
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.706666666666669e-06,
+      "loss": 1.315,
+      "step": 13830
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.697777777777778e-06,
+      "loss": 1.2577,
+      "step": 13840
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.68888888888889e-06,
+      "loss": 1.2138,
+      "step": 13850
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 1.2946,
+      "step": 13860
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.67111111111111e-06,
+      "loss": 1.238,
+      "step": 13870
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.662222222222224e-06,
+      "loss": 1.3104,
+      "step": 13880
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.653333333333333e-06,
+      "loss": 1.338,
+      "step": 13890
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.644444444444445e-06,
+      "loss": 1.2746,
+      "step": 13900
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.635555555555556e-06,
+      "loss": 1.2892,
+      "step": 13910
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.626666666666668e-06,
+      "loss": 1.1335,
+      "step": 13920
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.617777777777778e-06,
+      "loss": 1.2212,
+      "step": 13930
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.608888888888889e-06,
+      "loss": 1.312,
+      "step": 13940
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 1.1373,
+      "step": 13950
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.5911111111111115e-06,
+      "loss": 1.2674,
+      "step": 13960
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.582222222222223e-06,
+      "loss": 1.1686,
+      "step": 13970
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.573333333333333e-06,
+      "loss": 1.3738,
+      "step": 13980
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.564444444444446e-06,
+      "loss": 1.2727,
+      "step": 13990
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.555555555555556e-06,
+      "loss": 1.2919,
+      "step": 14000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.5466666666666675e-06,
+      "loss": 1.2957,
+      "step": 14010
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.537777777777778e-06,
+      "loss": 1.1067,
+      "step": 14020
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.52888888888889e-06,
+      "loss": 1.3171,
+      "step": 14030
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 1.2833,
+      "step": 14040
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.511111111111111e-06,
+      "loss": 1.3938,
+      "step": 14050
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.502222222222223e-06,
+      "loss": 1.3277,
+      "step": 14060
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.493333333333333e-06,
+      "loss": 1.3569,
+      "step": 14070
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.4844444444444455e-06,
+      "loss": 1.2262,
+      "step": 14080
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.475555555555556e-06,
+      "loss": 1.2948,
+      "step": 14090
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.4666666666666675e-06,
+      "loss": 1.3357,
+      "step": 14100
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.457777777777778e-06,
+      "loss": 1.2638,
+      "step": 14110
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.44888888888889e-06,
+      "loss": 1.2455,
+      "step": 14120
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.440000000000001e-06,
+      "loss": 1.2697,
+      "step": 14130
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.431111111111111e-06,
+      "loss": 1.3099,
+      "step": 14140
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.422222222222223e-06,
+      "loss": 1.3437,
+      "step": 14150
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.413333333333333e-06,
+      "loss": 1.3246,
+      "step": 14160
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.4044444444444455e-06,
+      "loss": 1.2499,
+      "step": 14170
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.395555555555556e-06,
+      "loss": 1.2036,
+      "step": 14180
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.386666666666667e-06,
+      "loss": 1.227,
+      "step": 14190
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.377777777777778e-06,
+      "loss": 1.3326,
+      "step": 14200
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.36888888888889e-06,
+      "loss": 1.3089,
+      "step": 14210
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 1.1983,
+      "step": 14220
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.351111111111112e-06,
+      "loss": 1.2575,
+      "step": 14230
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.342222222222223e-06,
+      "loss": 1.281,
+      "step": 14240
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.333333333333333e-06,
+      "loss": 1.3717,
+      "step": 14250
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.324444444444445e-06,
+      "loss": 1.2904,
+      "step": 14260
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.315555555555556e-06,
+      "loss": 1.2083,
+      "step": 14270
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.306666666666667e-06,
+      "loss": 1.1944,
+      "step": 14280
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.297777777777778e-06,
+      "loss": 1.345,
+      "step": 14290
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.28888888888889e-06,
+      "loss": 1.3647,
+      "step": 14300
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.280000000000001e-06,
+      "loss": 1.2096,
+      "step": 14310
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.271111111111112e-06,
+      "loss": 1.2166,
+      "step": 14320
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.2622222222222225e-06,
+      "loss": 1.3078,
+      "step": 14330
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.253333333333335e-06,
+      "loss": 1.3183,
+      "step": 14340
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.244444444444445e-06,
+      "loss": 1.2197,
+      "step": 14350
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.235555555555556e-06,
+      "loss": 1.3547,
+      "step": 14360
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.226666666666667e-06,
+      "loss": 1.1229,
+      "step": 14370
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.217777777777778e-06,
+      "loss": 1.1625,
+      "step": 14380
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.20888888888889e-06,
+      "loss": 1.2613,
+      "step": 14390
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 1.2235,
+      "step": 14400
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.191111111111112e-06,
+      "loss": 1.3275,
+      "step": 14410
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.1822222222222224e-06,
+      "loss": 1.3054,
+      "step": 14420
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.173333333333335e-06,
+      "loss": 1.2218,
+      "step": 14430
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.164444444444445e-06,
+      "loss": 1.3295,
+      "step": 14440
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.155555555555556e-06,
+      "loss": 1.2847,
+      "step": 14450
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.146666666666667e-06,
+      "loss": 1.2679,
+      "step": 14460
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.137777777777778e-06,
+      "loss": 1.2052,
+      "step": 14470
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.12888888888889e-06,
+      "loss": 1.276,
+      "step": 14480
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.1200000000000004e-06,
+      "loss": 1.2431,
+      "step": 14490
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.111111111111112e-06,
+      "loss": 1.2889,
+      "step": 14500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.102222222222222e-06,
+      "loss": 1.3095,
+      "step": 14510
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.093333333333335e-06,
+      "loss": 1.2753,
+      "step": 14520
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.084444444444445e-06,
+      "loss": 1.3078,
+      "step": 14530
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.0755555555555565e-06,
+      "loss": 1.2743,
+      "step": 14540
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.066666666666667e-06,
+      "loss": 1.2702,
+      "step": 14550
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.057777777777778e-06,
+      "loss": 1.3746,
+      "step": 14560
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.04888888888889e-06,
+      "loss": 1.2119,
+      "step": 14570
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.04e-06,
+      "loss": 1.2695,
+      "step": 14580
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.031111111111112e-06,
+      "loss": 1.3613,
+      "step": 14590
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.022222222222222e-06,
+      "loss": 1.1823,
+      "step": 14600
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.0133333333333345e-06,
+      "loss": 1.3189,
+      "step": 14610
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.004444444444445e-06,
+      "loss": 1.2656,
+      "step": 14620
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.9955555555555564e-06,
+      "loss": 1.3264,
+      "step": 14630
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.986666666666667e-06,
+      "loss": 1.2679,
+      "step": 14640
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.977777777777779e-06,
+      "loss": 1.2579,
+      "step": 14650
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.96888888888889e-06,
+      "loss": 1.1576,
+      "step": 14660
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.96e-06,
+      "loss": 1.2369,
+      "step": 14670
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.951111111111112e-06,
+      "loss": 1.2884,
+      "step": 14680
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.942222222222222e-06,
+      "loss": 1.1801,
+      "step": 14690
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.9333333333333344e-06,
+      "loss": 1.2305,
+      "step": 14700
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.924444444444445e-06,
+      "loss": 1.2369,
+      "step": 14710
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.915555555555556e-06,
+      "loss": 1.32,
+      "step": 14720
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.906666666666667e-06,
+      "loss": 1.3006,
+      "step": 14730
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.897777777777779e-06,
+      "loss": 1.1992,
+      "step": 14740
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.88888888888889e-06,
+      "loss": 1.2324,
+      "step": 14750
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.88e-06,
+      "loss": 1.3592,
+      "step": 14760
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.871111111111112e-06,
+      "loss": 1.2282,
+      "step": 14770
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.862222222222222e-06,
+      "loss": 1.244,
+      "step": 14780
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.853333333333334e-06,
+      "loss": 1.3236,
+      "step": 14790
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.844444444444445e-06,
+      "loss": 1.2188,
+      "step": 14800
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.835555555555556e-06,
+      "loss": 1.2541,
+      "step": 14810
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.826666666666667e-06,
+      "loss": 1.3408,
+      "step": 14820
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.817777777777779e-06,
+      "loss": 1.2765,
+      "step": 14830
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.80888888888889e-06,
+      "loss": 1.3218,
+      "step": 14840
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 1.2574,
+      "step": 14850
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.7911111111111115e-06,
+      "loss": 1.2121,
+      "step": 14860
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.782222222222222e-06,
+      "loss": 1.288,
+      "step": 14870
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.773333333333334e-06,
+      "loss": 1.2875,
+      "step": 14880
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.764444444444445e-06,
+      "loss": 1.2592,
+      "step": 14890
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.755555555555556e-06,
+      "loss": 1.3308,
+      "step": 14900
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.746666666666667e-06,
+      "loss": 1.3268,
+      "step": 14910
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.737777777777779e-06,
+      "loss": 1.3026,
+      "step": 14920
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.7288888888888895e-06,
+      "loss": 1.3352,
+      "step": 14930
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 1.3234,
+      "step": 14940
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.711111111111111e-06,
+      "loss": 1.2499,
+      "step": 14950
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.702222222222224e-06,
+      "loss": 1.192,
+      "step": 14960
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.693333333333334e-06,
+      "loss": 1.2889,
+      "step": 14970
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.684444444444445e-06,
+      "loss": 1.208,
+      "step": 14980
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.675555555555556e-06,
+      "loss": 1.3019,
+      "step": 14990
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.2796,
+      "step": 15000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.657777777777779e-06,
+      "loss": 0.7504,
+      "step": 15010
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.648888888888889e-06,
+      "loss": 0.8104,
+      "step": 15020
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.640000000000001e-06,
+      "loss": 0.8572,
+      "step": 15030
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.631111111111111e-06,
+      "loss": 0.8371,
+      "step": 15040
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.6222222222222236e-06,
+      "loss": 0.9228,
+      "step": 15050
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.613333333333334e-06,
+      "loss": 0.8783,
+      "step": 15060
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.604444444444445e-06,
+      "loss": 0.8029,
+      "step": 15070
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.595555555555556e-06,
+      "loss": 0.8786,
+      "step": 15080
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.5866666666666666e-06,
+      "loss": 0.8431,
+      "step": 15090
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.577777777777779e-06,
+      "loss": 0.8182,
+      "step": 15100
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.568888888888889e-06,
+      "loss": 0.8197,
+      "step": 15110
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 0.7368,
+      "step": 15120
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.551111111111111e-06,
+      "loss": 0.8777,
+      "step": 15130
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.5422222222222235e-06,
+      "loss": 0.8408,
+      "step": 15140
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.533333333333334e-06,
+      "loss": 0.8304,
+      "step": 15150
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.524444444444445e-06,
+      "loss": 0.8911,
+      "step": 15160
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.515555555555556e-06,
+      "loss": 0.8725,
+      "step": 15170
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.5066666666666665e-06,
+      "loss": 0.8376,
+      "step": 15180
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.497777777777779e-06,
+      "loss": 0.8669,
+      "step": 15190
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.488888888888889e-06,
+      "loss": 0.8835,
+      "step": 15200
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.480000000000001e-06,
+      "loss": 0.8586,
+      "step": 15210
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.471111111111111e-06,
+      "loss": 0.7967,
+      "step": 15220
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.462222222222223e-06,
+      "loss": 0.8172,
+      "step": 15230
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.453333333333334e-06,
+      "loss": 0.8763,
+      "step": 15240
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.444444444444445e-06,
+      "loss": 0.8762,
+      "step": 15250
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.435555555555556e-06,
+      "loss": 0.8458,
+      "step": 15260
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.426666666666668e-06,
+      "loss": 0.828,
+      "step": 15270
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.417777777777779e-06,
+      "loss": 0.8802,
+      "step": 15280
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.408888888888889e-06,
+      "loss": 0.8259,
+      "step": 15290
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 0.9081,
+      "step": 15300
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.391111111111111e-06,
+      "loss": 0.911,
+      "step": 15310
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.382222222222223e-06,
+      "loss": 0.886,
+      "step": 15320
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.373333333333334e-06,
+      "loss": 0.8639,
+      "step": 15330
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.364444444444445e-06,
+      "loss": 0.9236,
+      "step": 15340
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.355555555555556e-06,
+      "loss": 0.8687,
+      "step": 15350
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.346666666666668e-06,
+      "loss": 0.7954,
+      "step": 15360
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.3377777777777786e-06,
+      "loss": 0.8017,
+      "step": 15370
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.328888888888889e-06,
+      "loss": 0.8881,
+      "step": 15380
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.3200000000000005e-06,
+      "loss": 0.823,
+      "step": 15390
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.311111111111111e-06,
+      "loss": 0.7838,
+      "step": 15400
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.302222222222223e-06,
+      "loss": 0.8183,
+      "step": 15410
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.293333333333334e-06,
+      "loss": 0.8415,
+      "step": 15420
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.284444444444445e-06,
+      "loss": 0.8234,
+      "step": 15430
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.275555555555556e-06,
+      "loss": 0.7671,
+      "step": 15440
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.266666666666668e-06,
+      "loss": 0.8645,
+      "step": 15450
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.2577777777777785e-06,
+      "loss": 0.8737,
+      "step": 15460
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.24888888888889e-06,
+      "loss": 0.7653,
+      "step": 15470
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.24e-06,
+      "loss": 0.8244,
+      "step": 15480
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.231111111111111e-06,
+      "loss": 0.7947,
+      "step": 15490
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.222222222222223e-06,
+      "loss": 0.842,
+      "step": 15500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.213333333333334e-06,
+      "loss": 0.9095,
+      "step": 15510
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.204444444444445e-06,
+      "loss": 0.8083,
+      "step": 15520
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.195555555555556e-06,
+      "loss": 0.9521,
+      "step": 15530
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.186666666666668e-06,
+      "loss": 0.8497,
+      "step": 15540
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.177777777777778e-06,
+      "loss": 0.8331,
+      "step": 15550
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.16888888888889e-06,
+      "loss": 0.7741,
+      "step": 15560
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.16e-06,
+      "loss": 0.8873,
+      "step": 15570
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.1511111111111125e-06,
+      "loss": 0.86,
+      "step": 15580
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.142222222222223e-06,
+      "loss": 0.8701,
+      "step": 15590
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.133333333333334e-06,
+      "loss": 0.8363,
+      "step": 15600
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.124444444444445e-06,
+      "loss": 0.7626,
+      "step": 15610
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.1155555555555555e-06,
+      "loss": 0.8335,
+      "step": 15620
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.106666666666668e-06,
+      "loss": 0.8826,
+      "step": 15630
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.097777777777778e-06,
+      "loss": 0.8546,
+      "step": 15640
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.08888888888889e-06,
+      "loss": 0.8139,
+      "step": 15650
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.08e-06,
+      "loss": 0.8074,
+      "step": 15660
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0711111111111125e-06,
+      "loss": 0.7886,
+      "step": 15670
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.062222222222223e-06,
+      "loss": 0.8763,
+      "step": 15680
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0533333333333335e-06,
+      "loss": 0.9336,
+      "step": 15690
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.044444444444445e-06,
+      "loss": 0.8344,
+      "step": 15700
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0355555555555555e-06,
+      "loss": 0.8935,
+      "step": 15710
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.026666666666668e-06,
+      "loss": 0.8563,
+      "step": 15720
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.017777777777778e-06,
+      "loss": 0.8974,
+      "step": 15730
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.00888888888889e-06,
+      "loss": 0.7982,
+      "step": 15740
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6e-06,
+      "loss": 0.908,
+      "step": 15750
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.991111111111112e-06,
+      "loss": 0.8391,
+      "step": 15760
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.982222222222223e-06,
+      "loss": 0.8611,
+      "step": 15770
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.973333333333334e-06,
+      "loss": 0.9108,
+      "step": 15780
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.964444444444445e-06,
+      "loss": 0.7565,
+      "step": 15790
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.955555555555555e-06,
+      "loss": 0.8436,
+      "step": 15800
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.946666666666668e-06,
+      "loss": 0.9264,
+      "step": 15810
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.937777777777778e-06,
+      "loss": 0.7446,
+      "step": 15820
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9288888888888895e-06,
+      "loss": 0.8361,
+      "step": 15830
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.92e-06,
+      "loss": 0.7619,
+      "step": 15840
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.911111111111112e-06,
+      "loss": 0.8188,
+      "step": 15850
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.902222222222223e-06,
+      "loss": 0.8139,
+      "step": 15860
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.893333333333334e-06,
+      "loss": 0.7734,
+      "step": 15870
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.884444444444445e-06,
+      "loss": 0.7552,
+      "step": 15880
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.875555555555556e-06,
+      "loss": 0.7679,
+      "step": 15890
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8666666666666675e-06,
+      "loss": 0.8077,
+      "step": 15900
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.857777777777778e-06,
+      "loss": 0.7296,
+      "step": 15910
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8488888888888895e-06,
+      "loss": 0.7994,
+      "step": 15920
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.84e-06,
+      "loss": 0.8408,
+      "step": 15930
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.831111111111112e-06,
+      "loss": 0.8356,
+      "step": 15940
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.822222222222223e-06,
+      "loss": 0.876,
+      "step": 15950
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.813333333333334e-06,
+      "loss": 0.7415,
+      "step": 15960
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.804444444444445e-06,
+      "loss": 0.895,
+      "step": 15970
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.795555555555557e-06,
+      "loss": 0.8641,
+      "step": 15980
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7866666666666674e-06,
+      "loss": 0.8357,
+      "step": 15990
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.777777777777778e-06,
+      "loss": 0.8129,
+      "step": 16000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.768888888888889e-06,
+      "loss": 0.7297,
+      "step": 16010
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.76e-06,
+      "loss": 0.8158,
+      "step": 16020
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.751111111111112e-06,
+      "loss": 0.8921,
+      "step": 16030
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.742222222222223e-06,
+      "loss": 0.8021,
+      "step": 16040
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.733333333333334e-06,
+      "loss": 0.8179,
+      "step": 16050
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.724444444444445e-06,
+      "loss": 0.7971,
+      "step": 16060
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.715555555555557e-06,
+      "loss": 0.7824,
+      "step": 16070
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.706666666666667e-06,
+      "loss": 0.8772,
+      "step": 16080
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.697777777777779e-06,
+      "loss": 0.8576,
+      "step": 16090
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.688888888888889e-06,
+      "loss": 0.7725,
+      "step": 16100
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.68e-06,
+      "loss": 0.8302,
+      "step": 16110
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.671111111111112e-06,
+      "loss": 0.8072,
+      "step": 16120
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.662222222222223e-06,
+      "loss": 0.8582,
+      "step": 16130
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.653333333333334e-06,
+      "loss": 0.8676,
+      "step": 16140
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6444444444444445e-06,
+      "loss": 0.7829,
+      "step": 16150
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.635555555555557e-06,
+      "loss": 0.8062,
+      "step": 16160
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.626666666666667e-06,
+      "loss": 0.7902,
+      "step": 16170
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.617777777777779e-06,
+      "loss": 0.8327,
+      "step": 16180
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.608888888888889e-06,
+      "loss": 0.8609,
+      "step": 16190
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 0.8131,
+      "step": 16200
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.591111111111112e-06,
+      "loss": 0.8061,
+      "step": 16210
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.5822222222222225e-06,
+      "loss": 0.8494,
+      "step": 16220
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.573333333333334e-06,
+      "loss": 0.8691,
+      "step": 16230
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5644444444444444e-06,
+      "loss": 0.7922,
+      "step": 16240
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.555555555555557e-06,
+      "loss": 0.8658,
+      "step": 16250
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.546666666666667e-06,
+      "loss": 0.7882,
+      "step": 16260
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.537777777777779e-06,
+      "loss": 0.8414,
+      "step": 16270
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.528888888888889e-06,
+      "loss": 0.8258,
+      "step": 16280
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5200000000000005e-06,
+      "loss": 0.8152,
+      "step": 16290
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.511111111111112e-06,
+      "loss": 0.862,
+      "step": 16300
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5022222222222224e-06,
+      "loss": 0.8049,
+      "step": 16310
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.493333333333334e-06,
+      "loss": 0.7815,
+      "step": 16320
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.484444444444444e-06,
+      "loss": 0.7657,
+      "step": 16330
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.475555555555557e-06,
+      "loss": 0.8149,
+      "step": 16340
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.466666666666667e-06,
+      "loss": 0.7863,
+      "step": 16350
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4577777777777785e-06,
+      "loss": 0.7555,
+      "step": 16360
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.448888888888889e-06,
+      "loss": 0.8423,
+      "step": 16370
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 0.8865,
+      "step": 16380
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.431111111111112e-06,
+      "loss": 0.8595,
+      "step": 16390
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.422222222222223e-06,
+      "loss": 0.8932,
+      "step": 16400
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.413333333333334e-06,
+      "loss": 0.7722,
+      "step": 16410
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.404444444444444e-06,
+      "loss": 0.869,
+      "step": 16420
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.3955555555555565e-06,
+      "loss": 0.829,
+      "step": 16430
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.386666666666667e-06,
+      "loss": 0.7943,
+      "step": 16440
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.3777777777777784e-06,
+      "loss": 0.8216,
+      "step": 16450
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.368888888888889e-06,
+      "loss": 0.8164,
+      "step": 16460
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.36e-06,
+      "loss": 0.7457,
+      "step": 16470
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.351111111111112e-06,
+      "loss": 0.8391,
+      "step": 16480
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.342222222222223e-06,
+      "loss": 0.846,
+      "step": 16490
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 0.7716,
+      "step": 16500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.324444444444445e-06,
+      "loss": 0.8338,
+      "step": 16510
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3155555555555564e-06,
+      "loss": 0.9658,
+      "step": 16520
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.306666666666667e-06,
+      "loss": 0.8611,
+      "step": 16530
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.297777777777778e-06,
+      "loss": 0.8649,
+      "step": 16540
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.288888888888889e-06,
+      "loss": 0.7962,
+      "step": 16550
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.28e-06,
+      "loss": 0.8775,
+      "step": 16560
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.271111111111112e-06,
+      "loss": 0.8436,
+      "step": 16570
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.262222222222223e-06,
+      "loss": 0.7496,
+      "step": 16580
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2533333333333336e-06,
+      "loss": 0.787,
+      "step": 16590
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.244444444444445e-06,
+      "loss": 0.8345,
+      "step": 16600
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.235555555555556e-06,
+      "loss": 0.8504,
+      "step": 16610
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.226666666666667e-06,
+      "loss": 0.8282,
+      "step": 16620
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.217777777777778e-06,
+      "loss": 0.811,
+      "step": 16630
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.208888888888889e-06,
+      "loss": 0.8029,
+      "step": 16640
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.2e-06,
+      "loss": 0.8608,
+      "step": 16650
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.1911111111111116e-06,
+      "loss": 0.7647,
+      "step": 16660
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.182222222222223e-06,
+      "loss": 0.7712,
+      "step": 16670
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.1733333333333335e-06,
+      "loss": 0.8984,
+      "step": 16680
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.164444444444445e-06,
+      "loss": 0.8008,
+      "step": 16690
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.155555555555556e-06,
+      "loss": 0.882,
+      "step": 16700
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.146666666666668e-06,
+      "loss": 0.8265,
+      "step": 16710
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.137777777777778e-06,
+      "loss": 0.8459,
+      "step": 16720
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.128888888888889e-06,
+      "loss": 0.8008,
+      "step": 16730
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.12e-06,
+      "loss": 0.8676,
+      "step": 16740
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1111111111111115e-06,
+      "loss": 0.8191,
+      "step": 16750
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.102222222222223e-06,
+      "loss": 0.7605,
+      "step": 16760
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.093333333333333e-06,
+      "loss": 0.8374,
+      "step": 16770
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.084444444444445e-06,
+      "loss": 0.8522,
+      "step": 16780
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.075555555555556e-06,
+      "loss": 0.7552,
+      "step": 16790
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0666666666666676e-06,
+      "loss": 0.7798,
+      "step": 16800
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.057777777777778e-06,
+      "loss": 0.8763,
+      "step": 16810
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0488888888888895e-06,
+      "loss": 0.8432,
+      "step": 16820
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.04e-06,
+      "loss": 0.8978,
+      "step": 16830
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.031111111111111e-06,
+      "loss": 0.7981,
+      "step": 16840
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.022222222222223e-06,
+      "loss": 0.7606,
+      "step": 16850
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.013333333333333e-06,
+      "loss": 0.7978,
+      "step": 16860
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.004444444444445e-06,
+      "loss": 0.7653,
+      "step": 16870
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.995555555555556e-06,
+      "loss": 0.8516,
+      "step": 16880
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.986666666666667e-06,
+      "loss": 0.8663,
+      "step": 16890
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.977777777777778e-06,
+      "loss": 0.886,
+      "step": 16900
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.968888888888889e-06,
+      "loss": 0.824,
+      "step": 16910
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 0.7939,
+      "step": 16920
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.951111111111111e-06,
+      "loss": 0.7669,
+      "step": 16930
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.942222222222223e-06,
+      "loss": 0.8419,
+      "step": 16940
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.933333333333334e-06,
+      "loss": 0.786,
+      "step": 16950
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.924444444444445e-06,
+      "loss": 0.7915,
+      "step": 16960
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.915555555555556e-06,
+      "loss": 0.8393,
+      "step": 16970
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9066666666666666e-06,
+      "loss": 0.7761,
+      "step": 16980
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.897777777777778e-06,
+      "loss": 0.9133,
+      "step": 16990
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.888888888888889e-06,
+      "loss": 0.7557,
+      "step": 17000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 0.8233,
+      "step": 17010
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.871111111111111e-06,
+      "loss": 0.8668,
+      "step": 17020
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.862222222222223e-06,
+      "loss": 0.8814,
+      "step": 17030
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.853333333333334e-06,
+      "loss": 0.7857,
+      "step": 17040
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8444444444444446e-06,
+      "loss": 0.8391,
+      "step": 17050
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.835555555555556e-06,
+      "loss": 0.7768,
+      "step": 17060
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.826666666666667e-06,
+      "loss": 0.7996,
+      "step": 17070
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.817777777777779e-06,
+      "loss": 0.7801,
+      "step": 17080
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.808888888888889e-06,
+      "loss": 0.8352,
+      "step": 17090
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.8868,
+      "step": 17100
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.791111111111111e-06,
+      "loss": 0.8935,
+      "step": 17110
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.7822222222222226e-06,
+      "loss": 0.8162,
+      "step": 17120
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.773333333333334e-06,
+      "loss": 0.8463,
+      "step": 17130
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7644444444444445e-06,
+      "loss": 0.8191,
+      "step": 17140
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.755555555555556e-06,
+      "loss": 0.8515,
+      "step": 17150
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.746666666666667e-06,
+      "loss": 0.8806,
+      "step": 17160
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.737777777777779e-06,
+      "loss": 0.8262,
+      "step": 17170
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.728888888888889e-06,
+      "loss": 0.7989,
+      "step": 17180
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7200000000000005e-06,
+      "loss": 0.8395,
+      "step": 17190
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.711111111111111e-06,
+      "loss": 0.8362,
+      "step": 17200
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7022222222222225e-06,
+      "loss": 0.8672,
+      "step": 17210
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.693333333333334e-06,
+      "loss": 0.8261,
+      "step": 17220
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.684444444444444e-06,
+      "loss": 0.8341,
+      "step": 17230
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.675555555555556e-06,
+      "loss": 0.8509,
+      "step": 17240
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.666666666666667e-06,
+      "loss": 0.8656,
+      "step": 17250
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6577777777777785e-06,
+      "loss": 0.8361,
+      "step": 17260
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.648888888888889e-06,
+      "loss": 0.8321,
+      "step": 17270
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 0.8629,
+      "step": 17280
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.631111111111111e-06,
+      "loss": 0.7699,
+      "step": 17290
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.622222222222222e-06,
+      "loss": 0.873,
+      "step": 17300
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.613333333333334e-06,
+      "loss": 0.7825,
+      "step": 17310
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.604444444444444e-06,
+      "loss": 0.83,
+      "step": 17320
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.595555555555556e-06,
+      "loss": 0.8903,
+      "step": 17330
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.586666666666667e-06,
+      "loss": 0.8419,
+      "step": 17340
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.5777777777777785e-06,
+      "loss": 0.8148,
+      "step": 17350
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.568888888888889e-06,
+      "loss": 0.8523,
+      "step": 17360
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.56e-06,
+      "loss": 0.9085,
+      "step": 17370
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.551111111111112e-06,
+      "loss": 0.8006,
+      "step": 17380
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.542222222222223e-06,
+      "loss": 0.8137,
+      "step": 17390
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.533333333333334e-06,
+      "loss": 0.785,
+      "step": 17400
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.524444444444444e-06,
+      "loss": 0.8057,
+      "step": 17410
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.515555555555556e-06,
+      "loss": 0.7772,
+      "step": 17420
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.506666666666667e-06,
+      "loss": 0.8729,
+      "step": 17430
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.497777777777778e-06,
+      "loss": 0.8522,
+      "step": 17440
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.488888888888889e-06,
+      "loss": 0.8714,
+      "step": 17450
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.48e-06,
+      "loss": 0.8497,
+      "step": 17460
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.471111111111112e-06,
+      "loss": 0.8774,
+      "step": 17470
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.462222222222223e-06,
+      "loss": 0.9145,
+      "step": 17480
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.453333333333334e-06,
+      "loss": 0.8244,
+      "step": 17490
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 0.8455,
+      "step": 17500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.4355555555555555e-06,
+      "loss": 0.8221,
+      "step": 17510
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.426666666666667e-06,
+      "loss": 0.9013,
+      "step": 17520
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.417777777777778e-06,
+      "loss": 0.8473,
+      "step": 17530
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.408888888888889e-06,
+      "loss": 0.8395,
+      "step": 17540
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.4e-06,
+      "loss": 0.7817,
+      "step": 17550
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.391111111111112e-06,
+      "loss": 0.848,
+      "step": 17560
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.382222222222223e-06,
+      "loss": 0.7769,
+      "step": 17570
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.3733333333333335e-06,
+      "loss": 0.802,
+      "step": 17580
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.364444444444445e-06,
+      "loss": 0.7605,
+      "step": 17590
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3555555555555555e-06,
+      "loss": 0.8052,
+      "step": 17600
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.346666666666667e-06,
+      "loss": 0.8146,
+      "step": 17610
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.337777777777778e-06,
+      "loss": 0.8941,
+      "step": 17620
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.328888888888889e-06,
+      "loss": 0.8212,
+      "step": 17630
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.32e-06,
+      "loss": 0.7508,
+      "step": 17640
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3111111111111115e-06,
+      "loss": 0.7214,
+      "step": 17650
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.302222222222223e-06,
+      "loss": 0.8685,
+      "step": 17660
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2933333333333334e-06,
+      "loss": 0.8602,
+      "step": 17670
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.284444444444445e-06,
+      "loss": 0.8153,
+      "step": 17680
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.275555555555556e-06,
+      "loss": 0.8818,
+      "step": 17690
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.266666666666668e-06,
+      "loss": 0.836,
+      "step": 17700
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.257777777777778e-06,
+      "loss": 0.8321,
+      "step": 17710
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.248888888888889e-06,
+      "loss": 0.8192,
+      "step": 17720
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.24e-06,
+      "loss": 0.8141,
+      "step": 17730
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.2311111111111114e-06,
+      "loss": 0.7721,
+      "step": 17740
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.222222222222223e-06,
+      "loss": 0.8838,
+      "step": 17750
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.213333333333333e-06,
+      "loss": 0.9243,
+      "step": 17760
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.204444444444445e-06,
+      "loss": 0.7648,
+      "step": 17770
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.195555555555556e-06,
+      "loss": 0.8362,
+      "step": 17780
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.1866666666666675e-06,
+      "loss": 0.7992,
+      "step": 17790
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.177777777777778e-06,
+      "loss": 0.8351,
+      "step": 17800
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.168888888888889e-06,
+      "loss": 0.8035,
+      "step": 17810
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.16e-06,
+      "loss": 0.7604,
+      "step": 17820
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.151111111111111e-06,
+      "loss": 0.9146,
+      "step": 17830
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.142222222222223e-06,
+      "loss": 0.8055,
+      "step": 17840
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.133333333333333e-06,
+      "loss": 0.8951,
+      "step": 17850
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.124444444444445e-06,
+      "loss": 0.814,
+      "step": 17860
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.115555555555556e-06,
+      "loss": 0.814,
+      "step": 17870
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1066666666666674e-06,
+      "loss": 0.7765,
+      "step": 17880
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.097777777777778e-06,
+      "loss": 0.8106,
+      "step": 17890
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.088888888888889e-06,
+      "loss": 0.896,
+      "step": 17900
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.08e-06,
+      "loss": 0.8342,
+      "step": 17910
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.071111111111111e-06,
+      "loss": 0.8444,
+      "step": 17920
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.062222222222223e-06,
+      "loss": 0.8271,
+      "step": 17930
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.053333333333333e-06,
+      "loss": 0.8578,
+      "step": 17940
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.044444444444445e-06,
+      "loss": 0.7923,
+      "step": 17950
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.035555555555556e-06,
+      "loss": 0.843,
+      "step": 17960
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.026666666666667e-06,
+      "loss": 0.8322,
+      "step": 17970
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.017777777777778e-06,
+      "loss": 0.7879,
+      "step": 17980
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.008888888888889e-06,
+      "loss": 0.849,
+      "step": 17990
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.8924,
+      "step": 18000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.991111111111112e-06,
+      "loss": 0.8273,
+      "step": 18010
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.982222222222223e-06,
+      "loss": 0.8015,
+      "step": 18020
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.973333333333333e-06,
+      "loss": 0.8852,
+      "step": 18030
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9644444444444445e-06,
+      "loss": 0.8027,
+      "step": 18040
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.955555555555556e-06,
+      "loss": 0.8862,
+      "step": 18050
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.946666666666667e-06,
+      "loss": 0.907,
+      "step": 18060
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.937777777777778e-06,
+      "loss": 0.7465,
+      "step": 18070
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.928888888888889e-06,
+      "loss": 0.8386,
+      "step": 18080
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 0.7733,
+      "step": 18090
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.911111111111112e-06,
+      "loss": 0.9296,
+      "step": 18100
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9022222222222225e-06,
+      "loss": 0.8008,
+      "step": 18110
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.893333333333333e-06,
+      "loss": 0.7517,
+      "step": 18120
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8844444444444444e-06,
+      "loss": 0.9139,
+      "step": 18130
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.875555555555556e-06,
+      "loss": 0.7209,
+      "step": 18140
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.866666666666667e-06,
+      "loss": 0.8537,
+      "step": 18150
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.857777777777778e-06,
+      "loss": 0.7256,
+      "step": 18160
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.848888888888889e-06,
+      "loss": 0.7867,
+      "step": 18170
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 0.7737,
+      "step": 18180
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.831111111111112e-06,
+      "loss": 0.9487,
+      "step": 18190
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.8222222222222224e-06,
+      "loss": 0.8103,
+      "step": 18200
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.813333333333334e-06,
+      "loss": 0.8619,
+      "step": 18210
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.8044444444444443e-06,
+      "loss": 0.7318,
+      "step": 18220
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7955555555555557e-06,
+      "loss": 0.8405,
+      "step": 18230
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7866666666666667e-06,
+      "loss": 0.792,
+      "step": 18240
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.777777777777778e-06,
+      "loss": 0.8246,
+      "step": 18250
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.768888888888889e-06,
+      "loss": 0.8587,
+      "step": 18260
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 0.8668,
+      "step": 18270
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7511111111111114e-06,
+      "loss": 0.8428,
+      "step": 18280
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7422222222222228e-06,
+      "loss": 0.8274,
+      "step": 18290
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7333333333333337e-06,
+      "loss": 0.8453,
+      "step": 18300
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.724444444444445e-06,
+      "loss": 0.8106,
+      "step": 18310
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7155555555555557e-06,
+      "loss": 0.7368,
+      "step": 18320
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7066666666666666e-06,
+      "loss": 0.9053,
+      "step": 18330
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.697777777777778e-06,
+      "loss": 0.7942,
+      "step": 18340
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.688888888888889e-06,
+      "loss": 0.8066,
+      "step": 18350
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 0.801,
+      "step": 18360
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6711111111111113e-06,
+      "loss": 0.8932,
+      "step": 18370
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6622222222222227e-06,
+      "loss": 0.8557,
+      "step": 18380
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6533333333333336e-06,
+      "loss": 0.8394,
+      "step": 18390
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.644444444444445e-06,
+      "loss": 0.8468,
+      "step": 18400
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.635555555555556e-06,
+      "loss": 0.7558,
+      "step": 18410
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6266666666666674e-06,
+      "loss": 0.8375,
+      "step": 18420
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.617777777777778e-06,
+      "loss": 0.8555,
+      "step": 18430
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.608888888888889e-06,
+      "loss": 0.8171,
+      "step": 18440
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 0.8113,
+      "step": 18450
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.5911111111111112e-06,
+      "loss": 0.7679,
+      "step": 18460
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.5822222222222226e-06,
+      "loss": 0.7822,
+      "step": 18470
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.5733333333333336e-06,
+      "loss": 0.7701,
+      "step": 18480
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.564444444444445e-06,
+      "loss": 0.7458,
+      "step": 18490
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.555555555555556e-06,
+      "loss": 0.8055,
+      "step": 18500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5466666666666673e-06,
+      "loss": 0.8035,
+      "step": 18510
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5377777777777783e-06,
+      "loss": 0.7745,
+      "step": 18520
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.528888888888889e-06,
+      "loss": 0.8443,
+      "step": 18530
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.52e-06,
+      "loss": 0.8184,
+      "step": 18540
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.511111111111111e-06,
+      "loss": 0.8289,
+      "step": 18550
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5022222222222225e-06,
+      "loss": 0.7986,
+      "step": 18560
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4933333333333335e-06,
+      "loss": 0.8359,
+      "step": 18570
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.484444444444445e-06,
+      "loss": 0.7876,
+      "step": 18580
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.475555555555556e-06,
+      "loss": 0.8116,
+      "step": 18590
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4666666666666672e-06,
+      "loss": 0.8817,
+      "step": 18600
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.457777777777778e-06,
+      "loss": 0.7765,
+      "step": 18610
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4488888888888896e-06,
+      "loss": 0.7775,
+      "step": 18620
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.44e-06,
+      "loss": 0.8758,
+      "step": 18630
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.431111111111111e-06,
+      "loss": 0.9072,
+      "step": 18640
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4222222222222224e-06,
+      "loss": 0.8429,
+      "step": 18650
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4133333333333334e-06,
+      "loss": 0.7583,
+      "step": 18660
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.404444444444445e-06,
+      "loss": 0.806,
+      "step": 18670
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.3955555555555558e-06,
+      "loss": 0.8403,
+      "step": 18680
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.386666666666667e-06,
+      "loss": 0.7127,
+      "step": 18690
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.377777777777778e-06,
+      "loss": 0.7831,
+      "step": 18700
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.3688888888888895e-06,
+      "loss": 0.8528,
+      "step": 18710
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 0.8426,
+      "step": 18720
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.351111111111112e-06,
+      "loss": 0.8156,
+      "step": 18730
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3422222222222224e-06,
+      "loss": 0.9364,
+      "step": 18740
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.7374,
+      "step": 18750
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3244444444444447e-06,
+      "loss": 0.9053,
+      "step": 18760
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3155555555555557e-06,
+      "loss": 0.8288,
+      "step": 18770
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.306666666666667e-06,
+      "loss": 0.8866,
+      "step": 18780
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.297777777777778e-06,
+      "loss": 0.8698,
+      "step": 18790
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2888888888888894e-06,
+      "loss": 0.8481,
+      "step": 18800
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 0.8208,
+      "step": 18810
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2711111111111117e-06,
+      "loss": 0.7736,
+      "step": 18820
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2622222222222227e-06,
+      "loss": 0.8058,
+      "step": 18830
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2533333333333332e-06,
+      "loss": 0.8071,
+      "step": 18840
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2444444444444446e-06,
+      "loss": 0.8402,
+      "step": 18850
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2355555555555556e-06,
+      "loss": 0.8376,
+      "step": 18860
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.226666666666667e-06,
+      "loss": 0.831,
+      "step": 18870
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.217777777777778e-06,
+      "loss": 0.7468,
+      "step": 18880
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2088888888888893e-06,
+      "loss": 0.842,
+      "step": 18890
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.728,
+      "step": 18900
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.1911111111111117e-06,
+      "loss": 0.743,
+      "step": 18910
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.1822222222222226e-06,
+      "loss": 0.8161,
+      "step": 18920
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.173333333333334e-06,
+      "loss": 0.8466,
+      "step": 18930
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1644444444444445e-06,
+      "loss": 0.766,
+      "step": 18940
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1555555555555555e-06,
+      "loss": 0.7754,
+      "step": 18950
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.146666666666667e-06,
+      "loss": 0.8205,
+      "step": 18960
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.137777777777778e-06,
+      "loss": 0.844,
+      "step": 18970
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1288888888888892e-06,
+      "loss": 0.8461,
+      "step": 18980
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.12e-06,
+      "loss": 0.8228,
+      "step": 18990
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1111111111111116e-06,
+      "loss": 0.8773,
+      "step": 19000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1022222222222225e-06,
+      "loss": 0.8911,
+      "step": 19010
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.093333333333334e-06,
+      "loss": 0.9241,
+      "step": 19020
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.084444444444445e-06,
+      "loss": 0.7959,
+      "step": 19030
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0755555555555563e-06,
+      "loss": 0.8512,
+      "step": 19040
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.066666666666667e-06,
+      "loss": 0.7935,
+      "step": 19050
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0577777777777778e-06,
+      "loss": 0.8001,
+      "step": 19060
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.048888888888889e-06,
+      "loss": 0.8499,
+      "step": 19070
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.04e-06,
+      "loss": 0.7867,
+      "step": 19080
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0311111111111115e-06,
+      "loss": 0.8094,
+      "step": 19090
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0222222222222225e-06,
+      "loss": 0.8704,
+      "step": 19100
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.013333333333334e-06,
+      "loss": 0.8264,
+      "step": 19110
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.004444444444445e-06,
+      "loss": 0.7618,
+      "step": 19120
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.995555555555556e-06,
+      "loss": 0.8411,
+      "step": 19130
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.986666666666667e-06,
+      "loss": 0.776,
+      "step": 19140
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9777777777777777e-06,
+      "loss": 0.8921,
+      "step": 19150
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.968888888888889e-06,
+      "loss": 0.7897,
+      "step": 19160
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.96e-06,
+      "loss": 0.8343,
+      "step": 19170
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9511111111111114e-06,
+      "loss": 0.8287,
+      "step": 19180
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9422222222222224e-06,
+      "loss": 0.7859,
+      "step": 19190
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9333333333333338e-06,
+      "loss": 0.8222,
+      "step": 19200
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9244444444444447e-06,
+      "loss": 0.7744,
+      "step": 19210
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.915555555555556e-06,
+      "loss": 0.9075,
+      "step": 19220
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.906666666666667e-06,
+      "loss": 0.8316,
+      "step": 19230
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8977777777777785e-06,
+      "loss": 0.8619,
+      "step": 19240
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.888888888888889e-06,
+      "loss": 0.7999,
+      "step": 19250
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.88e-06,
+      "loss": 0.8379,
+      "step": 19260
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8711111111111113e-06,
+      "loss": 0.8196,
+      "step": 19270
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8622222222222223e-06,
+      "loss": 0.9262,
+      "step": 19280
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8533333333333337e-06,
+      "loss": 0.7689,
+      "step": 19290
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8444444444444446e-06,
+      "loss": 0.7483,
+      "step": 19300
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.835555555555556e-06,
+      "loss": 0.8292,
+      "step": 19310
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.826666666666667e-06,
+      "loss": 0.8049,
+      "step": 19320
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8177777777777784e-06,
+      "loss": 0.8726,
+      "step": 19330
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8088888888888893e-06,
+      "loss": 0.9292,
+      "step": 19340
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 0.9249,
+      "step": 19350
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7911111111111113e-06,
+      "loss": 0.7495,
+      "step": 19360
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7822222222222222e-06,
+      "loss": 0.7441,
+      "step": 19370
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7733333333333336e-06,
+      "loss": 0.8716,
+      "step": 19380
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7644444444444446e-06,
+      "loss": 0.7759,
+      "step": 19390
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.755555555555556e-06,
+      "loss": 0.9075,
+      "step": 19400
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.746666666666667e-06,
+      "loss": 0.7501,
+      "step": 19410
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7377777777777783e-06,
+      "loss": 0.8682,
+      "step": 19420
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7288888888888893e-06,
+      "loss": 0.8193,
+      "step": 19430
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 0.8806,
+      "step": 19440
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7111111111111116e-06,
+      "loss": 0.8633,
+      "step": 19450
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.702222222222222e-06,
+      "loss": 0.8159,
+      "step": 19460
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6933333333333335e-06,
+      "loss": 0.823,
+      "step": 19470
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6844444444444445e-06,
+      "loss": 0.7793,
+      "step": 19480
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.675555555555556e-06,
+      "loss": 0.7758,
+      "step": 19490
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 0.7891,
+      "step": 19500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6577777777777782e-06,
+      "loss": 0.7241,
+      "step": 19510
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.648888888888889e-06,
+      "loss": 0.8179,
+      "step": 19520
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.64e-06,
+      "loss": 0.8493,
+      "step": 19530
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6311111111111115e-06,
+      "loss": 0.8218,
+      "step": 19540
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6222222222222225e-06,
+      "loss": 0.8835,
+      "step": 19550
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6133333333333334e-06,
+      "loss": 0.9047,
+      "step": 19560
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6044444444444444e-06,
+      "loss": 0.8495,
+      "step": 19570
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5955555555555558e-06,
+      "loss": 0.8662,
+      "step": 19580
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5866666666666667e-06,
+      "loss": 0.7639,
+      "step": 19590
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.577777777777778e-06,
+      "loss": 0.8361,
+      "step": 19600
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.568888888888889e-06,
+      "loss": 0.8461,
+      "step": 19610
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.56e-06,
+      "loss": 0.7953,
+      "step": 19620
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5511111111111114e-06,
+      "loss": 0.8043,
+      "step": 19630
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5422222222222224e-06,
+      "loss": 0.7502,
+      "step": 19640
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5333333333333338e-06,
+      "loss": 0.8066,
+      "step": 19650
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5244444444444447e-06,
+      "loss": 0.8244,
+      "step": 19660
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5155555555555557e-06,
+      "loss": 0.7951,
+      "step": 19670
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5066666666666667e-06,
+      "loss": 0.7164,
+      "step": 19680
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.497777777777778e-06,
+      "loss": 0.8115,
+      "step": 19690
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.488888888888889e-06,
+      "loss": 0.825,
+      "step": 19700
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 0.7663,
+      "step": 19710
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4711111111111114e-06,
+      "loss": 0.7439,
+      "step": 19720
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4622222222222223e-06,
+      "loss": 0.8784,
+      "step": 19730
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4533333333333333e-06,
+      "loss": 0.8196,
+      "step": 19740
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4444444444444447e-06,
+      "loss": 0.8477,
+      "step": 19750
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4355555555555556e-06,
+      "loss": 0.8692,
+      "step": 19760
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.426666666666667e-06,
+      "loss": 0.837,
+      "step": 19770
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.417777777777778e-06,
+      "loss": 0.7935,
+      "step": 19780
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4088888888888894e-06,
+      "loss": 0.8376,
+      "step": 19790
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 0.7074,
+      "step": 19800
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3911111111111113e-06,
+      "loss": 0.7487,
+      "step": 19810
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3822222222222222e-06,
+      "loss": 0.8338,
+      "step": 19820
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3733333333333336e-06,
+      "loss": 0.7537,
+      "step": 19830
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3644444444444446e-06,
+      "loss": 0.8061,
+      "step": 19840
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3555555555555555e-06,
+      "loss": 0.8296,
+      "step": 19850
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.346666666666667e-06,
+      "loss": 0.7684,
+      "step": 19860
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.337777777777778e-06,
+      "loss": 0.8488,
+      "step": 19870
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3288888888888893e-06,
+      "loss": 0.9068,
+      "step": 19880
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 0.7854,
+      "step": 19890
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.311111111111111e-06,
+      "loss": 0.8545,
+      "step": 19900
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.302222222222222e-06,
+      "loss": 0.8685,
+      "step": 19910
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2933333333333335e-06,
+      "loss": 0.7214,
+      "step": 19920
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2844444444444445e-06,
+      "loss": 0.8166,
+      "step": 19930
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.275555555555556e-06,
+      "loss": 0.7984,
+      "step": 19940
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.266666666666667e-06,
+      "loss": 0.7398,
+      "step": 19950
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.257777777777778e-06,
+      "loss": 0.7805,
+      "step": 19960
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.248888888888889e-06,
+      "loss": 0.7772,
+      "step": 19970
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.24e-06,
+      "loss": 0.7705,
+      "step": 19980
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2311111111111115e-06,
+      "loss": 0.8715,
+      "step": 19990
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 0.8495,
+      "step": 20000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2133333333333335e-06,
+      "loss": 0.8954,
+      "step": 20010
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2044444444444444e-06,
+      "loss": 0.8071,
+      "step": 20020
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.195555555555556e-06,
+      "loss": 0.8287,
+      "step": 20030
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1866666666666668e-06,
+      "loss": 0.8425,
+      "step": 20040
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1777777777777777e-06,
+      "loss": 0.82,
+      "step": 20050
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.168888888888889e-06,
+      "loss": 0.8223,
+      "step": 20060
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.16e-06,
+      "loss": 0.8348,
+      "step": 20070
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1511111111111115e-06,
+      "loss": 0.7894,
+      "step": 20080
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1422222222222224e-06,
+      "loss": 0.7856,
+      "step": 20090
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.133333333333334e-06,
+      "loss": 0.8029,
+      "step": 20100
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1244444444444443e-06,
+      "loss": 0.8247,
+      "step": 20110
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1155555555555557e-06,
+      "loss": 0.8485,
+      "step": 20120
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1066666666666667e-06,
+      "loss": 0.8163,
+      "step": 20130
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.097777777777778e-06,
+      "loss": 0.8047,
+      "step": 20140
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.088888888888889e-06,
+      "loss": 0.7927,
+      "step": 20150
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.08e-06,
+      "loss": 0.8456,
+      "step": 20160
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0711111111111114e-06,
+      "loss": 0.7843,
+      "step": 20170
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0622222222222223e-06,
+      "loss": 0.7923,
+      "step": 20180
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0533333333333337e-06,
+      "loss": 0.8023,
+      "step": 20190
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0444444444444447e-06,
+      "loss": 0.8798,
+      "step": 20200
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0355555555555556e-06,
+      "loss": 0.7957,
+      "step": 20210
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0266666666666666e-06,
+      "loss": 0.877,
+      "step": 20220
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.017777777777778e-06,
+      "loss": 0.8563,
+      "step": 20230
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.008888888888889e-06,
+      "loss": 0.8301,
+      "step": 20240
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.8502,
+      "step": 20250
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9911111111111113e-06,
+      "loss": 0.8726,
+      "step": 20260
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9822222222222223e-06,
+      "loss": 0.7398,
+      "step": 20270
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9733333333333336e-06,
+      "loss": 0.733,
+      "step": 20280
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9644444444444446e-06,
+      "loss": 0.8428,
+      "step": 20290
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.955555555555556e-06,
+      "loss": 0.8025,
+      "step": 20300
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9466666666666665e-06,
+      "loss": 0.8166,
+      "step": 20310
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.937777777777778e-06,
+      "loss": 0.8525,
+      "step": 20320
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.928888888888889e-06,
+      "loss": 0.7615,
+      "step": 20330
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 0.8126,
+      "step": 20340
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9111111111111112e-06,
+      "loss": 0.8176,
+      "step": 20350
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9022222222222222e-06,
+      "loss": 0.7929,
+      "step": 20360
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8933333333333333e-06,
+      "loss": 0.8086,
+      "step": 20370
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8844444444444445e-06,
+      "loss": 0.8019,
+      "step": 20380
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8755555555555557e-06,
+      "loss": 0.8072,
+      "step": 20390
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8666666666666669e-06,
+      "loss": 0.8403,
+      "step": 20400
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8577777777777778e-06,
+      "loss": 0.8445,
+      "step": 20410
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.848888888888889e-06,
+      "loss": 0.7939,
+      "step": 20420
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 0.8084,
+      "step": 20430
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8311111111111113e-06,
+      "loss": 0.8025,
+      "step": 20440
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8222222222222225e-06,
+      "loss": 0.8008,
+      "step": 20450
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8133333333333337e-06,
+      "loss": 0.8089,
+      "step": 20460
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8044444444444444e-06,
+      "loss": 0.8157,
+      "step": 20470
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7955555555555556e-06,
+      "loss": 0.7951,
+      "step": 20480
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7866666666666668e-06,
+      "loss": 0.8361,
+      "step": 20490
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.777777777777778e-06,
+      "loss": 0.8473,
+      "step": 20500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7688888888888891e-06,
+      "loss": 0.8343,
+      "step": 20510
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.76e-06,
+      "loss": 0.8609,
+      "step": 20520
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7511111111111113e-06,
+      "loss": 0.8423,
+      "step": 20530
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7422222222222224e-06,
+      "loss": 0.9081,
+      "step": 20540
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7333333333333336e-06,
+      "loss": 0.8152,
+      "step": 20550
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7244444444444448e-06,
+      "loss": 0.8906,
+      "step": 20560
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7155555555555555e-06,
+      "loss": 0.8185,
+      "step": 20570
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7066666666666667e-06,
+      "loss": 0.7682,
+      "step": 20580
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6977777777777779e-06,
+      "loss": 0.7807,
+      "step": 20590
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.688888888888889e-06,
+      "loss": 0.769,
+      "step": 20600
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 0.7894,
+      "step": 20610
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6711111111111112e-06,
+      "loss": 0.841,
+      "step": 20620
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6622222222222224e-06,
+      "loss": 0.8328,
+      "step": 20630
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6533333333333335e-06,
+      "loss": 0.7972,
+      "step": 20640
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6444444444444447e-06,
+      "loss": 0.8104,
+      "step": 20650
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6355555555555559e-06,
+      "loss": 0.8485,
+      "step": 20660
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6266666666666666e-06,
+      "loss": 0.833,
+      "step": 20670
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6177777777777778e-06,
+      "loss": 0.8058,
+      "step": 20680
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.608888888888889e-06,
+      "loss": 0.7509,
+      "step": 20690
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 0.7888,
+      "step": 20700
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5911111111111113e-06,
+      "loss": 0.7974,
+      "step": 20710
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5822222222222223e-06,
+      "loss": 0.8466,
+      "step": 20720
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5733333333333334e-06,
+      "loss": 0.765,
+      "step": 20730
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5644444444444446e-06,
+      "loss": 0.8338,
+      "step": 20740
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5555555555555558e-06,
+      "loss": 0.7821,
+      "step": 20750
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.546666666666667e-06,
+      "loss": 0.8918,
+      "step": 20760
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5377777777777781e-06,
+      "loss": 0.7907,
+      "step": 20770
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5288888888888889e-06,
+      "loss": 0.8502,
+      "step": 20780
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.52e-06,
+      "loss": 0.8497,
+      "step": 20790
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5111111111111112e-06,
+      "loss": 0.8511,
+      "step": 20800
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5022222222222224e-06,
+      "loss": 0.8303,
+      "step": 20810
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4933333333333336e-06,
+      "loss": 0.8249,
+      "step": 20820
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4844444444444445e-06,
+      "loss": 0.7406,
+      "step": 20830
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4755555555555557e-06,
+      "loss": 0.8093,
+      "step": 20840
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4666666666666669e-06,
+      "loss": 0.8224,
+      "step": 20850
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.457777777777778e-06,
+      "loss": 0.8279,
+      "step": 20860
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4488888888888892e-06,
+      "loss": 0.7958,
+      "step": 20870
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.44e-06,
+      "loss": 0.8703,
+      "step": 20880
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4311111111111111e-06,
+      "loss": 0.8248,
+      "step": 20890
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4222222222222223e-06,
+      "loss": 0.8099,
+      "step": 20900
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4133333333333335e-06,
+      "loss": 0.8413,
+      "step": 20910
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4044444444444447e-06,
+      "loss": 0.7692,
+      "step": 20920
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3955555555555556e-06,
+      "loss": 0.7626,
+      "step": 20930
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3866666666666668e-06,
+      "loss": 0.7788,
+      "step": 20940
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.377777777777778e-06,
+      "loss": 0.7351,
+      "step": 20950
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3688888888888891e-06,
+      "loss": 0.7983,
+      "step": 20960
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 0.8482,
+      "step": 20970
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.351111111111111e-06,
+      "loss": 0.8397,
+      "step": 20980
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3422222222222222e-06,
+      "loss": 0.868,
+      "step": 20990
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 0.7908,
+      "step": 21000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3244444444444446e-06,
+      "loss": 0.8437,
+      "step": 21010
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3155555555555558e-06,
+      "loss": 0.8212,
+      "step": 21020
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3066666666666667e-06,
+      "loss": 0.8622,
+      "step": 21030
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2977777777777779e-06,
+      "loss": 0.9019,
+      "step": 21040
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.288888888888889e-06,
+      "loss": 0.7935,
+      "step": 21050
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.28e-06,
+      "loss": 0.824,
+      "step": 21060
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2711111111111112e-06,
+      "loss": 0.6767,
+      "step": 21070
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2622222222222224e-06,
+      "loss": 0.88,
+      "step": 21080
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2533333333333333e-06,
+      "loss": 0.8145,
+      "step": 21090
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2444444444444445e-06,
+      "loss": 0.8401,
+      "step": 21100
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2355555555555557e-06,
+      "loss": 0.7733,
+      "step": 21110
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2266666666666666e-06,
+      "loss": 0.7639,
+      "step": 21120
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2177777777777778e-06,
+      "loss": 0.785,
+      "step": 21130
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.208888888888889e-06,
+      "loss": 0.7948,
+      "step": 21140
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 0.8517,
+      "step": 21150
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1911111111111111e-06,
+      "loss": 0.8564,
+      "step": 21160
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1822222222222223e-06,
+      "loss": 0.8677,
+      "step": 21170
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1733333333333335e-06,
+      "loss": 0.8002,
+      "step": 21180
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1644444444444446e-06,
+      "loss": 0.7957,
+      "step": 21190
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1555555555555556e-06,
+      "loss": 0.8943,
+      "step": 21200
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1466666666666668e-06,
+      "loss": 0.7855,
+      "step": 21210
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.137777777777778e-06,
+      "loss": 0.7809,
+      "step": 21220
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.128888888888889e-06,
+      "loss": 0.9088,
+      "step": 21230
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.12e-06,
+      "loss": 0.8064,
+      "step": 21240
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.111111111111111e-06,
+      "loss": 0.8134,
+      "step": 21250
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1022222222222222e-06,
+      "loss": 0.8568,
+      "step": 21260
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0933333333333334e-06,
+      "loss": 0.761,
+      "step": 21270
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0844444444444446e-06,
+      "loss": 0.7866,
+      "step": 21280
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0755555555555557e-06,
+      "loss": 0.791,
+      "step": 21290
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.066666666666667e-06,
+      "loss": 0.8012,
+      "step": 21300
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0577777777777779e-06,
+      "loss": 0.7525,
+      "step": 21310
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.048888888888889e-06,
+      "loss": 0.7591,
+      "step": 21320
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.04e-06,
+      "loss": 0.7995,
+      "step": 21330
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0311111111111112e-06,
+      "loss": 0.7795,
+      "step": 21340
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0222222222222223e-06,
+      "loss": 0.8226,
+      "step": 21350
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0133333333333333e-06,
+      "loss": 0.7381,
+      "step": 21360
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0044444444444445e-06,
+      "loss": 0.8333,
+      "step": 21370
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.955555555555556e-07,
+      "loss": 0.7955,
+      "step": 21380
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.866666666666668e-07,
+      "loss": 0.8327,
+      "step": 21390
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.77777777777778e-07,
+      "loss": 0.8422,
+      "step": 21400
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.68888888888889e-07,
+      "loss": 0.8737,
+      "step": 21410
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 0.7847,
+      "step": 21420
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.511111111111111e-07,
+      "loss": 0.7823,
+      "step": 21430
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.422222222222223e-07,
+      "loss": 0.8462,
+      "step": 21440
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.333333333333334e-07,
+      "loss": 0.7742,
+      "step": 21450
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.244444444444445e-07,
+      "loss": 0.7643,
+      "step": 21460
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.155555555555557e-07,
+      "loss": 0.867,
+      "step": 21470
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.066666666666668e-07,
+      "loss": 0.7929,
+      "step": 21480
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.977777777777778e-07,
+      "loss": 0.8385,
+      "step": 21490
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.88888888888889e-07,
+      "loss": 0.8063,
+      "step": 21500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.8e-07,
+      "loss": 0.7531,
+      "step": 21510
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.711111111111112e-07,
+      "loss": 0.8165,
+      "step": 21520
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.622222222222224e-07,
+      "loss": 0.9315,
+      "step": 21530
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.533333333333334e-07,
+      "loss": 0.8534,
+      "step": 21540
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.444444444444445e-07,
+      "loss": 0.7863,
+      "step": 21550
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.355555555555556e-07,
+      "loss": 0.7896,
+      "step": 21560
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.266666666666668e-07,
+      "loss": 0.8015,
+      "step": 21570
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.177777777777779e-07,
+      "loss": 0.7227,
+      "step": 21580
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.088888888888889e-07,
+      "loss": 0.7587,
+      "step": 21590
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 0.8254,
+      "step": 21600
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.911111111111111e-07,
+      "loss": 0.8287,
+      "step": 21610
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.822222222222223e-07,
+      "loss": 0.8109,
+      "step": 21620
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.733333333333335e-07,
+      "loss": 0.7777,
+      "step": 21630
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.644444444444444e-07,
+      "loss": 0.8121,
+      "step": 21640
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.555555555555556e-07,
+      "loss": 0.849,
+      "step": 21650
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.466666666666668e-07,
+      "loss": 0.7384,
+      "step": 21660
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.377777777777779e-07,
+      "loss": 0.8534,
+      "step": 21670
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.28888888888889e-07,
+      "loss": 0.7182,
+      "step": 21680
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.2e-07,
+      "loss": 0.7404,
+      "step": 21690
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.111111111111112e-07,
+      "loss": 0.8923,
+      "step": 21700
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.022222222222223e-07,
+      "loss": 0.7663,
+      "step": 21710
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.933333333333334e-07,
+      "loss": 0.8139,
+      "step": 21720
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.844444444444446e-07,
+      "loss": 0.7931,
+      "step": 21730
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.755555555555555e-07,
+      "loss": 0.7545,
+      "step": 21740
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.666666666666667e-07,
+      "loss": 0.7211,
+      "step": 21750
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.577777777777779e-07,
+      "loss": 0.8763,
+      "step": 21760
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.488888888888889e-07,
+      "loss": 0.8207,
+      "step": 21770
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.4e-07,
+      "loss": 0.788,
+      "step": 21780
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.311111111111112e-07,
+      "loss": 0.8085,
+      "step": 21790
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.222222222222223e-07,
+      "loss": 0.8257,
+      "step": 21800
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.133333333333333e-07,
+      "loss": 0.7892,
+      "step": 21810
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.044444444444445e-07,
+      "loss": 0.7249,
+      "step": 21820
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.955555555555556e-07,
+      "loss": 0.8898,
+      "step": 21830
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.866666666666667e-07,
+      "loss": 0.7938,
+      "step": 21840
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.777777777777778e-07,
+      "loss": 0.773,
+      "step": 21850
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.68888888888889e-07,
+      "loss": 0.8356,
+      "step": 21860
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.6e-07,
+      "loss": 0.7823,
+      "step": 21870
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.511111111111111e-07,
+      "loss": 0.8303,
+      "step": 21880
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.422222222222223e-07,
+      "loss": 0.8017,
+      "step": 21890
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.333333333333335e-07,
+      "loss": 0.8064,
+      "step": 21900
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.244444444444445e-07,
+      "loss": 0.8344,
+      "step": 21910
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.155555555555556e-07,
+      "loss": 0.8667,
+      "step": 21920
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.066666666666667e-07,
+      "loss": 0.7859,
+      "step": 21930
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.977777777777778e-07,
+      "loss": 0.7767,
+      "step": 21940
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.88888888888889e-07,
+      "loss": 0.7409,
+      "step": 21950
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 0.7208,
+      "step": 21960
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.7111111111111113e-07,
+      "loss": 0.8506,
+      "step": 21970
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.6222222222222225e-07,
+      "loss": 0.8824,
+      "step": 21980
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.533333333333334e-07,
+      "loss": 0.8494,
+      "step": 21990
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.444444444444445e-07,
+      "loss": 0.7615,
+      "step": 22000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.355555555555556e-07,
+      "loss": 0.8258,
+      "step": 22010
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.266666666666667e-07,
+      "loss": 0.6931,
+      "step": 22020
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.177777777777778e-07,
+      "loss": 0.7577,
+      "step": 22030
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.0888888888888897e-07,
+      "loss": 0.7239,
+      "step": 22040
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 0.9302,
+      "step": 22050
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.9111111111111115e-07,
+      "loss": 0.8862,
+      "step": 22060
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.822222222222222e-07,
+      "loss": 0.7904,
+      "step": 22070
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.733333333333334e-07,
+      "loss": 0.7377,
+      "step": 22080
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.644444444444445e-07,
+      "loss": 0.8267,
+      "step": 22090
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.555555555555556e-07,
+      "loss": 0.8119,
+      "step": 22100
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.466666666666667e-07,
+      "loss": 0.8065,
+      "step": 22110
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3777777777777777e-07,
+      "loss": 0.735,
+      "step": 22120
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.2888888888888894e-07,
+      "loss": 0.7979,
+      "step": 22130
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.2e-07,
+      "loss": 0.8555,
+      "step": 22140
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.111111111111111e-07,
+      "loss": 0.7834,
+      "step": 22150
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0222222222222225e-07,
+      "loss": 0.883,
+      "step": 22160
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9333333333333337e-07,
+      "loss": 0.7805,
+      "step": 22170
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.844444444444445e-07,
+      "loss": 0.8173,
+      "step": 22180
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.7555555555555555e-07,
+      "loss": 0.822,
+      "step": 22190
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.666666666666667e-07,
+      "loss": 0.7168,
+      "step": 22200
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.577777777777778e-07,
+      "loss": 0.783,
+      "step": 22210
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.488888888888889e-07,
+      "loss": 0.7686,
+      "step": 22220
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 0.8544,
+      "step": 22230
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.3111111111111112e-07,
+      "loss": 0.8409,
+      "step": 22240
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.2222222222222224e-07,
+      "loss": 0.8705,
+      "step": 22250
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.1333333333333334e-07,
+      "loss": 0.8003,
+      "step": 22260
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0444444444444448e-07,
+      "loss": 0.7989,
+      "step": 22270
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9555555555555558e-07,
+      "loss": 0.8326,
+      "step": 22280
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.866666666666667e-07,
+      "loss": 0.7713,
+      "step": 22290
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.777777777777778e-07,
+      "loss": 0.7871,
+      "step": 22300
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.6888888888888888e-07,
+      "loss": 0.7972,
+      "step": 22310
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6e-07,
+      "loss": 0.8274,
+      "step": 22320
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5111111111111112e-07,
+      "loss": 0.8412,
+      "step": 22330
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4222222222222224e-07,
+      "loss": 0.7927,
+      "step": 22340
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3333333333333336e-07,
+      "loss": 0.8578,
+      "step": 22350
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2444444444444446e-07,
+      "loss": 0.7757,
+      "step": 22360
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1555555555555556e-07,
+      "loss": 0.8511,
+      "step": 22370
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0666666666666667e-07,
+      "loss": 0.7541,
+      "step": 22380
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.777777777777779e-08,
+      "loss": 0.815,
+      "step": 22390
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.88888888888889e-08,
+      "loss": 0.7566,
+      "step": 22400
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8e-08,
+      "loss": 0.8102,
+      "step": 22410
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.111111111111112e-08,
+      "loss": 0.7659,
+      "step": 22420
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.222222222222223e-08,
+      "loss": 0.8285,
+      "step": 22430
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.3333333333333334e-08,
+      "loss": 0.8284,
+      "step": 22440
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.444444444444445e-08,
+      "loss": 0.8904,
+      "step": 22450
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.555555555555556e-08,
+      "loss": 0.8546,
+      "step": 22460
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.6666666666666667e-08,
+      "loss": 0.7728,
+      "step": 22470
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.777777777777778e-08,
+      "loss": 0.7704,
+      "step": 22480
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.88888888888889e-09,
+      "loss": 0.658,
+      "step": 22490
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.8189,
+      "step": 22500
+    },
+    {
+      "epoch": 3.0,
+      "step": 22500,
+      "total_flos": 1.2708180506922394e+18,
+      "train_loss": 1.3105251697540283,
+      "train_runtime": 46807.5952,
+      "train_samples_per_second": 7.691,
+      "train_steps_per_second": 0.481
+    }
+  ],
+  "max_steps": 22500,
+  "num_train_epochs": 3,
+  "total_flos": 1.2708180506922394e+18,
+  "trial_name": null,
+  "trial_params": null
+}