diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8536 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.8933333333333333,
+  "global_step": 14200,
+  "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": 2.4048,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9982222222222224e-05,
+      "loss": 1.8775,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9973333333333334e-05,
+      "loss": 1.8612,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9964444444444447e-05,
+      "loss": 1.846,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9955555555555557e-05,
+      "loss": 1.7451,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9946666666666667e-05,
+      "loss": 1.7533,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.993777777777778e-05,
+      "loss": 1.7204,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.992888888888889e-05,
+      "loss": 1.6441,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9920000000000002e-05,
+      "loss": 1.6775,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9911111111111112e-05,
+      "loss": 1.6084,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9902222222222222e-05,
+      "loss": 1.6458,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9893333333333335e-05,
+      "loss": 1.7025,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9884444444444445e-05,
+      "loss": 1.6242,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9875555555555558e-05,
+      "loss": 1.6565,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9866666666666667e-05,
+      "loss": 1.5933,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.985777777777778e-05,
+      "loss": 1.6074,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.984888888888889e-05,
+      "loss": 1.4749,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9840000000000003e-05,
+      "loss": 1.6277,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9831111111111113e-05,
+      "loss": 1.584,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9822222222222226e-05,
+      "loss": 1.6242,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9813333333333336e-05,
+      "loss": 1.5364,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9804444444444445e-05,
+      "loss": 1.558,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.979555555555556e-05,
+      "loss": 1.4482,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9786666666666668e-05,
+      "loss": 1.4228,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.977777777777778e-05,
+      "loss": 1.4842,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.976888888888889e-05,
+      "loss": 1.4997,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.976e-05,
+      "loss": 1.5658,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9751111111111114e-05,
+      "loss": 1.5769,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9742222222222223e-05,
+      "loss": 1.4129,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9733333333333336e-05,
+      "loss": 1.4415,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9724444444444446e-05,
+      "loss": 1.5418,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9715555555555556e-05,
+      "loss": 1.5264,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.970666666666667e-05,
+      "loss": 1.5169,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.969777777777778e-05,
+      "loss": 1.4987,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.968888888888889e-05,
+      "loss": 1.4791,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.968e-05,
+      "loss": 1.4609,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.967111111111111e-05,
+      "loss": 1.5073,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9662222222222224e-05,
+      "loss": 1.3887,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9653333333333334e-05,
+      "loss": 1.4547,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9644444444444447e-05,
+      "loss": 1.3708,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9635555555555557e-05,
+      "loss": 1.372,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9626666666666666e-05,
+      "loss": 1.3724,
+      "step": 420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.961777777777778e-05,
+      "loss": 1.4475,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.960888888888889e-05,
+      "loss": 1.4078,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 1.3645,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9591111111111112e-05,
+      "loss": 1.3766,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9582222222222225e-05,
+      "loss": 1.547,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9573333333333335e-05,
+      "loss": 1.4228,
+      "step": 480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9564444444444444e-05,
+      "loss": 1.3612,
+      "step": 490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9555555555555557e-05,
+      "loss": 1.4907,
+      "step": 500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.954666666666667e-05,
+      "loss": 1.4219,
+      "step": 510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.953777777777778e-05,
+      "loss": 1.428,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.952888888888889e-05,
+      "loss": 1.4665,
+      "step": 530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9520000000000003e-05,
+      "loss": 1.4952,
+      "step": 540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9511111111111113e-05,
+      "loss": 1.4559,
+      "step": 550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9502222222222226e-05,
+      "loss": 1.3358,
+      "step": 560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9493333333333335e-05,
+      "loss": 1.3978,
+      "step": 570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9484444444444445e-05,
+      "loss": 1.4329,
+      "step": 580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9475555555555558e-05,
+      "loss": 1.4256,
+      "step": 590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9466666666666668e-05,
+      "loss": 1.4073,
+      "step": 600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.945777777777778e-05,
+      "loss": 1.3364,
+      "step": 610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.944888888888889e-05,
+      "loss": 1.3648,
+      "step": 620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.944e-05,
+      "loss": 1.3115,
+      "step": 630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9431111111111113e-05,
+      "loss": 1.3724,
+      "step": 640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9422222222222223e-05,
+      "loss": 1.3438,
+      "step": 650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9413333333333336e-05,
+      "loss": 1.4041,
+      "step": 660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9404444444444446e-05,
+      "loss": 1.3519,
+      "step": 670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9395555555555555e-05,
+      "loss": 1.3564,
+      "step": 680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.938666666666667e-05,
+      "loss": 1.3786,
+      "step": 690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9377777777777778e-05,
+      "loss": 1.3111,
+      "step": 700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.936888888888889e-05,
+      "loss": 1.3665,
+      "step": 710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.936e-05,
+      "loss": 1.3987,
+      "step": 720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.935111111111111e-05,
+      "loss": 1.3155,
+      "step": 730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9342222222222224e-05,
+      "loss": 1.3128,
+      "step": 740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9333333333333333e-05,
+      "loss": 1.3127,
+      "step": 750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9324444444444447e-05,
+      "loss": 1.2801,
+      "step": 760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9315555555555556e-05,
+      "loss": 1.3582,
+      "step": 770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.930666666666667e-05,
+      "loss": 1.3537,
+      "step": 780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.929777777777778e-05,
+      "loss": 1.2926,
+      "step": 790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.928888888888889e-05,
+      "loss": 1.2753,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9280000000000002e-05,
+      "loss": 1.3155,
+      "step": 810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9271111111111115e-05,
+      "loss": 1.3894,
+      "step": 820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9262222222222225e-05,
+      "loss": 1.2848,
+      "step": 830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9253333333333334e-05,
+      "loss": 1.3126,
+      "step": 840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9244444444444444e-05,
+      "loss": 1.33,
+      "step": 850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9235555555555557e-05,
+      "loss": 1.2493,
+      "step": 860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.922666666666667e-05,
+      "loss": 1.2834,
+      "step": 870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.921777777777778e-05,
+      "loss": 1.3542,
+      "step": 880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.920888888888889e-05,
+      "loss": 1.2733,
+      "step": 890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 1.3271,
+      "step": 900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9191111111111112e-05,
+      "loss": 1.4075,
+      "step": 910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9182222222222225e-05,
+      "loss": 1.2204,
+      "step": 920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9173333333333335e-05,
+      "loss": 1.3184,
+      "step": 930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9164444444444445e-05,
+      "loss": 1.3577,
+      "step": 940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9155555555555558e-05,
+      "loss": 1.2539,
+      "step": 950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9146666666666667e-05,
+      "loss": 1.3455,
+      "step": 960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.913777777777778e-05,
+      "loss": 1.3104,
+      "step": 970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.912888888888889e-05,
+      "loss": 1.3331,
+      "step": 980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.912e-05,
+      "loss": 1.2342,
+      "step": 990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9111111111111113e-05,
+      "loss": 1.3409,
+      "step": 1000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9102222222222223e-05,
+      "loss": 1.2472,
+      "step": 1010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9093333333333336e-05,
+      "loss": 1.1771,
+      "step": 1020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9084444444444445e-05,
+      "loss": 1.3049,
+      "step": 1030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9075555555555555e-05,
+      "loss": 1.2137,
+      "step": 1040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9066666666666668e-05,
+      "loss": 1.4173,
+      "step": 1050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9057777777777778e-05,
+      "loss": 1.2551,
+      "step": 1060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.904888888888889e-05,
+      "loss": 1.2043,
+      "step": 1070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.904e-05,
+      "loss": 1.2418,
+      "step": 1080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9031111111111114e-05,
+      "loss": 1.2922,
+      "step": 1090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9022222222222223e-05,
+      "loss": 1.2136,
+      "step": 1100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9013333333333333e-05,
+      "loss": 1.2552,
+      "step": 1110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9004444444444446e-05,
+      "loss": 1.2499,
+      "step": 1120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.899555555555556e-05,
+      "loss": 1.2998,
+      "step": 1130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.898666666666667e-05,
+      "loss": 1.2775,
+      "step": 1140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.897777777777778e-05,
+      "loss": 1.3368,
+      "step": 1150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.896888888888889e-05,
+      "loss": 1.2595,
+      "step": 1160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.896e-05,
+      "loss": 1.3476,
+      "step": 1170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8951111111111115e-05,
+      "loss": 1.2566,
+      "step": 1180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8942222222222224e-05,
+      "loss": 1.2813,
+      "step": 1190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8933333333333334e-05,
+      "loss": 1.2985,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8924444444444447e-05,
+      "loss": 1.2264,
+      "step": 1210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8915555555555557e-05,
+      "loss": 1.2706,
+      "step": 1220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.890666666666667e-05,
+      "loss": 1.1898,
+      "step": 1230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.889777777777778e-05,
+      "loss": 1.2292,
+      "step": 1240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.888888888888889e-05,
+      "loss": 1.2652,
+      "step": 1250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8880000000000002e-05,
+      "loss": 1.2528,
+      "step": 1260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8871111111111112e-05,
+      "loss": 1.271,
+      "step": 1270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8862222222222225e-05,
+      "loss": 1.1702,
+      "step": 1280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8853333333333335e-05,
+      "loss": 1.1904,
+      "step": 1290
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8844444444444444e-05,
+      "loss": 1.3074,
+      "step": 1300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8835555555555557e-05,
+      "loss": 1.2295,
+      "step": 1310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8826666666666667e-05,
+      "loss": 1.1498,
+      "step": 1320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881777777777778e-05,
+      "loss": 1.3108,
+      "step": 1330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880888888888889e-05,
+      "loss": 1.2615,
+      "step": 1340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88e-05,
+      "loss": 1.1936,
+      "step": 1350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8791111111111113e-05,
+      "loss": 1.2588,
+      "step": 1360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8782222222222222e-05,
+      "loss": 1.2175,
+      "step": 1370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8773333333333335e-05,
+      "loss": 1.3057,
+      "step": 1380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8764444444444445e-05,
+      "loss": 1.3412,
+      "step": 1390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8755555555555558e-05,
+      "loss": 1.148,
+      "step": 1400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8746666666666668e-05,
+      "loss": 1.2396,
+      "step": 1410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8737777777777778e-05,
+      "loss": 1.2624,
+      "step": 1420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872888888888889e-05,
+      "loss": 1.1955,
+      "step": 1430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8720000000000004e-05,
+      "loss": 1.1798,
+      "step": 1440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8711111111111113e-05,
+      "loss": 1.2366,
+      "step": 1450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8702222222222223e-05,
+      "loss": 1.1596,
+      "step": 1460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8693333333333333e-05,
+      "loss": 1.2025,
+      "step": 1470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8684444444444446e-05,
+      "loss": 1.1361,
+      "step": 1480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.867555555555556e-05,
+      "loss": 1.221,
+      "step": 1490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 1.1684,
+      "step": 1500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.865777777777778e-05,
+      "loss": 1.2633,
+      "step": 1510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8648888888888888e-05,
+      "loss": 1.2157,
+      "step": 1520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.864e-05,
+      "loss": 1.2302,
+      "step": 1530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8631111111111114e-05,
+      "loss": 1.1223,
+      "step": 1540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8622222222222224e-05,
+      "loss": 1.131,
+      "step": 1550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8613333333333334e-05,
+      "loss": 1.1905,
+      "step": 1560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8604444444444447e-05,
+      "loss": 1.2618,
+      "step": 1570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8595555555555556e-05,
+      "loss": 1.1637,
+      "step": 1580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.858666666666667e-05,
+      "loss": 1.2449,
+      "step": 1590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.857777777777778e-05,
+      "loss": 1.246,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.856888888888889e-05,
+      "loss": 1.2406,
+      "step": 1610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8560000000000002e-05,
+      "loss": 1.2184,
+      "step": 1620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.855111111111111e-05,
+      "loss": 1.1665,
+      "step": 1630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8542222222222225e-05,
+      "loss": 1.1181,
+      "step": 1640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8533333333333334e-05,
+      "loss": 1.1796,
+      "step": 1650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8524444444444444e-05,
+      "loss": 1.121,
+      "step": 1660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8515555555555557e-05,
+      "loss": 1.1749,
+      "step": 1670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8506666666666667e-05,
+      "loss": 1.1692,
+      "step": 1680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.849777777777778e-05,
+      "loss": 1.136,
+      "step": 1690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.848888888888889e-05,
+      "loss": 1.1391,
+      "step": 1700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8480000000000003e-05,
+      "loss": 1.2016,
+      "step": 1710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8471111111111112e-05,
+      "loss": 1.1369,
+      "step": 1720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8462222222222222e-05,
+      "loss": 1.1447,
+      "step": 1730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8453333333333335e-05,
+      "loss": 1.1595,
+      "step": 1740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8444444444444448e-05,
+      "loss": 1.1653,
+      "step": 1750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8435555555555558e-05,
+      "loss": 1.2229,
+      "step": 1760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8426666666666668e-05,
+      "loss": 1.1258,
+      "step": 1770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8417777777777777e-05,
+      "loss": 1.1931,
+      "step": 1780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.840888888888889e-05,
+      "loss": 1.1667,
+      "step": 1790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 1.101,
+      "step": 1800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8391111111111113e-05,
+      "loss": 1.1117,
+      "step": 1810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8382222222222223e-05,
+      "loss": 1.1763,
+      "step": 1820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8373333333333332e-05,
+      "loss": 1.1924,
+      "step": 1830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8364444444444446e-05,
+      "loss": 1.1929,
+      "step": 1840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.835555555555556e-05,
+      "loss": 1.1407,
+      "step": 1850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.834666666666667e-05,
+      "loss": 1.1641,
+      "step": 1860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8337777777777778e-05,
+      "loss": 1.2749,
+      "step": 1870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.832888888888889e-05,
+      "loss": 1.2064,
+      "step": 1880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.832e-05,
+      "loss": 1.1516,
+      "step": 1890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8311111111111114e-05,
+      "loss": 1.0882,
+      "step": 1900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8302222222222224e-05,
+      "loss": 1.2308,
+      "step": 1910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8293333333333333e-05,
+      "loss": 1.1474,
+      "step": 1920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8284444444444446e-05,
+      "loss": 1.2042,
+      "step": 1930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8275555555555556e-05,
+      "loss": 1.2158,
+      "step": 1940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.826666666666667e-05,
+      "loss": 1.2252,
+      "step": 1950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.825777777777778e-05,
+      "loss": 1.2847,
+      "step": 1960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.824888888888889e-05,
+      "loss": 1.1319,
+      "step": 1970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.824e-05,
+      "loss": 1.1339,
+      "step": 1980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.823111111111111e-05,
+      "loss": 1.1259,
+      "step": 1990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8222222222222224e-05,
+      "loss": 1.1442,
+      "step": 2000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8213333333333334e-05,
+      "loss": 1.1424,
+      "step": 2010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8204444444444447e-05,
+      "loss": 1.1209,
+      "step": 2020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8195555555555557e-05,
+      "loss": 1.1215,
+      "step": 2030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8186666666666666e-05,
+      "loss": 1.1604,
+      "step": 2040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.817777777777778e-05,
+      "loss": 1.1118,
+      "step": 2050
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8168888888888893e-05,
+      "loss": 1.0966,
+      "step": 2060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8160000000000002e-05,
+      "loss": 1.1598,
+      "step": 2070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8151111111111112e-05,
+      "loss": 1.1438,
+      "step": 2080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.814222222222222e-05,
+      "loss": 1.1719,
+      "step": 2090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8133333333333335e-05,
+      "loss": 1.1146,
+      "step": 2100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8124444444444448e-05,
+      "loss": 1.1549,
+      "step": 2110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8115555555555558e-05,
+      "loss": 1.1294,
+      "step": 2120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8106666666666667e-05,
+      "loss": 1.0684,
+      "step": 2130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8097777777777777e-05,
+      "loss": 1.1102,
+      "step": 2140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.808888888888889e-05,
+      "loss": 1.1921,
+      "step": 2150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8080000000000003e-05,
+      "loss": 1.0957,
+      "step": 2160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8071111111111113e-05,
+      "loss": 1.2035,
+      "step": 2170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8062222222222222e-05,
+      "loss": 1.1848,
+      "step": 2180
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8053333333333332e-05,
+      "loss": 1.1267,
+      "step": 2190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8044444444444445e-05,
+      "loss": 1.0981,
+      "step": 2200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.803555555555556e-05,
+      "loss": 1.1,
+      "step": 2210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8026666666666668e-05,
+      "loss": 1.0529,
+      "step": 2220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8017777777777778e-05,
+      "loss": 1.0171,
+      "step": 2230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.800888888888889e-05,
+      "loss": 1.1594,
+      "step": 2240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8e-05,
+      "loss": 1.0629,
+      "step": 2250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7991111111111114e-05,
+      "loss": 1.1984,
+      "step": 2260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7982222222222223e-05,
+      "loss": 1.1947,
+      "step": 2270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7973333333333333e-05,
+      "loss": 1.156,
+      "step": 2280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7964444444444446e-05,
+      "loss": 1.1862,
+      "step": 2290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7955555555555556e-05,
+      "loss": 1.0595,
+      "step": 2300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.794666666666667e-05,
+      "loss": 1.0802,
+      "step": 2310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.793777777777778e-05,
+      "loss": 1.0912,
+      "step": 2320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.792888888888889e-05,
+      "loss": 1.0775,
+      "step": 2330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.792e-05,
+      "loss": 1.048,
+      "step": 2340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.791111111111111e-05,
+      "loss": 1.1577,
+      "step": 2350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7902222222222224e-05,
+      "loss": 1.1265,
+      "step": 2360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7893333333333337e-05,
+      "loss": 1.1948,
+      "step": 2370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7884444444444447e-05,
+      "loss": 1.031,
+      "step": 2380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7875555555555556e-05,
+      "loss": 1.1495,
+      "step": 2390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7866666666666666e-05,
+      "loss": 1.0832,
+      "step": 2400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.785777777777778e-05,
+      "loss": 1.1345,
+      "step": 2410
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7848888888888892e-05,
+      "loss": 1.0734,
+      "step": 2420
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7840000000000002e-05,
+      "loss": 1.1869,
+      "step": 2430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.783111111111111e-05,
+      "loss": 1.1339,
+      "step": 2440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.782222222222222e-05,
+      "loss": 1.1147,
+      "step": 2450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7813333333333334e-05,
+      "loss": 1.1347,
+      "step": 2460
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7804444444444448e-05,
+      "loss": 1.116,
+      "step": 2470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7795555555555557e-05,
+      "loss": 1.0951,
+      "step": 2480
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7786666666666667e-05,
+      "loss": 1.0371,
+      "step": 2490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 1.1412,
+      "step": 2500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.776888888888889e-05,
+      "loss": 1.0725,
+      "step": 2510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7760000000000003e-05,
+      "loss": 1.106,
+      "step": 2520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7751111111111112e-05,
+      "loss": 1.1332,
+      "step": 2530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7742222222222222e-05,
+      "loss": 1.1655,
+      "step": 2540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7733333333333335e-05,
+      "loss": 1.0724,
+      "step": 2550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7724444444444445e-05,
+      "loss": 1.1051,
+      "step": 2560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7715555555555558e-05,
+      "loss": 1.0878,
+      "step": 2570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7706666666666668e-05,
+      "loss": 1.1064,
+      "step": 2580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7697777777777777e-05,
+      "loss": 1.0038,
+      "step": 2590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.768888888888889e-05,
+      "loss": 1.1117,
+      "step": 2600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.768e-05,
+      "loss": 1.0629,
+      "step": 2610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7671111111111113e-05,
+      "loss": 1.1432,
+      "step": 2620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7662222222222223e-05,
+      "loss": 1.0908,
+      "step": 2630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7653333333333336e-05,
+      "loss": 1.1247,
+      "step": 2640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7644444444444446e-05,
+      "loss": 1.0317,
+      "step": 2650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7635555555555555e-05,
+      "loss": 1.0392,
+      "step": 2660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.762666666666667e-05,
+      "loss": 1.0568,
+      "step": 2670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.761777777777778e-05,
+      "loss": 1.1636,
+      "step": 2680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.760888888888889e-05,
+      "loss": 1.1749,
+      "step": 2690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.76e-05,
+      "loss": 1.0609,
+      "step": 2700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.759111111111111e-05,
+      "loss": 1.0416,
+      "step": 2710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.7582222222222224e-05,
+      "loss": 1.0575,
+      "step": 2720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.7573333333333337e-05,
+      "loss": 1.1435,
+      "step": 2730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.7564444444444446e-05,
+      "loss": 1.1143,
+      "step": 2740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.7555555555555556e-05,
+      "loss": 1.0495,
+      "step": 2750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.7546666666666666e-05,
+      "loss": 1.1068,
+      "step": 2760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.753777777777778e-05,
+      "loss": 1.0669,
+      "step": 2770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.7528888888888892e-05,
+      "loss": 1.1336,
+      "step": 2780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.752e-05,
+      "loss": 1.0577,
+      "step": 2790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.751111111111111e-05,
+      "loss": 1.1104,
+      "step": 2800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.750222222222222e-05,
+      "loss": 1.0981,
+      "step": 2810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7493333333333334e-05,
+      "loss": 1.1212,
+      "step": 2820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7484444444444447e-05,
+      "loss": 1.1093,
+      "step": 2830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7475555555555557e-05,
+      "loss": 1.0886,
+      "step": 2840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7466666666666667e-05,
+      "loss": 1.0578,
+      "step": 2850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.745777777777778e-05,
+      "loss": 1.0253,
+      "step": 2860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.744888888888889e-05,
+      "loss": 1.2101,
+      "step": 2870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7440000000000002e-05,
+      "loss": 1.1014,
+      "step": 2880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7431111111111112e-05,
+      "loss": 1.0862,
+      "step": 2890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7422222222222222e-05,
+      "loss": 1.1147,
+      "step": 2900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7413333333333335e-05,
+      "loss": 1.1526,
+      "step": 2910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7404444444444445e-05,
+      "loss": 0.9954,
+      "step": 2920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7395555555555558e-05,
+      "loss": 1.0464,
+      "step": 2930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7386666666666667e-05,
+      "loss": 1.0722,
+      "step": 2940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.737777777777778e-05,
+      "loss": 1.1072,
+      "step": 2950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.736888888888889e-05,
+      "loss": 1.1118,
+      "step": 2960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.736e-05,
+      "loss": 1.115,
+      "step": 2970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7351111111111113e-05,
+      "loss": 1.0415,
+      "step": 2980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7342222222222226e-05,
+      "loss": 1.0142,
+      "step": 2990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 1.1325,
+      "step": 3000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7324444444444445e-05,
+      "loss": 1.0434,
+      "step": 3010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7315555555555555e-05,
+      "loss": 0.9897,
+      "step": 3020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7306666666666668e-05,
+      "loss": 1.0997,
+      "step": 3030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.729777777777778e-05,
+      "loss": 1.0362,
+      "step": 3040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.728888888888889e-05,
+      "loss": 1.0217,
+      "step": 3050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.728e-05,
+      "loss": 1.0983,
+      "step": 3060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.727111111111111e-05,
+      "loss": 1.0657,
+      "step": 3070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7262222222222223e-05,
+      "loss": 1.0473,
+      "step": 3080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7253333333333336e-05,
+      "loss": 1.0808,
+      "step": 3090
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7244444444444446e-05,
+      "loss": 1.077,
+      "step": 3100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7235555555555556e-05,
+      "loss": 1.1284,
+      "step": 3110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7226666666666665e-05,
+      "loss": 1.0361,
+      "step": 3120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.721777777777778e-05,
+      "loss": 1.071,
+      "step": 3130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.720888888888889e-05,
+      "loss": 1.0644,
+      "step": 3140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.72e-05,
+      "loss": 1.0926,
+      "step": 3150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.719111111111111e-05,
+      "loss": 1.1191,
+      "step": 3160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7182222222222224e-05,
+      "loss": 1.0015,
+      "step": 3170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7173333333333334e-05,
+      "loss": 1.019,
+      "step": 3180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7164444444444447e-05,
+      "loss": 1.0915,
+      "step": 3190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7155555555555557e-05,
+      "loss": 1.0894,
+      "step": 3200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7146666666666666e-05,
+      "loss": 1.0791,
+      "step": 3210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.713777777777778e-05,
+      "loss": 1.0386,
+      "step": 3220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.712888888888889e-05,
+      "loss": 1.0754,
+      "step": 3230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7120000000000002e-05,
+      "loss": 1.0645,
+      "step": 3240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7111111111111112e-05,
+      "loss": 0.9963,
+      "step": 3250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7102222222222225e-05,
+      "loss": 1.021,
+      "step": 3260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7093333333333335e-05,
+      "loss": 0.9769,
+      "step": 3270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7084444444444444e-05,
+      "loss": 1.0322,
+      "step": 3280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7075555555555557e-05,
+      "loss": 1.1114,
+      "step": 3290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.706666666666667e-05,
+      "loss": 1.0605,
+      "step": 3300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.705777777777778e-05,
+      "loss": 1.0789,
+      "step": 3310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.704888888888889e-05,
+      "loss": 1.0943,
+      "step": 3320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.704e-05,
+      "loss": 1.0243,
+      "step": 3330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7031111111111113e-05,
+      "loss": 1.0203,
+      "step": 3340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7022222222222226e-05,
+      "loss": 1.0241,
+      "step": 3350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7013333333333335e-05,
+      "loss": 1.0587,
+      "step": 3360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7004444444444445e-05,
+      "loss": 0.9933,
+      "step": 3370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6995555555555555e-05,
+      "loss": 1.0356,
+      "step": 3380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6986666666666668e-05,
+      "loss": 1.0042,
+      "step": 3390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.697777777777778e-05,
+      "loss": 1.0549,
+      "step": 3400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.696888888888889e-05,
+      "loss": 1.0179,
+      "step": 3410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.696e-05,
+      "loss": 1.0327,
+      "step": 3420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.695111111111111e-05,
+      "loss": 1.039,
+      "step": 3430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6942222222222223e-05,
+      "loss": 0.9851,
+      "step": 3440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6933333333333336e-05,
+      "loss": 1.0751,
+      "step": 3450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6924444444444446e-05,
+      "loss": 1.0844,
+      "step": 3460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6915555555555555e-05,
+      "loss": 1.0434,
+      "step": 3470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.690666666666667e-05,
+      "loss": 1.0278,
+      "step": 3480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6897777777777778e-05,
+      "loss": 0.9855,
+      "step": 3490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.688888888888889e-05,
+      "loss": 1.0691,
+      "step": 3500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.688e-05,
+      "loss": 1.0254,
+      "step": 3510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.687111111111111e-05,
+      "loss": 1.0265,
+      "step": 3520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6862222222222224e-05,
+      "loss": 1.0081,
+      "step": 3530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6853333333333333e-05,
+      "loss": 0.9902,
+      "step": 3540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6844444444444447e-05,
+      "loss": 1.0496,
+      "step": 3550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6835555555555556e-05,
+      "loss": 1.092,
+      "step": 3560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.682666666666667e-05,
+      "loss": 1.0257,
+      "step": 3570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.681777777777778e-05,
+      "loss": 0.9694,
+      "step": 3580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.680888888888889e-05,
+      "loss": 1.0551,
+      "step": 3590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 1.0301,
+      "step": 3600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6791111111111115e-05,
+      "loss": 1.0309,
+      "step": 3610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6782222222222225e-05,
+      "loss": 0.9837,
+      "step": 3620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6773333333333334e-05,
+      "loss": 1.0675,
+      "step": 3630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6764444444444444e-05,
+      "loss": 1.0518,
+      "step": 3640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6755555555555557e-05,
+      "loss": 0.9905,
+      "step": 3650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.674666666666667e-05,
+      "loss": 0.972,
+      "step": 3660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.673777777777778e-05,
+      "loss": 0.9871,
+      "step": 3670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.672888888888889e-05,
+      "loss": 1.0209,
+      "step": 3680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.672e-05,
+      "loss": 1.0496,
+      "step": 3690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6711111111111112e-05,
+      "loss": 1.0903,
+      "step": 3700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6702222222222225e-05,
+      "loss": 1.0351,
+      "step": 3710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6693333333333335e-05,
+      "loss": 1.0321,
+      "step": 3720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6684444444444445e-05,
+      "loss": 1.0223,
+      "step": 3730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6675555555555554e-05,
+      "loss": 1.0591,
+      "step": 3740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 1.005,
+      "step": 3750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.665777777777778e-05,
+      "loss": 1.0344,
+      "step": 3760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.664888888888889e-05,
+      "loss": 1.0726,
+      "step": 3770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.664e-05,
+      "loss": 1.0353,
+      "step": 3780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6631111111111113e-05,
+      "loss": 1.0594,
+      "step": 3790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6622222222222223e-05,
+      "loss": 1.0401,
+      "step": 3800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6613333333333336e-05,
+      "loss": 0.987,
+      "step": 3810
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6604444444444445e-05,
+      "loss": 1.0044,
+      "step": 3820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6595555555555555e-05,
+      "loss": 0.9659,
+      "step": 3830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6586666666666668e-05,
+      "loss": 0.984,
+      "step": 3840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6577777777777778e-05,
+      "loss": 0.9561,
+      "step": 3850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.656888888888889e-05,
+      "loss": 1.0197,
+      "step": 3860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.656e-05,
+      "loss": 1.0082,
+      "step": 3870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6551111111111114e-05,
+      "loss": 0.9926,
+      "step": 3880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6542222222222223e-05,
+      "loss": 1.1347,
+      "step": 3890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6533333333333333e-05,
+      "loss": 0.9225,
+      "step": 3900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6524444444444446e-05,
+      "loss": 1.0511,
+      "step": 3910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.651555555555556e-05,
+      "loss": 0.9633,
+      "step": 3920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.650666666666667e-05,
+      "loss": 1.0121,
+      "step": 3930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.649777777777778e-05,
+      "loss": 1.0054,
+      "step": 3940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.648888888888889e-05,
+      "loss": 0.9862,
+      "step": 3950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.648e-05,
+      "loss": 1.0617,
+      "step": 3960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6471111111111115e-05,
+      "loss": 1.0535,
+      "step": 3970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6462222222222224e-05,
+      "loss": 0.994,
+      "step": 3980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6453333333333334e-05,
+      "loss": 1.0628,
+      "step": 3990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6444444444444444e-05,
+      "loss": 0.9894,
+      "step": 4000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6435555555555557e-05,
+      "loss": 1.0635,
+      "step": 4010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.642666666666667e-05,
+      "loss": 1.0418,
+      "step": 4020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.641777777777778e-05,
+      "loss": 0.9938,
+      "step": 4030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.640888888888889e-05,
+      "loss": 0.9737,
+      "step": 4040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.64e-05,
+      "loss": 0.9671,
+      "step": 4050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6391111111111112e-05,
+      "loss": 0.9937,
+      "step": 4060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6382222222222225e-05,
+      "loss": 0.9477,
+      "step": 4070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6373333333333335e-05,
+      "loss": 1.1009,
+      "step": 4080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6364444444444444e-05,
+      "loss": 0.9271,
+      "step": 4090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6355555555555557e-05,
+      "loss": 0.9725,
+      "step": 4100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6346666666666667e-05,
+      "loss": 1.018,
+      "step": 4110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.633777777777778e-05,
+      "loss": 1.0247,
+      "step": 4120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.632888888888889e-05,
+      "loss": 0.9936,
+      "step": 4130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.632e-05,
+      "loss": 0.9915,
+      "step": 4140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6311111111111113e-05,
+      "loss": 0.9977,
+      "step": 4150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6302222222222222e-05,
+      "loss": 0.9974,
+      "step": 4160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6293333333333335e-05,
+      "loss": 0.9561,
+      "step": 4170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6284444444444445e-05,
+      "loss": 0.9646,
+      "step": 4180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6275555555555558e-05,
+      "loss": 1.0278,
+      "step": 4190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6266666666666668e-05,
+      "loss": 1.051,
+      "step": 4200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6257777777777778e-05,
+      "loss": 0.9725,
+      "step": 4210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.624888888888889e-05,
+      "loss": 0.9907,
+      "step": 4220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6240000000000004e-05,
+      "loss": 1.0095,
+      "step": 4230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6231111111111113e-05,
+      "loss": 1.0398,
+      "step": 4240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6222222222222223e-05,
+      "loss": 0.9242,
+      "step": 4250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6213333333333333e-05,
+      "loss": 0.971,
+      "step": 4260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6204444444444446e-05,
+      "loss": 0.9973,
+      "step": 4270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.619555555555556e-05,
+      "loss": 0.952,
+      "step": 4280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.618666666666667e-05,
+      "loss": 0.9942,
+      "step": 4290
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.617777777777778e-05,
+      "loss": 0.9642,
+      "step": 4300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6168888888888888e-05,
+      "loss": 0.9615,
+      "step": 4310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.616e-05,
+      "loss": 0.9425,
+      "step": 4320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6151111111111114e-05,
+      "loss": 0.9787,
+      "step": 4330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6142222222222224e-05,
+      "loss": 0.9459,
+      "step": 4340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6133333333333334e-05,
+      "loss": 0.9945,
+      "step": 4350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6124444444444443e-05,
+      "loss": 1.0766,
+      "step": 4360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6115555555555556e-05,
+      "loss": 0.9697,
+      "step": 4370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.610666666666667e-05,
+      "loss": 0.9975,
+      "step": 4380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.609777777777778e-05,
+      "loss": 1.0221,
+      "step": 4390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.608888888888889e-05,
+      "loss": 0.9796,
+      "step": 4400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6080000000000002e-05,
+      "loss": 0.9725,
+      "step": 4410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.607111111111111e-05,
+      "loss": 0.9555,
+      "step": 4420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6062222222222225e-05,
+      "loss": 1.0233,
+      "step": 4430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6053333333333334e-05,
+      "loss": 0.9116,
+      "step": 4440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6044444444444444e-05,
+      "loss": 0.9576,
+      "step": 4450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6035555555555557e-05,
+      "loss": 1.0401,
+      "step": 4460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6026666666666667e-05,
+      "loss": 0.9826,
+      "step": 4470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.601777777777778e-05,
+      "loss": 0.9551,
+      "step": 4480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.600888888888889e-05,
+      "loss": 0.9195,
+      "step": 4490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.936,
+      "step": 4500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5991111111111112e-05,
+      "loss": 0.9904,
+      "step": 4510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5982222222222222e-05,
+      "loss": 1.0146,
+      "step": 4520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5973333333333335e-05,
+      "loss": 0.9256,
+      "step": 4530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5964444444444448e-05,
+      "loss": 0.9898,
+      "step": 4540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5955555555555558e-05,
+      "loss": 1.0003,
+      "step": 4550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5946666666666668e-05,
+      "loss": 0.9904,
+      "step": 4560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5937777777777777e-05,
+      "loss": 1.0455,
+      "step": 4570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.592888888888889e-05,
+      "loss": 1.0103,
+      "step": 4580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5920000000000003e-05,
+      "loss": 0.9051,
+      "step": 4590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5911111111111113e-05,
+      "loss": 0.9406,
+      "step": 4600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5902222222222223e-05,
+      "loss": 0.9872,
+      "step": 4610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5893333333333333e-05,
+      "loss": 0.9281,
+      "step": 4620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5884444444444446e-05,
+      "loss": 0.9992,
+      "step": 4630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.587555555555556e-05,
+      "loss": 0.9596,
+      "step": 4640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.586666666666667e-05,
+      "loss": 0.9157,
+      "step": 4650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5857777777777778e-05,
+      "loss": 0.9478,
+      "step": 4660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5848888888888888e-05,
+      "loss": 0.903,
+      "step": 4670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.584e-05,
+      "loss": 0.9383,
+      "step": 4680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5831111111111114e-05,
+      "loss": 0.9767,
+      "step": 4690
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5822222222222224e-05,
+      "loss": 0.9972,
+      "step": 4700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5813333333333333e-05,
+      "loss": 0.9159,
+      "step": 4710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5804444444444446e-05,
+      "loss": 0.9773,
+      "step": 4720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5795555555555556e-05,
+      "loss": 0.907,
+      "step": 4730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.578666666666667e-05,
+      "loss": 0.9322,
+      "step": 4740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.577777777777778e-05,
+      "loss": 0.9645,
+      "step": 4750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.576888888888889e-05,
+      "loss": 0.9977,
+      "step": 4760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.576e-05,
+      "loss": 0.9669,
+      "step": 4770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.575111111111111e-05,
+      "loss": 0.9661,
+      "step": 4780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5742222222222224e-05,
+      "loss": 1.0359,
+      "step": 4790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5733333333333334e-05,
+      "loss": 0.9274,
+      "step": 4800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5724444444444447e-05,
+      "loss": 0.9323,
+      "step": 4810
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5715555555555557e-05,
+      "loss": 0.9501,
+      "step": 4820
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5706666666666666e-05,
+      "loss": 1.0361,
+      "step": 4830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.569777777777778e-05,
+      "loss": 0.9884,
+      "step": 4840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5688888888888893e-05,
+      "loss": 0.9659,
+      "step": 4850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5680000000000002e-05,
+      "loss": 0.9623,
+      "step": 4860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5671111111111112e-05,
+      "loss": 0.9322,
+      "step": 4870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5662222222222222e-05,
+      "loss": 0.9616,
+      "step": 4880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5653333333333335e-05,
+      "loss": 0.9586,
+      "step": 4890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5644444444444448e-05,
+      "loss": 0.9288,
+      "step": 4900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5635555555555558e-05,
+      "loss": 0.9597,
+      "step": 4910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5626666666666667e-05,
+      "loss": 0.9326,
+      "step": 4920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5617777777777777e-05,
+      "loss": 0.9742,
+      "step": 4930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.560888888888889e-05,
+      "loss": 0.9202,
+      "step": 4940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5600000000000003e-05,
+      "loss": 0.9748,
+      "step": 4950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5591111111111113e-05,
+      "loss": 1.0312,
+      "step": 4960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5582222222222222e-05,
+      "loss": 0.8969,
+      "step": 4970
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5573333333333332e-05,
+      "loss": 0.9671,
+      "step": 4980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5564444444444445e-05,
+      "loss": 0.9489,
+      "step": 4990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.555555555555556e-05,
+      "loss": 0.9591,
+      "step": 5000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5546666666666668e-05,
+      "loss": 0.8998,
+      "step": 5010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5537777777777778e-05,
+      "loss": 0.9953,
+      "step": 5020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.552888888888889e-05,
+      "loss": 0.8992,
+      "step": 5030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.552e-05,
+      "loss": 0.9277,
+      "step": 5040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5511111111111114e-05,
+      "loss": 0.9116,
+      "step": 5050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5502222222222223e-05,
+      "loss": 0.9418,
+      "step": 5060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5493333333333333e-05,
+      "loss": 1.0156,
+      "step": 5070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5484444444444446e-05,
+      "loss": 0.8966,
+      "step": 5080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5475555555555556e-05,
+      "loss": 0.9672,
+      "step": 5090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.546666666666667e-05,
+      "loss": 0.9586,
+      "step": 5100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.545777777777778e-05,
+      "loss": 0.9494,
+      "step": 5110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.544888888888889e-05,
+      "loss": 1.0093,
+      "step": 5120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.544e-05,
+      "loss": 0.9218,
+      "step": 5130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.543111111111111e-05,
+      "loss": 0.9788,
+      "step": 5140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5422222222222224e-05,
+      "loss": 1.0103,
+      "step": 5150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5413333333333337e-05,
+      "loss": 0.9994,
+      "step": 5160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5404444444444447e-05,
+      "loss": 1.0622,
+      "step": 5170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5395555555555556e-05,
+      "loss": 0.9772,
+      "step": 5180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5386666666666666e-05,
+      "loss": 1.0067,
+      "step": 5190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.537777777777778e-05,
+      "loss": 0.9603,
+      "step": 5200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5368888888888892e-05,
+      "loss": 0.9853,
+      "step": 5210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5360000000000002e-05,
+      "loss": 0.9613,
+      "step": 5220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5351111111111112e-05,
+      "loss": 0.9236,
+      "step": 5230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.534222222222222e-05,
+      "loss": 0.9349,
+      "step": 5240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5333333333333334e-05,
+      "loss": 0.9557,
+      "step": 5250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5324444444444448e-05,
+      "loss": 0.9329,
+      "step": 5260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5315555555555557e-05,
+      "loss": 0.9509,
+      "step": 5270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5306666666666667e-05,
+      "loss": 0.8865,
+      "step": 5280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5297777777777777e-05,
+      "loss": 0.991,
+      "step": 5290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.528888888888889e-05,
+      "loss": 0.8983,
+      "step": 5300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5280000000000003e-05,
+      "loss": 0.9348,
+      "step": 5310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5271111111111112e-05,
+      "loss": 0.9497,
+      "step": 5320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5262222222222222e-05,
+      "loss": 0.9707,
+      "step": 5330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5253333333333335e-05,
+      "loss": 0.9372,
+      "step": 5340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5244444444444447e-05,
+      "loss": 0.8714,
+      "step": 5350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5235555555555556e-05,
+      "loss": 0.9135,
+      "step": 5360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5226666666666668e-05,
+      "loss": 0.9346,
+      "step": 5370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5217777777777777e-05,
+      "loss": 0.965,
+      "step": 5380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.520888888888889e-05,
+      "loss": 0.8985,
+      "step": 5390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 0.8954,
+      "step": 5400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5191111111111112e-05,
+      "loss": 0.8545,
+      "step": 5410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5182222222222223e-05,
+      "loss": 0.9676,
+      "step": 5420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5173333333333336e-05,
+      "loss": 0.9198,
+      "step": 5430
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5164444444444446e-05,
+      "loss": 0.9798,
+      "step": 5440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5155555555555557e-05,
+      "loss": 0.9875,
+      "step": 5450
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5146666666666667e-05,
+      "loss": 0.9161,
+      "step": 5460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.513777777777778e-05,
+      "loss": 0.8654,
+      "step": 5470
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5128888888888891e-05,
+      "loss": 1.0018,
+      "step": 5480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5120000000000001e-05,
+      "loss": 0.8823,
+      "step": 5490
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5111111111111112e-05,
+      "loss": 0.9539,
+      "step": 5500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5102222222222222e-05,
+      "loss": 0.9626,
+      "step": 5510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5093333333333335e-05,
+      "loss": 0.9232,
+      "step": 5520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5084444444444446e-05,
+      "loss": 0.8346,
+      "step": 5530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5075555555555556e-05,
+      "loss": 1.0124,
+      "step": 5540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5066666666666668e-05,
+      "loss": 0.9096,
+      "step": 5550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.505777777777778e-05,
+      "loss": 0.9815,
+      "step": 5560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.504888888888889e-05,
+      "loss": 0.8835,
+      "step": 5570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5040000000000002e-05,
+      "loss": 0.9604,
+      "step": 5580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5031111111111111e-05,
+      "loss": 0.9602,
+      "step": 5590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5022222222222223e-05,
+      "loss": 0.9536,
+      "step": 5600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5013333333333336e-05,
+      "loss": 0.9988,
+      "step": 5610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5004444444444446e-05,
+      "loss": 0.9481,
+      "step": 5620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4995555555555557e-05,
+      "loss": 0.898,
+      "step": 5630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4986666666666667e-05,
+      "loss": 0.8502,
+      "step": 5640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.497777777777778e-05,
+      "loss": 0.9179,
+      "step": 5650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4968888888888891e-05,
+      "loss": 0.8196,
+      "step": 5660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.496e-05,
+      "loss": 0.9092,
+      "step": 5670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4951111111111112e-05,
+      "loss": 0.8639,
+      "step": 5680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4942222222222222e-05,
+      "loss": 0.9076,
+      "step": 5690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4933333333333335e-05,
+      "loss": 0.8628,
+      "step": 5700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4924444444444446e-05,
+      "loss": 0.9502,
+      "step": 5710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4915555555555556e-05,
+      "loss": 1.0408,
+      "step": 5720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4906666666666667e-05,
+      "loss": 0.855,
+      "step": 5730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.489777777777778e-05,
+      "loss": 0.9414,
+      "step": 5740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.488888888888889e-05,
+      "loss": 0.9403,
+      "step": 5750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4880000000000002e-05,
+      "loss": 0.8758,
+      "step": 5760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4871111111111111e-05,
+      "loss": 0.9094,
+      "step": 5770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4862222222222223e-05,
+      "loss": 0.8756,
+      "step": 5780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4853333333333336e-05,
+      "loss": 0.8516,
+      "step": 5790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4844444444444445e-05,
+      "loss": 0.9689,
+      "step": 5800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4835555555555557e-05,
+      "loss": 0.9742,
+      "step": 5810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4826666666666666e-05,
+      "loss": 0.9089,
+      "step": 5820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.481777777777778e-05,
+      "loss": 0.9663,
+      "step": 5830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4808888888888891e-05,
+      "loss": 0.8227,
+      "step": 5840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.48e-05,
+      "loss": 0.8964,
+      "step": 5850
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4791111111111112e-05,
+      "loss": 0.8939,
+      "step": 5860
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4782222222222225e-05,
+      "loss": 0.9705,
+      "step": 5870
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4773333333333335e-05,
+      "loss": 0.8779,
+      "step": 5880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4764444444444446e-05,
+      "loss": 0.9388,
+      "step": 5890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4755555555555556e-05,
+      "loss": 0.94,
+      "step": 5900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4746666666666667e-05,
+      "loss": 0.8964,
+      "step": 5910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.473777777777778e-05,
+      "loss": 0.908,
+      "step": 5920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.472888888888889e-05,
+      "loss": 0.8959,
+      "step": 5930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4720000000000001e-05,
+      "loss": 0.8989,
+      "step": 5940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4711111111111111e-05,
+      "loss": 0.8716,
+      "step": 5950
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4702222222222224e-05,
+      "loss": 0.832,
+      "step": 5960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4693333333333336e-05,
+      "loss": 0.8614,
+      "step": 5970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4684444444444445e-05,
+      "loss": 0.892,
+      "step": 5980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4675555555555557e-05,
+      "loss": 0.844,
+      "step": 5990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 0.9293,
+      "step": 6000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.465777777777778e-05,
+      "loss": 0.839,
+      "step": 6010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.464888888888889e-05,
+      "loss": 0.856,
+      "step": 6020
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.464e-05,
+      "loss": 0.8213,
+      "step": 6030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4631111111111112e-05,
+      "loss": 0.9093,
+      "step": 6040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4622222222222225e-05,
+      "loss": 0.8882,
+      "step": 6050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4613333333333335e-05,
+      "loss": 0.9232,
+      "step": 6060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4604444444444446e-05,
+      "loss": 0.9056,
+      "step": 6070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4595555555555556e-05,
+      "loss": 0.855,
+      "step": 6080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4586666666666667e-05,
+      "loss": 0.9247,
+      "step": 6090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.457777777777778e-05,
+      "loss": 0.9013,
+      "step": 6100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.456888888888889e-05,
+      "loss": 0.8943,
+      "step": 6110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4560000000000001e-05,
+      "loss": 0.8485,
+      "step": 6120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4551111111111111e-05,
+      "loss": 0.8636,
+      "step": 6130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4542222222222224e-05,
+      "loss": 0.9246,
+      "step": 6140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4533333333333335e-05,
+      "loss": 0.9055,
+      "step": 6150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4524444444444445e-05,
+      "loss": 0.8709,
+      "step": 6160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4515555555555556e-05,
+      "loss": 0.9288,
+      "step": 6170
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.450666666666667e-05,
+      "loss": 0.896,
+      "step": 6180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.449777777777778e-05,
+      "loss": 0.9383,
+      "step": 6190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.448888888888889e-05,
+      "loss": 0.8979,
+      "step": 6200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.448e-05,
+      "loss": 0.8791,
+      "step": 6210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4471111111111112e-05,
+      "loss": 0.8206,
+      "step": 6220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4462222222222225e-05,
+      "loss": 0.8948,
+      "step": 6230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4453333333333334e-05,
+      "loss": 0.9486,
+      "step": 6240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4444444444444446e-05,
+      "loss": 0.9089,
+      "step": 6250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4435555555555556e-05,
+      "loss": 0.9173,
+      "step": 6260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4426666666666669e-05,
+      "loss": 0.9715,
+      "step": 6270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.441777777777778e-05,
+      "loss": 0.8725,
+      "step": 6280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.440888888888889e-05,
+      "loss": 0.905,
+      "step": 6290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 0.8985,
+      "step": 6300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.439111111111111e-05,
+      "loss": 0.9293,
+      "step": 6310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4382222222222224e-05,
+      "loss": 0.9075,
+      "step": 6320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4373333333333335e-05,
+      "loss": 0.8554,
+      "step": 6330
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4364444444444445e-05,
+      "loss": 0.9562,
+      "step": 6340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4355555555555556e-05,
+      "loss": 0.8829,
+      "step": 6350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.434666666666667e-05,
+      "loss": 0.926,
+      "step": 6360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4337777777777779e-05,
+      "loss": 0.9242,
+      "step": 6370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.432888888888889e-05,
+      "loss": 0.9388,
+      "step": 6380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.432e-05,
+      "loss": 0.9885,
+      "step": 6390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4311111111111111e-05,
+      "loss": 0.9199,
+      "step": 6400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4302222222222225e-05,
+      "loss": 0.8749,
+      "step": 6410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4293333333333334e-05,
+      "loss": 0.8322,
+      "step": 6420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4284444444444446e-05,
+      "loss": 0.9338,
+      "step": 6430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4275555555555555e-05,
+      "loss": 0.8999,
+      "step": 6440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4266666666666668e-05,
+      "loss": 0.8232,
+      "step": 6450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.425777777777778e-05,
+      "loss": 0.8343,
+      "step": 6460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.424888888888889e-05,
+      "loss": 0.8645,
+      "step": 6470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4240000000000001e-05,
+      "loss": 0.9297,
+      "step": 6480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4231111111111114e-05,
+      "loss": 0.8599,
+      "step": 6490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4222222222222224e-05,
+      "loss": 0.9592,
+      "step": 6500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4213333333333335e-05,
+      "loss": 0.8855,
+      "step": 6510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4204444444444445e-05,
+      "loss": 0.9007,
+      "step": 6520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4195555555555556e-05,
+      "loss": 0.8942,
+      "step": 6530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.418666666666667e-05,
+      "loss": 0.8361,
+      "step": 6540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4177777777777779e-05,
+      "loss": 0.8866,
+      "step": 6550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.416888888888889e-05,
+      "loss": 0.8302,
+      "step": 6560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.416e-05,
+      "loss": 0.8951,
+      "step": 6570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4151111111111113e-05,
+      "loss": 0.8963,
+      "step": 6580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4142222222222224e-05,
+      "loss": 0.874,
+      "step": 6590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4133333333333334e-05,
+      "loss": 0.9074,
+      "step": 6600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4124444444444445e-05,
+      "loss": 0.8651,
+      "step": 6610
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4115555555555555e-05,
+      "loss": 0.8497,
+      "step": 6620
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4106666666666668e-05,
+      "loss": 0.8779,
+      "step": 6630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.409777777777778e-05,
+      "loss": 0.9283,
+      "step": 6640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.408888888888889e-05,
+      "loss": 0.959,
+      "step": 6650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.408e-05,
+      "loss": 0.9327,
+      "step": 6660
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4071111111111114e-05,
+      "loss": 0.9255,
+      "step": 6670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4062222222222223e-05,
+      "loss": 0.9078,
+      "step": 6680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4053333333333335e-05,
+      "loss": 0.8844,
+      "step": 6690
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4044444444444445e-05,
+      "loss": 0.8804,
+      "step": 6700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4035555555555556e-05,
+      "loss": 0.8899,
+      "step": 6710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4026666666666669e-05,
+      "loss": 0.7667,
+      "step": 6720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4017777777777779e-05,
+      "loss": 0.8689,
+      "step": 6730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.400888888888889e-05,
+      "loss": 0.8819,
+      "step": 6740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4e-05,
+      "loss": 0.9365,
+      "step": 6750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3991111111111113e-05,
+      "loss": 0.8558,
+      "step": 6760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3982222222222224e-05,
+      "loss": 0.8704,
+      "step": 6770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3973333333333334e-05,
+      "loss": 0.8887,
+      "step": 6780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3964444444444445e-05,
+      "loss": 0.9008,
+      "step": 6790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3955555555555558e-05,
+      "loss": 0.922,
+      "step": 6800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3946666666666668e-05,
+      "loss": 0.9465,
+      "step": 6810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.393777777777778e-05,
+      "loss": 0.8523,
+      "step": 6820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.392888888888889e-05,
+      "loss": 0.8503,
+      "step": 6830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.392e-05,
+      "loss": 0.8935,
+      "step": 6840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3911111111111114e-05,
+      "loss": 0.8598,
+      "step": 6850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3902222222222223e-05,
+      "loss": 0.9195,
+      "step": 6860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3893333333333335e-05,
+      "loss": 0.8922,
+      "step": 6870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3884444444444444e-05,
+      "loss": 0.882,
+      "step": 6880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3875555555555557e-05,
+      "loss": 0.8444,
+      "step": 6890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3866666666666669e-05,
+      "loss": 0.9444,
+      "step": 6900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3857777777777779e-05,
+      "loss": 0.8691,
+      "step": 6910
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.384888888888889e-05,
+      "loss": 0.9562,
+      "step": 6920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.384e-05,
+      "loss": 0.9621,
+      "step": 6930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3831111111111113e-05,
+      "loss": 0.9059,
+      "step": 6940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3822222222222224e-05,
+      "loss": 0.8453,
+      "step": 6950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3813333333333334e-05,
+      "loss": 0.907,
+      "step": 6960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3804444444444445e-05,
+      "loss": 0.8586,
+      "step": 6970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3795555555555558e-05,
+      "loss": 0.8168,
+      "step": 6980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3786666666666668e-05,
+      "loss": 0.8572,
+      "step": 6990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.377777777777778e-05,
+      "loss": 0.8161,
+      "step": 7000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3768888888888889e-05,
+      "loss": 0.869,
+      "step": 7010
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.376e-05,
+      "loss": 0.8421,
+      "step": 7020
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3751111111111113e-05,
+      "loss": 0.8431,
+      "step": 7030
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3742222222222223e-05,
+      "loss": 0.8565,
+      "step": 7040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3733333333333335e-05,
+      "loss": 0.8588,
+      "step": 7050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3724444444444444e-05,
+      "loss": 0.9456,
+      "step": 7060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3715555555555557e-05,
+      "loss": 0.9185,
+      "step": 7070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3706666666666669e-05,
+      "loss": 0.8514,
+      "step": 7080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3697777777777778e-05,
+      "loss": 0.8622,
+      "step": 7090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.368888888888889e-05,
+      "loss": 0.8476,
+      "step": 7100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3680000000000003e-05,
+      "loss": 0.8,
+      "step": 7110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3671111111111113e-05,
+      "loss": 0.7997,
+      "step": 7120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3662222222222224e-05,
+      "loss": 0.8403,
+      "step": 7130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3653333333333334e-05,
+      "loss": 0.83,
+      "step": 7140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3644444444444445e-05,
+      "loss": 0.817,
+      "step": 7150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3635555555555558e-05,
+      "loss": 0.8979,
+      "step": 7160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3626666666666668e-05,
+      "loss": 0.873,
+      "step": 7170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.361777777777778e-05,
+      "loss": 0.8684,
+      "step": 7180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3608888888888889e-05,
+      "loss": 0.8954,
+      "step": 7190
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 0.8569,
+      "step": 7200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3591111111111113e-05,
+      "loss": 0.8912,
+      "step": 7210
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3582222222222223e-05,
+      "loss": 0.8192,
+      "step": 7220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3573333333333334e-05,
+      "loss": 0.8306,
+      "step": 7230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3564444444444444e-05,
+      "loss": 0.9216,
+      "step": 7240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3555555555555557e-05,
+      "loss": 0.866,
+      "step": 7250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3546666666666669e-05,
+      "loss": 0.8389,
+      "step": 7260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3537777777777778e-05,
+      "loss": 0.8611,
+      "step": 7270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.352888888888889e-05,
+      "loss": 0.8438,
+      "step": 7280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3520000000000003e-05,
+      "loss": 0.8522,
+      "step": 7290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3511111111111112e-05,
+      "loss": 0.759,
+      "step": 7300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3502222222222224e-05,
+      "loss": 0.8135,
+      "step": 7310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3493333333333333e-05,
+      "loss": 0.8795,
+      "step": 7320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3484444444444445e-05,
+      "loss": 0.8669,
+      "step": 7330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3475555555555558e-05,
+      "loss": 0.9327,
+      "step": 7340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3466666666666668e-05,
+      "loss": 0.9265,
+      "step": 7350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3457777777777779e-05,
+      "loss": 0.876,
+      "step": 7360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3448888888888889e-05,
+      "loss": 0.8412,
+      "step": 7370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3440000000000002e-05,
+      "loss": 0.8775,
+      "step": 7380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3431111111111113e-05,
+      "loss": 0.8599,
+      "step": 7390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3422222222222223e-05,
+      "loss": 0.9101,
+      "step": 7400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3413333333333334e-05,
+      "loss": 0.8497,
+      "step": 7410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3404444444444447e-05,
+      "loss": 0.8807,
+      "step": 7420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3395555555555557e-05,
+      "loss": 0.8847,
+      "step": 7430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3386666666666668e-05,
+      "loss": 0.8731,
+      "step": 7440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3377777777777778e-05,
+      "loss": 0.85,
+      "step": 7450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.336888888888889e-05,
+      "loss": 0.836,
+      "step": 7460
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3360000000000003e-05,
+      "loss": 0.8322,
+      "step": 7470
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3351111111111112e-05,
+      "loss": 0.8306,
+      "step": 7480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3342222222222224e-05,
+      "loss": 0.8685,
+      "step": 7490
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.8619,
+      "step": 7500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3324444444444446e-05,
+      "loss": 0.7652,
+      "step": 7510
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3315555555555558e-05,
+      "loss": 0.8104,
+      "step": 7520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3306666666666667e-05,
+      "loss": 0.7767,
+      "step": 7530
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3297777777777779e-05,
+      "loss": 0.8019,
+      "step": 7540
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3288888888888889e-05,
+      "loss": 0.8629,
+      "step": 7550
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3280000000000002e-05,
+      "loss": 0.7541,
+      "step": 7560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3271111111111113e-05,
+      "loss": 0.7929,
+      "step": 7570
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3262222222222223e-05,
+      "loss": 0.8354,
+      "step": 7580
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3253333333333334e-05,
+      "loss": 0.8083,
+      "step": 7590
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3244444444444447e-05,
+      "loss": 0.7403,
+      "step": 7600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3235555555555557e-05,
+      "loss": 0.7754,
+      "step": 7610
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3226666666666668e-05,
+      "loss": 0.8386,
+      "step": 7620
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3217777777777778e-05,
+      "loss": 0.7527,
+      "step": 7630
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.320888888888889e-05,
+      "loss": 0.7943,
+      "step": 7640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3200000000000002e-05,
+      "loss": 0.8167,
+      "step": 7650
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3191111111111112e-05,
+      "loss": 0.8685,
+      "step": 7660
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3182222222222223e-05,
+      "loss": 0.8758,
+      "step": 7670
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3173333333333333e-05,
+      "loss": 0.8335,
+      "step": 7680
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3164444444444446e-05,
+      "loss": 0.8248,
+      "step": 7690
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3155555555555558e-05,
+      "loss": 0.8254,
+      "step": 7700
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3146666666666667e-05,
+      "loss": 0.8562,
+      "step": 7710
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3137777777777779e-05,
+      "loss": 0.8274,
+      "step": 7720
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3128888888888892e-05,
+      "loss": 0.7922,
+      "step": 7730
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3120000000000001e-05,
+      "loss": 0.7828,
+      "step": 7740
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3111111111111113e-05,
+      "loss": 0.8065,
+      "step": 7750
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3102222222222223e-05,
+      "loss": 0.8148,
+      "step": 7760
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3093333333333334e-05,
+      "loss": 0.7478,
+      "step": 7770
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3084444444444447e-05,
+      "loss": 0.776,
+      "step": 7780
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3075555555555557e-05,
+      "loss": 0.8066,
+      "step": 7790
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3066666666666668e-05,
+      "loss": 0.7365,
+      "step": 7800
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3057777777777778e-05,
+      "loss": 0.8159,
+      "step": 7810
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.304888888888889e-05,
+      "loss": 0.8735,
+      "step": 7820
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3040000000000002e-05,
+      "loss": 0.7734,
+      "step": 7830
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3031111111111112e-05,
+      "loss": 0.877,
+      "step": 7840
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3022222222222223e-05,
+      "loss": 0.8253,
+      "step": 7850
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3013333333333333e-05,
+      "loss": 0.7839,
+      "step": 7860
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3004444444444446e-05,
+      "loss": 0.7516,
+      "step": 7870
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2995555555555557e-05,
+      "loss": 0.8081,
+      "step": 7880
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2986666666666667e-05,
+      "loss": 0.8645,
+      "step": 7890
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2977777777777779e-05,
+      "loss": 0.8403,
+      "step": 7900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2968888888888892e-05,
+      "loss": 0.904,
+      "step": 7910
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2960000000000001e-05,
+      "loss": 0.7961,
+      "step": 7920
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2951111111111113e-05,
+      "loss": 0.7917,
+      "step": 7930
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2942222222222222e-05,
+      "loss": 0.8858,
+      "step": 7940
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2933333333333334e-05,
+      "loss": 0.81,
+      "step": 7950
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2924444444444447e-05,
+      "loss": 0.7789,
+      "step": 7960
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2915555555555557e-05,
+      "loss": 0.814,
+      "step": 7970
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2906666666666668e-05,
+      "loss": 0.8832,
+      "step": 7980
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2897777777777778e-05,
+      "loss": 0.7147,
+      "step": 7990
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.288888888888889e-05,
+      "loss": 0.8188,
+      "step": 8000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2880000000000002e-05,
+      "loss": 0.7606,
+      "step": 8010
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2871111111111112e-05,
+      "loss": 0.8368,
+      "step": 8020
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2862222222222223e-05,
+      "loss": 0.8728,
+      "step": 8030
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2853333333333336e-05,
+      "loss": 0.8504,
+      "step": 8040
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2844444444444446e-05,
+      "loss": 0.7784,
+      "step": 8050
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2835555555555557e-05,
+      "loss": 0.7533,
+      "step": 8060
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2826666666666667e-05,
+      "loss": 0.7817,
+      "step": 8070
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2817777777777778e-05,
+      "loss": 0.7948,
+      "step": 8080
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2808888888888891e-05,
+      "loss": 0.8063,
+      "step": 8090
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 0.7936,
+      "step": 8100
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2791111111111112e-05,
+      "loss": 0.8074,
+      "step": 8110
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2782222222222222e-05,
+      "loss": 0.8113,
+      "step": 8120
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2773333333333335e-05,
+      "loss": 0.7884,
+      "step": 8130
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2764444444444447e-05,
+      "loss": 0.8021,
+      "step": 8140
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2755555555555556e-05,
+      "loss": 0.7884,
+      "step": 8150
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2746666666666668e-05,
+      "loss": 0.8074,
+      "step": 8160
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2737777777777777e-05,
+      "loss": 0.8423,
+      "step": 8170
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.272888888888889e-05,
+      "loss": 0.8459,
+      "step": 8180
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2720000000000002e-05,
+      "loss": 0.7375,
+      "step": 8190
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2711111111111112e-05,
+      "loss": 0.8253,
+      "step": 8200
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2702222222222223e-05,
+      "loss": 0.7408,
+      "step": 8210
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2693333333333336e-05,
+      "loss": 0.8176,
+      "step": 8220
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2684444444444446e-05,
+      "loss": 0.8467,
+      "step": 8230
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2675555555555557e-05,
+      "loss": 0.7997,
+      "step": 8240
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2666666666666667e-05,
+      "loss": 0.8435,
+      "step": 8250
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2657777777777778e-05,
+      "loss": 0.6854,
+      "step": 8260
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2648888888888891e-05,
+      "loss": 0.8151,
+      "step": 8270
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2640000000000001e-05,
+      "loss": 0.8691,
+      "step": 8280
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2631111111111112e-05,
+      "loss": 0.7453,
+      "step": 8290
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2622222222222222e-05,
+      "loss": 0.8242,
+      "step": 8300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2613333333333335e-05,
+      "loss": 0.7078,
+      "step": 8310
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2604444444444446e-05,
+      "loss": 0.7312,
+      "step": 8320
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2595555555555556e-05,
+      "loss": 0.7741,
+      "step": 8330
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2586666666666668e-05,
+      "loss": 0.7878,
+      "step": 8340
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.257777777777778e-05,
+      "loss": 0.7667,
+      "step": 8350
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.256888888888889e-05,
+      "loss": 0.8244,
+      "step": 8360
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2560000000000002e-05,
+      "loss": 0.8027,
+      "step": 8370
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2551111111111111e-05,
+      "loss": 0.7907,
+      "step": 8380
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2542222222222223e-05,
+      "loss": 0.77,
+      "step": 8390
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2533333333333336e-05,
+      "loss": 0.7589,
+      "step": 8400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2524444444444446e-05,
+      "loss": 0.7713,
+      "step": 8410
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2515555555555557e-05,
+      "loss": 0.8287,
+      "step": 8420
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2506666666666667e-05,
+      "loss": 0.861,
+      "step": 8430
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.249777777777778e-05,
+      "loss": 0.8335,
+      "step": 8440
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2488888888888891e-05,
+      "loss": 0.8153,
+      "step": 8450
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.248e-05,
+      "loss": 0.8117,
+      "step": 8460
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2471111111111112e-05,
+      "loss": 0.7395,
+      "step": 8470
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2462222222222222e-05,
+      "loss": 0.7969,
+      "step": 8480
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2453333333333335e-05,
+      "loss": 0.7664,
+      "step": 8490
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2444444444444446e-05,
+      "loss": 0.788,
+      "step": 8500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2435555555555556e-05,
+      "loss": 0.7665,
+      "step": 8510
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2426666666666667e-05,
+      "loss": 0.7799,
+      "step": 8520
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.241777777777778e-05,
+      "loss": 0.7451,
+      "step": 8530
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.240888888888889e-05,
+      "loss": 0.7908,
+      "step": 8540
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2400000000000002e-05,
+      "loss": 0.7834,
+      "step": 8550
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2391111111111111e-05,
+      "loss": 0.8276,
+      "step": 8560
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2382222222222223e-05,
+      "loss": 0.7491,
+      "step": 8570
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2373333333333336e-05,
+      "loss": 0.8073,
+      "step": 8580
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2364444444444445e-05,
+      "loss": 0.7062,
+      "step": 8590
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2355555555555557e-05,
+      "loss": 0.8316,
+      "step": 8600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2346666666666666e-05,
+      "loss": 0.8742,
+      "step": 8610
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.233777777777778e-05,
+      "loss": 0.8407,
+      "step": 8620
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2328888888888891e-05,
+      "loss": 0.8387,
+      "step": 8630
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.232e-05,
+      "loss": 0.7871,
+      "step": 8640
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2311111111111112e-05,
+      "loss": 0.8154,
+      "step": 8650
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2302222222222225e-05,
+      "loss": 0.8002,
+      "step": 8660
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2293333333333335e-05,
+      "loss": 0.7993,
+      "step": 8670
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2284444444444446e-05,
+      "loss": 0.7214,
+      "step": 8680
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2275555555555556e-05,
+      "loss": 0.7596,
+      "step": 8690
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2266666666666667e-05,
+      "loss": 0.7683,
+      "step": 8700
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.225777777777778e-05,
+      "loss": 0.7787,
+      "step": 8710
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.224888888888889e-05,
+      "loss": 0.8293,
+      "step": 8720
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2240000000000001e-05,
+      "loss": 0.8355,
+      "step": 8730
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2231111111111111e-05,
+      "loss": 0.8168,
+      "step": 8740
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2222222222222224e-05,
+      "loss": 0.8321,
+      "step": 8750
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2213333333333336e-05,
+      "loss": 0.7942,
+      "step": 8760
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2204444444444445e-05,
+      "loss": 0.7354,
+      "step": 8770
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2195555555555557e-05,
+      "loss": 0.7821,
+      "step": 8780
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2186666666666666e-05,
+      "loss": 0.8045,
+      "step": 8790
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.217777777777778e-05,
+      "loss": 0.7925,
+      "step": 8800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.216888888888889e-05,
+      "loss": 0.7675,
+      "step": 8810
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.216e-05,
+      "loss": 0.8388,
+      "step": 8820
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2151111111111112e-05,
+      "loss": 0.7763,
+      "step": 8830
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2142222222222225e-05,
+      "loss": 0.725,
+      "step": 8840
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2133333333333335e-05,
+      "loss": 0.7409,
+      "step": 8850
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2124444444444446e-05,
+      "loss": 0.7972,
+      "step": 8860
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2115555555555556e-05,
+      "loss": 0.7559,
+      "step": 8870
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2106666666666667e-05,
+      "loss": 0.7916,
+      "step": 8880
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.209777777777778e-05,
+      "loss": 0.8488,
+      "step": 8890
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.208888888888889e-05,
+      "loss": 0.7573,
+      "step": 8900
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2080000000000001e-05,
+      "loss": 0.8603,
+      "step": 8910
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2071111111111111e-05,
+      "loss": 0.7329,
+      "step": 8920
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2062222222222224e-05,
+      "loss": 0.7988,
+      "step": 8930
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2053333333333335e-05,
+      "loss": 0.6915,
+      "step": 8940
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2044444444444445e-05,
+      "loss": 0.7726,
+      "step": 8950
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2035555555555556e-05,
+      "loss": 0.7653,
+      "step": 8960
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.202666666666667e-05,
+      "loss": 0.7435,
+      "step": 8970
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.201777777777778e-05,
+      "loss": 0.7668,
+      "step": 8980
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.200888888888889e-05,
+      "loss": 0.7759,
+      "step": 8990
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2e-05,
+      "loss": 0.7711,
+      "step": 9000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1991111111111112e-05,
+      "loss": 0.7968,
+      "step": 9010
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1982222222222225e-05,
+      "loss": 0.7736,
+      "step": 9020
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1973333333333334e-05,
+      "loss": 0.7461,
+      "step": 9030
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1964444444444446e-05,
+      "loss": 0.7298,
+      "step": 9040
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1955555555555556e-05,
+      "loss": 0.7281,
+      "step": 9050
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1946666666666669e-05,
+      "loss": 0.7728,
+      "step": 9060
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.193777777777778e-05,
+      "loss": 0.8506,
+      "step": 9070
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.192888888888889e-05,
+      "loss": 0.8115,
+      "step": 9080
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1920000000000001e-05,
+      "loss": 0.7715,
+      "step": 9090
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.191111111111111e-05,
+      "loss": 0.8234,
+      "step": 9100
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1902222222222224e-05,
+      "loss": 0.7911,
+      "step": 9110
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1893333333333335e-05,
+      "loss": 0.7786,
+      "step": 9120
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1884444444444445e-05,
+      "loss": 0.846,
+      "step": 9130
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1875555555555556e-05,
+      "loss": 0.7619,
+      "step": 9140
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.186666666666667e-05,
+      "loss": 0.7318,
+      "step": 9150
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1857777777777779e-05,
+      "loss": 0.8328,
+      "step": 9160
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.184888888888889e-05,
+      "loss": 0.7974,
+      "step": 9170
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.184e-05,
+      "loss": 0.8137,
+      "step": 9180
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1831111111111112e-05,
+      "loss": 0.8265,
+      "step": 9190
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1822222222222225e-05,
+      "loss": 0.7497,
+      "step": 9200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1813333333333334e-05,
+      "loss": 0.7859,
+      "step": 9210
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1804444444444446e-05,
+      "loss": 0.7464,
+      "step": 9220
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1795555555555555e-05,
+      "loss": 0.7343,
+      "step": 9230
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1786666666666668e-05,
+      "loss": 0.744,
+      "step": 9240
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.177777777777778e-05,
+      "loss": 0.914,
+      "step": 9250
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.176888888888889e-05,
+      "loss": 0.8342,
+      "step": 9260
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1760000000000001e-05,
+      "loss": 0.7114,
+      "step": 9270
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1751111111111112e-05,
+      "loss": 0.7673,
+      "step": 9280
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1742222222222224e-05,
+      "loss": 0.786,
+      "step": 9290
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1733333333333335e-05,
+      "loss": 0.7374,
+      "step": 9300
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1724444444444445e-05,
+      "loss": 0.7658,
+      "step": 9310
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1715555555555556e-05,
+      "loss": 0.7388,
+      "step": 9320
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.170666666666667e-05,
+      "loss": 0.7143,
+      "step": 9330
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1697777777777779e-05,
+      "loss": 0.7885,
+      "step": 9340
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.168888888888889e-05,
+      "loss": 0.76,
+      "step": 9350
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.168e-05,
+      "loss": 0.788,
+      "step": 9360
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1671111111111113e-05,
+      "loss": 0.7285,
+      "step": 9370
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1662222222222224e-05,
+      "loss": 0.7672,
+      "step": 9380
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1653333333333334e-05,
+      "loss": 0.7399,
+      "step": 9390
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1644444444444446e-05,
+      "loss": 0.8254,
+      "step": 9400
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1635555555555555e-05,
+      "loss": 0.746,
+      "step": 9410
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1626666666666668e-05,
+      "loss": 0.8151,
+      "step": 9420
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.161777777777778e-05,
+      "loss": 0.7813,
+      "step": 9430
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.160888888888889e-05,
+      "loss": 0.7144,
+      "step": 9440
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.16e-05,
+      "loss": 0.7607,
+      "step": 9450
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1591111111111114e-05,
+      "loss": 0.8035,
+      "step": 9460
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1582222222222224e-05,
+      "loss": 0.7826,
+      "step": 9470
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1573333333333335e-05,
+      "loss": 0.7327,
+      "step": 9480
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1564444444444445e-05,
+      "loss": 0.7162,
+      "step": 9490
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1555555555555556e-05,
+      "loss": 0.8058,
+      "step": 9500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1546666666666669e-05,
+      "loss": 0.7683,
+      "step": 9510
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1537777777777779e-05,
+      "loss": 0.8565,
+      "step": 9520
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.152888888888889e-05,
+      "loss": 0.7249,
+      "step": 9530
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.152e-05,
+      "loss": 0.7367,
+      "step": 9540
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1511111111111113e-05,
+      "loss": 0.7976,
+      "step": 9550
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1502222222222224e-05,
+      "loss": 0.808,
+      "step": 9560
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1493333333333334e-05,
+      "loss": 0.7246,
+      "step": 9570
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1484444444444445e-05,
+      "loss": 0.7858,
+      "step": 9580
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1475555555555557e-05,
+      "loss": 0.7846,
+      "step": 9590
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1466666666666668e-05,
+      "loss": 0.7237,
+      "step": 9600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.145777777777778e-05,
+      "loss": 0.8154,
+      "step": 9610
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.144888888888889e-05,
+      "loss": 0.8533,
+      "step": 9620
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.144e-05,
+      "loss": 0.7323,
+      "step": 9630
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1431111111111114e-05,
+      "loss": 0.8197,
+      "step": 9640
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1422222222222223e-05,
+      "loss": 0.7628,
+      "step": 9650
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1413333333333335e-05,
+      "loss": 0.7413,
+      "step": 9660
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1404444444444444e-05,
+      "loss": 0.7718,
+      "step": 9670
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1395555555555558e-05,
+      "loss": 0.7067,
+      "step": 9680
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1386666666666669e-05,
+      "loss": 0.8018,
+      "step": 9690
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1377777777777779e-05,
+      "loss": 0.7511,
+      "step": 9700
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.136888888888889e-05,
+      "loss": 0.7979,
+      "step": 9710
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.136e-05,
+      "loss": 0.7846,
+      "step": 9720
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1351111111111113e-05,
+      "loss": 0.7348,
+      "step": 9730
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1342222222222224e-05,
+      "loss": 0.7523,
+      "step": 9740
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1333333333333334e-05,
+      "loss": 0.8074,
+      "step": 9750
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1324444444444445e-05,
+      "loss": 0.7263,
+      "step": 9760
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1315555555555557e-05,
+      "loss": 0.8393,
+      "step": 9770
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1306666666666668e-05,
+      "loss": 0.7898,
+      "step": 9780
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.129777777777778e-05,
+      "loss": 0.8025,
+      "step": 9790
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1288888888888889e-05,
+      "loss": 0.7455,
+      "step": 9800
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.128e-05,
+      "loss": 0.7876,
+      "step": 9810
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1271111111111113e-05,
+      "loss": 0.7323,
+      "step": 9820
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1262222222222223e-05,
+      "loss": 0.7671,
+      "step": 9830
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1253333333333335e-05,
+      "loss": 0.8148,
+      "step": 9840
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1244444444444444e-05,
+      "loss": 0.7532,
+      "step": 9850
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1235555555555557e-05,
+      "loss": 0.7501,
+      "step": 9860
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1226666666666669e-05,
+      "loss": 0.7711,
+      "step": 9870
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1217777777777778e-05,
+      "loss": 0.7606,
+      "step": 9880
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.120888888888889e-05,
+      "loss": 0.7166,
+      "step": 9890
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 0.7462,
+      "step": 9900
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1191111111111113e-05,
+      "loss": 0.7213,
+      "step": 9910
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1182222222222224e-05,
+      "loss": 0.7864,
+      "step": 9920
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1173333333333334e-05,
+      "loss": 0.7229,
+      "step": 9930
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1164444444444445e-05,
+      "loss": 0.7744,
+      "step": 9940
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1155555555555556e-05,
+      "loss": 0.7246,
+      "step": 9950
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1146666666666668e-05,
+      "loss": 0.7975,
+      "step": 9960
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.113777777777778e-05,
+      "loss": 0.7355,
+      "step": 9970
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1128888888888889e-05,
+      "loss": 0.8357,
+      "step": 9980
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1120000000000002e-05,
+      "loss": 0.7834,
+      "step": 9990
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1111111111111113e-05,
+      "loss": 0.8293,
+      "step": 10000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1102222222222223e-05,
+      "loss": 0.7226,
+      "step": 10010
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1093333333333334e-05,
+      "loss": 0.8126,
+      "step": 10020
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1084444444444444e-05,
+      "loss": 0.8374,
+      "step": 10030
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1075555555555557e-05,
+      "loss": 0.786,
+      "step": 10040
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1066666666666669e-05,
+      "loss": 0.8358,
+      "step": 10050
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1057777777777778e-05,
+      "loss": 0.8163,
+      "step": 10060
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.104888888888889e-05,
+      "loss": 0.7775,
+      "step": 10070
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1040000000000001e-05,
+      "loss": 0.7185,
+      "step": 10080
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1031111111111112e-05,
+      "loss": 0.758,
+      "step": 10090
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1022222222222224e-05,
+      "loss": 0.7644,
+      "step": 10100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1013333333333333e-05,
+      "loss": 0.7676,
+      "step": 10110
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1004444444444445e-05,
+      "loss": 0.6974,
+      "step": 10120
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0995555555555556e-05,
+      "loss": 0.7701,
+      "step": 10130
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0986666666666668e-05,
+      "loss": 0.7988,
+      "step": 10140
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0977777777777779e-05,
+      "loss": 0.7663,
+      "step": 10150
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0968888888888889e-05,
+      "loss": 0.7213,
+      "step": 10160
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.0960000000000002e-05,
+      "loss": 0.8264,
+      "step": 10170
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.0951111111111113e-05,
+      "loss": 0.7377,
+      "step": 10180
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.0942222222222223e-05,
+      "loss": 0.7266,
+      "step": 10190
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.0933333333333334e-05,
+      "loss": 0.8125,
+      "step": 10200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.0924444444444446e-05,
+      "loss": 0.7755,
+      "step": 10210
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.0915555555555557e-05,
+      "loss": 0.773,
+      "step": 10220
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.0906666666666668e-05,
+      "loss": 0.7298,
+      "step": 10230
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.0897777777777778e-05,
+      "loss": 0.7758,
+      "step": 10240
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.088888888888889e-05,
+      "loss": 0.742,
+      "step": 10250
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.0880000000000001e-05,
+      "loss": 0.736,
+      "step": 10260
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.0871111111111112e-05,
+      "loss": 0.7613,
+      "step": 10270
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.0862222222222224e-05,
+      "loss": 0.7463,
+      "step": 10280
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.0853333333333333e-05,
+      "loss": 0.721,
+      "step": 10290
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.0844444444444446e-05,
+      "loss": 0.7498,
+      "step": 10300
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.0835555555555556e-05,
+      "loss": 0.7986,
+      "step": 10310
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0826666666666667e-05,
+      "loss": 0.808,
+      "step": 10320
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0817777777777779e-05,
+      "loss": 0.7459,
+      "step": 10330
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0808888888888889e-05,
+      "loss": 0.7394,
+      "step": 10340
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0800000000000002e-05,
+      "loss": 0.7931,
+      "step": 10350
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0791111111111113e-05,
+      "loss": 0.7159,
+      "step": 10360
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0782222222222223e-05,
+      "loss": 0.7146,
+      "step": 10370
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0773333333333334e-05,
+      "loss": 0.7166,
+      "step": 10380
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0764444444444445e-05,
+      "loss": 0.747,
+      "step": 10390
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0755555555555557e-05,
+      "loss": 0.7531,
+      "step": 10400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0746666666666668e-05,
+      "loss": 0.7861,
+      "step": 10410
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0737777777777778e-05,
+      "loss": 0.7888,
+      "step": 10420
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.072888888888889e-05,
+      "loss": 0.6578,
+      "step": 10430
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.072e-05,
+      "loss": 0.7542,
+      "step": 10440
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0711111111111112e-05,
+      "loss": 0.7091,
+      "step": 10450
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0702222222222223e-05,
+      "loss": 0.7238,
+      "step": 10460
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0693333333333333e-05,
+      "loss": 0.6912,
+      "step": 10470
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0684444444444446e-05,
+      "loss": 0.7583,
+      "step": 10480
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0675555555555558e-05,
+      "loss": 0.7198,
+      "step": 10490
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 0.7406,
+      "step": 10500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0657777777777779e-05,
+      "loss": 0.7321,
+      "step": 10510
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.064888888888889e-05,
+      "loss": 0.7009,
+      "step": 10520
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0640000000000001e-05,
+      "loss": 0.7136,
+      "step": 10530
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0631111111111113e-05,
+      "loss": 0.7258,
+      "step": 10540
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0622222222222223e-05,
+      "loss": 0.6963,
+      "step": 10550
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0613333333333334e-05,
+      "loss": 0.7559,
+      "step": 10560
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0604444444444445e-05,
+      "loss": 0.7619,
+      "step": 10570
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0595555555555557e-05,
+      "loss": 0.7853,
+      "step": 10580
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0586666666666668e-05,
+      "loss": 0.795,
+      "step": 10590
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0577777777777778e-05,
+      "loss": 0.6995,
+      "step": 10600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0568888888888891e-05,
+      "loss": 0.7825,
+      "step": 10610
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.056e-05,
+      "loss": 0.7539,
+      "step": 10620
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0551111111111112e-05,
+      "loss": 0.7691,
+      "step": 10630
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0542222222222223e-05,
+      "loss": 0.7297,
+      "step": 10640
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0533333333333333e-05,
+      "loss": 0.6817,
+      "step": 10650
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0524444444444446e-05,
+      "loss": 0.7288,
+      "step": 10660
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0515555555555557e-05,
+      "loss": 0.7351,
+      "step": 10670
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0506666666666667e-05,
+      "loss": 0.701,
+      "step": 10680
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0497777777777779e-05,
+      "loss": 0.7694,
+      "step": 10690
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.048888888888889e-05,
+      "loss": 0.779,
+      "step": 10700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0480000000000001e-05,
+      "loss": 0.7777,
+      "step": 10710
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0471111111111113e-05,
+      "loss": 0.7087,
+      "step": 10720
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0462222222222222e-05,
+      "loss": 0.726,
+      "step": 10730
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0453333333333334e-05,
+      "loss": 0.7361,
+      "step": 10740
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0444444444444445e-05,
+      "loss": 0.7797,
+      "step": 10750
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0435555555555557e-05,
+      "loss": 0.7506,
+      "step": 10760
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0426666666666668e-05,
+      "loss": 0.7607,
+      "step": 10770
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0417777777777778e-05,
+      "loss": 0.7281,
+      "step": 10780
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.040888888888889e-05,
+      "loss": 0.8274,
+      "step": 10790
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.04e-05,
+      "loss": 0.7909,
+      "step": 10800
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0391111111111112e-05,
+      "loss": 0.7153,
+      "step": 10810
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0382222222222223e-05,
+      "loss": 0.7852,
+      "step": 10820
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0373333333333335e-05,
+      "loss": 0.7045,
+      "step": 10830
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0364444444444446e-05,
+      "loss": 0.8618,
+      "step": 10840
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0355555555555557e-05,
+      "loss": 0.7632,
+      "step": 10850
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0346666666666667e-05,
+      "loss": 0.7132,
+      "step": 10860
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0337777777777778e-05,
+      "loss": 0.6889,
+      "step": 10870
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.032888888888889e-05,
+      "loss": 0.737,
+      "step": 10880
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0320000000000001e-05,
+      "loss": 0.6662,
+      "step": 10890
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0311111111111113e-05,
+      "loss": 0.6746,
+      "step": 10900
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0302222222222222e-05,
+      "loss": 0.6366,
+      "step": 10910
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0293333333333335e-05,
+      "loss": 0.6941,
+      "step": 10920
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0284444444444445e-05,
+      "loss": 0.7469,
+      "step": 10930
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0275555555555556e-05,
+      "loss": 0.7361,
+      "step": 10940
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0266666666666668e-05,
+      "loss": 0.7217,
+      "step": 10950
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0257777777777777e-05,
+      "loss": 0.6962,
+      "step": 10960
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.024888888888889e-05,
+      "loss": 0.7235,
+      "step": 10970
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.024e-05,
+      "loss": 0.7203,
+      "step": 10980
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0231111111111112e-05,
+      "loss": 0.7356,
+      "step": 10990
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0222222222222223e-05,
+      "loss": 0.6177,
+      "step": 11000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0213333333333334e-05,
+      "loss": 0.7063,
+      "step": 11010
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0204444444444446e-05,
+      "loss": 0.7497,
+      "step": 11020
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0195555555555557e-05,
+      "loss": 0.7654,
+      "step": 11030
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0186666666666667e-05,
+      "loss": 0.7089,
+      "step": 11040
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0177777777777778e-05,
+      "loss": 0.7678,
+      "step": 11050
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.016888888888889e-05,
+      "loss": 0.7753,
+      "step": 11060
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0160000000000001e-05,
+      "loss": 0.731,
+      "step": 11070
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0151111111111112e-05,
+      "loss": 0.7266,
+      "step": 11080
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0142222222222222e-05,
+      "loss": 0.7556,
+      "step": 11090
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0133333333333335e-05,
+      "loss": 0.6988,
+      "step": 11100
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0124444444444445e-05,
+      "loss": 0.7614,
+      "step": 11110
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0115555555555556e-05,
+      "loss": 0.7615,
+      "step": 11120
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0106666666666668e-05,
+      "loss": 0.7595,
+      "step": 11130
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0097777777777779e-05,
+      "loss": 0.7555,
+      "step": 11140
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.008888888888889e-05,
+      "loss": 0.759,
+      "step": 11150
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.008e-05,
+      "loss": 0.7215,
+      "step": 11160
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0071111111111111e-05,
+      "loss": 0.736,
+      "step": 11170
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0062222222222223e-05,
+      "loss": 0.7113,
+      "step": 11180
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0053333333333334e-05,
+      "loss": 0.6735,
+      "step": 11190
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0044444444444446e-05,
+      "loss": 0.6848,
+      "step": 11200
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0035555555555557e-05,
+      "loss": 0.7385,
+      "step": 11210
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0026666666666667e-05,
+      "loss": 0.6983,
+      "step": 11220
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.001777777777778e-05,
+      "loss": 0.7155,
+      "step": 11230
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.000888888888889e-05,
+      "loss": 0.789,
+      "step": 11240
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1e-05,
+      "loss": 0.6871,
+      "step": 11250
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.991111111111112e-06,
+      "loss": 0.7541,
+      "step": 11260
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.982222222222224e-06,
+      "loss": 0.6912,
+      "step": 11270
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.973333333333333e-06,
+      "loss": 0.7459,
+      "step": 11280
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.964444444444445e-06,
+      "loss": 0.7243,
+      "step": 11290
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.955555555555556e-06,
+      "loss": 0.7236,
+      "step": 11300
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.946666666666667e-06,
+      "loss": 0.6962,
+      "step": 11310
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.937777777777779e-06,
+      "loss": 0.7803,
+      "step": 11320
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.92888888888889e-06,
+      "loss": 0.7795,
+      "step": 11330
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 0.7613,
+      "step": 11340
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.911111111111113e-06,
+      "loss": 0.726,
+      "step": 11350
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.902222222222223e-06,
+      "loss": 0.7561,
+      "step": 11360
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.893333333333334e-06,
+      "loss": 0.712,
+      "step": 11370
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.884444444444445e-06,
+      "loss": 0.7257,
+      "step": 11380
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.875555555555557e-06,
+      "loss": 0.7219,
+      "step": 11390
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.866666666666668e-06,
+      "loss": 0.803,
+      "step": 11400
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.857777777777778e-06,
+      "loss": 0.6742,
+      "step": 11410
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.84888888888889e-06,
+      "loss": 0.6409,
+      "step": 11420
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.84e-06,
+      "loss": 0.6738,
+      "step": 11430
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.831111111111112e-06,
+      "loss": 0.6947,
+      "step": 11440
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.822222222222223e-06,
+      "loss": 0.7449,
+      "step": 11450
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.813333333333333e-06,
+      "loss": 0.726,
+      "step": 11460
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.804444444444444e-06,
+      "loss": 0.7434,
+      "step": 11470
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.795555555555556e-06,
+      "loss": 0.6992,
+      "step": 11480
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.786666666666667e-06,
+      "loss": 0.7759,
+      "step": 11490
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.777777777777779e-06,
+      "loss": 0.769,
+      "step": 11500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.76888888888889e-06,
+      "loss": 0.7542,
+      "step": 11510
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 0.7031,
+      "step": 11520
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.751111111111113e-06,
+      "loss": 0.6945,
+      "step": 11530
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.742222222222222e-06,
+      "loss": 0.7452,
+      "step": 11540
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.733333333333334e-06,
+      "loss": 0.766,
+      "step": 11550
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.724444444444445e-06,
+      "loss": 0.7297,
+      "step": 11560
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.715555555555557e-06,
+      "loss": 0.7682,
+      "step": 11570
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.706666666666668e-06,
+      "loss": 0.7813,
+      "step": 11580
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.697777777777778e-06,
+      "loss": 0.7848,
+      "step": 11590
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.688888888888889e-06,
+      "loss": 0.7779,
+      "step": 11600
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.68e-06,
+      "loss": 0.7297,
+      "step": 11610
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.671111111111112e-06,
+      "loss": 0.6557,
+      "step": 11620
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.662222222222223e-06,
+      "loss": 0.6723,
+      "step": 11630
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.653333333333335e-06,
+      "loss": 0.7018,
+      "step": 11640
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.644444444444444e-06,
+      "loss": 0.7004,
+      "step": 11650
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.635555555555557e-06,
+      "loss": 0.6964,
+      "step": 11660
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.626666666666667e-06,
+      "loss": 0.7217,
+      "step": 11670
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.617777777777778e-06,
+      "loss": 0.6883,
+      "step": 11680
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.60888888888889e-06,
+      "loss": 0.6794,
+      "step": 11690
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.8018,
+      "step": 11700
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.591111111111113e-06,
+      "loss": 0.7603,
+      "step": 11710
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.582222222222222e-06,
+      "loss": 0.7066,
+      "step": 11720
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.573333333333334e-06,
+      "loss": 0.7601,
+      "step": 11730
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.564444444444445e-06,
+      "loss": 0.6663,
+      "step": 11740
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.555555555555556e-06,
+      "loss": 0.7425,
+      "step": 11750
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.546666666666668e-06,
+      "loss": 0.7277,
+      "step": 11760
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.537777777777778e-06,
+      "loss": 0.646,
+      "step": 11770
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.528888888888889e-06,
+      "loss": 0.7021,
+      "step": 11780
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.52e-06,
+      "loss": 0.6693,
+      "step": 11790
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.511111111111112e-06,
+      "loss": 0.8546,
+      "step": 11800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.502222222222223e-06,
+      "loss": 0.7618,
+      "step": 11810
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.493333333333334e-06,
+      "loss": 0.7135,
+      "step": 11820
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.484444444444444e-06,
+      "loss": 0.6883,
+      "step": 11830
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.475555555555557e-06,
+      "loss": 0.6747,
+      "step": 11840
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.466666666666667e-06,
+      "loss": 0.7313,
+      "step": 11850
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.457777777777778e-06,
+      "loss": 0.688,
+      "step": 11860
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.44888888888889e-06,
+      "loss": 0.7064,
+      "step": 11870
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 0.7938,
+      "step": 11880
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.431111111111112e-06,
+      "loss": 0.7476,
+      "step": 11890
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.422222222222222e-06,
+      "loss": 0.7319,
+      "step": 11900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.413333333333334e-06,
+      "loss": 0.6986,
+      "step": 11910
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.404444444444445e-06,
+      "loss": 0.7206,
+      "step": 11920
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.395555555555556e-06,
+      "loss": 0.7049,
+      "step": 11930
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.386666666666668e-06,
+      "loss": 0.7671,
+      "step": 11940
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.377777777777779e-06,
+      "loss": 0.6748,
+      "step": 11950
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.368888888888889e-06,
+      "loss": 0.7162,
+      "step": 11960
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.360000000000002e-06,
+      "loss": 0.7036,
+      "step": 11970
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.351111111111112e-06,
+      "loss": 0.7187,
+      "step": 11980
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.342222222222223e-06,
+      "loss": 0.7259,
+      "step": 11990
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 0.6994,
+      "step": 12000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.324444444444444e-06,
+      "loss": 0.7618,
+      "step": 12010
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.315555555555557e-06,
+      "loss": 0.7888,
+      "step": 12020
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.306666666666667e-06,
+      "loss": 0.697,
+      "step": 12030
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.297777777777778e-06,
+      "loss": 0.7245,
+      "step": 12040
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.28888888888889e-06,
+      "loss": 0.7089,
+      "step": 12050
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 0.7039,
+      "step": 12060
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.271111111111112e-06,
+      "loss": 0.6872,
+      "step": 12070
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.262222222222222e-06,
+      "loss": 0.7602,
+      "step": 12080
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.253333333333333e-06,
+      "loss": 0.71,
+      "step": 12090
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.244444444444445e-06,
+      "loss": 0.7005,
+      "step": 12100
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.235555555555556e-06,
+      "loss": 0.7184,
+      "step": 12110
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.226666666666668e-06,
+      "loss": 0.6709,
+      "step": 12120
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.217777777777779e-06,
+      "loss": 0.7453,
+      "step": 12130
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.208888888888889e-06,
+      "loss": 0.6611,
+      "step": 12140
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 0.6782,
+      "step": 12150
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.191111111111111e-06,
+      "loss": 0.756,
+      "step": 12160
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.182222222222223e-06,
+      "loss": 0.7242,
+      "step": 12170
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.173333333333334e-06,
+      "loss": 0.7422,
+      "step": 12180
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.164444444444446e-06,
+      "loss": 0.6587,
+      "step": 12190
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.155555555555557e-06,
+      "loss": 0.6848,
+      "step": 12200
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.146666666666667e-06,
+      "loss": 0.7194,
+      "step": 12210
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.137777777777778e-06,
+      "loss": 0.6904,
+      "step": 12220
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.12888888888889e-06,
+      "loss": 0.7084,
+      "step": 12230
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.12e-06,
+      "loss": 0.7074,
+      "step": 12240
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.111111111111112e-06,
+      "loss": 0.7263,
+      "step": 12250
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.102222222222224e-06,
+      "loss": 0.7782,
+      "step": 12260
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.093333333333333e-06,
+      "loss": 0.7199,
+      "step": 12270
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.084444444444446e-06,
+      "loss": 0.7075,
+      "step": 12280
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.075555555555556e-06,
+      "loss": 0.7458,
+      "step": 12290
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.066666666666667e-06,
+      "loss": 0.6965,
+      "step": 12300
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.057777777777779e-06,
+      "loss": 0.7635,
+      "step": 12310
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.048888888888888e-06,
+      "loss": 0.7366,
+      "step": 12320
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.040000000000002e-06,
+      "loss": 0.7603,
+      "step": 12330
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.031111111111111e-06,
+      "loss": 0.7134,
+      "step": 12340
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.022222222222223e-06,
+      "loss": 0.6368,
+      "step": 12350
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.013333333333334e-06,
+      "loss": 0.7192,
+      "step": 12360
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.004444444444445e-06,
+      "loss": 0.8004,
+      "step": 12370
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.995555555555557e-06,
+      "loss": 0.6733,
+      "step": 12380
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.986666666666666e-06,
+      "loss": 0.7525,
+      "step": 12390
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.977777777777778e-06,
+      "loss": 0.7409,
+      "step": 12400
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.96888888888889e-06,
+      "loss": 0.6567,
+      "step": 12410
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.96e-06,
+      "loss": 0.7242,
+      "step": 12420
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.951111111111112e-06,
+      "loss": 0.6818,
+      "step": 12430
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.942222222222223e-06,
+      "loss": 0.7027,
+      "step": 12440
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.933333333333333e-06,
+      "loss": 0.6762,
+      "step": 12450
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.924444444444446e-06,
+      "loss": 0.7043,
+      "step": 12460
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.915555555555556e-06,
+      "loss": 0.6918,
+      "step": 12470
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.906666666666667e-06,
+      "loss": 0.6995,
+      "step": 12480
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.897777777777779e-06,
+      "loss": 0.6593,
+      "step": 12490
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 0.6886,
+      "step": 12500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.880000000000001e-06,
+      "loss": 0.6795,
+      "step": 12510
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.871111111111111e-06,
+      "loss": 0.6816,
+      "step": 12520
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.862222222222222e-06,
+      "loss": 0.6703,
+      "step": 12530
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.853333333333334e-06,
+      "loss": 0.7687,
+      "step": 12540
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.844444444444445e-06,
+      "loss": 0.7018,
+      "step": 12550
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.835555555555557e-06,
+      "loss": 0.6887,
+      "step": 12560
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.826666666666668e-06,
+      "loss": 0.7376,
+      "step": 12570
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.817777777777778e-06,
+      "loss": 0.6635,
+      "step": 12580
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.80888888888889e-06,
+      "loss": 0.6776,
+      "step": 12590
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.8e-06,
+      "loss": 0.7321,
+      "step": 12600
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.791111111111112e-06,
+      "loss": 0.6588,
+      "step": 12610
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.782222222222223e-06,
+      "loss": 0.7106,
+      "step": 12620
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.773333333333333e-06,
+      "loss": 0.7018,
+      "step": 12630
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.764444444444446e-06,
+      "loss": 0.7487,
+      "step": 12640
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.755555555555556e-06,
+      "loss": 0.7104,
+      "step": 12650
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.746666666666667e-06,
+      "loss": 0.7269,
+      "step": 12660
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.737777777777778e-06,
+      "loss": 0.7153,
+      "step": 12670
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.72888888888889e-06,
+      "loss": 0.7051,
+      "step": 12680
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.720000000000001e-06,
+      "loss": 0.7838,
+      "step": 12690
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.711111111111111e-06,
+      "loss": 0.712,
+      "step": 12700
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.702222222222222e-06,
+      "loss": 0.6718,
+      "step": 12710
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.693333333333334e-06,
+      "loss": 0.7334,
+      "step": 12720
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.684444444444445e-06,
+      "loss": 0.7287,
+      "step": 12730
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.675555555555556e-06,
+      "loss": 0.7179,
+      "step": 12740
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.666666666666668e-06,
+      "loss": 0.7112,
+      "step": 12750
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.657777777777778e-06,
+      "loss": 0.7557,
+      "step": 12760
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.64888888888889e-06,
+      "loss": 0.6514,
+      "step": 12770
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.64e-06,
+      "loss": 0.716,
+      "step": 12780
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.631111111111112e-06,
+      "loss": 0.7039,
+      "step": 12790
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.622222222222223e-06,
+      "loss": 0.6395,
+      "step": 12800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.613333333333333e-06,
+      "loss": 0.713,
+      "step": 12810
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.604444444444446e-06,
+      "loss": 0.7074,
+      "step": 12820
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.595555555555556e-06,
+      "loss": 0.672,
+      "step": 12830
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.586666666666667e-06,
+      "loss": 0.7612,
+      "step": 12840
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.577777777777778e-06,
+      "loss": 0.7159,
+      "step": 12850
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.56888888888889e-06,
+      "loss": 0.6793,
+      "step": 12860
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.560000000000001e-06,
+      "loss": 0.6674,
+      "step": 12870
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.551111111111112e-06,
+      "loss": 0.6803,
+      "step": 12880
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.542222222222222e-06,
+      "loss": 0.6837,
+      "step": 12890
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.533333333333335e-06,
+      "loss": 0.7543,
+      "step": 12900
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.524444444444445e-06,
+      "loss": 0.6482,
+      "step": 12910
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.515555555555556e-06,
+      "loss": 0.64,
+      "step": 12920
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.506666666666668e-06,
+      "loss": 0.6771,
+      "step": 12930
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.497777777777777e-06,
+      "loss": 0.6631,
+      "step": 12940
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.48888888888889e-06,
+      "loss": 0.7043,
+      "step": 12950
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.48e-06,
+      "loss": 0.7257,
+      "step": 12960
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.471111111111112e-06,
+      "loss": 0.7321,
+      "step": 12970
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.462222222222223e-06,
+      "loss": 0.6808,
+      "step": 12980
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.453333333333334e-06,
+      "loss": 0.7341,
+      "step": 12990
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.444444444444446e-06,
+      "loss": 0.6642,
+      "step": 13000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.435555555555555e-06,
+      "loss": 0.6515,
+      "step": 13010
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.426666666666667e-06,
+      "loss": 0.6671,
+      "step": 13020
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.417777777777778e-06,
+      "loss": 0.7053,
+      "step": 13030
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.40888888888889e-06,
+      "loss": 0.6431,
+      "step": 13040
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 0.693,
+      "step": 13050
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.391111111111112e-06,
+      "loss": 0.7597,
+      "step": 13060
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.382222222222222e-06,
+      "loss": 0.7125,
+      "step": 13070
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.373333333333335e-06,
+      "loss": 0.5933,
+      "step": 13080
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.364444444444445e-06,
+      "loss": 0.6834,
+      "step": 13090
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.355555555555556e-06,
+      "loss": 0.6817,
+      "step": 13100
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.346666666666668e-06,
+      "loss": 0.7137,
+      "step": 13110
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.337777777777777e-06,
+      "loss": 0.7056,
+      "step": 13120
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.32888888888889e-06,
+      "loss": 0.6912,
+      "step": 13130
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.32e-06,
+      "loss": 0.7214,
+      "step": 13140
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.311111111111111e-06,
+      "loss": 0.7583,
+      "step": 13150
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.302222222222223e-06,
+      "loss": 0.6606,
+      "step": 13160
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.293333333333334e-06,
+      "loss": 0.7134,
+      "step": 13170
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.284444444444446e-06,
+      "loss": 0.6941,
+      "step": 13180
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.275555555555557e-06,
+      "loss": 0.7008,
+      "step": 13190
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.266666666666667e-06,
+      "loss": 0.732,
+      "step": 13200
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.25777777777778e-06,
+      "loss": 0.6883,
+      "step": 13210
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.24888888888889e-06,
+      "loss": 0.7943,
+      "step": 13220
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.24e-06,
+      "loss": 0.6992,
+      "step": 13230
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.231111111111112e-06,
+      "loss": 0.6463,
+      "step": 13240
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.222222222222222e-06,
+      "loss": 0.7039,
+      "step": 13250
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.213333333333335e-06,
+      "loss": 0.7103,
+      "step": 13260
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.204444444444445e-06,
+      "loss": 0.6533,
+      "step": 13270
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.195555555555556e-06,
+      "loss": 0.7102,
+      "step": 13280
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.186666666666667e-06,
+      "loss": 0.6318,
+      "step": 13290
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.177777777777779e-06,
+      "loss": 0.6831,
+      "step": 13300
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.16888888888889e-06,
+      "loss": 0.6734,
+      "step": 13310
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.16e-06,
+      "loss": 0.6934,
+      "step": 13320
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.151111111111111e-06,
+      "loss": 0.6982,
+      "step": 13330
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.142222222222223e-06,
+      "loss": 0.7029,
+      "step": 13340
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.133333333333334e-06,
+      "loss": 0.6543,
+      "step": 13350
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.124444444444445e-06,
+      "loss": 0.6207,
+      "step": 13360
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.115555555555557e-06,
+      "loss": 0.6856,
+      "step": 13370
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.106666666666666e-06,
+      "loss": 0.731,
+      "step": 13380
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.09777777777778e-06,
+      "loss": 0.6858,
+      "step": 13390
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.08888888888889e-06,
+      "loss": 0.6667,
+      "step": 13400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.08e-06,
+      "loss": 0.701,
+      "step": 13410
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.071111111111112e-06,
+      "loss": 0.708,
+      "step": 13420
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.062222222222222e-06,
+      "loss": 0.6504,
+      "step": 13430
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.053333333333335e-06,
+      "loss": 0.6561,
+      "step": 13440
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.044444444444444e-06,
+      "loss": 0.6682,
+      "step": 13450
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.035555555555556e-06,
+      "loss": 0.71,
+      "step": 13460
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.026666666666667e-06,
+      "loss": 0.7156,
+      "step": 13470
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.017777777777779e-06,
+      "loss": 0.6964,
+      "step": 13480
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.00888888888889e-06,
+      "loss": 0.6996,
+      "step": 13490
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.7139,
+      "step": 13500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.991111111111111e-06,
+      "loss": 0.76,
+      "step": 13510
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.982222222222224e-06,
+      "loss": 0.6734,
+      "step": 13520
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.973333333333334e-06,
+      "loss": 0.723,
+      "step": 13530
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.964444444444445e-06,
+      "loss": 0.7115,
+      "step": 13540
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.955555555555557e-06,
+      "loss": 0.6493,
+      "step": 13550
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.946666666666666e-06,
+      "loss": 0.6918,
+      "step": 13560
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.93777777777778e-06,
+      "loss": 0.6698,
+      "step": 13570
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.928888888888889e-06,
+      "loss": 0.7333,
+      "step": 13580
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.92e-06,
+      "loss": 0.6791,
+      "step": 13590
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.911111111111112e-06,
+      "loss": 0.6802,
+      "step": 13600
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.902222222222223e-06,
+      "loss": 0.7073,
+      "step": 13610
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.893333333333335e-06,
+      "loss": 0.6823,
+      "step": 13620
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.884444444444444e-06,
+      "loss": 0.6737,
+      "step": 13630
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.875555555555556e-06,
+      "loss": 0.6729,
+      "step": 13640
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.866666666666667e-06,
+      "loss": 0.5965,
+      "step": 13650
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.857777777777778e-06,
+      "loss": 0.6568,
+      "step": 13660
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.84888888888889e-06,
+      "loss": 0.624,
+      "step": 13670
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.840000000000001e-06,
+      "loss": 0.6822,
+      "step": 13680
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.831111111111111e-06,
+      "loss": 0.7094,
+      "step": 13690
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.822222222222224e-06,
+      "loss": 0.6948,
+      "step": 13700
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.813333333333334e-06,
+      "loss": 0.6947,
+      "step": 13710
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.804444444444445e-06,
+      "loss": 0.637,
+      "step": 13720
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.795555555555556e-06,
+      "loss": 0.7252,
+      "step": 13730
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.786666666666666e-06,
+      "loss": 0.656,
+      "step": 13740
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.77777777777778e-06,
+      "loss": 0.6919,
+      "step": 13750
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.768888888888889e-06,
+      "loss": 0.6921,
+      "step": 13760
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.76e-06,
+      "loss": 0.7302,
+      "step": 13770
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.751111111111112e-06,
+      "loss": 0.7288,
+      "step": 13780
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.742222222222223e-06,
+      "loss": 0.6818,
+      "step": 13790
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.733333333333334e-06,
+      "loss": 0.6658,
+      "step": 13800
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.724444444444446e-06,
+      "loss": 0.7007,
+      "step": 13810
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.715555555555555e-06,
+      "loss": 0.6918,
+      "step": 13820
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.706666666666669e-06,
+      "loss": 0.6665,
+      "step": 13830
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.697777777777778e-06,
+      "loss": 0.7375,
+      "step": 13840
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.68888888888889e-06,
+      "loss": 0.6759,
+      "step": 13850
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 0.6705,
+      "step": 13860
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.67111111111111e-06,
+      "loss": 0.7336,
+      "step": 13870
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.662222222222224e-06,
+      "loss": 0.726,
+      "step": 13880
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.653333333333333e-06,
+      "loss": 0.7052,
+      "step": 13890
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.644444444444445e-06,
+      "loss": 0.7233,
+      "step": 13900
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.635555555555556e-06,
+      "loss": 0.6375,
+      "step": 13910
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.626666666666668e-06,
+      "loss": 0.6844,
+      "step": 13920
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.617777777777778e-06,
+      "loss": 0.7097,
+      "step": 13930
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.608888888888889e-06,
+      "loss": 0.6683,
+      "step": 13940
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 0.6622,
+      "step": 13950
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.5911111111111115e-06,
+      "loss": 0.65,
+      "step": 13960
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.582222222222223e-06,
+      "loss": 0.6841,
+      "step": 13970
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.573333333333333e-06,
+      "loss": 0.6802,
+      "step": 13980
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.564444444444446e-06,
+      "loss": 0.6763,
+      "step": 13990
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.555555555555556e-06,
+      "loss": 0.6184,
+      "step": 14000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.5466666666666675e-06,
+      "loss": 0.6764,
+      "step": 14010
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.537777777777778e-06,
+      "loss": 0.6801,
+      "step": 14020
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.52888888888889e-06,
+      "loss": 0.6836,
+      "step": 14030
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 0.6634,
+      "step": 14040
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.511111111111111e-06,
+      "loss": 0.6955,
+      "step": 14050
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.502222222222223e-06,
+      "loss": 0.6495,
+      "step": 14060
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.493333333333333e-06,
+      "loss": 0.7518,
+      "step": 14070
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.4844444444444455e-06,
+      "loss": 0.6673,
+      "step": 14080
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.475555555555556e-06,
+      "loss": 0.606,
+      "step": 14090
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.4666666666666675e-06,
+      "loss": 0.6734,
+      "step": 14100
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.457777777777778e-06,
+      "loss": 0.652,
+      "step": 14110
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.44888888888889e-06,
+      "loss": 0.668,
+      "step": 14120
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.440000000000001e-06,
+      "loss": 0.7122,
+      "step": 14130
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.431111111111111e-06,
+      "loss": 0.6522,
+      "step": 14140
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.422222222222223e-06,
+      "loss": 0.6663,
+      "step": 14150
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.413333333333333e-06,
+      "loss": 0.6331,
+      "step": 14160
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.4044444444444455e-06,
+      "loss": 0.6312,
+      "step": 14170
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.395555555555556e-06,
+      "loss": 0.7119,
+      "step": 14180
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.386666666666667e-06,
+      "loss": 0.7851,
+      "step": 14190
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.377777777777778e-06,
+      "loss": 0.6033,
+      "step": 14200
+    }
+  ],
+  "max_steps": 22500,
+  "num_train_epochs": 3,
+  "total_flos": 4.623791925392179e+16,
+  "trial_name": null,
+  "trial_params": null
+}